Browse the corpus

R and β2-AR in baseline renal function cannot be solved by the current study as we cannot compare the urine-concentrating ability of β3-AR−/− and β1-2-AR−/− mice. The 2 strains result from a different genetic background, and early studies showed that renal parameters significantly differ in mice of different strains.37 In line with the stimulatory effect of β3-AR activation on AQP2 subcellular localization and NKCC2 phosphorylation, BRL37344 exerts a potent antidiuretic effect in β3-AR+/+ mice but not in β3AR−/− mice, thus confirming our ex vivo data on its specific action at the β3-AR. The additional finding that in β3-AR+/+ mice, BRL37344 reduces urinary excretion of Na+, K+, and Cl− but not Ca++ and induces a 70% reduction of the urine output, whereas urine osmolality is increased by ∼40% may be explained by assuming that β3-AR stimulation promotes not only water but also salt reabsorption in the kidney. In line with this possibility, the strong reduction of urine output observed in BRL37344-treated mice is independent of the decrease in the GFR.

Premature cardiovascular disease is the leading cause of death in patients with end-stage renal disease (ESRD). Excessive cardiac mortality is thought to be secondary to nonatherosclerotic processes, with sudden cardiac death being a predominant feature.1, 2 Left ventricular hypertrophy (LVH) is common in these patients and is convincingly associated with cardiovascular and all-cause mortality.3, 4, 5 Along with other risk factors, subclinical ischemia and hemodynamic perturbations associated with hemodialysis (HD) are likely to contribute to the ultimate development of LVH, ventricular dilation, cardiac dysfunction, and myocardial fibrosis.6, 7, 8 The development of these adverse features reflects a cardiomyopathy specific to uremia that develops early in chronic kidney disease (CKD).9, 10 Detection and ultimately reversal of the development of this cardiomyopathy are important goals for improving the morbidity and mortality of CKD patients.

ial fibrosis.6, 7, 8 The development of these adverse features reflects a cardiomyopathy specific to uremia that develops early in chronic kidney disease (CKD).9, 10 Detection and ultimately reversal of the development of this cardiomyopathy are important goals for improving the morbidity and mortality of CKD patients. Cardiac magnetic resonance (CMR) imaging is a useful tool in the detection of cardiac disease. One of the advantages of CMR imaging is tissue characterization.11 We and others previously demonstrated myocardial fibrosis in the ESRD population using contrast-enhanced CMR imaging.9, 12 More recently, imaging and quantifying myocardial fibrosis in ESRD have been challenging as gadolinium contrast agent use has been curtailed due to the association between gadolinium and nephrogenic systemic fibrosis.13 Noncontrast native T1 relaxation time is emerging as a viable alternative to gadolinium contrast use and has been shown to correlate with cardiac fibrosis found on tissue histology.14, 15 T1 relaxation time reflects the longitudinal recovery time of hydrogen atoms following their excitation. At any given magnetic field strength, each type of tissue will have its own normal range of values. Normal cardiac tissue will produce a specific range of values; a significant departure from the normal range is thought to represent tissue pathology.16 Increased native T1 time is a surrogate marker of myocardial fibrosis in other disease states such as amyloidosis and hypertrophic obstructive cardiomyopathy.17, 18, 19

ormal cardiac tissue will produce a specific range of values; a significant departure from the normal range is thought to represent tissue pathology.16 Increased native T1 time is a surrogate marker of myocardial fibrosis in other disease states such as amyloidosis and hypertrophic obstructive cardiomyopathy.17, 18, 19 Although increased myocardial native T1 times have been demonstrated in early CKD,20 until now the assessment of native T1 times in ESRD and their comparison with those in a normal healthy population has not been performed. If T1 times are greater in the ESRD population than in healthy volunteers, then they warrant further exploration to determine whether they might also be a marker of cardiac fibrosis in the renal population. Limiting the evolvement of myocardial fibrosis could be an exciting and worthwhile future end point for renal clinical trials.

re greater in the ESRD population than in healthy volunteers, then they warrant further exploration to determine whether they might also be a marker of cardiac fibrosis in the renal population. Limiting the evolvement of myocardial fibrosis could be an exciting and worthwhile future end point for renal clinical trials. In this study, we compared native myocardial global and septal T1 relaxation times as potential markers of diffuse myocardial fibrosis in incident HD patients with those of healthy volunteers (HVs). Ejection fraction is often preserved late into the development of cardiomyopathy.21 Myocardial strain is a more useful early marker of abnormal cardiac muscle structure and function as myocardial tissue compliance will in theory be reduced with increasing fibrosis. As echocardiographic global longitudinal strain (GLS) is predictive of histologic findings of uremic cardiomyopathy and fibrosis in rat CKD models22 as well as being independently predictive of increased mortality,23 we also compared feature-tracking CMR imaging–derived GLS between groups. In the incident HD group, we examined the relationship between these variables and an emerging blood biomarker, galectin-3, which is potentially a surrogate of myocardial fibrosis.24 We also assessed the relationship between 12-lead electrocardiographic abnormalities, CMR imaging–derived cardiac indices, and other laboratory parameters including the more traditional markers of increased cardiac risk—N-terminal β-natriuretic peptide 1 (NT-proBNP) and highly sensitive troponin T (hs-tropT).

brosis.24 We also assessed the relationship between 12-lead electrocardiographic abnormalities, CMR imaging–derived cardiac indices, and other laboratory parameters including the more traditional markers of increased cardiac risk—N-terminal β-natriuretic peptide 1 (NT-proBNP) and highly sensitive troponin T (hs-tropT). Results Participants A total of 61 aged- and sex-matched subjects were enrolled: 28 HVs and 33 HD patients. Baseline demographic characteristics of the HD population and the prevalence of traditional cardiovascular risk factors in this group are shown in Table 1. The HVs had no cardiovascular or systemic disease and had a normal electrocardiogram. They were not treated for hypertension or hypercholesterolemia and were taking no regular medications. Table 2 lists prescribed medication use by the HD patients (Table 2). Left ventricular mass and function Left ventricular (LV) mass was significantly greater in the HD group: the median LV mass indexed to body surface area (LVMI) in the HD group was 69.8 g/m2 (interquartile range, 61.3–88) versus the HV group (55.0 g/m2) (interquartile range, 50.7–62.2) (P < 0.001). One participant in the HV group had LVH (defined as an LVMI >84.1 g/m2 for male participants and >76.4 g/m2 for female participants).25 In the HD group, 14 participants (42.4% of all HD participants) had LVH (P = 0.001).

rtile range, 61.3–88) versus the HV group (55.0 g/m2) (interquartile range, 50.7–62.2) (P < 0.001). One participant in the HV group had LVH (defined as an LVMI >84.1 g/m2 for male participants and >76.4 g/m2 for female participants).25 In the HD group, 14 participants (42.4% of all HD participants) had LVH (P = 0.001). LV ejection fraction was similar between the groups (Table 3). Five participants (15.2%) in the HD group had LV systolic dysfunction, defined as an LV ejection fraction <55%)25 compared with no participants in the HV group (P = 0.056). LV end-diastolic volumes and end-systolic volumes were similar between the groups. Full cardiac parameters of both groups are detailed in Table 3. Native T1 times Global T1 time and septal and midseptal T1 times were greater in the HD group compared with the HV group (Table 3): global T1 time in the HD group, 1171 ± 27 ms versus the HV group, 1154 ± 32 ms, P = 0.025; septal T1 time in the HD group, 1184 ± 29 ms versus the HV group, 1163 ± 30 ms, P = 0.007 (Figure 1); midseptal T1 time in the HD group, 1184 ± 34 ms versus the HV group, 1161 ± 29 ms, P = 0.006.

ared with the HV group (Table 3): global T1 time in the HD group, 1171 ± 27 ms versus the HV group, 1154 ± 32 ms, P = 0.025; septal T1 time in the HD group, 1184 ± 29 ms versus the HV group, 1163 ± 30 ms, P = 0.007 (Figure 1); midseptal T1 time in the HD group, 1184 ± 34 ms versus the HV group, 1161 ± 29 ms, P = 0.006. Correlation of myocardial native T1 times (in milliseconds) with LV mass indices and function In the HD group, LVMI correlated consistently with all measures of T1 times: Pearson’s R for global T1 with an LVMI = 0.452 (P = 0.008), for septal T1 with an LVMI = 0.449 (P = 0.009) (Figure 2), for midseptal T1 with an LVMI = 0.498 (P = 0.003). Septal T1 times positively correlated with end-diastolic volumes: Pearson’s R for global T1 = 0.323, P = 0.067, for septal T1 = 0.380 (P = 0.029), for midseptal T1 = 0.462 (P = 0.007). T1 times did not relate to ejection fraction. Feature tracking–derived strain Peak GLS was reduced in the HD group compared with the HV group (HD group: GLS, −17.7 ± 5.3% vs. HV group, −21.8 ± 6.2%, P = 0.008). In the HD group, GLS correlated with LVMI (Spearman’s R = 0.426, P = 0.013) and negatively correlated with ejection fraction (Pearson’s R = −0.535, P = 0.001). In the HD group, GLS also correlated with increasing end-systolic volume (Spearman’s R = 0.440, P = 0.01), which is associated with a poorer prognosis in HD.26

Feature tracking–derived strain Peak GLS was reduced in the HD group compared with the HV group (HD group: GLS, −17.7 ± 5.3% vs. HV group, −21.8 ± 6.2%, P = 0.008). In the HD group, GLS correlated with LVMI (Spearman’s R = 0.426, P = 0.013) and negatively correlated with ejection fraction (Pearson’s R = −0.535, P = 0.001). In the HD group, GLS also correlated with increasing end-systolic volume (Spearman’s R = 0.440, P = 0.01), which is associated with a poorer prognosis in HD.26 Unlike a previous study of early CKD using feature tracking–derived strain methods,20 we found no difference in early diastolic strain rate or strain rate between the HD and HV groups. There was no correlation between any marker of strain and native T1 values. Relationship of CMR imaging findings to hs-tropT, NT-proBNP, and galectin-3 In the HD group, LVMI correlated with NT-proBNP (Spearman’s R = 0.365, P = 0.044). LVMI did not correlate with hs-tropT or galectin-3. Septal T1 correlated with predialysis hs-tropT (Spearman’s R = 0.397, P = 0.027) but not with NT-proBNP or galectin-3. GLS showed a trend toward correlation with galectin-3 (Spearman’s R = 0.344, P = 0.05). There was no correlation between GLS and NT-proBNP or hs-tropT.

MI did not correlate with hs-tropT or galectin-3. Septal T1 correlated with predialysis hs-tropT (Spearman’s R = 0.397, P = 0.027) but not with NT-proBNP or galectin-3. GLS showed a trend toward correlation with galectin-3 (Spearman’s R = 0.344, P = 0.05). There was no correlation between GLS and NT-proBNP or hs-tropT. 12-Lead electrocardiogram and CMR imaging findings Twenty-eight (85%) HD participants had a pre-HD electrocardiogram available for analysis, and 27 of these subjects (96.4%) were in sinus rhythm. One participant had right bundle branch block. The median corrected QT interval was 435 ms. Significantly, there were interrelations of the Q-T interval time (QTc) with septal T1 (Spearman’s R = 0.376, P = 0.045) and NT-proBNP (Spearman’s R = 0.472, P = 0.011). There were no demonstrated relationships between LVMI or strain and QTc.

t had right bundle branch block. The median corrected QT interval was 435 ms. Significantly, there were interrelations of the Q-T interval time (QTc) with septal T1 (Spearman’s R = 0.376, P = 0.045) and NT-proBNP (Spearman’s R = 0.472, P = 0.011). There were no demonstrated relationships between LVMI or strain and QTc. Relationship of interdialytic fluid gains and ultrafiltration volumes with CMR imaging findings The weight difference between the “dry weight” at the end of dialysis the day before imaging and the weight at the time of scanning correlated with LVM (but not LVMI) (Spearman’s R = 0.384, P = 0.027). There was no correlation between T1 times and this postdialysis weight gain (global T1 and prescan weight change: Spearman’s R = 0.052, P = 0.776). Strain also was not related to this weight change. An average of participants’ HD ultrafiltration volumes for the 30 days before imaging correlated with LVM, but not LVMI (Spearman’s R = 0.422 P = 0.015). There was no relationship between any measure of T1 time and 30-day mean ultrafiltration (global T1 and mean ultrafiltration Spearman’s R = 0.114, P = 0.529). There was no correlation demonstrated between any measure of strain and ultrafiltration volumes.

related with LVM, but not LVMI (Spearman’s R = 0.422 P = 0.015). There was no relationship between any measure of T1 time and 30-day mean ultrafiltration (global T1 and mean ultrafiltration Spearman’s R = 0.114, P = 0.529). There was no correlation demonstrated between any measure of strain and ultrafiltration volumes. Influence of blood pressure and dialysis adequacy on CMR imaging findings In the HD group LVM (but not LVMI) correlated with diastolic blood pressure, correlation of LVM with systolic BP did not reach statistical significance (systolic blood pressure: Spearman’s R = 0.334, P = 0.057; diastolic blood pressure: R = 0.403, P = 0.02). There was no correlation demonstrated between global, septal, or midseptal T1 times or strain and blood pressure. There was a negative correlation between dialysis urea reduction ratio and LVMI (Pearson’s R = −0.560, P = 0.001). Further supporting the implication that HD adequacy was strongly correlated with LVMI, there were correlations between adjusted calcium and phosphate values and LVMI (calcium: Spearman’s R = −0.462, P = 0.007; phosphate: Pearson’s R = 0.358, P = 0.041). T1 times did not relate to the urea reduction ratio. GLS negatively correlated with dialysis urea reduction ratios (Spearman’s R = −0.348, P = 0.047). T1 times and strain were not related to calcium or phosphate.

Influence of blood pressure and dialysis adequacy on CMR imaging findings In the HD group LVM (but not LVMI) correlated with diastolic blood pressure, correlation of LVM with systolic BP did not reach statistical significance (systolic blood pressure: Spearman’s R = 0.334, P = 0.057; diastolic blood pressure: R = 0.403, P = 0.02). There was no correlation demonstrated between global, septal, or midseptal T1 times or strain and blood pressure. There was a negative correlation between dialysis urea reduction ratio and LVMI (Pearson’s R = −0.560, P = 0.001). Further supporting the implication that HD adequacy was strongly correlated with LVMI, there were correlations between adjusted calcium and phosphate values and LVMI (calcium: Spearman’s R = −0.462, P = 0.007; phosphate: Pearson’s R = 0.358, P = 0.041). T1 times did not relate to the urea reduction ratio. GLS negatively correlated with dialysis urea reduction ratios (Spearman’s R = −0.348, P = 0.047). T1 times and strain were not related to calcium or phosphate. Relationship of other factors known to contribute to LVH in ESRD to CMR imaging findings HD patients with anemia (defined as hemoglobin <100 mg/dl) had a higher LVMI than those without. The median LVMI in the 6 anemic HD patients was 92.1 g/m2 compared with 65.7 g/m2 in those with a hemoglobin ≥100 mg/dl (P = 0.008). Our study was not powered to detect a potential small difference in T1 times in anemic patients. Median global T1 time in patients with anemia was 1190 ms (IQR, 1164.9–1209.9) versus 1171 ms (IQR, 1150.9–1187.1) (P = 0.189). GLS was reduced in anemic patients: GLS: anemia = −13.5 ± 3.9% versus −18.6 ± 5.1% (P = 0.025).

Our study was not powered to detect a potential small difference in T1 times in anemic patients. Median global T1 time in patients with anemia was 1190 ms (IQR, 1164.9–1209.9) versus 1171 ms (IQR, 1150.9–1187.1) (P = 0.189). GLS was reduced in anemic patients: GLS: anemia = −13.5 ± 3.9% versus −18.6 ± 5.1% (P = 0.025). Of the HD participants, 81.8% were receiving dialysis through an arteriovenous fistula. There was no demonstrated difference in LVMI, strain, or T1 times across different types of dialysis access. No correlation was seen between age or dialysis vintage and LVMI, strain, or T1 time; 24.2% of participants were diabetic. There was no statistical difference in LVMI, any T1 measurements, or strain between those HD participants with and without diabetes. T1 image quality and reproducibility Of a total 976 T1 regions of interest (Figure 3), 808 (82.8%) were considered suitable for analysis. The greatest reliability of measurement was seen for septal T1 measurement. Septal T1 segments were less affected by artifact than other segments. The intraclass correlation coefficient for reliability of global T1 measurement was 0.872 (95% confidence interval 0.630–0.914, P < 0.001); for septal T1, it was 0.941 (95% CI 0.871–0.974, P < 0.001); and for midseptal T1, it was 0.901 (95% CI 0.786–0.955, P < 0.001).

were less affected by artifact than other segments. The intraclass correlation coefficient for reliability of global T1 measurement was 0.872 (95% confidence interval 0.630–0.914, P < 0.001); for septal T1, it was 0.941 (95% CI 0.871–0.974, P < 0.001); and for midseptal T1, it was 0.901 (95% CI 0.786–0.955, P < 0.001). Discussion This study is the first to compare native T1 times in HD participants with those of HVs. We demonstrated that T1 times are significantly prolonged in the HD population and correlate with increased LVMI. The increased T1 times demonstrated in the HD population may be representative of the cardiac fibrosis known to be found in ESRD. Native T1 mapping might be a novel way to quantify cardiac tissue abnormalities in HD. T1 times could be further investigated as a future surrogate end point in renal clinical trials. The utility of T1 times is further evidenced by their association with LVMI, suggesting that as LVH progresses in severity, the underlying tissue abnormalities that lead to longer T1 times increase in parallel. Although it is our assertion that these prolonged T1 times might be representative of myocardial fibrosis, without a tissue diagnosis, this cannot be proved. It should be noted that in our study that the proportion of LVH was lower than in some other HD studies. These were HD patients with a short dialysis vintage with a mean duration of renal replacement therapy of 5.5 months.

ight be representative of myocardial fibrosis, without a tissue diagnosis, this cannot be proved. It should be noted that in our study that the proportion of LVH was lower than in some other HD studies. These were HD patients with a short dialysis vintage with a mean duration of renal replacement therapy of 5.5 months. Rationale for global and septal T1 analysis The majority of previous studies considering native T1 times in other populations (as well as a single study in early CKD)20 consider increased septal T1 times to be representative of a diffuse fibrotic process. In most populations, it is possible to verify this through correlation with diffuse fibrosis seen after gadolinium contrast administration. As routine gadolinium-based contrast agent use has been curtailed in the ESRD population, verification of this assumption by this method is not practical. Additionally, although the correlation of native septal T1 times with diffuse cardiac fibrosis has also been borne out by cardiac tissue biopsy in several populations14, 15 and a previous LV biopsy study confirmed histologic evidence of diffuse fibrosis in ESRD,27 no study relating native septal T1 time in ESRD to cardiac tissue histology has ever been performed.

ive septal T1 times with diffuse cardiac fibrosis has also been borne out by cardiac tissue biopsy in several populations14, 15 and a previous LV biopsy study confirmed histologic evidence of diffuse fibrosis in ESRD,27 no study relating native septal T1 time in ESRD to cardiac tissue histology has ever been performed. Therefore, in order to ensure that we did not falsely assume that a localized high septal T1 time was representative of a diffuse process, we measured T1 times throughout 3 short-axis slices: basal, mid, and apical. Using the 16-segment model of the American Heart Association,28 we calculated an average global T1 time from all 16 regions of interest. Using this method, we were able to demonstrate that truly global T1 times were higher in the HD population than in healthy controls. Global T1 times were very closely correlated with septal T1 times (Pearson’s R = 0.885, P < 0.001). We analyzed midseptal T1 times alone on the assumption that the apical short-axis slice might be more susceptible to motion artifact than the midslice; however, midseptal T1 times were not of any greater value than septal T1 time. Furthermore, using only 2 segments to calculate midseptal T1 increases the risk of regional ischemia influencing results.

lone on the assumption that the apical short-axis slice might be more susceptible to motion artifact than the midslice; however, midseptal T1 times were not of any greater value than septal T1 time. Furthermore, using only 2 segments to calculate midseptal T1 increases the risk of regional ischemia influencing results. Influence of traditional ischemic heart disease on T1 times Native T1 times are known to be locally higher in areas of cardiac injury, for example, after a chronic myocardial infarction.29 Our HD population was unselected, and a proportion of patients had known ischemic heart disease. Before any knowledge of the association of gadolinium contrast use and nephrogenic systemic fibrosis in ESRD, it was described using gadolinium contrast that there are 2 patterns of cardiac fibrosis seen in ESRD: that of the traditional ischemic heart disease with subendocardial fibrosis and that of a more diffuse uremic cardiomyopathy.9, 12 We do not believe that the prolonged T1 times in the HD group were a consequence of traditional ischemic heart disease. Four HD participants had a history of myocardial infarction. Two observers independently reviewed cine images of these participants to exclude any patients who had thinned akinetic myocardial segments, which is commonly accepted as a surrogate of transmural chronic myocardial infarction. The observers agreed that there was no evidence of myocardial wall thinning on any of their images. Furthermore, if these 4 participants were excluded entirely, then the average global, septal, and midseptal T1 times for the HD group were essentially unaltered (global T1 with myocardial infarction patients excluded = 1170 ± 27 ms vs. 1171 ± 27 ms with myocardial infarction patients included).

on any of their images. Furthermore, if these 4 participants were excluded entirely, then the average global, septal, and midseptal T1 times for the HD group were essentially unaltered (global T1 with myocardial infarction patients excluded = 1170 ± 27 ms vs. 1171 ± 27 ms with myocardial infarction patients included). Arrhythmia, hs-tropT, and T1 time Interestingly, we saw a relationship between septal T1 and predialysis hs-tropT. Septal times also related to QTc. Increased corrected Q-T interval times are associated with an increased risk of sudden cardiac death.30 The increased incidence of sudden cardiac death is a poorly understood phenomenon in ESRD that cannot be entirely attributed to electrolyte disturbances.30 A study in rats showed that rats with induced CKD develop both LVH and increased susceptibility to arrhythmia.31 In this study, there were also some early signs that cardiac fibrosis began to develop in the rats. The demonstrated correlation between septal T1 times and troponin T and septal times and QTc in our study warrants further exploration. GLS, ejection fraction, and early myocardial dysfunction Our study showed no difference in ejection fraction between HV and HD groups. It is well-known that ejection fraction is often preserved until late in the development of cardiomyopathy, and thus it is not a reliable primary end point for renal clinical trials.21 Recent studies showed that echocardiographic myocardial strain is an independent predictor of increased mortality in CKD populations.22, 23

s well-known that ejection fraction is often preserved until late in the development of cardiomyopathy, and thus it is not a reliable primary end point for renal clinical trials.21 Recent studies showed that echocardiographic myocardial strain is an independent predictor of increased mortality in CKD populations.22, 23 In this study, there was a trend toward a correlation of GLS with galectin-3 (Spearman’s R = 0.344, P = 0.05). Given our small sample size, we consider this result to be of some interest. Galectin-3 is an established biomarker of myocardial fibrosis. The trend identified makes sense as GLS measured by speckle tracking echocardiography is predictive of histologic findings of uremic cardiomyopathy and cardiac fibrosis in rats with CKD.22

r small sample size, we consider this result to be of some interest. Galectin-3 is an established biomarker of myocardial fibrosis. The trend identified makes sense as GLS measured by speckle tracking echocardiography is predictive of histologic findings of uremic cardiomyopathy and cardiac fibrosis in rats with CKD.22 Assessment of GLS using speckle tracking echocardiography is recommended in recent guidelines for the quantitative assessment of LV function.32 However, echocardiography is not an ideal tool for clinical trials in ESRD because dialysis-associated fluid shifts can lead to overestimation of ventricular indices and variations in assessment of cardiac function.33 Use of a single imaging modality to assess outcomes in the HD population is preferable, and we consider CMR imaging to be the modality of choice. In particular because, more recently, myocardial strain quantified by CMR imaging has been found to be associated with adverse outcomes.34 Although there was no correlation of T1 times with GLS, our findings of a difference in GLS between the HD and HV groups, as well as the correlation of GLS with LVM in HD, support the assertion that altered myocardial strain mechanics have some role in the process of development of LVH and potentially fibrosis in CKD patients. Whether abnormal GLS is a cause or a consequence of cardiomyopathy development needs to be further investigated. Similarly, the lack of a correlation of T1 times with GLS in this study should be considered further. At present, it is not clear whether our study was underpowered to detect a relationship between the 2 or whether they reflect slightly different aspects of pathologic processes in the heart. In any case, they both have their own merits; GLS is a dynamic measure, whereas T1 time is arguably a more fixed quantity.

rther. At present, it is not clear whether our study was underpowered to detect a relationship between the 2 or whether they reflect slightly different aspects of pathologic processes in the heart. In any case, they both have their own merits; GLS is a dynamic measure, whereas T1 time is arguably a more fixed quantity. Limitations Although our study numbers are relatively small and all participants came from a single center, we have been able to present a well-characterized group of dialysis patients with a short dialysis vintage. The incidence of diabetes in our study population (24.2%) was less than that in many typical HD populations, and our study population was not racially diverse. We were unable to perform multivariable adjustments to assess for independence of any of our findings. Our study would have been strengthened by the addition of a control population of hypertensive patients with LVH and without renal disease. However, there is some evidence from other studies that included a control group that in contrast to LVMI, T1 times are independent of blood pressure.20 The fact that T1 times and strain were not related to blood pressure in our study provides some further reassurance in this regard.

with LVH and without renal disease. However, there is some evidence from other studies that included a control group that in contrast to LVMI, T1 times are independent of blood pressure.20 The fact that T1 times and strain were not related to blood pressure in our study provides some further reassurance in this regard. Without biopsy confirmation, we cannot be certain that the cardiac abnormalities identified in this population are representative of fibrosis. It is recognized that water content can prolong T1 times, and this is undoubtedly a limitation of our study as we cannot be sure of the precise influence of this on our results.35 However, we saw no correlation with weight gain between HD and imaging with T1 times. Similarly, there was no correlation between T1 time and ultrafiltration volumes. Both of these markers, which are representative of changing fluid status, correlated with LVM. Overall, we believe that these findings support our assertion that increased T1 times in the HD population are likely to reflect tissue abnormalities (potentially fibrosis). Previous biopsies of the uremic hearts have shown extensive myocardial fibrosis, which would be consistent with our conclusions.27

lated with LVM. Overall, we believe that these findings support our assertion that increased T1 times in the HD population are likely to reflect tissue abnormalities (potentially fibrosis). Previous biopsies of the uremic hearts have shown extensive myocardial fibrosis, which would be consistent with our conclusions.27 Implications of our study findings T1 times are a potentially novel way to quantify cardiac tissue abnormalities in ESRD. In the future, they may be conclusively demonstrated to be a surrogate marker of fibrosis. Nephrologists should note that normal values will vary from scanner to scanner. The next steps for development of T1 mapping as a widely applicable clinical tool will involve each center developing robust normal values for each of their scanners. Phantom work will be required before any future multicenter collaborations run successfully. We showed that T1 times are abnormally high in the HD population. Perhaps in the future, change in T1 times could potentially be a primary outcome measure in renal clinical trials. However, a number of questions need to be answered before this can happen: what is the natural history of T1 times throughout the progression of CKD, do T1 times change after renal transplantation, how exactly are T1 times affected by fluid status variation, and are increased T1 times associated with increased risk of future cardiovascular events and mortality. To date, increased T1 times have been shown to be predictive of mortality in amyloidosis36; the utility of T1 times to predict mortality in other populations remains to be proved.

affected by fluid status variation, and are increased T1 times associated with increased risk of future cardiovascular events and mortality. To date, increased T1 times have been shown to be predictive of mortality in amyloidosis36; the utility of T1 times to predict mortality in other populations remains to be proved. Methods Participants Sixty-one participants were included in the study. All participants were older than 18 years of age and had no contraindications to CMR scanning. HD participants were eligible if they had commenced dialysis within the past 12 months; baseline scans for all eligible patients in 2 concurrent studies recruiting in Glasgow were used. In total, 33 HD patients were recruited from the Cardiac Uraemic fibrosis Detection in DiaLysis patiEnts study (CUDDLE study, ISRCTN99591655), and the ALTERED study (Does ALlopurinol Regress LefT Ventricular Hypertrophy in End Stage REnal Disease, NCT01951404). A total of 28 age- and sex-matched healthy volunteers were recruited from a Glasgow-based study examining variations in native T1 relaxation times in healthy adults.37 Healthy volunteers within 5 years of age of each HD participant were blindly selected from a list of HV participants that detailed only the HV participants’ age and sex. All participants provided written informed consent, and all studies were approved by local ethics committees (CUDDLE: West of Scotland Research Ethics Service, reference 13/WS/0301, healthy volunteers: 11/AL/0190; ALTERED: East of Scotland Research Ethics Service, 13/ES/0051).

Methods Participants Sixty-one participants were included in the study. All participants were older than 18 years of age and had no contraindications to CMR scanning. HD participants were eligible if they had commenced dialysis within the past 12 months; baseline scans for all eligible patients in 2 concurrent studies recruiting in Glasgow were used. In total, 33 HD patients were recruited from the Cardiac Uraemic fibrosis Detection in DiaLysis patiEnts study (CUDDLE study, ISRCTN99591655), and the ALTERED study (Does ALlopurinol Regress LefT Ventricular Hypertrophy in End Stage REnal Disease, NCT01951404). A total of 28 age- and sex-matched healthy volunteers were recruited from a Glasgow-based study examining variations in native T1 relaxation times in healthy adults.37 Healthy volunteers within 5 years of age of each HD participant were blindly selected from a list of HV participants that detailed only the HV participants’ age and sex. All participants provided written informed consent, and all studies were approved by local ethics committees (CUDDLE: West of Scotland Research Ethics Service, reference 13/WS/0301, healthy volunteers: 11/AL/0190; ALTERED: East of Scotland Research Ethics Service, 13/ES/0051). Inclusion and exclusion criteria The HV group had no cardiovascular or systemic disease and had normal electrocardiograms. They were not treated for hypertension or hypercholesterolemia and were taking no regular medications. HD patients were excluded if they had atrial fibrillation (as this makes magnetic resonance imaging and electrocardiographic gating difficult). All HD patients had been receiving renal replacement therapy for less than 1 year. ALTERED patients were not taking allopurinol, had a life expectancy >1 year, and did not have echocardiogram-defined LV systolic dysfunction.

rillation (as this makes magnetic resonance imaging and electrocardiographic gating difficult). All HD patients had been receiving renal replacement therapy for less than 1 year. ALTERED patients were not taking allopurinol, had a life expectancy >1 year, and did not have echocardiogram-defined LV systolic dysfunction. Magnetic resonance image acquisition All participants underwent 3-T CMR imaging (MAGNETOM Verio, Siemens Healthcare, Erlangen, Germany) at the British Heart Foundation Clinical Research Centre, University of Glasgow. In the HD patients, CMR imaging was consistently performed on a postdialysis day. Baseline ALTERED study magnetic resonance imaging scans were used before any study intervention. A double radiofrequency coil array (anterior and posterior) was used. The scans were electrocardiographically gated. The imaging protocol included cine magnetic resonance with steady-state free precession and T1 mapping sequences.35 The full left ventricle was captured in a cine short-axis stack. Cine acquisition parameters included 41.4-ms repetition time, 40° flip angle, 1.51-ms echo time, 256 × 173-pixel matrix, 1.5 × 1.3 × 8-mm voxel size, 8-mm slice thickness, and 977-Hz/pixel bandwidth.

teady-state free precession and T1 mapping sequences.35 The full left ventricle was captured in a cine short-axis stack. Cine acquisition parameters included 41.4-ms repetition time, 40° flip angle, 1.51-ms echo time, 256 × 173-pixel matrix, 1.5 × 1.3 × 8-mm voxel size, 8-mm slice thickness, and 977-Hz/pixel bandwidth. Basal, mid, and apical T1 maps were acquired in 3 short-axis slices using a motion-corrected, optimized, modified Look-Locker inversion recovery investigational prototype sequence without contrast administration (Siemens Healthcare, works-in-progress method 448). T1 imaging parameters included 267.84-ms repetition time, 35° flip angle, 1.06-ms echo time, 100-ms T1 of the first experiment, 80-ms T1 increment, 124 × 192 pixel matrix, 2.2 × 1.8 × 8.0-mm spatial resolution, 930-Hz/pixel bandwidth, and 17-heartbeat (range, 12–18 s) scan time. Image analysis LVM and function A single observer analyzed anonymized images in a random order to determine cardiac indices including LVM and function by manually tracing endocardial and epicardial borders at end-systole and end-diastole on the short-axis cine images as according to well-established protocols.37 End-systolic and end-diastolic volumes and LVM were calculated and indexed to body surface area (using a weight acquired immediately prescan) using the Siemens Argus analysis software.

dial and epicardial borders at end-systole and end-diastole on the short-axis cine images as according to well-established protocols.37 End-systolic and end-diastolic volumes and LVM were calculated and indexed to body surface area (using a weight acquired immediately prescan) using the Siemens Argus analysis software. T1 maps On the raw T1 images, LV contours were defined and copied onto the color-enhanced spatially coregistered maps. Using the anterior right ventricular-LV insertion post as a reference, T1 maps were segmented according to the American Heart Association 16-segment model.28 Segmental American Heart Association regions of interest were delineated by user-defined semiautomated border delineation (Siemens Argus analysis software). The regions of interest were standardized to be of similar size and shape. T1 times were measured in each of the 16 segments, with care taken to delineate regions of interest with adequate margins of separation from tissue interfaces such as between the blood pool and myocardium. A typical T1 map is shown in Figure 3.

The regions of interest were standardized to be of similar size and shape. T1 times were measured in each of the 16 segments, with care taken to delineate regions of interest with adequate margins of separation from tissue interfaces such as between the blood pool and myocardium. A typical T1 map is shown in Figure 3. Each individual segment was assessed for the presence or absence of susceptibility and motion artifacts. After removal of any segments affected by artifact, a global T1 time was calculated from the mean of all remaining segments. A septal T1 time was calculated by averaging the acceptable anteroseptal, inferoseptal, and septal American Heart Association segments (segment numbers 2, 3, 8, 9, and 14). Midseptal T1 time was derived from an average of included septal segments from the midshort-axis slice (segment numbers 8 and 9). T1 map reproducibility was assessed by blinded reanalysis of 25 randomly selected images by the same observer. Strain Dedicated feature-tracking software (Diogenes Image Arena, Munich, Germany) was used on the horizontal long-axis cine acquisition. In a method previously described,38, 39 the end-diastolic frame was identified for each image, and endocardial borders were delineated. The delineated contour was then automatically propagated throughout the cardiac cycle. GLS, strain rate, and early diastolic strain rate were then calculated. Intraobserver reproducibility was checked by blinded reanalysis of a proportion of images (n = 20).

fied for each image, and endocardial borders were delineated. The delineated contour was then automatically propagated throughout the cardiac cycle. GLS, strain rate, and early diastolic strain rate were then calculated. Intraobserver reproducibility was checked by blinded reanalysis of a proportion of images (n = 20). 12-Lead electrocardiograms In the HD group, predialysis 12-lead electrocardiograms were obtained for 28 participants. The underlying rhythm was recorded. Corrected QTc was noted from the electronic print out of each electrocardiogram. HD patient biomarkers and other clinical parameters Frozen stored predialysis blood samples were obtained in the HD group for analysis of hs-tropT, NT-proBNP (e411, Roche Diagnostics, Burgess Hill, UK), and galectin-3 (Bio-Techne, Abingdon, UK). All were analyzed using the manufacturer-recommended protocols and calibrations. Available collected blood tests from the time of imaging, including hemoglobin, urea reduction ratios, albumin, C-reactive protein, phosphate, parathyroid hormone, glucose, predialysis creatinine and potassium, lipid profile, as well as each HD participant’s medical and dialysis history including ultrafiltration volumes and postdialysis weights were obtained from electronic records.

ng hemoglobin, urea reduction ratios, albumin, C-reactive protein, phosphate, parathyroid hormone, glucose, predialysis creatinine and potassium, lipid profile, as well as each HD participant’s medical and dialysis history including ultrafiltration volumes and postdialysis weights were obtained from electronic records. Statistics All statistical analyses were performed using SPSS version 22 (Armonk, NY) and STATA 13 (StataCorp, College Station, TX). Paired t tests (for parametric data) and Mann-Whitney U tests (for nonparametric data) were used to compare continuous indices between HVs and HD patients. Categorical data were assessed using the χ2 test or the Fisher exact test, as appropriate. Correlations within each group between continuous indices were assessed using Pearson’s and Spearman’s correlation coefficients for parametric and nonparametric data, respectively. At the time of the start of this study, there were few data to inform a power calculation using native T1 times at 3-T CMR imaging. Based on our normal volunteer study,37 to detect a 30-ms difference in mean native T1 time between a group of HVs and patients with ESRD, at 90% power and a probability of a type 1 error of 0.05 would require 22 per group, based on a standard deviation in T1 time of 30 ms in each group. Disclosure The University of Glasgow holds a research agreement with Siemens Healthcare (UK). All the other authors declared no competing interests.

At the time of the start of this study, there were few data to inform a power calculation using native T1 times at 3-T CMR imaging. Based on our normal volunteer study,37 to detect a 30-ms difference in mean native T1 time between a group of HVs and patients with ESRD, at 90% power and a probability of a type 1 error of 0.05 would require 22 per group, based on a standard deviation in T1 time of 30 ms in each group. Disclosure The University of Glasgow holds a research agreement with Siemens Healthcare (UK). All the other authors declared no competing interests. Acknowledgments This study was funded by Kidney Research UK (Research Innovation Grant IN02/2013) and the British Heart Foundation (Project Grant PG/12/72/29743). ER is funded through the same British Heart Foundation Project Grant. KM is supported by a Fellowship from the British Heart Foundation (FS/15/5431639). see commentary on page 729 Figure 1 Boxplot comparing septal T1 times in healthy volunteers and hemodialysis (HD) patients. Figure 2 Scatterplot of septal T1 times against left ventricular mass indexed to body surface area (LVMI) in hemodialysis patients. g m-2, grams per meter squared. Figure 3 A typically segmented T1 map of a basal myocardial slice in a hemodialysis patient. Min/Max, minimum/maximum. Table 1 Baseline Demographic Characteristics and Clinical Data for HD Patients

Figure 2 Scatterplot of septal T1 times against left ventricular mass indexed to body surface area (LVMI) in hemodialysis patients. g m-2, grams per meter squared. Figure 3 A typically segmented T1 map of a basal myocardial slice in a hemodialysis patient. Min/Max, minimum/maximum. Table 1 Baseline Demographic Characteristics and Clinical Data for HD Patients Variable All HD Patients (N = 33) Primary renal diagnosis (%) Diabetic nephropathy 24.2 [8] Polycystic kidney disease 15.2 [5] Renovascular disease 12.1 [4] Glomerulonephritis 24.2 [8] Unknown cause 12.1 [4] Other known 12.1 [4] Length of time on HD (mo) 5.5 ± 2.7 Mean UF volume (l) 1.7 ± 1.0 Dialysis access (%) Fistula 81.8 [27] Graft 3 [1] Tunneled line 15.2 [5] Diabetes [%] 24.2 [8] Hypertension [%] 60.6 [20] Myocardial infarction [%] 12.1 [4] Ischemic heart disease [%] 18.2 [6] Stroke [%] 15.2 [5] Peripheral vascular disease [%] 9.1 [3] Systolic blood pressure (mm Hg) 138 (131–155) Diastolic blood pressure (mm Hg) 78 (67–83) Hemoglobin (mg/dl) 111 (103–118) Urea reduction ratio (%) 73 (68–78) Albumin (g/l) 35 (32–36) C-reactive protein (mmol/l) 8 (3–14) Phosphate (mmol/l) 1.73 (1.33–2.18) Corrected calcium (mmol/l) 2.35 (2.24–2.39) Parathyroid hormone (mmol/l) 48.1 (24.5–85.9) Galectin-3 (ng/ml) 17.5 (13.4–22.0) hs-Troponin Ta (pg/ml) 33.7 (23.5–46.9) NT-BNPa (pg/ml) 1934 (1111–5111.5) ECG QTcb (m/s) 435 (412.8–453) ECG, electrocardiogram; HD, hemodialysis; hs-Troponin T, highly sensitive Troponin T; NT-proBNP, N-terminal β-natriuretic peptide 1; UF, ultrafiltration.

mmol/l) 48.1 (24.5–85.9) Galectin-3 (ng/ml) 17.5 (13.4–22.0) hs-Troponin Ta (pg/ml) 33.7 (23.5–46.9) NT-BNPa (pg/ml) 1934 (1111–5111.5) ECG QTcb (m/s) 435 (412.8–453) ECG, electrocardiogram; HD, hemodialysis; hs-Troponin T, highly sensitive Troponin T; NT-proBNP, N-terminal β-natriuretic peptide 1; UF, ultrafiltration. All data presented as percentage [number of participants], median (interquartile range), or mean ± SD, as appropriate. a hs-Troponin T and NT-BNP values were available for 31 HD participants. b QTc was available for 28 HD participants. Table 2 Prescribed Medications in the HD Participants Medication HD Participants Taking (N = 33) Erythropoietin 78.8 (26) Beta-blocker 57.6 (19) Aspirin 24.2 (8) Clopidogrel 24.2 (8) ACE inhibitor 9.1 (3) Diuretics 27.3 (9) Calcium channel blockers 39.4 (13) Alpha-blockers 6.1 (2) Statin 54.5 (18) ACE, angiotensin-converting enzyme; HD, hemodialysis. Data shown as percentage (number of participants). Table 3 Patient Characteristics and Cardiac Parameters of HV and HD Patients

Medication HD Participants Taking (N = 33) Erythropoietin 78.8 (26) Beta-blocker 57.6 (19) Aspirin 24.2 (8) Clopidogrel 24.2 (8) ACE inhibitor 9.1 (3) Diuretics 27.3 (9) Calcium channel blockers 39.4 (13) Alpha-blockers 6.1 (2) Statin 54.5 (18) ACE, angiotensin-converting enzyme; HD, hemodialysis. Data shown as percentage (number of participants). Table 3 Patient Characteristics and Cardiac Parameters of HV and HD Patients Variable Healthy Volunteers (N = 28) HD Patients (N = 33) P Value Age (yr) 60 (53.8–72.3) 56 (50–71) 0.562 Male [%] 57.1 [16] 57.6 [19] 0.973 Weight (kg) 79 (68.8–89) 73.9 (63–83) 0.343 BMI (kg/m2) 25.6 (24.1–29.5) 27.7 (23.1–30.5) 0.772 Ethnicity White 96.4 [27] 90.9 [30] 0.618 South Asian 3.6 [1] 9.1 [3] 0.618 Global T1 (ms) 1154 ± 32 1171± 27 0.025 Septal T1 (ms) 1163 ± 30 1184 ± 29 0.007 Midseptal T1 (ms) 1161 ± 29 1184 ± 34 0.006 LVM (g) 107.7 (89.9–115.6) 131.7 (105.8–152.6) 0.001 LVMI (g/m2) 55.0 (50.7–62.2) 69.8 (61.3–88) 0.001 LVH (%) 3.6 [1] 42.4 [14] 0.001 EDV (ml) 148.4 (128.2–168.9) 142.9 (133.6–163.8) 0.856 EDVI (ml/m2) 77.4 ± 9.7 83.8 ± 23.4 0.180 ESV (ml) 56.8 (43.4–62.4) 51.2 (41.8–64.1) 0.783 ESVI (ml/m2) 28.4 ± 6.0 32.0 ± 17.5 0.307 LV Dilation 0 [0] 9.1 [3] 0.243 Stroke Volume (ml) 94.4 (75.8–103.4) 92.8 (74.2–112.2) 0.954 Ejection Fraction (%) 63.3 ± 5.2 63.2 ± 9.3 0.963 LVSD (%) 0 [0] 15.2 [5] 0.056 GLS (%) -21.8 ± 6.2 -17.7 ± 5.3 0.007 Strain Rate (s−1) 1.06 ± 0.37 0.95 ± 0.24 0.140 EDSR (s-1) 0.97 ± 0.36 1.03 ± 0.36 0.473 BMI, body mass index; EDSR, early diastolic strain rate; EDV, end-diastolic volume; EDVI, end-diastolic volume indexed to body surface area; ESV, end-systolic volume; ESVI, end-systolic volume indexed to body surface area; GLS, global longitudinal strain; HD, hemodialysis; HV, health volunteer; LV dilation, left ventricular dilation; LVH, left ventricular hypertrophy; LVM, left ventricular mass; LVMI, left ventricular mass indexed to body surface area; LVSD, left ventricular systolic dysfunction.

olic volume indexed to body surface area; GLS, global longitudinal strain; HD, hemodialysis; HV, health volunteer; LV dilation, left ventricular dilation; LVH, left ventricular hypertrophy; LVM, left ventricular mass; LVMI, left ventricular mass indexed to body surface area; LVSD, left ventricular systolic dysfunction. All data are shown as mean ± SD, median (interquartile range), or percentage [number of participants], as appropriate.

Urinary tract infections (UTIs) remain among the most common human infectious diseases worldwide (∼150–250 million cases globally per year).1, 2, 3 UTIs present as a wide spectrum of diseases, including bladder infection (cystitis), kidney infection (pyelonephritis), and associated renal damage (e.g., renal fibrosis). Kidney infection is usually caused by ascending infections of the lower urinary tract. Recurrent or long-standing (chronic) kidney infections can result in renal tubulointerstitial fibrosis, which is a much more common situation in children or in patients with diabetes mellitus or urinary obstructions. Although antibiotics are available to treat the disease, a number of challenges remain, including frequent recurrence, persistence of infection, and the increasing risk for resistance to antibiotics.1, 4 It is therefore imperative to improve our current understanding of the pathogenesis of UTIs and develop novel therapeutic strategies to improve current treatment.

e disease, a number of challenges remain, including frequent recurrence, persistence of infection, and the increasing risk for resistance to antibiotics.1, 4 It is therefore imperative to improve our current understanding of the pathogenesis of UTIs and develop novel therapeutic strategies to improve current treatment. Uropathogenic Escherichia coli (UPEC) is the primary cause of UTIs, and most UPEC express a variety of fimbriae (e.g., P, type 1) that enable them to bind and invade uroepithelial cells.5 Although innate immunity plays an essential role in the first line of host defense against pathogens, in UTIs most human UPEC strains are resistant to complement-mediated killing.6, 7 Bacteria-mediated acute inflammatory responses can cause renal tissue inflammation and epithelium destruction, allowing bacteria to enter the underlying tissue,8, 9, 10 and persistent bacterial colonization and chronic inflammation can lead to tubular atrophy and tubulointerstitial fibrosis.11

ent-mediated killing.6, 7 Bacteria-mediated acute inflammatory responses can cause renal tissue inflammation and epithelium destruction, allowing bacteria to enter the underlying tissue,8, 9, 10 and persistent bacterial colonization and chronic inflammation can lead to tubular atrophy and tubulointerstitial fibrosis.11 C5a receptor 1 (C5aR1) is a 350–amino acid glycoprotein and member of the G-protein–coupled receptor superfamily of proteins that is expressed in myeloid cells (e.g., neutrophils and monocytes/macrophages [MO/MΦs]) and nonmyeloid cells, including renal tubular epithelial cells.12 The well-known ligand for C5aR1 is C5a (also called an anaphylatoxin), which is a 74–amino acid glycopolypeptide fragment generated during complement activation by cleavage of complement C5. The interaction of C5aR1 with C5a mediates a broad spectrum of proinflammatory reactions, such as an increase in vascular permeability, recruitment of leukocytes to sites of injury or infection, generation of cytotoxic oxygen radicals (by granulocytes), and generation of proinflammatory mediators (by myeloid and nonmyeloid cells). A large body of research has demonstrated that C5a/C5aR1 signaling contributes to the pathogenesis of a wide range of inflammatory pathologies, including renal disorders.12, 13, 14 Furthermore, there is compelling evidence from sepsis studies indicating that C5a/C5aR1 signaling can provide counterregulatory effects in host defense through impairment of innate immune cell function and induce excessive inflammatory responses.15 Pathogenic roles for C5a/C5aR1 signaling have also been reported in a number of other animal models of infectious disease, such as malaria, acute pneumococcal otitis media, and gram-negative bacteremia.16, 17, 18 However, the roles for C5aR1 in chronic kidney disease, particularly under conditions of infection, are largely unknown.

les for C5a/C5aR1 signaling have also been reported in a number of other animal models of infectious disease, such as malaria, acute pneumococcal otitis media, and gram-negative bacteremia.16, 17, 18 However, the roles for C5aR1 in chronic kidney disease, particularly under conditions of infection, are largely unknown. Given that (i) C5aR1 is expressed in renal resident and inflammatory cells and is up-regulated under pathological conditions,12, 19, 20, 21, 22 (ii) C5a/C5aR1 signaling is a strong driver of tissue inflammation,13, 14 and (iii) C5a/C5aR1 signaling has a negative impact on phagocyte function,23, 24 together with the pathological features of chronic kidney infection (i.e., persistent bacterial colonization, tissue inflammation, and tubulointerstitial fibrosis),11, 25 we hypothesized that C5aR1 may play a pathogenic role in chronic kidney infection. To test this hypothesis, we used a well-established murine model of chronic pyelonephritis induced by the UPEC strain IH11128 and C5aR1-deficient (C5aR1-/-) mice, as well as a C5aR1 antagonist, to determine the role of C5aR1 in chronic kidney infection (i.e., bacterial load, tissue inflammation, and tubulointerstitial fibrosis). We also investigated the cellular basis of the C5a/C5aR1 axis, which contributes to the pathogenesis of chronic kidney infection, by examining the influence of C5aR1 on cellular infiltration of the kidney following renal infection. In addition, we measured the effects of C5a/C5aR1 on the production of proinflammatory and profibrogenic factors by primary cultured renal tubular epithelial cells (RTECs) and MO/MΦs in response to bacterial stimulation and assessed the impact of C5a/C5aR1 on the phagocytic function of MO/MΦs. Our data demonstrate that following infection, early and persistent bacterial colonization, renal inflammation, and tubulointerstitial fibrosis are dependent on C5aR1, suggesting that C5aR1 facilities the pathogenesis of chronic kidney infection by enhancement of bacterial colonization of tubular epithelium, promotion of local inflammatory responses, and impairment of phagocytic function of MO/MΦs.

onization, renal inflammation, and tubulointerstitial fibrosis are dependent on C5aR1, suggesting that C5aR1 facilities the pathogenesis of chronic kidney infection by enhancement of bacterial colonization of tubular epithelium, promotion of local inflammatory responses, and impairment of phagocytic function of MO/MΦs. Results C5aR1-/- mice have reduced bacterial load in the kidney and bladder following bladder inoculation with UPEC Previous studies in a murine model of chronic kidney infection induced by IH11128 have shown that bacterial colonization of the kidney can be detected at 1 to 2 days after infection and persist for months.11 We therefore assessed bacterial load in the kidney and bladder of wild-type (WT) and C5aR1-/- mice at three stages of infection, namely, early (day 2), intermediate (day 14), and late (day 56) after infection, by counting bacterial colony-forming units (CFUs) recovered from kidney or bladder tissue samples on agar plates.26 After bacterial inoculation into the bladder (postinfection), bacterial colonies in the kidney and bladder peaked at day 1 to day 2; afterwords, there was a trend. However, C5aR1-/- mice had significantly lower bacterial colony counts in the kidneys at all time points after infection (day 2, day 4, and day 56) and in the bladder at day 2 and day 14 after infection compared with WT mice (Figure 1a and b). We also performed fluorescence microscopy analysis of bacterial colonies in infected kidney tissues from WT and C5aR1-/- mice at day 2 after inoculation with fluorescence-labeled bacteria. Consistent with the results of the agar plate assay, bacterial colonies in the renal tubular epithelium were significantly lower in C5aR1-/- mice compared with in WT mice (Figure 1c and d). Collectively, these data demonstrate that C5aR1 deficiency reduces bacterial load in the kidney and bladder.

nce-labeled bacteria. Consistent with the results of the agar plate assay, bacterial colonies in the renal tubular epithelium were significantly lower in C5aR1-/- mice compared with in WT mice (Figure 1c and d). Collectively, these data demonstrate that C5aR1 deficiency reduces bacterial load in the kidney and bladder. C5aR1 deficiency attenuates renal pathology following infection The UPEC strain IH11128 used in this study lacks expression of P fimbriae; therefore, they do not cause severe tissue destruction. However, they do mediate the chronic inflammatory process within the renal parenchyma, leading to progressive tissue injury.11 We therefore assessed the renal histopathology of infected WT and C5aR1-/- mice at different stages of infection. Renal histopathological changes, including cellular infiltration, tubular damage, and interstitial inflammation, were observed at all time points after infection. Tubular atrophy and interstitial inflammation became more apparent at the later time points. The changes were predominantly located within the corticomedullary junction but were also observed in other areas (e.g., the outer cortex and inner medulla). On the basis of these changes, we performed histological scoring of periodic acid–Schiff– and hematoxylin and eosin–stained kidney sections from individual mice in the WT and C5aR1-/- groups. Compared with WT mice, C5aR1-/- mice exhibited attenuated renal histopathological lesions at all time points (days 2, 14, and 56) (Figure 2a and b). We also assessed renal function in WT and C5aR1-/- mice after infection by measuring blood urea nitrogen (BUN) levels. Blood urea nitrogen was not significantly elevated in all infected mice when compared with normal mice, but there was a trend toward increased blood urea nitrogen in WT mice at the late time point after infection (Figure 2c). These data demonstrate that renal histopathological lesions and late functional impairment were reduced in C5aR1-/- mice after infection.

levated in all infected mice when compared with normal mice, but there was a trend toward increased blood urea nitrogen in WT mice at the late time point after infection (Figure 2c). These data demonstrate that renal histopathological lesions and late functional impairment were reduced in C5aR1-/- mice after infection. C5aR1 deficiency influences the extent and phenotype of cellular infiltrates in the kidney in response to infection Cellular infiltration was further analyzed by flow cytometry and immunohistochemistry at early and intermediate time points after infection, as described previously.10 Flow cytometry analysis of renal cell suspensions showed that when compared with WT, C5aR1-/- kidneys had lower numbers of leukocytes (CD45+) at day 2 after infection and a lower proportion of MO/MΦs (Ly6G-CD11b+) within CD45+ cells at day 14 after infection (Figure 3a–c). In addition, compared with WT, C5aR1-/- kidneys had a lower Ly6Chi population and reduced ratios of Ly6Chi to Ly6Clo populations within the MO/MΦs (Ly6G-CD11b+) compartment at both day 2 and day 14 after infection (Figure 3a, d, and e). However, C5aR1-/- kidneys had a similar proportion of neutrophils (Ly6G+) within CD45+ cells compared with WT kidneys at day 2 and day 14 after infection (Figure 3a and f). Immunohistochemistry showed that when compared with WT, C5aR1-/- kidneys had lower numbers of CD45+ cells (day 2 after infection) and F4/80+ cells (day 14 after infection) than WT kidneys (Figure 3g–3i), which is in agreement with the results of flow cytometry analysis. Collectively, these results indicate that C5aR1 deficiency not only caused a general reduction in cellular infiltration and MO/MΦ accumulation but also specifically reduced inflammatory monocytes' infiltration of the kidney in response to renal infection. However, C5aR1 deficiency did not cause a reduction of neutrophil infiltrate at the time points studied in this model.

ency not only caused a general reduction in cellular infiltration and MO/MΦ accumulation but also specifically reduced inflammatory monocytes' infiltration of the kidney in response to renal infection. However, C5aR1 deficiency did not cause a reduction of neutrophil infiltrate at the time points studied in this model. C5aR1 deficiency attenuates renal tissue inflammation and fibrogenesis following renal infection Persistent tissue inflammation and fibrogenesis play an important role in the progression of chronic kidney disease. We therefore used semiquantitative reverse transcriptase polymerase chain reaction to assess renal tissue inflammation and fibrogenesis of WT and C5aR1-/- mice following renal infection. Gene expression of proinflammatory cytokines (tumor necrosis factor-α [TNF-α] and interleukin-1β [IL-1β]), chemokines (keratinocyte-derived protein chemokine [KC], monocyte chemoattactant protein-1 [MCP-1]), chemokine receptor (C-C motif chemokine receptor 2 [CCR2]), and profibrogenic factors (transforming growth factor-β [TFG-β] and platelet-derived growth factor [PDGF]) was significantly reduced in C5aR1-/- mice compared with in WT mice at all time points studied (Figure 4a–4c). In contrast, gene expression of antifibrogenic factor (hepatocyte growth factor) was significantly higher in C5aR1-/- mice than in WT mice at day 2 and day 14 after infection (Figure 4d). Collectively, these results indicate that absence of C5aR1 attenuates renal tissue inflammation and fibrogenesis following renal infection.

rast, gene expression of antifibrogenic factor (hepatocyte growth factor) was significantly higher in C5aR1-/- mice than in WT mice at day 2 and day 14 after infection (Figure 4d). Collectively, these results indicate that absence of C5aR1 attenuates renal tissue inflammation and fibrogenesis following renal infection. Less severe renal fibrosis develops in C5aR1-/- mice after infection To further investigate the impact of C5aR1 on the development of renal fibrosis, we assessed the extent of collagen deposition and extracellular matrix production in kidneys from WT and C5aR1-/- mice following renal infection. Sirius red staining was performed on kidney sections at day 14 and day 56 after infection. Compared with WT kidneys, C5aR1-/- kidneys exhibited a significant reduction of Sirius red staining (Figure 5a and b). We also examined gene expression of several major extracellular matrix and cytoskeletal proteins in kidneys from WT and C5aR1-/- mice at day 14 and day 56 after infection using semiquantitative reverse transcriptase polymerase chain reaction. Intrarenal expression of mRNA encoding for extracellular matrix proteins (collagen I and fibronectin) and cytoskeleton proteins (α-smooth muscle and vimentin) was significantly reduced in C5aR1-/- mice compared with in WT mice. In contrast, C5aR1-/- mice had significantly higher levels of mRNA for collagen IV (which is required for renal tubular epithelial cell structural and functional integrity27) than did WT mice (Figure 5c). Taken together, these findings indicate that C5aR1 deficiency protects mice from renal fibrosis and parenchymal loss following renal infection.

ad significantly higher levels of mRNA for collagen IV (which is required for renal tubular epithelial cell structural and functional integrity27) than did WT mice (Figure 5c). Taken together, these findings indicate that C5aR1 deficiency protects mice from renal fibrosis and parenchymal loss following renal infection. Antagonizing C5aR1 reduces renal inflammation and fibrosis following renal infection In addition to studying C5aR1-/- mice, we used a well-known C5aR1 antagonist (PMX53)28 to assess whether this approach can curtail renal infection, tissue inflammation, and fibrosis. WT mice were administered PMX53 or control agent daily for 13 days. Consistent with observations made in C5aR1-/- mice, renal histopathology, tissue inflammation/fibrogenesis, collagen deposition, and bacterial load were significantly reduced in the kidneys of PMX53-treated mice compared with in the control group at day 14 after infection (Figure 6a–e). These results further confirmed the pathogenic role for C5aR1 in this model and also suggest a therapeutic potential for targeting C5aR1 in chronic kidney infection. To further investigate whether the major effects of C5aR1 on chronic inflammation and renal fibrosis are dependent on the initial impact of C5aR1 or the subsequent responses, we performed an additional set of in vivo experiments. WT mice were given PMX53 or control agent at the following different stages of infection: (i) starting at day 0 (2 hours before the inoculation) and continuing up to day 2 (early administration) and (ii) starting at day 3 after the inoculation and continuing up to day 13 (late administration). Early bacterial colonization was assessed at day 3 after infection following early administration. Mice that received PMX53 exhibited significantly fewer CFUs in the kidney compared with mice that received control agent (Figure 6f). Chronic injury was assessed at day 14 after infection. Both early and late administration of PMX53 reduced renal fibrosis and tissue bacterial load, but early administration led to a more profound reduction compared with in the control group (Figure 6g–i). These results suggest that the initial impact of C5aR1 plays an important role in the limit of bacterial load and subsequent chronic inflammation and renal fibrosis.

uced renal fibrosis and tissue bacterial load, but early administration led to a more profound reduction compared with in the control group (Figure 6g–i). These results suggest that the initial impact of C5aR1 plays an important role in the limit of bacterial load and subsequent chronic inflammation and renal fibrosis. C5a/C5aR1 interaction amplifies bacteria-induced production of proinflammatory and profibrogenic factors by RTECs and MO/MΦs Our in vivo data presented in this article suggest a critical role for C5aR1 in renal tissue inflammation and fibrogenesis following renal infection. To explore the cellular basis for the C5a/C5aR1 axis contributing to renal tissue inflammation and fibrogenesis, we used primary cell culture systems for RTECs and MO/MΦs and then used reverse transcriptase polymerase chain reaction to assess the effects of C5a/C5aR1 interactions on production of proinflammatory and profibrogenic factors by these cells in response to bacterial stimulation. Incubation with C5a alone led to only a small increase in proinflammatory and profibrogenic factor mRNA expression by RTECs and MO/MΦs. Incubation with heat-killed UPEC alone clearly increased mRNA expression of proinflammatory factors (i.e., tumor necrosis factor-α, interleukin-1β, monocyte chemoattactant protein-1, and keratinocyte-derived protein chemokine) by RTECs and MO/MΦs. In the presence of C5a, mRNA expression levels for those proinflammatory factors were further increased in RTECs and MO/MΦs compared with in the absence of C5a. Combined heat-killed UPEC and C5a treatment also increased mRNA expression of profibrotic factors (i.e., transforming growth factor-β and platelet-derived growth factor) by RTECs and MO/MΦs compared with heat-killed UPEC or C5a treatment alone (Table 1). Thus, our data indicate that engagement of C5aR1 amplifies UPEC-induced production of proinflammatory and profibrogenic factors by RTECs and MO/MΦs.

brotic factors (i.e., transforming growth factor-β and platelet-derived growth factor) by RTECs and MO/MΦs compared with heat-killed UPEC or C5a treatment alone (Table 1). Thus, our data indicate that engagement of C5aR1 amplifies UPEC-induced production of proinflammatory and profibrogenic factors by RTECs and MO/MΦs. C5a/C5aR1 interaction has a negative effect on the phagocytic function of phagocytes In addition to the effects on cytokine production by MO/MΦs, we assessed the impact of C5a/C5aR1 interactions on the ability of phagocytes to kill bacteria by using primary MO/MΦs prepared from WT mice. MO/MΦs were pretreated with C5a (10 nM or 50 nM) for 4 hours and then used for bacterial uptake and intracellular killing assays. The uptake of UPEC was comparable in the control and C5a-treated MO/MΦs, whereas survival of intracellular UPEC was significantly increased in C5a-treated MO/MΦs compared with in the control group, indicating that C5a treatment caused impairment of the bactericidal activity of MO/MΦs but had no apparent effect on bacterial uptake (Figure 7).

comparable in the control and C5a-treated MO/MΦs, whereas survival of intracellular UPEC was significantly increased in C5a-treated MO/MΦs compared with in the control group, indicating that C5a treatment caused impairment of the bactericidal activity of MO/MΦs but had no apparent effect on bacterial uptake (Figure 7). Discussion Our current understanding of the roles of C5a/C5aR1 in the pathogenesis of kidney disease is primarily based on studies in acute or non–pathogen-related injury. In this study we used a well-characterized murine model of ascending UTI to investigate the role of C5aR1 in chronic kidney injury. Our data demonstrate that C5aR1 deficiency or blockade not only reduces renal bacterial load at all stages of infection but also attenuates tissue inflammation and tubulointerstitial fibrosis, suggesting a pathogenic role for C5aR1 in experimental chronic kidney infection. Mechanistic studies suggest that C5aR1-mediated bacterial colonization of tubular epithelium, persistent local inflammatory responses, and impairment of the phagocytic function of MO/MΦs could contribute to the pathogenesis of chronic kidney infection. Thus, our data further support the notion that excessive or persistent tissue inflammation represents an important pathogenic mechanism in acute and chronic kidney infection.8, 10, 11

mmatory responses, and impairment of the phagocytic function of MO/MΦs could contribute to the pathogenesis of chronic kidney infection. Thus, our data further support the notion that excessive or persistent tissue inflammation represents an important pathogenic mechanism in acute and chronic kidney infection.8, 10, 11 One of the important observations in this study is that C5aR1 deficiency or blockade C5aR1 reduced bacterial load at all stages of infection studied, even at the early stage of infection (day 2 or 3 after inoculation). This suggests that C5aR1 could have an impact on early UPEC colonization of renal tract epithelium. In support of this hypothesis, our recent work (in a separate study) has revealed that bacterial adhesion and tissue colonization mediated by the expression of mannosyl residues—a ligand for type 1 fimbriae that we detected on the luminal surface of the renal tubular and bladder epithelium—is increased by C5aR1-induced signaling (unpublished data). In addition to the impact on bacterial colonization, in the present study, we have also found that treatment of MO/MΦs with C5a significantly reduced their bactericidal activity, suggesting that C5a/C5aR1 interaction has a negative regulatory effect on the phagocytic function of MO/MΦs, which is consistent with the well-recognized phenomenon of C5a/C5aR1 signaling being a strong driver of inflammation and having a negative impact on the phagocytic function of phagocytes.24, 29 Therefore, C5a/C5aR1 interaction–mediated enhancement of UPEC adhesion/colonization of renal tract epithelium (possibly through upregulation of expression of mannosyl residue on the luminal surface of renal tubular and bladder epithelium) and impairment of phagocytic function of phagocytes could contribute to the early and persistent bacterial colonization of the kidney that we observed in this model.

of renal tract epithelium (possibly through upregulation of expression of mannosyl residue on the luminal surface of renal tubular and bladder epithelium) and impairment of phagocytic function of phagocytes could contribute to the early and persistent bacterial colonization of the kidney that we observed in this model. Another important observation from this study is that following renal infection, C5aR1 deficiency or blockade resulted in a reduction in renal tissue inflammation (i.e., cellular infiltration, tubular atrophy, and intrarenal gene expression of proinflammatory factors). These findings strongly suggest that C5aR1 has a critical role in upregulating local inflammatory responses to UPEC, which contributes to chronic inflammation of the kidney and development of renal scarring. To explore the cellular basis of the C5a/C5aR1 axis contributing to tissue inflammation, we examined the impact of C5aR1 on inflammatory cell infiltration and found that C5aR1-/- deficiency not only reduced the accumulation of leukocytes and MO/MΦs but specifically reduced the population of Ly6Chi proinflammatory MO/MΦs following renal infection. These findings suggest that C5aR1 not only promotes chemotaxis of leukocytes but also plays an important role in modulating the phenotype of infiltrating cells. C5aR1 deficiency resulting in overall inhibition of cellular infiltration and accumulation of proinflammatory MO/MΦs could contribute to the control and resolution of local inflammation.

C5aR1 not only promotes chemotaxis of leukocytes but also plays an important role in modulating the phenotype of infiltrating cells. C5aR1 deficiency resulting in overall inhibition of cellular infiltration and accumulation of proinflammatory MO/MΦs could contribute to the control and resolution of local inflammation. In addition to cellular infiltration, we also examined the impact of C5a/C5aR1 interaction on cellular responses to UPEC. We focused on RTECs and MO/MΦs, as both types of cells express C5aR1 and also are important sources of proinflammatory and profibrogenic factors in the kidney. Our findings that C5a stimulation up-regulated UPEC-induced production of proinflammatory and profibrogenic factors by cultured RTECs and MO/MΦs suggest that (i) engagement of C5aR1 on both inflammatory and renal parenchymal cells could contribute to local tissue inflammation and (ii) the additive effect of C5a and UPEC on cytokine production by RTECs and MO/MΦs could be explained by the interaction of C5aR1 and Toll-like receptor signaling, as suggested by a previous study in a lipopolysaccharide-induced cytokine production model.30

hymal cells could contribute to local tissue inflammation and (ii) the additive effect of C5a and UPEC on cytokine production by RTECs and MO/MΦs could be explained by the interaction of C5aR1 and Toll-like receptor signaling, as suggested by a previous study in a lipopolysaccharide-induced cytokine production model.30 Additionally, our study found that C5aR1 has a big impact on the development of renal fibrosis. Less severe renal tubular interstitial fibrosis developed in C5aR1-/- mice than in WT mice following renal infection. The pathogenesis of renal fibrosis is complex, as multiple cell types and molecular pathways are involved. However, it is becoming increasingly clear that the inflammatory microenvironment of the kidney after sustained injury is a key determinant of the dynamic balance between tissue destruction (tubular atrophy and interstitial fibrosis) and repair (tubular cell growth and resolution of renal inflammation and fibrosis).31, 32 Previous studies have shown that increased intrarenal expression of proinflammatory and profibrogenic factors are strongly associated with renal fibrosis.32, 33, 34 Our finding of an association of reduced renal fibrosis with significant reduction of renal expression of an array of proinflammatory and profibrogenic factors in C5aR1-/- mice suggests that C5a/C5aR1-mediated proinflammatory and profibrotic responses could contribute to the pathogenesis of renal fibrosis. In addition, previous studies have shown that the phenotype of the inflammatory cell infiltrate has an impact on chronic renal inflammation and fibrosis, suggesting that renal Ly6Chi proinflammatory MO/MΦs have profibrotic effects. Our findings that C5aR1-/- deficiency specifically reduced the population of Ly6Chi proinflammatory MO/MΦs following renal infection suggest that C5a/C5aR1 could contribute to the development of renal fibrosis by modulating the phenotype of infiltrating leukocytes.

y6Chi proinflammatory MO/MΦs have profibrotic effects. Our findings that C5aR1-/- deficiency specifically reduced the population of Ly6Chi proinflammatory MO/MΦs following renal infection suggest that C5a/C5aR1 could contribute to the development of renal fibrosis by modulating the phenotype of infiltrating leukocytes. Intrarenal gene expression profiles from this study revealed that absence of C5aR1 not only reduced the expression of proinflammatory and profibrogenic factors and mesenchymal markers but also led to increased expression of hepatocyte growth factor and collagen IV following renal infection. Hepatocyte growth factor was suggested as an important antifibrotic factor in renal fibrosis, promoting renal cell survival, proliferation, migration, and tubulogenesis and directly antagonizing the profibrotic actions of transforming growth factor-β.35 Although the role for collagen IV in renal fibrosis is not well defined, upregulation of intrarenal expression of collagen IV mRNA was found to be associated with reduced renal fibrosis in a unilateral ureteral obstruction model,36 thus supporting an antifibrotic role for collagen IV in renal fibrosis. Our finding that C5aR1 has opposing effects on renal expression of profibrogenic and antifibrogenic factors supports the concept that C5aR1 influences the dynamic balance between tissue destruction and repair after renal infection, favoring the promotion of tissue destruction and impairing the repair process.

fibrosis. Our finding that C5aR1 has opposing effects on renal expression of profibrogenic and antifibrogenic factors supports the concept that C5aR1 influences the dynamic balance between tissue destruction and repair after renal infection, favoring the promotion of tissue destruction and impairing the repair process. To conclude, this study is the first to define a novel and important role of C5aR1 in the pathogenesis of experimental chronic pyelonephritis. It also suggests that C5aR1-dependent enhancement of UPEC colonization of renal tract epithelium and excessive local inflammatory responses, as well as impairment of phagocytic function of phagocytes, contribute to the chronic inflammation and renal fibrosis. Our observation that blocking the C5a/C5aR1 axis curtailed the pathology suggests a therapeutic potential for targeting C5aR1 in chronic kidney infection. Although antibiotics can be used to treat the UTI, there is the emerging threat of multidrug-resistant gram-negative bacteria in urology in addition to the well-known frequent recurrence and persistence of infection. Targeting the C5a/C5aR1 axis may offer alternative or combination therapies to reduce the use of antibiotics. Methods Mice Homozygous C5aR1-/- mice were generated by homologous recombination in embryonic stem cells37 and backcrossed onto the C576BL/6 (H-2b) parental strain for at least 12 generations. C5aR1-/- mice and their WT littermates were used in all experiments, and animal procedures were carried out in accordance with the Animals Scientific Procedures Act 1986.

generated by homologous recombination in embryonic stem cells37 and backcrossed onto the C576BL/6 (H-2b) parental strain for at least 12 generations. C5aR1-/- mice and their WT littermates were used in all experiments, and animal procedures were carried out in accordance with the Animals Scientific Procedures Act 1986. Induction of chronic pyelonephritis A previously described model of ascending UTI leading to chronic pyelonephritis was used. The infection was induced in WT and C5aR1-/- mice (females 8–10 weeks old) by bladder inoculation of the human UPEC strain IH11128 (108 CFUs in 50 μl phosphate-buffered sailine) per urethram. The IH11128 strain (075:K5:H–) was isolated from patients with chronic pyelonephritis. It is a mannose-resistant strain expressing both Dr and type 1 fimbriae but lacking P fimbriae and hemolytic activity (a kind gift from Dr. B. Nowicki, University of Texas, Galveston, Texas, USA).38 Mice were killed at day 2, day 14, and day 56 after infection for evaluating bacterial load and renal histopathology. In some experiments, tetrarhodamine isothiocyanate–labeled IH11128 was used for imaging bacteria in kidney sections. For C5aR1 antagonist treatment experiments, WT mice were administered C5aR1 peptide antagonist (PMX53, Ac-Phe-cyclo[Orn-Pro-dCha-Trp-Arg]) or control agent (random sequence peptide) (synthesized by GenScript, Shanghai, China) by subcutaneous injection (1 mg/kg daily) starting at different stages of infection and for different times.

gonist treatment experiments, WT mice were administered C5aR1 peptide antagonist (PMX53, Ac-Phe-cyclo[Orn-Pro-dCha-Trp-Arg]) or control agent (random sequence peptide) (synthesized by GenScript, Shanghai, China) by subcutaneous injection (1 mg/kg daily) starting at different stages of infection and for different times. Assessment of renal histopathological features Paraffin sections (4 μm) were stained with hematoxylin and eosin, periodic acid–Schiff stain, or Sirius red. Stained kidney sections were scanned with a Hamamatsu Nanozoomer 2.0 HT slide scanner (Hamamatsu Photonics, Hamamtsu, Japan) and viewed using NDP.view2 software. Renal histopathological changes were assessed on periodic acid–Schiff– and hematoxylin and eosin–stained sections using a 6-point scale in which 0, 1, 2, 3, 4, and 5 indicated normal, very little, very mild, mild, moderate, and severe histological lesions, respectively. The assessment was based on histopathological changes (i.e., cellular infiltration, tubular atrophy, and interstitial inflammation) that were mainly located at the corticomedullary junction area. Renal fibrosis was assessed on Sirius red–stained sections. The positively stained areas were quantified by imaging analysis (ImageJ software; National Institutes of Health, Bethesda, MD, USA). Briefly, 6 to 8 corticomedullary junction viewing fields selected from appropriate areas within each kidney were examined. Positively stained areas were expressed as a percentage of the whole field area (1.92 mm2). All the aforementioned quantitative analyses were performed in a blinded fashion by 2 experienced persons.

riefly, 6 to 8 corticomedullary junction viewing fields selected from appropriate areas within each kidney were examined. Positively stained areas were expressed as a percentage of the whole field area (1.92 mm2). All the aforementioned quantitative analyses were performed in a blinded fashion by 2 experienced persons. Measurement of bacterial load in the kidney and bladder Total bacterial load in kidney and bladder tissue was measured by the agar plate assay as previously described, with modifications.26 In brief, the tissue was weighed and subsequently homogenized in 2 ml (for the kidney) or 1 ml (for the bladder) of phosphate-buffered saline. A quantity of 100 μl of a series dilution of homogenates were plated on cysteine-, lactose-, and electrolyte-deficient plates. After incubation of the plates for 24 hours, bacterial CFUs on the agar plates were manually counted and expressed as CFU per gram of tissue. Statistical analysis Data are shown either as the mean ± SEM or the readout of individual mice. Unpaired Student’s t test was used for comparison between two groups. All the analyses were performed using Graphpad Prism Software version 5 (GraphPad Software, LaJolla, CA, USA). Disclosure All the authors declared no competing interests.

Statistical analysis Data are shown either as the mean ± SEM or the readout of individual mice. Unpaired Student’s t test was used for comparison between two groups. All the analyses were performed using Graphpad Prism Software version 5 (GraphPad Software, LaJolla, CA, USA). Disclosure All the authors declared no competing interests. Supplementary Material Supplementary Methods The following additional information for methods is given in supplementary methods: materials, cell cultures, assessment of effects of C5a/C5aR1 interaction on cytokine production by RTECs and MO/MΦs in response to bacteria stimulation, assessment of bacterial uptake and intracellular killing by MO/MΦs, assessment of renal inflammatory cell infiltration by flow cytometry, immunohistochemistry, semiquantitative real-time reverse transcriptase polymerase chain reaction, and detection of fluorescence-labeled IH11128 in kidney tissues. Acknowledgments This study was supported by the Medical Research Council of the UK (G1001141 to WZ and SS) and the National Natural Science Foundation of China (NSFC 81170644 to KL). We thank Dr. Bao Lu and Professor Craig Gerard for providing the C5aR1 knockout mice. We thank Dr. Daxin Chen for helpful advice on image analysis. see commentary on page 469 Supplementary Methods.

Acknowledgments This study was supported by the Medical Research Council of the UK (G1001141 to WZ and SS) and the National Natural Science Foundation of China (NSFC 81170644 to KL). We thank Dr. Bao Lu and Professor Craig Gerard for providing the C5aR1 knockout mice. We thank Dr. Daxin Chen for helpful advice on image analysis. see commentary on page 469 Supplementary Methods. The following additional information for methods is given in supplementary methods: materials, cell cultures, assessment of effects of C5a/C5aR1 interaction on cytokine production by RTECs and MO/MΦs in response to bacteria stimulation, assessment of bacterial uptake and intracellular killing by MO/MΦs, assessment of renal inflammatory cell infiltration by flow cytometry, immunohistochemistry, semiquantitative real-time reverse transcriptase polymerase chain reaction, and detection of fluorescence-labeled IH11128 in kidney tissues. Supplementary material is linked to the online version of the paper at www.kidney-international.org.

ine output, whereas urine osmolality is increased by ∼40% may be explained by assuming that β3-AR stimulation promotes not only water but also salt reabsorption in the kidney. In line with this possibility, the strong reduction of urine output observed in BRL37344-treated mice is independent of the decrease in the GFR. β3-ARs are also expressed in the hypothalamus38; thus, the possibility exists that the antidiuretic effect of BRL37344 may involve hypothalamic regulation of AVP release. Our results in AVPR2-null mice31 seem to exclude this possibility. In these mice, the classic symptoms of XNDI develop.39, 40, 41 As shown here, a single i.p. injection of BRL37344 greatly reduces the diuresis and increases urine osmolality, supporting the notion that, in vivo, β3-AR agonism triggers AVP-independent antidiuresis. In addition, results in live kidney slices, demonstrating that BRL37344 induces cAMP production, AQP2 plasma membrane accumulation, and NKCC2 phosphorylation/activation in the thick ascending limb of Henle, provide additional, although indirect, evidence that BRL37344 triggers its effect independently of central β3-AR activation. The current results cannot exclude that the effects of BRL37344 on urine output may be related to the systemic effects of the drug on arterial pressure. However, it has been shown in rats that BRL37344 reduces arterial pressure by ∼14%42; therefore, it is unlikely that such an effect may be responsible for the observed 70% reduction in urine output.

The following additional information for methods is given in supplementary methods: materials, cell cultures, assessment of effects of C5a/C5aR1 interaction on cytokine production by RTECs and MO/MΦs in response to bacteria stimulation, assessment of bacterial uptake and intracellular killing by MO/MΦs, assessment of renal inflammatory cell infiltration by flow cytometry, immunohistochemistry, semiquantitative real-time reverse transcriptase polymerase chain reaction, and detection of fluorescence-labeled IH11128 in kidney tissues. Supplementary material is linked to the online version of the paper at www.kidney-international.org. Figure 1 C5aR1-/- mice have reduced bacterial load in the kidney and bladder after bladder inoculation with urethropathogenic Escherichia coli (UPEC). Bacterial loads in the kidney (a) and bladder (b) from wild-type (WT) and C5aR1-/- mice were examined at days 2, 14, and 56 after infection. Each dot represents colony-forming units (CFUs) recovered from an individual mouse and is shown as average CFUs from 2 replicate agar plates. Data were analyzed by Student’s t test (n = 6–11 mice per group). ***P < 0.001. (c) Representative fluorescence microscope images of kidney sections from WT and C5aR1-/- mice at day after infection, taken at the corticomedullary junction and showing bacterial colonization of renal tubular epithelium (arrows) (blue = 4,6-diamino-2-phenylindole, green = lotus tetragonolobus lectin, red = bacteria) (bar = 25 μm). (d) Quantification of bacterial colonies in the kidneys of infected WT and C5aR1-/- mice. Data were analyzed by Student’s t test (60 viewing fields [×200 magnification] from 4 mice per group). ***P < 0.001. A representative of 2 independent experiments is shown.

tus tetragonolobus lectin, red = bacteria) (bar = 25 μm). (d) Quantification of bacterial colonies in the kidneys of infected WT and C5aR1-/- mice. Data were analyzed by Student’s t test (60 viewing fields [×200 magnification] from 4 mice per group). ***P < 0.001. A representative of 2 independent experiments is shown. Figure 2 C5aR1 attenuates renal pathology following renal infection. (a) Representative images of periodic acid-Schiff–stained kidney sections from noninfected and infected wild-type (WT) and C5aR1-/- mice at days 2, 14, and 56 after infection, taken at the corticomedullary junction. Arrows show renal tubular lesions. Bar = 100 μm. (b) Histological scores in the mice illustrated in panel (a). Each dot represents an individual mouse. Data were analyzed by Student’s t test (n = 5 or 6 mice per group). ***P < 0.001. (c) Blood urea nitrogen (BUN) levels in infected WT and C5aR1-/- mice at days 2, 14, and 56 after infection. The dotted line on the graph represents BUN level in normal mice. Data were analyzed by Student’s t test (n = 6–11 mice per group). *P < 0.05. A representative of 2 independent experiments is shown.

**P < 0.001. (c) Blood urea nitrogen (BUN) levels in infected WT and C5aR1-/- mice at days 2, 14, and 56 after infection. The dotted line on the graph represents BUN level in normal mice. Data were analyzed by Student’s t test (n = 6–11 mice per group). *P < 0.05. A representative of 2 independent experiments is shown. Figure 3 C5a receptor (C5aR) deficiency influences the extent and phenotype of cellular infiltrates in the kidneys following renal infection. Renal inflammatory cell infiltration was analyzed in infected wild-type (WT) and C5aR1-/- mice at days 2 and 14 after infection by flow cytometry. (a) Stepwise gating strategy used in flow cytometric analysis of leukocytes, neutrophils, monocytes/macrophages (MO/MΦs), and Ly6chi MO/MΦs in kidney tissues. (b–f) Quantification of leukocytes (CD45+), MO/MΦ (Ly6G-CD11b+), Ly6chi population, and ratio of Ly6chi to Ly6clo populations within the Ly6G-CD11b+ compartment and neutrophils (Ly6G+), respectively. Each dot represents an individual mouse. Data were analyzed by Student’s t test (n = 5 mice per group). **P < 0.05. ***P < 0.005. (g–i) Immunohistochemistry. (g) Representative images of CD45- and F4/80-stained kidney sections from infected WT and C5aR1-/- mice (n = 4 mice per group). Arrows show positively stained cells. Bar = 100 μm. (h,i) Quantification of CD45+ and F4/80+ cells. Data were analyzed by Student’s t test (40–50 viewing fields [0.04 mm2 per field] from 4 mice per group). ***P < 0.001. A representative of 2 independent experiments is shown.

WT and C5aR1-/- mice (n = 4 mice per group). Arrows show positively stained cells. Bar = 100 μm. (h,i) Quantification of CD45+ and F4/80+ cells. Data were analyzed by Student’s t test (40–50 viewing fields [0.04 mm2 per field] from 4 mice per group). ***P < 0.001. A representative of 2 independent experiments is shown. Figure 4 Absence of C5a receptor (C5aR1) reduces intrarenal gene expression of proinflammatory and profibrogenic factors in response to renal infection. Relative mRNA levels of proinflammatory and profibrogenic factors in infected kidney tissues from wild-type WT and C5aR1-/- mice at the indicated stages of infection by quantitative reverse transcriptase polymerase chain reaction. (a) Proinflammatory cytokines (tumor necrosis factor-α [TNF-α] and interleukin-6 [IL-6]). (b) Proinflammatory chemokines (keratinocyte-derived protein chemokine [KC] and monocyte chemoattactant protein-1 [MCP-1]) and MCP-1 receptor C-C motif chemokine receptor 2 (CCR2). (c) Profibrogenic factors (transforming growth factor-β [TGF-β] and platelet-derived growth factor [PDGF]). (d) Hepatocyte growth factor (HGF). Each dot represents an individual mouse and is shown as the mean of 2 replicate polymerase chain reaction results. Data were representative of 3 separate cDNA preparations tested in duplicate for each mouse. The dotted line on the graph represents the gene expression level in normal kidney tissue. Data were analyzed by Student’s t test (n = 6 mice per group). *P < 0.05, **P < 0.005, ***P < 0.001. A representative of 2 independent experiments is shown.

tive of 3 separate cDNA preparations tested in duplicate for each mouse. The dotted line on the graph represents the gene expression level in normal kidney tissue. Data were analyzed by Student’s t test (n = 6 mice per group). *P < 0.05, **P < 0.005, ***P < 0.001. A representative of 2 independent experiments is shown. Figure 5 Less severe renal fibrosis develops in C5a receptor 1–deficient (C5aR1-/-) mice after infection. (a) Representative images of Sirius red–stained kidney sections from wild-type (WT) and C5aR1-/- mice at days 14 and 56 after infection, taken at the corticomedullary junction. Arrows show positively stained area. Bar = 200 μm. (b) Quantification of collagen deposition (Sirius red staining) in kidney sections of the mice in panel (a). Data were analyzed by Student’s t test (20–30 viewing fields [1.92 mm2 per field] from 4 mice per group). ***P < 0.001. (c) Semiquantitative analysis of mRNA expression of extracellular matrix and cytoskeletal proteins in infected WT and C5aR1-/- kidney tissue (n = 6 mice per group). Each dot represents an individual mouse and is shown as the mean of 2 replicate polymerase chain reaction results. The dotted line on the graph represents the gene expression level in normal kidney tissue. Data were analyzed by Student’s t test. **P < 0.005, ***P < 0.001. A representative of 2 independent experiments is shown.

represents an individual mouse and is shown as the mean of 2 replicate polymerase chain reaction results. The dotted line on the graph represents the gene expression level in normal kidney tissue. Data were analyzed by Student’s t test. **P < 0.005, ***P < 0.001. A representative of 2 independent experiments is shown. Figure 6 Antagonizing C5a receptor (C5aR1) reduces renal inflammation and fibrosis following renal infection. (a–e) Wild-type mice were administered PMX53 or control (ctrl) agent daily for 13 days starting on the day of induction of renal infection. Assessment of chronic inflammation, renal fibrosis, and renal bacterial load was performed at day 14 after infection. (a) Representative images of periodic acid–Schiff (PAS)- or Sirius red (SR)-stained kidney sections taken at the corticomedullary junction. Arrows show tubular lesions or areas positively stained with Sirius red. Bar = 200 μm. (b) Histological scores in the mice illustrated in panel (a). Each dot represents an individual mouse. Data were analyzed by Student’s t test (n = 5 mice per group). (c) Quantification of collagen deposition (Sirius red staining) in kidney sections of the mice in panel (a). Data were analyzed by Student’s t test (20 to 30 viewing fields [1.92 mm2 per field] from 5 mice per group). (d) Bacterial loads in the kidneys. Each dot represents colony-forming units recovered from an individual mouse and is shown as average colony-forming units from 2 replicate agar plates. (e) Semiquantitative analysis of mRNA expression of extracellular matrix and cytoskeletal proteins in infected kidney tissue. Each dot represents an individual mouse and is shown as the mean of two replicate polymerase chain reaction results. (d,e) Data were analyzed by Student’s t test (n = 5 mice per group). A representative of 2 independent experiments is shown. (f–i) Wild-type mice were given PMX53 or control agent at different stages of infection starting at day 0 (2 hours before the inoculation) and continuing up to day 2 (early administration) or starting at day 3 after the inoculation and continuing up to day 13 (late administration). (f) Colony-forming units recovered from the kidney of individual mice at day 3 after infection. Each dot represents an individual mouse. Data were analyzed by Student’s t test (n = 5 mice per group). (g) Representative images of Sirius red–stained kidney sections., taken at the corticomedullary junction. Arrows show area positively stained with Sirius red. Bar = 100 μm.

f individual mice at day 3 after infection. Each dot represents an individual mouse. Data were analyzed by Student’s t test (n = 5 mice per group). (g) Representative images of Sirius red–stained kidney sections., taken at the corticomedullary junction. Arrows show area positively stained with Sirius red. Bar = 100 μm. (h) Quantification of Sirius red staining in kidney sections of the mice in panel (g). Data were analyzed by Student’s t test (20–35 viewing fields [1.92 mm2 per field] from 5–7 mice per group). (i) Colony-forming units recovered from the kidney of individual mice at day 14 after infection. Each dot represents an individual mouse. Data were analyzed by Student’s t test (n = 7 mice per group). *P < 0.05, **P < 0.005, ***P < 0.001. Figure 7 C5a/C5a receptor 1 (C5aR1) interaction has a negative effect on phagocytic function of phagocytes. Freshly prepared peritoneal monocytes/macrophages from wild-type mice were pretreated with control agent or C5a (10 nM or 50 nM) for 4 hours and used for uropathogenic Escherichia coli phagocytosis and intracellular killing assays. (a) Uptake of uropathogenic E. coli by monocytes/macrophages. (b) Survival of intracellular (phagocytosed) bacteria in monocytes/macrophages. (a,b) Data were analyzed by Student’s t test (n = 5 per group). A representative of 3 independent experiments is shown. **P < 0.005. Table 1 C5a/C5aR1 interaction amplifies bacteria-induced production of proinflammatory and profibrogenic factors by RTEC and MO/MΦ

Figure 7 C5a/C5a receptor 1 (C5aR1) interaction has a negative effect on phagocytic function of phagocytes. Freshly prepared peritoneal monocytes/macrophages from wild-type mice were pretreated with control agent or C5a (10 nM or 50 nM) for 4 hours and used for uropathogenic Escherichia coli phagocytosis and intracellular killing assays. (a) Uptake of uropathogenic E. coli by monocytes/macrophages. (b) Survival of intracellular (phagocytosed) bacteria in monocytes/macrophages. (a,b) Data were analyzed by Student’s t test (n = 5 per group). A representative of 3 independent experiments is shown. **P < 0.005. Table 1 C5a/C5aR1 interaction amplifies bacteria-induced production of proinflammatory and profibrogenic factors by RTEC and MO/MΦ mRNA RTECsa MO/MΦsa C5a UPEC C5a/UPEC C5a UPEC C5a/UPEC TNF-α 1.1 ± 0.1 5.1 ± 1.2c 17.4 ± 3.3c,e 1.1 ± 0.3 4.0 ± 1.0b 6.7 ± 1.0b,e IL-6 2.0 ± 0.1b 48.3 ± 9.9c 77.3 ± 8.7d,e 2.9 ± 1.9 14.8 ± 3.1c 34.2 ± 5.4c,e MCP 1.9 ± 0.3b 48.2 ± 8.3d 78.1 ± 10.1c,e 2.1 ± 0.4b 2.7 ± 0.5b 4.6 ± 0.6d,e KC 1.2 ± 0.1 14.1 ± 2.8c 121.1 ± 16.1d,e 1.8 ± 0.2b 1.9 ± 0.1b 6.9 ± 1.5b,e TGF-β 1.2 ± 0.2 0.8 ± 0.3 2.8 ± 0.7b,e 2.2 ± 0.4b 1.8 ± 0.2b 5.7 ± 0.9c,e PDGF 1.4 ± 0.2 1.0 ± 0.2 3.5 ± 0.6c,e 1.4 ± 0.2 1.4 ± 0.2 4.2 ± 0.7c,e Ca5R1, C5a receptor 1; IL-6, interleukin-6; KC, keratinocyte-derived protein chemokine; MCP, monocyte chemoattactant protein-1; MO/MΦ, monocyte/macrophage; PDGF, platelet-derived growth factor; RTEC, renal tubular epithelial cell; TGF-β, transforming growth factor-β; TNF-α, tumor necrosis factor-α; UPEC, uropathogenic Escherichia coli.

or 1; IL-6, interleukin-6; KC, keratinocyte-derived protein chemokine; MCP, monocyte chemoattactant protein-1; MO/MΦ, monocyte/macrophage; PDGF, platelet-derived growth factor; RTEC, renal tubular epithelial cell; TGF-β, transforming growth factor-β; TNF-α, tumor necrosis factor-α; UPEC, uropathogenic Escherichia coli. a Primary cultured RTECs and freshly prepared peritoneal MO/MΦs from wild-type mice were incubated with heat-killed UPEC and C5a, alone and combined, for 24 hours and subjected to reverse transcriptase polymerase chain reaction analysis. mRNA expression was expressed as fold change over the control (absence of C5a and UPEC). Superscript letters b–d indicate significant difference versus control (bP < 0.05 cP < 0.005 dP < 0.001). e Significant difference versus UPEC alone (P < 0.05). Data were analyzed by Student’s t test (n = 5 per group). A representative of 3 independent experiments is shown.

In the kidney, the antidiuretic hormone arginine vasopressin (AVP) is a critical regulator of water and electrolyte homeostasis. AVP is released from the pituitary gland into the bloodstream and binds to the type 2 vasopressin receptor (AVPR2),1 a G protein–coupled receptor localized in the thick ascending limb of Henle, the distal convolute tubule, and the collecting duct, acting mainly through the cAMP–protein kinase A pathway. In the thick ascending limb of Henle, AVP stimulates NaCl reabsorption across the Na-K-Cl cotransporter (NKCC2), increasing its phosphorylation,2 thus generating the corticomedullary osmotic gradient providing the driving force for water reabsorption in the kidney tubules. In the CD, AVP stimulates the exocytosis of the water channel aquaporin 2 (AQP2)3 at the apical membrane of the principal cells, dramatically increasing water reabsorption (for a review, see ref. 4). Inactivating mutations of the AVPR2 gene cause X-linked nephrogenic diabetes insipidus (XNDI), characterized by constant diuresis and the risk of severe dehydration.5 Many studies have shown that hormones other than AVP also exhibit antidiuretic effect,6, 7, 8, 9, 10 suggesting novel strategies to manage XNDI. The β-adrenergic system controls several renal functions. In particular, types 1 and 2 β-adrenoreceptors (β1-2-AR)11 regulate renal blood flow, glomerular filtration rate (GFR), sodium and water reabsorption, acid-base balance, and secretion of renin (for a review, see Johns et al.12).

In the CD, AVP stimulates the exocytosis of the water channel aquaporin 2 (AQP2)3 at the apical membrane of the principal cells, dramatically increasing water reabsorption (for a review, see ref. 4). Inactivating mutations of the AVPR2 gene cause X-linked nephrogenic diabetes insipidus (XNDI), characterized by constant diuresis and the risk of severe dehydration.5 Many studies have shown that hormones other than AVP also exhibit antidiuretic effect,6, 7, 8, 9, 10 suggesting novel strategies to manage XNDI. The β-adrenergic system controls several renal functions. In particular, types 1 and 2 β-adrenoreceptors (β1-2-AR)11 regulate renal blood flow, glomerular filtration rate (GFR), sodium and water reabsorption, acid-base balance, and secretion of renin (for a review, see Johns et al.12). Among β-ARs, the β3-AR is the last identified member of this family. At first, it was shown to regulate lipolysis and thermogenesis in adipose tissue,13 whereas subsequently it was shown to play important roles in the pathophysiology of the cardiovascular14 and urinary15 systems. However, its expression and possible physiologic role in the kidney remains to be fully clarified. There are indications in mice that β3-AR mRNA is expressed by renal arteries.16 In addition, in the rat kidney, a cDNA microarray screening showed that β3-AR is expressed in the kidney medulla.17 Moreover, in humans, β3-AR polymorphisms seem to be associated with the effect of thiazide diuretics,16, 18 suggesting a role for β3-AR in regulating renal water reabsorption. In this respect, demonstrating this novel role of β3-AR in renal physiology is particularly intriguing in light of potential therapeutic applications of β3-AR–acting drugs in diseases characterized by altered diuresis. Moreover, β3-AR is relatively resistant to agonist-induced desensitization,19 which would ensure prolonged pharmacologic stimulation in vivo. In addition, due to the limited number of tissues expressing β3-AR, compared with β1-2-AR, β3-AR agonists are supposed to show a low systemic off-target effect.14

tered diuresis. Moreover, β3-AR is relatively resistant to agonist-induced desensitization,19 which would ensure prolonged pharmacologic stimulation in vivo. In addition, due to the limited number of tissues expressing β3-AR, compared with β1-2-AR, β3-AR agonists are supposed to show a low systemic off-target effect.14 Results β3-AR expression in the mouse kidney Reverse transcriptase polymerase chain reaction revealed that β3-AR mRNA was clearly detectable in the RNA samples from the mouse kidney, brown adipose tissue, and bladder (Figure 1a). In particular, the intron-spanning primers amplified 2 bands of 234 bp and 337 bp, representing β3a-AR and β3b-AR transcripts, respectively.20 Sequencing confirmed the specificity of the obtained bands (data not shown). Immunoblotting analysis revealed that mouse kidney cortex and total medulla expressed a band of 44 kDa for the core protein and 1 at 68 kDa for the glycosylated form in all samples (Figure 1b). Both bands were also revealed in β3-AR–expressing control tissues.

Results β3-AR expression in the mouse kidney Reverse transcriptase polymerase chain reaction revealed that β3-AR mRNA was clearly detectable in the RNA samples from the mouse kidney, brown adipose tissue, and bladder (Figure 1a). In particular, the intron-spanning primers amplified 2 bands of 234 bp and 337 bp, representing β3a-AR and β3b-AR transcripts, respectively.20 Sequencing confirmed the specificity of the obtained bands (data not shown). Immunoblotting analysis revealed that mouse kidney cortex and total medulla expressed a band of 44 kDa for the core protein and 1 at 68 kDa for the glycosylated form in all samples (Figure 1b). Both bands were also revealed in β3-AR–expressing control tissues. Immunolocalization of β3-ARs in the mouse kidney As shown in Figure 2, β3-AR was expressed at the apical and basolateral membrane of the epithelial cells of the thin ascending limb, identified by the presence of the kidney-specific chloride channel ClC-K1.21, 22 β3-AR was also localized at the basolateral membrane of the epithelial cells of (i) the thick ascending limb of Henle, expressing the apical NKCC2 cotransporter23; (ii) the distal convolute tubule, expressing the apical thiazide-sensitive NaCl symporter (NCC)24; and (iii) the cortical CD and the outer medullary CD (the latter not shown), expressing AQP2 at the apical membrane.25 The staining for β3-AR completely disappeared when the anti–β3-AR antibody used for immunofluorescence was preadsorbed on its immunizing peptide (Supplementary Figure S1).

itive NaCl symporter (NCC)24; and (iii) the cortical CD and the outer medullary CD (the latter not shown), expressing AQP2 at the apical membrane.25 The staining for β3-AR completely disappeared when the anti–β3-AR antibody used for immunofluorescence was preadsorbed on its immunizing peptide (Supplementary Figure S1). We also demonstrated that β3-AR was neither expressed in the proximal convolute tubule nor in the thin descending limb of Henle’s loop, the inner medullary CD, and the vasa recta (Supplementary Figure S2). Overall, the current data show that β3-AR is localized in those nephron tracts also expressing AVPR2. Effect of β3-AR activation on cAMP production, AQP2 trafficking, and NKCC2 phosphorylation: ex vivo experiments Our finding that β3-AR is expressed in the AVPR2-positive kidney segments prompted us to investigate whether β3-AR activation may mimic the effect of AVP on cAMP production, AQP2 intracellular trafficking, and NKCC2 activation. Using an ex vivo model consisting of freshly isolated mouse kidney tubule suspensions, we measured changes in intracellular cAMP concentrations in response to either the specific β3-AR agonist BRL37344 (1, 10, 100 μM) or the AVP analog 1-deamino-8-d-arginine-vasopressin (dDAVP), 10−7 M, used as positive control for cAMP production (Figure 3a). Results are reported as the percentage of the cAMP concentration measured in resting tubules. Treatment with BRL37344 led to a concentration-dependent increase in intracellular cAMP levels, with the maximal effect observed at 10 μM (+173%, P < 0.0001).

, 10−7 M, used as positive control for cAMP production (Figure 3a). Results are reported as the percentage of the cAMP concentration measured in resting tubules. Treatment with BRL37344 led to a concentration-dependent increase in intracellular cAMP levels, with the maximal effect observed at 10 μM (+173%, P < 0.0001). Accordingly, we used 10 μM BRL37344 for all the following experiments performed in freshly isolated live mouse kidney slices, untreated (resting) or incubated with either dDAVP or BRL37344 (Figure 3b). Confocal microscopy showed that both BRL37344 and dDAVP promoted AQP2 accumulation at the luminal plasma membrane of cortical collecting duct cells (Figure 3b, white arrows) compared with the cytoplasmic localization of AQP2 observed in control slices (Figure 3b, white arrowheads). In line with the absence of β3-AR in the inner medullary CD, BRL37344 failed to induce AQP2 apical accumulation in this portion of the CD (not shown). Importantly, the effect of BRL37344 was prevented by preincubation with either the β3-AR-selective antagonist L748,33726 or the protein kinase A inhibitor H89.27

heads). In line with the absence of β3-AR in the inner medullary CD, BRL37344 failed to induce AQP2 apical accumulation in this portion of the CD (not shown). Importantly, the effect of BRL37344 was prevented by preincubation with either the β3-AR-selective antagonist L748,33726 or the protein kinase A inhibitor H89.27 Next, we evaluated the level of NKCC2 phosphorylation under the same experimental conditions using an antibody against the regulatory phosphothreonine residues in the N-terminus of NKCC2.28 Western blotting (Figure 3c and d) showed that p-NKCC2 increased by about 5-fold after BRL37344 treatment compared with resting conditions, an effect comparable to that obtained with dDAVP. Pretreatment with either L748,337 or H89 significantly prevented this effect of BRL37344. Of note, incubation of kidney slices with either L748,337 or H89 alone did not change AQP2 subcellular localization or NKCC2 phosphorylation compared with resting slices (not shown). To confirm that these effects of BRL37344 were ascribable to β3-AR stimulation, we repeated these experiments on live kidney slices from β3-AR–null mice (β3-AR−/−).29 Importantly, in the absence of β3-AR functional expression, BRL37344 promoted neither AQP2 apical accumulation (Figure 4a) nor NKCC2 phosphorylation (Figure 4b and c). In addition, in β3-AR−/− mice, 10 μM BRL37344 was unable to promote intracellular cAMP elevation in isolated kidney tubules (not shown).

e (β3-AR−/−).29 Importantly, in the absence of β3-AR functional expression, BRL37344 promoted neither AQP2 apical accumulation (Figure 4a) nor NKCC2 phosphorylation (Figure 4b and c). In addition, in β3-AR−/− mice, 10 μM BRL37344 was unable to promote intracellular cAMP elevation in isolated kidney tubules (not shown). Effect of β3-AR knockout on water and electrolyte handling in the mouse kidney The effect of β3-AR agonism on AQP2 and NKCC2, the major players involved in antidiuresis, prompted us to investigate whether β3-AR inactivation may affect water and electrolyte handling in the kidney in vivo. To this end, we evaluated these parameters in β3-AR−/− mice29 lacking β3-AR functional expression and β1-2-AR−/− knockout mice,30 in which β3-AR is the only expressed β-AR. Age-matched wild-type (wt) mice of each strain were used as controls (β3-AR+/+ and β1-2-AR+/+). Strikingly, in β3-AR−/−, diuresis was higher (by 77%), urine osmolality was lower (by 30%), and water intake was increased (by 40%) compared with β3-AR+/+ (Figure 5a). In contrast, urine parameters and water intake were comparable between β1-2-AR+/+ and β1-2-AR−/− mice (Figure 5b). No significant differences in food intake were observed between mouse strains (not shown). In line with these results, immunofluorescence analysis showed that, compared with control β3-AR+/+ mice, β3-AR−/− mice have reduced AQP2 plasma membrane expression and increased subapical localization (Figure 5c).

Effect of β3-AR knockout on water and electrolyte handling in the mouse kidney The effect of β3-AR agonism on AQP2 and NKCC2, the major players involved in antidiuresis, prompted us to investigate whether β3-AR inactivation may affect water and electrolyte handling in the kidney in vivo. To this end, we evaluated these parameters in β3-AR−/− mice29 lacking β3-AR functional expression and β1-2-AR−/− knockout mice,30 in which β3-AR is the only expressed β-AR. Age-matched wild-type (wt) mice of each strain were used as controls (β3-AR+/+ and β1-2-AR+/+). Strikingly, in β3-AR−/−, diuresis was higher (by 77%), urine osmolality was lower (by 30%), and water intake was increased (by 40%) compared with β3-AR+/+ (Figure 5a). In contrast, urine parameters and water intake were comparable between β1-2-AR+/+ and β1-2-AR−/− mice (Figure 5b). No significant differences in food intake were observed between mouse strains (not shown). In line with these results, immunofluorescence analysis showed that, compared with control β3-AR+/+ mice, β3-AR−/− mice have reduced AQP2 plasma membrane expression and increased subapical localization (Figure 5c). Analysis of urine electrolytes, reported in Table 1, showed that β3-AR−/− mice have significantly higher urine excretion of Na+, K+, and Cl− compared with their age-matched β3-AR+/+. Instead, the plasma concentration of the same electrolytes and the GFR were comparable between β3-AR−/− and β3-AR+/+ mice. These results suggest reduced activity of the NKCC2 transporter in β3-AR−/− mice.

-AR−/− mice have significantly higher urine excretion of Na+, K+, and Cl− compared with their age-matched β3-AR+/+. Instead, the plasma concentration of the same electrolytes and the GFR were comparable between β3-AR−/− and β3-AR+/+ mice. These results suggest reduced activity of the NKCC2 transporter in β3-AR−/− mice. Immunofluorescence analysis showed that in β3-AR−/− mice, the antibody against phosphorylated NKCC2 detected a lower amount of activated NKCC2 in the outer medulla compared with β3-AR+/+ mice (Figure 5d). To further support this evidence, we analyzed the effects of bumetanide injection on natriuresis in both β3-AR−/− and β3-AR+/+ mice (Supplementary Figure 3c and d). Natriuresis was higher in β3-AR+/+ than in β3-AR−/− mice (355.4 ± 21.55% vs. 287 ± 27.5%; P < 0.0001), confirming that β3-AR−/− mice have less basal NKCC2 cotransporter activity to inhibit. Of note, the maximal urine concentrating ability of β3-AR−/− mice on a water deprivation test was comparable to that of β3-AR+/+ mice (Supplementary Figure S3a and b).

To further support this evidence, we analyzed the effects of bumetanide injection on natriuresis in both β3-AR−/− and β3-AR+/+ mice (Supplementary Figure 3c and d). Natriuresis was higher in β3-AR+/+ than in β3-AR−/− mice (355.4 ± 21.55% vs. 287 ± 27.5%; P < 0.0001), confirming that β3-AR−/− mice have less basal NKCC2 cotransporter activity to inhibit. Of note, the maximal urine concentrating ability of β3-AR−/− mice on a water deprivation test was comparable to that of β3-AR+/+ mice (Supplementary Figure S3a and b). Effect of β3-AR stimulation on urine output Next, to uncover the possible antidiuretic effect of pharmacologic stimulation of β3-AR in mice, we examined whether BRL37344 could per se induce antidiuresis. β3-AR+/+ and β3-AR−/− mice received a single i.p. injection of BRL37344 (0.6 mg/kg) or phosphate-buffered saline (PBS) alone (vehicle). Urine samples were collected for 4 hours after injections, the first time point at which all BRL37344-treated animals began to urinate. Diuresis, urine osmolality, and urine electrolyte excretion were analyzed and are shown in Figure 6. Notwithstanding the differing diuresis in β3-AR−/− and β3-AR+/+ mice, we expressed our results as a percentage of the values measured in vehicle-treated animals of each genotype. Strikingly, BRL37344 greatly reduced the diuresis in β3-AR+/+ mice but not in β3-AR−/− mice (Figure 6a). Concomitantly, BRL37344 significantly increased urine osmolality only in β3-AR+/+ mice (Figure 6b). Interestingly, in β3-AR+/+ mice, urine excretion of Na+, K+, and Cl−, normalized to the volume of diuresis, were significantly reduced by BRL37344 (Figure 6c–e). Of note, the GFR in β3-AR+/+ mice, measured at 1, 2, 3, and 4 hours after BRL37344 treatment, was not affected (Figure 6f).

only in β3-AR+/+ mice (Figure 6b). Interestingly, in β3-AR+/+ mice, urine excretion of Na+, K+, and Cl−, normalized to the volume of diuresis, were significantly reduced by BRL37344 (Figure 6c–e). Of note, the GFR in β3-AR+/+ mice, measured at 1, 2, 3, and 4 hours after BRL37344 treatment, was not affected (Figure 6f). Effect of β3-AR stimulation on diuresis of mice lacking AVPR2 Next, we investigated whether the potent antidiuretic effect of BRL37344 observed in β3-AR+/+ mice could bypass the inactivation of the AVP signaling in mice lacking AVPR2.12, 31 Mice received a single i.p. injection of BRL37344 (0.6 mg/kg) or PBS alone (vehicle). Urine samples were collected every hour for 3 hours, and diuresis (Figure 7a) and urine osmolality (Figure 7b) were reported. Strikingly, 1 hour after the injection, the urine output of all BRL37344-treated mice was reduced to zero compared with vehicle-treated mice (Figure 7a, 1 hour). Therefore, we could not measure urine osmolality at this time point (Figure 7b, 1 hour). Two hours after injection, the diuresis of BRL37344-treated mice was still dramatically reduced compared with vehicle-treated animals (Figure 7a, 2 hours) and urine osmolality increased (Figure 7b, 2 hours). Three hours postinjection, the effect BRL37344 on diuresis still persisted (Figure 7a, 3 hours), whereas that on urine osmolality partially reversed (Figure 7b, 3 hours).

d mice was still dramatically reduced compared with vehicle-treated animals (Figure 7a, 2 hours) and urine osmolality increased (Figure 7b, 2 hours). Three hours postinjection, the effect BRL37344 on diuresis still persisted (Figure 7a, 3 hours), whereas that on urine osmolality partially reversed (Figure 7b, 3 hours). Discussion The possible expression and physiologic role of β3-AR in the kidney has not been investigated in depth thus far. The present results show β3-AR expression in the same AVPR2-expressing tubules. Because it is known that, similar to AVPR2, β3-AR activates the cAMP pathway,32 we hypothesized that pharmacologic stimulation of β3-AR might regulate the trafficking/activity of AQP2 and NKCC2 involved in the AVP-elicited antidiuresis in the kidney.2, 3 We first demonstrated that BRL37344 significantly increases cAMP production and promotes both AQP2 apical accumulation and NKCC2 phosphorylation/activation, suggesting that, similar to AVP, β3-AR agonists may increase reabsorption of water and solutes in the kidney. The pharmacologic profile of BRL37344 indicates that it may have an intrinsic activity at β1-ARs or β2-ARs.33 As shown by the current results, BRL37344 effects on AQP2 and NKCC2 are prevented by the β3-AR antagonist L748,337 and are not observed in β3-AR−/− mice, thus supporting the notion that BRL37344 acts selectively at β3-AR at the dose used and excluding an off-target effect.

e an intrinsic activity at β1-ARs or β2-ARs.33 As shown by the current results, BRL37344 effects on AQP2 and NKCC2 are prevented by the β3-AR antagonist L748,337 and are not observed in β3-AR−/− mice, thus supporting the notion that BRL37344 acts selectively at β3-AR at the dose used and excluding an off-target effect. Here we also show that β3-AR−/− mice are characterized by mild polyuria, lower urine osmolality, and increased urinary excretion of Na+, K+, and Cl− but not Ca++. Increased water excretion is in line with the observed reduced plasma membrane expression of AQP2 in the cortical collecting duct of β3-AR−/−. In addition, increased Na+, K+, and Cl− excretion in β3-AR−/− is in line with decreased NKCC2 activity, as also supported by the findings that β3-AR−/− mice show less activated NKCC2 at the plasma membrane and a less pronounced natriuretic response to bumetanide. The fact that food consumption in β3-AR−/− mice is comparable to that of β3-AR+/+ mice (Table 1) seems to exclude that solute diuresis can explain the polyuria of β3-AR−/− mice. Neither defect of AVP release (central polydipsia) can explain the polyuria of β3-AR−/− mice because these mice show normal urine-concentrating abilities under a water deprivation challenge.

is comparable to that of β3-AR+/+ mice (Table 1) seems to exclude that solute diuresis can explain the polyuria of β3-AR−/− mice. Neither defect of AVP release (central polydipsia) can explain the polyuria of β3-AR−/− mice because these mice show normal urine-concentrating abilities under a water deprivation challenge. We also show that β3-AR−/− mice have normal plasma levels of Na+, K+, Cl−, and Ca++ indicating that their polyuric phenotype is neither induced by hypercalciuria/hypercalcemia nor by hypokalemia.34, 35, 36 In addition, the polyuria in β3-AR−/− mice is not a consequence of an increased GFR, which is comparable to that in β3-AR+/+ mice. On the other hand, β1-2-AR−/− do not show alterations of urine output and osmolality, suggesting that β3-AR, rather β1-AR and β2-AR, regulate these urine parameters. However, the question whether β3-AR is more important than β1-AR and β2-AR in baseline renal function cannot be solved by the current study as we cannot compare the urine-concentrating ability of β3-AR−/− and β1-2-AR−/− mice. The 2 strains result from a different genetic background, and early studies showed that renal parameters significantly differ in mice of different strains.37

ude that the effects of BRL37344 on urine output may be related to the systemic effects of the drug on arterial pressure. However, it has been shown in rats that BRL37344 reduces arterial pressure by ∼14%42; therefore, it is unlikely that such an effect may be responsible for the observed 70% reduction in urine output. In conclusion, our experimental data indicate that (i) in mice, β3-ARs are expressed in most of the AVP-sensitive nephron segments; (ii) β3-AR stimulation promotes AQP2 plasma membrane accumulation and NKCC2 activation, thus increasing water and salt reabsorption in the kidney tubule; (iii) this effect is likely mediated by an increase of intracellular cAMP; and (iv) β3-AR agonism induces antidiuresis in mice lacking AVPR2. Taken together, these data suggest an unexplored role of sympathetic stimulation via the β3-AR in promoting antidiuresis under physiologic conditions. Some evidence indicates that there is a synaptic contact between renal sympathetic varicosities and renal tubular epithelial cell basolateral membranes.12, 43 In this respect, the current data support the hypothesis that sympathetic stimulation of β3-ARs, upregulating NKCC2 and AQP2 activity, can enhance solutes and water reabsorption in the nephron, thus eliciting an antidiuretic effect. Although we restricted our investigation to the regulatory role of β3-ARs on AQP2 and NKCC2, the possible effect of β3-AR stimulation on other Na/Cl transporters or additional AQPs, participating in the countercurrent multiplier system, is worth further investigation.

the nephron, thus eliciting an antidiuretic effect. Although we restricted our investigation to the regulatory role of β3-ARs on AQP2 and NKCC2, the possible effect of β3-AR stimulation on other Na/Cl transporters or additional AQPs, participating in the countercurrent multiplier system, is worth further investigation. The observation that β3-AR−/− mice are polyuric but show normal urine-concentrating ability during water deprivation suggest that, under physiologic conditions, β3-AR activation by sympathetic nerves does not provide an additional mechanism, corroborating the kidney antidiuretic response to AVP. Although much work remains to be done to fully understand the role of β3-ARs in water and salt reabsorption during sympathetic activation, the current results are potentially relevant for the development of novel pharmacologic approaches to the treatment of diseases caused by AVPR2-altered signaling, including XNDI, polycystic kidney diseases, and the syndrome of inappropriate secretion of AVP. For instance, in XNDI patients, β3-AR agonists may bypass the lack of AVPR2 function, restore NKCC2 and AQP2 activity, and improve the unpaired urine concentration mechanism. It must be emphasized that patients with autosomal forms of NDI40 due to mutations of the AQP2 gene would not benefit from this potential treatment.

instance, in XNDI patients, β3-AR agonists may bypass the lack of AVPR2 function, restore NKCC2 and AQP2 activity, and improve the unpaired urine concentration mechanism. It must be emphasized that patients with autosomal forms of NDI40 due to mutations of the AQP2 gene would not benefit from this potential treatment. Further studies are needed to verify this proof-of-concept, but the ameliorative effect of BRL37344 on renal concentrating abilities of AVPR2-null mice strongly encourages studies in this direction. In particular, we suggest that agonists of the human β3-AR, such as mirabegron,44 already used to treat an overactive bladder, may either improve the impaired concentrating ability of the kidney or increase the beneficial effects of the current XNDI therapy.

VPR2-null mice strongly encourages studies in this direction. In particular, we suggest that agonists of the human β3-AR, such as mirabegron,44 already used to treat an overactive bladder, may either improve the impaired concentrating ability of the kidney or increase the beneficial effects of the current XNDI therapy. Materials and Methods Antibodies and reagents Polyclonal antibodies against β3-AR (cat. nos. sc-50436 and sc-1473) were obtained from Santa Cruz Biotechnology (Dallas, TX) and were previously validated for Western blotting and immunofluorescence analysis.45, 46 Antibodies against AQP1 (cat. no. sc-20810), and CD-31 (cat. no. sc-1506), BRL37344 (cat. no. sc-200154), L748,337 (cat. no. sc-204044) were from Santa Cruz Biotechnology. H-89 (cat. no. B1427), and [deamino-Cys1, D-Arg8]-vasopressin (dDAVP, cat. no. V-1005) were from Sigma (St. Louis, MO). Antibody anti-CLC-K (cat. no. ACL-004) was from Alomone Labs (Jerusalem, Israel). Antibodies anti-NKCC2 (cat. no. AB3562P) were from Merck Millipore (Billerica, MA). Antibodies anti-NCC (cat. no. SPC-402D) were from StressMarq Biosciences Inc. (Victoria, BC, Canada). The antibody against human AQP2 was previously described.47 The antibody against the phosphorylated threonines 96 and 101 of phosphorylated mouse NKCC228 was kindly provided by Prof. Biff Forbush, Yale University.

MA). Antibodies anti-NCC (cat. no. SPC-402D) were from StressMarq Biosciences Inc. (Victoria, BC, Canada). The antibody against human AQP2 was previously described.47 The antibody against the phosphorylated threonines 96 and 101 of phosphorylated mouse NKCC228 was kindly provided by Prof. Biff Forbush, Yale University. β3-AR pharmacology BRL37344 is a well-known β3-AR agonist48 that has been previously used in mice.49, 50, 51 BRL37344 displays a rank order of potency at the human β-ARs, that is, β3-AR > β2-AR > β1-AR, with an approximately 20-fold and 100-fold higher selectivity for β3-AR versus β2-AR and β1-AR, respectively.52 BRL37344 has been found to be effective at 10 μM in the human isolated internal anal sphincter model,53 in human retinal endothelial cells,54 and in mouse retinal explants.55 L748,337 has been reported as one of the very few antagonists with a high selectivity for β3-AR.56 Nonetheless, L748,337 remains the most suitable β3-AR antagonist currently available.26 RNA isolation and reverse transcriptase polymerase chain reaction Total RNA was extracted from mouse brown adipose tissue, the bladder, and the kidney by the TRIzol reagent and reverse-transcribed into cDNA using SuperScript VILO cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA).

L748,337 has been reported as one of the very few antagonists with a high selectivity for β3-AR.56 Nonetheless, L748,337 remains the most suitable β3-AR antagonist currently available.26 RNA isolation and reverse transcriptase polymerase chain reaction Total RNA was extracted from mouse brown adipose tissue, the bladder, and the kidney by the TRIzol reagent and reverse-transcribed into cDNA using SuperScript VILO cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA). The mouse β3-AR intron-spanning primers were previously reported.20 As a positive control, mouse β-actin cDNA was amplified using specific primers. Polymerase chain reaction was performed using Taq DNA polymerase recombinant (Life Technologies, Carlsbad, CA) according to the following: (94°C, 3 minutes) × 1 cycle and (94°C, 45 seconds; 55°C, 30 seconds; 72°C, 1 minute) × 40 cycles. Amplified products were analyzed on 3% agarose gel. Sequencing was performed by BMR Genomics (Padova, Italy), using the method of Sanger. Cell and tissue fractionation and immunoblotting Brown adipose tissue, bladder, and kidney cortex/total medulla were isolated from male C57BL/6J mice and homogenized in radioimmunoprecipitation assay buffer.57 Where reported, kidney slices were lysed in antiphosphatase buffer.2

The mouse β3-AR intron-spanning primers were previously reported.20 As a positive control, mouse β-actin cDNA was amplified using specific primers. Polymerase chain reaction was performed using Taq DNA polymerase recombinant (Life Technologies, Carlsbad, CA) according to the following: (94°C, 3 minutes) × 1 cycle and (94°C, 45 seconds; 55°C, 30 seconds; 72°C, 1 minute) × 40 cycles. Amplified products were analyzed on 3% agarose gel. Sequencing was performed by BMR Genomics (Padova, Italy), using the method of Sanger. Cell and tissue fractionation and immunoblotting Brown adipose tissue, bladder, and kidney cortex/total medulla were isolated from male C57BL/6J mice and homogenized in radioimmunoprecipitation assay buffer.57 Where reported, kidney slices were lysed in antiphosphatase buffer.2 Fifteen micrograms of each lysate were separated by sodium dodecylsulfate-polyacrylamide gel electrophoresis and analyzed by Western blotting. After blocking with 3% bovine serum albumin, blots were incubated with anti–β3-AR antibody (sc-50436, 1:200) and anti–p-NKCC2 antibodies (1:1000) followed by horseradish peroxidase–conjugated secondary antibody. Blots were revealed by enhanced chemiluminescence, with Chemidoc XRS, equipped with Image Lab Software (Bio-Rad, Hercules, CA) and quantified with ImageJ software.

Fifteen micrograms of each lysate were separated by sodium dodecylsulfate-polyacrylamide gel electrophoresis and analyzed by Western blotting. After blocking with 3% bovine serum albumin, blots were incubated with anti–β3-AR antibody (sc-50436, 1:200) and anti–p-NKCC2 antibodies (1:1000) followed by horseradish peroxidase–conjugated secondary antibody. Blots were revealed by enhanced chemiluminescence, with Chemidoc XRS, equipped with Image Lab Software (Bio-Rad, Hercules, CA) and quantified with ImageJ software. Immunofluorescence Mouse kidneys were fixed with 4% paraformaldehyde in PBS at 4°C, dehydrated in graded ethanol, and embedded in paraffin wax. Serial sections, 5 μm thick, were deparaffinized, rehydrated, and subjected to immunofluorescence analysis. Antigen retrieval was performed by boiling sections in citrate buffer (10 mM sodium citrate, pH 6). After blocking with 1% bovine serum albumin in PBS for 30 minutes, sections were incubated with the primary antibodies β3-AR (sc-1473), AQP2, AQP1, CLC-K, NKCC2, CD31, NCC, and phosphorylated NKCC2. Sections were incubated with AlexaFluor-conjugated secondary antibodies (Life Technologies). Confocal images were obtained with a confocal microscope (TSC-SP2, Leica; Wetzlar, Germany).

Immunofluorescence Mouse kidneys were fixed with 4% paraformaldehyde in PBS at 4°C, dehydrated in graded ethanol, and embedded in paraffin wax. Serial sections, 5 μm thick, were deparaffinized, rehydrated, and subjected to immunofluorescence analysis. Antigen retrieval was performed by boiling sections in citrate buffer (10 mM sodium citrate, pH 6). After blocking with 1% bovine serum albumin in PBS for 30 minutes, sections were incubated with the primary antibodies β3-AR (sc-1473), AQP2, AQP1, CLC-K, NKCC2, CD31, NCC, and phosphorylated NKCC2. Sections were incubated with AlexaFluor-conjugated secondary antibodies (Life Technologies). Confocal images were obtained with a confocal microscope (TSC-SP2, Leica; Wetzlar, Germany). Preparation of kidney tubule suspensions and cAMP assay Kidneys from FVB/C57/129/DBA mice (10-week old males) were minced and enzymatically digested as previously reported.10 Aliquots of tubule suspensions were preincubated with the phosphodiesterase inhibitor IBMX for 10 minutes at 37°C. Subsequently, BRL37344 (1, 10, and 100 μM) or dDAVP (100 nM) were added, and reactions were carried out for 45 minutes at 37°C. Total intracellular cAMP was determined by enzyme-linked immunosorbent assay, as previously reported.10

ions were preincubated with the phosphodiesterase inhibitor IBMX for 10 minutes at 37°C. Subsequently, BRL37344 (1, 10, and 100 μM) or dDAVP (100 nM) were added, and reactions were carried out for 45 minutes at 37°C. Total intracellular cAMP was determined by enzyme-linked immunosorbent assay, as previously reported.10 Kidney tissue slices: preparation and treatment C57BL/6J male mice were anesthetized with tribromoethanol (250 mg/kg) and killed by cervical dislocation. Kidneys were excised, and thin transversal slices (250 μm) were cut using a McIlwain Tissue Chopper (Ted Pella Inc.; Redding, CA, United States). Slices were left at 37°C for 15 minutes in Dulbecco’s Modified Eagle Medium/F12 medium pre-equilibrated with 5% CO2, then stimulated for 40 minutes with dDAVP (10−7 M) or BRL37344 (10−5 M), the latter alone or after 30 minutes of preincubation with either L748,337 (10−7 M) or H89 (10−5 M). Slices were either processed for immunoblotting analysis or fixed in 4% paraformaldehyde and processed for immunofluorescence as described previously. Animal studies All animal experiments were approved by the Institutional Committee on Research Animal Care, in accordance with the Italian Institute of Health Guide for the Care and Use of Laboratory Animals. Mice were maintained on a 12-hour light/12-hour dark cycle, with free access to water and food.

Kidney tissue slices: preparation and treatment C57BL/6J male mice were anesthetized with tribromoethanol (250 mg/kg) and killed by cervical dislocation. Kidneys were excised, and thin transversal slices (250 μm) were cut using a McIlwain Tissue Chopper (Ted Pella Inc.; Redding, CA, United States). Slices were left at 37°C for 15 minutes in Dulbecco’s Modified Eagle Medium/F12 medium pre-equilibrated with 5% CO2, then stimulated for 40 minutes with dDAVP (10−7 M) or BRL37344 (10−5 M), the latter alone or after 30 minutes of preincubation with either L748,337 (10−7 M) or H89 (10−5 M). Slices were either processed for immunoblotting analysis or fixed in 4% paraformaldehyde and processed for immunofluorescence as described previously. Animal studies All animal experiments were approved by the Institutional Committee on Research Animal Care, in accordance with the Italian Institute of Health Guide for the Care and Use of Laboratory Animals. Mice were maintained on a 12-hour light/12-hour dark cycle, with free access to water and food. β3-AR−/− and β3-AR+/+ mice58 were purchased from Jackson Laboratory (Bar Harbor, ME, United States). β1-2-AR−/− and β1-2-AR+/+ mice30 were generated as previously described.59, 60 Metabolic cages were used to measure urine output, osmolality, and water intake. Mice received a single i.p. injection of BRL37344 (0.6 mg/kg) or PBS alone. Electrolytes were measured using the ion selective electrode method. The GFR of conscious mice was measured as previously reported.61 AVPR2 knockout mice (V2Rfl/fl and V2Rfl/y Esr1-Cre mice) were previously described.31

β3-AR−/− and β3-AR+/+ mice58 were purchased from Jackson Laboratory (Bar Harbor, ME, United States). β1-2-AR−/− and β1-2-AR+/+ mice30 were generated as previously described.59, 60 Metabolic cages were used to measure urine output, osmolality, and water intake. Mice received a single i.p. injection of BRL37344 (0.6 mg/kg) or PBS alone. Electrolytes were measured using the ion selective electrode method. The GFR of conscious mice was measured as previously reported.61 AVPR2 knockout mice (V2Rfl/fl and V2Rfl/y Esr1-Cre mice) were previously described.31 V2Rfl/y Esr1-Cre mice received a single i.p. injection of BRL37344 (0.6 mg/kg) or PBS alone and urine output and osmolality were monitored every hour for 3 hours. Urine osmolality was measured using a vapor pressure osmometer. Statistical analysis For statistical analysis, GraphPad Prism software (La Jolla, CA) was used. The statistical analysis performed is indicated in the figure legends. Disclosure All the authors declared no competing interests. Supplementary Material Figure S1 Pre-adsorption of anti β3-AR antibody on its immunizing peptide completely abolished the immunostaining of β3-AR in mouse kidney sections. Goat antiβ3-AR (#SC1473) was preadsorbed on its immunizing peptide (SC1473p) and used to immunostained paraffin-embedded mouse kidney sections. Compared to whole antibody, immunodepleted antibody failed to detect β3-AR-positive tubule in both kidney cortex and medulla (bar = 50 μm).

g of β3-AR in mouse kidney sections. Goat antiβ3-AR (#SC1473) was preadsorbed on its immunizing peptide (SC1473p) and used to immunostained paraffin-embedded mouse kidney sections. Compared to whole antibody, immunodepleted antibody failed to detect β3-AR-positive tubule in both kidney cortex and medulla (bar = 50 μm). Figure S2 Immunolocalization of β3-AR in mouse kidney. Paraffin-embedded kidney sections (C57BL6/J, wt) were immunostained with anti β3-AR antibodies (green) and co-stained with antibodies against specific markers of different segments of the kidney tubule or vasculature: AQP1 for the proximal tubule (PT) and the thin descending limb (TDL), AQP2 for the inner medullary collecting duct (IMCD) and CD31 for the endothelium of Vasa Recta (all in red). Overlay of the each double staining experiment indicated that β3-AR was neither expressed in the PT nor in the TDL nor in the IMCD nor in the Vasa Recta. Drawings of the nephron on the right column indicated in red the β3-AR-negative tubule or vascular portions. β3-AR was expressed at the basolateral plasma membrane in all the tubules where it is expressed. Same results were obtained in at least 5 animals (bar = 20 μm).

in the TDL nor in the IMCD nor in the Vasa Recta. Drawings of the nephron on the right column indicated in red the β3-AR-negative tubule or vascular portions. β3-AR was expressed at the basolateral plasma membrane in all the tubules where it is expressed. Same results were obtained in at least 5 animals (bar = 20 μm). Figure S3 Water deprivation test and bumetanide-induced diuresis and natriuresis. β3-AR-null mice (β3-AR-/-) and their age-matched controls (β3-AR+/+) (N = 8 for each group), were individually housed in metabolic cages for 24 hours, then 4 animals per group were subjected to water deprivation for 24 hours, while 4 animals had free access to water (basal). The 24-hour urine output (A) and urine osmolality (B) of control animals were set as 100%. Urine output of water-deprived animals was reduced of 32% in β3-AR+/+ mice and of 40% in β3-AR-/- mice. Urine osmolality of water-deprived animals was increased of 27% in β3-AR+/+ mice and of 28% in β3-AR-/- mice. Data are provided as mean ± SEM. Significant differences between means were tested by one-way analysis of variance ANOVA with Newman-Keuls’s post-test. **P < 0.001, *P <0.05 No significant interstrain differences were observed. Urine output (C) and natriuresis (D) after the i.p. injection of vehicle or 40 mg/kg bumetanide for 4 hours in β3-AR+/+ and β3-AR-/- mice. The 4-hour urine output and urine Na+ excretion of control animals was set as 100% (n = 5). Data are provided as mean ± SEM. Significant differences between means were tested by one-way analysis of variance ANOVA with Newman-Keuls’s post-test. ***P < 0.0001 for intrastrain differences between vehicle and bumetanide treatments; §P < 0.01 for interstrain differences in the effects of bumetanide. The bumetanide-induced natriuresis was significantly attenuated in β3-AR-/- mice compared with β3-AR+/+.

-way analysis of variance ANOVA with Newman-Keuls’s post-test. ***P < 0.0001 for intrastrain differences between vehicle and bumetanide treatments; §P < 0.01 for interstrain differences in the effects of bumetanide. The bumetanide-induced natriuresis was significantly attenuated in β3-AR-/- mice compared with β3-AR+/+. Acknowledgments This work was funded by grants GGP12040 and GGP15083 from the Fondazione Telethon to MS, from grant MRAR08P011 from the Agenzia Italiana del Farmaco (AIFA) to MS, and grant RF02351158 from the Ministero della Salute to PB. We are grateful to J. H. Wess, National Institute of Diabetes and Digestive and Kidney Diseases, for providing the V2Rfl/y and V2Rfl/fl mice. The antibody against the phosphorylated threonines 96 and 101 of mouse NKCC2 (p-NKCC2) was kindly provided by Prof. B. Forbush, Yale University. We are also grateful to Silvia Torretta, Gaetano De Vito, Maurizio Cammalleri, Filippo Locri, Vincenzo Calderone, Alma Martelli, and Dominga Lapi for technical assistance with the animal experiments and maintenance of transgenic mouse colonies. see commentary on page 471 Figure S1. Pre-adsorption of anti β3-AR antibody on its immunizing peptide completely abolished the immunostaining of β3-AR in mouse kidney sections. Goat antiβ3-AR (#SC1473) was preadsorbed on its immunizing peptide (SC1473p) and used to immunostained paraffin-embedded mouse kidney sections. Compared to whole antibody, immunodepleted antibody failed to detect β3-AR-positive tubule in both kidney cortex and medulla (bar = 50 μm).

g of β3-AR in mouse kidney sections. Goat antiβ3-AR (#SC1473) was preadsorbed on its immunizing peptide (SC1473p) and used to immunostained paraffin-embedded mouse kidney sections. Compared to whole antibody, immunodepleted antibody failed to detect β3-AR-positive tubule in both kidney cortex and medulla (bar = 50 μm). Figure S2. Immunolocalization of β3-AR in mouse kidney. Paraffin-embedded kidney sections (C57BL6/J, wt) were immunostained with anti β3-AR antibodies (green) and co-stained with antibodies against specific markers of different segments of the kidney tubule or vasculature: AQP1 for the proximal tubule (PT) and the thin descending limb (TDL), AQP2 for the inner medullary collecting duct (IMCD) and CD31 for the endothelium of Vasa Recta (all in red). Overlay of the each double staining experiment indicated that β3-AR was neither expressed in the PT nor in the TDL nor in the IMCD nor in the Vasa Recta. Drawings of the nephron on the right column indicated in red the β3-AR-negative tubule or vascular portions. β3-AR was expressed at the basolateral plasma membrane in all the tubules where it is expressed. Same results were obtained in at least 5 animals (bar = 20 μm).

in the TDL nor in the IMCD nor in the Vasa Recta. Drawings of the nephron on the right column indicated in red the β3-AR-negative tubule or vascular portions. β3-AR was expressed at the basolateral plasma membrane in all the tubules where it is expressed. Same results were obtained in at least 5 animals (bar = 20 μm). Figure S3. Water deprivation test and bumetanide-induced diuresis and natriuresis. β3-AR-null mice (β3-AR-/-) and their age-matched controls (β3-AR+/+) (N = 8 for each group), were individually housed in metabolic cages for 24 hours, then 4 animals per group were subjected to water deprivation for 24 hours, while 4 animals had free access to water (basal). The 24-hour urine output (A) and urine osmolality (B) of control animals were set as 100%. Urine output of water-deprived animals was reduced of 32% in β3-AR+/+ mice and of 40% in β3-AR-/- mice. Urine osmolality of water-deprived animals was increased of 27% in β3-AR+/+ mice and of 28% in β3-AR-/- mice. Data are provided as mean ± SEM. Significant differences between means were tested by one-way analysis of variance ANOVA with Newman-Keuls’s post-test. **P < 0.001, *P <0.05 No significant interstrain differences were observed. Urine output (C) and natriuresis (D) after the i.p. injection of vehicle or 40 mg/kg bumetanide for 4 hours in β3-AR+/+ and β3-AR-/- mice. The 4-hour urine output and urine Na+ excretion of control animals was set as 100% (n = 5). Data are provided as mean ± SEM. Significant differences between means were tested by one-way analysis of variance ANOVA with Newman-Keuls’s post-test. ***P < 0.0001 for intrastrain differences between vehicle and bumetanide treatments; §P < 0.01 for interstrain differences in the effects of bumetanide. The bumetanide-induced natriuresis was significantly attenuated in β3-AR-/- mice compared with β3-AR+/+.

-way analysis of variance ANOVA with Newman-Keuls’s post-test. ***P < 0.0001 for intrastrain differences between vehicle and bumetanide treatments; §P < 0.01 for interstrain differences in the effects of bumetanide. The bumetanide-induced natriuresis was significantly attenuated in β3-AR-/- mice compared with β3-AR+/+. Supplementary material is linked to the online version of the paper at www.kidney-international.org. Figure 1 Expression of β3-ARs mRNA and protein in mouse kidney. (a) Total RNA from mouse kidney was probed for the presence of mRNA coding β3-adrenergic receptors (β3-ARs). Brown adipose tissue and bladder were used as positive controls. Two amplicons corresponding to β3b-AR (337 bp) and β3a-AR (234 bp) were visualized in all samples. Control reverse transcriptase polymerase chain reaction was performed using primers amplifying mouse β-actin. (b) Total protein extracts from mouse kidney cortex and total medulla were analyzed by Western blotting using anti–β3-AR antibodies. Two bands, corresponding to the core and the glycosylated protein, were detected in the kidney fractions at the same molecular size as those revealed in brown adipose and urinary bladder. Experiments were repeated 3 times with comparable results. RT-PCR, reverse transcriptase polymerase chain reaction.

3-AR antibodies. Two bands, corresponding to the core and the glycosylated protein, were detected in the kidney fractions at the same molecular size as those revealed in brown adipose and urinary bladder. Experiments were repeated 3 times with comparable results. RT-PCR, reverse transcriptase polymerase chain reaction. Figure 2 Immunolocalization of β3-AR in mouse kidney. Paraffin-embedded kidney sections (C57BL6/J, wild type) were immunostained with anti–β3-adrenergic receptor (β3-AR) antibodies (green) and costained with antibodies against specific markers of different segments of the kidney tubule: kidney-specific chloride channel (CLC-K) channel for the thin ascending limb (tAL), Na-K-Cl cotransporter (NKCC2) for the thick ascending limb (TAL), NaCl symporter (NCC) for the distal convolute tubule (DCT), and aquaporin 2 (AQP2) for the cortical collecting duct (CCD) (all in red). Overlay of the each double-staining experiment indicated significant expression of β3-AR in the tAL, TAL, DCT, and CCD. Drawings of the nephron on the right show in light green the β3-AR–positive segments. The same results were obtained in 5 different animals (bar = 20 μm).

cortical collecting duct (CCD) (all in red). Overlay of the each double-staining experiment indicated significant expression of β3-AR in the tAL, TAL, DCT, and CCD. Drawings of the nephron on the right show in light green the β3-AR–positive segments. The same results were obtained in 5 different animals (bar = 20 μm). Figure 3 Ex vivo β3-AR activation in a kidney tubule: intracellular cAMP measurements, AQP2 subcellular localization, and NKCC2 phosphorylation. (a) 1-Deamino-8-d-arginine-vasopressin (dDAVP)– and BRL37344 (BRL)-induced cAMP production in mouse kidney tubule suspensions. Freshly isolated tubule suspensions from wild-type mice (12-week-old males, 3 per individual experiment) were pooled and equally distributed into 24-well plates. Samples were treated with dDAVP (10−7 M) or with the indicated concentrations of BRL for 60 minutes at 37 °C. Total cAMP-generated in each well was normalized to the protein content. Three independent experiments were carried out. Data are expressed as the percentage of the cAMP content measured in resting cells ± SEM. Significant differences between means were tested by 1-way analysis of variance with the Newman-Keuls posttest. Significance was accepted for P values < 0.05. ***P < 0.0001, **P < 0.001 compared with resting tubules. ##P < 0.001 compared with 1 μM BRL. (b) Freshly isolated kidney slices (250 μm) were rapidly cut after sacrifice, maintained in CO2-equilibrated culture medium at 37 °C, left untreated (resting), or incubated with dDAVP (10−7 M) or with BRL (10 μM BRL). BRL was also incubated after preincubation with either the β3-AR–antagonist (L748,337, 10−7 M) or the protein kinase A inhibitor (H89) (10−5 M). Slices were treated as described and fixed, and ultrathin sections were stained for aquaporin 2 (AQP2) and β3-adrenergic receptor (β3-AR) and subjected to confocal microscopy. BRL was as effective as dDAVP in promoting AQP2 expression at the apical plasma membrane of cortical and outer medullary collecting duct cells (white arrows) compared with the intracellular localization of AQP2 observed in untreated samples (resting) or samples incubated with BRL after preincubation with L748,337 or H89 (white arrowheads) (bar = 15 μm). (c) Kidney slices were treated as described, then lysed, and total protein extracts underwent Western blotting analysis using the anti–phosphorylated Na-K-Cl cotransporter (pNKCC2) and the anti–total NKCC2 antibodies.

samples incubated with BRL after preincubation with L748,337 or H89 (white arrowheads) (bar = 15 μm). (c) Kidney slices were treated as described, then lysed, and total protein extracts underwent Western blotting analysis using the anti–phosphorylated Na-K-Cl cotransporter (pNKCC2) and the anti–total NKCC2 antibodies. (d) Densitometric analysis showed a five-fold increase in pNKCC2 (normalized to total NKCC2) in samples treated with BRL or dDAVP, and the effect of BRL was significantly prevented by L748,337 and H89. Data are provided as mean ± SEM and expressed as a percentage of the resting condition. Significant differences between means were tested by 1-way analysis of variance with the Newman-Keuls posttest. ***P < 0.001. Comparable results were obtained in 3 different mice. NKCC2, Na-K-Cl cotransporter; NS, not significant.

. Data are provided as mean ± SEM and expressed as a percentage of the resting condition. Significant differences between means were tested by 1-way analysis of variance with the Newman-Keuls posttest. ***P < 0.001. Comparable results were obtained in 3 different mice. NKCC2, Na-K-Cl cotransporter; NS, not significant. Figure 4 BRL37344 failed to induce AQP2 apical expression and NKCC2 phosphorylation in the kidney of β3-AR–null mice. (a) Freshly isolated kidney slices (250 μm) were obtained from β3−/−AR (β3-adrenergic receptor) mice, maintained in CO2-equilibrated culture medium at 37 °C, and left untreated (resting) or incubated with desmopressin (dDAVP, 10−7 M) or BRL37344 (BRL, 10 μM). Slices were fixed, and ultrathin sections (5 μm) were stained for AQP2 (aquaporin 2) and subjected to confocal laser-scanning microscopy. In β3−/−AR mice, BRL was unable to promote AQP2 expression at the apical plasma membrane of cortical and outer medullary collecting duct cells. dDAVP was used as an internal control to promote AQP2 apical expression (bar = 10 μm). Arrows indicate apical plasma membrane staining. Arrowheads indicate intracellular staining. (b) Slices were also lysed and protein extracts subjected to Western blotting analysis with antiphosphorylated Na-K-Cl cotransporter (pNKCC2) and total Na-K-Cl cotransporter (NKCC2) antibodies. (c) Densitometric analysis of pNKCC2, normalized to total NKCC2, showed that in β3−/−AR mice, BRL was unable to increase NKCC2 phosphorylation compared with dDAVP. Data are provided as mean ± SEM and expressed as the percentage of the resting condition. Significant differences between means were tested by 1-way analysis of variance with the Newman-Keuls posttest. ***P < 0.0001. Comparable results were obtained in 3 different mice.

crease NKCC2 phosphorylation compared with dDAVP. Data are provided as mean ± SEM and expressed as the percentage of the resting condition. Significant differences between means were tested by 1-way analysis of variance with the Newman-Keuls posttest. ***P < 0.0001. Comparable results were obtained in 3 different mice. Figure 5 Mice lacking functional expression of β3-AR showed mild polyuria and reduced urine osmolality. (a) β3-adrenergic receptor–null mice (β3-AR−/−) and their age-matched controls (β3-AR+/+) (8 in each group) were individually housed in metabolic cages for 5 days, and 24-hour urine output, urine osmolality, and water intake were measured daily. The analysis reports the mean ± SEM values relative to 24-hour urine collection. In β3-AR−/− mice, urine output was nearly 77% higher, urine osmolality 30% was lower, and water intake was 41% higher compared with control β3-AR+/+ mice. Statistical analysis was performed by unpaired t-test. *P < 0.05. (b) The same experimental protocol was applied to β1-2-AR-null mice (β1-2-AR−/−) and their age-matched controls (β1-2-AR+/+) (8 in each group). No statistically significant difference was observed in urine parameters and water intake between the 2 experimental groups. (c) Immunofluorescence analysis showed that β3-AR−/− mice have reduced plasma membrane expression and higher subapical localization of AQP2 compared with control β3-AR+/+ mice (bar = 10 μm). (d) Immunofluorescence analysis with the antiphosphorylated Na-K-Cl cotransporter (pNKCC2) showed also that β3-AR−/− mice have reduced levels of activated NKCC2 (pNKCC2) (bar = 30 μm). Comparable results were obtained in 3 different mice.

subapical localization of AQP2 compared with control β3-AR+/+ mice (bar = 10 μm). (d) Immunofluorescence analysis with the antiphosphorylated Na-K-Cl cotransporter (pNKCC2) showed also that β3-AR−/− mice have reduced levels of activated NKCC2 (pNKCC2) (bar = 30 μm). Comparable results were obtained in 3 different mice. Figure 6 Effect of β3-AR stimulation on urine concentrating ability in β3-AR+/+mice. β3-adrenergic receptor (β3-AR)–null (β3-AR−/−) mice and their age-matched controls (β3-AR+/+) (10 of each genotype) were individually acclimatized in metabolic cages for 48 hours; 5 received a single i.p. injection of BRL37344 (BRL) (0.6 mg/kg), whereas 5 control animals received phosphate-buffered saline alone (vehicle). Urine samples were collected for 4 hours after injection. Urine output (a) and urine osmolality (b) were measured in β3-AR+/+ and β3-AR−/− mice and expressed as a percentage of control values measured in vehicle-injected animals ± SEM. Urine output of β3-AR+/+ mice decreased ∼70% and urine osmolality increased ∼40% after BRL37344 (BRL) injection. No significant effect was seen in β3-AR−/− mice. Significant differences between means were tested by 1-way analysis of variance with the Newman-Keuls posttest. *P < 0.05, ***P < 0.0001. (c,d,e) Urine excretion of Na+, K+, and Cl–, normalized for the urine volume, measured in β3-AR+/+ mice. Data are reported as a percentage of the values measured in vehicle-injected mice ± SEM. Significant differences between means were tested by the Mann-Whitney U test. *P < 0.05. (f) Glomerular filtration rate (GFR) of β3-AR+/+ conscious mice was measured at 1, 2, 3, and 4 hours after injection of BRL or vehicle alone. No significant difference was found at each time point between BRL- and vehicle-injected mice. Significant differences between means were tested by 2-way analysis of variance with a Bonferroni posttest.

n rate (GFR) of β3-AR+/+ conscious mice was measured at 1, 2, 3, and 4 hours after injection of BRL or vehicle alone. No significant difference was found at each time point between BRL- and vehicle-injected mice. Significant differences between means were tested by 2-way analysis of variance with a Bonferroni posttest. Figure 7 β3-Adrenergic receptor stimulation promotes antidiuresis in mice lacking functional expression of the arginine vasopressin receptor type 2. Ten V2Rfl/yEsr1-Cre mice were acclimatized in mouse metabolic cages for 48 hours; 5 received a single i.p. injection of BRL37344 (0.6 mg/kg), whereas 5 control mice received phosphate-buffered saline alone (vehicle). Urine samples were collected every hour for 3 hours from both groups, and urine output (a) and urine osmolality (b) at each time point were reported. One hour after the injection, the urine output of BRL37344-treated mice was reduced to zero compared with vehicle-injected animals (vehicle) (diuresis, 1 hour). Two hours after injection, urine output of treated mice was still dramatically reduced compared with control animals (diuresis, 2 hours). Urine osmolality increased in BRL37344-injected animals (urine osmolality, 2 hours). At 3 hours postinjection, the effect of BRL on the urine output still persisted (urine output, 3 hours), whereas the effect on urine osmolality partially reversed (urine osmolality, 3 hours). The analysis reports the mean ± SEM values. Significant differences between measurements were tested by 2-way analysis of variance with a Bonferroni posttest for diuresis and by 1-way analysis of variance with a Bonferroni posttest. *P < 0.05; ***P < 0.001.

ine osmolality partially reversed (urine osmolality, 3 hours). The analysis reports the mean ± SEM values. Significant differences between measurements were tested by 2-way analysis of variance with a Bonferroni posttest for diuresis and by 1-way analysis of variance with a Bonferroni posttest. *P < 0.05; ***P < 0.001. Table 1 Plasma electrolyte concentrations, renal 24-h electrolyte excretion, GFRs, and food intake in β3-AR+/+ and β3-AR−/− mice Electrolytes β3-AR+/+ β3-AR−/− P Value Plasma Na+ (mEq/l) 139.0 ± 5.57 141.3 ± 0.67 NS K+ (mEq/l) 6.73 ± 0.29 6.43 ± 0.19 NS Cl− (mEq/l) 108.0 ± 3.22 106.3 ± 2.67 NS Ca2+ (mEq/l) 3.16 ± 0.53 3.34 ± 0.44 NS Urine Na+ (mEq/24 hr) 019 ± 0.02 0.27 ± 0.01 P < 0.01 K+ (mEq/24 hr) 0.18 ± 0.02 0.23 ± 0.01 P < 0.05 Cl− (mEq/24 hr) 0.45 ± 0.04 0.59 ± 0.03 P < 0.01 Ca2+ (mEq/24 hr) 0.005 ± 0.0006 0.005 ± 0.0005 NS GFR (μl/min) 235.5 ± 20.76 253.8 ± 30.94 NS Food intake (g) 5.07 ± 0.08 5.14 ± 0.08 NS Values are means ± SEM of measurements in 8 mice/genotype. Statistical analysis was performed using an unpaired t test. GFR, glomerular filtration rate; NS, not significant.

Kidney transplantation is the best treatment for kidney failure, in terms of length and quality of life and cost-effectiveness,1, 2 but a significant number of patients keep their transplants for less than 10 years,3 returning to dialysis as the transplant fails. The single biggest cause is immune-mediated injury.4 The association between antibody (Ab) against donor human leukocyte antigen (HLA) (donor-specific Ab [DSA]) and graft failure,5 and description of specific histological features constituting antibody-mediated rejection (AMR),6 have advanced our understanding of this problem. Graft failure is usually preceded by a progressive decline in glomerular filtration rate (GFR), although many patients with DSA have stable graft function, and the immunological factors that influence decline in GFR are unknown.

tuting antibody-mediated rejection (AMR),6 have advanced our understanding of this problem. Graft failure is usually preceded by a progressive decline in glomerular filtration rate (GFR), although many patients with DSA have stable graft function, and the immunological factors that influence decline in GFR are unknown. We recently reported the findings of a long-term observational study in patients with a transplant biopsy diagnosis of chronic AMR (CAMR),7 describing the activity of antidonor T cells recognizing donor antigen via the indirect pathway.8 For the first time, we showed that donor antigen presentation by B cells in enzyme-linked immunosorbent spot (ELISPOT) assays of interferon (IFN)-γ production by CD8-depleted peripheral blood mononuclear cells (PBMCs) was seen preferentially in patients with CAMR compared with controls. Importantly, two-thirds of nonreactive samples had evidence of suppression of antidonor IFN-γ production by CD19+ B lymphocytes or CD25+ T cells, challenging the prevalent hypothesis that patients with chronic rejection have lost the ability to regulate antidonor cellular immunity.9 In this report, we expand our findings from the same cohort by describing the dynamic changes in ELISPOT patterns in individual patients and report an association with changes in estimated GFR (eGFR), testing the hypothesis that progression of renal dysfunction is influenced by the activity of antidonor cell-mediated responses. We provide evidence of the predictive accuracy of ELISPOT, above that provided by other clinical factors alone. Finally, in attempting to demonstrate the role of interleukin (IL)-10 in patients with regulated ELISPOT responses, we discovered evidence that B cells activated a well-defined IL-10 autocrine regulatory mechanism in T helper 1 (Th-1) cells, which was involved in suppressing antidonor responses. Further investigation of the importance of cellular immune responses in AMR may promote a deeper understanding of how to treat chronic rejection.

covered evidence that B cells activated a well-defined IL-10 autocrine regulatory mechanism in T helper 1 (Th-1) cells, which was involved in suppressing antidonor responses. Further investigation of the importance of cellular immune responses in AMR may promote a deeper understanding of how to treat chronic rejection. Results Patient groups and outcomes This report concerns 52 patients included in our recent publication7 who had either a protocol (PROTCL, n = 15) or “for-cause” biopsy (BFC, n = 37). Reasons for exclusions and relevant details of those included are provided in the Supplementary Material. Blood samples were collected within a month of biopsy (time point 1) and 9 to 12 months later (time point 2) for analysis of DSA and antidonor IFN-γ responses. There were no graft failures in the PROTCL group, whereas 11 grafts from the BFC group failed. There was no statistically significant difference in median eGFR at time of first ELISPOT between the 11 who had graft failure (39.3 ml/min per 1.73 m2 [interquartile range (IQR) 16.8]) and the other 41 who maintained graft function during the course of the study (45.7 ml/min per 1.73 m2 [IQR 23.9] P = 0.1 Mann-Whitney U test). To assist interpretation of some analyses, changes in eGFR (ΔeGFR) were dichotomized into “deteriorating” (n = 27) and “stable” (n = 25), based on relationship to the median in each of the PROTCL or BFC groups (Figure 1, Supplementary Tables S1B and S1C), and the 2 subgroups created had statistically significant differences in ΔeGFR, despite having similar eGFRs at the time of first ELISPOT (Supplementary Tables S1B and S1C). All 11 patients who lost graft function were in the “deteriorating” subgroup.

e PROTCL or BFC groups (Figure 1, Supplementary Tables S1B and S1C), and the 2 subgroups created had statistically significant differences in ΔeGFR, despite having similar eGFRs at the time of first ELISPOT (Supplementary Tables S1B and S1C). All 11 patients who lost graft function were in the “deteriorating” subgroup. Associations with ΔeGFR Proteinuria, biopsy features, and DSA (Table 1) Proteinuria at the time of biopsy was strongly associated with graft failure, and a protein-to-creatinine ratio >50 was a sensitive marker of graft failure, whereas protein-to-creatinine ratio <50 was highly predictive of graft survival (see legend to Table 1). Protein-to-creatinine ratio was also associated with ΔeGFR, although was relatively insensitive and poorly predictive of whether a patient was “stable” or “deteriorating.” Two specific biopsy features were associated with graft failure, but both appeared relatively insensitive and poorly predictive within the follow-up period. There was no association between ΔeGFR and either of these biopsy features (Supplementary Figure S1 and Supplementary Table S2). The mean fluorescence intensities of serum DSA and lack of association with ELISPOT activity were presented in Shiu et al.7 DSAs were associated with graft failure, but with poor sensitivity, and although they were also associated with ΔeGFR at time point 1 (Supplementary Figure S2), they could not discriminate between stable and deteriorating subgroups, all suggesting that DSA presence was not sensitive, predictive, or relatively specific at discriminating patients with outcomes based on ΔeGFR.

ensitivity, and although they were also associated with ΔeGFR at time point 1 (Supplementary Figure S2), they could not discriminate between stable and deteriorating subgroups, all suggesting that DSA presence was not sensitive, predictive, or relatively specific at discriminating patients with outcomes based on ΔeGFR. ELISPOT patterns (Table 2, Table 3, Table 4, Table 5) Donor-specific reactivity (DSR) was defined as antidonor reactivity above threshold (see Tables 2 and 3) by CD8-depleted PBMCs in IFN-γ ELISPOT assays, after incubation with whole donor-derived proteins, whereas no DSR (NDSR) refers to a subthreshold response. Tables 2 and 3 show the different types of patterns obtained and the numbers of samples with each pattern at time points 1 and 2. In the whole cohort, the greatest loss of GFR was seen in patients with DSR at time point 2 (Figure 2a) and approximately 65% of patients with DSR at this time point appeared in the “deteriorating subgroup”; these relationships just failed to reach statistical significance. However, when analyzed separately, DSR at time point 2 was associated with deterioration in the BFC but not the PROTCL subgroup (Supplementary Figure S3 and Supplementary Table S3).

s with DSR at this time point appeared in the “deteriorating subgroup”; these relationships just failed to reach statistical significance. However, when analyzed separately, DSR at time point 2 was associated with deterioration in the BFC but not the PROTCL subgroup (Supplementary Figure S3 and Supplementary Table S3). ELISPOT patterns also could be defined according to functional B-cell phenotype and in particular whether there was evidence of a B-cell–dependent antidonor response. Using this definition, associations between ΔeGFR and ELISPOT patterns were maintained (Figure 2b, Table 5). Subgroup analysis suggested a strengthening of associations in the PROTCL patients at time point 1, whereas in the BFC subgroup, associations at time point 2 were weakened and failed to reach statistical significance (Supplementary Figure S4 and Supplementary Table S4).

OT patterns were maintained (Figure 2b, Table 5). Subgroup analysis suggested a strengthening of associations in the PROTCL patients at time point 1, whereas in the BFC subgroup, associations at time point 2 were weakened and failed to reach statistical significance (Supplementary Figure S4 and Supplementary Table S4). Multivariate logistic regression analysis The independent predictive value of the ELISPOT assay at time point 1 was considered separately in PROTCL and BFC subgroups, by comparison with other factors potentially associated with graft dysfunction. We estimated a series of multivariate logistic regression models, each of which included groups of related predictive variables: demographics, recipient factors from time of transplantation, donor factors, HLA antibody, protocol biopsy features, and ELISPOT results (Supplementary Tables S5 and S6). The probability of graft dysfunction as estimated by each of the models was then used to build receiver operator characteristic curves to evaluate performance differences across the different models. In the PROTCL group, ELISPOT assay results alone were able to predict development of graft dysfunction better than any other set of risk factors (Supplementary Table S5). Subsequently, elastic net with leave-group-out cross-validation was used to select the optimal model for classification, considering all predictors in a combined model. Results showed that a predictive algorithm that included B-dependent DSR (from the ELISPOT assay) as the only predictor provided the best performance, with an area under the curve (AUC) of 0.84 (95% confidence interval 0.61–1, specificity 0.88, sensitivity 0.80) (Figure 3a). The cross-validated estimate of the AUC was 0.89.

ts showed that a predictive algorithm that included B-dependent DSR (from the ELISPOT assay) as the only predictor provided the best performance, with an area under the curve (AUC) of 0.84 (95% confidence interval 0.61–1, specificity 0.88, sensitivity 0.80) (Figure 3a). The cross-validated estimate of the AUC was 0.89. A similar approach was used for the BFC subgroup (Supplementary Table S6), but the optimal model generated by elastic net and leave-group-out cross-validation identified 5 factors, including B-dependent DSR on ELISPOT assay (the others were HLA Ab status [including DSA mean fluorescence intensity at time of biopsy], C4d in peritubular capillaries (PTC), degree of interstitial fibrosis/tubular atrophy (IF/TA) on biopsy, and proteinuria). This combined model produced a receiver operating characteristic curve with an AUC of 0.85 (95% confidence interval 0.72–1) with a peak of 89% sensitivity and 77% specificity, which was better than any of the individual models (Figure 3b). The cross-validated estimate of the AUC was 0.73. These data indicate that the patterns of antidonor T-cell IFN-γ production, from around the time of biopsy, do have prognostic influence on progression of renal dysfunction, particularly in the PROTCL group.

hich was better than any of the individual models (Figure 3b). The cross-validated estimate of the AUC was 0.73. These data indicate that the patterns of antidonor T-cell IFN-γ production, from around the time of biopsy, do have prognostic influence on progression of renal dysfunction, particularly in the PROTCL group. Dynamic changes in antidonor IFN-γ production and association with eGFR Loss of responsiveness/regulation and dysfunction In contrast to when individual time points were considered in isolation, changes in antidonor ELISPOT reactivity in individuals were strongly associated with ΔeGFR (Tables 4 and 5). To assess this further, we estimated generalized linear mixed models separately for those patients from the BFC cohort (with 2 viable ELISPOT samples), who were NDSR (Figure 4a) or DSR (Figure 4b) at time of biopsy. Results showed a statistically significant interaction between the presence of DSR on follow-up samples, and the time of eGFR assessment in both groups (P = 0.003 for baseline NDSR, and P = 0.0001 for baseline DSR, respectively), indicating that the change in antidonor responses was significantly associated with different patterns of eGFR over time. Remarkably, Figure 4b shows how those patients who had DSR at baseline, but then changed to NDSR, maintained stable function, as opposed to those who remained DSR and showed significant decline. Similarly, patients who developed DSR over follow-up, showed a steeper decline than those who remained DSR negative (Figure 4a).

arkably, Figure 4b shows how those patients who had DSR at baseline, but then changed to NDSR, maintained stable function, as opposed to those who remained DSR and showed significant decline. Similarly, patients who developed DSR over follow-up, showed a steeper decline than those who remained DSR negative (Figure 4a). To assess the whole cohort, and to further address the importance of B-cell phenotype, we performed an analysis of changes in individual patients who had 2 interpretable ELISPOTs. A detailed descriptive analysis is provided in the supplementary file (Results section and Supplementary Table S7) and only a concise interpretation is presented here (Table 6). This analysis showed a significant association between maintenance of nonreactivity or development of regulated donor reactivity at time point 2 and graft stability (Fisher exact P = 0.0417). When analyzed by ΔeGFR, the differences between these groups was significant (Figure 4c). These associations between antidonor ELISPOT pattern changes and eGFR were antigen specific, as an analysis of the responses to control cytomegalovirus and varicella-zoster virus proteins revealed no significant associations with ΔeGFR (Figure 4d), despite the antiviral antigen ELISPOT patterns themselves showing similar changes as those to antidonor proteins (Supplementary Tables S8 and S9). Altogether, these data support the conclusion that a change from a nonresponsive or regulated antidonor response to unregulated, B-cell–dependent antidonor IFN-γ production is associated with a significant decline in eGFR.

These associations between antidonor ELISPOT pattern changes and eGFR were antigen specific, as an analysis of the responses to control cytomegalovirus and varicella-zoster virus proteins revealed no significant associations with ΔeGFR (Figure 4d), despite the antiviral antigen ELISPOT patterns themselves showing similar changes as those to antidonor proteins (Supplementary Tables S8 and S9). Altogether, these data support the conclusion that a change from a nonresponsive or regulated antidonor response to unregulated, B-cell–dependent antidonor IFN-γ production is associated with a significant decline in eGFR. Treatment-associated nonresponsiveness/regulation and stability (Table 7) Changes in antidonor reactivity occurred spontaneously in PROTCL patients, but followed treatment in patients with BFC. To address whether treatment could influence outcome, we selected a homogeneous subgroup of 18 patients with BFC-CAMR, chosen for 3 reasons: (i) they had no tubulitis on biopsy; (ii) they all had ongoing and progressive rises in creatinine, as determined by analysis of reciprocal creatinine plots at the time of first ELISPOT (i.e., this group excluded 5 patients who presented with isolated proteinuria only); and (iii) they were all treated with a protocolized treatment regimen (determined clinically), consisting of addition or optimization of tacrolimus and mycophenolate mofetil followed by i.v. rituximab when oral immunotherapy was thought to have been maximally optimized.

ts who presented with isolated proteinuria only); and (iii) they were all treated with a protocolized treatment regimen (determined clinically), consisting of addition or optimization of tacrolimus and mycophenolate mofetil followed by i.v. rituximab when oral immunotherapy was thought to have been maximally optimized. Three had an eGFR <20 ml/min at the time of first ELISPOT, so were excluded. For the remaining 15, 7 became stable (all in “stable” subgroup) and remained stable after treatment (Supplementary Figure S5). All 7 had time point 2 samples showing nonresponsiveness or regulation, and in 5 of 7, it was clear there had been a shift involving loss of B-dependent responses or development of regulated antidonor reactivity or nonresponsiveness. In the 8 patients who showed a continued decline in eGFR (all in the “deteriorating” subgroup), the picture was more complex. Interpretable time point 2 ELISPOTs were available on 6. Three had unregulated B-dependent antidonor activity at time point 2, and 2 of these had clearly lost evidence of regulation that had been present at time point 1, including a patient (ID 635) in whom loss of regulation by B cells followed treatment with rituximab. The 3 remaining were nonresponsive or had regulated antidonor activity at time point 2, but it is notable that 2 of these had infectious complications beyond the time point 2 ELISPOT that necessitated immunosuppression reduction, perhaps confounding an association between their ELISPOT patterns and outcome.

rituximab. The 3 remaining were nonresponsive or had regulated antidonor activity at time point 2, but it is notable that 2 of these had infectious complications beyond the time point 2 ELISPOT that necessitated immunosuppression reduction, perhaps confounding an association between their ELISPOT patterns and outcome. Multivariate logistic regression analysis on those patients with viable time point 1 ELISPOTs in this subgroup was performed. The combined model in this case included age, sex, previous acute rejection, pretransplant dialysis time, HLA Ab and MHC class I polypeptide-related sequence A (MICA) status, C4d on PTC on biopsy, proteinuria, IF/TA, B-cell–dependent antidonor IFN-γ production assay, and treatment with tacrolimus/mycophenolate mofetil/rituximab and generated a receiver operating characteristic curve with 100% sensitivity and 100% specificity (Figure 3c). In this uniformly treated BFC-CAMR subgroup, ELISPOT pattern was a better predictor of outcome than in the whole cohort and was significantly better than HLA Ab status.

rolimus/mycophenolate mofetil/rituximab and generated a receiver operating characteristic curve with 100% sensitivity and 100% specificity (Figure 3c). In this uniformly treated BFC-CAMR subgroup, ELISPOT pattern was a better predictor of outcome than in the whole cohort and was significantly better than HLA Ab status. Importance of IL-10 and regulation of IFN-γ in Th-1 CD4+ cells As described in detail in the supplementary materials, experiments to assess whether functional Breg (increase in spot count of ≥20% when CD19+ cells depleted) activity involved IL-10 secretion indicated that there was an additional source of IL-10 in ELISPOTs besides B cells. To explore whether T cells might themselves be making IL-10,10, 11 we selected samples with sufficient cells available, stimulated PBMC with donor material, and assessed IL-10 and IFN-γ single or coexpression by CD4+ cells (Figure 5a and 5b). In samples showing evidence of B-cell regulation without any B-dependent antidonor responses on ELISPOT (n = 3), all CD4+ T cells expressing IFN-γ also expressed IL-10, whereas cells expressing IFN-γ alone were evident only in samples in which there was evidence of a B-dependent antidonor response (n = 9). The frequencies of cytokine-positive antidonor CD4+ T cells revealed by these analyses were consistent with those seen in ELISPOT assays. These data suggest that “pure” B-cell regulation in ELISPOT associates with IL-10 expression by IFN-γ–producing CD4+ T cells.

as evidence of a B-dependent antidonor response (n = 9). The frequencies of cytokine-positive antidonor CD4+ T cells revealed by these analyses were consistent with those seen in ELISPOT assays. These data suggest that “pure” B-cell regulation in ELISPOT associates with IL-10 expression by IFN-γ–producing CD4+ T cells. A failure of this autocrine mechanism, resulting in Th-1 cells that produce large amounts of IFN-γ, with a proportional reduction in IL-10 coexpression, has been associated with active rheumatoid arthritis (RA).12 We polyclonally stimulated all samples from which we had sufficient PBMCs (n = 16) to assess if such cells were present (Figure 5c–5e). CD4+ cells from 4 samples produced significantly more IFN-γ than IL-10, consistent with the proinflammatory phenotype seen in patients with RA12; these samples had a high frequency (>20%) of double-positive T cells expressing both IFN-γ and IL-10. The other 12 samples had low frequencies of double-positive CD4+ T cells that secreted as much IL-10 as IFN-γ (Figure 5c–5e). Importantly, all 4 samples containing the highly inflammatory Th-1 cells showed an identical pattern on antidonor ELISPOT assay (B-cell–dependent reactivity without any evidence of regulation by B or T cells), whereas the others showed either no evidence of B-dependency, or regulated B-dependent antidonor activity and appeared to respond similarly to polyclonal stimulation (Figure 5c–5e).

s showed an identical pattern on antidonor ELISPOT assay (B-cell–dependent reactivity without any evidence of regulation by B or T cells), whereas the others showed either no evidence of B-dependency, or regulated B-dependent antidonor activity and appeared to respond similarly to polyclonal stimulation (Figure 5c–5e). All these data suggest that regulation by B cells in antidonor ELISPOT involves activation of an autocrine IL-10 regulatory pathway, to tip the balance from IFN-γ expression by Th-1 cells to preferential IL-10 production, and that a failure of this regulatory pathway associates with unregulated B-cell–dependent antidonor responses. Discussion The association between DSA and graft failure is well established5, 13, 14; however, the significant variability in clinical phenotype associated with DSA15, 16, 17, 18 is difficult to explain. Differences in the functional characteristics of DSA, such as the subclass of IgG19 or the ability to fix complement,20 offer a potential explanation. However, other factors associated with the presence of DSA might influence the progression of pathology, rate of functional deterioration, and timing of eventual graft failure.

nces in the functional characteristics of DSA, such as the subclass of IgG19 or the ability to fix complement,20 offer a potential explanation. However, other factors associated with the presence of DSA might influence the progression of pathology, rate of functional deterioration, and timing of eventual graft failure. HLA Abs are a marker of B-cell activation, which is a T-cell–dependent process involving cognate interactions between B and T cells, so it therefore follows that DSAs are markers of “indirect” CD4+ T-cell sensitization to donor antigens. “Indirect” in this context refers to a specific pathway of allorecognition in which graft antigens are processed into peptide fragments and presented on recipient HLA class II molecules by professional antigen-presenting cells. Indirect responses to mismatched donor HLA have been associated with graft dysfunction and chronic rejection in both renal21, 22, 23, 24, 25 and cardiac allografts.26, 27 Our previous report confirmed that B cells acted as antigen-presenting cells for indirect alloresponses in patients with CAMR,7 and also described the complexities of antidonor reactivity, such that a significant proportion had evidence of active regulation of their antidonor responses. This report addresses the hypothesis that the activity of these cellular immune responses is one of the significant “other factors” that influence the progression of graft dysfunction.

omplexities of antidonor reactivity, such that a significant proportion had evidence of active regulation of their antidonor responses. This report addresses the hypothesis that the activity of these cellular immune responses is one of the significant “other factors” that influence the progression of graft dysfunction. We confirmed, as reported by others, that DSA,28 peritubular capillaritis,29 and IF/TA30 on biopsy, along with proteinuria, were all associated with graft failure. Of these traditional factors, only DSA and proteinuria were associated with ΔeGFR, although with relatively poor sensitivity and predictive value. With regard to ELISPOT patterns, B-cell–dependent antidonor IFN-γ production was the factor most strongly correlated with graft dysfunction in the PROTCL subgroup. Within the BFC subgroup, correlations between B-cell–dependent antidonor reactivity at time point 1 and ΔeGFR were weaker; in this subgroup, time point 2 samples appeared to have a stronger association with outcome, as they did in analysis of the whole cohort. Most impressively, this was evident in patients in whom antidonor reactivity changed from nonresponsiveness or a regulated response at time point 1, to an unregulated B-cell–dependent response at time point 2: these patients showed the greatest loss of GFR. Conversely, those with B-cell–dependent antidonor reactivity who became nonresponsive or developed evidence of T- or B-cell regulation, appeared to stabilize and maintain GFR over the course of the study. This was seen clearly as a “treatment effect” in a subset of selected patients with BFC-CAMR who received an optimized treatment protocol for “creeping creatinine,” in whom the predictive value of the time point 1 ELISPOT pattern was enhanced and was better than HLA Ab status at predicting ΔeGFR. We believe the main reason why univariate associations were not seen in subgroup analyses at all time points was the small number of patients in our analysis, compounded by the fact that ELISPOT interpretations were complex, so that when they were defined according to the patterns revealed by sequential depletion of CD25 /CD19 cells, associations were mostly evident only in analysis of the whole group.

yses at all time points was the small number of patients in our analysis, compounded by the fact that ELISPOT interpretations were complex, so that when they were defined according to the patterns revealed by sequential depletion of CD25 /CD19 cells, associations were mostly evident only in analysis of the whole group. Nevertheless, multivariate testing indicated the ELISPOT pattern to be an independent factor that predicted changes in eGFR in both PROTCL and BFC groups, as demonstrated by the superior AUCs obtained from our prediction modeling when ELISPOT patterns were included. Our statistical methodology was chosen so we could estimate the predictive accuracy of all variables irrespective of their P values; where individual variables were selected for testing, elastic net regression, tuned via cross-validation, was used because of the small sample sizes.

when ELISPOT patterns were included. Our statistical methodology was chosen so we could estimate the predictive accuracy of all variables irrespective of their P values; where individual variables were selected for testing, elastic net regression, tuned via cross-validation, was used because of the small sample sizes. These data provide the first potential explanation for the findings of Wiebe et al.,31 who described patients with DSA, who were compliant with immunosuppressive medication and had stable graft function, suggesting that maintaining “control” of T and B cells with conventional immunosuppression is sufficient to achieve stability of function in some patients with DSA. They are also compatible with the recent report from Shabir et al.,32 who described a link between stable graft function and preserved peripheral transitional B-cell proportions, even in a small number of patients who developed de novo DSA, although we were unable to correlate a specific surface B-cell phenotype with the functional phenotype revealed by ELISPOT.7 Our results also provide a basis for understanding the reports of how enhanced immunosuppression can stabilize function in patients with CAMR. Theruvath et al.33 reported 12-month stabilization of kidney function in 3 of 4 patients with CAMR after transfer onto tacrolimus and mycophenolate mofetil and a short course of prednisolone. In addition, several studies have reported successful stabilization after B-cell depletional therapy,34, 35, 36 supporting the hypothesis that underlying cellular responses are contributing to functional deterioration in these patients.

nsfer onto tacrolimus and mycophenolate mofetil and a short course of prednisolone. In addition, several studies have reported successful stabilization after B-cell depletional therapy,34, 35, 36 supporting the hypothesis that underlying cellular responses are contributing to functional deterioration in these patients. Eight of our patients received rituximab, but no definitive conclusions from these small numbers can be made. However, as well as patient 635 (highlighted previously in this article), who lost evidence of B-cell regulation after rituximab, another patient (326) stabilized eGFR after rituximab in association with development of a regulated antidonor response (there was no change in the median fluorescence intensity of DSA assessed by Luminex [xMAP assay (Thermofisher, California, USA)] in either patient [Supplementary Table S1c and c]). Both cases suggest that response to rituximab might be informed by knowledge of ELISPOT patterns pretreatment.

egulated antidonor response (there was no change in the median fluorescence intensity of DSA assessed by Luminex [xMAP assay (Thermofisher, California, USA)] in either patient [Supplementary Table S1c and c]). Both cases suggest that response to rituximab might be informed by knowledge of ELISPOT patterns pretreatment. It is important to state, as we have before,7 that we were not able to purify putative regulatory T or B cells from patients due to lack of cells. Therefore, the conclusions made are based on indirect evidence of regulation, on depleting specific lymphocyte subsets, rather than a direct demonstration of suppressive activity from purified, and then in vitro assessed cell subpopulations. We presented our flow cytometric analysis of these samples in our earlier article last year, and found no associations between the proportions of B-cell subsets, including transitional or naïve cells and ELISPOT patterns or outcome,7 although we acknowledge that others working in this field have shown associations between rejection and particular B-cell subsets, including transitional B cells.32, 37, 38 In addition, the associations reported in this article need to be validated in different cohorts of patients, ideally with work to assess the reproducibility of the assays within patients and to more carefully document how patterns change over time.

d particular B-cell subsets, including transitional B cells.32, 37, 38 In addition, the associations reported in this article need to be validated in different cohorts of patients, ideally with work to assess the reproducibility of the assays within patients and to more carefully document how patterns change over time. Importantly, for the first time, we have attempted to link antigen-specific indirect alloresponses with mechanisms by which IFN-γ production is regulated in Th-1 CD4+ T cells. Physiological regulation of these cells, by IL-10, is known to be essential, as unchecked IFN-γ production results in severe tissue damage, as illustrated by the responses to Listeria or Trypanosoma disease in IL-10–deficient mice.10 Although multiple cell types can make IL-10, that made by the Th-1 cells themselves39 is the major in vivo source40 and critical to providing regulatory feedback, via antigen-presenting cells and T cells themselves, to prevent inappropriate Th-1–driven immunopathology.41 Aligning with this model, we showed that an anti–IL-10 antibody caused significant increase in the frequency of IFN-γ producing spots in ELISPOT, even in the absence of B cells, and additionally, that in CD8-depleted samples showing evidence of only a B-regulated response, Th-1 cells stimulated by donor antigen only made IFN-γ in the context of coexpression with IL-10. We have not attempted to assess the predominant source of IL-10 in our assays, and our data cannot exclude an important role for B-cell–derived IL-10 in regulation of antidonor alloresponses, as others working in this area have shown.37, 42, 43

by donor antigen only made IFN-γ in the context of coexpression with IL-10. We have not attempted to assess the predominant source of IL-10 in our assays, and our data cannot exclude an important role for B-cell–derived IL-10 in regulation of antidonor alloresponses, as others working in this area have shown.37, 42, 43 Abnormalities of IL-10 switching mediated by CD46 signaling have been associated with excessive IFN-γ production by Th1 cells from synovial fluid of patients with RA.12 T cells from patients with active RA fail to shut down IFN-γ production on CD46-activation, and, perhaps counterintuitively, have high proportions of IFN-γ + IL-10 + double-positive cells, but these express a very large amount of IFN-γ compared with T cells from healthy individuals.10, 12 Using an in vitro system involving polyclonal stimulation through CD46,44 we found cells similar to those described in patients with RA in 4 of 16 samples, all of which demonstrated the same pattern of antidonor response (unregulated B-cell–dependent reactivity), whereas the other 12 samples showing evidence of regulation resembled responses seen in healthy controls.12

lation through CD46,44 we found cells similar to those described in patients with RA in 4 of 16 samples, all of which demonstrated the same pattern of antidonor response (unregulated B-cell–dependent reactivity), whereas the other 12 samples showing evidence of regulation resembled responses seen in healthy controls.12 These data imply, for the first time, that functional suppression by B cells in ELISPOT activates this IL-10 autocrine pathway of regulation to restrict IFN-γ production by Th-1 cells. Moreover, inability to switch off IFN-γ production via this regulatory mechanism associates with a specific pattern of unregulated antidonor response that, as we have demonstrated here, is associated with a greater loss of renal function over time. Better understanding of this regulatory mechanism may lead to the development of more sophisticated treatments for chronic rejection.

gulatory mechanism associates with a specific pattern of unregulated antidonor response that, as we have demonstrated here, is associated with a greater loss of renal function over time. Better understanding of this regulatory mechanism may lead to the development of more sophisticated treatments for chronic rejection. In summary, our analysis of the cell-mediated IFN-γ production by PBMCs against donor antigens has generated 2 important and novel findings. First, a significant association between patterns of antidonor ELISPOT reactivity and eGFR outcomes, with evidence that treatment to influence antidonor responses can affect patient outcomes: these findings support the hypothesis we set out to test, that cell-mediated immunity has a strong influence on deterioration in patients with CAMR. Second, we have defined a novel link between B-cell regulation of IFN-γ production and an IL-10–dependent autocrine mechanism regulating Th-1 CD4+ T cells, with the implication that manipulation of this mechanism might significantly affect the evolution of indirect alloresponses, and ultimately, on long-term allograft survival. Materials and Methods Methodology is exactly as described in a previous report.7 Full details are given in the Methods section of the Supplementary Materials and Methods. Brief descriptions are given here.

In summary, our analysis of the cell-mediated IFN-γ production by PBMCs against donor antigens has generated 2 important and novel findings. First, a significant association between patterns of antidonor ELISPOT reactivity and eGFR outcomes, with evidence that treatment to influence antidonor responses can affect patient outcomes: these findings support the hypothesis we set out to test, that cell-mediated immunity has a strong influence on deterioration in patients with CAMR. Second, we have defined a novel link between B-cell regulation of IFN-γ production and an IL-10–dependent autocrine mechanism regulating Th-1 CD4+ T cells, with the implication that manipulation of this mechanism might significantly affect the evolution of indirect alloresponses, and ultimately, on long-term allograft survival. Materials and Methods Methodology is exactly as described in a previous report.7 Full details are given in the Methods section of the Supplementary Materials and Methods. Brief descriptions are given here. Experimental design and recruitment The study was performed as part of a large observational study looking at the importance of HLA antibodies posttransplantation, the protocol of which was approved by the Hammersmith, Queen Charlotte’s, and Chelsea and Acton Hospitals Research Ethics Committees (2002/6452) and conformed to the 1964 Declaration of Helsinki and subsequent amendments. All participants gave written informed consent before inclusion. Calculation of eGFR, ΔeGFR, and details of blood collection and processing is described in the Methods section of the supplementary materials.

earch Ethics Committees (2002/6452) and conformed to the 1964 Declaration of Helsinki and subsequent amendments. All participants gave written informed consent before inclusion. Calculation of eGFR, ΔeGFR, and details of blood collection and processing is described in the Methods section of the supplementary materials. ELISPOT assay IFN-γ ELISPOT plates (Mabtech AB, Nacka, Sweden) precoated with primary IFN-γ Ab were blocked for 2 hours with “complete medium” (AIM-V medium/10% human AB serum from Life Technologies [Paisley, UK]) before addition of 4 x 105 responder PBMCs per well in 100 μl complete medium with donor antigens. PBMCs were prepared according to standard laboratory protocols. Controls and source of donor antigen are described in detail in the Methods section of the supplementary materials.

rum from Life Technologies [Paisley, UK]) before addition of 4 x 105 responder PBMCs per well in 100 μl complete medium with donor antigens. PBMCs were prepared according to standard laboratory protocols. Controls and source of donor antigen are described in detail in the Methods section of the supplementary materials. Statistical analysis Statistical analyses were performed by using R.45 Two-sided tests were used throughout, and a P < 0.05 was considered statistically significant in univariate statistical testing. Group differences were assessed by using Fisher χ2 test for categorical variables, Wilcoxon rank-sum (Mann-Whitney) or Kruskal-Wallis tests for non-normally distributed continuous variables, and t-test for normally distributed continuous variables (for paired or unpaired samples as appropriate). For prediction analysis, we estimated generalized linear models for predefined groups of predictive variables. Only baseline variables were added to generalized linear prediction models, and thus fixed effects only models were fitted. The estimated predicted probability of outcome was then used to build a receiver operating characteristic curve, and estimate the AUC, sensitivity, and specificity.46 To obtain the optimal combination of predictors of outcome, we used elastic net models. Elastic net is a regularized regression method in which a penalty is imposed on the regression coefficients, which is a combination of the penalties used in lasso and ridge regression. Elastic net enables selection of predictors (unlike ridge regression, which would moderate coefficients but not make them 0), and can handle and select groups of correlated predictors (unlike lasso, which would select only 1 of a group of correlated predictors, and drop the rest).47

used in lasso and ridge regression. Elastic net enables selection of predictors (unlike ridge regression, which would moderate coefficients but not make them 0), and can handle and select groups of correlated predictors (unlike lasso, which would select only 1 of a group of correlated predictors, and drop the rest).47 For the analysis of patterns of change in eGFR over time in patients with and without baseline DSR, we fitted linear mixed-effects models. Separate models were fitted for patients with and without baseline DSR. The model included an intercept, main effects, and interaction between study time point (from prebiopsy to 3-years’ follow-up) and DSR status at time point 2 as fixed effects, as well as a random intercept for the subject. Disclosure All the authors declared no competing interests.

For the analysis of patterns of change in eGFR over time in patients with and without baseline DSR, we fitted linear mixed-effects models. Separate models were fitted for patients with and without baseline DSR. The model included an intercept, main effects, and interaction between study time point (from prebiopsy to 3-years’ follow-up) and DSR status at time point 2 as fixed effects, as well as a random intercept for the subject. Disclosure All the authors declared no competing interests. Supplementary Material Supplementary Methods Figure S1 Lack of association between PTC score and ΔeGFR and between IF/TA and ΔeGFR over the course of the study in combined PROTCL and BFC group. (A) Box plots show median with IQR with whiskers showing upper and lower limits of ΔeGFR. Combined group includes all PROTCL and BFC patients, except 9 patients with BFC who either had missing follow-up data (n = 2) or eGFR ≤20 ml/min per 1.73 m2 at time of biopsy (n = 7). Patients with PTC score <1 have median ΔeGFR of −6.85 ml/min per 1.73 m2 (IQR 12.4) and mean ΔeGFR of −8.7 ml/min per 1.73 m2 (SD ±13.1). Patients with PTC score ≥1 have median ΔeGFR of −8.0 ml/min per 1.73 m2 (IQR 16.7) and mean ΔeGFR of −7.2 ml/min per 1.73 m2 (SD ±15.1). *Mann-Whitney U test. (B) Graph shows lack of correlation between IF/TA % on biopsy and ΔeGFR for each of the patients (n = 52) included in this analysis.

/min per 1.73 m2 (SD ±13.1). Patients with PTC score ≥1 have median ΔeGFR of −8.0 ml/min per 1.73 m2 (IQR 16.7) and mean ΔeGFR of −7.2 ml/min per 1.73 m2 (SD ±15.1). *Mann-Whitney U test. (B) Graph shows lack of correlation between IF/TA % on biopsy and ΔeGFR for each of the patients (n = 52) included in this analysis. Figure S2 Association between DSA and ΔeGFR over the course of the study in combined cohort. Box plots show median with IQR with whiskers showing upper and lower limits of ΔeGFR. Combined group includes all PROTCL and BFC patients, except 9 patients with BFC who either had missing follow-up data (n = 2), or eGFR ≤20 ml/min 1.73 m2 at time of biopsy (n = 7). Time point 1: Patients with No DSA or DSA with cumulative mean fluorescence intensity of <1000 (n = 35) have median ΔeGFR of −3.19 ml/min per 1.73 m2 (IQR 12.3) and mean ΔeGFR of −6.24 ml/min 1.73 m2 (SD ±13.4). Patients with DSA with cumulative mean fluorescence intensity >1000 (n = 17) have median ΔeGFR of −11.4 ml/min per 1.73 m2 (IQR 10.7) and mean ΔeGFR of −12.4 ml/min per 1.73 m2 (SD ±13.3). Time point 2: Patients with No DSA or DSA with cumulative mean fluorescence intensity of <1000 (n = 34) have median ΔeGFR of −3.69 ml/min per 1.73 m2 (IQR 12.4) and mean ΔeGFR of −5.68 ml/min/1.73 m2 (SD ±11.9). Patients with DSA with cumulative mean fluorescence intensity >1000 have median ΔeGFR of −9.9 ml/min per 1.73 m2 (IQR 15) and mean ΔeGFR of −13.0 ml/min per 1.73 m2 (SD ±16). **Mann-Whitney U test.

<1000 (n = 34) have median ΔeGFR of −3.69 ml/min per 1.73 m2 (IQR 12.4) and mean ΔeGFR of −5.68 ml/min/1.73 m2 (SD ±11.9). Patients with DSA with cumulative mean fluorescence intensity >1000 have median ΔeGFR of −9.9 ml/min per 1.73 m2 (IQR 15) and mean ΔeGFR of −13.0 ml/min per 1.73 m2 (SD ±16). **Mann-Whitney U test. Figure S3 Box plots show median with IQR with whiskers showing upper and lower limits of ΔeGFR in PROTCL (A) and BFC (B) subgroups. Time point 1: PROTCL patients with DSR have median ΔeGFR of −8.34 ml/min per 1.73 m2 (IQR 17.4) and mean ΔeGFR of −17.3 ml/min per 1.73 m2 (SD ±17.4), compared with those with NDSR who have a median ΔeGFR of −2.03 ml/min per 1.73 m2 (IQR 6.2) and mean ΔeGFR of −2.38 ml/min per 1.73 m2 (SD ±4.9). Patients with BFC with DSR have median ΔeGFR of −6.4 ml/min per 1.73 m2 (IQR 11.7) and mean ΔeGFR of −7.1 ml/min per 1.73 m2 (SD ±10.7), compared with those with NDSR who have median ΔeGFR of –9.9 ml/min per 1.73 m2 (IQR 20.1) and mean ΔeGFR of −10.3 ml/min per 1.73 m2 (SD ±17.1). Actual P = 0.46. Time point 2: PROTCL patients with DSR have median ΔeGFR of −4.5 ml/min per 1.73 m2 (IQR 7.2) and mean ΔeGFR of −4.6 ml/min per 1.73 m2 (SD ±5.4), compared to those with NDSR who have median ΔeGFR of −4.2 ml/min per 1.73 m2 (IQR 15.8) and mean ΔeGFR of −10.9 ml/min per 1.73 m2 (SD ±15.5). Patients with BFC with DSR have median ΔeGFR of −12.7ml/min per 1.73 m2 (IQR 13.5) and mean ΔeGFR of −13.5 ml/min per 1.73 m2 (SD ±14.6), compared with those with NDSR who have median ΔeGFR of –2.5 ml/min per 1.73 m2 (IQR 12.3) and mean ΔeGFR of −2.4 ml/min per 1.73 m2 (SD ±12.1). Actual P = 0.01. **Mann-Whitney U test.

atients with BFC with DSR have median ΔeGFR of −12.7ml/min per 1.73 m2 (IQR 13.5) and mean ΔeGFR of −13.5 ml/min per 1.73 m2 (SD ±14.6), compared with those with NDSR who have median ΔeGFR of –2.5 ml/min per 1.73 m2 (IQR 12.3) and mean ΔeGFR of −2.4 ml/min per 1.73 m2 (SD ±12.1). Actual P = 0.01. **Mann-Whitney U test. Figure S4 Box plots show median with IQR with whiskers showing upper and lower limits of ΔeGFR in PROTCL (upper panel) and BFC (lower panel) subgroups. Time point 1: PROTCL patients with no evidence of B-dependent antidonor responses have median ΔeGFR of +1.5 ml/min per 1.73 m2 (IQR 3) and mean ΔeGFR of −0.54 ml/min per 1.73 m2 (SD ±2.6), compared with those with evidence of B-dependent antidonor reactivity, who have a median ΔeGFR of −7.7 ml/min per 1.73 m2 (IQR 11.7) and mean ΔeGFR of −13.8 ml/min per 1.73 m2 (SD ±15.4). Patients with BFC with no evidence of B-dependent antidonor responses have median ΔeGFR of −11.1 ml/min per 1.73 m2 (IQR 27.6) and mean ΔeGFR of −6.9 ml/min per 1.73 m2 (SD ±19.7), compared with those with evidence of B-dependent antidonor reactivity, who have median ΔeGFR of –9.7 ml/min per 1.73 m2 (IQR 12) and mean ΔeGFR of −10.3 ml/min per 1.73 m2 (SD ±12.4). Actual P = 0.88 by Mann-Whitney U. Time point 2: PROTCL patients with no evidence of B-dependent antidonor responses have median ΔeGFR of −1.7 ml/min per 1.73 m2 (IQR 3.9) and mean ΔeGFR of −7.7 ml/min per 1.73 m2 (SD ±18), compared with those with evidence of B-dependent antidonor reactivity, who have a median ΔeGFR of −6.8 ml/min per 1.73 m2 (IQR 5.7) and mean ΔeGFR of −7.7 ml/min per 1.73 m2 (SD ±8.2). Actual P = 0.22 by Mann-Whitney U. Patients with BFC with no evidence of B-dependent antidonor responses have median ΔeGFR of −0.27 ml/min per 1.73 m2 (IQR 13.4) and mean ΔeGFR of −2 ml/min per 1.73 m2 (SD ±14.8), compared with those with evidence of B-dependent antidonor reactivity, who have median ΔeGFR of –9.1 ml/min per 1.73 m2 (IQR 13.6) and mean ΔeGFR of −9.6 ml/min per 1.73 m2 (SD ±13.8). Actual P = 0.1 by Mann-Whitney U. **Mann-Whitney U test.

min per 1.73 m2 (IQR 13.4) and mean ΔeGFR of −2 ml/min per 1.73 m2 (SD ±14.8), compared with those with evidence of B-dependent antidonor reactivity, who have median ΔeGFR of –9.1 ml/min per 1.73 m2 (IQR 13.6) and mean ΔeGFR of −9.6 ml/min per 1.73 m2 (SD ±13.8). Actual P = 0.1 by Mann-Whitney U. **Mann-Whitney U test. Figure S5 Changes in eGFR in CAMR subgroup who received protocolized treatment. Box plots showing median with IQR with whiskers showing upper and lower limits of the Modification of Diet in Renal Disease (MDRD) eGFR (A) and ΔeGFR (B) in the subgroup of patients with CAMR (n=15) characterized by having eGFR >20 at time of biopsy, no tubulitis on histological examination of biopsy, and identified as having an ongoing and progressive rise in creatinine, as determined by analysis of reciprocal creatinine plots at the time of first ELISPOT. All were treated with a protocolized treatment regimen, details of which are shown in Table 5. Seven patients stabilized (identified by boxes joined with coarse dotted line). Eight patients failed to stabilize (identified by boxes joined by fine dotted line). Analysis excludes 3 patients who had eGFR <20 at time of biopsy (see Table 5). The differences in the ELISPOT patterns in these 2 subgroups is described in the text. *Points at which values are statistically significant (P < 0.05) by Mann-Whitney U test.

ed to stabilize (identified by boxes joined by fine dotted line). Analysis excludes 3 patients who had eGFR <20 at time of biopsy (see Table 5). The differences in the ELISPOT patterns in these 2 subgroups is described in the text. *Points at which values are statistically significant (P < 0.05) by Mann-Whitney U test. Figure S6 Experiments to address the role of IL-10 in control of IFN-γ production. Anti–IL-10 monoclonal antibody (to inhibit IL-10) or isotype control was added into the CD8-depleted leukocyte “cone” samples (white, individual 1; gray, individual 2; black, individual 3), 2 showing consistent suppression of IFN-γ production by B cells (white, gray bars) and the third showing B-dependent IFN-γ production (black bars). Frequencies >50/million CD4+ T cells (dotted line on graph) were defined as positive. SFC, spotforming cells. The impact of the antibody after B-cell depletion suggests there is an additional source of IL-10, other than B cells, in the PBMC. Table S1a Patients excluded from analysis of outcomes because eGFR at first ELISPOT ≤20 ml/min per 1.73 m2 OR because follow-up data missing. Table S1b. Basic demographics, biopsy results, and eGFR data on the “stable” subgroup with ΔeGFR >median in either PROTCL or BFC group. Table S1c. Basic demographics, biopsy results, and eGFR data on the “deteriorating” subgroup with ΔeGFR ≤median in either PROTCL or BFC group. Table S2a Impact of reducing threshold for positive DSA to >0. Table S2b. Lack of association between DSA and outcomes in BFC cohort only. Table S2c. Lack of association between DSA and outcomes in PROTCL cohort only.

Table S1c. Basic demographics, biopsy results, and eGFR data on the “deteriorating” subgroup with ΔeGFR ≤median in either PROTCL or BFC group. Table S2a Impact of reducing threshold for positive DSA to >0. Table S2b. Lack of association between DSA and outcomes in BFC cohort only. Table S2c. Lack of association between DSA and outcomes in PROTCL cohort only. Table S3 Associations between ELISPOT pattern and outcomes in PROTCL (A) and BFC (B) subgroups. Table S4 Association between antidonor reactivity based on functional B-cell phenotype and patient outcomes in PROTCL (A) and BFC (B) cohorts. Table S5 PROTCL cohort – factors used for prediction modeling. Table S6 BFC cohort – factors used for prediction modeling. Table S7 Dynamic changes in ELISPOT patterns. Table S8 Dynamic changes in antiviral antigen ELISPOT patterns and lack of association with outcome. Table S9 Dynamic changes in antiviral ELISPOT patterns.

Table S4 Association between antidonor reactivity based on functional B-cell phenotype and patient outcomes in PROTCL (A) and BFC (B) cohorts. Table S5 PROTCL cohort – factors used for prediction modeling. Table S6 BFC cohort – factors used for prediction modeling. Table S7 Dynamic changes in ELISPOT patterns. Table S8 Dynamic changes in antiviral antigen ELISPOT patterns and lack of association with outcome. Table S9 Dynamic changes in antiviral ELISPOT patterns. Acknowledgments Thanks to the physicians, nursing staff, and the patients and donors at the Hammersmith Hospital. We acknowledge Phil Peacock and colleagues at NHS Blood and Transplant, for their assistance with obtaining transplant clinical data. Staff based at King's College London acknowledge that the research was funded/supported by the National Institute for Health Research Biomedical Research Centre based at Guy's and St Thomas' NHS Foundation Trust and King's College London. The views expressed are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research, or the Department of Health. CR and HTC are grateful for funding from the National Institute for Health Research Imperial College Biomedical Research Centre. MPH-F received funding from the Medical Research Council (grants G0801537/ID: 88245 grant and Medical Research Council Centre for Transplantation, Medical Research Council grant no. MR/J006742/1) and Guy’s and St Thomas’ Charity (grants R080530 and R090782). MPH-F and IRM received funding from the European Union, project number 305147: BIO-DrIM.

the Medical Research Council (grants G0801537/ID: 88245 grant and Medical Research Council Centre for Transplantation, Medical Research Council grant no. MR/J006742/1) and Guy’s and St Thomas’ Charity (grants R080530 and R090782). MPH-F and IRM received funding from the European Union, project number 305147: BIO-DrIM. This study was funded by the Medical Research Council (Clinical Fellowship K.Y. Shiu G84/6713; Project Awards G0401591 and G0801965); Centre Award MR/J006742/1; Roche Organ Transplant Research Foundation (ROTRF 53331024), Novartis Pharmaceuticals UK Ltd (nonpromotional grant TRA10–087), Kidney Research UK (Project Award RP3/2011), and the Wellcome Trust (Investigator award C. Kemper – 102932/Z/13/Z). Supplementary Methods. Figure S1. Lack of association between PTC score and ΔeGFR and between IF/TA and ΔeGFR over the course of the study in combined PROTCL and BFC group. (A) Box plots show median with IQR with whiskers showing upper and lower limits of ΔeGFR. Combined group includes all PROTCL and BFC patients, except 9 patients with BFC who either had missing follow-up data (n = 2) or eGFR ≤20 ml/min per 1.73 m2 at time of biopsy (n = 7). Patients with PTC score <1 have median ΔeGFR of −6.85 ml/min per 1.73 m2 (IQR 12.4) and mean ΔeGFR of −8.7 ml/min per 1.73 m2 (SD ±13.1). Patients with PTC score ≥1 have median ΔeGFR of −8.0 ml/min per 1.73 m2 (IQR 16.7) and mean ΔeGFR of −7.2 ml/min per 1.73 m2 (SD ±15.1). *Mann-Whitney U test. (B) Graph shows lack of correlation between IF/TA % on biopsy and ΔeGFR for each of the patients (n = 52) included in this analysis.

/min per 1.73 m2 (SD ±13.1). Patients with PTC score ≥1 have median ΔeGFR of −8.0 ml/min per 1.73 m2 (IQR 16.7) and mean ΔeGFR of −7.2 ml/min per 1.73 m2 (SD ±15.1). *Mann-Whitney U test. (B) Graph shows lack of correlation between IF/TA % on biopsy and ΔeGFR for each of the patients (n = 52) included in this analysis. Figure S2. Association between DSA and ΔeGFR over the course of the study in combined cohort. Box plots show median with IQR with whiskers showing upper and lower limits of ΔeGFR. Combined group includes all PROTCL and BFC patients, except 9 patients with BFC who either had missing follow-up data (n = 2), or eGFR ≤20 ml/min 1.73 m2 at time of biopsy (n = 7). Time point 1: Patients with No DSA or DSA with cumulative mean fluorescence intensity of <1000 (n = 35) have median ΔeGFR of −3.19 ml/min per 1.73 m2 (IQR 12.3) and mean ΔeGFR of −6.24 ml/min 1.73 m2 (SD ±13.4). Patients with DSA with cumulative mean fluorescence intensity >1000 (n = 17) have median ΔeGFR of −11.4 ml/min per 1.73 m2 (IQR 10.7) and mean ΔeGFR of −12.4 ml/min per 1.73 m2 (SD ±13.3). Time point 2: Patients with No DSA or DSA with cumulative mean fluorescence intensity of <1000 (n = 34) have median ΔeGFR of −3.69 ml/min per 1.73 m2 (IQR 12.4) and mean ΔeGFR of −5.68 ml/min/1.73 m2 (SD ±11.9). Patients with DSA with cumulative mean fluorescence intensity >1000 have median ΔeGFR of −9.9 ml/min per 1.73 m2 (IQR 15) and mean ΔeGFR of −13.0 ml/min per 1.73 m2 (SD ±16). **Mann-Whitney U test.

<1000 (n = 34) have median ΔeGFR of −3.69 ml/min per 1.73 m2 (IQR 12.4) and mean ΔeGFR of −5.68 ml/min/1.73 m2 (SD ±11.9). Patients with DSA with cumulative mean fluorescence intensity >1000 have median ΔeGFR of −9.9 ml/min per 1.73 m2 (IQR 15) and mean ΔeGFR of −13.0 ml/min per 1.73 m2 (SD ±16). **Mann-Whitney U test. Figure S3. Box plots show median with IQR with whiskers showing upper and lower limits of ΔeGFR in PROTCL (A) and BFC (B) subgroups. Time point 1: PROTCL patients with DSR have median ΔeGFR of −8.34 ml/min per 1.73 m2 (IQR 17.4) and mean ΔeGFR of −17.3 ml/min per 1.73 m2 (SD ±17.4), compared with those with NDSR who have a median ΔeGFR of −2.03 ml/min per 1.73 m2 (IQR 6.2) and mean ΔeGFR of −2.38 ml/min per 1.73 m2 (SD ±4.9). Patients with BFC with DSR have median ΔeGFR of −6.4 ml/min per 1.73 m2 (IQR 11.7) and mean ΔeGFR of −7.1 ml/min per 1.73 m2 (SD ±10.7), compared with those with NDSR who have median ΔeGFR of –9.9 ml/min per 1.73 m2 (IQR 20.1) and mean ΔeGFR of −10.3 ml/min per 1.73 m2 (SD ±17.1). Actual P = 0.46. Time point 2: PROTCL patients with DSR have median ΔeGFR of −4.5 ml/min per 1.73 m2 (IQR 7.2) and mean ΔeGFR of −4.6 ml/min per 1.73 m2 (SD ±5.4), compared to those with NDSR who have median ΔeGFR of −4.2 ml/min per 1.73 m2 (IQR 15.8) and mean ΔeGFR of −10.9 ml/min per 1.73 m2 (SD ±15.5). Patients with BFC with DSR have median ΔeGFR of −12.7ml/min per 1.73 m2 (IQR 13.5) and mean ΔeGFR of −13.5 ml/min per 1.73 m2 (SD ±14.6), compared with those with NDSR who have median ΔeGFR of –2.5 ml/min per 1.73 m2 (IQR 12.3) and mean ΔeGFR of −2.4 ml/min per 1.73 m2 (SD ±12.1). Actual P = 0.01. **Mann-Whitney U test.

atients with BFC with DSR have median ΔeGFR of −12.7ml/min per 1.73 m2 (IQR 13.5) and mean ΔeGFR of −13.5 ml/min per 1.73 m2 (SD ±14.6), compared with those with NDSR who have median ΔeGFR of –2.5 ml/min per 1.73 m2 (IQR 12.3) and mean ΔeGFR of −2.4 ml/min per 1.73 m2 (SD ±12.1). Actual P = 0.01. **Mann-Whitney U test. Figure S4. Box plots show median with IQR with whiskers showing upper and lower limits of ΔeGFR in PROTCL (upper panel) and BFC (lower panel) subgroups. Time point 1: PROTCL patients with no evidence of B-dependent antidonor responses have median ΔeGFR of +1.5 ml/min per 1.73 m2 (IQR 3) and mean ΔeGFR of −0.54 ml/min per 1.73 m2 (SD ±2.6), compared with those with evidence of B-dependent antidonor reactivity, who have a median ΔeGFR of −7.7 ml/min per 1.73 m2 (IQR 11.7) and mean ΔeGFR of −13.8 ml/min per 1.73 m2 (SD ±15.4). Patients with BFC with no evidence of B-dependent antidonor responses have median ΔeGFR of −11.1 ml/min per 1.73 m2 (IQR 27.6) and mean ΔeGFR of −6.9 ml/min per 1.73 m2 (SD ±19.7), compared with those with evidence of B-dependent antidonor reactivity, who have median ΔeGFR of –9.7 ml/min per 1.73 m2 (IQR 12) and mean ΔeGFR of −10.3 ml/min per 1.73 m2 (SD ±12.4). Actual P = 0.88 by Mann-Whitney U. Time point 2: PROTCL patients with no evidence of B-dependent antidonor responses have median ΔeGFR of −1.7 ml/min per 1.73 m2 (IQR 3.9) and mean ΔeGFR of −7.7 ml/min per 1.73 m2 (SD ±18), compared with those with evidence of B-dependent antidonor reactivity, who have a median ΔeGFR of −6.8 ml/min per 1.73 m2 (IQR 5.7) and mean ΔeGFR of −7.7 ml/min per 1.73 m2 (SD ±8.2). Actual P = 0.22 by Mann-Whitney U. Patients with BFC with no evidence of B-dependent antidonor responses have median ΔeGFR of −0.27 ml/min per 1.73 m2 (IQR 13.4) and mean ΔeGFR of −2 ml/min per 1.73 m2 (SD ±14.8), compared with those with evidence of B-dependent antidonor reactivity, who have median ΔeGFR of –9.1 ml/min per 1.73 m2 (IQR 13.6) and mean ΔeGFR of −9.6 ml/min per 1.73 m2 (SD ±13.8). Actual P = 0.1 by Mann-Whitney U. **Mann-Whitney U test.

min per 1.73 m2 (IQR 13.4) and mean ΔeGFR of −2 ml/min per 1.73 m2 (SD ±14.8), compared with those with evidence of B-dependent antidonor reactivity, who have median ΔeGFR of –9.1 ml/min per 1.73 m2 (IQR 13.6) and mean ΔeGFR of −9.6 ml/min per 1.73 m2 (SD ±13.8). Actual P = 0.1 by Mann-Whitney U. **Mann-Whitney U test. Figure S5. Changes in eGFR in CAMR subgroup who received protocolized treatment. Box plots showing median with IQR with whiskers showing upper and lower limits of the Modification of Diet in Renal Disease (MDRD) eGFR (A) and ΔeGFR (B) in the subgroup of patients with CAMR (n=15) characterized by having eGFR >20 at time of biopsy, no tubulitis on histological examination of biopsy, and identified as having an ongoing and progressive rise in creatinine, as determined by analysis of reciprocal creatinine plots at the time of first ELISPOT. All were treated with a protocolized treatment regimen, details of which are shown in Table 5. Seven patients stabilized (identified by boxes joined with coarse dotted line). Eight patients failed to stabilize (identified by boxes joined by fine dotted line). Analysis excludes 3 patients who had eGFR <20 at time of biopsy (see Table 5). The differences in the ELISPOT patterns in these 2 subgroups is described in the text. *Points at which values are statistically significant (P < 0.05) by Mann-Whitney U test.

ed to stabilize (identified by boxes joined by fine dotted line). Analysis excludes 3 patients who had eGFR <20 at time of biopsy (see Table 5). The differences in the ELISPOT patterns in these 2 subgroups is described in the text. *Points at which values are statistically significant (P < 0.05) by Mann-Whitney U test. Figure S6. Experiments to address the role of IL-10 in control of IFN-γ production. Anti–IL-10 monoclonal antibody (to inhibit IL-10) or isotype control was added into the CD8-depleted leukocyte “cone” samples (white, individual 1; gray, individual 2; black, individual 3), 2 showing consistent suppression of IFN-γ production by B cells (white, gray bars) and the third showing B-dependent IFN-γ production (black bars). Frequencies >50/million CD4+ T cells (dotted line on graph) were defined as positive. SFC, spotforming cells. The impact of the antibody after B-cell depletion suggests there is an additional source of IL-10, other than B cells, in the PBMC. Table S1a. Patients excluded from analysis of outcomes because eGFR at first ELISPOT ≤20 ml/min per 1.73 m2 OR because follow-up data missing. Table S1b. Basic demographics, biopsy results, and eGFR data on the “stable” subgroup with ΔeGFR >median in either PROTCL or BFC group. Table S1c. Basic demographics, biopsy results, and eGFR data on the “deteriorating” subgroup with ΔeGFR ≤median in either PROTCL or BFC group. Table S2a. Impact of reducing threshold for positive DSA to >0. Table S2b. Lack of association between DSA and outcomes in BFC cohort only. Table S2c. Lack of association between DSA and outcomes in PROTCL cohort only.

Table S1c. Basic demographics, biopsy results, and eGFR data on the “deteriorating” subgroup with ΔeGFR ≤median in either PROTCL or BFC group. Table S2a. Impact of reducing threshold for positive DSA to >0. Table S2b. Lack of association between DSA and outcomes in BFC cohort only. Table S2c. Lack of association between DSA and outcomes in PROTCL cohort only. Table S3. Associations between ELISPOT pattern and outcomes in PROTCL (A) and BFC (B) subgroups. Table S4. Association between antidonor reactivity based on functional B-cell phenotype and patient outcomes in PROTCL (A) and BFC (B) cohorts. Table S5. PROTCL cohort – factors used for prediction modeling. Table S6. BFC cohort – factors used for prediction modeling. Table S7. Dynamic changes in ELISPOT patterns. Table S8. Dynamic changes in antiviral antigen ELISPOT patterns and lack of association with outcome. Table S9. Dynamic changes in antiviral ELISPOT patterns. Supplementary material is linked to the online version of the paper at www.kidney-international.org.

Table S6. BFC cohort – factors used for prediction modeling. Table S7. Dynamic changes in ELISPOT patterns. Table S8. Dynamic changes in antiviral antigen ELISPOT patterns and lack of association with outcome. Table S9. Dynamic changes in antiviral ELISPOT patterns. Supplementary material is linked to the online version of the paper at www.kidney-international.org. Figure 1 ΔeGFR in PROTCL and BFC subgroups and combined group. Box plots show median and IQR, with whiskers representing data within 1.5 the IQR of the upper and lower quartiles, with outliers >1.5 and <3.0 IQR as + and >3.0 IQR as *. Horizontal lines to right of box plots indicate the mean value. Nine patients with BFC who either had missing follow-up data (n = 2), or eGFR <20 ml/min per 1.73 m2 at time of biopsy (n = 7) have been excluded (Supplementary Table S1c). The “combined” group includes all patients with PROTCL and BFC, and have been split into “deteriorating” and “stable” subgroups, based on the relationship to the median ΔeGFR in each of the PROTCL and BFC subgroups. *Deteriorating group contains patients with ΔeGFR below or equal to the median in each of PROTCL and BFC groups (n = 27). The median ΔeGFR in this subgroup is −14.4 ml/min per 1.73 m2 (IQR 15.5). Stable group contains patients with ΔeGFR above the median from each of the PROTCL and BFC groups (n = 25). The median ΔeGFR for this group is 0.3 ml/min per 1.73 m2 (IQR 6.0). **Mann-Whitney U test.

Figure 1 ΔeGFR in PROTCL and BFC subgroups and combined group. Box plots show median and IQR, with whiskers representing data within 1.5 the IQR of the upper and lower quartiles, with outliers >1.5 and <3.0 IQR as + and >3.0 IQR as *. Horizontal lines to right of box plots indicate the mean value. Nine patients with BFC who either had missing follow-up data (n = 2), or eGFR <20 ml/min per 1.73 m2 at time of biopsy (n = 7) have been excluded (Supplementary Table S1c). The “combined” group includes all patients with PROTCL and BFC, and have been split into “deteriorating” and “stable” subgroups, based on the relationship to the median ΔeGFR in each of the PROTCL and BFC subgroups. *Deteriorating group contains patients with ΔeGFR below or equal to the median in each of PROTCL and BFC groups (n = 27). The median ΔeGFR in this subgroup is −14.4 ml/min per 1.73 m2 (IQR 15.5). Stable group contains patients with ΔeGFR above the median from each of the PROTCL and BFC groups (n = 25). The median ΔeGFR for this group is 0.3 ml/min per 1.73 m2 (IQR 6.0). **Mann-Whitney U test. Figure 2 Association between ELISPOT patterns and ΔeGFR over the course of the study in combined PROTCL and BFC group. Box plots show median and IQR, with whiskers representing data within 1.5 of the IQR of the upper and lower quartiles, with outliers >1.5 and <3.0 IQR as + and >3.0 IQR as *. Horizontal lines to right of box plots indicate the mean value. (a) Patterns grouped according to DSR versus NDSR status. Time point 1: Patients with ELISPOT showing DSR have median ΔeGFR of −8.0 ml/min (IQR 11.6) and mean ΔeGFR of −9.6 ml/min (SD ±12.6). Patients with NDSR have median ΔeGFR of −5.8 ml/min (IQR 17.9) and mean ΔeGFR of −8.0 ml/min (SD ±14.4). P = 0.70 Mann-Whitney U. Time point 2: Patients with ELISPOT showing DSR have median ΔeGFR of −10.1 ml/min (IQR 15.5) and mean ΔeGFR of −11.3 (±SD 10.9) ml/min. Patients with NDSR have median ΔeGFR of −3.1 ml/min (IQR 11.9) and mean ΔeGFR of −5.1 ml/min (SD ±13.4). P = 0.05 Mann-Whitney U. NB: Analysis with 2 outliers at time point 2 removed (ΔeGFR −33.7 [ID 392] and 44.2 [ID 958] both in NDSR group) and replaced with missing data reveal P = 0.015. (b) Patterns group according to evidence on ELISPOT of B-cell–dependent antidonor reactivity. Time point 1: Patients with ELISPOTs showing evidence of B-dependent antidonor IFN-γ production have median ΔeGFR of −8.3 ml/min per 1.73 m2 (IQR 15.2) and mean ΔeGFR of −11.5 (SD ±15.0 ml/min per 1.73 m2). Patients with ELISPOT showing no evidence of B-dependent IFN-γ production have median ΔeGFR of −0.9 ml/min (IQR 17.9) and mean ΔeGFR of −4.0 (SD ±15.6) ml/min. P > 0.11 Mann-Whitney U. Time point 2: Patients with ELISPOTs showing evidence of B-dependent IFN-γ production have median ΔeGFR of −7.9 ml/min per 1.73m2 (IQR 11.7) and mean ΔeGFR of −9.6 (SD ±10.9 ml/min per 1.73 m2). Patients with ELISPOT showing no evidence of B-dependent IFN-γ production have median ΔeGFR of −0.9 ml/min (IQR 11.6) and mean ΔeGFR of −4.1 (SD ±15.4) ml/min. P = 0.053 Mann-Whitney U.

ependent IFN-γ production have median ΔeGFR of −7.9 ml/min per 1.73m2 (IQR 11.7) and mean ΔeGFR of −9.6 (SD ±10.9 ml/min per 1.73 m2). Patients with ELISPOT showing no evidence of B-dependent IFN-γ production have median ΔeGFR of −0.9 ml/min (IQR 11.6) and mean ΔeGFR of −4.1 (SD ±15.4) ml/min. P = 0.053 Mann-Whitney U. NB: Analysis with 3 outliers at time point 2 removed (ΔeGFR −33.7 [ID 392] and −44.2 [ID 958] both in “No evidence of B-dependency” group, and ΔeGFR −35.3 [ID 635] in “Evidence of B-dependency” group) and replaced with missing data reveal P = 0.01. Figure 3 Multivariate logistic regression models in patient subgroups. ROC curves corresponding to the multivariate logistic regression models for linked groups of predictive variable in the PROTCL biopsy (a), BFC (b), and the optimized treatment BFC-CAMR subgroup with deteriorating creatinines (c), using generalized linear models to estimate each of the models, followed by elastic net estimate the optimal combined algorithm, with cross validation for parameter tuning. The predictive variables included in each of the models are listed in Supplementary Tables S5 and S6.

subgroup with deteriorating creatinines (c), using generalized linear models to estimate each of the models, followed by elastic net estimate the optimal combined algorithm, with cross validation for parameter tuning. The predictive variables included in each of the models are listed in Supplementary Tables S5 and S6. Figure 4 Associations between patterns on ELISPOT and changes in eGFR in BFC cohort. Box plots showing the association between the results of ELISPOT assays at time of biopsy and follow-up sample with graft outcome in patients who had viable PBMC samples after thawing at both times (n = 27). (a) Patients with BFC with NDSR at time of biopsy (n = 16), showing stability for those who remained NDSR (n = 11) compared with those who became DSR (n = 5, P = 0.003) and (b) patients with BFC with DSR at time of biopsy, showing stable eGFR for those patients who were DSR at time of biopsy (n = 11) but converted to NDSR (n = 8), compared with progressive decline among those who remained DSR (n = 3, P = 0.0001). (c,d) Box plots show median and IQR, with whiskers representing data within 1.5 the IQR of the upper and lower quartiles, with outliers >1.5 and <3.0 IQR as + and >3 IQR as *. Horizontal lines to right of box plots indicate the mean value. Graphs shows the association between the changes in ELISPOT assays from time point 1 to time point 2 with graft outcome in patients who had 2 viable PBMC samples that could be fully interpreted (i.e., had results from CD8-, CD19-, CD25-, and CD8–CD25-depleted PBMC) (n = 37). (c) Antidonor responses. Groups correspond to those shown in Table 6 and Supplementary Table S8. Patients at time point 1 with no evidence of B-dependent antidonor responses who maintained evidence of regulated responses at time point 2 had a median ΔeGFR of 1.8 ml/min per 1.73 m2 (IQR 6.6) and mean ΔeGFR of 1.2ml/min per 1.73 m2 (SD ±12.3). Patients with evidence of B-dependent antidonor responses at time point 1 who maintained evidence of regulated responses at time point 2 had a median ΔeGFR of −5.5 ml/min per 1.73 m2 (IQR 12.9) and mean ΔeGFR of −8.6 ml/min per 1.73 m2 (SD ±13.7). Finally, patients who had unregulated B-cell–dependent antidonor responses at time point 2 had a median ΔeGFR of −10.1 ml/min per 1.73 m2 (IQR 13.7) and mean ΔeGFR of −14 ml/min per 1.73 m2 (SD ±12) irrespective of the pattern they had at time point 1. P = 0.036. (d) Antiviral responses. Groups correspond to those shown in Supplementary Tables S9 and S10.

B-cell–dependent antidonor responses at time point 2 had a median ΔeGFR of −10.1 ml/min per 1.73 m2 (IQR 13.7) and mean ΔeGFR of −14 ml/min per 1.73 m2 (SD ±12) irrespective of the pattern they had at time point 1. P = 0.036. (d) Antiviral responses. Groups correspond to those shown in Supplementary Tables S9 and S10. Groups compared by Kruskal-Wallis test. MDRD, Modification of Diet in Renal Disease.

B-cell–dependent antidonor responses at time point 2 had a median ΔeGFR of −10.1 ml/min per 1.73 m2 (IQR 13.7) and mean ΔeGFR of −14 ml/min per 1.73 m2 (SD ±12) irrespective of the pattern they had at time point 1. P = 0.036. (d) Antiviral responses. Groups correspond to those shown in Supplementary Tables S9 and S10. Groups compared by Kruskal-Wallis test. MDRD, Modification of Diet in Renal Disease. Figure 5 Flow cytometric analysis of Th-1 cytokine production. (a,b) Donor antigen-specific IFN-γ production by CD4+ T cells: comparison of subgroups according to functional B-cell phenotype on ELISPOT. CD8-depleted PBMCs were stimulated with donor antigen under same conditions as in ELISPOT, then assayed by flow cytometry by using a cytokine capture system. White bars: Samples (n = 8) from patients with ELISPOT pattern showing evidence of B-dependent antidonor IFN-γ production (with or without evidence of regulation). Black bars: Samples (n = 3) from patients with ELISPOT pattern showing only suppression of antidonor IFN-γ production by B cells with NO evidence of B-dependent responses. (a) Shows the percentage of CD4+ cells expressing only IFN-γ (IFN-γ + IL-10−) or coexpressing with IL-10 (IFN-γ + IL-10+). (b) Shows the comparison of the percentage of total cells expressing IFN-γ/% total cells expressing IL-10. (c–e) Polyclonal stimulation with anti-CD3/anti-CD46 monoclonal antibodies: comparison of subgroups according to functional B-cell phenotype on ELISPOT. White bars: Samples (n = 4) in which antidonor-specific ELISPOT showed only suppression of antidonor IFN-γ production by B cells with NO evidence of B-dependent responses. Black bars: Samples (n = 8) in which antidonor-specific ELISPOT showed evidence of a regulated B-dependent antidonor response. Gray bars: Samples (n = 4) in which antidonor-specific ELISPOT showed evidence of an unregulated B-dependent antidonor response. (c) Percentage of CD4+ cells staining for IFN-γ alone (IFN-γ + IL-10−) compared with cells staining for both (IFNγ + IL-10+). (d) Median fluorescence intensity of staining for IFN-γ or IL-10 in the single-positive (IFNγ + IL-10−) or double-positive (IFNγ + IL-10+) CD4+ populations as indicated. (e) Ratio of mean fluorescence intensity of IFN-γ staining to IL-10 staining in the double-positive (IFN-γ + IL-10+) population in (d). *P < 0.05 by Mann-Whitney U test.

ence intensity of staining for IFN-γ or IL-10 in the single-positive (IFNγ + IL-10−) or double-positive (IFNγ + IL-10+) CD4+ populations as indicated. (e) Ratio of mean fluorescence intensity of IFN-γ staining to IL-10 staining in the double-positive (IFN-γ + IL-10+) population in (d). *P < 0.05 by Mann-Whitney U test. Table 1 Associations between clinical and biopsy features, DSA and patient outcomes

ence intensity of staining for IFN-γ or IL-10 in the single-positive (IFNγ + IL-10−) or double-positive (IFNγ + IL-10+) CD4+ populations as indicated. (e) Ratio of mean fluorescence intensity of IFN-γ staining to IL-10 staining in the double-positive (IFN-γ + IL-10+) population in (d). *P < 0.05 by Mann-Whitney U test. Table 1 Associations between clinical and biopsy features, DSA and patient outcomes Clinical variable Result in whole cohort Number of biopsies/samples Pa Number of biopsies/samples Pa Graft failure No graft failure Deteriorating eGFR (≤median) Stable eGFR (>median) PCR >50 at time of biopsy Yes (n = 25) 10 15 0.002 17 8 0.03 No (n = 27) 1 26 10 17 Biopsy subgroup BFC (n = 37) 11 26 0.02 19 18 1 PROTCL (n = 15) 0 15 8 7 Gross biopsy features AMR (n = 45) 11 34 0.3 25 20 0.24 Control (n = 7) 0 7 2 5 Tubulitis Positive (n = 4) 2 2 0.2 3 1 0.6 Negative (n = 48) 9 39 24 24 C4d (PTC) Positive (n = 26) 8 18 0.1 16 10 0.27 Negative (n = 26) 3 23 11 15 C4d (g) Positive (n = 30) 9 21 0.09 17 13 0.58 Negative (n = 22) 2 20 10 12 G score ≥1 (n = 21) 6 15 0.3 14 7 0.1 0 (n = 31) 5 26 13 18 PTC score ≥1 (n = 15) 7 8 0.008 8 7 1 0 (n = 37) 4 33 19 18 CG score ≥1 (n = 20) 7 13 0.08 10 10 1 0 (n = 32) 4 28 17 15 CV score ≥1 (n = 22) 6 16 0.5 13 9 0.57 0 (n = 29) 5 24 14 15 % Median IF/TA score 30 15 <0.05b 20 15 >0.05 DSA time point 1 >1000 (n = 18) 7 11 0.03 12 6 0.15 0 or <1000 (n = 34) 4 30 15 19 DSA time point 2 >1000 (n = 18) 7 11 0.04 11 7 0.39 0 or <1000 (n = 32) 4 28 15 17 DSA overallc >1000 (n = 20) 7 13 0.08 12 8 0.4 0 or <1000 (n = 32) 4 28 15 17 Proteinuria: Graft failure – Sensitivity: 10/11 = 91%; PPV: 10/25 = 40%; NPV: 26/27 = 96%; Specificity: 26/41 = 63%.

3 12 6 0.15 0 or <1000 (n = 34) 4 30 15 19 DSA time point 2 >1000 (n = 18) 7 11 0.04 11 7 0.39 0 or <1000 (n = 32) 4 28 15 17 DSA overallc >1000 (n = 20) 7 13 0.08 12 8 0.4 0 or <1000 (n = 32) 4 28 15 17 Proteinuria: Graft failure – Sensitivity: 10/11 = 91%; PPV: 10/25 = 40%; NPV: 26/27 = 96%; Specificity: 26/41 = 63%. Proteinuria: Deteriorating function – Sensitivity: 17/27 = 63%; PPV 17/25 = 68%; NPV 17/27 = 63%; Specificity 68%. PTC score: Graft failure – Sensitivity: 7/11 = 64%; PPV: 7/15 = 67%; NPV: 33/37 = 89%; Specificity: 33/41 = 80%. DSA >1000 time point 1: Graft failure – Sensitivity: 7/11 = 64%; PPV: 7/18 = 39%; NPV: 30/34 = 88%; Specificity: 30/41 = 77%. DSA >1000 time point 2: Graft failure – Sensitivity: 7/11 = 64%; PPV: 7/18 = 39%: NPV: 28/32 = 88%; Specificity: 28/39 = 72%. Bold P values are statistically significant. AMR, antibody-mediated rejection; BFC, for-cause biopsy; DSA, donor-specific antibody; eGFR, estimated glomerular filtration rate; NPV, negative predictive value; PCR, protein-to-creatinine ratio; PPV, positive predictive value; PROTCL, protocol; PTC, peritubular capillary; g, glomerulitis; cg/cv, BANFF chronic glomerulopathy and vascular scores; IF/TA, interstitial fibrosis/tubular atrophy. a Fisher exact test. b Mann-Whitney U test. c DSA at either time point 1 or 2 or both. Table 2 ELISPOT patterns, classified by response to donor antigens when CD8+ cells depleted as DSR or NDSR and, for the latter by the response after depletion of CD25+ cells (“Treg”) or CD19+ cells (Breg)

AMR, antibody-mediated rejection; BFC, for-cause biopsy; DSA, donor-specific antibody; eGFR, estimated glomerular filtration rate; NPV, negative predictive value; PCR, protein-to-creatinine ratio; PPV, positive predictive value; PROTCL, protocol; PTC, peritubular capillary; g, glomerulitis; cg/cv, BANFF chronic glomerulopathy and vascular scores; IF/TA, interstitial fibrosis/tubular atrophy. a Fisher exact test. b Mann-Whitney U test. c DSA at either time point 1 or 2 or both. Table 2 ELISPOT patterns, classified by response to donor antigens when CD8+ cells depleted as DSR or NDSR and, for the latter by the response after depletion of CD25+ cells (“Treg”) or CD19+ cells (Breg) ELISPOT patterns Interpretation based on reactivity of CD8-depleted PBMC (DSR) versus nonreactivity (NDSR)a Interpretation based on B-cell phenotype CD25 present CD25 depleted CD8-deplete CD8- and CD19-deplete CD8-deplete CD8 and CD19-deplete − − − − NDSR No regulation No regulation No response − +b − − Breg Breg: only when CD25+ cells present Regulated antidonor response without evidence of B-dependency − + − + Breg: when CD25+ cells present or absent − − − + Treg, Breg Breg: only when CD25+ cells absent − + + − Breg when CD25 present BUT Bdep when CD25 absent B-dependent antidonor response with evidence of regulation

− +b − − Breg Breg: only when CD25+ cells present Regulated antidonor response without evidence of B-dependency − + − + Breg: when CD25+ cells present or absent − − − + Treg, Breg Breg: only when CD25+ cells absent − + + − Breg when CD25 present BUT Bdep when CD25 absent B-dependent antidonor response with evidence of regulation − − + − Treg Bdep: only when CD25+ cells absent + − − + DSR Bdep Bdep when CD25 present, Breg when CD25+ cells absent + − + − Bdep: when CD25+ cells present and absent Unregulated B-dependent antidonor response + − − − Bdep: only when CD25+ cells present An alternative way to interpret these patterns is by the functional B-cell phenotype in the presence or of CD25+ cells. Note that some samples defined in Shiu et al.7 as DSR Bdep showed evidence of Breg activity in absence of CD25+ cells. Bdep, decrease in spot count of ≥20% when CD19+ cells depleted; Breg, increase in spot count of ≥20% when CD19+ cells depleted; DSR, donor-specific reactivity; ELISPOT, enzyme-linked immunosorbent spot; NDSR, no donor-specific reactivity; PBMC, peripheral blood mononuclear cell; Treg, increase in spot count of ≥20% when CD25+ cells depleted. a As used in Shiu et al.7 b Threshold for defining positive antidonor interferon-γ production was 25 or more spots per million CD4+ cells on donor antigen plate compared with background. Therefore, “+” = spot count ≥25; “−” =spot count <25. Table 3 Number of samples ELISPOT patterns interpreted by B-cell phenotype in PROTCL and BFC by time

a As used in Shiu et al.7 b Threshold for defining positive antidonor interferon-γ production was 25 or more spots per million CD4+ cells on donor antigen plate compared with background. Therefore, “+” = spot count ≥25; “−” =spot count <25. Table 3 Number of samples ELISPOT patterns interpreted by B-cell phenotype in PROTCL and BFC by time Interpretation based on B-cell phenotype Number of ELISPOTs showing the defined pattern PROTCL BFC Total Time point 1 Time point 2 Time point 1 Time point 2 Time point 1 Time point 2 No evidence of B-dependent antidonor response No response No regulation 3 1 5 11 8 12 Evidence of regulation Regulated antidonor response without evidence of B-dependency Breg: only when CD25+ cells present 2 2a 0 1 5 7 Breg: when CD25+ cells present and absent 0 2 2 0 Breg: only when CD25+ cells absent 0 1 1 1 Evidence of B-dependent antidonor response B-dependent antidonor response with evidence of regulation Breg when CD25 present BUT Bdep when CD25 absent, 1 0 5b 3c 12 13 Bdep: only when CD25+ cells absent 1 3 4 3 Bdep when CD25 present, Breg when CD25+ cells absent 0 1 1 3 No evidence of regulation Unregulated B-dependent antidonor response Bdep: when CD25+ cells present and absent 3 0 7 3 13 11 Bdep: only when CD25+ cells present 2 4 1 4 Not done / Not viable / Not interpretabled 2 ND 1 NDSR 1 NDSR 5 ND 3 NDSR 3 DSR 5 ND 3 NDSR 14 9 Several other viable samples at later time points were collected and analyzed and included in Shiu et al.7 but are not considered further here.

ls present and absent 3 0 7 3 13 11 Bdep: only when CD25+ cells present 2 4 1 4 Not done / Not viable / Not interpretabled 2 ND 1 NDSR 1 NDSR 5 ND 3 NDSR 3 DSR 5 ND 3 NDSR 14 9 Several other viable samples at later time points were collected and analyzed and included in Shiu et al.7 but are not considered further here. Bdep, decrease in spot count of ≥20% when CD19+ cells depleted; BFC, for-cause biopsy; DSR, donor-specific reactivity; Breg, increase in spot count of ≥20% when CD19+ cells depleted; ELISPOT, enzyme-linked immunosorbent spot; IFN-γ, interferon-γ; NDSR, no donor-specific reactivity; PBMC, peripheral blood mononuclear cells; PROTCL, protocol. a One of these samples was DSR (i.e., IFN-γ produced by CD8-depleted PBMCs but spot count increased by >20% with B-cell depletion. Therefore, there were 7 DSR and 8 NDSR in PROTCL at time point 2. b Three of these samples were DSR (i.e., IFN-γ produced by CD8-depleted PBMCs but spot count increased by >20% with B-cell depletion. All 3 also showed increases >20% with depletion of CD25+ cells followed by reduction (>20%) in spot count when CD19+ cells additionally depleted. Therefore, there were 14 DSR and 18 NDSR in BFC at time point 1. c One of these samples was DSR (i.e., IFN-γ produced by CD8-depleted PBMCs but spot count increased by >20% with B-cell depletion. Both also showed increases >20% with depletion of CD25+ cells followed by reduction (>20%) in spot count when CD19+ cells additionally depleted. Therefore, there were 11 DSR and 21 NDSR in BFC at time point 2.

es was DSR (i.e., IFN-γ produced by CD8-depleted PBMCs but spot count increased by >20% with B-cell depletion. Both also showed increases >20% with depletion of CD25+ cells followed by reduction (>20%) in spot count when CD19+ cells additionally depleted. Therefore, there were 11 DSR and 21 NDSR in BFC at time point 2. d In some samples, it was not possible to perform all 4 depletion combinations to enable interpretation based on B-cell phenotype, but classification as DSR/NDSR was possible. Table 4 Association between antidonor reactivity (DSR/NDSR) and patient outcomes in whole cohort ELISPOT variable ELISPOT pattern Number of samples Pa Number of samples Pa Graft failure No graft failure Deteriorating eGFR (≤median) Stable eGFR (>median) Time point 1 DSR (n = 20) 3 17 0.47 11 9 0.80 NDSR (n = 25) 7 18 12 13 Time point 2 DSR (n = 17) 5 12 0.23 11 6 0.08 NDSR (n = 30) 3 27 11 19 Change to or maintenance ofb: DSR (n = 15) 5 10 0.08 10 5 0.05 NDSR (n = 25) 2 23 8 17 Bold P values are those that are statistically significant. DSR, donor-specific reactivity; eGFR, estimated glomerular filtration rate; ELISPOT, enzyme-linked immunosorbent spot; NDSR, no donor-specific reactivity. a Fisher exact test. b In paired samples only (i.e., samples in which ELISPOTS available at both time points); in relation to graft failure, only 7 of the 11 with graft failure had ELISPOTS at both time points. Table 5 Association between antidonor reactivity based on functional B-cell phenotype and patient outcomes in whole cohort

a Fisher exact test. b In paired samples only (i.e., samples in which ELISPOTS available at both time points); in relation to graft failure, only 7 of the 11 with graft failure had ELISPOTS at both time points. Table 5 Association between antidonor reactivity based on functional B-cell phenotype and patient outcomes in whole cohort ELISPOT variable ELISPOT pattern Number of samples Pa Number of samples Pa Graft failure No graft failure Deteriorating eGFR (≤median) Stable eGFR (>median) Time point 1 No evidence of B-dependence (n = 13) 2 11 0.69 4 9 0.09 Evidence of B-dependence (n = 25) 6 19 16 9 Time point 2 No evidence of B-dependence (n = 19) 1 18 0.11 5 14 0.03 Evidence of B-dependence (n = 24) 6 18 15 9 Change to or maintenance of:b No evidence of B-dependence (n = 16) 1 15 0.20 4 12 0.04 Evidence of B-dependence (n = 21) 5 16 13 8 Bold P values are those that are statistically significant. eGFR, estimated glomerular filtration rate; ELISPOT, enzyme-linked immunosorbent spot. a Fisher exact test. b In paired samples only (i.e., samples in which ELISPOTs available at both time points). Table 6 Dynamic changes in antidonor ELISPOT patterns and association with outcome

ELISPOT variable ELISPOT pattern Number of samples Pa Number of samples Pa Graft failure No graft failure Deteriorating eGFR (≤median) Stable eGFR (>median) Time point 1 No evidence of B-dependence (n = 13) 2 11 0.69 4 9 0.09 Evidence of B-dependence (n = 25) 6 19 16 9 Time point 2 No evidence of B-dependence (n = 19) 1 18 0.11 5 14 0.03 Evidence of B-dependence (n = 24) 6 18 15 9 Change to or maintenance of:b No evidence of B-dependence (n = 16) 1 15 0.20 4 12 0.04 Evidence of B-dependence (n = 21) 5 16 13 8 Bold P values are those that are statistically significant. eGFR, estimated glomerular filtration rate; ELISPOT, enzyme-linked immunosorbent spot. a Fisher exact test. b In paired samples only (i.e., samples in which ELISPOTs available at both time points). Table 6 Dynamic changes in antidonor ELISPOT patterns and association with outcome Interpretation based on B-cell phenotype Time point 2a No response Evidence of regulation No evidence of regulation Regulated antidonor response without evidence of B-dependency B-dependent antidonor response with evidence of regulation Unregulated B-dependent antidonor response Time point 1b No evidence of B-dependent antidonor response No response 10 patients: 8 stable, 2 deteriorating ΔeGFRc 1.79 (IQR 6.63) 9 patients: 2 stable, 7 deteriorating ΔeGFR −10.1 (IQR 13.7)d Regulated antidonor response without evidence of B-dependency

dence of regulation Unregulated B-dependent antidonor response Time point 1b No evidence of B-dependent antidonor response No response 10 patients: 8 stable, 2 deteriorating ΔeGFRc 1.79 (IQR 6.63) 9 patients: 2 stable, 7 deteriorating ΔeGFR −10.1 (IQR 13.7)d Regulated antidonor response without evidence of B-dependency Evidence of B-dependent antidonor response B-dependent antidonor response with evidence of regulation 18 patients: 10 stable, 8 deteriorating ΔeGFR −5.54 (IQR 12.9) Unregulated B-dependent antidonor response Refer to Supplementary Table S7 for full details of all patients. eGFR, estimated glomerular filtration rate; ELISPOT, enzyme-linked immunosorbent spot; IQR, interquartile range. a Six additional patients with time point 1 samples had time point 2 samples that were either not done or not fully interpretable, so they are not included in this analysis; 3 patients had neither time point 1 or 2 samples that could be interpreted by B-cell phenotype so they are not included here. b Six additional patients with time point 2 samples had time point 1 samples that were either not done or not fully interpretable, so they are not included in this analysis. c eGFR in ml/min per 1.73m2. d Comparison of stable and deteriorating patients in each group: P = 0.047 Fisher Exact Probability 3 x 2 test. Table 7 Summary details of demographics, biopsy, and immunosuppressive treatment of optimized CAMR patients: details of the 18 patients treated with optimization (Tac/MMF ± rituximab) for deteriorating creatinine

d Comparison of stable and deteriorating patients in each group: P = 0.047 Fisher Exact Probability 3 x 2 test. Table 7 Summary details of demographics, biopsy, and immunosuppressive treatment of optimized CAMR patients: details of the 18 patients treated with optimization (Tac/MMF ± rituximab) for deteriorating creatinine Patient ID PCR >50 Changes in treatment postbiopsy B phenotype in ELISPOT T1 B phenotype in ELISPOT T2 Renal outcome at 3 yr Adverse events 165 No CsA to Tac switch Regulated Bdep Breg Stable GFR, no proteinuria 0 326 Yes MMF and rituximab Bdep - no reg Regulated Bdep Stable GFR, ongoing proteinuria 0 392 Yes CsA to Tac switch Regulated Bdep NR Graft loss, 2 yr after biopsy Staph sepsis/joint infection, 7 mo after switch to Tac 397 Yes MMF Bdep - no reg Regulated Bdep Stable GFR, proteinuria resolved 0 399 No Rituximab NR NR Continued deterioration, no proteinuria 0 438 Yes MMF Bdep - no reg Regulated Bdep Stable GFR, continued proteinuria 0 635 Yes Rituximab Breg Bdep - no reg Graft loss, 22 mo after biopsy 0 739 No CsA to Tac switch NR NR Stable GFR, no proteinuria Recurrent UTI. No serious infections. 807 (<20) Yes CsA to Tac switch - - Graft loss at 15 mo after biopsy 0 835 Yes CsA to Tac, Aza to MMF switch Breg Bdep - no reg Graft loss at 12 mo postbiopsy. Rituximab (9 mo postbiopsy) 0 841 No CsA to Tac switch Bdep - no reg NR Stable GFR, no proteinuria 0 861 Yes CsA to Tac, Aza to MMF switch, rituximab Nonviable Bdep - no reg Continued deterioration, no proteinuria Nausea and vomiting 1 mo after rituximab. No cause found. Settled spontaneously. 965 No Optimized Tac, MMF levels Nonviable NR Stable GFR, no proteinuria 0 1364 Yes Rituximab NR NR Continued deterioration, continued proteinuria Aspergillus and Stenotrophomonas lung infection 2 wk after first dose of rituximab. Not given second dose. 1404 Yes MMF. Steroids (3 mo postbiopsy) Rituximab (8 mo postbiopsy) Regulated Bdep Nonviable Graft loss at 22 mo postbiopsy Pseudomonas and Klebsiella soft tissue infection (orbital cellulitis) 13 mo postbiopsy (5 mo after rituximab). 2002 Yes Rituximab Bdep - no reg Not done Graft loss at 7 mo postbiopsy/rituximab 0 2006 (<20) Yes Tac, MMF, rituximab - - Graft loss at 9 mo postbiopsy 0 2010 (<20) Yes CsA to Tac switch - - Graft loss at 16 mo postbiopsy 0 Rows highlighted in gray are the patients who stabilized their renal function after treatment. Refer to Supplementary Figure S6 for the plots of Modification of Diet in Renal Disease and change in estimated GFR.

ab - - Graft loss at 9 mo postbiopsy 0 2010 (<20) Yes CsA to Tac switch - - Graft loss at 16 mo postbiopsy 0 Rows highlighted in gray are the patients who stabilized their renal function after treatment. Refer to Supplementary Figure S6 for the plots of Modification of Diet in Renal Disease and change in estimated GFR. Deteriorating GFR at time of biopsy confirmed by analysis of 1/creatinine plot. Stability and continued deterioration after 3 years were confirmed also on analysis of 1/creatinine plots. Bdep - no reg, unregulated B-dependent antidonor response; Breg, regulated antidonor response without evidence of B-dependency; CAMR, chronic antibody-mediated rejection; ELISPOT, enzyme-linked immunosorbent spot; GFR, glomerular filtration rate; MMF, mycophenolate mofetil; NR, nonresponsive; PCR, protein-to-creatinine ratio; regulated Bdep, B-dependent antidonor response with evidence of regulation; T1, time point 1; T2, time point 2; CsA, ciclosporin A; Tac, tacrolimus; Aza, azathioprine; UTI, urinary tract infection. ELISPOT responses for patients 807, 2006, 2010 are not provided at T1 or T2 as they had eGFR <20 and were therefore excluded from analysis.

ts of the top 5 biomarkers in patients with culture-negative peritonitis and with infectious (other) episodes of peritonitis, as assessed by Mann-Whitney tests (***P < 0.001). (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IL, interleukin; MMP, matrix metalloproteinase. Different types of Gram-positive bacteria induce distinct immune responses that allow discrimination between organism subgroups We next sought to define immune fingerprints in PD patients with confirmed infections caused by Gram-positive bacteria (Supplementary Table S4A). Here, feature elimination models were able to discriminate between Gram-positive infections and other episodes, yet required combinations of ≥30 biomarkers for optimal performance and failed to reach satisfactory AUCs (Supplementary Figure S1). The best 5 biomarkers together only reached an AUC of 0.711 in the RF model, and none of the individual markers on their own—IL-17A, IL-12p40, interferon-γ, IL-1β, and total cell count—exceeded an AUC of 0.72 or stayed well below that value (Supplementary Table S4B).

The immune system is an intricate network of specialized cell types and molecular structures evolved to sense and target invading pathogens, control and clear the infection, and repair and restore the integrity of affected tissues and organs. The human body is constantly exposed to a plethora of pathogenic, opportunistic, commensal, and environmental microorganisms and has developed mechanisms to discriminate between harmful and harmless colonization through receptors and pathways that specifically recognize pathogen and danger-associated molecular patterns and unique antigenic epitopes.1, 2, 3, 4 However, unequivocal evidence that the human immune system distinguishes between different types of organisms in a physiologic context and mounts appropriate responses that are distinct enough to be exploited as rapid diagnostic indicators driving appropriate therapy is lacking.5, 6, 7, 8, 9, 10

nic epitopes.1, 2, 3, 4 However, unequivocal evidence that the human immune system distinguishes between different types of organisms in a physiologic context and mounts appropriate responses that are distinct enough to be exploited as rapid diagnostic indicators driving appropriate therapy is lacking.5, 6, 7, 8, 9, 10 Individuals with end-stage kidney disease receiving peritoneal dialysis (PD) serve as well-defined exemplar of a clinical infection requiring immediate medical intervention. Peritonitis is a common complication of PD and remains a major cause of early dropout, technical failure, and mortality.11, 12 In addition to its clinical relevance for individuals with end-stage kidney failure who depend on dialysis as life-saving renal replacement therapy, PD offers unparalleled insights into complex local cell interactions and molecular mechanisms that underpin the clinical severity of infectious episodes and that are readily translatable to improve patient management and outcomes.13, 14, 15 Importantly, peritoneal effluent can be sampled repeatedly and noninvasively, thus providing early and continuous access to the site of infection, even before antibiotic treatment is initiated. Moreover, PD-related peritonitis is caused by a wide spectrum of bacterial species, thereby allowing the study of acute responses to defined groups of organisms under closely related conditions.6, 15 However, although highly elevated white cell counts with a proportion of >50% granulocytes in the peritoneal effluent are used as indicators of peritonitis, only little progress has been made with regard to reliable discrimination between infection and noninfectious inflammation. Culture-based diagnosis of infection is slow and unsatisfactory, and rapid identification of disease-causing organisms using molecular techniques with sufficient sensitivity and specificity remains a challenge.11, 12, 16 Treatment of peritonitis therefore continues to be largely empirical, and early but untargeted treatment with broad-spectrum antibiotics and antifungals is recommended.12, 17

identification of disease-causing organisms using molecular techniques with sufficient sensitivity and specificity remains a challenge.11, 12, 16 Treatment of peritonitis therefore continues to be largely empirical, and early but untargeted treatment with broad-spectrum antibiotics and antifungals is recommended.12, 17 As alternative to organism-based diagnostics, we aimed at exploiting the human host response and used a systematic approach based on machine learning algorithms to identify diagnostically relevant, pathogen-specific local immune fingerprints in PD patients who presented with acute peritonitis. The introduction of “big data” technologies in biomedical sciences to address the complexity of the molecular and cellular mechanisms underlying disease has brought about an increasing need for advanced statistical models, machine learning, and pattern recognition techniques. In particular, wrapped feature selection methods have proved highly efficient for finding the best feature combination compared with time-consuming exhaustive searches.18 Support Vector Machines (SVMs) are data-driven methods that try to find a separating hyperplane with the maximal “margin” for classification problems and that can also be used for regression or density estimation.19, 20, 21 Artificial neural networks (ANNs) are inspired by biological neural networks with data processing from the input through a network of multiple nodes that are connected with each other in different layers.22, 23, 24 Random Forests (RFs) are ensemble methods constructed on multiple decision trees for classification and regression.25, 26, 27 By combining biomarker measurements during acute peritonitis and feature selection approaches based on SVMs, ANNs, and RFs, our findings demonstrate the power of advanced mathematical models to analyze complex biomedical datasets and highlight key pathways involved in pathogen-specific inflammatory responses at the site of infection. The observation that different infecting bacteria induce consistent and unique local immune responses has immediate diagnostic implications at the point of care by directing appropriate antibiotic treatment before conventional microbiological culture results become available.

rcellularity (denoted E). (b) Serum FHR-5 levels in IgAN patients with diagnostic renal biopsy MEST scores of 1 or less (gray box) or at least 4 (white box). The bar represents the median value, box represents the interquartile range, and whiskers represent the range of values. P values derived using Mann-Whitney test. Discussion The mechanism underlying the robust association between the delCFHR3-1 allele and reduced susceptibility to IgAN is unclear. While the role of fH as a critical negative regulator of C3 activation through the complement alternative pathway is well established, the biological roles of the FHR proteins have, until recently, been poorly understood.17, 18, 19 In fact, the common occurrence of the complete absence of FHR-1 and FHR-3 in healthy individuals with the delCFHR3-1 allele in homozygosity has suggested that these 2 FHR proteins are biologically redundant. Insights into key roles of FHR-1 and FHR-3 in diseases have been derived from the association between delCFHR3-1 alleles and protection from not only IgAN but also age-related macular degeneration.25 Furthermore, abnormal FHR proteins are associated with familial cases of C3G, a condition wherein complement-mediated renal injury is derived from abnormal regulation of alternative pathway activation.20, 26, 29, 30, 31 In vitro data have shown that FHR-117 and FHR-519 compete with fH for binding to activated C3 (termed C3b). Unlike fH, interaction of C3b with either FHR-1 or FHR-5 allows continued complement activation, preventing the inhibitory actions of fH, a phenomenon referred to as fH deregulation. We considered that the association between delCFHR3-1 alleles and IgAN can be explained by fH deregulation. We hypothesized that reduced (or absent) FHR-1 levels result in a reduction or absence of fH deregulation and consequently less complement-mediated renal injury.

mmatory responses at the site of infection. The observation that different infecting bacteria induce consistent and unique local immune responses has immediate diagnostic implications at the point of care by directing appropriate antibiotic treatment before conventional microbiological culture results become available. Results Local immune biomarkers form distinct hierarchical clusters In order to define combinations of local biomarkers that would constitute relevant disease-specific immune fingerprints, we measured a broad range of cellular and soluble biomarkers in 83 PD patients presenting with microbiologically well-defined episodes of acute peritonitis (Table 1). To cover the breadth and the complexity of local inflammatory and regulatory immune responses during early infection, these biomarkers included frequencies and total numbers of infiltrating leukocytes as well as levels of common cytokines, chemokines, and tissue damage–associated molecules, the majority of which were elevated during acute peritonitis compared with baseline parameters in stable individuals (Supplementary Table S1). Perhaps not surprisingly, due to the redundant roles of many inflammatory mediators within the human immune system, some of the 49 biomarkers correlated with each other and formed 5 distinct hierarchical clusters during acute peritonitis (Figure 1). These data suggested that a signature comprising as few as 5 parameters might already suffice to define a reliable immune fingerprint.Figure 1 Correlation analysis of local biomarkers in a total of 83 peritoneal dialysis patients on the day of presentation with acute peritonitis. Ellipses depict the correlation coefficients for each pair of biomarkers in the corresponding cell of the matrix, with the direction of the dip and the color of the shading representing positive and negative correlations, respectively. Only pairs with significant correlations (P < 0.05) are shown. Analyses were performed using the corrplot R and Hmisc R packages. GM-CSF, granulocyte macrophage colony-stimulating factor; HNE, human neutrophil elastase; IFN-γ, interferon-γ; IL, interleukin; MMP, matrix metalloproteinase; sIL-6R, soluble IL-6 receptor; TGF-β, transforming growth factor-β; TNF-α, tumor necrosis factor-α; VEGF, vascular endothelial growth factor.

d Hmisc R packages. GM-CSF, granulocyte macrophage colony-stimulating factor; HNE, human neutrophil elastase; IFN-γ, interferon-γ; IL, interleukin; MMP, matrix metalloproteinase; sIL-6R, soluble IL-6 receptor; TGF-β, transforming growth factor-β; TNF-α, tumor necrosis factor-α; VEGF, vascular endothelial growth factor. Table 1 Characteristics of the patient cohort investigated in this study Stable No growth Gram-positive Gram-negative no. % no. % no. % no. % Age, yr, mean ± SEM 70.1 ± 2.2 63.8 ± 2.9 65.9 ± 2.1 68.6 ± 3.3 18–40 1 4.2 1 5.3 4 8.5 1 5.9 40–50 1 4.2 1 5.3 3 6.4 0 0 50–60 0 0 1 5.3 6 12.8 3 17.7 60–70 8 33.3 11 57.9 14 29.8 3 17.7 70–80 10 41.7 3 15.8 12 25.5 8 47.1 ≥80 4 16.7 2 10.5 8 17.0 2 11.8 Sex Male 18 75 11 57.9 32 68.1 9 52.9 Female 6 25 8 42.1 15 31.9 8 47.1 Days on PD, mean ± SEM 1165.1 ± 186.5 638.8 ± 163.5 1022.9 ± 144.2 768.5 ± 245.8 0–30 0 0 3 15.8 1 2.1 0 0 30–360 1 4.2 4 21.1 15 31.9 9 52.9 360–720 9 37.5 6 31.6 8 17.0 3 17.7 720–1800 8 33.3 4 21.1 16 34.0 3 17.7 1800–3600 5 20.8 2 10.5 6 12.8 1 5.9 ≥3600 1 4.2 0 0 1 2.1 1 5.9 Hypertension Yes 12 50 2 10.5 14 29.8 7 41.2 No 12 50 17 89.5 33 70.2 10 58.8 Diabetes Yes 7 29.2 6 31.6 13 27.7 7 41.2 No 17 70.8 13 68.4 34 72.3 10 58.8 Technique failure Days 0–14 N/A 1 5.3 2 4.3 2 11.8 Days 14–30 N/A 2 10.5 3 6.5 2 11.8 Days 30–90 N/A 1 5.3 6 12.9 4 23.5 Mortality Days 0–14 N/A 1 5.3 0 0 1 5.9 Days 14–90 N/A 1 5.3 0 0 1 5.9 N/A, not available; PD, peritoneal dialysis.

0 58.8 Diabetes Yes 7 29.2 6 31.6 13 27.7 7 41.2 No 17 70.8 13 68.4 34 72.3 10 58.8 Technique failure Days 0–14 N/A 1 5.3 2 4.3 2 11.8 Days 14–30 N/A 2 10.5 3 6.5 2 11.8 Days 30–90 N/A 1 5.3 6 12.9 4 23.5 Mortality Days 0–14 N/A 1 5.3 0 0 1 5.9 Days 14–90 N/A 1 5.3 0 0 1 5.9 N/A, not available; PD, peritoneal dialysis. Feature selection methods define local fingerprints associated with Gram-negative infections We next divided the patients into groups according to the type of infecting organism. We initially attempted to define immune fingerprints that would reliably discriminate patients presenting with Gram-negative infections against all other cases of peritonitis (Supplementary Table S2A), based on our earlier observation of certain differences between Gram-negative and Gram-positive infections using logistic regression analyses.6 To this end, recursive feature elimination was used by evaluating the model performance according to the area under the receiver operating characteristic curve achieved and eliminating the least important features in each step. To reduce variability, 5 rounds of resampling methods were applied in the outer layer of the iteration, and cross-validation was used to avoid overfitting. These steps clearly demonstrated that Gram-negative infections were associated with unique different immune fingerprints. Figure 2a shows the number of features changing during feature elimination and the corresponding performance based on 3 different models, using SVMs, ANNs, and RFs. Whilst all 3 models successfully discriminated between Gram-negative infections and all other causes of peritonitis, RF-based feature elimination showed the best average performance, with the optimum biomarker combination comprising 8 features (area under receiver operating characteristic curve [AUC] = 0.993; sensitivity = 98.5% and specificity = 92.6%). In comparison, SVMs and ANNs were far less powerful for the recursive elimination of pathogen-related biomarkers, reaching overall lower degrees of sensitivity and specificity and requiring combinations comprising 10 and 30 features, respectively (Figure 2a).Figure 2 Identification of local immune fingerprints associated with peritonitis caused by Gram-negative bacteria.

or the recursive elimination of pathogen-related biomarkers, reaching overall lower degrees of sensitivity and specificity and requiring combinations comprising 10 and 30 features, respectively (Figure 2a).Figure 2 Identification of local immune fingerprints associated with peritonitis caused by Gram-negative bacteria. (a) Performance of recursive feature elimination models based on Random Forest (RF), Support Vector Machines (SVM), and artificial neural networks (ANN) for the prediction of Gram-negative infections (N = 17) against all other episodes of peritonitis (N = 66), shown as area under the receiver operating characteristic curve (AUC) depending on the number of biomarkers. Red symbols depict the maximum AUC achieved for each model. (b) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (c) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (d) Tukey plots of the top 5 biomarkers in patients with confirmed Gram-negative infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (**P < 0.01). (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IL, interleukin; TNF-α, tumor necrosis factor-α; VEGF, vascular endothelial growth factor.

with confirmed Gram-negative infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (**P < 0.01). (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IL, interleukin; TNF-α, tumor necrosis factor-α; VEGF, vascular endothelial growth factor. The top 5 and 10 individual biomarkers selected by the 3 different models and the corresponding average performance of the models based on combinations of these biomarkers are listed in Supplementary Table S2B. Of note, although the 3 models yielded different sets of biomarkers, the frequencies of Vγ9+ and Vδ2+ T cells within peritoneal T cells featured prominently in each. These findings appear to concur with our previous data suggesting a key role for Vγ9/Vδ2 T cells in Gram-negative infections and emphasize the diagnostic potential of those cells at the point of care.15, 28 Soluble biomarkers of particular interest for the prediction of Gram-negative infections using the RF model included local levels of tumor necrosis factor (TNF)-α, interleukin (IL)-12p40, and vascular endothelial growth factor (VEGF). Taken together, our findings demonstrate that RF models showed striking performance at combinations of 5 and fewer biomarkers, thereby making it the algorithm of choice for a clinically viable prediction of the causative pathogen in individuals presenting with PD-related peritonitis.

r endothelial growth factor (VEGF). Taken together, our findings demonstrate that RF models showed striking performance at combinations of 5 and fewer biomarkers, thereby making it the algorithm of choice for a clinically viable prediction of the causative pathogen in individuals presenting with PD-related peritonitis. Internal validation of individual biomarkers constituting Gram-negative immune fingerprints We next sought to validate the findings from the feature elimination process by assessing the distribution and performance of the top 5 biomarkers that had been selected by the RF model. Using skewness as a measure of symmetry and kurtosis as a measure of peakedness, all 5 biomarkers showed very low positive skewness and only limited or even negative kurtosis in patients presenting with Gram-negative infections (Figure 2b). In contrast, these 5 biomarkers had generally higher skewness and considerable positive kurtosis in patients with other episodes of peritonitis, especially in the case of Vδ2+ T-cell frequencies and levels of IL-12p40 and VEGF. Among the top 5 biomarkers, the local frequencies of Vγ9+ and Vδ2+ T cells and the levels of TNF-α on their own already showed relatively good sensitivities and specificities to identify Gram-negative infections with AUCs ≥0.75 for each biomarker, much more so than levels of IL-12p40 and VEGF (Figure 2c), but far lower than the full signature that reached an AUC of 0.99 (Supplementary Table S2B). This conclusion was supported by classic analyses using the Mann-Whitney U test showing that levels of Vγ9+ and Vδ2+ T cells and TNF-α, but not of IL-12p40 and VEGF, were markedly different between patients with Gram-negative infections and patients with other episodes of peritonitis (Figure 2d). These results confirmed the importance of the biomarkers selected by recursive feature elimination, especially of local levels of Vγ9+ and Vδ2+ T cells and of TNF-α, in identifying Gram-negative organisms. However, our analyses also identified shortcomings of conventional statistical methods that were especially apparent when visualizing the individual readings across all patients in the form of a heat map, where the overall differences between Gram-negative and other episodes of peritonitis were not very pronounced (Figure 2e).

our analyses also identified shortcomings of conventional statistical methods that were especially apparent when visualizing the individual readings across all patients in the form of a heat map, where the overall differences between Gram-negative and other episodes of peritonitis were not very pronounced (Figure 2e). Overall, the performances of the individual biomarkers lagged behind their combined performance in an RF model, demonstrating the importance of defining complex signatures comprising distinct biomarkers and of assessing their relationships in nonlinear models.

our analyses also identified shortcomings of conventional statistical methods that were especially apparent when visualizing the individual readings across all patients in the form of a heat map, where the overall differences between Gram-negative and other episodes of peritonitis were not very pronounced (Figure 2e). Overall, the performances of the individual biomarkers lagged behind their combined performance in an RF model, demonstrating the importance of defining complex signatures comprising distinct biomarkers and of assessing their relationships in nonlinear models. Patients with culture-negative episodes of peritonitis show distinct local immune fingerprints associated with milder inflammation Although microbiological culture remains the method of choice for diagnosis of infection, a considerable proportion of samples does not yield any culture results, thereby not allowing a reliable designation of the underlying cause of the inflammatory episode, which may or may not be infectious.29 Patients with culture-negative episodes often show less severe inflammation and have better clinical outcomes, suggesting that local biomarkers might aid in the diagnosis, treatment, and prognosis of such episodes. Here, RF models selected biomarker signatures that reliably distinguished samples with no growth from cases with confirmed bacterial infection (Figure 3a, Supplementary Tables S3A and S3B). The top 5 biomarkers all showed great potential in identifying culture-negative episodes that were characterized by relatively low total cell counts, an increased proportion of CD14+ monocytes/macrophages in the cellular infiltrate, and lower levels of IL-1β, matrix metalloproteinase (MMP)-8 and the chemokine CCL4 compared with confirmed infections (Figure 3b–e, Supplementary Table S3B). Despite the heterogeneity of this patient group, in which the failure to grow organisms might be due to inappropriate sampling, poor handling and culture techniques, low organism numbers, ongoing treatment with antibiotics for unrelated infections, or nonmicrobial disease such as sterile inflammation or viral infection, our findings indicate that culture-negative episodes are immunologically distinct from confirmed cases of bacterial peritonitis and are characterized by a less severe inflammatory response.Figure 3 Local immune fingerprints in culture-negative episodes of peritonitis. (a) Performance of Random Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models for the prediction of culture-negative episodes (no growth, N = 19) against microbiologically confirmed infections (other, N = 64), shown as area under the curve (AUC) depending on the number of biomarkers.

m Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models for the prediction of culture-negative episodes (no growth, N = 19) against microbiologically confirmed infections (other, N = 64), shown as area under the curve (AUC) depending on the number of biomarkers. Red symbols depict the maximum AUC for each model. (b) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (c) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (d) Tukey plots of the top 5 biomarkers in patients with culture-negative peritonitis and with infectious (other) episodes of peritonitis, as assessed by Mann-Whitney tests (***P < 0.001). (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IL, interleukin; MMP, matrix metalloproteinase.

o reach satisfactory AUCs (Supplementary Figure S1). The best 5 biomarkers together only reached an AUC of 0.711 in the RF model, and none of the individual markers on their own—IL-17A, IL-12p40, interferon-γ, IL-1β, and total cell count—exceeded an AUC of 0.72 or stayed well below that value (Supplementary Table S4B). Of note, there is considerable heterogeneity in the Gram-positive organisms causing PD-related peritonitis, comprising streptococci, staphylococci, coryneforms, and other bacteria that cause clinically distinct diseases with different outcomes and require different antibiotics.12 We therefore attempted to define the pathogen-specific immune responses to subtypes of Gram-positive organisms. These analyses demonstrated that in the Gram-positive group, streptococcal infections caused by Streptococcus and Enterococcus species were associated with markedly distinct immune responses compared with all other cases of peritonitis (Figure 4a, Supplementary Table S5A). The most promising biomarker combination consisted of the local levels of IL-1β, TNF-β, and IL-15 together with the enzymatic activity of MMPs in the PD effluent, as measured by specific cleavage of a fluorogenic MMP substrate and by gelatin zymography on gel electrophoretic separation (Figure 4b). Individually, IL-1β, TNF-β, and zymography showed differences between streptococcal infections and all other patients, yet only an RF-based model revealed their full diagnostic potential with an AUC of 0.969 (Figure 4c–e, Supplementary Table S5B).Figure 4 Local immune fingerprints in streptococcal (Strep) infections. (a) Performance of Random Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models for the prediction of infections caused by streptococcal species (Streptococcus spp. and Enterococcus spp., N = 16) against all other episodes of peritonitis (N = 67), shown as area under the curve (AUC) depending on the number of biomarkers. One episode of peritonitis classified as streptococcal infection was a coinfection caused by Enterococcus sp. with light growth of coagulase-negative Staphylococcus spp. Red symbols depict the maximum AUC for each model. (b) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (c) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers.

occus sp. with light growth of coagulase-negative Staphylococcus spp. Red symbols depict the maximum AUC for each model. (b) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (c) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (d) Tukey plots of the top 5 biomarkers in patients with confirmed streptococcal infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01). (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IL, interleukin; MMP, matrix metalloproteinase; TNF-β, tumor necrosis factor-β. Similar to the definition of Streptococcus/Enterococcus-specific immune fingerprints, nonstreptococcal Gram-positive infections caused by Staphylococcus aureus, coagulase-negative Staphylococcus spp. (CNS) and Corynebacterium spp. were associated with biomarker signatures that distinguished them from all other episodes of peritonitis (Supplementary Figure S2, Supplementary Table S6A). Yet, despite a statistically significant difference between such nonstreptococcal Gram-positive infections and other episodes, especially for local levels of IL-17A, interferon-γ, and IL-15, RF-based algorithms were not as powerful in this case as for the above predictions of Gram-negative or streptococcal infections, most likely due to the remaining heterogeneity of the organisms in that patient group (Supplementary Table S6B).

ther episodes, especially for local levels of IL-17A, interferon-γ, and IL-15, RF-based algorithms were not as powerful in this case as for the above predictions of Gram-negative or streptococcal infections, most likely due to the remaining heterogeneity of the organisms in that patient group (Supplementary Table S6B). Given that CNS species such as Staphylococcus epidermidis are the major cause of peritonitis in PD patients and are also clinically associated with a relatively benign outcome,30 we finally determined immune fingerprints that would specifically define CNS infections. Despite having only marginal or no statistical significance as individual biomarkers on conventional tests, the combination of IL-15, IL-16, and soluble IL-6 receptor (sIL-6R) levels, total cell count, and MMP substrate turnover showed excellent performance in the RF model, with an AUC of 0.961 (Figure 5, Supplementary Tables S7A+B), demonstrating that CNS infections are sufficiently immunologically distinct to allow a pathogen-specific diagnosis in PD patients.Figure 5 Local immune fingerprints in coagulase-negative Staphylococcus (CNS) infections. (a) Performance of Random Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models for the prediction of infections caused by CNS (Staphylococcus epidermidis and related species; N = 21) against all other episodes of peritonitis (N = 62), shown as area under the curve (AUC) depending on the number of biomarkers. Red symbols depict the maximum AUC for each model. (b) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (c) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (d) Tukey plots of the top 5 biomarkers in patients with confirmed CNS infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05). (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IL, interleukin; MMP, matrix metalloproteinase; sIL-6R, soluble IL-6 receptor.

biomarkers in patients with confirmed CNS infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05). (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IL, interleukin; MMP, matrix metalloproteinase; sIL-6R, soluble IL-6 receptor. Local biomarkers on the day of presentation correlate with subsequent clinical outcomes over the following 90 days The nature of the causative pathogen and the underlying inflammatory response profoundly affect clinical outcomes of peritonitis in PD patients,6, 15 indicating a need for prognostic biomarkers in early disease. When applying recursive feature elimination methods to our dataset, certain models accurately predicted the risk of downstream complications, defined as technique failure over the following 90 days, including catheter removal, transfer to hemodialysis, and death (Figure 6). Patients experiencing technique failure after an episode of acute peritonitis had marginally higher levels of calprotectin, MMP-8, sIL-6R, and transforming growth factor-β in their dialysis effluent as well as lower CD4+ : CD8+ T-cell ratios compared with uncomplicated cases. Although not being significantly different on their own, the combination of these 5 parameters in RF models yielded an AUC of 0.911 with excellent sensitivity (Supplementary Tables S8A and S8B), implying that the development of prognostic tests is feasible.Figure 6 Local immune fingerprints associated with poor clinical outcomes. (a) Performance of Random Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models for the prediction of technique failure over the next 90 days (catheter removal, transfer to hemodialysis, or peritonitis-related death; N = 23) against all other episodes of peritonitis (N = 60), shown as area under the curve (AUC) depending on the number of biomarkers. Red symbols depict the maximum AUC for each model. (b) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (c) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (d) Tukey plots of the top 5 biomarkers in patients with subsequent technique failure and all other patients, as assessed by Mann-Whitney tests. (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis.

teristic analysis showing specificity and sensitivity of the top 5 biomarkers. (d) Tukey plots of the top 5 biomarkers in patients with subsequent technique failure and all other patients, as assessed by Mann-Whitney tests. (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. MMP, matrix metalloproteinase; sIL-6R, soluble IL-6 receptor; TGF-β, transforming growth factor-β.

teristic analysis showing specificity and sensitivity of the top 5 biomarkers. (d) Tukey plots of the top 5 biomarkers in patients with subsequent technique failure and all other patients, as assessed by Mann-Whitney tests. (e) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. MMP, matrix metalloproteinase; sIL-6R, soluble IL-6 receptor; TGF-β, transforming growth factor-β. Taken together, our findings show that combinations of local biomarkers readily identify clinically meaningful subgroups of peritonitis patients, depending on the culture results and subsequent clinical outcomes (Figure 7). With microbiologically distinct infections also displaying immunologically distinct immune responses, most individual parameters constituting meaningful fingerprints were only associated with single patient groups. However, certain biomarkers featured more prominently in these mathematical models than others, suggesting that they are particularly important factors in acute peritonitis, such as IL-1β and IL-15, each of which contributed to 3 different biomarker signatures, and the total cell count that was included in 4 different signatures and is thus likely to be of the utmost relevance in the diagnosis of PD patients.Figure 7 Summary of disease-specific immune fingerprints in patients presenting with acute peritonitis. Shown are the top 5 biomarkers associated with the type of causative organism as indicated or with the risk of technique failure over the next 90 days. IFN-γ, interferon-γ; IL, interleukin; MMP, matrix metalloproteinase; sIL-6R, soluble IL-6 receptor; TGF-β, transforming growth factor-β; TNF-α, tumor necrosis factor-α; VEGF, vascular endothelial growth factor.

with the type of causative organism as indicated or with the risk of technique failure over the next 90 days. IFN-γ, interferon-γ; IL, interleukin; MMP, matrix metalloproteinase; sIL-6R, soluble IL-6 receptor; TGF-β, transforming growth factor-β; TNF-α, tumor necrosis factor-α; VEGF, vascular endothelial growth factor. Discussion This study demonstrates that different groups of bacteria induce qualitatively distinct local immune responses in infected patients. Specific biomarker signatures were associated with acute infections by Gram-negative and Gram-positive organisms, respectively, and with culture-negative episodes of peritonitis of unclear etiology. Taking advantage of the unique access to local inflammatory responses that is possible in PD patients via the peritoneal catheter and drained dialysate, we were also able for the first time to characterize pathogen-specific local immune responses to defined genera of organisms such as streptococcal species and coagulase-negative staphylococci. These findings demonstrate the feasibility of developing rapid pathogen-specific diagnostics that exploit the exquisite responsiveness and specificity of the human immune system for different types of organisms. Such rapid methods might have greater utility than microbiological and molecular methods that aim at directly detecting such organisms but that are too slow, subject to confounding contaminants and often lack sufficient sensitivity to inform early patient management and target antibiotic therapy on the day of presentation with acute disease.31

have greater utility than microbiological and molecular methods that aim at directly detecting such organisms but that are too slow, subject to confounding contaminants and often lack sufficient sensitivity to inform early patient management and target antibiotic therapy on the day of presentation with acute disease.31 Our study also demonstrates the power of using nonlinear approaches for mining complex biomedical datasets where conventional statistical methods fail to yield satisfactory results and where individual biomarkers on their own are unlikely to reach sufficient sensitivity and specificity to change clinical practice. Notably, the nature of the signatures identified in this study varied according to the mathematical model applied. By directly comparing 3 different approaches and assessing their performance when predicting microbiological and clinical endpoints in PD patients, we identified RFs as the most suitable models in this study and patient cohort, yielding superior performances and at fewer biomarkers than SVMs and ANNs, in accordance with investigations in other fields of research.32, 33, 34

ing their performance when predicting microbiological and clinical endpoints in PD patients, we identified RFs as the most suitable models in this study and patient cohort, yielding superior performances and at fewer biomarkers than SVMs and ANNs, in accordance with investigations in other fields of research.32, 33, 34 Gram-negative bacteria, streptococci, and CNS are the major types of bacteria causing peritonitis in individuals receiving PD.11 Strikingly, clinical outcomes of infections by those organisms differ, implying differences in their pathogenicity, their susceptibility to antibiotics, and/or the pathophysiology of the host responses they trigger.12, 15 Our findings demonstrate that in addition to their microbiological differences, Gram-negative bacteria, streptococci, and CNS elicit fundamentally distinct immune responses, which is not only of relevance for the development of novel diagnostics but potentially also highlights key factors and cell types involved in sensing and fighting such infections. In this context, Vγ9/Vδ2 T cells and TNF-α appear to be particularly relevant in Gram-negative infections, IL-1β and MMPs in streptococcal infections, and IL-16 and sIL-6R in CNS infections. IL-15 featured in prediction models for streptococcal, nonstreptococcal, and CNS infection, indicating that this cytokine may play a role in all Gram-positive infections. In addition, the fact that IL-17A and interferon-γ featured prominently in nonstreptococcal Gram-positive infections (which included CNS), but not in CNS infections themselves, may argue for a particular role of both cytokines in response to Gram-positive bacteria other than streptococci and CNS (i.e., in infections by Staphylococcus aureus and/or coryneform bacteria). The organism-specific contributions of these soluble mediators during acute disease, the cell types that produce them, and the targets that respond to them can now be addressed in appropriate cellular assays, suitable animal models that mimic the situation in patients as closely as possible, and well-defined cohorts with bacterial peritonitis and other infections with access to the site of inflammation. This will ultimately further our understanding of antimicrobial immune responses and how to exploit such knowledge diagnostically and therapeutically.

mimic the situation in patients as closely as possible, and well-defined cohorts with bacterial peritonitis and other infections with access to the site of inflammation. This will ultimately further our understanding of antimicrobial immune responses and how to exploit such knowledge diagnostically and therapeutically. The roles of calprotectin, MMP-8, sIL-6R, and transforming growth factor-β as well as the balance between CD4+ and CD8+ T cells may deserve special attention with regard to their involvement in regulating pathologic processes in the peritoneal cavity, and their contribution to predicting clinical outcomes.

mimic the situation in patients as closely as possible, and well-defined cohorts with bacterial peritonitis and other infections with access to the site of inflammation. This will ultimately further our understanding of antimicrobial immune responses and how to exploit such knowledge diagnostically and therapeutically. The roles of calprotectin, MMP-8, sIL-6R, and transforming growth factor-β as well as the balance between CD4+ and CD8+ T cells may deserve special attention with regard to their involvement in regulating pathologic processes in the peritoneal cavity, and their contribution to predicting clinical outcomes. Taken together, we successfully applied different machine learning models to complex biomedical datasets and identified key pathways involved in pathogen-specific immune responses at the site of infection. It is apparent that the nature of the signatures identified depends on both biological and analytical parameters. However, our current findings demonstrate that such methodologies have immediate diagnostic and prognostic implications at the point of care, by informing patient management and the choice of treatment before traditional culture results become available. Being based on a relatively small population in a single hospital, the biomarkers identified in this study and the corresponding algorithms now await external validation in larger patient cohorts at multiple sites35 in order to demonstrate the applicability of the chosen approach to other centers where the spectrum of the infecting organisms and the previous infection history as well as patient demographics and health care settings may vary. Validated biomarker combinations can then be incorporated into appropriate diagnostic tests to be used in central laboratories or at the point of care and into new patient management and treatment guidelines based on such test results. The final choice of biomarkers to be taken forward will depend on the desired performance requirements, with soluble proteins being equally suitable for automated immunodiagnostic analyzers and bedside or home tests, whilst assessments of immune cell subsets such as Vγ9/Vδ2 T cells would require standardized flow cytometric protocols. In the meantime, our study reaffirms the importance of correctly interpreting simple parameters such as the total cell count (contributing to 4 different immune fingerprints) and the differential leukocyte count (reflected in the proportion of CD14+ cells among total cells), which already convey vital information about the nature of the causative pathogen.

he importance of correctly interpreting simple parameters such as the total cell count (contributing to 4 different immune fingerprints) and the differential leukocyte count (reflected in the proportion of CD14+ cells among total cells), which already convey vital information about the nature of the causative pathogen. Materials and Methods Patient samples This study was approved by the South East Wales Local Ethics Committee (04WSE04/27) and registered on the UK Clinical Research Network Study Portfolio under reference number #11838 “Patient Immune Responses to Infection in PD.” All individuals provided written informed consent. The local cohort comprised 83 adults PD patients admitted between 2008 and 2016 to the University Hospital of Wales, Cardiff, on day 1 of acute peritonitis. Twenty-four age- and sex-matched stable PD patients with no infection in the previous 3 months were included as controls. Subjects positive for HIV or hepatitis C virus were excluded.

prised 83 adults PD patients admitted between 2008 and 2016 to the University Hospital of Wales, Cardiff, on day 1 of acute peritonitis. Twenty-four age- and sex-matched stable PD patients with no infection in the previous 3 months were included as controls. Subjects positive for HIV or hepatitis C virus were excluded. Clinical diagnosis of acute peritonitis was based on the presence of abdominal pain and cloudy peritoneal effluent with >100 white blood cells per cubic millimeter. According to the microbiological analysis of the effluent from preinoculated blood culture bottles by the routine Microbiology Laboratory, Public Health Wales, peritonitis episodes were defined as culture-negative (N = 19, after incubation of up to 5 days) or as confirmed bacterial infections by Gram-positive (N = 47) and Gram-negative organisms (N = 17) (Table 1). Cases of fungal infection and mixed or unclear culture results were excluded from the study. Clinical outcomes were recorded by following patients for 90 days after presenting with peritonitis. Technique failure was defined as catheter removal, transfer to hemodialysis or death within 90 days, and occurred in 21.1% of culture-negative episodes, 23.4% of Gram-positive infections, and 47.1% of Gram-negative infections (Table 1). Samples from ≥8-hour dwells with volumes of 1 to 2.5 L were collected for biomarker measurements and processed as previously described.6, 13, 14, 15

modialysis or death within 90 days, and occurred in 21.1% of culture-negative episodes, 23.4% of Gram-positive infections, and 47.1% of Gram-negative infections (Table 1). Samples from ≥8-hour dwells with volumes of 1 to 2.5 L were collected for biomarker measurements and processed as previously described.6, 13, 14, 15 Cellular biomarkers Cells from cloudy peritoneal effluents were acquired on an 8-color FACSCanto II (BD Biosciences, San Diego, CA) and analyzed with FlowJo 10.1 (TreeStar, Ashland, OR), using monoclonal antibodies against CD3 (SK7), CD4 (RPA-T4), CD8 (RPA-T8), CD15 (HI98 or HIM1), and TCR-Vδ2 from BD Biosciences; anti-TCR-Vγ9 (Immu360) from Beckman Coulter (Brea, CA); and anti-CD14 (61D3) from eBioscience (San Diego, CA); together with appropriate isotype controls. Leukocyte populations were gated based on their appearance in side scatter and forward scatter area/height and exclusion of live/dead staining (Fixable Aqua; Invitrogen, Carlsbad, CA). Biomarkers determined were the total cell counts; the percentages of CD3+ T cells, CD14+ monocytes/macrophages, and CD15+ neutrophils among total cells; the frequencies of CD4+, CD8+, Vγ9+, and Vδ2+ cells within the CD3+ T cell gate; and the ratios of CD4+ to CD8+ T cells.

ning (Fixable Aqua; Invitrogen, Carlsbad, CA). Biomarkers determined were the total cell counts; the percentages of CD3+ T cells, CD14+ monocytes/macrophages, and CD15+ neutrophils among total cells; the frequencies of CD4+, CD8+, Vγ9+, and Vδ2+ cells within the CD3+ T cell gate; and the ratios of CD4+ to CD8+ T cells. Soluble biomarkers Cell-free peritoneal effluents were analyzed on a SECTOR Imager 6000 (Meso Scale Discovery, Rockville, MD) using the V-PLEX Human Cytokine 30-Plex Kit to measure levels of IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-10, IL-12p40, IL-12p70, IL-13, IL-15, IL-16, IL-17A, interferon-γ, TNF-α, TNF-β, granulocyte macrophage colony-stimulating factor, and VEGF as well as the chemokines CCL2, CCL3, CCL4, CCL11, CCL13, CCL17, CCL22, CCL26, CXCL8, and CXCL10; ultrasensitive single-plex assays for sIL-6R and IL-18 (Meso Scale Discovery); and a customer-made single-plex assay for IL-22 using capture (MAB7822) and biotinylated detection antibodies (BAM7821) and recombinant human IL-22 from R&D Systems. Conventional enzyme-linked immunosorbent assay kits were used to measure transforming growth factor-β, total MMP-8, total MMP-9, and surfactant protein D (R&D Systems); calprotectin (Hycult Biotech, Inc., Plymouth Meeting, PA); and CCL2 (BD Biosciences). Human neutrophil elastase was measured using a B.I.T.S. enzyme-linked immunosorbent assay kit (Mologic, Bedford, UK). Active human neutrophil elastase was measured using the fluorogenic substrate MeOSuc-Ala-Ala-Pro-Val-AMC (Bachem, Bubendorf, Switzerland); active MMP was measured using the fluorogenic substrate Mca-Lys-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 (Enzo Life Sciences, Farmingdale, NY) and by zymography using precast Novex gelatin zymogram gels (Invitrogen) scanned on a Bio-Rad GS800 densitometer (Bio-Rad Laboratories, Berkeley, CA) and analyzed using ImageJ software. Total protein was measured using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA). Measurements that were below or above the detection limit were replaced by the lowest and highest detectable values for each biomarker, respectively.

Laboratories, Berkeley, CA) and analyzed using ImageJ software. Total protein was measured using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA). Measurements that were below or above the detection limit were replaced by the lowest and highest detectable values for each biomarker, respectively. Data preprocessing All analyses were performed using R software version 3.2.5 (R Foundation, Vienna, Austria). Before applying machine learning models, missing data imputation was applied to fit gaps due to missing or failed measurements by adopting Multivariate Imputation by Chained Equations,36 which impute an incomplete feature by generating synthetic values taking into account their relationship with other biomarkers, using RF models (Supplementary Table S9). Data were then standardized to a mean of 0 and a variance of 1 to reduce the effect of large feature range variation. After preprocessing, the samples in the minority groups were unsampled so that minority and majority groups had equal frequencies.

onship with other biomarkers, using RF models (Supplementary Table S9). Data were then standardized to a mean of 0 and a variance of 1 to reduce the effect of large feature range variation. After preprocessing, the samples in the minority groups were unsampled so that minority and majority groups had equal frequencies. Feature elimination The caret package in R37 was adopted for the implementation of recursive feature elimination methods using 3 different machine learning models: SVMs with radial basis function kernel in the kernlab R package,38 RFs with Breiman’s algorithm in the randomForest R package,25 and single-hidden-layer ANNs in the nnet R package.39 To reduce variability, resampling methods were applied in the outer layer of the iteration, and cross-validation was used in the model fitting and parameter tuning to avoid overfitting. During parameter tuning, the search regions for hyperparameters in different classification models were set as follows: the penalty factor C that controls the tradeoff between learning errors and the complexity term and the radial basis function kernel parameter σ both ranged from 2−5 to 25 with steps of 2; the number of trees in RFs were selected from [100, 300, 500, 1000, 3000], and the size of ANNs ranged from 2 to 210 hidden nodes with steps of 2 and a weight decay from 10−4 to 10−10 with steps of 10. The number of repeats for resampling was set as 5 and for cross-validation and model selection as 10.

with steps of 2; the number of trees in RFs were selected from [100, 300, 500, 1000, 3000], and the size of ANNs ranged from 2 to 210 hidden nodes with steps of 2 and a weight decay from 10−4 to 10−10 with steps of 10. The number of repeats for resampling was set as 5 and for cross-validation and model selection as 10. Basic statistical analyses Correlations between all 49 biomarkers were determined using the corrplot R package,40 based on correlation calculations using the Hmisc R package.41 Means, SEs, skewness, and kurtosis were determined with plotrix R42 and e1071 R packages.43 Mann-Whitney tests (2-sample Wilcoxon tests) were applied to assess the relationship between two patient groups. Heat maps were visualized using the gplots R package.44 Disclosure C-YL, NT, and ME are inventors on patent applications filed by University College Cardiff Consultants Ltd. (Cardiff University) in Europe and the United States on the identification of bacterial infections in peritoneal dialysis patients. PD is the cofounder and chief scientific officer of Mologic Ltd.; PD and GP are shareholders/option holders in Mologic. All the other authors declared no competing interests.

sultants Ltd. (Cardiff University) in Europe and the United States on the identification of bacterial infections in peritoneal dialysis patients. PD is the cofounder and chief scientific officer of Mologic Ltd.; PD and GP are shareholders/option holders in Mologic. All the other authors declared no competing interests. Supplementary Material Figure S1 Local immune fingerprints in Gram-positive infections. (A) Performance of Random Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models for the prediction of Gram-positive infections (N = 47) against all other episodes of peritonitis (N = 36), shown as area under the curve (AUC) depending on the number of biomarkers. Red symbols depict the maximum AUC for each model. (B) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (C) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (D) Tukey plots of the top 5 biomarkers in patients with confirmed Gram-positive infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01; ***P < 0.001). (E) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IFN-γ, interferon-γ; IL, interleukin.

top 5 biomarkers in patients with confirmed Gram-positive infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01; ***P < 0.001). (E) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IFN-γ, interferon-γ; IL, interleukin. Figure S2 Local immune fingerprints in nonstreptococcal Gram-positive infections. (A) Performance of Random Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models, for the prediction of infections caused by nonstreptococcal Gram-positive species (Staphylococcus aureus, coagulase-negative Staphylococcus spp., Corynebacterium spp.; N = 31) against all other episodes of peritonitis (N = 62), shown as area under the curve (AUC) depending on the number of biomarkers. One episode of peritonitis classified as nonstreptococcal (Non-strep) infection was a coinfection caused by Corynebacterium sp. and coagulase-negative Staphylococcus sp. Red symbols depict the maximum AUC for each model. (B) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (C) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (D) Tukey plots of the top 5 biomarkers in patients with confirmed nonstreptococcal Gram-positive infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01; ***P < 0.001). (E) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis.

) Tukey plots of the top 5 biomarkers in patients with confirmed nonstreptococcal Gram-positive infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01; ***P < 0.001). (E) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. Table S1 Local biomarkers in stable PD patients and in patients presenting with acute peritonitis. Table S2A Local biomarkers in patients presenting with acute peritonitis caused by Gram-negative organisms or with other episodes. Table S2B Performance of local biomarkers in predicting Gram-negative infections in PD patients against all other episodes of peritonitis. Table S3A Local biomarkers in patients presenting with culture-negative peritonitis or with other episodes. Table S3B Performance of local biomarkers in predicting culture-negative episodes in PD patients against all other microbiologically confirmed infections. Table S4A Local biomarkers in patients presenting with acute peritonitis caused by Gram-positive organisms or with other episodes. Table S4B Performance of local biomarkers in predicting Gram-positive infections in PD patients against all other episodes of peritonitis. Table S5A Local biomarkers in patients presenting with acute peritonitis caused by streptococcal species (Streptococcus spp. and Enterococcus spp.) or with other episodes. Table S5B Performance of local biomarkers in predicting infections caused by streptococcal species (Streptococcus spp. and Enterococcus spp.) against all other episodes of peritonitis.

Table S5A Local biomarkers in patients presenting with acute peritonitis caused by streptococcal species (Streptococcus spp. and Enterococcus spp.) or with other episodes. Table S5B Performance of local biomarkers in predicting infections caused by streptococcal species (Streptococcus spp. and Enterococcus spp.) against all other episodes of peritonitis. Table S6A Local biomarkers in patients presenting with acute peritonitis caused by nonstreptococcal Gram-positive species (Staphylococcus aureus, coagulase-negative Staphylococcus spp., Corynebacterium spp.) or with other episodes. Table S6B Performance of local biomarkers in predicting infections caused by nonstreptococcal Gram-positive species (Staphylococcus aureus, coagulase-negative Staphylococcus spp., Corynebacterium spp.) against all other episodes of peritonitis. Table S7A Local biomarkers in patients presenting with acute peritonitis caused by coagulase-negative staphylococci or with other episodes. Table S7B Performance of local biomarkers in predicting infections caused by coagulase-negative staphylococci against all other episodes of peritonitis. Table S8A Local biomarkers in patients technique failure over the next 90 days or with other episodes. Table S8B Performance of local biomarkers in predicting technique failure over the next 90 days against all other episodes of peritonitis. Table S9 Proportion of missing values.

Table S7B Performance of local biomarkers in predicting infections caused by coagulase-negative staphylococci against all other episodes of peritonitis. Table S8A Local biomarkers in patients technique failure over the next 90 days or with other episodes. Table S8B Performance of local biomarkers in predicting technique failure over the next 90 days against all other episodes of peritonitis. Table S9 Proportion of missing values. Acknowledgments We are grateful to all patients and volunteers for participating in this study and to the clinicians and nurses for their cooperation. We especially thank Delyth Davies, Sally Jones, Billy Keogh, Chia-Te Liao, and Sharron Tatchell for help with patient recruitment and sampling and Eryl Francis for her support throughout this study. This research received support from the Wales Kidney Research Unit (WKRU), UK Clinical Research Network (UKCRN) Study Portfolio, NISCHR/MRC Health Research Partnership Award HA09-009, Kidney Research UK grant RP6/2014, MRC grant MR/N023145/1, NIHR i4i Product Development Award II-LA-0712-20006, NISCHR/Wellcome Trust Institutional Strategic Support Fund (ISSF), MRC Confidence in Concept scheme, SARTRE/SEWAHSP Health Technology Challenge scheme, and EU-FP7 Initial Training Network 287813 “European Training & Research in Peritoneal Dialysis” (EuTRiPD). see commentary on page 16

Acknowledgments We are grateful to all patients and volunteers for participating in this study and to the clinicians and nurses for their cooperation. We especially thank Delyth Davies, Sally Jones, Billy Keogh, Chia-Te Liao, and Sharron Tatchell for help with patient recruitment and sampling and Eryl Francis for her support throughout this study. This research received support from the Wales Kidney Research Unit (WKRU), UK Clinical Research Network (UKCRN) Study Portfolio, NISCHR/MRC Health Research Partnership Award HA09-009, Kidney Research UK grant RP6/2014, MRC grant MR/N023145/1, NIHR i4i Product Development Award II-LA-0712-20006, NISCHR/Wellcome Trust Institutional Strategic Support Fund (ISSF), MRC Confidence in Concept scheme, SARTRE/SEWAHSP Health Technology Challenge scheme, and EU-FP7 Initial Training Network 287813 “European Training & Research in Peritoneal Dialysis” (EuTRiPD). see commentary on page 16 Figure S1. Local immune fingerprints in Gram-positive infections. (A) Performance of Random Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models for the prediction of Gram-positive infections (N = 47) against all other episodes of peritonitis (N = 36), shown as area under the curve (AUC) depending on the number of biomarkers. Red symbols depict the maximum AUC for each model. (B) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (C) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (D) Tukey plots of the top 5 biomarkers in patients with confirmed Gram-positive infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01; ***P < 0.001). (E) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IFN-γ, interferon-γ; IL, interleukin.

top 5 biomarkers in patients with confirmed Gram-positive infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01; ***P < 0.001). (E) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. IFN-γ, interferon-γ; IL, interleukin. Figure S2. Local immune fingerprints in nonstreptococcal Gram-positive infections. (A) Performance of Random Forest (RF), Support Vector Machine (SVM), and artificial neural network (ANN)–based feature elimination models, for the prediction of infections caused by nonstreptococcal Gram-positive species (Staphylococcus aureus, coagulase-negative Staphylococcus spp., Corynebacterium spp.; N = 31) against all other episodes of peritonitis (N = 62), shown as area under the curve (AUC) depending on the number of biomarkers. One episode of peritonitis classified as nonstreptococcal (Non-strep) infection was a coinfection caused by Corynebacterium sp. and coagulase-negative Staphylococcus sp. Red symbols depict the maximum AUC for each model. (B) Kurtosis and skewness of the top 5 biomarkers selected by RF-based feature elimination. (C) Receiver operating characteristic analysis showing specificity and sensitivity of the top 5 biomarkers. (D) Tukey plots of the top 5 biomarkers in patients with confirmed nonstreptococcal Gram-positive infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01; ***P < 0.001). (E) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis.

) Tukey plots of the top 5 biomarkers in patients with confirmed nonstreptococcal Gram-positive infections and with all other episodes of peritonitis, as assessed by Mann-Whitney tests (*P < 0.05; **P < 0.01; ***P < 0.001). (E) Heat map showing the top 5 biomarkers across all patients presenting with acute peritonitis. Table S1. Local biomarkers in stable PD patients and in patients presenting with acute peritonitis. Table S2A. Local biomarkers in patients presenting with acute peritonitis caused by Gram-negative organisms or with other episodes. Table S2B. Performance of local biomarkers in predicting Gram-negative infections in PD patients against all other episodes of peritonitis. Table S3A. Local biomarkers in patients presenting with culture-negative peritonitis or with other episodes. Table S3B. Performance of local biomarkers in predicting culture-negative episodes in PD patients against all other microbiologically confirmed infections. Table S4A. Local biomarkers in patients presenting with acute peritonitis caused by Gram-positive organisms or with other episodes. Table S4B. Performance of local biomarkers in predicting Gram-positive infections in PD patients against all other episodes of peritonitis. Table S5A. Local biomarkers in patients presenting with acute peritonitis caused by streptococcal species (Streptococcus spp. and Enterococcus spp.) or with other episodes. Table S5B. Performance of local biomarkers in predicting infections caused by streptococcal species (Streptococcus spp. and Enterococcus spp.) against all other episodes of peritonitis.

Table S5A. Local biomarkers in patients presenting with acute peritonitis caused by streptococcal species (Streptococcus spp. and Enterococcus spp.) or with other episodes. Table S5B. Performance of local biomarkers in predicting infections caused by streptococcal species (Streptococcus spp. and Enterococcus spp.) against all other episodes of peritonitis. Table S6A. Local biomarkers in patients presenting with acute peritonitis caused by nonstreptococcal Gram-positive species (Staphylococcus aureus, coagulase-negative Staphylococcus spp., Corynebacterium spp.) or with other episodes. Table S6B. Performance of local biomarkers in predicting infections caused by nonstreptococcal Gram-positive species (Staphylococcus aureus, coagulase-negative Staphylococcus spp., Corynebacterium spp.) against all other episodes of peritonitis. Table S7A. Local biomarkers in patients presenting with acute peritonitis caused by coagulase-negative staphylococci or with other episodes. Table S7B. Performance of local biomarkers in predicting infections caused by coagulase-negative staphylococci against all other episodes of peritonitis. Table S8A. Local biomarkers in patients technique failure over the next 90 days or with other episodes. Table S8B. Performance of local biomarkers in predicting technique failure over the next 90 days against all other episodes of peritonitis. Table S9. Proportion of missing values. Supplementary material is linked to the online version of the paper at www.kidney-international.org.

Acute kidney injury (AKI) is common and associated with poor renal outcomes,1 but the clinical course is not well understood.2, 3, 4 One reason for the increase in advanced chronic kidney disease (CKD) after AKI (vs. no AKI) is “nonrecovery,” that is, the occurrence of a step drop in estimated glomerular filtration rate (eGFR) during the AKI episode, which does not return to baseline once the episode has ended (Figure 1, pink dashed line). However, another path to advanced CKD after AKI may be a trajectory of subsequent renal decline after the episode has ended (Figure 1, red solid line). This distinction between subsequent progression and nonrecovery is crucial in clinical practice. At the time of a post-discharge clinical review, future subsequent renal decline is uncertain, whereas the extent of nonrecovery can already be observed. Moreover, because the trajectory of renal decline can vary from a gradual to a catastrophic loss of function,5 both hard outcomes (e.g., de novo long-term renal replacement therapy [RRT] or CKD stage 4) and intermediate outcomes (e.g., a 30% drop in kidney function)6 are important for clinicians and their patients to understand when planning care.Figure 1 Renal progression after acute kidney injury (AKI) caused by renal decline (red solid line) or nonrecovery (pink dashed line). A patient with AKI who has incomplete post-episode recovery has a high risk of developing advanced chronic kidney disease (CKD) even if subsequent renal decline is slow (pink dashed line). However, the risk of advanced CKD in a patient with AKI who has near-complete recovery depends on whether he or she experiences subsequent decline at a rapid trajectory (red solid line). In both cases at a post-AKI reassessment review (time d), renal recovery and post-episode kidney function are already observable, but the risk of subsequent decline is uncertain. The vertical black dashed line at time d represents the start of follow-up in this study. eGFR, estimated glomerular filtration rate.

line). In both cases at a post-AKI reassessment review (time d), renal recovery and post-episode kidney function are already observable, but the risk of subsequent decline is uncertain. The vertical black dashed line at time d represents the start of follow-up in this study. eGFR, estimated glomerular filtration rate. The Kidney Disease: Improving Global Outcomes AKI guidelines provide advice for post-AKI management based on expert opinion but without graded evidence.7 They state that people with AKI should be re-evaluated for resolution of kidney function and receive care based on CKD guidelines if they have developed CKD. However, this guidance does not apply to those who have had an episode of AKI and recovered to normal levels of kidney function after the episode. The relevance of post-episode recovery to baseline as a stratifying risk factor for AKI outcomes has been previously recognized in some studies but not in others.8 This is because previous studies have dichotomized recovery as being present or absent, with each using different cutoff values, but in reality, a spectrum of renal recovery exists.9, 10 Moreover, even if patients could be adequately grouped by recovery status, the use of a pre-episode baseline for determining outcomes would not separate the initial progression caused by incomplete recovery (however slight) from the subsequent progression caused by ongoing decline. The solution in clinical practice is that a clinician will wait to see where a post-episode eGFR finally settles (which becomes the “new baseline”) before evaluating risk and planning care from that point on. Therefore this is the approach we adopted in our analysis.

he subsequent progression caused by ongoing decline. The solution in clinical practice is that a clinician will wait to see where a post-episode eGFR finally settles (which becomes the “new baseline”) before evaluating risk and planning care from that point on. Therefore this is the approach we adopted in our analysis. In this large population study, we evaluated whether a completed AKI episode was still associated with subsequent renal decline, after allowing for a variable extent of initial renal recovery to baseline once the episode has ended. We isolated subsequent renal decline by using post-episode eGFR as the reference for subsequent renal outcomes. We hypothesized that more patients with AKI (vs. no AKI) would experience ongoing renal decline (a 30% eGFR drop), resulting in more patients with AKI having CKD stage 4.

baseline once the episode has ended. We isolated subsequent renal decline by using post-episode eGFR as the reference for subsequent renal outcomes. We hypothesized that more patients with AKI (vs. no AKI) would experience ongoing renal decline (a 30% eGFR drop), resulting in more patients with AKI having CKD stage 4. Results Population Of 17,630 patients with an index hospital admission in 2003, 14,651 patients were alive and not receiving long-term RRT 1 year after index hospital admission (Figure 2). This included 1966 with AKI and 12,685 without AKI. For the study of de novo CKD stage 4, an additional 545 patients who already had eGFR < 30 ml/min per 1.73 m2 at study entry were excluded. Thus 14,651 patients were available for the study of renal decline and 14,106 for the study of de novo CKD stage 4. The follow-up period of the study extended up to 10 years after the hospital admission, including 93,419 patient-years and a median of 8.6 years of follow-up.Figure 2 Flow diagram showing derivation of the cohort from the Grampian population. AKI, acute kidney injury; RRT, renal replacement therapy.

e novo CKD stage 4. The follow-up period of the study extended up to 10 years after the hospital admission, including 93,419 patient-years and a median of 8.6 years of follow-up.Figure 2 Flow diagram showing derivation of the cohort from the Grampian population. AKI, acute kidney injury; RRT, renal replacement therapy. Key findings From a reference point of the eGFR 1 year after a hospital admission episode, sustained subsequent 30% eGFR decline developed in 1660 of 14,651 patients (11.3%; 14.8% for AKI, 10.8% for no AKI), and sustained new CKD stage 4 occurred in 632 of 14,106 patients (4.5%; 7.1% for AKI, 4.1% for no AKI). Overall, patients were more likely to die than experience subsequent renal progression, whether defined as 30% renal decline (37.5% vs. 11.3%) or as new CKD stage 4 (38.8% vs. 4.5%).

14.8% for AKI, 10.8% for no AKI), and sustained new CKD stage 4 occurred in 632 of 14,106 patients (4.5%; 7.1% for AKI, 4.1% for no AKI). Overall, patients were more likely to die than experience subsequent renal progression, whether defined as 30% renal decline (37.5% vs. 11.3%) or as new CKD stage 4 (38.8% vs. 4.5%). Characteristics of patients with and without AKI Table 1 describes the characteristics of patients with and without AKI during the index hospital admission. Those with AKI (vs. no AKI) were older and were more frequently admitted on an emergency basis or to a critical care setting. They had more comorbidities. Although pre-hospital episode baseline eGFR was higher among those with AKI, post-discharge eGFR was lower, and a greater proportion of patients had a 30% decline in eGFR from the pre-hospital episode to the post-hospital episode (25.7% AKI, 2.3% no AKI) (i.e., nonrecovery). Nonrecovery was especially common among patients with AKI and post-episode eGFR < 60 ml/min per 1.73 m2 (42.3%). The proportion of patients with post-episode proteinuria was also 3-fold higher among those with AKI.Table 1 Baseline characteristics for patients with and without acute kidney injury

I) (i.e., nonrecovery). Nonrecovery was especially common among patients with AKI and post-episode eGFR < 60 ml/min per 1.73 m2 (42.3%). The proportion of patients with post-episode proteinuria was also 3-fold higher among those with AKI.Table 1 Baseline characteristics for patients with and without acute kidney injury Characteristic Overall AKI No AKI N % N % N % N 14,651 1966 (13.4% of cohort) 12,685 (86.6% of cohort) Age in years (median & IQR) 69 (54–78) 73 (63–81) 68 (53–78) Female 8317 (56.8) 1011 (51.4) 7306 (57.6) Residential care 433 (3.0) 111 (5.6) 322 (2.5) Deprived home locationa 1215 (8.3) 169 (8.6) 1046 (8.2) Rural home location 4014 (27.4) 551 (28.0) 3463 (27.3) Emergency hospital admission 8689 (59.3) 1580 (80.4) 7109 (56.0) Medical specialty admission 7203 (49.2) 1336 (68.0) 5867 (46.3) Critical care admission 1288 (8.8) 529 (26.9) 759 (6.0) Intensive care admission 428 (2.9) 276 (14.0) 152 (1.2) Length of hospital stay in days (median & IQR) 3 (1–9) 14 (7–31) 2 (1–7) Cancer 1011 (6.9) 201 (10.2) 810 (6.4) Cardiac failure 668 (4.6) 181 (9.2) 487 (3.8) Cerebrovascular disease 613 (4.2) 124 (6.3) 489 (3.9) Dementia 150 (1.0) 30 (1.5) 120 (0.9) Diabetes 917 (6.3) 255 (13.0) 662 (5.2) Liver disease 189 (1.3) 49 (2.5) 140 (1.1) Myocardial infarction 735 (5.0) 182 (9.3) 553 (4.4) Neurologic disease 76 (0.5) 20 (1.0) 56 (0.4) Peptic disease 304 (2.1) 66 (3.4) 238 (1.9) Peripheral vascular disease 487 (3.3) 140 (7.1) 347 (2.7) Pulmonary disease 836 (5.7) 199 (10.1) 637 (5.0) Rheumatic disease 312 (2.1) 68 (3.5) 244 (1.9) Baseline (pre-episode) eGFR (median & IQR) 66.8 (53.0–88.2) 75.3 (53.9–91.8) 65.8 (52.9–87.3) Post-episode eGFRb ≥60 9004 (61.5) 955 (48.6) 8049 (63.5) 45–59 3369 (23.0) 444 (22.6) 2925 (23.1) 30–44 1733 (11.8) 374 (19.0) 1359 (10.7) <30 545 (3.7) 193 (9.8) 352 (2.8) Intra-episode background change in eGFRc >30% rise 1135 (7.7) 67 (3.4) 1068 (8.4) 10%–30% rise 2069 (14.1) 120 (6.1) 1949 (15.4) No change 7654 (52.2) 517 (26.3) 7137 (56.3) 10%–30% fall 2990 (20.4) 757 (38.5) 2233 (17.6) >30% fall 803 (5.5) 505 (25.7) 298 (2.3) Post-episode proteinuriad Untested 13069 (89.2) 1550 (78.8) 11519 (90.8) Normal 753 (5.1) 136 (6.9) 617 (4.9) Abnormal 829 (5.7) 280 (14.2) 549 (4.3) AKI stage 0 12685 (86.6) n/a – 12685 (100.0) 1 1355 (9.2) 1355 (68.9) n/a – 2 410 (2.8) 410 (20.9) n/a – 3 201 (1.4) 201 (10.2) n/a – Prior AKI episodes 1358 (9.3) 356 (18.1) 1002 (7.9) AKI, acute kidney injury; eGFR, estimated glomerular filtration rate; IQR, interquartile range; n/

rmal 829 (5.7) 280 (14.2) 549 (4.3) AKI stage 0 12685 (86.6) n/a – 12685 (100.0) 1 1355 (9.2) 1355 (68.9) n/a – 2 410 (2.8) 410 (20.9) n/a – 3 201 (1.4) 201 (10.2) n/a – Prior AKI episodes 1358 (9.3) 356 (18.1) 1002 (7.9) AKI, acute kidney injury; eGFR, estimated glomerular filtration rate; IQR, interquartile range; n/ a, not applicable. a Most deprived quintile of the Scottish Index of Multiple Deprivation. b Post-episode eGFR was the most recent available eGFR at a time point 1 year after discharge from the index hospital admission. This was used as the reference for determining subsequent renal outcomes. c Intra-episode background change in eGFR was the change between pre-episode baseline and post-episode eGFR (that can occur irrespective of the presence of AKI). d Post-episode proteinuria was based on proteinuria measurements taken during or within 1 year of the index hospital admission.

b Post-episode eGFR was the most recent available eGFR at a time point 1 year after discharge from the index hospital admission. This was used as the reference for determining subsequent renal outcomes. c Intra-episode background change in eGFR was the change between pre-episode baseline and post-episode eGFR (that can occur irrespective of the presence of AKI). d Post-episode proteinuria was based on proteinuria measurements taken during or within 1 year of the index hospital admission. Crude proportions and cumulative rates of subsequent renal progression Figure 3 shows the crude proportions of people alive 1 year after hospital discharge who subsequently experienced renal decline (top plot), de novo CKD stage 4 (bottom plot), and death before progression during the study follow-up period until 10 years after discharge. After AKI (vs. no AKI), there was an excess of renal decline, or decline and death outcomes combined, among those with post-episode eGFR ≥ 60 ml/min per 1.73 m2. This excess was not present among those with a lower post-episode eGFR. The outcome of de novo CKD stage 4 was also more common among those with AKI, but uncommon among those with post-episode eGFR ≥ 60 ml/min per 1.73 m2 (1.5% AKI, 0.7% no AKI) and eGFR 45–59 ml/min per 1.73 m2 (6.1% AKI, 4.8% no AKI), compared with eGFR 30–44 ml/min per 1.73 m2 (22.5% AKI, 22.7% no AKI). Supplementary Figure S1 shows crude outcomes at 5 years after hospital discharge, with similar relationships, but with fewer deaths.Figure 3 Crude long-term renal outcomes after a hospital admission episode with or without acute kidney injury (AKI). CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate.

pplementary Figure S1 shows crude outcomes at 5 years after hospital discharge, with similar relationships, but with fewer deaths.Figure 3 Crude long-term renal outcomes after a hospital admission episode with or without acute kidney injury (AKI). CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate. Figure 4 shows the cumulative incidences of subsequent renal decline (a and b) and de novo CKD stage 4 (c and d) stratified by AKI and post-episode eGFR, accounting for the competing risk of death. Follow-up in all plots starts 1 year after discharge (i.e., study entry). Death without progression was more common than either progression outcome, and in the absence of a post-episode eGFR < 60 ml/min per 1.73 m2, de novo CKD stage 4 was rare.Figure 4 Cumulative incidences of subsequent renal progression (solid line) for those with (red) and without (blue) an acute kidney injury (AKI) admission in 2003, grouped by postdischarge estimated glomerular rate (eGFR) and accounting for the competing risk of death (dashed line). (a,b) Subsequent sustained 30% renal decline; (c,d) new chronic kidney disease (CKD) stage 4.

ssion (solid line) for those with (red) and without (blue) an acute kidney injury (AKI) admission in 2003, grouped by postdischarge estimated glomerular rate (eGFR) and accounting for the competing risk of death (dashed line). (a,b) Subsequent sustained 30% renal decline; (c,d) new chronic kidney disease (CKD) stage 4. Characteristics of patients with and without subsequent renal progression Table 2 describes the baseline characteristics of patients who progressed, died, or were alive without subsequent progression at the end of the study. Those with subsequent progression by either definition were older and had more comorbidities, including baseline renal impairment, as were those who died. A greater proportion of patients who experienced progression had diabetes as compared with those who died without progression, whereas the reverse was true for the other comorbidities.Table 2 Baseline characteristics for each progression group

comorbidities, including baseline renal impairment, as were those who died. A greater proportion of patients who experienced progression had diabetes as compared with those who died without progression, whereas the reverse was true for the other comorbidities.Table 2 Baseline characteristics for each progression group Characteristic Overall No renal decline or death New sustained 30% renal decline Dead before 30% renal decline No CKD stage 4 or death De novo CKD stage 4 Dead before de novo CKD stage 4 N % N % N % N % N % N % N % N 14,651 7497 (51.2% of cohort) 1660 (11.3% of cohort) 5494 (37.5% of cohort) 7996 (56.7% of cohort) 632 (4.5% of cohort) 5478 (38.8% of cohort) Age in years (median & IQR) 69 (54–78) 57 (42–69) 73 (65–79) 78 (71–84) 58 (43–69) 75 (69–82) 78 (70–84) Female 8317 (56.8) 4168 (55.6) 959 (57.8) 3190 (58.1) 4485 (56.1) 353 (55.9) 3150 (57.5) Residential care 433 (3.0) 36 (0.5) 25 (1.5) 372 (6.8) 36 (0.5) 11 (1.7) 351 (6.4) Deprived home locationa 1215 (8.3) 624 (8.3) 126 (7.6) 465 (8.5) 646 (8.1) 50 (7.9) 476 (8.7) Rural home location 4014 (27.4) 2137 (28.5) 447 (26.9) 1430 (26.0) 2274 (28.4) 158 (25.0) 1432 (26.1) Emergency hospital admission 8689 (59.3) 4027 (53.7) 939 (56.6) 3723 (67.8) 4281 (53.5) 371 (58.7) 3653 (66.7) Medical specialty admission 7203 (49.2) 3158 (42.1) 880 (53.0) 3165 (57.6) 3399 (42.5) 346 (54.7) 3139 (57.3) Critical care admission 1288 (8.8) 634 (8.5) 166 (10.0) 488 (8.9) 692 (8.7) 63 (10.0) 482 (8.8) Intensive care admission 428 (2.9) 217 (2.9) 60 (3.6) 151 (2.7) 236 (3.0) 20 (3.2) 160 (2.9) Cancer 1011 (6.9) 302 (4.0) 119 (7.2) 590 (10.7) 330 (4.1) 52 (8.2) 579 (10.6) Cardiac failure 668 (4.6) 113 (1.5) 100 (6.0) 455 (8.3) 120 (1.5) 53 (8.4) 408 (7.4) Cerebrovascular disease 613 (4.2) 113 (1.5) 85 (5.1) 415 (7.6) 126 (1.6) 36 (5.7) 402 (7.3) Dementia 150 (1.0) 6 (0.1) 8 (0.5) 136 (2.5) 6 (0.1) 5 (0.8) 128 (2.3) Diabetes 917 (6.3) 242 (3.2) 183 (11.0) 492 (9.0) 271 (3.4) 92 (14.6) 453 (8.3) Liver disease 189 (1.3) 72 (1.0) 25 (1.5) 92 (1.7) 75 (0.9) 9 (1.4) 97 (1.8) Myocardial infarction 735 (5.0) 226 (3.0) 96 (5.8) 413 (7.5) 241 (3.0) 60 (9.5) 379 (6.9) Neurologic disease 76 (0.5) 21 (0.3) 9 (0.5) 46 (0.8) 21 (0.3) 5 (0.8) 47 (0.9) Peptic disease 304 (2.1) 103 (1.4) 46 (2.8) 155 (2.8) 111 (1.4) 17 (2.7) 157 (2.9) Peripheral vascular disease 487 (3.3) 113 (1.5) 80 (4.8) 294 (5.4) 118 (1.5) 40 (6.3) 273 (5.0) Pulmonary disease 836 (5.7) 229 (3.1) 108 (6.5) 499 (9.1) 245 (3.1) 46 (7.3) 512 (9.3) Rheumatic disease 312 (2.1) 91 (1.2)

5 (0.8) 47 (0.9) Peptic disease 304 (2.1) 103 (1.4) 46 (2.8) 155 (2.8) 111 (1.4) 17 (2.7) 157 (2.9) Peripheral vascular disease 487 (3.3) 113 (1.5) 80 (4.8) 294 (5.4) 118 (1.5) 40 (6.3) 273 (5.0) Pulmonary disease 836 (5.7) 229 (3.1) 108 (6.5) 499 (9.1) 245 (3.1) 46 (7.3) 512 (9.3) Rheumatic disease 312 (2.1) 91 (1.2) 37 (2.2) 184 (3.3) 106 (1.3) 14 (2.2) 177 (3.2) Baseline (pre-episode) eGFR (median & IQR) 66.8 (53.0–88.2) 79.6 (61.8–98.7) 57.7 (46.2–70.5) 57.3 (45.8–73.7) 78.8 (61.6–97.9) 45.8 (37.4–55.4) 59.5 (49.0–75.4) There are 14,106 patients in de novo CKD stage 4 analysis because 545 patients had eGFR<30 at study entry. AKI, acute kidney injury; eGFR, estimated glomerular filtration rate; IQR, interquartile range. a Most deprived quintile of the Scottish Index of Multiple Deprivation. Table 3 describes renal measurements within each progression group. Those with subsequent progression by either definition had a greater proportion with a low post-episode eGFR, proteinuria, AKI, prior AKI, and any change in eGFR (whether a rise or fall) during the admission episode. The proportion with post-episode proteinuria was 2-fold higher in those with progression than in those who died without progression. The increased progression among those with AKI did not vary by AKI stage.Table 3 Renal measurements for each progression group

eGFR (whether a rise or fall) during the admission episode. The proportion with post-episode proteinuria was 2-fold higher in those with progression than in those who died without progression. The increased progression among those with AKI did not vary by AKI stage.Table 3 Renal measurements for each progression group Renal measurement Overall No renal decline or death New sustained 30% renal decline Dead before 30% renal decline No CKD stage 4 or death De novo CKD stage 4 Dead before de novo CKD stage 4 N % N % N % N % N % N % N % n 14,651 7497 (51.2% of cohort) 1660 (11.3% of cohort) 5494 (37.5% of cohort) 7996 (56.7% of cohort) 632 (4.5% of cohort) 5478 (38.8% of cohort) Post-episode eGFRa ≥60 9004 (61.5) 5854 (78.1) 825 (49.7) 2325 (42.3) 6282 (78.6) 73 (11.6) 2649 (48.4) 45–59 3369 (23.0) 1257 (16.8) 450 (27.1) 1662 (30.3) 1402 (17.5) 167 (26.4) 1800 (32.9) 30–44 1733 (11.8) 344 (4.6) 263 (15.8) 1126 (20.5) 312 (3.9) 392 (62.0) 1029 (18.8) <30 545 (3.7) 42 (0.6) 122 (7.3) 381 (6.9) Intra-episode background change in eGFRb >30% rise 1135 (7.7) 420 (5.6) 226 (13.6) 489 (8.9) 498 (6.2) 45 (7.1) 572 (10.4) 10%–30% rise% 2069 (14.1) 923 (12.3) 277 (16.7) 869 (15.8) 1014 (12.7) 102 (16.1) 909 (16.6) No change 7654 (52.2) 4659 (62.1) 703 (42.3) 2292 (41.7) 4933 (61.7) 243 (38.4) 2344 (42.8) 10%–30% fall 2990 (20.4) 1282 (17.1) 362 (21.8) 1346 (24.5) 1346 (16.8) 179 (28.3) 1302 (23.8) >30% fall 803 (5.5) 213 (2.8) 92 (5.5) 498 (9.1) 205 (2.6) 63 (10.0) 351 (6.4)

77 (16.7) 869 (15.8) 1014 (12.7) 102 (16.1) 909 (16.6) No change 7654 (52.2) 4659 (62.1) 703 (42.3) 2292 (41.7) 4933 (61.7) 243 (38.4) 2344 (42.8) 10%–30% fall 2990 (20.4) 1282 (17.1) 362 (21.8) 1346 (24.5) 1346 (16.8) 179 (28.3) 1302 (23.8) >30% fall 803 (5.5) 213 (2.8) 92 (5.5) 498 (9.1) 205 (2.6) 63 (10.0) 351 (6.4) Post-episode proteinuriac Untested 13069 (89.2) 6985 (93.2) 1302 (78.4) 4782 (87.0) 7416 (92.7) 469 (74.2) 4809 (87.8) Normal 753 (5.1) 340 (4.5) 122 (7.3) 291 (5.3) 386 (4.8) 51 (8.1) 291 (5.3) Abnormal 829 (5.7) 172 (2.3) 236 (14.2) 421 (7.7) 194 (2.4) 112 (17.7) 378 (6.9) AKI stage 0 12685 (86.6) 6912 (92.2) 1369 (82.5) 4404 (80.2) 7359 (92.0) 507 (80.2) 4467 (81.5) 1 1355 (9.2) 387 (5.2) 201 (12.1) 767 (14.0) 432 (5.4) 86 (13.6) 711 (13.0) 2 410 (2.8) 124 (1.7) 57 (3.4) 229 (4.2) 128 (1.6) 26 (4.1) 219 (4.0) 3 201 (1.4) 74 (1.0) 33 (2.0) 94 (1.7) 77 (1.0) 13 (2.1) 81 (1.5) Prior AKI episodes 1358 (9.3) 355 (4.7) 230 (13.9) 773 (14.1) 379 (4.7) 110 (17.4) 705 (12.9) There are 14,106 patients in de novo CKD stage 4 analysis because 545 patients had eGFR<30 at study entry. AKI, acute kidney injury; CI, confidence interval; eGFR, estimated glomerular filtration rate. a Post-episode eGFR was the most recent available eGFR at a time point 1 year after discharge from the index hospital admission. This was used as the reference for determining subsequent renal outcomes. b Intra-episode background change in eGFR was the change between pre-episode baseline and post-episode eGFR (that can occur irrespective of the presence of AKI).

a Post-episode eGFR was the most recent available eGFR at a time point 1 year after discharge from the index hospital admission. This was used as the reference for determining subsequent renal outcomes. b Intra-episode background change in eGFR was the change between pre-episode baseline and post-episode eGFR (that can occur irrespective of the presence of AKI). c Post-episode proteinuria was based on proteinuria measurements taken during or within 1 year of the index hospital admission. Independent association between AKI and subsequent renal progression Table 4 describes the multivariable-adjusted relationship between AKI and renal decline stratified by post-episode eGFR (interaction P < 0.001). The plain text indicates relative risks compared with a reference group with no AKI and eGFR ≥ 60 ml/min per 1.73 m2. Bold text indicates AKI versus no AKI at each level of post-episode eGFR. The relative risk of renal decline for AKI (vs. no AKI) was greater in those with otherwise normal function than in those with lower post-episode eGFR: hazard ratio (HR) (for AKI vs. no AKI) 2.29 (1.88–2.41), 1.50 (1.13–2.00), 0.84 (0.68–1.32) and 0.95 (0.64–1.41) for post-episode eGFR ≥ 60, 45–59, 30–45, and <30 ml/min per 1.73 m2, respectively. Table 5 shows a similar relationship between AKI and de novo CKD stage 4, but the magnitude of the variation with post-episode eGFR was greater.Table 4 Relative risk of subsequent sustained 30% renal decline after acute kidney injury

1) for post-episode eGFR ≥ 60, 45–59, 30–45, and <30 ml/min per 1.73 m2, respectively. Table 5 shows a similar relationship between AKI and de novo CKD stage 4, but the magnitude of the variation with post-episode eGFR was greater.Table 4 Relative risk of subsequent sustained 30% renal decline after acute kidney injury Post-hospital episode eGFR AKI or no AKI N Cause-specific renal decline; age-sex adjusted (HR, 95% CI) Cause-specific renal decline; fully adjusted (HR, 95% CI) Competing event of death without renal decline; age-sex adjusted (HR, 95% CI) Competing event of death without renal decline; fully adjusted (HR, 95% CI) eGFR ≥ 60 No AKI (reference) 8049 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) AKI 955 2.01 (1.68-2.41) 2.29 (1.88-2.78) 1.77 (1.59-1.97) 1.51 (1.35-1.70) AKI vs. no AKI 2.01 (1.68-2.41) 2.29 (1.88-2.78) 1.77 (1.59-1.97) 1.51 (1.35-1.70) eGFR 45–59 No AKI 2925 1.14 (1.00-1.30) 1.22 (1.07-1.40) 0.98 (0.91-1.05) 0.97 (0.90-1.04) AKI 444 1.51 (1.16-1.97) 1.84 (1.38-2.45) 1.52 (1.33-1.73) 1.21 (1.05-1.39) AKI vs. no AKI 1.32 (1.01-1.73) 1.50 (1.13-2.00) 1.56 (1.37-1.77) 1.25 (1.09-1.44) eGFR 30–44 No AKI 1359 1.63 (1.38-1.93) 1.71 (1.44-2.02) 1.29 (1.18-1.40) 1.18 (1.08-1.29) AKI 374 1.38 (1.01-1.87) 1.61 (1.16-2.24) 1.64 (1.43-1.87) 1.24 (1.07-1.44) AKI vs. no AKI 0.84 (0.61-1.16) 0.94 (0.68-1.32) 1.27 (1.11-1.46) 1.05 (0.91-1.22)

eGFR 45–59 No AKI 2925 1.14 (1.00-1.30) 1.22 (1.07-1.40) 0.98 (0.91-1.05) 0.97 (0.90-1.04) AKI 444 1.51 (1.16-1.97) 1.84 (1.38-2.45) 1.52 (1.33-1.73) 1.21 (1.05-1.39) AKI vs. no AKI 1.32 (1.01-1.73) 1.50 (1.13-2.00) 1.56 (1.37-1.77) 1.25 (1.09-1.44) eGFR 30–44 No AKI 1359 1.63 (1.38-1.93) 1.71 (1.44-2.02) 1.29 (1.18-1.40) 1.18 (1.08-1.29) AKI 374 1.38 (1.01-1.87) 1.61 (1.16-2.24) 1.64 (1.43-1.87) 1.24 (1.07-1.44) AKI vs. no AKI 0.84 (0.61-1.16) 0.94 (0.68-1.32) 1.27 (1.11-1.46) 1.05 (0.91-1.22) eGFR < 30 No AKI 352 3.78 (2.98-4.80) 3.81 (2.97-4.88) 1.87 (1.62-2.15) 1.65 (1.42-1.90) AKI 193 3.36 (2.40-4.69) 3.63 (2.52-5.22) 2.20 (1.85-2.63) 1.57 (1.29-1.90) AKI vs. no AKI 0.89 (0.69-1.30) 0.95 (0.64-1.41) 1.18 (0.96-1.45) 0.95 (0.77-1.18) Multivariable cause-specific Cox regression with interaction terms between AKI and baseline eGFR. Adjusted estimates are reported with reference to no AKI and eGFR > 60 (plain type), and for AKI versus no AKI within each eGFR group calculated using the interaction terms (bold type). The “fully adjusted” model included adjustment for social, demographic, admission circumstances, each separate nonrenal Charlson comorbidity, and renal measurements as described in the Covariates section. AKI, acute kidney injury; CI, confidence interval; eGFR, estimated glomerular filtration rate (ml/min per 1.73 m2); HR, hazard ratio. Table 5 Relative risk of de novo CKD stage 4 after acute kidney injury

eGFR < 30 No AKI 352 3.78 (2.98-4.80) 3.81 (2.97-4.88) 1.87 (1.62-2.15) 1.65 (1.42-1.90) AKI 193 3.36 (2.40-4.69) 3.63 (2.52-5.22) 2.20 (1.85-2.63) 1.57 (1.29-1.90) AKI vs. no AKI 0.89 (0.69-1.30) 0.95 (0.64-1.41) 1.18 (0.96-1.45) 0.95 (0.77-1.18) Multivariable cause-specific Cox regression with interaction terms between AKI and baseline eGFR. Adjusted estimates are reported with reference to no AKI and eGFR > 60 (plain type), and for AKI versus no AKI within each eGFR group calculated using the interaction terms (bold type). The “fully adjusted” model included adjustment for social, demographic, admission circumstances, each separate nonrenal Charlson comorbidity, and renal measurements as described in the Covariates section. AKI, acute kidney injury; CI, confidence interval; eGFR, estimated glomerular filtration rate (ml/min per 1.73 m2); HR, hazard ratio. Table 5 Relative risk of de novo CKD stage 4 after acute kidney injury Post-hospital episode eGFR AKI or no AKI N Cause-specific new CKD 4; age-sex adjusted (HR, 95% CI) Cause-specific new CKD 4; fully adjusted (HR, 95% CI) Competing event of death without renal decline; age-sex adjusted (HR, 95% CI) Competing event of death without renal decline; fully adjusted (HR, 95% CI) eGFR ≥ 60 No AKI (reference) 8049 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) AKI 955 2.36 (1.31-4.24) 2.55 (1.41-4.64) 1.70 (1.54-1.88) 1.47 (1.32-1.63) AKI vs. no AKI 2.36 (1.31-4.24) 2.55 (1.41-4.64) 1.70 (1.54-1.88) 1.47 (1.32-1.63)

of death without renal decline; fully adjusted (HR, 95% CI) eGFR ≥ 60 No AKI (reference) 8049 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) AKI 955 2.36 (1.31-4.24) 2.55 (1.41-4.64) 1.70 (1.54-1.88) 1.47 (1.32-1.63) AKI vs. no AKI 2.36 (1.31-4.24) 2.55 (1.41-4.64) 1.70 (1.54-1.88) 1.47 (1.32-1.63) eGFR 45–59 No AKI 2925 7.09 (5.07-9.90) 7.18 (5.14-10.02) 0.93 (0.87-1.00) 0.94 (0.87-1.00) AKI 444 10.96 (6.82-17.62) 12.60 (7.63-20.81) 1.46 (1.29-1.65) 1.17 (1.02-1.34) AKI vs. no AKI 1.55 (1.02-2.34) 1.75 (1.13-2.71) 1.56 (1.38-1.77) 1.25 (1.09-1.43) eGFR 30–44 No AKI 1359 48.54 (35.22-66.91) 50.21 (36.31-69.43) 1.18 (1.08-1.29) 1.10 (1.00-1.20) AKI 374 52.40 (36.36-75.52) 61.17 (40.73-91.87) 1.47 (1.28-1.69) 1.12 (0.95-1.31) AKI vs. no AKI 1.08 (0.85-1.38) 1.22 (0.92-1.61) 1.25 (1.08-1.45) 1.02 (0.87-1.19) Multivariable cause-specific Cox regression with interaction terms between AKI and baseline eGFR. Adjusted estimates are reported with reference to no AKI and eGFR > 60 (plain type) and for AKI versus no AKI within each eGFR group calculated using the interaction terms (bold type). The “fully adjusted” model included adjustment for social, demographic, admission circumstances, each separate nonrenal Charlson comorbidity, and renal measurements as described in the Covariates section. AKI, acute kidney injury; CI, confidence interval; eGFR, estimated glomerular filtration rate (ml/min per 1.73 m2); HR, hazard ratio.

eGFR 30–44 No AKI 1359 48.54 (35.22-66.91) 50.21 (36.31-69.43) 1.18 (1.08-1.29) 1.10 (1.00-1.20) AKI 374 52.40 (36.36-75.52) 61.17 (40.73-91.87) 1.47 (1.28-1.69) 1.12 (0.95-1.31) AKI vs. no AKI 1.08 (0.85-1.38) 1.22 (0.92-1.61) 1.25 (1.08-1.45) 1.02 (0.87-1.19) Multivariable cause-specific Cox regression with interaction terms between AKI and baseline eGFR. Adjusted estimates are reported with reference to no AKI and eGFR > 60 (plain type) and for AKI versus no AKI within each eGFR group calculated using the interaction terms (bold type). The “fully adjusted” model included adjustment for social, demographic, admission circumstances, each separate nonrenal Charlson comorbidity, and renal measurements as described in the Covariates section. AKI, acute kidney injury; CI, confidence interval; eGFR, estimated glomerular filtration rate (ml/min per 1.73 m2); HR, hazard ratio. Sensitivity and subgroup analyses Supplementary Table 2 shows the interactions tested in further analyses for the renal decline endpoint. The role of AKI (vs. no AKI) was modified by age (interaction P value = 0.01) with greater relative risk in the young than in the elderly, but significant in both groups. It was not modified by gender (P value = 0.86), diabetes (P value = 0.90), cancer (P value = 0.78), or cardiac failure (P value = 0.23). A statistically significant time interaction (i.e., nonproportionality) was also present (P value = 0.04), and therefore in a sensitivity analysis we split follow-up at 5 years after discharge. As reported in Supplementary Table S2, the HR for AKI up to 5 years among those alive at 1 year (1.69, 1.40–2.03) was greater than the HR for AKI up to 10 years for those alive and at risk 5 years after discharge (1.45, 1.12–1.88). As reported in Supplementary Table S3, we also repeated the analysis for AKI versus no AKI excluding those with post-episode proteinuria, with similar pattern results. We also repeated the analysis excluding those who only had a discharge eGFR available for post-episode eGFR, and the findings were unchanged. Further sensitivity analyses included additional adjustment for acute hospital diagnoses (Supplementary Table S4) and reanalysis with the use of a Fine and Gray competing risks model (Supplementary Table S5). Both showed similar results.

a discharge eGFR available for post-episode eGFR, and the findings were unchanged. Further sensitivity analyses included additional adjustment for acute hospital diagnoses (Supplementary Table S4) and reanalysis with the use of a Fine and Gray competing risks model (Supplementary Table S5). Both showed similar results. Discussion This large analysis of hospital survivors after AKI isolates the risk of long-term subsequent progression of kidney disease from progression that has already arisen because of an initial step drop in kidney function (incomplete recovery). When this novel approach was used, AKI during a hospital admission was associated with increased subsequent renal progression irrespective of how progression was defined, irrespective of proteinuria or AKI severity, and even if post-episode kidney function was normal. In one of the longest observation periods of any renal progression study, the excess risk after AKI diminished over time but persisted throughout the 10 years of the study.

sion irrespective of how progression was defined, irrespective of proteinuria or AKI severity, and even if post-episode kidney function was normal. In one of the longest observation periods of any renal progression study, the excess risk after AKI diminished over time but persisted throughout the 10 years of the study. Previous studies have also shown an association between AKI and long-term CKD,1, 2, 3, 4 but our analysis extends the current understanding by providing greater detail and precision. First, to the best of our knowledge, no previous studies of AKI prognosis have presented renal progression endpoints both defined in terms of an intermediate outcome (30% renal decline) and a hard outcome (de novo CKD stage 4 or long-term RRT). This analysis shows that no matter how renal progression is defined, AKI is associated with poorer long-term outcomes, although we recognize that de novo CKD stage 4 was most common among those with AKI who already had a low eGFR. Second, previous analyses have used an all-or-nothing “renal recovery” dichotomy as a risk factor for prognostic study in AKI.8, 9 However, grouping patients in this way does not separate the initial renal decline (i.e., nonrecovery to baseline that is already observable after the episode) from subsequent renal decline (the uncertainty of what happens next), nor does it account for intra-episode changes in eGFR that can occur irrespective of AKI.2 Our analysis provides the following important detail: whereas 25.7% of people with AKI experienced a 30% decline between their pre-episode and post-discharge eGFR values from the post-discharge eGFR value, 14.8% of people with AKI experienced subsequent renal decline. This represented a relative risk of up to 2.5-fold from AKI (vs. no AKI), which varied depending on the level of post-episode eGFR.

th AKI experienced a 30% decline between their pre-episode and post-discharge eGFR values from the post-discharge eGFR value, 14.8% of people with AKI experienced subsequent renal decline. This represented a relative risk of up to 2.5-fold from AKI (vs. no AKI), which varied depending on the level of post-episode eGFR. The interaction between AKI and post-episode eGFR is a unique finding of this analysis. The increased relative risk from AKI (vs. no AKI) was greatest (more than 2-fold) among those who experienced recovery to normal levels (eGFR ≥ 60), even when those with post-episode proteinuria were excluded. This finding is in contrast to the Kidney Disease: Improving Global Outcomes AKI guidelines, which recommend prioritization of those with de novo CKD at a 90- day clinical reassessment.7 However, regarding the lack of elevated relative risk from AKI among those with a low post-episode eGFR, we note for the reader that subsequent decline represents only one mode of progression. Indeed, a step 30% eGFR drop during admission was particularly common among those with AKI and low post-episode eGFR, indicating that nonrecovery with low subsequent progression was common. The interaction between AKI and eGFR on progression also complements previous studies, which have demonstrated a similar interaction between AKI and eGFR on mortality.11, 12 Finally, a complementary explanation for the poorer outcomes among those with a normal post-episode eGFR could be that AKI indicates a “failed stress test” unmasking subclinical renal disease. This is biologically plausible because some people with ostensibly “normal” kidney function nevertheless lack functional glomerular filtration reserve.13 The important clinical implication is that such patients will be more vulnerable to future decline, even though currently available metrics of renal function remain “normal.” In this situation a recent AKI episode yields important prognostic information that may otherwise be overlooked if not clearly communicated at any transitions in care.

implication is that such patients will be more vulnerable to future decline, even though currently available metrics of renal function remain “normal.” In this situation a recent AKI episode yields important prognostic information that may otherwise be overlooked if not clearly communicated at any transitions in care. A strength of this analysis is the use of routinely collected data within a large regional population with long follow-up. This “real-life” situation maximizes the generalizability of our findings. Another strength is the distinction of 2 perspectives on progression by using both intermediate (30% renal decline) and hard (CKD stage 4) endpoints. Moreover, by defining the study entry eGFR at a post-discharge time point, our analysis provides a precise separation of incomplete initial recovery and subsequent renal decline that has not previously been studied. A limitation is that data collection was not protocolized. This means there may be ascertainment biases in our determination of renal progression, but these will have been partially offset by our requirement for the outcomes to be sustained for at least 90 days. We also conducted a sensitivity analysis excluding patients with only a discharge creatinine value available in the first post-episode year, with unchanged results. Similarly, quantified proteinuria was frequently not tested in the cohort. In most cases this will have been because of a low level of suspicion for proteinuria or a negative urinalysis, but some cases of proteinuria may have been missed. We also recognize that post-AKI proteinuria will frequently reflect underlying renal disease rather than a new derangement consequent to AKI. Finally, as with all observational studies, there will be residual confounding, which means that the long-term role of AKI may have been overestimated. Our analyses included adjustment for confounders including social, demographic, and renal measurements; comorbid factors; and acute diagnoses in a sensitivity analysis. However, we recognize that hospital episode International Classification of Diseases, 10th Revision (ICD-10) codes for comorbidities may lack granularity. We note that prospective recruitment and protocolized follow-up could overcome some of these issues but would be at the expense of “real-life” generalizability.

sis. However, we recognize that hospital episode International Classification of Diseases, 10th Revision (ICD-10) codes for comorbidities may lack granularity. We note that prospective recruitment and protocolized follow-up could overcome some of these issues but would be at the expense of “real-life” generalizability. Overall, this study shows that no matter how severe an AKI episode is, irrespective of proteinuria and even if post-episode function is apparently preserved, an episode of AKI is associated with increased subsequent renal decline that persists for up to 10 years. Recommendations for follow-up should therefore be formulated carefully to avoid false reassurance when eGFR after AKI appears to have returned to the normal range.

if post-episode function is apparently preserved, an episode of AKI is associated with increased subsequent renal decline that persists for up to 10 years. Recommendations for follow-up should therefore be formulated carefully to avoid false reassurance when eGFR after AKI appears to have returned to the normal range. Methods Population The Grampian Laboratory Outcomes Morbidity and Mortality Study12, 14, 15 is a population cohort achieved by linking national and regional data sources for a single United Kingdom health authority (Grampian resident adult population 438,332).16 Nonresidents have been excluded. Because data linkage avoids the need for active recruitment, this virtual cohort is not affected by the selection biases inherent in patient enrollment. The region includes a large tertiary center (∼1000 beds) and 2 outlying hospitals (combined ∼500 beds). All biochemistry testing was provided by a single biochemistry service (1999–2013), regardless of clinical location (inpatient, outpatient, community). This minimizes any loss of baseline and follow-up data, which are vital in renal clinical research.17 Information on mortality, hospital admission episodes, morbidity events, and long-term RRT are available by linkage to hospital episode data, the local renal information management system, and the Scottish Renal Registry.

s minimizes any loss of baseline and follow-up data, which are vital in renal clinical research.17 Information on mortality, hospital admission episodes, morbidity events, and long-term RRT are available by linkage to hospital episode data, the local renal information management system, and the Scottish Renal Registry. Study entry This study includes patients from the Grampian Laboratory Outcomes Morbidity and Mortality Study who survived to the completion of a hospital admission episode in 2003.12 We chose 1 year after discharge as the time point when the index episode was considered complete and subjects “entered” the study. The most recent eGFR available at study entry (1 year after an episode) was taken as the reference value for determining all subsequent renal outcomes. For the rest of this article, we refer to this as the “post-episode eGFR.” Those who were dead or receiving long-term RRT at study entry were excluded. We selected “1 year after episode” to optimize opportunity for renal recovery and a completeness of post-episode testing. This testing was performed a median 264 days after discharge. As shown in Supplementary Table S1, had study entry been earlier, post-episode tests would have been unavailable in 39% of patients with AKI and 55% of patients who did not have AKI. For the minority of patients who still did not have a test at 1 year, the eGFR at discharge was used as the reference. These patients were also removed in a sensitivity analysis to ensure this did not affect the results.

s would have been unavailable in 39% of patients with AKI and 55% of patients who did not have AKI. For the minority of patients who still did not have a test at 1 year, the eGFR at discharge was used as the reference. These patients were also removed in a sensitivity analysis to ensure this did not affect the results. Exposure AKI during the index hospital admission was identified and staged from 1 to 3 by using AKI criteria based on the Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group guidelines.7 Baseline creatinine values for identifying AKI were determined by a modified “e-alert” algorithm with a hierarchy of criteria for creatinine changes from the previous 48 hours and 7, 90, and 365 days as stated elsewhere.12 Briefly, this definition for AKI involves 1 of 3 criteria: serum creatinine level ≥ 1.5 times higher than the median of all creatinine values 8 to 90 days earlier, or 91 to 365 days earlier if no tests were done between 8 and 90 days; serum creatinine level ≥ 1.5 times higher than the lowest creatinine value within 7 days; or serum creatinine level > 26 μmol/l higher than the lowest creatinine value within 48 hours. We have previously described this definition in more detail for studying the prognosis of AKI in hospitals.12

ne between 8 and 90 days; serum creatinine level ≥ 1.5 times higher than the lowest creatinine value within 7 days; or serum creatinine level > 26 μmol/l higher than the lowest creatinine value within 48 hours. We have previously described this definition in more detail for studying the prognosis of AKI in hospitals.12 Outcomes Two subsequent renal progression outcomes are defined for this analysis: sustained 30% eGFR decline (in all 14,651 patients) and de novo CKD stage 4 (in 14,106 patients with post-episode eGFR ≥ 30 ml/min per 1.73 m2). For either outcome, progression was sustained if it lasted at least 90 days or if the patient started long-term RRT. Notably, as illustrated in Supplementary Table S1, if a pre-episode eGFR (instead of the post-episode eGFR) had been used as reference value for subsequent renal outcomes, 25.7% of patients with AKI would already have had a 30% eGFR decline on the first day of the study (i.e., nonrecovery from AKI misclassified as post-episode renal decline). Follow-up Follow-up began at study entry, 1 year after discharge from the index hospital episode. It lasted until the date of renal progression (respectively sustained 30% renal decline or new CKD stage 4 for each progression subanalysis), death, or the end of study follow-up in July 2013.

Outcomes Two subsequent renal progression outcomes are defined for this analysis: sustained 30% eGFR decline (in all 14,651 patients) and de novo CKD stage 4 (in 14,106 patients with post-episode eGFR ≥ 30 ml/min per 1.73 m2). For either outcome, progression was sustained if it lasted at least 90 days or if the patient started long-term RRT. Notably, as illustrated in Supplementary Table S1, if a pre-episode eGFR (instead of the post-episode eGFR) had been used as reference value for subsequent renal outcomes, 25.7% of patients with AKI would already have had a 30% eGFR decline on the first day of the study (i.e., nonrecovery from AKI misclassified as post-episode renal decline). Follow-up Follow-up began at study entry, 1 year after discharge from the index hospital episode. It lasted until the date of renal progression (respectively sustained 30% renal decline or new CKD stage 4 for each progression subanalysis), death, or the end of study follow-up in July 2013. Covariates We adjusted for all other observable renal measurements at study entry to isolate the post-episode role of an AKI episode. These included the most recent post-episode eGFR available at study entry, the “intra-episode background change in eGFR,” the presence of any other prior AKI episodes in the previous 3 years, and quantified proteinuria during the 1-year post-episode period.

tudy entry to isolate the post-episode role of an AKI episode. These included the most recent post-episode eGFR available at study entry, the “intra-episode background change in eGFR,” the presence of any other prior AKI episodes in the previous 3 years, and quantified proteinuria during the 1-year post-episode period. We defined “intra-episode background change in eGFR” as the difference between the pre-episode baseline eGFR and the post-episode eGFR (Figure 1). The reason for adjusting for this change in addition to all other renal measurements is that background rise or fall in eGFR can occur during any admission episode, irrespective of AKI (e.g., because of loss of muscle mass or slow background renal progression). All eGFR measures were reported using the Chronic Kidney Disease Epidemiology Collaboration creatinine equation.18 We grouped post-episode eGFR in 4 categories: ≥60, 45–59, 30–44, and <30 ml/min per 1.73 m2. To allow for nonlinearity, we also grouped intra-episode eGFR change in 5 categories: >30% rise, 10%–30% rise, no change, 10%–30% decline, and >30% decline.19 Proteinuria measures recorded as “abnormal” were albumin creatinine ratio ≥ 3 mg/mmol or protein creatinine ratio ≥ 15 mg/mmol.20

44, and <30 ml/min per 1.73 m2. To allow for nonlinearity, we also grouped intra-episode eGFR change in 5 categories: >30% rise, 10%–30% rise, no change, 10%–30% decline, and >30% decline.19 Proteinuria measures recorded as “abnormal” were albumin creatinine ratio ≥ 3 mg/mmol or protein creatinine ratio ≥ 15 mg/mmol.20 Nonrenal comorbidities were determined using ICD-10 codes for Charlson comorbidities from the 5 years before admission as previously described and validated.21 Social and demographic measures included age, sex, whether the patient was in residential care, and home address–based measures of deprivation (most deprived quintile versus the other 4 quintiles of the Scottish Index of Multiple Deprivation), and rural location (settlement of less than 3000 people).22 Metrics of admission circumstances were whether the index hospital admission was an emergency or elective and whether the index admission included a stay in a medical (vs. surgical) ward or a critical care or intensive care unit. Statistical analyses For both renal progression outcomes (first for 30% eGFR decline, then for de novo CKD stage 4), we reported patient characteristics grouped as follows: those who were alive without renal progression at the end of follow-up, those who experienced renal progression, and those who died without experiencing renal progression (a competing risk).

sion outcomes (first for 30% eGFR decline, then for de novo CKD stage 4), we reported patient characteristics grouped as follows: those who were alive without renal progression at the end of follow-up, those who experienced renal progression, and those who died without experiencing renal progression (a competing risk). We plotted crude outcomes both for 30% renal decline and de novo CKD stage 4 during follow-up by AKI and eGFR category. We also estimated and plotted the cumulative incidence of renal progression, accounting for the competing risk of death using the Stata command “stcompet” with Stata SE 13.0 software (StataCorp LLC, College Station, TX) as described elsewhere.23 We estimated the independent association of AKI (vs. no AKI) with long-term renal progression in multivariable analysis using both cause-specific Cox models for progression and death without progression. The fully adjusted model included adjustment for social, demographic, admission circumstances, each separate nonrenal Charlson comorbidity, and renal measurements as described previously in the Covariates section. Because an interaction exists between eGFR (in categories ≥60, 45–59, 30–45, <30) and AKI (vs. no AKI) on mortality,12, 24 we included an eGFR AKI interaction term for renal progression in this analysis. All analyses were conducted in Stata SE 13.0 software (StataCorp LLC).

scribed previously in the Covariates section. Because an interaction exists between eGFR (in categories ≥60, 45–59, 30–45, <30) and AKI (vs. no AKI) on mortality,12, 24 we included an eGFR AKI interaction term for renal progression in this analysis. All analyses were conducted in Stata SE 13.0 software (StataCorp LLC). Subgroup and sensitivity analyses We tested for interactions of AKI and progression with old age (≥70 years), sex, cancer, and diabetes. We also tested an interaction with follow-up time (per year of follow-up completed) to assess the proportionality assumption for AKI. In the main analysis, we presented cause-specific HRs for individuals alive and at risk, which is the preferred approach for estimating the effect of covariates on outcomes as HRs.25 However, in sensitivity analyses, we also estimated “subdistribution HRs” according to the Fine and Gray model (in which those who died without progression remain in the “risk set”).25, 26 Because it is not possible to distinguish whether acute diagnoses recorded during the index admission were a cause or consequence of AKI, we did not adjust for acute diagnoses in the primary multivariable analysis, but in a sensitivity analysis, we compared the findings after acute hospital diagnoses (extracted from ICD-10 codes) were added to the models.12, 27 Because the most recent available post-episode eGFR was at discharge for 20% of the patients, we also repeated the analysis excluding these patients. Finally, because it is possible for patients to have a normal eGFR (≥ 60 ml/min per 1.73 m2) but still have proteinuric evidence of nonrecovery,20 we repeated the analysis excluding those with abnormal proteinuria measurements.

t discharge for 20% of the patients, we also repeated the analysis excluding these patients. Finally, because it is possible for patients to have a normal eGFR (≥ 60 ml/min per 1.73 m2) but still have proteinuric evidence of nonrecovery,20 we repeated the analysis excluding those with abnormal proteinuria measurements. Disclosure All the authors declared no competing interests. Supplementary Material Figure S1 Crude long-term renal outcomes up to 5 years after a hospital admission episode with or without acute kidney injury (AKI). Table S1 Post-episode estimated glomerular filtration rate in the first year after discharge from the index admission episode. Table S2 Analyses of interactions. Table S3 Relative risk of renal progression after excluding those with post-episode proteinuria. Table S4 Relative risk of subsequent sustained 30% renal decline and new chronic kidney disease stage 4 after acute kidney injury with additional adjustment for acute hospital diagnoses. Table S5 Relative risk of subsequent sustained 30% renal decline and new chronic kidney disease stage 4 after acute kidney injury using Fine and Gray model.

Table S4 Relative risk of subsequent sustained 30% renal decline and new chronic kidney disease stage 4 after acute kidney injury with additional adjustment for acute hospital diagnoses. Table S5 Relative risk of subsequent sustained 30% renal decline and new chronic kidney disease stage 4 after acute kidney injury using Fine and Gray model. Acknowledgments We acknowledge the data management support of Grampian Data Safe Haven (DaSH) and the associated financial support of NHS Research Scotland, through NHS Grampian investment in the Grampian DaSH. SS is supported by a Clinical Research Training Fellowship from the Wellcome Trust (Ref 102729/Z/13/Z). We also acknowledge the support from The Farr Institute of Health Informatics Research. The Farr Institute is supported by a 10-funder consortium: Arthritis Research UK, the British Heart Foundation, Cancer Research UK, the Economic and Social Research Council, the Engineering and Physical Sciences Research Council, the Medical Research Council, the National Institute of Health Research, the National Institute for Social Care and Health Research (Welsh Assembly Government), the Chief Scientist Office (Scottish Government Health Directorates), and the Wellcome Trust (MRC Grant Nos: Scotland MR/K007017/1). The funders of this study had no role in study design; collection, analysis, and interpretation of data; writing the report; or the decision to submit the report for publication. see commentary on page 288 Figure S1. Crude long-term renal outcomes up to 5 years after a hospital admission episode with or without acute kidney injury (AKI).

Acknowledgments We acknowledge the data management support of Grampian Data Safe Haven (DaSH) and the associated financial support of NHS Research Scotland, through NHS Grampian investment in the Grampian DaSH. SS is supported by a Clinical Research Training Fellowship from the Wellcome Trust (Ref 102729/Z/13/Z). We also acknowledge the support from The Farr Institute of Health Informatics Research. The Farr Institute is supported by a 10-funder consortium: Arthritis Research UK, the British Heart Foundation, Cancer Research UK, the Economic and Social Research Council, the Engineering and Physical Sciences Research Council, the Medical Research Council, the National Institute of Health Research, the National Institute for Social Care and Health Research (Welsh Assembly Government), the Chief Scientist Office (Scottish Government Health Directorates), and the Wellcome Trust (MRC Grant Nos: Scotland MR/K007017/1). The funders of this study had no role in study design; collection, analysis, and interpretation of data; writing the report; or the decision to submit the report for publication. see commentary on page 288 Figure S1. Crude long-term renal outcomes up to 5 years after a hospital admission episode with or without acute kidney injury (AKI). Table S1. Post-episode estimated glomerular filtration rate in the first year after discharge from the index admission episode. Table S2. Analyses of interactions. Table S3. Relative risk of renal progression after excluding those with post-episode proteinuria.

Figure S1. Crude long-term renal outcomes up to 5 years after a hospital admission episode with or without acute kidney injury (AKI). Table S1. Post-episode estimated glomerular filtration rate in the first year after discharge from the index admission episode. Table S2. Analyses of interactions. Table S3. Relative risk of renal progression after excluding those with post-episode proteinuria. Table S4. Relative risk of subsequent sustained 30% renal decline and new chronic kidney disease stage 4 after acute kidney injury with additional adjustment for acute hospital diagnoses. Table S5. Relative risk of subsequent sustained 30% renal decline and new chronic kidney disease stage 4 after acute kidney injury using Fine and Gray model. Supplementary material is linked to the online version of the paper at www.kidney-international.org.

Anti-glomerular basement membrane (GBM) disease and the anti-neutrophil cytoplasm antibody (ANCA)-associated vasculitides (AAV) are rare conditions, with estimated incidences in Europe of 1 and 20 per million population per year, respectively.1, 2 The concurrence of both ANCA and anti-GBM antibodies in individual patients, however, is well-recognized, and occurs at a much higher frequency than would be expected by chance alone. This phenomenon was first reported within a few years of the first description of ANCA in the 1980s,3, 4 and has been observed in several series from around the world over the subsequent 30 years.5, 6, 7, 8 It is clear that the 2 antibody populations associated with these diseases are antigenically distinct,9 and that this phenomenon is not due to cross-reactivity, although the mechanisms of the association are not fully understood.

n observed in several series from around the world over the subsequent 30 years.5, 6, 7, 8 It is clear that the 2 antibody populations associated with these diseases are antigenically distinct,9 and that this phenomenon is not due to cross-reactivity, although the mechanisms of the association are not fully understood. Several studies have reported the outcomes of these patients who are double positive, although with conflicting findings; some have observed better outcomes compared with those with single-positive anti-GBM disease,4, 10, 11 while others have suggested that patients who are double positive have comparable or worse outcomes.5, 6, 12, 13, 14, 15, 16 These studies, however, have generally been limited by small size (many describing fewer than 20 cases) and variations in the severity of disease at presentation, with between 0% and 100% of patients being dependent on dialysis at diagnosis.8, 15 Furthermore, in the largest series to date, from Chinese centers, fewer than 25% of patients were treated with plasma exchange, and so the applicability of the findings to European patients treated with substantially different therapeutic regimens is limited.7, 16

patients being dependent on dialysis at diagnosis.8, 15 Furthermore, in the largest series to date, from Chinese centers, fewer than 25% of patients were treated with plasma exchange, and so the applicability of the findings to European patients treated with substantially different therapeutic regimens is limited.7, 16 The aim of the present study is to describe the clinical features and long-term outcomes of a contemporary cohort of patients with double-positive ANCA and anti-GBM disease. Given the rarity of these patients, we have identified cases from 4 large Northern European nephrology centers, which employ comparable treatment protocols for these cases, including plasma exchange, cyclophosphamide, and steroids, unless contraindicated. We have compared clinical features and outcomes to those for single-positive AAV and single-positive anti-GBM disease. Because patients with double-positive disease more closely resemble those with single-positive anti-GBM disease at presentation, we have also compared histopathology and treatment in these 2 groups.

d. We have compared clinical features and outcomes to those for single-positive AAV and single-positive anti-GBM disease. Because patients with double-positive disease more closely resemble those with single-positive anti-GBM disease at presentation, we have also compared histopathology and treatment in these 2 groups. Results Case identification and demographics Between 2000 and 2013, a total of 646 cases were identified at 4 centers in 3 countries, including 568 patients with single-positive AAV, 41 with single-positive anti-GBM disease, and 37 patients who were double positive for anti-GBM antibodies and ANCA (hereafter AAV, anti-GBM, and double-positive groups, respectively) (Table 1). The ratio of double-positive to single-positive anti-GBM cases was similar in all 3 countries (47% overall); however, patients who were double positive represented a variable proportion of the AAV cases (3% to 10.5%; 6.1% overall). The demographic features of the cohort are summarized in Table 1. The single-positive anti-GBM group demonstrated the typical bimodal age distribution of this disease, whereas patients who were double positive had an age distribution similar to patients with isolated AAV (Figure 1). There was no significant difference in gender ratio between the 3 groups. Notably, 1 patient in the double-positive group had a previous diagnosis of isolated anti-myeloperoxidase (MPO) and AAV 2 years prior to presenting with double-positive disease.Figure 1 Age distribution of patients with anti–glomerular basement membrane (GBM) disease, anti-neutrophil cytoplasm antibody–associated vasculitis (AAV), and double positive disease at presentation.

vious diagnosis of isolated anti-myeloperoxidase (MPO) and AAV 2 years prior to presenting with double-positive disease.Figure 1 Age distribution of patients with anti–glomerular basement membrane (GBM) disease, anti-neutrophil cytoplasm antibody–associated vasculitis (AAV), and double positive disease at presentation. Table 1 Case identification, demographics, clinical features, and serology AAV Anti-GBM Double positive P value AAV versus DP versus GBM AAV versus DP GBM versus DP AAV versus GBM Cases, n 568 41 37 – – – – • United Kingdom 171 19 20 • Sweden 100 13 8 • Czech Republic 297 9 9 Cases, % 87.9% 6.3% 5.7% Demographics Age, yr (range) 62.3 (11–95) 58.3 (13–91) 63.6 (17–88) 0.17 0.99 0.31 0.21 Gender • Male 54% 46% 38% 0.11 0.06 0.49 0.34 • Female 46% 54% 62% Clinical Features Duration of symptoms,a wk (range) 12 (0–56) 2 (0–20) 10 (1–26) <0.01 0.99 <0.01 <0.01 Lung hemorrhage 131/568 23% 16/41 40% 14/37 38% 0.01 0.04 0.85 0.02 Required RRT at presentation 132/568 23% 26/41 63% 21/37 57% <0.01 <0.01 0.55 <0.01 eGFR,b ml/min (range) 29 (5–90) 20 (5–90) 19 (6–76) 0.06 0.11 0.99 0.67 Serum creatinine,b μmol/l (range) 186 (39–693) 275 (62–667) 309 (71–606) 0.06 0.18 0.99 0.37 Serology Anti-GBM level, xULN (range) – 5.4 (1–29.1) 14.2 (1–50.4) – 0.06 – Proportion seronegative for anti-GBM, % – 4/41 11% 4/37 11% – 1.00 – ANCA serology, % <0.01 – – • Anti-MPO 48% 70% • Anti-PR3 51% 27% • Anti-MPO & PR3

46% 54% 62% Clinical Features Duration of symptoms,a wk (range) 12 (0–56) 2 (0–20) 10 (1–26) <0.01 0.99 <0.01 <0.01 Lung hemorrhage 131/568 23% 16/41 40% 14/37 38% 0.01 0.04 0.85 0.02 Required RRT at presentation 132/568 23% 26/41 63% 21/37 57% <0.01 <0.01 0.55 <0.01 eGFR,b ml/min (range) 29 (5–90) 20 (5–90) 19 (6–76) 0.06 0.11 0.99 0.67 Serum creatinine,b μmol/l (range) 186 (39–693) 275 (62–667) 309 (71–606) 0.06 0.18 0.99 0.37 Serology Anti-GBM level, xULN (range) – 5.4 (1–29.1) 14.2 (1–50.4) – 0.06 – Proportion seronegative for anti-GBM, % – 4/41 11% 4/37 11% – 1.00 – ANCA serology, % <0.01 – – • Anti-MPO 48% 70% • Anti-PR3 51% 27% • Anti-MPO & PR3 <1% (n = 2) 3% AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; DP, double-positive; eGFR, estimated glomerular filtration rate; GBM, glomerular basement membrane; MPO, myeloperoxidase; PR3, proteinase 3; RRT, renal replacement therapy; xULN, multiples of upper limit of normal. Results expressed as median ± range. Comparison between groups by Kruskall–Wallis test with Dunn’s post-test to ascertain differences between individual groups (for continuous data), or by chi-square test (for categorical data). a Calculated for a sample of 48 ANCA cases. b Censored for patients on RRT.

<1% (n = 2) 3% AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; DP, double-positive; eGFR, estimated glomerular filtration rate; GBM, glomerular basement membrane; MPO, myeloperoxidase; PR3, proteinase 3; RRT, renal replacement therapy; xULN, multiples of upper limit of normal. Results expressed as median ± range. Comparison between groups by Kruskall–Wallis test with Dunn’s post-test to ascertain differences between individual groups (for continuous data), or by chi-square test (for categorical data). a Calculated for a sample of 48 ANCA cases. b Censored for patients on RRT. Clinical presentation and serology Table 1 summarizes key clinical features and serological findings at presentation. The duration of symptoms prior to receiving a diagnosis was similar in the AAV and double-positive groups (median: 10–12 weeks), and this was significantly longer than in single-positive anti-GBM patients (2 weeks; P < 0.01). Despite shorter duration of symptoms, the severity of disease—whether defined as need for hemodialysis or, in patients who were not dialysis-dependent, by GFR estimations or serum creatinine measurements—in anti-GBM patients was similar to that of patients who were double positive. The frequency of alveolar hemorrhage was similar in anti-GBM and double-positive groups, occurring in about one-third of patients. Severe disease manifestations (dialysis requirement and lung hemorrhage) were less common in patients with AAV, each occurring in approximately one-quarter of cases. Patients who were double positive had additional extrarenal manifestations, including nonhemorrhage lower respiratory tract disease (in 26%), otorhinolaryngological involvement (18%), musculoskeletal symptoms (18%), cutaneous features (13%), neurological (8%), gastrointestinal (5%), and ocular symptoms (3%).

of cases. Patients who were double positive had additional extrarenal manifestations, including nonhemorrhage lower respiratory tract disease (in 26%), otorhinolaryngological involvement (18%), musculoskeletal symptoms (18%), cutaneous features (13%), neurological (8%), gastrointestinal (5%), and ocular symptoms (3%). Because a variety of anti-GBM assays were used during the study period and at different hospital sites, results were standardized by expressing them as multiples of the upper limit of normal (xULN) for each assay. Patients who were double positive tended to have lower levels of circulating anti-GBM antibodies than did anti-GBM patients who were single positive, although the difference between groups was not statistically significant. A similar proportion of patients in both these groups (approximately 10%) were seronegative for circulating anti-GBM antibodies. In these cases, there was convincing evidence of linear IgG deposition on renal biopsy, in the absence of another attributable cause, in keeping with our definition of anti-GBM disease.

cant. A similar proportion of patients in both these groups (approximately 10%) were seronegative for circulating anti-GBM antibodies. In these cases, there was convincing evidence of linear IgG deposition on renal biopsy, in the absence of another attributable cause, in keeping with our definition of anti-GBM disease. In the AAV group, the proportion of patients who were positive for anti-proteinase 3 (PR3) versus anti-MPO antibody was approximately equal. However, in the double-positive groups there was a comparative over-representation of patients with anti-MPO antibodies (70% vs. 48%, P < 0.01). One patient in the double-positive group was triple positive for anti-MPO, anti-PR3, and anti-GBM antibodies. Notably, this patient had a history of recreational drug use and was positive for hepatitis C virus. Because a variety of methodologies, including indirect immunofluorescence and antigen-specific assays, were used to confirm ANCA positivity over the period of the study, it was not possible to standardize comparisons of ANCA titer.

s patient had a history of recreational drug use and was positive for hepatitis C virus. Because a variety of methodologies, including indirect immunofluorescence and antigen-specific assays, were used to confirm ANCA positivity over the period of the study, it was not possible to standardize comparisons of ANCA titer. Histopathology Because the anti-GBM and double-positive groups had similar disease severity at presentation, we performed more detailed analysis to compare histopathological and therapeutic differences in these 2 cohorts. Approximately two-thirds of patients in both groups underwent renal biopsy, as described in Table 2. The severity of renal disease in those who underwent biopsy was similar in both groups and equivalent to the severity of renal disease in the parent cohort, suggesting that biopsy findings may be representative and comparable between groups. It was not possible, however, to retrospectively identify the reasons why renal biopsy was not undertaken in all remaining patients. In some cases, this was due to the need for immediate treatment with plasma exchange, and others were considered too clinically unstable for biopsy. In keeping with the age difference between the parent groups, the mean age in the subset of patients who underwent biopsy was lower in the anti-GBM group compared with the double-positive group.Table 2 Histopathology Anti-GBM Double positive P value Underwent biopsy, n (%) 29 (71%) 25 (68%) 0.81 Mean age at biopsy, yr (range) 46 (13–91) 62 (46–76) <0.01 Renal status at biopsy • Required RRT 52% 54% 1.00 • eGFR,a ml/min (range)

Histopathology Because the anti-GBM and double-positive groups had similar disease severity at presentation, we performed more detailed analysis to compare histopathological and therapeutic differences in these 2 cohorts. Approximately two-thirds of patients in both groups underwent renal biopsy, as described in Table 2. The severity of renal disease in those who underwent biopsy was similar in both groups and equivalent to the severity of renal disease in the parent cohort, suggesting that biopsy findings may be representative and comparable between groups. It was not possible, however, to retrospectively identify the reasons why renal biopsy was not undertaken in all remaining patients. In some cases, this was due to the need for immediate treatment with plasma exchange, and others were considered too clinically unstable for biopsy. In keeping with the age difference between the parent groups, the mean age in the subset of patients who underwent biopsy was lower in the anti-GBM group compared with the double-positive group.Table 2 Histopathology Anti-GBM Double positive P value Underwent biopsy, n (%) 29 (71%) 25 (68%) 0.81 Mean age at biopsy, yr (range) 46 (13–91) 62 (46–76) <0.01 Renal status at biopsy • Required RRT 52% 54% 1.00 • eGFR,a ml/min (range) 21 (5–90) 16 (8–73) 0.78 • Serum creatininea μmol/l (range) 275 (62–677) 315 (71–606) 0.75 Glomerular findings • Crescentic glomeruli, % 64% (0–100) 64% (25–100) 0.98 • Sclerotic glomeruli, % 0% (0–80) 15% (0–100) 0.19 • Normal glomeruli, %

Anti-GBM Double positive P value Underwent biopsy, n (%) 29 (71%) 25 (68%) 0.81 Mean age at biopsy, yr (range) 46 (13–91) 62 (46–76) <0.01 Renal status at biopsy • Required RRT 52% 54% 1.00 • eGFR,a ml/min (range) 21 (5–90) 16 (8–73) 0.78 • Serum creatininea μmol/l (range) 275 (62–677) 315 (71–606) 0.75 Glomerular findings • Crescentic glomeruli, % 64% (0–100) 64% (25–100) 0.98 • Sclerotic glomeruli, % 0% (0–80) 15% (0–100) 0.19 • Normal glomeruli, % 5% (0–100) 0% (0–67) 0.56 Tubular atrophy, % (range) 5% (0%–30%) 27% (0%–80%) <0.01 Immunofluorescence pattern 0.69 • Linear IgG 79% 80% • Pauci-immune 3% 8% • Technically inadequate 17% 12% eGFR, estimated glomerular filtration rate; GBM, glomerular basement membrane; RRT, renal replacement therapy. Results expressed as median ± range. Comparison between groups by Mann-Whitney test (for continuous data), or by chi-square test (for categorical data). a Censored for patients on RRT.

3% 8% • Technically inadequate 17% 12% eGFR, estimated glomerular filtration rate; GBM, glomerular basement membrane; RRT, renal replacement therapy. Results expressed as median ± range. Comparison between groups by Mann-Whitney test (for continuous data), or by chi-square test (for categorical data). a Censored for patients on RRT. The median number of glomeruli in each biopsy was 15 (interquartile range: 10–20). There was no difference in the proportion of crescentic glomeruli between the 2 groups. There was, however, a tendency for more sclerotic glomeruli to be observed in patients who were double positive (median: 15% vs. 0%; P = 0.188). Likewise, the finding of “synchronous” crescent formation tended to be more commonly observed in patients with anti-GBM disease (73% vs. 33% in patients who were double positive; P = 0.092). There was a highly significant difference in the degree of interstitial fibrosis and tubular atrophy between these 2 groups, with more evidence of chronic damage in double-positive cases (median: 27% vs. 5%, P < 0.001). In those cases in which adequate tissue was available for analysis, all but 3 cases (2 double-positive, 1 single-positive for anti-GBM, and all with circulating anti-GBM antibodies) had linear deposition of IgG by immunofluorescence analysis of the kidney biopsy.

n double-positive cases (median: 27% vs. 5%, P < 0.001). In those cases in which adequate tissue was available for analysis, all but 3 cases (2 double-positive, 1 single-positive for anti-GBM, and all with circulating anti-GBM antibodies) had linear deposition of IgG by immunofluorescence analysis of the kidney biopsy. Treatment There was no detectable difference between single-positive anti-GBM and double-positive groups with regard to initial treatment administered, the majority receiving standard of care with steroids (97% vs. 100% in double-positive and anti-GBM cases, respectively; P = 0.47), cyclophosphamide (100% vs. 92%; P = 0.24), and plasma exchange (80% vs. 89%; P = 0.33). In total, 10 patients did not undergo plasma exchange for various reasons. In the double-positive group, 7 patients did not receive plasma exchange. Of these, 2 were dialysis dependent at presentation with 100% crescent formation on kidney biopsy, and in the absence of lung hemorrhage, plasma exchange was deemed futile. These patients received cytotoxic therapy and steroids for nonrenal manifestations. Of the other 5 patients, none had alveolar hemorrhage or required dialysis, and their initial treatment was typical of those presenting with isolated AAV. In the single-positive anti-GBM group, 1 patient was dialysis dependent with no normal glomeruli on kidney biopsy, and so treatment was deemed futile; 1 patient had well-preserved renal function (serum creatinine: 78 μmol/l) and so plasma exchange was initially reserved for nonresponse to immunosuppressive treatment alone, and was ultimately not required; and 1 patient was clinically unstable and therefore unable to undergo plasma exchange.

eatment was deemed futile; 1 patient had well-preserved renal function (serum creatinine: 78 μmol/l) and so plasma exchange was initially reserved for nonresponse to immunosuppressive treatment alone, and was ultimately not required; and 1 patient was clinically unstable and therefore unable to undergo plasma exchange. At 6 months, 74% of patients who were double positive were receiving ongoing immunosuppressive treatment with or without corticosteroids (71% with azathioprine, 21% with mycophenolate mofetil, and 8% with methotrexate), whereas only 14% of patients with single-positive anti-GBM disease received ongoing therapy (P < 0.001), of whom 80% received azathioprine and 20% received MMF.

eceiving ongoing immunosuppressive treatment with or without corticosteroids (71% with azathioprine, 21% with mycophenolate mofetil, and 8% with methotrexate), whereas only 14% of patients with single-positive anti-GBM disease received ongoing therapy (P < 0.001), of whom 80% received azathioprine and 20% received MMF. Outcomes Patient and renal survival for all 3 cohorts at 3 and 12 months is summarized in Table 3. Overall patient survival was similar in all groups at both time points. Renal survival was favorable in the AAV group at both time points, although there was no significant difference in the proportion of patients who required dialysis in the anti-GBM and double-positive group at either time point. The proportion of patients who presented with dialysis-dependent renal failure and who recovered renal function and were alive at 1 year was significantly different between groups, varying from 17% in patients with single-positive anti-GBM disease to 29% in patients who were double positive and 49% in AAV cases. As Figure 2 demonstrates, this was due in part to cross-over from dialysis dependence to independence, and vice versa, particularly within the double-positive group, in which a substantial proportion recovered renal function in the first 3 months of follow-up (35% recovery vs. 10% recovery in patients who were single positive; P = 0.11). There was no significant difference in age (mean: 65 vs. 64 years; P = 0.88) or receipt of plasma exchange (100% vs. 82%; P = 0.52) between those patients and who were double-positive and who recovered and those who did not, respectively, although those who recovered renal function tended to have lower levels of anti-GBM antibodies (3.8 vs. 10 xULN, respectively; P = 0.07). Only one-half of the patients who were double-positive who recovered renal function underwent renal biopsy, and so it was not possible to reliably identify histopathological predictors of treatment response.Figure 2 Transition to and from dialysis dependence in the first 3 months (mo) following treatment, in double-positive and single-positive anti–glomerular basement membrane disease cases. Censored for death in the first 3 months.

t was not possible to reliably identify histopathological predictors of treatment response.Figure 2 Transition to and from dialysis dependence in the first 3 months (mo) following treatment, in double-positive and single-positive anti–glomerular basement membrane disease cases. Censored for death in the first 3 months. Table 3 Patient and renal survival at 3 and 12 months after diagnosis Diagnosis 0 Months 3 months 12 months Renal recovery at 1 yearb Independent of RRT Patient survival Renal survivala Patient survival Renal survivala AAV 437/568 77% 540/568 95% 490/540 91% 512/568 90% 452/512 88% 64/131 49% Anti-GBM 15/41 37% 37/41 90% 15/36 42% 36/41 87% 15/34 44% 4/24 17% Double positive 16/37 43% 33/37 89% 16/32 50% 31/37 83% 16/30 53% 6/21 29% P value <0.01 0.13 <0.01 0.38 <0.01 <0.01 AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; GBM, glomerular basement membrane; RRT, renal replacement therapy. Comparison between groups by chi-square test. a Censored for death. b Proportion of patients requiring RRT at presentation who were alive with independent renal function at 1 year.

Diagnosis 0 Months 3 months 12 months Renal recovery at 1 yearb Independent of RRT Patient survival Renal survivala Patient survival Renal survivala AAV 437/568 77% 540/568 95% 490/540 91% 512/568 90% 452/512 88% 64/131 49% Anti-GBM 15/41 37% 37/41 90% 15/36 42% 36/41 87% 15/34 44% 4/24 17% Double positive 16/37 43% 33/37 89% 16/32 50% 31/37 83% 16/30 53% 6/21 29% P value <0.01 0.13 <0.01 0.38 <0.01 <0.01 AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; GBM, glomerular basement membrane; RRT, renal replacement therapy. Comparison between groups by chi-square test. a Censored for death. b Proportion of patients requiring RRT at presentation who were alive with independent renal function at 1 year. The median duration of follow-up was 4.8 years (range: 0–15 years). Long-term patient and renal survival is summarized in Figure 3a and b, respectively. No difference in unadjusted overall patient survival was observed during the study (P = 0.49). Renal survival was favorable in the AAV group compared with both the anti-GBM group (P < 0.01) and the double double-positive groups (P < 0.01). Patients who were double positive tended to have better renal survival than those who were single positive with anti-GBM disease, although this difference was not statistically significant in unadjusted analysis (P = 0.17).Figure 3 Unadjusted Kaplan–Meier survival functions describing long-term patient, renal, and relapse-free survival rates of the study cohort during 10 years’ follow-up. (a) Overall patient survival. (b) End-stage renal disease-free survival. (c) Relapse-free survival (censored for death). AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; DP, double positive; ESRD, end-stage renal disease; GBM, glomerular basement membrane.

s of the study cohort during 10 years’ follow-up. (a) Overall patient survival. (b) End-stage renal disease-free survival. (c) Relapse-free survival (censored for death). AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; DP, double positive; ESRD, end-stage renal disease; GBM, glomerular basement membrane. Relapse Relapse data were available in 316 AAV patients and all patients with anti-GBM disease and double positivity until last follow-up. Within the follow-up period, 116 of 316 patients with AAV (37%) and 8 of 37 patients who were double positive (22%) had a relapse. Two patients with single-positive anti-GBM disease had early recrudescence of anti-GBM antibodies requiring augmented treatment in the first 6 months following initial diagnosis, which we believe reflected inadequate initial disease control rather than disease relapse. No patients with single-positive anti-GBM disease exhibited evidence of relapse or recurrent antibodies beyond 6 months. During long-term follow-up, a significant proportion (n = 8; approximately 50%) of the surviving patients who were double positive developed recurrent disease (Figure 3c). These relapses tended to occur late (median time to first relapse: 4.4 years; range: 1.1–7.9 years), and the majority (5 of 8) were in patients with ANCA directed against PR3, in keeping with the natural history of isolated AAV. Two patients had relapses associated with MPO-ANCA, and 1 notable patient had a relapse of both MPO-ANCA and anti-GBM antibodies. All relapses occurred in patient who were ANCA positive, and the majority (6 of 8) were associated with a rise in ANCA titer of more than 25%, or seroconversion from ANCA-negative to ANCA-positive status, in the 6 months prior to diagnosis of relapsing disease. Table 4 summarizes the clinical features of each relapse and its relation to immunosuppressive treatment. Notably, the majority of patients were not receiving maintenance treatment other than corticosteroids at the time of relapse. As shown in Figure 3c, the frequency of relapse in the double-positive cohort was comparable to that in AAV cases (P = 0.29), whereas there were no relapses in the anti-GBM group (P < 0.01). Two patients who were double positive have undergone renal transplantation. To date, no disease recurrence related to either ANCA or anti-GBM disease has been described in the renal allografts.Table 4 Details of double-positive patients with relapse

.29), whereas there were no relapses in the anti-GBM group (P < 0.01). Two patients who were double positive have undergone renal transplantation. To date, no disease recurrence related to either ANCA or anti-GBM disease has been described in the renal allografts.Table 4 Details of double-positive patients with relapse Case Time to relapse (months) ANCA GBM Ab Organ involvement Treatment at time of relapse Treatment for relapse Subsequent relapse 1 13 PR3 Neg Renal, skin CS only; CYC stopped 9 mo prior CYC, CS Yes 2 16 PR3 Neg LRT CS only; AZA stopped 4 mo prior AZA, CS No 3 22 MPO Neg LRT CS only; AZA stopped 18 mo prior CS Yes 4 36 PR3 Neg Renal, Skin MMF, CS RTX, CYC, CS Yes 5 71 PR3 Neg Renal AZA AZA, CS No 6 81 PR3 Neg Constitutional CS only; AZA stopped 10 mo prior AZA, CS Yes 7 87 MPO Neg Renal None; off treatment 5 yr CS No 8 95 MPO Positive LRT, Renal None; off treatment 12 mo RTX, CYC, CS No ANCA, anti-neutrophil cytoplasm antibody; AZA, azathioprine; CS, corticosteroids; CYC, cyclophosphamide; GBM, glomerular basement membrane; LRT, lower respiratory tract; MPO, myeloperoxidase; MMF, mycophenolate mofetil; mo, months; PR3, proteinase 3; RTX, rituximab; y, years.

l None; off treatment 12 mo RTX, CYC, CS No ANCA, anti-neutrophil cytoplasm antibody; AZA, azathioprine; CS, corticosteroids; CYC, cyclophosphamide; GBM, glomerular basement membrane; LRT, lower respiratory tract; MPO, myeloperoxidase; MMF, mycophenolate mofetil; mo, months; PR3, proteinase 3; RTX, rituximab; y, years. Predictors of death, ESRD, and relapse There were significant differences in age, proportion of patients requiring renal replacement therapy (RRT), and proportion of patients with lung hemorrhage at presentation between patients diagnosed with AAV, anti-GBM disease, and double positivity. We therefore performed regression analysis to identify predictors of death and end-stage renal disease (ESRD) in all patients, correcting for these differences in baseline variables (Table 5; Figure 4).Figure 4 Cox proportional hazards regression curves describing long-term risk of (a) death, (b) end-stage renal disease (ESRD), and (c) death or ESRD. Measures being controlled for include diagnosis, age, requirement for renal replacement therapy at presentation, and presence of lung hemorrhage at presentation. AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; DP, double positive; GBM, anti-glomerular basement membrane disease. Table 5 Predictors of death and end-stage renal disease

Predictors of death, ESRD, and relapse There were significant differences in age, proportion of patients requiring renal replacement therapy (RRT), and proportion of patients with lung hemorrhage at presentation between patients diagnosed with AAV, anti-GBM disease, and double positivity. We therefore performed regression analysis to identify predictors of death and end-stage renal disease (ESRD) in all patients, correcting for these differences in baseline variables (Table 5; Figure 4).Figure 4 Cox proportional hazards regression curves describing long-term risk of (a) death, (b) end-stage renal disease (ESRD), and (c) death or ESRD. Measures being controlled for include diagnosis, age, requirement for renal replacement therapy at presentation, and presence of lung hemorrhage at presentation. AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; DP, double positive; GBM, anti-glomerular basement membrane disease. Table 5 Predictors of death and end-stage renal disease Unadjusted analysis Death ESRD Death or ESRD HR (CI) P value HR (CI) P value HR (CI) P value Diagnosis – 0.25 – <0.01 – <0.01 DPa versus AAV 0.72 (0.406–1.26) 0.25 0.31 (0.19–0.51) <0.01 0.33 (0.42–0.52) <0.01 DPa versus GBM 0.78 (0.34–1.79) 0.56 1.542 (0.85–2.80) 0.16 1.38 (0.78–2.45) 0.27 AAVa versus GBM 1.10 (0.58–2.08) 0.78 4.98 (3.25–7.64) <0.01 4.21 (2.78–6.37) <0.01 Age at presentation 1.05 (1.04–1.06) <0.01 1.01 (0.99–1.02) 0.18 1.01 (1.00–1.02) 0.01 RRT at presentation 2.20 (1.63–2.97) <0.01 9.34 (6.53–13.33) <0.01 5.71 (4.17–15.38) <0.01 LH at presentation 1.45 (1.06–1.99) 0.02 1.89 (1.36–2.63) <0.01 1.79 (1.32–2.38) <0.01 Multivariable analysis Death ESRD Death or ESRD HR (CI) P value HR (CI) P value HR (CI) P value Diagnosis – 0.95 <0.01 – <0.01 DPa versus AAV 0.97 (0.52–1.81) 0.92 0.62 (0.36–1.06) 0.08 0.57 (0.35–0.93) 0.02 DPa versus GBM 0.94 (0.39–2.28) 0.89 1.66 (0.88–3.12) 0.12 1.49 (0.82–2.72) 0.19 AAVa versus GBM 0.97 (0.48–1.95) 0.93 2.66 (1.69–4.19) <0.01 2.63 (1.69–4.09) <0.01 Age at presentation 1.05 (1.04–1.06) <0.01 1.01 (1.00–1.02) 0.11 1.02 (1.01–1.03) 0.03 RRT at presentation 2.04 (1.49–2.78) <0.01 7.69 (5.26–11.1) <0.01 4.55 (3.33–6.25) <0.01 LH at presentation 1.37 (0.99–1.89) 0.06 1.15 (0.82–1.61) 0.42 1.21 (0.88–1.64) 0.23 AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; DP, double positive; ESRD, end-stage renal disease; GBM, anti–glomerular basement membrane disease; HR, hazard ratio; LH, lung hemorrhage; RRT, renal replacement therapy.

1 LH at presentation 1.37 (0.99–1.89) 0.06 1.15 (0.82–1.61) 0.42 1.21 (0.88–1.64) 0.23 AAV, anti-neutrophil cytoplasm antibody–associated vasculitis; DP, double positive; ESRD, end-stage renal disease; GBM, anti–glomerular basement membrane disease; HR, hazard ratio; LH, lung hemorrhage; RRT, renal replacement therapy. a Reference group for estimates of hazard ratios. Unadjusted predictors of death included RRT at presentation (hazard ratio [HR]: 2.2 [1.63–2.97]; P < 0.01), lung hemorrhage (HR: 1.45 [1.06–1.99]; P = 0.02), and age (HR: 1.05 [1.04–1.06]; P < 0.01), but not diagnosis. In multivariable analysis, RRT at presentation (HR: 2.04 [1.49–2.78], P < 0.01) and age (HR: 1.05 [1.04–1.06]; P < 0.01) predicted death. The influence of lung hemorrhage at presentation (HR: 1.37 [0.99–1.89]; P = 0.06) approached statistical significance, while diagnosis had no influence.

P < 0.01), but not diagnosis. In multivariable analysis, RRT at presentation (HR: 2.04 [1.49–2.78], P < 0.01) and age (HR: 1.05 [1.04–1.06]; P < 0.01) predicted death. The influence of lung hemorrhage at presentation (HR: 1.37 [0.99–1.89]; P = 0.06) approached statistical significance, while diagnosis had no influence. Unadjusted predictors of progression to ESRD included diagnosis (P < 0.01), lung hemorrhage at presentation (HR: 1.89 [1.32–2.63]; P < 0.01), and RRT at presentation (HR: 9.34 [6.53–13.33]; P < 0.01). Age was not associated with progression to ESRD (P = 0.18). In multivariable analysis, RRT at presentation (HR: 7.69 [5.26–11.10]; P < 0.01) and diagnosis (P < 0.01) increased the HR of progression to ESRD. The risk of ESRD was increased in anti-GBM disease compared with AAV (HR: 2.66 [1.69–4.19], P < 0.01), though the risk in patients who were double positive versus those with AAV was not significantly different (HR: 0.62 [0.36–1.06]; P = 0.08), suggesting an intermediate risk of ESRD in patients who were double positive. Age (P = 0.11) and lung hemorrhage at presentation (P = 0.42) had no influence on progression to ESRD.

hough the risk in patients who were double positive versus those with AAV was not significantly different (HR: 0.62 [0.36–1.06]; P = 0.08), suggesting an intermediate risk of ESRD in patients who were double positive. Age (P = 0.11) and lung hemorrhage at presentation (P = 0.42) had no influence on progression to ESRD. Unadjusted predictors of a composite outcome of death or progression to ESRD included age (HR: 1.01 [1.00–1.02]; P = 0.01), diagnosis (P < 0.01), RRT at presentation (HR: 5.71 [4.17–15.38]; P < 0.01), and lung hemorrhage at presentation (HR: 1.79 [1.08–2.94]; P < 0.01). In multivariable analysis, diagnosis (P < 0.01), RRT at presentation (HR: 4.55 [3.33– 6.25]; P < 0.01), and age (HR: 1.02 [1.01–1.03]; P < 0.01) were associated with progression to death or ESRD. Lung hemorrhage at presentation was not associated with progression to death or ESRD (P = 0.23). Death or progression to ESRD was similar between double-positive and single-positive anti-GBM disease cases. However, patients with AAV had a lower hazard ratio of progression to ESRD or death compared with patients who were double positive (HR: 0.57 [0.35–0.93]; P = 0.02). Unadjusted predictors of relapse included diagnosis (P < 0.01), and RRT dependence at presentation was associated with a lower risk of relapse (P = 0.02). Further analysis was not conducted due to the relative small number of relapse episodes.

Unadjusted predictors of a composite outcome of death or progression to ESRD included age (HR: 1.01 [1.00–1.02]; P = 0.01), diagnosis (P < 0.01), RRT at presentation (HR: 5.71 [4.17–15.38]; P < 0.01), and lung hemorrhage at presentation (HR: 1.79 [1.08–2.94]; P < 0.01). In multivariable analysis, diagnosis (P < 0.01), RRT at presentation (HR: 4.55 [3.33– 6.25]; P < 0.01), and age (HR: 1.02 [1.01–1.03]; P < 0.01) were associated with progression to death or ESRD. Lung hemorrhage at presentation was not associated with progression to death or ESRD (P = 0.23). Death or progression to ESRD was similar between double-positive and single-positive anti-GBM disease cases. However, patients with AAV had a lower hazard ratio of progression to ESRD or death compared with patients who were double positive (HR: 0.57 [0.35–0.93]; P = 0.02). Unadjusted predictors of relapse included diagnosis (P < 0.01), and RRT dependence at presentation was associated with a lower risk of relapse (P = 0.02). Further analysis was not conducted due to the relative small number of relapse episodes. Discussion This is the largest published series to compare the outcomes of patients with both ANCA and anti-GBM autoantibodies with patients with single-positive AAV and anti-GBM disease. As such, it provides several important observations: the phenomenon of double positivity is common, these patients experience the early morbidity and mortality typical of anti-GBM disease, and they carry the long-term risk of relapse typical of AAV.

ntibodies with patients with single-positive AAV and anti-GBM disease. As such, it provides several important observations: the phenomenon of double positivity is common, these patients experience the early morbidity and mortality typical of anti-GBM disease, and they carry the long-term risk of relapse typical of AAV. Patients who are double positive accounted for approximately one-half of all anti-GBM disease cases seen at our centers since 2000, and over 10% of AAV patients with renal involvement seen at the UK site over the same time period. The proportion of AAV cases was variable at the other sites, perhaps reflecting differences in referral pattern at each (with varying proportions of patients with extrarenal vasculitis) and differences in sensitivity of assays used to detect ANCA, or geographical variations in disease frequency. Notably, a recent study reported that more than 60% of patients with anti-GBM disease had autoantibodies reactive to linear epitopes of MPO, versus 24% who had antibodies to native protein detected by conventional assays. Our study highlights this common concurrence, and our observations reiterate the need to determine the alternative antibody type in all cases of AAV- or anti-GBM disease.

i-GBM disease had autoantibodies reactive to linear epitopes of MPO, versus 24% who had antibodies to native protein detected by conventional assays. Our study highlights this common concurrence, and our observations reiterate the need to determine the alternative antibody type in all cases of AAV- or anti-GBM disease. In this large series, we observed comparable severity of disease at presentation between single-positive anti-GBM and double-positive cases, with approximately 60% of patients requiring renal replacement therapy at presentation, and one-third developing lung hemorrhage, in both groups. Regression analysis suggests that it is these severe disease manifestations, rather than diagnosis per se, that affect overall patient survival, and because they are equally prevalent in patients with anti-GBM disease and those who are double positive (and less frequent than in AAV), this suggests that anti-GBM disease is the “dominant” early disease phenotype in double-positive cases. These patients, however, also demonstrate clinical features of AAV, such as an older age distribution, a longer prodrome of systemic symptoms, and features of chronic damage on renal biopsy (in excess of what would be expected for the age difference between groups). In addition, regression analysis suggests that patients who are double positive have an intermediate risk of progression to ESRD compared with single-positive AAV or anti-GBM cases. This may be related to the observation that more than one-third of the surviving patients who were double positive and required dialysis at presentation regained independent renal function by 3 months. This propensity to renal recovery was more in keeping with AAV than anti-GBM disease, where regaining renal function from dialysis is very uncommon,17 and consistent with some previous reports of double-positive cases.4, 10, 11

itive and required dialysis at presentation regained independent renal function by 3 months. This propensity to renal recovery was more in keeping with AAV than anti-GBM disease, where regaining renal function from dialysis is very uncommon,17 and consistent with some previous reports of double-positive cases.4, 10, 11 That our patients who were double positive had this tendency to renal recovery and intermediate long-term renal survival, despite more chronic renal damage on kidney biopsy and more advanced age at presentation, is striking. Responders tended to have lower levels of anti-GBM antibodies, suggesting that they may have been identified early in the course of anti-GBM disease, or that they may have a “milder” form of disease. Of note, there are several recent reports of “atypical” variants of anti-GBM disease, each characterized by less severe renal involvement than is usually observed.18, 19, 20 Differences in antibody subclass or in antigen or epitope specificity may account for these variable presentations. With regard to patients who are double positive, 1 previous study found that they had a broader spectrum of anti-GBM antibodies and lower reactivity to a3(IV)NC1 than patients who were single positive,21 although an earlier study did not report differences in antigen specificity.22 It is therefore possible that differences in antigen or epitope specificity account for the difference in treatment response seen in our cohort; however, we were unable to analyze this in detail in a retrospective study. Likewise, we have been unable to identify histopathological predictors of recovery, a significant limitation of our analysis. Previous studies have shown that the proportion of crescentic and normal glomeruli in anti-GBM disease is predictive of prognosis,17, 23 and a prognostic classification based on histopathological findings has been proposed for ANCA-associated glomerulonephritis.24, 25 It is unclear, however, whether these observations apply to patients who are double positive. Given the overall rarity of anti-GBM disease, a larger, multicenter analysis of histopathological lesions in these cases is likely needed in order to infer reliable prognostic indicators.

A-associated glomerulonephritis.24, 25 It is unclear, however, whether these observations apply to patients who are double positive. Given the overall rarity of anti-GBM disease, a larger, multicenter analysis of histopathological lesions in these cases is likely needed in order to infer reliable prognostic indicators. Pending such information, our observations suggest that while patients who are double positive behave primarily like those with isolated anti-GBM disease, a subset of patients who are dialysis dependent at presentation may be more responsive to therapy, and aggressive treatment may be warranted in some cases.

A-associated glomerulonephritis.24, 25 It is unclear, however, whether these observations apply to patients who are double positive. Given the overall rarity of anti-GBM disease, a larger, multicenter analysis of histopathological lesions in these cases is likely needed in order to infer reliable prognostic indicators. Pending such information, our observations suggest that while patients who are double positive behave primarily like those with isolated anti-GBM disease, a subset of patients who are dialysis dependent at presentation may be more responsive to therapy, and aggressive treatment may be warranted in some cases. The other striking characteristic of AAV retained by patients who are double positive is a risk of disease relapse. The long-term follow-up offered by our study suggests that almost one-half of surviving patients who are double positive will experience disease relapse at some point, at a frequency comparable to that observed in our single-positive AAV cohort. As might be expected, these relapses were more likely in patients who were anti-PR3 positive, and were associated with preceding increases in ANCA titer. Of note, 1 patient relapsed with both ANCA and anti-GBM antibodies. These observations suggest that while the dominant early disease phenotype in these patients is anti-GBM disease, unlike patients with isolated anti-GBM disease these cases require frequent long-term follow-up and consideration of maintenance immunosuppression. That 1 of our patients had an earlier diagnosis of isolated AAV prior to presenting with a double-positive “relapse” also suggests that anti-GBM antibodies should be determined in relapsing AAV cases, particularly if there is evidence of renal involvement.

m follow-up and consideration of maintenance immunosuppression. That 1 of our patients had an earlier diagnosis of isolated AAV prior to presenting with a double-positive “relapse” also suggests that anti-GBM antibodies should be determined in relapsing AAV cases, particularly if there is evidence of renal involvement. The mechanism of association between AAV and anti-GBM disease in unclear. Studies in animal models suggest that administration of the alternate antibody type may augment the severity of renal disease in models of either vasculitis or anti-GBM nephritis;26, 27, 28 however, these in vivo studies have shed little light on the spontaneous development of both antibody types, or on the sequence in which they develop in clinical disease. An elegant clinical study by Olson and colleagues, using stored sera from the US Department of Defense, suggested that the majority of patients with anti-GBM disease had detectable ANCA prior to the development of anti-GBM antibodies, which in turn predated the development of clinical disease, suggesting that AAV may act as trigger for anti-GBM disease.29 Our observations support this hypothesis: patients who were double positive had the age restriction of isolated AAV cases, a longer prodrome of systemic symptoms prior to diagnosis, and more features of chronicity on their renal biopsy, suggesting that ANCA-mediated glomerular inflammation may precede and contribute to the development of anti-GBM disease, perhaps by disrupting the quaternary structure of the GBM.30 This could lead to the exposure of normally sequestered epitopes in a pro-inflammatory milieu, resulting in a fulminant anti-GBM response. Conversely, it has been shown that aberrant extracellular expression of myeloperoxidase, as a constituent of neutrophil extracellular traps, may predispose to the development of anti-MPO antibodies,31 and that neutrophil extracellular traps are formed in experimental anti-GBM disease.32 Thus, it is possible that glomerular neutrophil recruitment and activation in anti-GBM disease similarly contributes to the development of ANCA.

trophil extracellular traps, may predispose to the development of anti-MPO antibodies,31 and that neutrophil extracellular traps are formed in experimental anti-GBM disease.32 Thus, it is possible that glomerular neutrophil recruitment and activation in anti-GBM disease similarly contributes to the development of ANCA. The recent observation that a high proportion of patients with anti-GBM disease have autoantibodies reactive to linear epitopes of MPO might support this hypothesis, as it suggests reactivity to conformational MPO epitopes might arise as a consequence of inter- and intramolecular epitope spreading initiated by anti-GBM disease.33 Whether additional environmental or genetic factors predispose to forming both antibodies is unclear. The genetic associations of both anti-GBM disease and AAV are increasingly well-described,34, 35 and both conditions have strong associations with certain HLA genes. Notably, both conditions have reported associations with HLA-DRB1*1501, and a previous small study observed a DRB1-15 genotype in 4 of the 5 patients who were double positive that were examined.36

disease and AAV are increasingly well-described,34, 35 and both conditions have strong associations with certain HLA genes. Notably, both conditions have reported associations with HLA-DRB1*1501, and a previous small study observed a DRB1-15 genotype in 4 of the 5 patients who were double positive that were examined.36 In this descriptive, retrospective study, we have been unable to analyze the genetic or detailed serological and pathological features of our cohort. Strengths of our study, however, include its large size, its long follow-up period beyond 10 years for many patients, the inclusion of all single-positive anti-GBM and AAV cases by way of controlled comparison, and that it is multicenter, from international sites that utilize comparable treatment regimens. We highlight several important clinical practice points—in particular that while anti-GBM disease is the predominant disease phenotype in these patients, their ANCA status should neither be ignored nor forgotten, because a subset may be more responsive to initial immunosuppressive treatment, and they have a significant risk of relapse requiring careful long-term follow-up and monitoring.

that while anti-GBM disease is the predominant disease phenotype in these patients, their ANCA status should neither be ignored nor forgotten, because a subset may be more responsive to initial immunosuppressive treatment, and they have a significant risk of relapse requiring careful long-term follow-up and monitoring. Methods This is a retrospective analysis of patients diagnosed with AAV, anti-GBM disease, and double-positive ANCA and anti-GBM antibody disease from 4 European centers: Hammersmith Hospital, London, UK; Charles University Hospital, Prague, Czech Republic; Skånes University Hospital, Lund, Sweden; and Linköping University Hospital, Linköping, Sweden. All patients diagnosed between 2000 and 2013 with at least 1 year of follow-up were included in analysis. Patients with a diagnosis of systemic vasculitis consistent with the Chapel Hill Consensus Conference37 and positive ANCA serology were included in the AAV group. Anti-GBM disease was defined by either (i) the presence of circulating anti-GBM antibodies in association with clinical manifestations of alveolar hemorrhage and/or rapidly progressive glomerulonephritis, or (ii) biopsy-proven crescentic glomerulonephritis with linear deposition of IgG along the GBM in the absence of another attributable cause (such as diabetes mellitus or paraproteinaemia). The double-positive cohort included patients who met this diagnosis of anti-GBM disease and in addition had positive ANCA serology.

r (ii) biopsy-proven crescentic glomerulonephritis with linear deposition of IgG along the GBM in the absence of another attributable cause (such as diabetes mellitus or paraproteinaemia). The double-positive cohort included patients who met this diagnosis of anti-GBM disease and in addition had positive ANCA serology. Circulating anti-GBM antibodies were identified using conventional commercially available assays, which varied between site and over time at each center. Antigen substrates included purified bovine or human GBM preparations and recombinant human α3(IV) collagen chain. ANCA was detected either by indirect immunofluorescence using ethanol–fixed human neutrophils, or by antigen specific assays using commercially available ELISA or bead-based multiplex assays, which used purified human ANCA antigens. Patients who are ANCA positive were subclassified by ANCA specificity to either myeloperoxidase or PR3 antigens. In patients who tested positive by fluorescence testing but negative by antigen-specific assay, those with a perinuclear indirect immunofluorescence pattern were assigned to the MPO group and those with a cytoplasmic pattern to the PR3 group.

subclassified by ANCA specificity to either myeloperoxidase or PR3 antigens. In patients who tested positive by fluorescence testing but negative by antigen-specific assay, those with a perinuclear indirect immunofluorescence pattern were assigned to the MPO group and those with a cytoplasmic pattern to the PR3 group. Following identification of cases, the case notes, pathology, and laboratory records were reviewed to collect data using an electronic database on details of clinical presentation, treatment, and outcomes. Patients were followed up from presentation until last clinical encounter prior to December 31, 2014. RRT at presentation was defined by the need for acute dialysis during the first hospital admission. ESRD was defined by a sustained requirement for RRT that did not recover during follow-up or before death. GFR was estimated by the Modified Diet in Renal Disease calculation.38 Relapses were defined by an increase or recurrence in disease activity requiring augmented immunotherapy. For histopathological analysis, we reviewed original renal biopsy reports. We defined “crescentic” glomeruli by the presence of cellular, fibrocellular, or fibrous crescents. Synchronous crescent formation was defined by the presence of uniformly aged glomerular crescents in the biopsy, whereas a mix of cellular, fibrocellular, or fibrous crescents defined “asynchronous” crescent formation. Obsolete glomeruli, and those with segmental scars, were included in the category of “sclerotic” glomeruli. “Normal” glomeruli included those with minor mesangial or ischemic changes only, without significant proliferation, scarring, or crescent formation.

or fibrous crescents defined “asynchronous” crescent formation. Obsolete glomeruli, and those with segmental scars, were included in the category of “sclerotic” glomeruli. “Normal” glomeruli included those with minor mesangial or ischemic changes only, without significant proliferation, scarring, or crescent formation. We compared baseline clinical features and long-term outcomes between all 3 diagnoses. In addition, we performed more detailed comparison of histopathology and treatment in the single-positive anti-GBM and double-positive groups. Continuous data were regarded as nonparametric. For comparison of continuous variables, Mann-Whitney (2 groups) and Kruskal–Wallis with post hoc Dunn’s test (more than 2 groups) were used to ascertain differences between individual groups. For comparison of categorical variables between groups, chi-square test was used. Log-rank test was used to ascertain unadjusted survival differences and plotted as Kaplan–Meier curves. Cox proportional regression analysis was used to ascertain proportional hazards ratio of factors associated with categorical outcomes (death, ESRD progression, and death or ESRD progression as a composite outcome). Covariates included in the multivariable Cox proportional regression analysis included diagnosis, age, RRT at presentation, and lung hemorrhage at presentation. For the diagnosis subgroup in the Cox regression analyses, diagnosis was entered into the model as a categorical predictor, with double-positive chosen as the reference. The analysis was repeated with AAV as the reference subgroup to evaluate differences between AAV and anti-GBM disease. Proportionality assumption was met for covariates included in the Cox regression analysis. Data are presented as hazards ratios (confidence interval; P value). Graphs were constructed and statistical analysis performed using Prism 5.0 (GraphPad Software, La Jolla, CA) and SPSS version 22.0 (IBM Corp., Armonk, NY).

ase. Proportionality assumption was met for covariates included in the Cox regression analysis. Data are presented as hazards ratios (confidence interval; P value). Graphs were constructed and statistical analysis performed using Prism 5.0 (GraphPad Software, La Jolla, CA) and SPSS version 22.0 (IBM Corp., Armonk, NY). As this was a retrospective study and all treatment decisions were made prior to our assessment, research ethics approval was not required for this report. Disclosure All the authors declared no competing interests. Acknowledgments The data reported here were presented in abstract form at the 17th International Vasculitis and ANCA Workshop in London, UK, in May 2015, and at the American Society of Nephrology Renal Week meeting in Philadelphia, PA, USA, in November 2015. SPM is in receipt of a UK National Institute for Health Research (NIHR) Academic Clinical Lectureship. AT holds a Wellcome Trust Clinical Research Training Fellowship. This work was supported by the NIHR Imperial Biomedical Research Centre. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health. We acknowledge support from research project RVO-VFN64165 at Charles University Hospital, Prague, Czech Republic. see commentary on page 544

IgA nephropathy (IgAN) is the most common primary glomerulopathy worldwide and is an important cause of chronic kidney disease, especially in young people.1 IgAN is associated with a wide range of clinical outcomes. Although 40% of patients will have reached end-stage renal disease within 20 years of diagnosis, 20% of patients will have preserved renal function with only minor urinary abnormalities.2 IgAN is characterized by dominant or codominant IgA-containing immune deposits on renal biopsy.3 The pathophysiology of IgAN associated with characteristic galactose-deficient IgA1 (gd-IgA1)-containing immune deposits is considered to be a 4-hit mechanism.4 However, the link between mesangial IgA deposition and the spectrum of clinical outcomes that are characteristic of IgAN is poorly understood. Therefore, we are limited in our ability to both identify patients for whom immunosuppression therapy is appropriate and develop novel therapeutic strategies. The current repertoire of clinical tools available to predict outcome and guide treatment strategies of IgAN, such as proteinuria or estimated glomerular filtration rate (eGFR), may reflect progressive glomerular scarring as well as active immunologically driven disease.

priate and develop novel therapeutic strategies. The current repertoire of clinical tools available to predict outcome and guide treatment strategies of IgAN, such as proteinuria or estimated glomerular filtration rate (eGFR), may reflect progressive glomerular scarring as well as active immunologically driven disease. Systemic and renal complement activation in IgAN is well documented. However, the role of complement activation in IgAN pathogenesis remains poorly defined. Complement C3 accompanies IgA immunostaining in most diagnostic IgAN biopsies.5 Evidence of complement alternative pathway activation was identified in the plasma of 30% to 75% of IgAN adults6, 7 and correlated with proteinuria and the rate of renal function loss in a cohort of 50 IgAN patients.7 An association was demonstrated between mesangial C3 deposition, decreased serum C3 levels, and doubling of serum creatinine or reaching end-stage renal disease in a cohort of 343 IgAN patients.8 This demonstrated a link between histology and serum markers of complement activation in IgAN pathogenesis. Complement activation by the mannose-binding lectin (MBL) complement pathway in IgAN has also been demonstrated. MBL and MBL-associated serine protease (MASP)-1 were detected in 24% of renal biopsies from IgAN patients, but were identified in less than 3% of biopsies from patients with other forms of glomerulonephritis.9 Deposited MBL and MASP-1 were associated with C3b, C3c, and C5b9, which are markers of alternative and terminal complement pathway activity.9 Although this study did not show associations with markers of clinical outcomes, glomerular MBL and other MBL complement pathway components, including MASP-1/-3 and MASP-2, were found in 25% of a separate cohort of 60 IgAN patients, and these findings correlated with proteinuria and histologic features of severity.10

9 Although this study did not show associations with markers of clinical outcomes, glomerular MBL and other MBL complement pathway components, including MASP-1/-3 and MASP-2, were found in 25% of a separate cohort of 60 IgAN patients, and these findings correlated with proteinuria and histologic features of severity.10 The ability of IgA to activate complement in vitro, predominantly through the alternative pathway, has also been demonstrated. IgA isolated from pooled human plasma triggers complement-dependent lysis of, and properdin deposition on, erythrocytes coated with mouse monoclonal antihuman IgA.11 Furthermore, a rat model of IgA-mediated glomerular inflammation demonstrated that polymeric but not monomeric IgA triggered mesangial C3 deposition and not C4 or C1q deposition.12

uman plasma triggers complement-dependent lysis of, and properdin deposition on, erythrocytes coated with mouse monoclonal antihuman IgA.11 Furthermore, a rat model of IgA-mediated glomerular inflammation demonstrated that polymeric but not monomeric IgA triggered mesangial C3 deposition and not C4 or C1q deposition.12 Recent genetic studies implicate a role for the complement factor H related (FHR) proteins in IgAN.13, 14, 15, 16 The FHR proteins may interfere with the regulatory functions of factor H (fH), the major negative regulator of complement C3 activation.17, 18 This process is referred to as fH deregulation.19, 20 The deletion polymorphism of the genes coding FHR-3 and FHR-1 (delCFHR3-1) is associated with protection from IgAN.13, 14, 16 A meta-analysis of approximately 20,000 individuals of different ethnicities estimated that the inheritance of the minor A allele at single-nucleotide polymorphism rs6677604, which tags the delCFHR3-1 allele, reduced the risk of IgAN disease by 26% in heterozygosity and by 45% in homozygosity.14 Across populations worldwide, delCFHR3-1 frequency exhibits marked differences in a pattern inverse to that of IgAN prevalence.14 An association has also been demonstrated between rs6677604 and histologic IgAN markers. In Chinese patients with IgAN, the rs6677604-A allele was associated with reduced mesangial C3 deposition, high serum fH levels, and low complement C3a levels but was not associated with clinical outcomes.15 Xie et al. showed an association of delCFHR3-1 with reduced segmental sclerosis and tubular atrophy in IgAN.16 Although this was independent of eGFR and proteinuria, no other associations with clinical parameters or outcomes were demonstrated.

d low complement C3a levels but was not associated with clinical outcomes.15 Xie et al. showed an association of delCFHR3-1 with reduced segmental sclerosis and tubular atrophy in IgAN.16 Although this was independent of eGFR and proteinuria, no other associations with clinical parameters or outcomes were demonstrated. By enhancing complement activation in response to mesangial gd-IgA1-containing immune complexes, we hypothesized that fH deregulation influences disease severity in IgAN. In this study, we assessed circulating fH, FHR-1, and FHR-5 levels in IgAN patients who were stratified into cohorts with stable and progressive disease.

nhancing complement activation in response to mesangial gd-IgA1-containing immune complexes, we hypothesized that fH deregulation influences disease severity in IgAN. In this study, we assessed circulating fH, FHR-1, and FHR-5 levels in IgAN patients who were stratified into cohorts with stable and progressive disease. Results Patient cohort The patient cohort characteristics are summarized in Table 1. Using the criteria detailed in the Materials and Methods, 179 patients had progressive IgAN and 89 had stable IgAN. We had insufficient data to categorize 26 patients as having either progressive or stable IgAN. The cohort of patients meeting the criteria for progressive IgAN showed lower eGFR and higher systolic blood pressure than the stable IgAN cohort (Table 1). Consistent with previous reports,21, 22, 23 the median serum IgA and gd-IgA1 levels were higher in patients than in controls (Table 1). The progressive IgAN cohort showed lower median serum IgA and higher median serum gd-IgA1 levels than stable IgAN (Table 1). Similarly, of the patients who had received immunosuppression therapy, serum gd-IgA1 levels were higher in patients with progressive IgAN than stable IgAN after treatment (0.54 units [AU] vs. 0.44 AU, P = 0.04, Supplementary Figure S1A). Consistent with these data, we detected a negative correlation between serum gd-IgA1 levels and eGFRs at the sampling time point (Supplementary Figure S1B). The delCFHR3-1 allele is associated with protection from IgAN.13 However, we did not observe any difference in CFHR3 and CFHR1 copy numbers between patients with stable IgAN and those with progressive IgAN (Table 1).Table 1 Cohort characteristics

1 levels and eGFRs at the sampling time point (Supplementary Figure S1B). The delCFHR3-1 allele is associated with protection from IgAN.13 However, we did not observe any difference in CFHR3 and CFHR1 copy numbers between patients with stable IgAN and those with progressive IgAN (Table 1).Table 1 Cohort characteristics Variable IgA nephropathy patients Healthy controls (n = 161) Entire IgAN (n = 294) Stable IgAN (n = 89) Progressive IgAN (n = 179) Clinical features Male/Female 195/99 53/36 120/59 Caucasian/Non-Caucasian 244/50 78/11 144/35 Median age (range), yr 48.2 (18–84) 47.5 (18–82) 48.3 (19–84) Median eGFR, ml/min per 1.73 m2 52.7 (28.7–82.7) 74.8 (47.7–106) 45.8 (17.4–70.9)a Median urine PCR, mg/mmol 44 (16–117) 39 (13.5–89) 50 (17–144.5) Median antihypertensive drug classes per patient 1.6 1.4 1.6 Patients with ACEi/ARB at enrolment, excluding dialysis patients, % (n) 73.7 (n = 199) 79.8 (n = 71) 77.3 (n = 109) Median systolic/diastolic blood pressure, mm Hg 134 (122–145)/79 (70–88) (n = 286) 128 (116–140)/77.5 (70–85) (n = 88) 136 (124–146)b/79 (70–88) (n = 174) Median follow-up duration, mo 55.0 (22.5–100.8) 71.4 (22.9–177.6) 50.7 (22.2–91.6) Reached ESRD, % 34.9 (n = 103) 0 57.5c (n = 103) History of macroscopic hematuria, % 27.8 (n = 82) 44.9 (n = 40) 20.1c (n = 36) Diagnosis of Henoch-Schonlein purpura, % 6.4 (n = 19) 0 8.9c (n = 16) Laboratory measurements Median serum IgA, g/l 3.4 (2.9–4) (n = 293) 3.7 (3.2–4.3) (n = 88) 3.3 (2.7–3.8)d (n = 178) 2.8 (2.2–3.2)e (n = 57) Median serum gd-IgA1, AU 0.50 (0.41–0.58) (n = 293) 0.47 (0.37–0.55) (n = 88) 0.52 (0.42–0.59)f (n = 178) 0.42 (0.33–0.55)g (n = 57) CFHR1–2 copies/1 copy/no copiesh n 183/101/9 (n = 293) 58/28/3 111/63/4 (n = 178) 85/45/3 (n = 133) % 62.4/34.5/3.1 65.9/31.8/3.4 62.4/35.4/2.2 63.9/33.8/2.3 CFHR3–2 copies/1 copy/no copies n 185/100/8 (n = 293) 59/27/3 112/63/3 (n = 178) % 63.1/34.1/2.7 67/30.7/3.4 62.9/35.4/1.7 Median plasma FHR-1, μg/ml 126.8 (86.4–158.6) 112.5 (74.7–149.3) 132.0 (88.6–162.8) 94.4 (70.5–119.6)i Median plasma fH, μg/ml 153.8 (130.6–187.6) 155.1 (137.4–187.8) 150.1 (126.0–186.9) 152.5 (122.9–189.8) Median plasma FHR-1:fH ratio 0.85 (0.55–1.10) 0.77 (0.46–0.98) 0.89 (0.59–1.16)j 0.68 (0.40–0.86)k Median serum FHR-5, μg/ml 2.74 (2.07–3.64) 2.80 (2.07–3.49) 2.79 (2.08–4.03) 2.46 (1.79–3.67)l (n = 158) ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; AU, arbitrary units; CFHR, complement factor H-related; CI, confidence interval; ESRD, end-stage renal di

59–1.16)j 0.68 (0.40–0.86)k Median serum FHR-5, μg/ml 2.74 (2.07–3.64) 2.80 (2.07–3.49) 2.79 (2.08–4.03) 2.46 (1.79–3.67)l (n = 158) ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; AU, arbitrary units; CFHR, complement factor H-related; CI, confidence interval; ESRD, end-stage renal di sease; fH, factor H; FHR-1/-5, factor H-related protein 1/5; gd-IgA1, galactose-deficient IgA1; PCR, protein-to-creatinine ratio; no, number. Values within parentheses represent interquartile range and number analyzed if less than the respective cohort numbers. a P < 0.0001 versus stable disease. b P = 0.0014 versus stable disease. c P < 0.0001 versus stable disease. d P = 0.0007 versus stable disease. e P < 0.0001 versus entire IgAN cohort. f P = 0.01 versus stable disease. g P = 0.015 versus entire IgAN cohort. h 1 patient had 3 copies of CFHR3 and was excluded from the analysis. i P < 0.0001 versus entire IgAN cohort (difference between medians, 32.4 μg/ml; 95% CI, 19.9–37.6 μg/ml). j P = 0.019 versus stable disease (difference between medians, 0.12; 95% CI, 0.02–0.2). k P < 0.0001 versus entire IgAN cohort (difference between medians, 0.17; 95% CI, 0.12–0.24). l P = 0.041 versus entire IgAN cohort (difference between medians, 0.28 μg/ml; 95% CI, 0.01–0.48 μg/ml).

i P < 0.0001 versus entire IgAN cohort (difference between medians, 32.4 μg/ml; 95% CI, 19.9–37.6 μg/ml). j P = 0.019 versus stable disease (difference between medians, 0.12; 95% CI, 0.02–0.2). k P < 0.0001 versus entire IgAN cohort (difference between medians, 0.17; 95% CI, 0.12–0.24). l P = 0.041 versus entire IgAN cohort (difference between medians, 0.28 μg/ml; 95% CI, 0.01–0.48 μg/ml). Plasma FHR-1 levels and the FHR-1:fH ratio were elevated in IgAN and were associated with progressive disease While fH is the major negative regulator of C3 activation via the complement alternative pathway, FHR-1 is postulated to act as a positive regulator by antagonizing the effect of fH, a process termed fH deregulation.17, 18, 19, 24 To investigate fH deregulation in IgAN, we measured both plasma fH and FHR-1 levels. The median plasma FHR-1 level was increased in IgAN patients compared with that in healthy controls, whereas the plasma fH level did not differ (Table 1). Notably, the relative abundance of these proteins differed between patients and healthy controls, and the FHR-1:fH ratio was significantly increased in IgAN patients (Table 1).

plasma FHR-1 level was increased in IgAN patients compared with that in healthy controls, whereas the plasma fH level did not differ (Table 1). Notably, the relative abundance of these proteins differed between patients and healthy controls, and the FHR-1:fH ratio was significantly increased in IgAN patients (Table 1). Because the presence of the delCFHR3-1 allele will influence circulating FHR-1 levels (most clearly the absence of the protein in delCFHR3-1 homozygotes), we stratified patients according to the CFHR1 gene copy number. Patients with 2 copies of the CFHR1 gene had higher FHR-1 levels compared with genotype-matched healthy controls (Figure 1a). This difference was not observed in patients with 1 CFHR1 gene copy number (Figure 1a). However, the FHR-1:fH ratio was significantly higher in patients than in healthy controls, irrespective of the CFHR1 gene copy number (Figure 1b). FHR-1 was undetectable in deletion homozygotes (n = 12, including 9 patients). Plasma fH levels remained similar between patients and controls when stratified according to the CFHR1 gene copy number (Figure 1c). As previously reported,15, 25 fH levels were higher in delCFHR3-1 homozygotes (Figure 1c).Figure 1 The plasma factor H-related protein 1 (FHR-1) levels and the FHR-1:factor H (fH) ratio are elevated in IgA nephropathy (IgAN) and are associated with a progressive disease. (a) The plasma FHR-1 levels, (b) FHR-1:fH ratio, and (c) fH levels in healthy controls (gray boxes) and IgAN patients (white boxes) stratified according to the CFHR1 gene copy number. Plasma FHR-1 was undetectable in patients with 0 CFHR1 gene copy number. The plasma FHR-1 levels and FHR-1:fH ratio were significantly different (P < 0.001) between individuals with either 1 or 2 CFHR1 gene copy numbers in both the control and IgAN cohorts. (d) Comparison of plasma FHR-1 levels and (e) FHR-1:fH ratio between patients with stable (dashed gray boxes) and those with progressive (dashed white boxes) IgAN stratified according to the CFHR1 gene copy number. (f) Comparison of the FHR-1:fH ratio between patients with stable (dashed gray boxes) and those with progressive (dashed white boxes) IgAN after immunosuppression therapy. The bar represents the median value, box represents the interquartile range, and whiskers represent the range of values. P values derived using Mann-Whitney test.

of the FHR-1:fH ratio between patients with stable (dashed gray boxes) and those with progressive (dashed white boxes) IgAN after immunosuppression therapy. The bar represents the median value, box represents the interquartile range, and whiskers represent the range of values. P values derived using Mann-Whitney test. We next examined if FHR-1 levels and the FHR-1:fH ratio differed between patients with stable IgAN and those with progressive IgAN. Irrespective of the CFHR1 gene copy number, compared with patients with stable IgAN, those with progressive IgAN had significantly elevated plasma FHR-1 levels (Figure 1d) and FHR-1:fH ratios (Figure 1e), whereas fH levels did not differ (Table 1). In addition, the FHR-1:fH ratio was elevated in patients who had progressive compared with stable disease following immunosuppression treatment (Figure 1f).

those with progressive IgAN had significantly elevated plasma FHR-1 levels (Figure 1d) and FHR-1:fH ratios (Figure 1e), whereas fH levels did not differ (Table 1). In addition, the FHR-1:fH ratio was elevated in patients who had progressive compared with stable disease following immunosuppression treatment (Figure 1f). Plasma FHR-1 was negatively correlated with eGFR but remained elevated in IgAN patients with normal eGFR Given that we observed higher FHR-1 levels in patients with progressive IgAN, we next determined if FHR-1 levels were influenced by renal impairment. We stratified patients by the CFHR1 gene copy number and assessed the correlation between FHR-1 levels and eGFR (Figure 2). We detected a negative correlation between plasma FHR-1 levels and eGFR in patients with either 1 (Figure 2a) or 2 (Figure 2b) CFHR1 gene copy number. When we stratified patients into those with eGFR of <30 ml/min per 1.73 m2 and those with eGFR of >60 ml/min per 1.73 m2, FHR-1 levels were significantly higher in those with eGFR of <30 ml/min per 1.73 m2 (Figure 2c and d, 1 and 2 CFHR1 copy numbers, respectively). To determine if FHR-1 levels were higher in IgAN patients before the development of renal impairment, we compared FHR-1 levels between IgAN patients with normal eGFR and healthy controls (Figure 2e). Higher FHR-1 levels were observed in patients with 2 CFHR1 gene copy numbers and normal eGFR than in genotype-matched healthy controls (Figure 2e). This difference was not observed in patients with 1 CFHR1 gene copy number (Figure 2e). We next assessed FHR-1 levels before and after renal transplantation in patients with either biopsy-proven IgAN or autosomal dominant polycystic kidney disease (ADPKD). The cohorts had comparable pretransplant characteristics (Supplementary Table S1), and no patients had a clinical diagnosis of delayed graft function, transplant rejection, or disease recurrence at the sampling time. Both groups showed significant reduction in serum FHR-1 levels after renal transplantation (Figure 2f). Altogether, our data indicate that both the diagnosis of IgAN and eGFR are independently associated with higher FHR-1 levels. Notably, there was no significant correlation between eGFR and plasma fH levels in our IgAN cohort (Supplementary Figure S1C).Figure 2 Factor H-related protein 1 (FHR-1) is negatively correlated with the estimated glomerular filtration rate (eGFR) but remains elevated in IgA nephropathy (IgAN) patients with normal eGFR.

otably, there was no significant correlation between eGFR and plasma fH levels in our IgAN cohort (Supplementary Figure S1C).Figure 2 Factor H-related protein 1 (FHR-1) is negatively correlated with the estimated glomerular filtration rate (eGFR) but remains elevated in IgA nephropathy (IgAN) patients with normal eGFR. Correlation between plasma FHR-1 levels and eGFR after logarithmic transformation in IgAN patients with either 1 (a) or 2 (b) CFHR1 gene copy numbers. P values derived from Spearman’s rank correlation. Plasma FHR-1 levels in IgAN patients with eGFR of <30 (gray boxes) or >60 (white boxes) ml/min per 1.73 m2 stratified according to 1 (c) or 2 (d) CFHR1 gene copy number. (e) Comparison of the plasma FHR-1 levels between healthy controls (gray boxes) and IgAN patients with eGFR of >60 ml/min per 1.73 m2 (white boxes) stratified according to the CFHR1 gene copy number. The bar represents the median value, box represents the interquartile range, and whiskers represent the range of values. P values derived from Mann-Whitney test. (f) Paired FHR-1 levels before (Pre) and after (Post) renal transplantation in a cohort of patients with autosomal dominant polycystic kidney disease (ADPKD) (gray circles, n = 25) and IgAN (white circles, n = 23). All patients were homozygous for the major allele rs6677604, consistent with 2 CFHR1 gene copy numbers.

. (f) Paired FHR-1 levels before (Pre) and after (Post) renal transplantation in a cohort of patients with autosomal dominant polycystic kidney disease (ADPKD) (gray circles, n = 25) and IgAN (white circles, n = 23). All patients were homozygous for the major allele rs6677604, consistent with 2 CFHR1 gene copy numbers. Serum FHR-5 was slightly elevated in IgAN but was not correlated with eGFR Abnormalities in FHR-5 have been shown to be associated with C3 glomerulopathy (C3G), a complement-mediated kidney disease with phenotypic similarities to IgAN.26, 27 In addition, FHR-5 is associated with fH deregulation in vitro.19 We measured serum FHR-5 levels in our IgAN cohort. The median FHR-5 level was higher in IgAN patients than in healthy controls (Table 1), but the magnitude of the difference was small. The median FHR-5 levels were 2.46 and 2.74 μg/ml in the healthy controls and IgAN patients, respectively (difference between medians, 0.28 μg/ml, 95% confidence interval, 0.01–0.48 μg/ml). Furthermore, we did not detect any difference in FHR-5 levels between stable and progressive IgAN patients (Figure 3a and Table 1). However, patients with progressive disease following immunosuppression treatment had significantly higher FHR-5 levels than patients who improved to meet stable IgAN criteria after immunosuppression (Figure 3a). Unlike FHR-1, we did not find an association between serum FHR-5 levels and eGFR (Figure 3b and c). Moreover, we found no significant difference in the FHR-5:fH ratio between patients and healthy controls or between progressive and stable IgAN (Supplementary Figure S2).Figure 3 Serum factor H-related protein 5 (FHR-5) levels are associated with IgA nephropathy (IgAN) severity following immunosuppression therapy and are not correlated with estimated glomerular filtration rate (eGFR). (a) Left panel, Comparison of serum FHR-5 levels between stable (dashed gray boxes) and progressive (dashed white boxes) IgAN. Right panel, Comparison of serum FHR-5 between patients with stable (dashed gray boxes) and those with progressive (dashed white boxes) IgAN after immunosuppression therapy. (b) Correlation between serum FHR-5 levels and eGFR in IgAN patients. (c) Serum FHR-5 levels in IgAN patients with eGFR of <30 (gray boxes) or >60 (white boxes) ml/min per 1.73 m2. The bar represents the median value, box represents the interquartile range, and whiskers represent the range of values. P values derived using Mann-Whitney test.

serum FHR-5 levels and eGFR in IgAN patients. (c) Serum FHR-5 levels in IgAN patients with eGFR of <30 (gray boxes) or >60 (white boxes) ml/min per 1.73 m2. The bar represents the median value, box represents the interquartile range, and whiskers represent the range of values. P values derived using Mann-Whitney test. Serum FHR-5 levels correlated with histologic markers of renal injury To explore the significance of the changes in serum FHR-5 levels, we assessed the correlation between FHR-5 levels and validated markers of histologic injury in IgAN, according to the Oxford classification.28 Serum FHR-5 levels at recruitment were significantly higher in patients with endocapillary hypercellularity score E1 at diagnosis than in those with no endocapillary hypercellularity (E0) (Figure 4a, right panel). Serum FHR-5 levels in IgAN patients with (M1) and without (M0) renal biopsy evidence of mesangial hypercellularity did not differ (Figure 4a, left panel). The total MEST score is calculated by adding the scores for mesangial hypercellularity (M), endocapillary hypercellularity (E), segmental sclerosis (S), and tubular atrophy (T). Serum FHR-5 levels were higher in IgAN patients with MEST score of 4 than in those with MEST score of 1 (Figure 4b). Unlike FHR-5, our data demonstrated that FHR-1 levels were influenced by eGFR. Therefore, we did not consider it valid to assess associations between histologic markers and plasma FHR-1 levels because eGFR would differ between the diagnostic renal biopsy and study plasma in many patients.Figure 4 Serum factor H-related protein 5 (FHR-5) levels are correlated with histologic markers of disease severity in IgA nephropathy (IgAN). (a) Left panel, Serum FHR-5 levels in IgAN patients without (M0, gray box) and with (M1, white box) renal biopsy evidence of mesangial hypercellularity (denoted M). Right panel, Serum FHR-5 levels in IgAN patients without (E0, gray box) and with (E1, white box) biopsy evidence of endocapillary hypercellularity (denoted E). (b) Serum FHR-5 levels in IgAN patients with diagnostic renal biopsy MEST scores of 1 or less (gray box) or at least 4 (white box). The bar represents the median value, box represents the interquartile range, and whiskers represent the range of values. P values derived using Mann-Whitney test.

f fH, a phenomenon referred to as fH deregulation. We considered that the association between delCFHR3-1 alleles and IgAN can be explained by fH deregulation. We hypothesized that reduced (or absent) FHR-1 levels result in a reduction or absence of fH deregulation and consequently less complement-mediated renal injury. In this study, we first explored the association between FHR-1 and fH levels in patients with either stable or progressive IgAN. We found that FHR-1 levels and importantly, the FHR-1:fH ratio, were higher in IgAN patients than in healthy controls. These data are replicated in a separate IgAN cohort (Tortajada A, Gutierrez E, Goicoechea de Jorge E, et al. Elevated factor H-related 1 and occurrence of factor H pathogenic variants in IgA nephropathy. Kidney International, submitted for publication). In addition, irrespective of the CFHR1 gene copy number, patients with progressive IgAN had significantly elevated plasma FHR1/fH ratios compared to patients with stable IgAN. These data are consistent with our hypothesis that reduced fH deregulation is associated with favorable outcomes in IgAN. We categorized the patient cohort into those with progressive IgAN and those with stable IgAN. Our criteria were designed to enable us to identify patients with immunologically active disease. The criteria included eGFR loss without additional renal pathology and histologic features of glomerular inflammation, such as the presence of endocapillary hypercellularity and cellular crescents. Although this inevitably excluded a subset of patients (those we could not reliably categorize as either stable or progressive at study entry), our robust classification enabled us to detect differences in FHR-1 levels and the FHR-1:fH ratio. Although changes in these parameters correlated with worse disease outcomes, the levels showed large overlap between the groups, precluding their use for patient stratification. In addition, our cohort included 9 individuals (3 with stable IgAN, 4 with progressive IgAN, and 2 who did not fulfil the criteria) with delCFHR3-1 allele homozygosity, demonstrating that even in the complete absence of FHR-1 and FHR-3, renal injury that requires renal biopsy and hospital follow-up can occur. This is consistent with previous studies16 and indicates that factors independent of fH deregulation can drive the disease.

fil the criteria) with delCFHR3-1 allele homozygosity, demonstrating that even in the complete absence of FHR-1 and FHR-3, renal injury that requires renal biopsy and hospital follow-up can occur. This is consistent with previous studies16 and indicates that factors independent of fH deregulation can drive the disease. We did not measure FHR-3 levels because of the lack of a reliable assay. FHR-3 is not clearly associated with fH deregulation, and its role remains unclear; it can interact with the meningococcus fH-binding protein and through competition with fH, influence the complement-mediated clearance of meningococcal strains and hence meningococcal disease severity.32 Through interaction with its ligand C3d, FHR-3 may influence B-cell regulation through a B-cell receptor complex (CD19/CD21/CD81), but the association with IgAN is unclear.33 Notably, despite the interaction with C3d, this effect was not observed for FHR-1.

ingococcal strains and hence meningococcal disease severity.32 Through interaction with its ligand C3d, FHR-3 may influence B-cell regulation through a B-cell receptor complex (CD19/CD21/CD81), but the association with IgAN is unclear.33 Notably, despite the interaction with C3d, this effect was not observed for FHR-1. We found a negative correlation between eGFR and plasma FHR-1 levels measured at study recruitment. When patients with normal eGFR were compared with genotype-matched healthy controls, patients with 2 CFHR1 gene copy numbers and normal eGFR had higher FHR-1 levels. Both IgAN patients and ADPKD patients showed reduced FHR-1 levels coincidental with increased eGFRs after renal transplantation. This indicated that the diagnosis of IgAN and eGFR are independently associated with higher FHR-1 levels. Moreover, this conclusion is supported by the findings in a study that assessed FHR-1 levels in IgAN and polycystic renal disease (Tortajada A, Gutierrez E, Goicoechea de Jorge E, et al. Elevated factor H-related 1 and occurrence of factor H pathogenic variants in IgA nephropathy. Kidney International, submitted for publication). Although we consider the association to be robust, we currently do not know why FHR-1 increases as eGFR decreases. Nevertheless, this would be predicted to further enhance fH deregulation and aggravate disease. It is predicted that this could be applicable to any glomerular pathology in which there is complement-mediated injury and that there are more general implications beyond IgAN. Because FHR-1 levels were influenced by eGFR and diagnostic renal biopsy was not coincident with the timing of the study blood sample in our cohort, we did not analyze the association between FHR-1 and histologic changes in the renal biopsy. A prospective study will be required to clarify the contribution of FHR-1 to histologic changes before decreases in eGFR.

R and diagnostic renal biopsy was not coincident with the timing of the study blood sample in our cohort, we did not analyze the association between FHR-1 and histologic changes in the renal biopsy. A prospective study will be required to clarify the contribution of FHR-1 to histologic changes before decreases in eGFR. As published data showed that FHR-5 can mediate fH deregulation19 and studies indicated the strong association between FHR-5 mutation and C3G,26, 30, 31 we measured serum FHR-5 levels and the FHR-5:fH ratio. Notably, there are strong phenotypic similarities between familial C3G associated with FHR-5 mutations and IgAN.27 Despite a significant increase in serum FHR-5 levels between IgAN patients and healthy controls, the magnitude of the difference was very small and of doubtful biological significance. In addition, we did not identify any difference in FHR-5 levels between patients with stable disease and those with progressive disease. However, when we analyzed only patients treated with immunosuppression therapy, FHR-5 levels were higher in those with ongoing progressive IgAN. The significance of this is unclear and requires confirmation. Unlike the FHR-1:fH ratio, we did not detect any correlation between the FHR-5:fH ratio and disease outcome. Because eGFR did not influence the serum FHR-5 levels, we assessed the correlations between serum FHR-5 levels and validated histologic markers of renal injury in IgAN. Serum FHR-5 levels were associated with endocapillary hypercellularity independent of mesangial hypercellularity, a histologic marker of active inflammation.34 They were also associated with higher overall Oxford classification of IgAN MEST scores.35 To further understand these associations, it would be necessary to analyze the pattern and degree of FHR-5 deposition in renal tissues. In this respect, it is notable that FHR-5 was detected in association with complement C3 in patients with IgAN. Notably, the association between FHR-5 and C3 was also observed in membranous nephropathy, lupus nephritis, and postinfectious nephritis, indicating that this phenomenon is not specific to IgAN.36

al tissues. In this respect, it is notable that FHR-5 was detected in association with complement C3 in patients with IgAN. Notably, the association between FHR-5 and C3 was also observed in membranous nephropathy, lupus nephritis, and postinfectious nephritis, indicating that this phenomenon is not specific to IgAN.36 The limitations of our dataset include the difference in timing between diagnostic renal biopsy and study sample collection and the lack of serial blood samples from patients. Additionally, it would be important to investigate the relative amounts of C3, fH, FHR-1, and FHR-5 in renal tissues in IgAN.

al tissues. In this respect, it is notable that FHR-5 was detected in association with complement C3 in patients with IgAN. Notably, the association between FHR-5 and C3 was also observed in membranous nephropathy, lupus nephritis, and postinfectious nephritis, indicating that this phenomenon is not specific to IgAN.36 The limitations of our dataset include the difference in timing between diagnostic renal biopsy and study sample collection and the lack of serial blood samples from patients. Additionally, it would be important to investigate the relative amounts of C3, fH, FHR-1, and FHR-5 in renal tissues in IgAN. Current assessment of IgAN patients involves the use of nonspecific clinical and histologic markers to identify patients likely to improve with immunosuppression therapy. However, the benefit of the currently recommended 6-month corticosteroid treatment for persistently proteinuric, noncrescentic IgAN37 has not been conclusively demonstrated.38 Our data, together with those of a separate cohort (Tortajada A, Gutierrez E, Goicoechea de Jorge E, et al. Elevated factor H-related 1 and occurrence of factor H pathogenic variants in IgA nephropathy. Kidney International, submitted for publication), associate circulating FHR-1 levels and the FHR-1:fH ratio with IgAN severity. Our findings need corroboration with histologic data but nevertheless suggest that fH deregulation contributes to IgAN progression and support further investigation into this hypothesis. If it can be shown that complement-mediated kidney injury in IgAN is influenced by FHR-1 and mechanistically understood, the modulation of FHR-1 might be therapeutic in progressive IgAN.

but nevertheless suggest that fH deregulation contributes to IgAN progression and support further investigation into this hypothesis. If it can be shown that complement-mediated kidney injury in IgAN is influenced by FHR-1 and mechanistically understood, the modulation of FHR-1 might be therapeutic in progressive IgAN. Materials and Methods Study cohort and clinical measurements The Causes and Predictors of Outcome in IgA Nephropathy study is a retrospective cohort UK study of patients with biopsy-proven IgAN ethically approved by the UK National Research Ethics Service Committee (14/LO/0155). The inclusion criteria were availability of the renal biopsy report and relevant clinical data. In this study, 334 patients were recruited, and after excluding 40 patients, 294 were analyzed. The reasons for exclusion were alternative diagnosis (n = 4), no biopsy immunostaining completed (n = 5), no biopsy report available (n = 3), and no creatinine measurement at enrollment (n = 28).

and relevant clinical data. In this study, 334 patients were recruited, and after excluding 40 patients, 294 were analyzed. The reasons for exclusion were alternative diagnosis (n = 4), no biopsy immunostaining completed (n = 5), no biopsy report available (n = 3), and no creatinine measurement at enrollment (n = 28). Progressive disease was considered if at least one of the following criteria was present: (i) progression to end-stage renal disease without histologic evidence of a second pathology causing renal impairment, (ii) renal biopsy evidence of endocapillary hypercellularity, (iii) renal biopsy evidence of cellular and/or fibrocellular crescents; (iv) treatment with immunosuppressants (including corticosteroids) for native IgAN, (v) clinical Henoch-Schonlein purpura, unless spontaneous resolution and >20 years of follow-up with “stable” criteria, (vi) 50% loss of eGFR or an average annual eGFR loss of >5 ml/min per 1.73 m2 without evidence of a second pathology causing renal impairment. Stable disease was considered if none of the criteria for progressive IgAN were met and if all of the following criteria were met: (i) urine protein-to-creatinine ratio of <100 units or daily proteinuria of <1 g/24 h, (ii) combined Oxford classification MEST score of <3, and (iii) an average annual eGFR loss of <3 ml/min per 1.73 m2.

was considered if none of the criteria for progressive IgAN were met and if all of the following criteria were met: (i) urine protein-to-creatinine ratio of <100 units or daily proteinuria of <1 g/24 h, (ii) combined Oxford classification MEST score of <3, and (iii) an average annual eGFR loss of <3 ml/min per 1.73 m2. All patients in the transplantation cohorts received a renal transplant and underwent clinical follow-up at Imperial College Healthcare NHS Trust. They received posttransplant immunosuppression therapy and clinical care as per local guidelines. All ADPKD patients had a radiological diagnosis. All transplant patients provided consent for the storage of serum and plasma samples that were surplus to clinical diagnostic requirements and their subsequent use for research. Blood samples were obtained within 4 weeks before and between 12 and 16 weeks after transplantation. Two transplant ADPKD patients, both of whom had the AA rs6677604 single-nucleotide polymorphism genotype, had undetectable FHR-1 levels before and after transplantation. Healthy control samples were obtained from healthy volunteer donors and from members of the TwinsUK cohort,39 a cohort of twins of Caucasian ethnicity. We randomly selected samples from only 1 individual of each pair of twins included in the cohort. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration Creatinine Equation.40

Healthy control samples were obtained from healthy volunteer donors and from members of the TwinsUK cohort,39 a cohort of twins of Caucasian ethnicity. We randomly selected samples from only 1 individual of each pair of twins included in the cohort. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration Creatinine Equation.40 Assessment of CFHR3 and CFHR1 gene copy number DNA was extracted from whole blood samples using QIAamp DNA Blood Mini Kits (Qiagen, Hilden, Germany). Quantitative real-time polymerase chain reaction (PCR) was performed using the ViiA Real-Time PCR System (Applied Biosystems, Foster City, CA). Copy number variation within the CFHR3 and CFHR1 genes was assessed using the Taqman Copy Number Real-Time Detection System (Applied Biosystems). Copy number variation calls were determined using the Copy Caller Software (Applied Biosystems). Assay readings were normalized to control samples, and the values were presented as mean ± SD. All probes were validated using genomic DNA from healthy controls with either heterozygous or homozygous polymorphic deletion of the CFHR1 and CFHR3 genes. The CFHR1 gene copy number in the renal transplant cohort was inferred from the rs6677604 genotype, which is in linkage disequilibrium with the CFHR1 and CFHR3 gene copy number.16 Genotyping was performed using the Taqman genotyping assays (Applied Biosystems).

ozygous polymorphic deletion of the CFHR1 and CFHR3 genes. The CFHR1 gene copy number in the renal transplant cohort was inferred from the rs6677604 genotype, which is in linkage disequilibrium with the CFHR1 and CFHR3 gene copy number.16 Genotyping was performed using the Taqman genotyping assays (Applied Biosystems). Measurement of serum IgA and gd-IgA levels Serum IgA levels were measured using enzyme-linked immunosorbent assay (ELISA) as previously described.41 The capture antibody was the F(ab’)2 fragment goat antihuman IgA (Jackson ImmunoResearch, West Grove, PA), and the detection antibody was the F(ab’)2 fragment biotinylated goat antihuman IgA1 (Jackson ImmunoResearch). Serum gd-IgA1 levels were measured using a lectin-based ELISA as previously described.41 The capture antibody was a polyclonal rabbit antihuman IgA (Dako, Glostrup, Denmark). The detection involved Helix aspersa agglutinin-biotin (Sigma, Darmstadt, Germany), followed by poly-streptavidin horseradish peroxidase (Pierce, Waltham, MA). The intraclass correlation coefficient for the IgA assay was 0.74 (95% confidence interval, 0.63–0.83), and that for the gd-IgA1 assay was 0.89 (95% confidence interval, 0.73–0.95).

d Helix aspersa agglutinin-biotin (Sigma, Darmstadt, Germany), followed by poly-streptavidin horseradish peroxidase (Pierce, Waltham, MA). The intraclass correlation coefficient for the IgA assay was 0.74 (95% confidence interval, 0.63–0.83), and that for the gd-IgA1 assay was 0.89 (95% confidence interval, 0.73–0.95). Measurement of plasma fH and FHR-1 levels and serum FHR-5 levels We used a sandwich ELISA designed by Tortajada et al. (Tortajada A, Gutierrez E, Goicoechea de Jorge E, et al. Elevated factor H-related 1 and occurrence of factor H pathogenic variants in IgA nephropathy. Kidney International, submitted for publication) to measure the plasma fH and FHR-1 levels. Although FHR-1 and FHR-2 can form homodimers, heterodimers, and heterooligomeric molecules in vivo, the levels detected by this ELISA are referred to as FHR-1 levels because FHR-1 is the major component in these complexes. The capture antibody was a rabbit polyclonal antibody that recognizes both fH and FHR-1. fH was detected with a mouse monoclonal anti-fH antibody that recognizes SCR10 and SCR11 of fH. FHR-1 was detected using a mouse monoclonal antibody that recognizes an epitope within SCR1 and SCR2 of FHR-1 (both provided by Professor Santiago Rodriguez de Cordoba, Madrid). The interassay coefficient of variation was 8.7 for fH ELISA and 9.9 for FHR-1 ELISA. FHR-5 levels were measured using ELISA. The capture antibody was a rabbit monoclonal anti-FHR-5 antibody (Abcam, Cambridge, UK). FHR-5 was detected using a mouse monoclonal anti-FHR-5 antibody (Abcam). The interassay coefficient of variation was 12.1.

Measurement of plasma fH and FHR-1 levels and serum FHR-5 levels We used a sandwich ELISA designed by Tortajada et al. (Tortajada A, Gutierrez E, Goicoechea de Jorge E, et al. Elevated factor H-related 1 and occurrence of factor H pathogenic variants in IgA nephropathy. Kidney International, submitted for publication) to measure the plasma fH and FHR-1 levels. Although FHR-1 and FHR-2 can form homodimers, heterodimers, and heterooligomeric molecules in vivo, the levels detected by this ELISA are referred to as FHR-1 levels because FHR-1 is the major component in these complexes. The capture antibody was a rabbit polyclonal antibody that recognizes both fH and FHR-1. fH was detected with a mouse monoclonal anti-fH antibody that recognizes SCR10 and SCR11 of fH. FHR-1 was detected using a mouse monoclonal antibody that recognizes an epitope within SCR1 and SCR2 of FHR-1 (both provided by Professor Santiago Rodriguez de Cordoba, Madrid). The interassay coefficient of variation was 8.7 for fH ELISA and 9.9 for FHR-1 ELISA. FHR-5 levels were measured using ELISA. The capture antibody was a rabbit monoclonal anti-FHR-5 antibody (Abcam, Cambridge, UK). FHR-5 was detected using a mouse monoclonal anti-FHR-5 antibody (Abcam). The interassay coefficient of variation was 12.1. Statistical analysis Normally distributed continuous variables were tested using unpaired t-test, and pre- and posttransplant levels were compared using paired t-test. Continuous variables with skewed distribution were tested using Mann-Whitney U test, and confidence intervals for differences between medians were calculated using the Hodges-Lehmann method. Categorical data was tested using chi-square test and Fisher’s exact test (samples, <10). We applied Pearson’s correlations and simple linear regression to log eGFR and FHR-1, and Spearman’s rank correlation to eGFR and gd-IgA1. A P value <0.05 was considered to be statistically significant. We used GraphPad Prism version 6.00 for Windows (GraphPad Software, La Jolla, CA) for all analyses.

t (samples, <10). We applied Pearson’s correlations and simple linear regression to log eGFR and FHR-1, and Spearman’s rank correlation to eGFR and gd-IgA1. A P value <0.05 was considered to be statistically significant. We used GraphPad Prism version 6.00 for Windows (GraphPad Software, La Jolla, CA) for all analyses. Disclosure All the authors declare no competing interests. Collaborator List and Affiliations Dr. A. Ahmed, Royal Preston Hospital, Lancashire, UK. Dr. L. Baines, Freeman Hospital, Newcastle upon Tyne, UK. Dr. C. Bingham, Royal Devon and Exeter Hospital, Exeter, UK. Prof S. Bhandari, Hull Royal Infirmary, Hull, UK. Dr. A. Bow, Cumberland Infirmary, Carlisle, UK. Dr. A. Chan, Broomfield Hospital, Chelmsford, Essex, UK. Dr. G. Cserep, Colchester General Hospital, Colchester, Essex, UK. Dr. D. de Takats, Royal Stoke University Hospital, Stoke-on-Trent, UK. Dr. S. Dickinson, Royal Cornwall Hospital, Truro, Cornwall, UK. Dr. C. Goldsmith, Aintree University Hospital, Liverpool, UK. Dr. S. Lawman, Royal Sussex County Hospital, Brighton, UK. Dr. S. Mitra, Manchester Royal Infirmary, Manchester, UK. Dr. R. Montero, Guys Hospital, London, UK. Dr. A. Mooney, St James’s University Hospital, Leeds, UK. Dr. A. Power, Southmead Hospital, Bristol, UK. Dr. C. Wroe, James Cook University Hospital, Middlesbrough, UK.

Dr. C. Goldsmith, Aintree University Hospital, Liverpool, UK. Dr. S. Lawman, Royal Sussex County Hospital, Brighton, UK. Dr. S. Mitra, Manchester Royal Infirmary, Manchester, UK. Dr. R. Montero, Guys Hospital, London, UK. Dr. A. Mooney, St James’s University Hospital, Leeds, UK. Dr. A. Power, Southmead Hospital, Bristol, UK. Dr. C. Wroe, James Cook University Hospital, Middlesbrough, UK. Supplementary Material Figure S1 (A) Serum IgA (left panel) and galactose-deficient IgA1 (gd-IgA1; right panel) levels in patients with stable (gray dashed box) and progressive (white dashed box) IgA nephropathy (IgAN) after immunosuppression therapy. The progressive IgAN cohort shows higher median serum gd-IgA1 levels. (B) Correlation of estimated glomerular filtration rate (eGFR) and serum gd-IgA1 levels in IgAN patients (n = 293). A negative correlation was calculated with Spearman’s rank correlation, and the black line represents the correlation equation. (C) Comparison of serum eGFR and plasma factor H (fH) at the same sampling point in IgAN patients (n = 294) shows a small and insignificant correlation between eGFR and plasma fH levels.

293). A negative correlation was calculated with Spearman’s rank correlation, and the black line represents the correlation equation. (C) Comparison of serum eGFR and plasma factor H (fH) at the same sampling point in IgAN patients (n = 294) shows a small and insignificant correlation between eGFR and plasma fH levels. Figure S2 The plasma factor H-related protein 5 (FHR-5):factor H (fH) ratio is not associated with IgA nephropathy (IgAN). (A) The FHR-5:fH ratio in healthy controls and IgAN patients. We did not identify a significant difference in plasma FHR-5:fH ratio between healthy controls (gray box) and IgAN patients (white box). (B) The FHR-5:fH ratio in stable and progressive IgAN. There was no significant difference in the plasma FHR-5:fH ratio between patients with stable (gray boxes) and those with progressive (white boxes) IgAN for the same CFHR1 genotype. Table S1 Cohort characteristics for patients with IgA nephropathy (IgAN) or autosomal dominant polycystic kidney disease (ADPKD) who received a renal transplant. Values within parentheses represent interquartile range. ESRD, end-stage renal disease. aP < 0.001 compared with the pretransplant levels. *rs6677604 tags the CFHR3-1 deletion.

aracteristics for patients with IgA nephropathy (IgAN) or autosomal dominant polycystic kidney disease (ADPKD) who received a renal transplant. Values within parentheses represent interquartile range. ESRD, end-stage renal disease. aP < 0.001 compared with the pretransplant levels. *rs6677604 tags the CFHR3-1 deletion. Acknowledgments We acknowledge the support from the National Institute for Health Research (NIHR) Biomedical Research Centre based at Imperial College Healthcare National Health Service Trust and Imperial College London and from the NIHR Clinical Research Network. The views expressed are those of the authors and not necessarily those of the National Health Service, the National Institute for Health research, or the Department of Health. This work was supported by funds recovered from the Medical Research Council (MR/K01353X/1). MCP is a Wellcome Trust Senior Fellow in Clinical Science (fellowship WT082291MA). NMT is funded by a Kidney Research UK PhD Clinical Research Fellowship. TwinsUK is funded by the Wellcome Trust, the Medical Research Council, the European Union and the National Institute for Health Research–funded BioResource, the Clinical Research Facility, and the Biomedical Research Centre based at Guy's and St Thomas' National Health Service Foundation Trust in partnership with King's College London. We would like to thank all the Causes and Predictors of Outcome in IgA Nephropathy study patients and local research staff.

This work was supported by funds recovered from the Medical Research Council (MR/K01353X/1). MCP is a Wellcome Trust Senior Fellow in Clinical Science (fellowship WT082291MA). NMT is funded by a Kidney Research UK PhD Clinical Research Fellowship. TwinsUK is funded by the Wellcome Trust, the Medical Research Council, the European Union and the National Institute for Health Research–funded BioResource, the Clinical Research Facility, and the Biomedical Research Centre based at Guy's and St Thomas' National Health Service Foundation Trust in partnership with King's College London. We would like to thank all the Causes and Predictors of Outcome in IgA Nephropathy study patients and local research staff. We would like to thank and acknowledge Professor Santiago Rodriguez de Cordoba and Dr. Agustin Torajada, Department of Cellular and Molecular Medicine, Center for Biological Research and Center for Biomedical Network Research on Rare Diseases, Madrid, Spain and Dr. Elena Goicoechea de Jorge, Departamento de Inmunologia, Facultad de Medicina, Universidad Complutense, Madrid, Spain for providing monoclonal antibodies, standards, and methods for the fH, FHR-1, and FHR-5 ELISA. see commentary on page 790

We would like to thank and acknowledge Professor Santiago Rodriguez de Cordoba and Dr. Agustin Torajada, Department of Cellular and Molecular Medicine, Center for Biological Research and Center for Biomedical Network Research on Rare Diseases, Madrid, Spain and Dr. Elena Goicoechea de Jorge, Departamento de Inmunologia, Facultad de Medicina, Universidad Complutense, Madrid, Spain for providing monoclonal antibodies, standards, and methods for the fH, FHR-1, and FHR-5 ELISA. see commentary on page 790 Figure S1. (A) Serum IgA (left panel) and galactose-deficient IgA1 (gd-IgA1; right panel) levels in patients with stable (gray dashed box) and progressive (white dashed box) IgA nephropathy (IgAN) after immunosuppression therapy. The progressive IgAN cohort shows higher median serum gd-IgA1 levels. (B) Correlation of estimated glomerular filtration rate (eGFR) and serum gd-IgA1 levels in IgAN patients (n = 293). A negative correlation was calculated with Spearman’s rank correlation, and the black line represents the correlation equation. (C) Comparison of serum eGFR and plasma factor H (fH) at the same sampling point in IgAN patients (n = 294) shows a small and insignificant correlation between eGFR and plasma fH levels.

293). A negative correlation was calculated with Spearman’s rank correlation, and the black line represents the correlation equation. (C) Comparison of serum eGFR and plasma factor H (fH) at the same sampling point in IgAN patients (n = 294) shows a small and insignificant correlation between eGFR and plasma fH levels. Figure S2. The plasma factor H-related protein 5 (FHR-5):factor H (fH) ratio is not associated with IgA nephropathy (IgAN). (A) The FHR-5:fH ratio in healthy controls and IgAN patients. We did not identify a significant difference in plasma FHR-5:fH ratio between healthy controls (gray box) and IgAN patients (white box). (B) The FHR-5:fH ratio in stable and progressive IgAN. There was no significant difference in the plasma FHR-5:fH ratio between patients with stable (gray boxes) and those with progressive (white boxes) IgAN for the same CFHR1 genotype. Table S1. Cohort characteristics for patients with IgA nephropathy (IgAN) or autosomal dominant polycystic kidney disease (ADPKD) who received a renal transplant. Values within parentheses represent interquartile range. ESRD, end-stage renal disease. aP < 0.001 compared with the pretransplant levels. *rs6677604 tags the CFHR3-1 deletion. Supplementary material is linked to the online version of the paper at www.kidney-international.org.

In the 1960s, it was established that the assembly of collagen into its mature, triple helical form requires prolyl hydroxylation, which is the conversion of proline residues in procollagen to hydroxyproline. The enzymes responsible were subsequently identified as collagen prolyl-4-hydroxylases (CP4Hs),1, 2 which are members of the superfamily of 2-oxoglutarate (2-OG)–dependent dioxygenases that have an Fe(II) atom at the active site and require oxygen and 2-OG as cosubstrates. Another important group of prolyl 4-hydroxylases has more recently been identified: prolyl hydroxylase domain (PHD) enzymes. These are structurally related to CP4Hs and belong to the superfamily of 2-OG−dependent dioxygenases. They act as key oxygen sensors in metazoans, controlling the master regulator hypoxia-inducible factor (HIF) through the hydroxylation of HIF-α subunits.3 Consistent with their role as oxygen sensors, the Km for oxygen of the PHD enzymes is substantially higher than that of CP4Hs.4 PHDs are under intensive investigation as potential therapeutic targets to promote erythropoiesis and ameliorate ischemic injury.5, 6 Several companies have developed PHD inhibitors with structural similarity to 2-OG, and these are being tested in clinical trials. The extent to which these may have off-target effects on CP4Hs is largely unexplored.

tigation as potential therapeutic targets to promote erythropoiesis and ameliorate ischemic injury.5, 6 Several companies have developed PHD inhibitors with structural similarity to 2-OG, and these are being tested in clinical trials. The extent to which these may have off-target effects on CP4Hs is largely unexplored. Prolyl 4-hydroxylation also occurs in other proteins besides collagen and HIF-α subunits but has been less extensively investigated. One important example is the complement protein complex C1q.7, 8, 9 C1q comprises 18 polypeptide chains, assembled as 6 heterotrimers containing C1q A, B, and C chains, with an N-terminal triple helical collagen-like structure and a C-terminal globular region.10 C1q interacts with complement fixation sites of Igs to initiate complement activation and is mainly produced by bone marrow–derived cells.11, 12, 13 The importance of C1q is underlined by the fact that C1q deficiency causes susceptibility to infections and the autoimmune disease systemic lupus erythematosus.14 In this study, we aimed to determine whether PHD enzymes were responsible for the hydroxylation of C1q and whether small molecules developed as HIF activators may affect C1q.

Prolyl 4-hydroxylation also occurs in other proteins besides collagen and HIF-α subunits but has been less extensively investigated. One important example is the complement protein complex C1q.7, 8, 9 C1q comprises 18 polypeptide chains, assembled as 6 heterotrimers containing C1q A, B, and C chains, with an N-terminal triple helical collagen-like structure and a C-terminal globular region.10 C1q interacts with complement fixation sites of Igs to initiate complement activation and is mainly produced by bone marrow–derived cells.11, 12, 13 The importance of C1q is underlined by the fact that C1q deficiency causes susceptibility to infections and the autoimmune disease systemic lupus erythematosus.14 In this study, we aimed to determine whether PHD enzymes were responsible for the hydroxylation of C1q and whether small molecules developed as HIF activators may affect C1q. Results Prolyl hydroxylase inhibitors, but not hypoxia, reduce C1q secretion Although it is established that active C1q requires prolyl hydroxylation,7, 15 the enzymes responsible have not been previously identified. This raised the possibility that PHD enzymes are involved in C1q hydroxylation, which would have important implications for pursuing PHDs as therapeutic targets. Macrophages are a major source of C1q. To investigate the potential role of PHDs in C1q secretion by human cells, we stimulated human THP-1–derived macrophages (TDMs) with interferon-γ under normoxic or hypoxic conditions. Interestingly, we found that exposure to 1% O2, a hypoxia level that results in substantial HIF stabilization by decreasing the activity of PHDs, did not significantly reduce the secretion of C1q into the culture medium relative to that found in cultures under normoxia (Figure 1a). Next, we investigated the effect of dimethyloxalylglycine (DMOG), a cell-permeable 2-OG analog that is widely used as an HIF activator.16 DMOG is a prodrug; its hydrolysis product N-oxalylglycine is a broad spectrum 2-OG oxygenase inhibitor, which competes with 2-OG for binding to both PHDs and CP4Hs. DMOG potently inhibited the secretion of C1q; notably, this was observed at a DMOG concentration that did not result in detectable HIF-1α stabilization in normoxia by Western blot (Figure 1b). We also examined primary monocyte-derived macrophages (MDMs) from human blood and observed that DMOG efficiently reduced C1q levels in response to interferon-γ or lipopolysaccharide (Supplementary Figure S1A and B).Figure 1 C1q secretion by THP 1–derived macrophages (TDMs) is inhibited by prolyl hydroxylase domain inhibitors. (a) TDMs were treated with interferon (IFN)-γ (10 ng/ml) or left untreated (U) and exposed to either normoxia (20% O2) or hypoxia (1% O2) for 16 hours. C1q secretion in the culture supernatants was measured using enzyme-linked immunosorbent assay (ELISA). Each point represents 1 experiment, and data are shown as means ± SEM. No significant differences were observed in IFN-γ–treated TDMs under normoxia versus hypoxia.

normoxia (20% O2) or hypoxia (1% O2) for 16 hours. C1q secretion in the culture supernatants was measured using enzyme-linked immunosorbent assay (ELISA). Each point represents 1 experiment, and data are shown as means ± SEM. No significant differences were observed in IFN-γ–treated TDMs under normoxia versus hypoxia. (b–e) C1q production in TDMs was induced with IFN-γ (10 ng/ml) in the presence or absence of 16-hour treatment with dimethyloxalylglycine (DMOG; 62 μM, b), L-mimosine (250 μM, c), FG0041 (10 μM, d), or roxadustat (2.5 μM, e), as measured using ELISA. Protein levels relative to IFN-γ–treated controls are shown. Bar graphs show means ± SEM of 4–6 experiments. *** P < 0.001 versus IFN-γ alone. Hypoxia-inducible factor (HIF)-1α proteins were analyzed using Western blotting, with α-tubulin or β-actin used as the loading controls. We next examined 3 other small molecules that have been shown to inhibit PHD enzymes and stabilize HIF-1α: L-mimosine, an iron chelator,17, 18 and two compounds structurally related to N-oxalylglycine, roxadustat (FG4592)19, 20 and FG0041.21 All 3 molecules decreased C1q secretion by macrophages in a concentration-dependent manner, and C1q levels were significantly reduced at doses that were lower than (L-mimosine and FG0041) or similar (roxadustat) to those required to stabilize HIF-1α efficiently (Figure 1c–e).

ne, roxadustat (FG4592)19, 20 and FG0041.21 All 3 molecules decreased C1q secretion by macrophages in a concentration-dependent manner, and C1q levels were significantly reduced at doses that were lower than (L-mimosine and FG0041) or similar (roxadustat) to those required to stabilize HIF-1α efficiently (Figure 1c–e). Collagen prolyl hydroxylases are required for C1q secretion C1q secretion is relatively resistant to hypoxia and is sensitive to some small molecule prolyl hydroxylase inhibitors at doses below those required to stabilize HIF-1α. These findings implied that C1q hydroxylation was unlikely to be catalyzed by PHDs and that it may be mediated by CP4Hs. To further investigate this possibility, we examined whether the CP4H1-specific subunit prolyl 4-hydroxylase alpha-1 (P4HA1) was expressed in TDMs and MDMs and compared with those in fibroblasts, which are specialized collagen-producing cells. Protein analysis by Western blotting revealed that PHD1 to PHD3 and P4HA1 were expressed in all these cells (Figure 2a).Figure 2 C1q secretion is reduced after prolyl 4-hydroxylase alpha-1 (P4HA1) knockdown. (a) Expression levels of prolyl hydroxylase domains (PHDs) in different cell types were examined using Western blotting, with α-tubulin used as the loading control. A blot representative of 3 independent experiments is shown. (b) Effect of knocking down the 2 subunits of collagen prolyl-4-hydroxylase 1 (CP4H1), P4HA and P4HB, or PHD2 in 293 cells stably transfected to express C1q. C1q expression in the culture supernatants was measured using enzyme-linked immunosorbent assay (ELISA). Bar graph shows means ± SEM of 3 experiments, expressed as fold changes over control: ** P < 0.01 versus control. Knockdown was confirmed by reverse transcription-polymerase chain reaction, with EE1A1 transcript levels used as the control. EE1A1, Eukaryotic translation elongation factor 1 alpha 1; HSF, human skin fibroblasts; HUVEC, human umbilical vein endothelial cell; MDM monocyte-derived macrophage; shRNA, short hairpin RNA. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org.

the control. EE1A1, Eukaryotic translation elongation factor 1 alpha 1; HSF, human skin fibroblasts; HUVEC, human umbilical vein endothelial cell; MDM monocyte-derived macrophage; shRNA, short hairpin RNA. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org. Although TDMs and MDMs are accepted models for studying C1q production, in our study, the observed secretion of C1q was modest. Furthermore, we reasoned that in this context, C1q production might be indirectly influenced, for example, through the activation of HIF that influences the cell-specific transcription of several hundred genes, including those encoding certain CP4H and PHD enzymes.3, 22, 23 To address these issues, we also examined the effect of decreasing PHD2 and P4HA1 levels in 293 cells engineered to produce recombinant C1q. Using lentiviral short hairpin RNA, we found that P4HA1 knockdown in these cells inhibited C1q secretion, whereas PHD2 knockdown did not (Figure 2b). To explore the fate of C1q that was not assembled into a macromolecular complex and secreted, we examined the effect of inhibiting potential degradation pathways. We found that the lysosomal inhibitor bafilomycin substantially increased the amount of C1q in cell lysates, but the proteasomal inhibitor MG132 did not. This effect was observed even under standard culture conditions, likely reflecting imperfect stoichiometry of C1q components in the overexpression system (Supplementary Figure S2).

the lysosomal inhibitor bafilomycin substantially increased the amount of C1q in cell lysates, but the proteasomal inhibitor MG132 did not. This effect was observed even under standard culture conditions, likely reflecting imperfect stoichiometry of C1q components in the overexpression system (Supplementary Figure S2). To examine the hydroxylation status of recombinant C1q secreted into the supernatants, we performed mass spectrometry analysis, which displayed a high degree of prolyl hydroxylation in collagen-like domains, consistent with previous reports for serum-derived C1q analyzed using amino acid sequencing (Supplementary Figure S3). To further characterize the manner in which roxadustat decreased the secretion of C1q, we examined intracellular C1q in the presence of bafilomycin to block degradation. We compared the hydroxylation status of intracellular C1q with and without roxadustat treatment using stable isotope labeling with amino acids in cell culture–based quantitative mass spectrometry. As predicted, a significant reduction of prolyl hydroxylation at multiple sites of intracellular C1q in treated samples was observed (Supplementary Table S1).

s of intracellular C1q with and without roxadustat treatment using stable isotope labeling with amino acids in cell culture–based quantitative mass spectrometry. As predicted, a significant reduction of prolyl hydroxylation at multiple sites of intracellular C1q in treated samples was observed (Supplementary Table S1). To directly test the ability of CP4H and PHD enzymes to hydroxylate C1q, we performed in vitro enzyme assays using peptides derived from HIF-1α, C1q A chain (C1q-4Pro), and procollagen [(Pro-Pro-Gly)10; PPG10] as templates, with recombinant preparations of PHD2 and CP4H1 (Figure 3a). The C1q peptide was a substrate for CP4H1, as evidenced by the increased conversion of the cosubstrate 2-OG to succinate (Figure 3b), but it was not a substrate for PHD2 (Figure 3c). Mass spectrometry analysis of the peptide substrates confirmed that CP4H1 was able to hydroxylate both PPG10 and C1q-4Pro peptide on multiple sites (Figure 3d and 3e, respectively). Control reactions simultaneously performed using mass spectrometry samples showed concomitant conversion of 2-OG to succinate (Figure 3f).Figure 3 C1q peptides are substrates for collagen prolyl-4-hydroxylase 1 (CP4H1) in vitro. (a) Schematic depicting the role of CP4H and prolyl hydroxylase domain (PHD) enzymes in the hydroxylation of proline residues within target proteins and concomitant conversion of the essential cosubstrate 2-oxoglutarate (2-OG) to succinate. (b–c) Peptides derived from hypoxia-inducible factor (HIF)-1α, C1q A chain, or collagen (PPG10) were incubated with enzyme for 2.5 hours to stimulate 2-OG conversion to succinate. (b) Reactions contained 35 nM CP4H1 enzyme, 250 μM 2-OG, and 10 μM ferrous sulfate (FeSO4) and 50 μM peptides. C1q peptides that substituted all 4 prolines with 4-hydroxyproline (C1q-4HyP) or dehydroproline (C1q-4dHP) did not support CP4H1 enzyme activity, indicating that the observed reaction did not occur because of substrate uncoupled turnover of 2-OG and was specific to proline residues. (c) Reactions contained 120 nM PHD2, 20 μM 2-OG, and 100 μM FeSO4. Percent conversion of 2-OG to succinate was normalized to either PPG10 peptide (b) or HIF-1α peptide (c). Values are expressed as averages ± SD (n = 3 from a single experiment). Findings were reproduced in at least 1 additional experiment for each peptide and enzyme. (d,e) Mass spectrometry analysis of PPG10 and C1q-4Pro peptides following incubation with CP4H1 as described in (b) except nonradiolabeled 2-OG was used.

(c). Values are expressed as averages ± SD (n = 3 from a single experiment). Findings were reproduced in at least 1 additional experiment for each peptide and enzyme. (d,e) Mass spectrometry analysis of PPG10 and C1q-4Pro peptides following incubation with CP4H1 as described in (b) except nonradiolabeled 2-OG was used. (d) For PPG10, nonhydroxylated peptide (M + H = 2532), along with species showing 1, 2, 3, and 4–6 OH events (M + H = 16 per OH), are evident in the spectrum. (e) For C1q-4Pro peptide, nonhydroxylated peptide (M + H = 2985), along with species showing 1 and 2 OH events are evident in the spectrum. (f) Succinate produced in parallel reactions to those shown in panels (d) and (e), except radiolabeled 2-OG was used. Experiments contained 35 nM CP4H1, 10 μM Fe, 250 μM 2-OG, and 50 μM peptide. Reactions were incubated for 3 hours.

+ H = 2985), along with species showing 1 and 2 OH events are evident in the spectrum. (f) Succinate produced in parallel reactions to those shown in panels (d) and (e), except radiolabeled 2-OG was used. Experiments contained 35 nM CP4H1, 10 μM Fe, 250 μM 2-OG, and 50 μM peptide. Reactions were incubated for 3 hours. Prolyl hydroxylase inhibitors reduce C1q plasma levels in vivo To establish whether the inhibitory effect of small molecule HIF activators on C1q secretion translated to the in vivo setting, C57BL/6 mice were treated with DMOG, FG0041, roxadustat, or vehicle every 12 hours for 6 days, and their effects on circulating C1q were assessed. The doses selected were similar to those reported to confer significant therapeutic effects in preclinical models of ischemic injury.18, 20, 24, 25 Plasma C1q was reduced by 28% with DMOG, 46% with FG0041, and 49% with roxadustat (Figure 4a). These data demonstrated the potential for small molecules that activate HIF to influence C1q and the complement pathway in vivo at therapeutically relevant doses. On the basis of the above evidence, this was likely an off-target effect, mediated via CP4H inhibition.Figure 4 Prolyl hydroxylase domain (PHD) inhibitors reduce circulating C1q in mice. (a) C57BL/6 mice were treated every 12 hours with dimethyloxalylglycine (DMOG; 20 mg/kg), FG0041 (25 mg/kg), or roxadustat (10 mg/kg) for 6 days. Serum C1q levels were measured using enzyme-linked immunosorbent assay (ELISA). Blood was sampled prior to treatment (baseline) and 6 days after treatment. Changes in plasma C1q levels are expressed as fold changes from baseline. Plasma C1q levels were significantly reduced in mice treated with PHD inhibitors. Vehicle-treated mice were unaffected (* P < 0.05, ** P < 0.01 vs. baseline). Data represent means ± SEM of 5–6 mice per group. (b) C1q expression in mice with germline deletion of different PHD enzymes or C1q was measured by ELISA. Each symbol represents 1 mouse, with means ± SEM shown. WT, wild-type.

h PHD inhibitors. Vehicle-treated mice were unaffected (* P < 0.05, ** P < 0.01 vs. baseline). Data represent means ± SEM of 5–6 mice per group. (b) C1q expression in mice with germline deletion of different PHD enzymes or C1q was measured by ELISA. Each symbol represents 1 mouse, with means ± SEM shown. WT, wild-type. To provide further evidence that inhibiting PHD enzymes per se was not sufficient to decrease C1q, we examined mice with genetic defects in PHD enzymes (phd1−/−, phd2+/−, and phd−/−) and observed that these mice did not show reduced circulating C1q levels (Figure 4b). Homozygous phd2−/− mice were not viable; thus, we could not exclude the possibility that the residual PHD2 activity in heterozygotes influenced C1q secretion. Finally, we did not observe any significant changes in circulating C1q levels in mice exposed to hypoxia (10% O2) for 3 days (Supplementary Figure S4). Discussion Together, our findings are consistent with the hydroxylation of C1q being mediated by CP4H and not by the PHD subfamily of prolyl hydroxylases. FG0041, DMOG and L-mimosine are all effective inhibitors of collagen synthesis,26 which is entirely consistent with our observation that they are effective inhibitors of C1q secretion and our conclusion that CP4H catalyzes C1q modification. Interestingly, P4HA1 expression is increased in hypoxia via HIF-promoted transcription.22 This, together with the fact that CP4H enzymes have a lower Km for oxygen than PHD enzymes,4 likely explains why C1q hydroxylation is maintained under hypoxic conditions (1% O2).

ur conclusion that CP4H catalyzes C1q modification. Interestingly, P4HA1 expression is increased in hypoxia via HIF-promoted transcription.22 This, together with the fact that CP4H enzymes have a lower Km for oxygen than PHD enzymes,4 likely explains why C1q hydroxylation is maintained under hypoxic conditions (1% O2). Our results revealed that CP4H1 was able to hydroxylate proline residues in the collagen domain of C1q, was expressed in macrophages, and was necessary for C1q secretion. We found that the 4 small molecules used to activate HIF also prevented C1q secretion by macrophages. We further showed that 2 inhibitors (DMOG and FG0041), which have been used in preclinical models, and one inhibitor (roxadustat, FG4592), which is currently in phase III clinical trials, significantly reduced circulating C1q levels in mice. These molecules are beneficial in models of ischemic disease, which has been attributed to HIF activation.18 Our findings further suggest that the inhibition of C1q secretion contributed to the observed benefit because C1q contributes to hypoxic-ischemic injury of several organs, including the brain and the heart,27, 28, 29 and the prevention of complement activation is considered a promising therapeutic strategy to limit ischemia-reperfusion injury to the kidneys.30, 31

on of C1q secretion contributed to the observed benefit because C1q contributes to hypoxic-ischemic injury of several organs, including the brain and the heart,27, 28, 29 and the prevention of complement activation is considered a promising therapeutic strategy to limit ischemia-reperfusion injury to the kidneys.30, 31 In some circumstances such as ischemic injury, a multitargeted prolyl hydroxylase inhibitor that activates HIF and decreases C1q secretion might be attractive. In contrast, long-term reduction of C1q levels could occur when a nonselective prolyl hydroxylase inhibitor such as roxadustat is used to treat renal anemia. This might be harmful because genetic defects in C1q are associated with chronic infections and systemic lupus erythematosus.14 Similarly, Schindler et al. reported increased mortality in a sepsis model of kidney injury, possibly contributed to by off-target effects of the prolyl hydroxylase inhibitors used.32 Importantly, the fact that C1q secretion and HIF inactivation utilize structurally related oxygenases, which have different substrate selectivities and active sites, suggests that the development of PHD inhibitors that selectively activate HIF without compromising C1q secretion is possible.

droxylase inhibitors used.32 Importantly, the fact that C1q secretion and HIF inactivation utilize structurally related oxygenases, which have different substrate selectivities and active sites, suggests that the development of PHD inhibitors that selectively activate HIF without compromising C1q secretion is possible. Materials and Methods Cell culture and reagents THP-1 cells were obtained from ATCC, Maryland, USA. Human monocytes were obtained from single-donor plateletpheresis residues, as previously described.33 The 293 C1q cells were a kind gift from Nicole Thielens (IBS, France).34 DMOG was purchased from Frontier Scientific (Logan, UT). FG0041 was produced in-house.21 Roxadustat (FG4592) was purchased from ApexBio (Houston, TX). All other chemicals were obtained from Sigma Chemical Company (Poole, England, UK), unless otherwise indicated. Differentiation of monocytes into macrophages TDMs were produced, as previously described.35 Human monocytes were differentiated into MDMs using macrophage colony-stimulating factor (PeproTech, London, UK) as previously described.33 Stimulation of cells for C1q expression C1q expression was induced in TDMs and MDMs by stimulation with interferon-γ (R&D systems, Minneapolis, MN) or lipopolysaccharide in Opti-MEM (Invitrogen, Paisley, UK) with or without PHD inhibitors. For hypoxic experiments, cells were exposed to 1% O2 using a hypoxic incubator (Galaxy R; Biotech, Palo Alto, CA).

ssion C1q expression was induced in TDMs and MDMs by stimulation with interferon-γ (R&D systems, Minneapolis, MN) or lipopolysaccharide in Opti-MEM (Invitrogen, Paisley, UK) with or without PHD inhibitors. For hypoxic experiments, cells were exposed to 1% O2 using a hypoxic incubator (Galaxy R; Biotech, Palo Alto, CA). Enzyme-linked immunosorbent assay for C1q Media were harvested after 48 hours and stored at −80°C. C1q was measured using a commercial human C1q enzyme-linked immunosorbent assay (ELISA; eBioscience, Hatfield, UK), according to the manufacturer’s instructions or an in-house sandwich ELISA. In brief, microtiter plates (Nunc-Immunoplate II, BRL, Middlesex, UK) were coated with 100 μl (1:1000) human sheep polyclonal anti-C1q (The Binding Site Group Ltd, Birmingham, UK) in phosphate-buffered saline (PBS) by incubating overnight at 4°C. Plates were then washed twice with PBS containing 0.075% Tween 20, followed by 2 more washes with PBS only, and were then blocked with 200 μl PBS-2% bovine serum albumin (BSA) for 2 hours at 37°C. Mouse monoclonal anti-C1q (clone 1A4, Santa Cruz Biotechnology, Santa Cruz, CA) at 1:1000 dilution in PBS-2% BSA was then used as the detection antibody, followed by alkaline phosphatase–conjugated anti-mouse IgG (Sigma-Aldrich, Poole, UK) at 1:1000 in PBS-2% BSA. The substrate p-nitrophenyl phosphate (Sigma-Aldrich) dissolved in Tris-buffer was added to the plate and incubated for 30 minutes at room temperature in the dark. The plates ware read at 405 nm on a spectrophotometric ELISA plate reader.

ase–conjugated anti-mouse IgG (Sigma-Aldrich, Poole, UK) at 1:1000 in PBS-2% BSA. The substrate p-nitrophenyl phosphate (Sigma-Aldrich) dissolved in Tris-buffer was added to the plate and incubated for 30 minutes at room temperature in the dark. The plates ware read at 405 nm on a spectrophotometric ELISA plate reader. Mouse plasma samples were analyzed using an in-house mouse C1q ELISA adapted from Petry et al.13 Microtiter plates (MaxiSorb; Nunc, Germany) were coated with 1 μg/ml C1q antibody (clone 7H8; Hycult Biotech, The Netherlands) and incubated overnight at 4°C. The wells were blocked with 1% BSA in PBS-0.05% Tween 20 for 1 hour at room temperature and then washed thrice with PBS-0.05% Tween 20 and once with PBS. Next, 100 μl of samples (diluted at 1:400 and 1:800) were added to wells and incubated for 1.5 hours at room temperature, washed as before, and incubated with 100 μl biotinylated C1q antibody (clone JL-1; Hycult Biotech) diluted at 1:4000 in PBS for 1 hour at room temperature. The wells were washed, as previously described, and incubated with 100 μl horseradish peroxidase–conjugated streptavidin (Biolegend, UK) diluted at 1:2000 in PBS for 1 hour at room temperature. The wells were washed and incubated with 100 μl 3,3′,5,5′-tetramethylbenzidine substrate solution for 10 to 20 minutes, and the reaction was stopped with 0.5 M H2SO4 (50 μl/well). The optical density was read at 450 nm (with the wavelength correction set at 540 and 570 nm).

1:2000 in PBS for 1 hour at room temperature. The wells were washed and incubated with 100 μl 3,3′,5,5′-tetramethylbenzidine substrate solution for 10 to 20 minutes, and the reaction was stopped with 0.5 M H2SO4 (50 μl/well). The optical density was read at 450 nm (with the wavelength correction set at 540 and 570 nm). Stable prolyl hydroxylase knockdown MISSION short hairpin RNA constructs for P4HA1 (TRCN0000303934), P4HB (TRCN000 O296675), and PHD2 (TRCN0000001043) were purchased from Sigma-Aldrich. Lentiviral particles were produced in 293T cells by cotransfection of the short hairpin RNA vector and the packaging vectors pMD2.G and pCMV-dR8.2 (Addgene, Cambridge, MA) using TransIT-293 transfection reagent (Mirus Bio, Madison, WI), according to the manufacturer’s instructions. Filtered cell culture supernatants were added to 293 cells stably expressing C1q, followed by selection with puromycin. Stable knockdown was confirmed by reverse transcription-polymerase chain reaction using TRIzol-extracted RNA and SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. Reverse transcription-polymerase chain reaction primer sequences 5′-3′ P4HA1F1 AGACCTAGCAAAACCAAGGCT P4HA1R1 TTTCATAGCCAGAGAGCCAGG P4HBF2 CTTCAAGGACGTGGAGTCGG P4HBR2 ACCCCATCTTTGTCGAGCTG PHD2F2 ACTGGGATGCCAAGGTAAGTG PHD2R2 CTCGTGCTCTCTCATCTGCAT EE1A1F AAGTGCTAACATGCCTTGGTTC EE1A1R AGGAACAGTACCAATACCACCA

Stable prolyl hydroxylase knockdown MISSION short hairpin RNA constructs for P4HA1 (TRCN0000303934), P4HB (TRCN000 O296675), and PHD2 (TRCN0000001043) were purchased from Sigma-Aldrich. Lentiviral particles were produced in 293T cells by cotransfection of the short hairpin RNA vector and the packaging vectors pMD2.G and pCMV-dR8.2 (Addgene, Cambridge, MA) using TransIT-293 transfection reagent (Mirus Bio, Madison, WI), according to the manufacturer’s instructions. Filtered cell culture supernatants were added to 293 cells stably expressing C1q, followed by selection with puromycin. Stable knockdown was confirmed by reverse transcription-polymerase chain reaction using TRIzol-extracted RNA and SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. Reverse transcription-polymerase chain reaction primer sequences 5′-3′ P4HA1F1 AGACCTAGCAAAACCAAGGCT P4HA1R1 TTTCATAGCCAGAGAGCCAGG P4HBF2 CTTCAAGGACGTGGAGTCGG P4HBR2 ACCCCATCTTTGTCGAGCTG PHD2F2 ACTGGGATGCCAAGGTAAGTG PHD2R2 CTCGTGCTCTCTCATCTGCAT EE1A1F AAGTGCTAACATGCCTTGGTTC EE1A1R AGGAACAGTACCAATACCACCA Immunoblotting Tissues and cells were homogenized in protein extraction buffer, and protein analysis was performed, as previously described.36 The following antibodies were used: mouse monoclonal anti-HIF-1α (clone 54; Transduction Labs, Lexington, KY), rabbit polyclonal anti-HIF-1α (NB100-479; Novus Biologicals, Littleton, CO), rabbit polyclonal anti-PHD-1, rabbit polyclonal anti-PHD-2, and goat polyclonal anti-P4HA1 (Abcam, Cambridge, MA). Alkaline phosphatase–conjugated anti-mouse IgG (γ chain specific), horseradish peroxidase–conjugated donkey anti-mouse and anti-rabbit IgG antibodies (Bethyl Laboratories, Montgomery, TX), and mouse monoclonal anti-human α-tubulin and β-actin antibodies were purchased from Sigma-Aldrich.

i-P4HA1 (Abcam, Cambridge, MA). Alkaline phosphatase–conjugated anti-mouse IgG (γ chain specific), horseradish peroxidase–conjugated donkey anti-mouse and anti-rabbit IgG antibodies (Bethyl Laboratories, Montgomery, TX), and mouse monoclonal anti-human α-tubulin and β-actin antibodies were purchased from Sigma-Aldrich. PHD2 and CP4H1 enzymatic assays The CP4H1 and PHD2 enzyme assays measured the conversion of 5-[14C]-2-oxoglutarate to [14C]-succinic acid. For CP4H1 reactions, final conditions included 50 mM Tris-HCl pH 7.5, 0.2 mM dithiothreitol, 1 mM ascorbate, 4% dimethylsulfoxide, and 0.5 mg/ml catalase. For PHD2 reactions, final conditions included 30 mM 2-(N-morpholino) ethanesulfonic acid pH 6.0, 10 mM NaCl, 5 mM CaCl2, 2.5 mM dithiothreitol, 0.25% Brij-35, 0.05 mg/ml BSA, 4% dimethylsulfoxide, and 2 mM ascorbate. Concentrations of FeSO4, enzyme, peptide substrates, and 2-OG are shown in the legend of Figure 2. In brief, FeSO4, enzyme, and peptide were sequentially added to the assay buffer and gently mixed for 10 minutes at room temperature. 2-OG was then added to initiate the reaction. Reactions were gently mixed at room temperature and terminated after 3 or 3.5 hours by adding an equal volume of 0.02 N H2SO4. A portion of each terminated reaction was injected into a Polypore H column (PerkinElmer, Waltham, MA) at a rate of 0.3 ml/min with 0.01 N H2SO4 as the mobile phase. Substrate and product peaks were detected at 210 nm (Agilent 1100, Agilent Technologies, Santa Clara, CA), and the radioactivity associated with each was captured using a Beta-RAM Model 2 radiation detector and the In-Flow 2:1 scintillation cocktail (IN/US Systems Inc., Tampa, FL). Laura Lite software (IN/US Systems) was used to collect and analyze radiometric data. This high-performance liquid chromatography method exploits the difference in pKa of 2-OG and succinic acid carboxylates to chromatographically separate substrate from product at low pH using ion exchange resin, as described by Cunliffe et al. and Kaule and Günzler.37, 38

s used to collect and analyze radiometric data. This high-performance liquid chromatography method exploits the difference in pKa of 2-OG and succinic acid carboxylates to chromatographically separate substrate from product at low pH using ion exchange resin, as described by Cunliffe et al. and Kaule and Günzler.37, 38 The sequence of C1q A chain peptides with C-terminal amides (C1q-4 Pro) was H-GEAGR[Pro]GRRGR[Pro] GLKGEQGE[Pro][GA[Pro]GIRTGI-OH. For C1q-4 dHP and C1q-4 HyP, the proline motifs were substituted with dehydroproline and 4-hydroxyproline, respectively. The sequence selected for testing was based on the findings of Reid.8 The sequence of HIF-1α peptides (HIF-1α Pro) was H-DLDLEMLA[Pro] YIPMDDD-FQL-OH. For HIF-1α dHP and HIF-1α HyP, the proline motifs were substituted with dehydroproline and 4-hydroxyproline, respectively. The collagen peptide PPG10 was purchased from Peptides International (Louisville, KY). The following materials were obtained from their respective manufacturers: 5-[14C]-2-oxoglutarate (2-OG), PerkinElmer LAS (Shelton, CT) or Moravek Biochemicals (Brea, CA); complement C1q peptides, MidWest Biotech (Fishers, IN); and catalase (Sigma-Aldrich). HIF peptides were internally generated (Amgen) or obtained from MidWest Biotech.

rials were obtained from their respective manufacturers: 5-[14C]-2-oxoglutarate (2-OG), PerkinElmer LAS (Shelton, CT) or Moravek Biochemicals (Brea, CA); complement C1q peptides, MidWest Biotech (Fishers, IN); and catalase (Sigma-Aldrich). HIF peptides were internally generated (Amgen) or obtained from MidWest Biotech. Mass spectrometry of peptides Reactions with unlabeled 2-OG were run in parallel with reactions containing radiolabeled 2-OG and allowed to proceed for 3 hours. Reactions containing unlabeled 2-OG were prepared for matrix-assisted laser desorption/ionization-mass spectrometry as follows: 1 μl sample was diluted to 10 μl α-cyano-4-hydroxycinnamic acid (10 mg/ml in 50% acetonitrile containing 0.05% trifluoroacetic acid), and 1 μl of the resultant solution was air dried on stainless steel. The sample mass was measured using a Waters MALDI Micro MX mass spectrometer (Waters, Milford, MA). The reflectron mode (using an acceleration voltage of 25 kV) in combination with delayed extraction was employed. The mass spectrometer was externally calibrated using a mixture of neurotensin, angiotensin, and 3 adrenocorticotropic hormone fragments (1–17, 18–39, and 7–38). The 14C-labeled control reactions were analyzed by high-performance liquid chromatography, as described above.

mbination with delayed extraction was employed. The mass spectrometer was externally calibrated using a mixture of neurotensin, angiotensin, and 3 adrenocorticotropic hormone fragments (1–17, 18–39, and 7–38). The 14C-labeled control reactions were analyzed by high-performance liquid chromatography, as described above. Stable isotope labeling with amino acids in cell culture and mass spectrometry Cells were labeled with heavy or light arginine and lysine for at least 5 passages. Cells were treated with roxadustat or vehicle (dimethylsulfoxide) and bafilomycin A to inhibit lysosomal acidification and prevent any loss of nonhydroxylated protein marked for degradation. After 20 hours, roxadustat-treated cells were combined with controls and lysed in 1% Triton X-100 before C1q protein was purified by FLAG pull-down and separated using a polyacrylamide gel. Samples were reduced, alkylated, and digested in-gel using trypsin, with the resulting peptides eluted for analysis using liquid chromatography-tandem mass spectrometry. Peptides were analyzed using a Q Exactive mass spectrometer (Thermo Fisher Scientific, Waltham, MA) coupled to an RSLC nano3000 UPLC (Thermo Fisher Scientific). Data were acquired in a DDA fashion. Raw files were processed in PEAKS Studio 8.0 (Bioinformatics Solutions Inc., Waterloo, ON, Canada) or Proteome Discoverer 1.4 software (Thermo Fisher Scientific). Data were searched using the Uniprot database (465,339 sequences, downloaded 11/07/16) and the human Uniprot database (30,510 sequences, downloaded 06/19/16). Using the Proteome Discoverer software, variable modifications were set as 13C(6)15H(2) lysine, 13C(6)15N(4) arginine, and oxidized methionine and proline. The PEAKS software was used to search for posttranslational modifications, with confident modification sites requiring a minimum ion intensity of 5%.

ded 06/19/16). Using the Proteome Discoverer software, variable modifications were set as 13C(6)15H(2) lysine, 13C(6)15N(4) arginine, and oxidized methionine and proline. The PEAKS software was used to search for posttranslational modifications, with confident modification sites requiring a minimum ion intensity of 5%. In vivo studies All procedures were ethically approved and complied with the Scientific Procedures Act 1986 and the Guidelines for the Welfare and Use of Animals in Cancer 2010,39 under the authority of the Home Office, UK. Mice with defective phd and c1q alleles were previously described.40, 41 For inhibitor studies, male and female C57BL/6 mice (aged 12 weeks) were randomly assigned to treatment groups (4 groups, 5–6/group; calculated using the resource equation). Prior to treatment, blood was sampled for analyzing baseline plasma C1q levels. Mice were then treated with 20 mg/kg DMOG, 25 mg/kg FG0041, 10 mg/kg roxadustat, or vehicle (4% dimethylsulfoxide in PBS) and dosed with 0.1 ml/10 g body weight with an i.p. injection every 12 hours for 6 days. A second blood sample was collected for analyzing plasma C1q after killing the mice. Statistical analysis All data were analyzed by 1-way analysis of variance with Bonferroni or Tukey post hoc correction. Disclosures PHM and CJS are scientific founders and equity holders of ReOx, which aims to develop PHD inhibitors as therapies. All the other authors declared no competing interests.

In vivo studies All procedures were ethically approved and complied with the Scientific Procedures Act 1986 and the Guidelines for the Welfare and Use of Animals in Cancer 2010,39 under the authority of the Home Office, UK. Mice with defective phd and c1q alleles were previously described.40, 41 For inhibitor studies, male and female C57BL/6 mice (aged 12 weeks) were randomly assigned to treatment groups (4 groups, 5–6/group; calculated using the resource equation). Prior to treatment, blood was sampled for analyzing baseline plasma C1q levels. Mice were then treated with 20 mg/kg DMOG, 25 mg/kg FG0041, 10 mg/kg roxadustat, or vehicle (4% dimethylsulfoxide in PBS) and dosed with 0.1 ml/10 g body weight with an i.p. injection every 12 hours for 6 days. A second blood sample was collected for analyzing plasma C1q after killing the mice. Statistical analysis All data were analyzed by 1-way analysis of variance with Bonferroni or Tukey post hoc correction. Disclosures PHM and CJS are scientific founders and equity holders of ReOx, which aims to develop PHD inhibitors as therapies. All the other authors declared no competing interests. Supplementary Material Figure S1 C1q expression in human monocyte-derived macrophages (MDMs) is inhibited by dimethyloxalylglycine (DMOG). MDMs were treated with (A) lipopolysaccharide (LPS, 10 ng/ml) or (B) interferon (IFN)-γ (10 ng/ml), with or without DMOG for 16 hours. C1q expression in culture supernatants was measured using enzyme-linked immunosorbent assay (ELISA). Histograms show the means ± SEM of 3 experiments. *** P < 0.001 versus cells treated with either LPS or IFN-γ.

lipopolysaccharide (LPS, 10 ng/ml) or (B) interferon (IFN)-γ (10 ng/ml), with or without DMOG for 16 hours. C1q expression in culture supernatants was measured using enzyme-linked immunosorbent assay (ELISA). Histograms show the means ± SEM of 3 experiments. *** P < 0.001 versus cells treated with either LPS or IFN-γ. Figure S2 Recombinant C1q is degraded by the lysosomal pathway. Two hundred and ninety-three cells expressing recombinant C1q were treated with the lysosomal inhibitor bafilomycin A (Bafilo, 100 nM) or the proteasomal inhibitor MG132 (MG, 10 μg/ml) for 16 hours, with and without roxadustat treatment (Rox, 10 μM). Immunoblotting of cell lysates shows accumulation of C1q A, B, and C chains after bafilomycin treatment. Hypoxia-inducible factor (HIF)-1α, which was previously shown to be rescued by all 3 treatments (i.e., bafilomycin, MG132, and roxadustat) serves as the positive control, and β-actin serves as the loading control. Figure S3 Posttranslational modification of recombinant C1q A chain. FLAG-tagged C1q protein was purified from the supernatants of 293 cells expressing all 3 C1q chains. The purified protein was separated using a polyacrylamide gel and digested in-gel using trypsin. The coverage is shown in gray highlights. The PEAKS software (Bioinformatics Solutions Inc.) was used to identify posttranslational modifications, as indicated above the modified amino acids. Hydroxylated proline residues are circled in red.

in was separated using a polyacrylamide gel and digested in-gel using trypsin. The coverage is shown in gray highlights. The PEAKS software (Bioinformatics Solutions Inc.) was used to identify posttranslational modifications, as indicated above the modified amino acids. Hydroxylated proline residues are circled in red. Figure S4 Mice exposed to hypoxia for 3 days exhibit no changes in circulating C1q levels. Mice (3 males and 3 females in each group, 70–77-day-old) were exposed to 10% O2 for 3 days. The serum C1q level at day 3 was measured with an in-house enzyme-linked immunosorbent assay (ELISA) and given as arbitrary units (AU). Table S1 Reduced prolyl hydroxylation of C1q A chain after roxadustat treatment. Acknowledgments We are grateful to John Hui (Amgen) for matrix-assisted laser desorption/ionization-mass spectrometry analysis of reactions shown in Figure 3c and d; Drs. Peter J Norsworthy and Giannis Deligiannis for assistance with the C1q ELISAs and in vivo studies; and Professors Ken GC Smith, Nick W Morrell, and Dr. Stephen Moore for their support in the in vivo studies. We gratefully acknowledge the support of our core facilities, especially Robin Antrobus for the mass spectrometry work. This work was supported by the Wellcome Trust and the NIHR Cambridge Biomedical Research Centre Senior Investigator Awards (to PHM, supporting SSH and NB), the Wellcome Trust Scientific strategic award [100140]. SK was supported by a British Heart Foundation grant awarded to PHM as well as a Kennedy Institute of Rheumatology trustees fund.

ed by the Wellcome Trust and the NIHR Cambridge Biomedical Research Centre Senior Investigator Awards (to PHM, supporting SSH and NB), the Wellcome Trust Scientific strategic award [100140]. SK was supported by a British Heart Foundation grant awarded to PHM as well as a Kennedy Institute of Rheumatology trustees fund. Figure S1. C1q expression in human monocyte-derived macrophages (MDMs) is inhibited by dimethyloxalylglycine (DMOG). MDMs were treated with (A) lipopolysaccharide (LPS, 10 ng/ml) or (B) interferon (IFN)-γ (10 ng/ml), with or without DMOG for 16 hours. C1q expression in culture supernatants was measured using enzyme-linked immunosorbent assay (ELISA). Histograms show the means ± SEM of 3 experiments. *** P < 0.001 versus cells treated with either LPS or IFN-γ. Figure S2. Recombinant C1q is degraded by the lysosomal pathway. Two hundred and ninety-three cells expressing recombinant C1q were treated with the lysosomal inhibitor bafilomycin A (Bafilo, 100 nM) or the proteasomal inhibitor MG132 (MG, 10 μg/ml) for 16 hours, with and without roxadustat treatment (Rox, 10 μM). Immunoblotting of cell lysates shows accumulation of C1q A, B, and C chains after bafilomycin treatment. Hypoxia-inducible factor (HIF)-1α, which was previously shown to be rescued by all 3 treatments (i.e., bafilomycin, MG132, and roxadustat) serves as the positive control, and β-actin serves as the loading control.

. Immunoblotting of cell lysates shows accumulation of C1q A, B, and C chains after bafilomycin treatment. Hypoxia-inducible factor (HIF)-1α, which was previously shown to be rescued by all 3 treatments (i.e., bafilomycin, MG132, and roxadustat) serves as the positive control, and β-actin serves as the loading control. Figure S3. Posttranslational modification of recombinant C1q A chain. FLAG-tagged C1q protein was purified from the supernatants of 293 cells expressing all 3 C1q chains. The purified protein was separated using a polyacrylamide gel and digested in-gel using trypsin. The coverage is shown in gray highlights. The PEAKS software (Bioinformatics Solutions Inc.) was used to identify posttranslational modifications, as indicated above the modified amino acids. Hydroxylated proline residues are circled in red. Figure S4. Mice exposed to hypoxia for 3 days exhibit no changes in circulating C1q levels. Mice (3 males and 3 females in each group, 70–77-day-old) were exposed to 10% O2 for 3 days. The serum C1q level at day 3 was measured with an in-house enzyme-linked immunosorbent assay (ELISA) and given as arbitrary units (AU). Table S1. Reduced prolyl hydroxylation of C1q A chain after roxadustat treatment. Supplementary material is linked to the online version of the paper at www.kidney-international.org.

Chronic kidney disease (CKD) affects ∼10% of adults in high-income countries,1, 2 and its treatment is burdensome and costly, accounting for substantial proportions of health budgets.3, 4 It is now recognized as an important cause of death, particularly in low- and middle-income countries where diabetes is becoming common and resources for treatment are limited.5, 6 Other conditions affecting the kidney, such as acute kidney injury7, 8 and kidney stone disease,9 also contribute a substantial burden of morbidity and mortality. Despite the high individual patient and societal burden of kidney diseases, the amount of reliable information available to guide kidney patient care is very limited.10 Although some treatments have been assessed in randomized trials, most of these have been too small to detect treatment effects of a magnitude that can realistically be achieved with a single intervention (e.g., reductions of 15%–20% in major outcomes such as death or disability). Experience to date indicates that conducting trials in patients with kidney disease on the sort of scale that has led to major therapeutic advances in other specialties (e.g., oncology, diabetes, and cardiology11) is challenging.

intervention (e.g., reductions of 15%–20% in major outcomes such as death or disability). Experience to date indicates that conducting trials in patients with kidney disease on the sort of scale that has led to major therapeutic advances in other specialties (e.g., oncology, diabetes, and cardiology11) is challenging. Conference structure In order to discuss the most significant barriers to conducting trials in patients with kidney disease and to propose potential solutions, KDIGO (Kidney Disease: Improving Global Outcomes) convened an international multidisciplinary Controversies Conference in Paris, France, titled “Challenges in the Conduct of Clinical Trials in Nephrology” in September 2016. The meeting began with plenary talks that aimed to identify the key themes for discussion related to trial design (including the specific topics of how to measure kidney and nonkidney outcomes) and trial execution before 4 breakout groups considered the key issues in detail. After each of the breakout sessions, ongoing deliberations were reported and discussed in plenary sessions. This paper synthesizes the main areas of discussion, agreement, and remaining controversies addressed at the conference. The conference agenda and selected presentations from the meeting are available on the KDIGO website (http://kdigo.org/conferences/clinical-trials/).

were reported and discussed in plenary sessions. This paper synthesizes the main areas of discussion, agreement, and remaining controversies addressed at the conference. The conference agenda and selected presentations from the meeting are available on the KDIGO website (http://kdigo.org/conferences/clinical-trials/). The discussion was based on the general principle that, in order to conduct a successful randomized trial, there are 4 main requirements: randomization of a sufficient number of patients (to ensure sufficient numbers of outcomes); assurance of adherence to the allocated treatment; reliable ascertainment of relevant study outcomes; and appropriate statistical analysis.12 These requirements can be met through a combination of sound trial design and efficient conduct. Much of the discussion at the conference centered on the concept of “streamlined” trial design, which focuses on the main determinants of trial quality and avoids unnecessary elements of design or conduct that increase cost and complexity.13

can be met through a combination of sound trial design and efficient conduct. Much of the discussion at the conference centered on the concept of “streamlined” trial design, which focuses on the main determinants of trial quality and avoids unnecessary elements of design or conduct that increase cost and complexity.13 Trial design (analogous to experimental design) and trial conduct were considered separately. However, it should be noted that many aspects of a trial’s design influence its conduct and can therefore be decisive in determining whether a trial is successful. Because of their large scope, discussions about trial design were organized into 3 breakout groups that considered the design of trials in kidney disease, how to measure different kidney-specific outcomes, and assessment of other outcomes. The fourth breakout group focused on optimizing trial conduct. Table 1 brings together key findings from the breakout groups with regard to challenges in the design and conduct of randomized trials in patients with kidney disease. More complete details are provided in the following.Table 1 Objectives in designing and conducting randomized trials and issues in the context of kidney disease

brings together key findings from the breakout groups with regard to challenges in the design and conduct of randomized trials in patients with kidney disease. More complete details are provided in the following.Table 1 Objectives in designing and conducting randomized trials and issues in the context of kidney disease Trial objectives Elements that help to achieve trial objectives Difficulties in kidney disease Study design Study procedures Answer an important question reliably Differences between treatment(s) must be important to patients Sample size: needs to be determined by realistic assumptions (treatment effects, rates of adherence [drop-out/drop-in], and contemporary event rates), and is generally best determined by a fixed number of primary outcomes (i.e., event driven) Study duration: sufficient time is required for study treatments to exert benefit (or for any known hazards to emerge) Outcome selection:• Relevant to patients and must be measurable without undue burden on them • Sensitive to the main benefits and hazards of the trial treatment(s) Many treatments are already in use despite a lack of reliable evidence of safety or efficacy. Nephrologists may be reluctant to compare such treatments with placebo. Because of the difficulties of identifying large numbers of eligible patients (especially in rare diseases), trialists may:• assume unrealistically large relative risk reductions (sometimes based on implausible results of systematic reviews or nonrandomized studies) • fail to allow for often substantial nonadherence, which severely diminishes statistical power

Many treatments are already in use despite a lack of reliable evidence of safety or efficacy. Nephrologists may be reluctant to compare such treatments with placebo. Because of the difficulties of identifying large numbers of eligible patients (especially in rare diseases), trialists may:• assume unrealistically large relative risk reductions (sometimes based on implausible results of systematic reviews or nonrandomized studies) • fail to allow for often substantial nonadherence, which severely diminishes statistical power Many kidney disease trials have not measured suitable outcomes. Common problems include:• Lack of relevance to patients, prescribers, and payers • Lack of consistency with outcomes in other pivotal trials • Use of total mortality, either alone or as a component of a composite primary outcome, resulting in a lack of statistical power Effective recruitment Population selection:• Trials should be relevant to a wide range of patients who might in the future be treated with the study intervention • Avoid unnecessary exclusions

• Lack of consistency with outcomes in other pivotal trials • Use of total mortality, either alone or as a component of a composite primary outcome, resulting in a lack of statistical power Effective recruitment Population selection:• Trials should be relevant to a wide range of patients who might in the future be treated with the study intervention • Avoid unnecessary exclusions Availability of large numbers of potentially eligible patients from routine databases, with prescreening (if feasible) Pilot study experience Trials of patients with kidney disease often exclude large proportions of patients, resulting in both difficulty with recruitment and a lack of generalizability Achieving good adherence Exclude participants likely to drop out or drop in at screening or before randomization Use of a run-in Procedures that are not onerous for trial participants Allow flexibility in determining nontrial treatments Patients with kidney disease have a high burden of medication and intervention Nephrologists can become certain about benefit or harm before treatments are adequately tested Complete recording of outcomes Outcome definition does not require complex procedures or difficulties for patients Simple case report forms recording outcomes Multimodal sources of patient data (patients, family members, primary care physicians) Maintaining contact with patients who no longer wish to take study treatment or attend clinics (e.g., by telephone follow-up) Use of registry data and electronic health care records Unbiased analysis Statistical analysis plan, including:• Intention-to-treat analyses as primary

, family members, primary care physicians) Maintaining contact with patients who no longer wish to take study treatment or attend clinics (e.g., by telephone follow-up) Use of registry data and electronic health care records Unbiased analysis Statistical analysis plan, including:• Intention-to-treat analyses as primary • Limited number of subgroup analyses and only when hypotheses can be stated in advance • Clear demarcation of primary, secondary, and exploratory analyses Underpowered trials and low adherence resulting in “negative” results for the primary outcome have previously led to inappropriate emphasis on underpowered subgroup analyses and potentially biased on-treatment analyses Optimizing trial design General trial design considerations The discussions identified important trial design principles that are generic but inconsistently applied in trials among patients with kidney disease, as well as issues that are specific to such trials. Formulating the trial question A key first step when planning a trial is to formulate the trial question. This should be both clinically relevant (i.e., addressing a major area of clinical uncertainty) and relevant to patients (i.e., aiming to avoid an outcome or condition that patients themselves consider significant). Patients should be involved in discussions when planning trials.

ial is to formulate the trial question. This should be both clinically relevant (i.e., addressing a major area of clinical uncertainty) and relevant to patients (i.e., aiming to avoid an outcome or condition that patients themselves consider significant). Patients should be involved in discussions when planning trials. An important underlying principle in trials testing superiority of an intervention is that the aim should always be to compare treatment arms that differ substantially in their expected effects on the primary outcome and to preserve this separation for the duration of the trial. The likely magnitude of this difference can sometimes be usefully tracked through assessing differences in a surrogate variable between randomized treatment groups (e.g., plasma low-density lipoprotein cholesterol in the SHARP trial14), provided that blinding at an individual participant level is strictly enforced. Failure to maintain adequate adherence to study treatment leads to a loss of separation between groups and loss of statistical power. This may be a particular problem in trials among patients with CKD who are typically receiving multiple treatments, and are often required to attend clinics frequently, and are particularly prone to drug toxicity and intercurrent illnesses that may require the trial treatment to be modified (as observed in the EVOLVE trial in which study drug discontinuation was much higher than anticipated).15, 16

lly receiving multiple treatments, and are often required to attend clinics frequently, and are particularly prone to drug toxicity and intercurrent illnesses that may require the trial treatment to be modified (as observed in the EVOLVE trial in which study drug discontinuation was much higher than anticipated).15, 16 Selecting a suitable trial population As a general principle, eligibility criteria should be practical and broad. This ensures a widely generalizable result and facilitates efficient recruitment. Inclusion criteria should identify a suitable population of patients who are likely to have the type of outcome that the treatment is anticipated to prevent. Guideline committees may help to maximize the size of eligible populations by highlighting areas of uncertainty, by avoiding making recommendations where evidence is weak, and by stating where placebo-controlled trials are needed.

who are likely to have the type of outcome that the treatment is anticipated to prevent. Guideline committees may help to maximize the size of eligible populations by highlighting areas of uncertainty, by avoiding making recommendations where evidence is weak, and by stating where placebo-controlled trials are needed. Exclusion criteria should be constructed that exclude patients who have a definite indication or contraindication for at least 1 of the study treatments, who are likely to be nonadherent to the trial protocol, or who are not expected to survive for the duration of the trial. Therefore, patients who would have obvious safety issues (from relevant comorbidity) or who are at high risk of a potential pharmacologic interaction related to the intervention should generally be excluded (although this might not always be necessary if a Data and Safety Monitoring Board is charged with monitoring safety in specific patient groups). An important group to exclude is those who are likely to stop allocated treatment because of intolerance or nonadherence (“drop-out”) or who are likely to start the study drug (or a drug with a similar mode of action) outside of the trial (“drop-in”). It may be possible to identify patients who are likely to be nonadherent during a prerandomization “run-in” period so that they can be withdrawn before being randomized. This helps to preserve study power.17 Run-in periods may be particularly beneficial in trials of dialysis patients, a complex group for which maintaining adherence to generally well-tolerated medications can be difficult.14, 15

ng a prerandomization “run-in” period so that they can be withdrawn before being randomized. This helps to preserve study power.17 Run-in periods may be particularly beneficial in trials of dialysis patients, a complex group for which maintaining adherence to generally well-tolerated medications can be difficult.14, 15 Calculating an appropriate sample size Kidney disease has a wide variety of causes, and many different pathophysiologic mechanisms are responsible for disease initiation and progression, as well as for its complications. This makes it unlikely that a single treatment would have a large relative effect on major outcomes (particularly on outcomes such as death or progression to end-stage kidney disease). In determining sample size, therefore, particularly for large Phase 3 outcome trials, it is important to avoid overly optimistic assumptions about treatment effect size, even when there are apparently large effects on disease biomarkers. In practice, this means that relative risk reductions in major outcomes of more than ∼20% are unlikely to be observed. Nonetheless, in high-risk patients such as those with kidney disease, small relative risk reductions may still correspond to large reductions in absolute risk and therefore be clinically and economically worthwhile.

his means that relative risk reductions in major outcomes of more than ∼20% are unlikely to be observed. Nonetheless, in high-risk patients such as those with kidney disease, small relative risk reductions may still correspond to large reductions in absolute risk and therefore be clinically and economically worthwhile. It is inappropriate to estimate relative risk reductions (for major outcomes) from systematic reviews composed solely of small randomized trials; such trials can only achieve statistically significant results (and hence publication) if—by the play of chance—their effect estimates are larger than the truth, which means that such reviews will tend to yield inflated effect size estimates.18 When the evidence from previous small trials is limited in this way, a trial that is capable of detecting a realistically moderate effect (e.g., a 15%–20% reduction in a major outcome) is more rational than one that aims to detect an effect estimated in a systematic review. When designing kidney trials, there may be considerable uncertainty when estimating event rates owing to a lack of contemporary data from population-based studies.19 In addition, event rates derived from population-based registries may not reflect those of an enrolled trial population. The potential for high levels of nonadherence to study treatment, which can have a particularly detrimental effect on study power, should also be considered. Adequate drug exposure and study power may therefore be achieved by planning on a predefined total number of primary outcomes and a minimum duration of follow-up in a study population of approximately the correct size. A Data and Safety Monitoring Board can advise on early termination of a trial if convincing evidence of harm or efficacy emerges before trial completion or if there are other reasons (e.g., irremediable failure to recruit or futility) that make it inappropriate to continue.

population of approximately the correct size. A Data and Safety Monitoring Board can advise on early termination of a trial if convincing evidence of harm or efficacy emerges before trial completion or if there are other reasons (e.g., irremediable failure to recruit or futility) that make it inappropriate to continue. Statistical analysis The principles of statistical analysis for randomized trials are well documented and should be applied in trials among patients with kidney disease.12 For example, intention-to-treat analyses should be specified for the primary analysis. Several particular problems are commonly encountered. First, where sample sizes are limited, there has been a temptation to overinterpret subgroup findings or where nonadherence is a problem, to conduct potentially biased on-treatment analyses. In superiority trials, such analyses should be considered exploratory rather than confirmatory, except where the analysis plan provides a clear justification for them before database lock and consideration has been given to the number of comparisons being made. In reporting the results of a clinical trial, the primary outcome should be emphasized, even if the results do not support the intervention. However, the full interpretation of a trial’s results should relate to the totality of evidence, meaning the primary outcome plus secondary and safety outcomes.20 If evidence is indeed inconclusive, this is an acceptable and valuable conclusion, and further randomized trials may still be useful if uncertainty remains.

However, the full interpretation of a trial’s results should relate to the totality of evidence, meaning the primary outcome plus secondary and safety outcomes.20 If evidence is indeed inconclusive, this is an acceptable and valuable conclusion, and further randomized trials may still be useful if uncertainty remains. Selection of outcomes for assessing treatment effects Measurement of kidney-specific outcomes The choice of outcome measures in trials designed to assess kidney disease status (i.e., function, damage, or disease activity) should depend on the disease setting and phase of clinical development. Consideration should be given both to the stage of the disease and how rapidly it is progressing. Table 2 provides a matrix of general approaches to and strategies for selecting endpoints. The measurement of activity of disease and kidney structure, if available, may add useful information to measures of kidney function. Disease-specific markers (or outcomes) that reflect the underlying pathophysiology or molecular pathways of disease may be helpful in the setting of primary kidney diseases such as systemic lupus erythematosus, IgA nephropathy, and polycystic kidney disease. However, in settings where progression of kidney disease is a multifactorial process, such as with hypertension or diabetes mellitus, a kidney-specific outcome may be best assessed using the estimated glomerular filtration rate (eGFR).21Table 2 Suggested outcomes in measuring kidney disease status in randomized trials

However, in settings where progression of kidney disease is a multifactorial process, such as with hypertension or diabetes mellitus, a kidney-specific outcome may be best assessed using the estimated glomerular filtration rate (eGFR).21Table 2 Suggested outcomes in measuring kidney disease status in randomized trials CKD stage Progression of CKD Slow Rapida Early stage: CKD G1-G3a (eGFR ≥45 ml/min per 1.73 m2) • Slope of mGFR or eGFR or • Surrogate outcomeb or • Combinations of outcomes 30%−40% decline in eGFR using repeat measurements to rule out transient acute effectsc Late stage: CKD G3b-G5 (eGFR <45 ml/min per 1.73 m2) End-stage kidney disease or 30%−40% decline in eGFRc End-stage kidney disease or doubling of serum creatinine level (or 40%–57% decline in eGFR)c CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; mGFR, measured glomerular filtration rate. a For example, in patients with macroalbuminuria. b Surrogates may include measures of activity of disease (e.g., in lupus nephritis) or kidney structure (e.g., in adult polycystic kidney disease). c The added value of eGFRs outside the routine study visit schedule has not yet been demonstrated and they may be unnecessary.

30%−40% decline in eGFR using repeat measurements to rule out transient acute effectsc Late stage: CKD G3b-G5 (eGFR <45 ml/min per 1.73 m2) End-stage kidney disease or 30%−40% decline in eGFRc End-stage kidney disease or doubling of serum creatinine level (or 40%–57% decline in eGFR)c CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; mGFR, measured glomerular filtration rate. a For example, in patients with macroalbuminuria. b Surrogates may include measures of activity of disease (e.g., in lupus nephritis) or kidney structure (e.g., in adult polycystic kidney disease). c The added value of eGFRs outside the routine study visit schedule has not yet been demonstrated and they may be unnecessary. When outcome measures reflect the underlying pathophysiology of a disease, the effects of an intervention may be large enough to detect reliably in small to medium sized trials. Therefore, surrogate outcomes are suited to smaller Phase 2 clinical trials used to establish proof-of-concept, optimal drug dose, and information on tolerability. Markers of glomerular or tubular damage, inflammation, fibrosis, etc., or a combination of markers may be considered at this phase. A demonstrable difference in the average eGFR (or change in the eGFR) between treatment groups may also be possible in small- to medium-size trials.

dose, and information on tolerability. Markers of glomerular or tubular damage, inflammation, fibrosis, etc., or a combination of markers may be considered at this phase. A demonstrable difference in the average eGFR (or change in the eGFR) between treatment groups may also be possible in small- to medium-size trials. In larger Phase 3 trials, it may also occasionally be appropriate for outcome measures in support of the primary outcome to include measures of structural damage or disease-specific markers. Acceptable surrogates include measuring total kidney volume in autosomal dominant polycystic kidney disease.22, 23 However, such trials must be sufficiently large to also provide reliable information on the safety of the intervention. Changes in eGFR over time may remain a more practical and acceptable method for assessing progression of kidney disease in many trials (Table 2).24

autosomal dominant polycystic kidney disease.22, 23 However, such trials must be sufficiently large to also provide reliable information on the safety of the intervention. Changes in eGFR over time may remain a more practical and acceptable method for assessing progression of kidney disease in many trials (Table 2).24 Directly measured GFR may occasionally be necessary if a treatment might influence variables used to measure eGFR through mechanisms other than effects on glomerular filtration (e.g., muscle mass changes or tubular secretion of creatinine). In the pediatric population, most clinical research is conducted in the setting of rare diseases, which, by definition, have small disease populations. Even so, studies in pediatric populations are subject to the same principles for evaluating progression of CKD as in adults. However, using creatinine as a marker of GFR is problematic in children younger than 2 years of age because creatinine levels rise rapidly during infancy. The European Medicines Agency provides guidance on extrapolating efficacy and safety data from adults to children to inform pediatric investigation plans.25

adults. However, using creatinine as a marker of GFR is problematic in children younger than 2 years of age because creatinine levels rise rapidly during infancy. The European Medicines Agency provides guidance on extrapolating efficacy and safety data from adults to children to inform pediatric investigation plans.25 An important current issue is whether change in albuminuria is acceptable as a surrogate marker of CKD progression.26 In the context of nephrotic syndrome, large changes in albuminuria are an acceptable marker of kidney disease activity. Albuminuria may also be appropriate in the setting of structural damage, but it may not be in the setting of hemodynamic dysfunction or acute reversible disease. Albuminuria may also be an appropriate surrogate if there is evidence that the effects of treatment are durable. Possible endpoints for evaluating treatments include prevention of incident macroalbuminuria, remission from macroalbuminuria to normoalbuminuria, and a predetermined decrease, such as a set quantitative change. More data are needed to better understand whether and how changes in albuminuria correspond to disease progression and how changes can be meaningfully applied to designing trial endpoints.27 The US National Kidney Foundation, the US Food and Drug Administration, and the European Medicines Agency will convene a meeting on March 15 to 16, 2018, to discuss these issues.28

nd how changes in albuminuria correspond to disease progression and how changes can be meaningfully applied to designing trial endpoints.27 The US National Kidney Foundation, the US Food and Drug Administration, and the European Medicines Agency will convene a meeting on March 15 to 16, 2018, to discuss these issues.28 Measurement of comorbidity and mortality Outcomes that measure aspects of disease status other than kidney structure or function may be disease-specific morbidity or mortality and can include assessments of quality of life or everyday functioning or economic impact. Outcomes should capture the expected, plausible treatment effects (benefits and harms), be relevant to patients and health care providers, and be appropriate for the phase of clinical development. Excess burden to patients, their families, health care providers, and research staff should be avoided. When available, measurement instruments that have operating characteristics known to be within acceptable limits should be used.

ts and health care providers, and be appropriate for the phase of clinical development. Excess burden to patients, their families, health care providers, and research staff should be avoided. When available, measurement instruments that have operating characteristics known to be within acceptable limits should be used. In determining outcome measures, patient and caregiver perspectives should be sought and considered for all clinical trials in nephrology. Standard definitions for key core outcomes identified as being important to patients on dialysis or with kidney transplants are being developed by the Standardised Outcomes in Nephrology initiative.29, 30 For example, outcomes that the Standardised Outcomes in Nephrology initiative has identified to be important to patients on hemodialysis include mortality, functioning of vascular access, and cardiovascular disease, but also include symptoms such as fatigue, pruritus, cognitive function, and functional limitations. The US Food and Drug Administration has provided specific guidance on how Patient Reported Outcomes, which include assessments of quality of life, can be developed and validated to assess these symptoms to support labeling claims.31 Examples for which streamlined Patient Reported Outcomes may be particularly important include trials of treatments for anemia (alongside outcomes related to clinical safety and efficacy).

which include assessments of quality of life, can be developed and validated to assess these symptoms to support labeling claims.31 Examples for which streamlined Patient Reported Outcomes may be particularly important include trials of treatments for anemia (alongside outcomes related to clinical safety and efficacy). Composite outcomes should combine components that make sense for the specific intervention, patient population, and disease state and should be of approximately comparable clinical importance. All-cause mortality is rarely an appropriate primary outcome in kidney trials because it is neither sensitive to any real effects on particular causes of death nor generalizable to different types of patients. It is preferable to create composite outcomes comprising related events that are all likely to be influenced favorably by treatment (and assess the effects on safety outcomes separately). If kidney and cardiovascular outcomes are to be combined in a single composite primary outcome, it is important to ensure that sufficient information will be available on both disease components to be able to guide treatment decisions. Similarly, the use of co-primary endpoints (for which an effect has only to be demonstrated on one of them) is not generally appropriate unless each is in some way relevant to patients and analyses of all such endpoints are adequately powered.

vailable on both disease components to be able to guide treatment decisions. Similarly, the use of co-primary endpoints (for which an effect has only to be demonstrated on one of them) is not generally appropriate unless each is in some way relevant to patients and analyses of all such endpoints are adequately powered. When available and appropriate for the particular clinical context under study, standardized disease outcome definitions that are feasible to apply at scale should be considered. However, if needed, new definitions should be considered in special situations such as heart failure in the context of dialysis. When a composite outcome is used, it is important to assess the effects of treatment on its components and related outcomes. Analyses of events that recur (e.g., vascular access procedures or hospitalization for heart failure) may be important to patients and payers, and appropriate statistical methodology to analyze recurrent events is available.32 Where continuous measures are used, clinically important differences should be defined and justified. Instruments assessing health-related quality of life can assess the economic impact of a treatment to inform payers. Application of such instruments should follow the same principles as those for other patient populations without kidney disease, and appropriately streamlined methods for gathering such data should be included in trial designs.

alth-related quality of life can assess the economic impact of a treatment to inform payers. Application of such instruments should follow the same principles as those for other patient populations without kidney disease, and appropriately streamlined methods for gathering such data should be included in trial designs. Optimizing trial conduct Increasing the number of large, important, and relevant clinical trials will require a culture shift within the nephrology community. A multipronged approach is required to improve study conduct and help all stakeholders to better understand the ways in which high-quality clinical trials improve patient care. One practical solution for streamlining the conduct of trials is integrating research processes and procedures into routine care (as has been done so successfully in oncology, diabetes, and cardiology33), and success in this depends on improved community awareness.

ich high-quality clinical trials improve patient care. One practical solution for streamlining the conduct of trials is integrating research processes and procedures into routine care (as has been done so successfully in oncology, diabetes, and cardiology33), and success in this depends on improved community awareness. Inefficiencies in clinical trial conduct jeopardize the ability to address important clinical questions and are a disservice to trial participants. Table 3 lists potential strategies to improve the efficiency and effectiveness of clinical trials in nephrology and in other disciplines. Such strategies would be expected to enhance the rights of participants (e.g., by ensuring that consent procedures provide information in an accessible form) as well as their safety and well-being (e.g., by minimizing the requirements for study visits and invasive tests or by more effective methods of pharmacovigilance). This work builds on the Quality by Design approaches developed by the Clinical Trials Transformation Initiative.34, 35Table 3 Strategies for minimizing issues that have a meaningful impact on the rights, safety, and well-being of trial participants or on the reliability of the trial conclusions (which will influence the care of future patients)

ality by Design approaches developed by the Clinical Trials Transformation Initiative.34, 35Table 3 Strategies for minimizing issues that have a meaningful impact on the rights, safety, and well-being of trial participants or on the reliability of the trial conclusions (which will influence the care of future patients) 1. Facilitating efficient and rapid recruitment 2. Streamlining the process of high-quality data collection (by assessing a limited number of critical data elements) 3. Maximizing adherence to study treatment and minimizing loss to follow-up 4. Improving the efficiency and appropriateness of trial monitoring (including using risk-based central statistical processes) 5. Rationalizing safety monitoring and pharmacovigilance activity (with more focus on the review of randomized comparisons of aggregated data by the unblinded Data and Safety Monitoring Boards) 6. Tailoring adjudication methods to focus on those events in which adjudication may materially influence interpretation of the results

alizing safety monitoring and pharmacovigilance activity (with more focus on the review of randomized comparisons of aggregated data by the unblinded Data and Safety Monitoring Boards) 6. Tailoring adjudication methods to focus on those events in which adjudication may materially influence interpretation of the results Recruitment Increasing participation in clinical trials is a major goal and requires a range of strategies (Table 4). Patients and clinicians should be made aware of the value of research participation and the importance of randomization. This could be done through education via videos, webinars, targeted advertising strategies, or peer group discussions. Educational efforts, both formal and informal, need to be dedicated, consistent, and constant. Patient advocacy groups for rare diseases have been successful at engaging patients and providers about the meaning and value of research and are a potential resource for ideas and collaboration.Table 4 Strategies to improve recruitment into kidney disease trials

nformal, need to be dedicated, consistent, and constant. Patient advocacy groups for rare diseases have been successful at engaging patients and providers about the meaning and value of research and are a potential resource for ideas and collaboration.Table 4 Strategies to improve recruitment into kidney disease trials Education ● Demonstrate to the kidney health community the value of research participation using visual media (i.e., social media, charity/patient advocacy group websites, webinars) and peer group discussion ● Provide nephrologists with examples of the importance of uncertainty ● Develop annual kidney clinical trials education for the community (providers and patients) • Improve knowledge of the principles of clinical trial design and conduct • Identify global and local barriers to conducting quality trials • Share successes/tools ● Increase trial awareness through local advertising and patient advocacy groups ● Develop systems for peer review of protocols for new trialists Improve information on potential trials ● Within individual health care systems and clinic settings: • Create a readily accessible repository of current and planned trials • Create or use existing electronic health care records or registries to identify eligible patients (particularly for rare diseases) Improve trial infrastructure ● Widen the type of health care services participating in trials Incentivize trial participation ● Acknowledge clinical research activities (e.g., using continuing medical education credits, “awards”) ● Nationally audit trial participation as a marker of quality of care ● Payers to reward randomization Make randomization easy ● Simplify consent procedures ● Integrate trial systems into the electronic systems used in routine practice Cross-collaborate with other specialties ● Develop trials with diabetologists, cardiologists, and other specialists

ipation as a marker of quality of care ● Payers to reward randomization Make randomization easy ● Simplify consent procedures ● Integrate trial systems into the electronic systems used in routine practice Cross-collaborate with other specialties ● Develop trials with diabetologists, cardiologists, and other specialists Processes for informing patients and health care team members about specific clinical trials should be systemically embedded in health care communities. At individual centers, repositories of current and planned studies should be accessible via various internet portals, and information about studies should be displayed in the waiting areas and offices of medical facilities. For recruiting into specific trials, it may be helpful to institute a process of “prescreening” whereby research coordinators develop lists of potentially eligible patients and, where permissible, provide those patients with information about the trial and “preconsent” them. This then enables recruitment to proceed rapidly once the trial receives full ethical and regulatory approval (and also helps to identify sites that will not have sufficient patients to contribute). Electronic health records may provide an opportunity to identify large numbers of potential participants who may be eligible and should be invited. Widespread invitation ensures that patients are empowered to decide whether they want to participate in a trial rather than waiting for their doctor to hand-select them.

). Electronic health records may provide an opportunity to identify large numbers of potential participants who may be eligible and should be invited. Widespread invitation ensures that patients are empowered to decide whether they want to participate in a trial rather than waiting for their doctor to hand-select them. Research champions within countries and regions should be recognized and identifiable. Clinical research organizations, academic research organizations, and networks of trialists should be encouraged to share information about the enrollment performance of individual study sites. This could speed completion of enrollment and reduce wasting of resources on poor performing sites. Excellence in research should also be recognized by national audits and by payers, with rewards given both for randomization and success in achieving quality data and completeness of follow-up.

of individual study sites. This could speed completion of enrollment and reduce wasting of resources on poor performing sites. Excellence in research should also be recognized by national audits and by payers, with rewards given both for randomization and success in achieving quality data and completeness of follow-up. Data collection The amount and type of data collected in a clinical trial affect the ability to recruit patients, follow their progress, and complete the trial. Inefficiencies in data collection increase trial costs, labor, and burden of participation (for both participants and the research team). To date, trials in patients with kidney disease have tended to collect too many data fields, most of which do not contribute to answering the main clinical question and lead to unnecessary complexity and difficulty in recruiting.34, 35 Researchers should identify small core datasets required for each study, minimize the frequency of measurements, and simplify the collection of data. The specific processes developed for an individual study will vary depending on the study’s purpose, type of intervention, available resources, and stage of development. Well-established national dialysis and transplant registries in many countries provide an opportunity to streamline data collection in trials. Focused kidney-specific templates that use consistent terminology, definitions, and sets of variables could be of value if these templates were integrated into registries or electronic health records.

al dialysis and transplant registries in many countries provide an opportunity to streamline data collection in trials. Focused kidney-specific templates that use consistent terminology, definitions, and sets of variables could be of value if these templates were integrated into registries or electronic health records. Depending on the circumstances, expensive central laboratory analyses may not be essential where an outcome is a measure of difference in a biomarker between randomized arms. Variation in calibration for a biomarker between laboratories has little impact on the magnitude of differences between randomized groups. For example, in the SHARP trial, analyses of differences in routine plasma (or serum) creatinine measured every 6 months at local hospital laboratories allowed low-cost, but reliable, assessment of the effects of ezetimibe/simvastatin on the progression of CKD.14 When using local laboratories, however, it is important to know their reporting units, reference ranges, and analytic methods to ensure that summary analyses are meaningful. There is increasing evidence that verification of clinical outcome data (usually referred to as outcome adjudication) may have little effect on the relative risk reductions reported by trials and that this process could also be streamlined in certain situations.36

Depending on the circumstances, expensive central laboratory analyses may not be essential where an outcome is a measure of difference in a biomarker between randomized arms. Variation in calibration for a biomarker between laboratories has little impact on the magnitude of differences between randomized groups. For example, in the SHARP trial, analyses of differences in routine plasma (or serum) creatinine measured every 6 months at local hospital laboratories allowed low-cost, but reliable, assessment of the effects of ezetimibe/simvastatin on the progression of CKD.14 When using local laboratories, however, it is important to know their reporting units, reference ranges, and analytic methods to ensure that summary analyses are meaningful. There is increasing evidence that verification of clinical outcome data (usually referred to as outcome adjudication) may have little effect on the relative risk reductions reported by trials and that this process could also be streamlined in certain situations.36 Maximizing adherence to treatment and follow-up procedures Approaches such as using “run-in” periods to identify patients who are unlikely to adhere to study treatment and/or attend clinics,17 minimizing unnecessary data collection, limiting excessive numbers of study visits (perhaps by arranging follow-up by telephone or electronic health records where possible), and expediting in-person visits (e.g., by avoiding lengthy waits in the hospital pharmacy) may all help maintain adherence and follow-up (Table 1). Adherence should be monitored centrally, and each study treatment “dropout” should prompt investigation into the reason and discussion as to whether study treatment can be restarted.

ting in-person visits (e.g., by avoiding lengthy waits in the hospital pharmacy) may all help maintain adherence and follow-up (Table 1). Adherence should be monitored centrally, and each study treatment “dropout” should prompt investigation into the reason and discussion as to whether study treatment can be restarted. In clinical studies, the term withdrawal of consent is problematic because of its lack of specificity and because it is frequently confused with a participant’s wish to stop study treatment or not undergo a certain trial procedure. Specific levels of withdrawal from the protocol-specified follow-up, which range from a patient not attending clinic visits but perhaps agreeing to clinical note review to an absolute withdrawal with no further data being provided, should be embedded in case report forms. To better capture follow-up data for patients who are no longer participating in a study, patients could be asked to agree that electronic health records can be accessed. This would allow for capturing information about their outcomes, adverse events, and concomitant medications during and after study participation.37 Engaging general practitioners and primary care providers could also help in establishing streamlined methods for complete study participant follow-up.

ds can be accessed. This would allow for capturing information about their outcomes, adverse events, and concomitant medications during and after study participation.37 Engaging general practitioners and primary care providers could also help in establishing streamlined methods for complete study participant follow-up. Trial monitoring Trial monitoring can be time-consuming and resource intensive, and therefore simplifying these processes can have a profound impact on reducing the burden of the trial conduct. When using direct electronic data entry, processes for trial monitoring and source data verification can become much more efficient.38 For example, risk-based approaches to monitoring (focusing on those data that are critical to trial quality) and central statistical monitoring (using study data to identify unusual patterns of performance) can reduce and prioritize site visits.

monitoring and source data verification can become much more efficient.38 For example, risk-based approaches to monitoring (focusing on those data that are critical to trial quality) and central statistical monitoring (using study data to identify unusual patterns of performance) can reduce and prioritize site visits. Safety reporting should be tailored for each trial protocol. In early phase development, rigorous detailed adverse event ascertainment is necessary, but this level of event recording may not be necessary when the safety profile of a treatment is well-known. During protocol development, regulators such as the US Food and Drug Administration and European Medicines Agency can advise on which specific adverse events need to be collected and which do not. Regulators can also advise on the level of information that needs to be collected. Clinical narratives are burdensome, may reduce trial participation, and should be focused only on those serious adverse events where such data may be informative, such as suspected unexpected serious adverse reactions.39 Reliable review of safety during a trial is best achieved by examining randomized comparisons of aggregated data by the unblinded Data and Safety Monitoring Boards.

n, and should be focused only on those serious adverse events where such data may be informative, such as suspected unexpected serious adverse reactions.39 Reliable review of safety during a trial is best achieved by examining randomized comparisons of aggregated data by the unblinded Data and Safety Monitoring Boards. Conclusions The lack of adequately powered randomized trials in nephrology has led to a problematic imbalance between the clinical need of patients with kidney disease and the amount of reliable evidence to inform practice. This KDIGO conference highlighted some of the key challenges faced by those trying to perform large trials. These include a lack of uncertainty among nephrologists who have often adopted treatments before adequate evidence of efficacy or safety is available, smaller treatment effects than were predicted from effects on surrogate biomarkers, inappropriate selection of outcomes including those with little relevance to patients or unlikely to be affected by treatment (e.g., all-cause mortality), difficulty in identifying large numbers of eligible patients, high levels of nonadherence to study treatment by overburdened patients, and overcomplicated trial conduct (Table 1). Adoption of the approaches outlined in this report has the potential to dramatically improve the quality of clinical trials in nephrology and substantially enhance the evidence base for the safe and effective treatment of patients with kidney disease.

overburdened patients, and overcomplicated trial conduct (Table 1). Adoption of the approaches outlined in this report has the potential to dramatically improve the quality of clinical trials in nephrology and substantially enhance the evidence base for the safe and effective treatment of patients with kidney disease. Disclosure This conference was sponsored by KDIGO and was in part supported by unrestricted educational grants from Abbvie, Achillion, Akebia Therapeutics, Alexion, Amgen, AstraZeneca, Bayer HealthCare, Fresenius Medical Care, KBP Biosciences, Keryx Biopharmaceuticals, Merck, Omeros, Relypsa, Roche, and Vifor Fresenius Medical Care – Renal Pharma. Appendix Other conference participants

Disclosure This conference was sponsored by KDIGO and was in part supported by unrestricted educational grants from Abbvie, Achillion, Akebia Therapeutics, Alexion, Amgen, AstraZeneca, Bayer HealthCare, Fresenius Medical Care, KBP Biosciences, Keryx Biopharmaceuticals, Merck, Omeros, Relypsa, Roche, and Vifor Fresenius Medical Care – Renal Pharma. Appendix Other conference participants Ali Abu-Alfa, Lebanon; Patrick Archdeacon, United States; Geoffrey A. Block, United States; Fergus J. Caskey, United Kingdom; Alfred K. Cheung, United States; Bruce Cooper, Australia; Jonathan C. Craig, Australia; Laura M. Dember, United States; Garabed Eknoyan, United States; Ron T. Gansevoort, The Netherlands; John S. Gill, Canada; Barbara Gillespie, United States; Tom Greene, United States; David C. Harris, Australia; Richard Haynes, United Kingdom; Brenda R. Hemmelgarn, Canada; Charles A. Herzog, United States; Thomas F. Hiemstra, United Kingdom; Lesley A. Inker, United States; Meg J. Jardine, Australia; Vivekanand Jha, India; Lixin Jiang, China; Kirsten L. Johansen, United States; Reshma Kewalramani, United States; Hiddo J. Lambers Heerspink, The Netherlands; Martin Lefkowitz, United States; Charmaine E. Lok, Canada; Fiona Loud, United Kingdom; Romaldas Mačiulaitis, United Kingdom; Dugan W. Maddux, United States; Franklin W. Maddux, United States; Magdalena Madero, Mexico; Segundo Mariz, United Kingdom; Michael Mauer, United States; Joseph V. Nally, Jr., United States; Masaomi Nangaku, Japan; Ikechi G. Okpechi, South Africa; Patrick S. Parfrey, Canada; Roberto Pecoits-Filho, Brazil; Brian J. G. Pereira, United States; Michael V. Rocco, United States; Patrick Rossignol, France; Franz Schaefer, Germany; Francesca Tentori, United States; Aliza Thompson, United States; Marcello Tonelli, Canada; Allison Tong, Australia; Robert D. Toto, United States; Katherine R. Tuttle, United States; Thorsten Vetter, United Kingdom; Angela Yee Moon Wang, Hong Kong; Faiez Zannad, France.

trick Rossignol, France; Franz Schaefer, Germany; Francesca Tentori, United States; Aliza Thompson, United States; Marcello Tonelli, Canada; Allison Tong, Australia; Robert D. Toto, United States; Katherine R. Tuttle, United States; Thorsten Vetter, United Kingdom; Angela Yee Moon Wang, Hong Kong; Faiez Zannad, France. Acknowledgments We thank all the attendees for their contributions to the Conference discussions and Jennifer King and Michael Cheung for their support in developing the manuscript.

see commentary on page 22 With over 250 million people affected worldwide, CKD is a common disease,1, 2 and its prevalence is expected to increase further with rising levels of obesity and diabetes and an aging population. In the US, approximately 7% of adults have stage 3-5 CKD.3 People with reduced kidney function have increased cardiovascular risk,4, 5 which is a key treatment target.

ide, CKD is a common disease,1, 2 and its prevalence is expected to increase further with rising levels of obesity and diabetes and an aging population. In the US, approximately 7% of adults have stage 3-5 CKD.3 People with reduced kidney function have increased cardiovascular risk,4, 5 which is a key treatment target. The SHARP trial showed that lowering LDL cholesterol with a combination of simvastatin 20 mg plus ezetimibe 10 mg daily for about 5 years safely reduced the risk of major atherosclerotic events (i.e., nonfatal myocardial infarction or coronary death, nonhemorrhagic stroke, or arterial revascularization procedure) in nondialysis patients with moderate-to-advanced CKD.6 Subsequently, a large meta-analysis of individual participant data from 28 trials of statin therapy, or, in the case of SHARP, statin plus ezetimibe, was coordinated by the Cholesterol Treatment Trialists’ (CTT) Collaboration.7 This showed that the relative reduction in major vascular events (i.e., major atherosclerotic event, noncoronary cardiac death, or hemorrhagic stroke) per 1 mmol/L reduction in LDL cholesterol ranged from 22% among participants with estimated glomerular filtration rate (eGFR) of ≥60 ml/min per 1.73 m2 to 15% among nondialysis participants with eGFR <30 ml/min per 1.73 m2, with a trend toward smaller risk reductions with lower eGFR and no evidence of clinical efficacy in dialysis patients. By design, SHARP did not assess the separate clinical effects of ezetimibe, but evidence of the effects on vascular outcomes of adding ezetimibe to simvastatin, albeit in patients with an acute coronary syndrome and without established CKD, has been provided by the Improved Reduction of Outcomes: Vytorin Efficacy International Trial (IMPROVE-IT).8 In this trial, for each mmol/L reduction in LDL cholesterol, the relative reduction in vascular events resulting from ezetimibe was consistent with that predicted by a meta-analysis of randomized trials of statin therapy alone.9

by the Improved Reduction of Outcomes: Vytorin Efficacy International Trial (IMPROVE-IT).8 In this trial, for each mmol/L reduction in LDL cholesterol, the relative reduction in vascular events resulting from ezetimibe was consistent with that predicted by a meta-analysis of randomized trials of statin therapy alone.9 Based on all randomized trial evidence, the Kidney Disease: Improving Global Outcomes (KDIGO) Clinical Practice Guideline for Lipid Management in CKD recommended use of a statin or statin/ezetimibe combination for adults ≥50 years old with an eGFR ≤60 ml/min per 1.73 m2 who are not on renal replacement therapy (RRT: chronic dialysis or kidney transplantation), and a statin alone for other adults with non-dialysis-dependent CKD.10 The 2018 ACC/AHA Multisociety Guideline on cholesterol management recommends initiating statin or statin/ezetimibe combination in ≥40-year-old CKD patients with an increased (>7.5%) 10-year risk of atherosclerotic cardiovascular disease.11, 12, 13 In the UK, the National Institute for Health and Care Excellence recommends atorvastatin 20 mg daily for the primary prevention of cardiovascular disease in people with eGFR <60 ml/min per 1.73 m2 and suggests considering a higher dose, and/or combination with ezetimibe, in more advanced CKD (e.g., eGFR <30 ml/min per 1.73 m2), for secondary cardiovascular disease prevention, or when the desired cholesterol reduction has not been achieved.14, 15

f cardiovascular disease in people with eGFR <60 ml/min per 1.73 m2 and suggests considering a higher dose, and/or combination with ezetimibe, in more advanced CKD (e.g., eGFR <30 ml/min per 1.73 m2), for secondary cardiovascular disease prevention, or when the desired cholesterol reduction has not been achieved.14, 15 Despite these recommendations, it remains unclear which statin/ezetimibe combination is the most cost-effective treatment in moderate-to-advanced CKD (i.e., offers the greatest benefits and is affordable). A cost-effectiveness study of 5-year simvastatin plus ezetimibe treatment in SHARP concluded that high-intensity generic treatments, rather than more expensive proprietary treatments, are cost-effective.16 However, a more pertinent question for health care providers is the cost-effectiveness of long-term treatments. We used the SHARP CKD-CVD policy model17 to project lifetime risks of CVD and CKD progression and the net effects and cost-effectiveness of long-term statin/ezetimibe treatments in categories of patients with CKD.

er, a more pertinent question for health care providers is the cost-effectiveness of long-term treatments. We used the SHARP CKD-CVD policy model17 to project lifetime risks of CVD and CKD progression and the net effects and cost-effectiveness of long-term statin/ezetimibe treatments in categories of patients with CKD. Results CKD patients Results are presented only for nondialysis patients, since there is no clear evidence that LDL cholesterol-lowering therapy is effective in dialysis patients.7 The mean age of the 6235 patients in the SHARP study6, 18 was 63 years (SD, 12); 62% were male; 23% had diabetes; and 15% had a prior history of (noncoronary) vascular disease. There were 2020 participants with stage 3B disease; 2767 with stage 4 disease; and 1448 with stage 5 disease not on dialysis. The participants’ median 5-year cardiovascular risk ranged from 10% in those with stage 3B to 20% in nondialysis participants in stage 5, and from 6% to 31% across the 3 categories of baseline risk (Table 1). Within each stage of CKD, participants at higher risk at baseline were older and more likely to have previous vascular disease or diabetes (Supplementary Table S1).Table 1 Characteristics of nondialysis SHARP participants by CKD stage and cardiovascular disease risk at baseline

3 categories of baseline risk (Table 1). Within each stage of CKD, participants at higher risk at baseline were older and more likely to have previous vascular disease or diabetes (Supplementary Table S1).Table 1 Characteristics of nondialysis SHARP participants by CKD stage and cardiovascular disease risk at baseline By CKD stage at baseline By 5-year risk of cardiovascular disease at baseline CKD stage 3B* CKD stage 4 CKD stage 5, not on dialysis Low (<10%) Medium (10%-20%) High (≥20%) n=2020 n=2767 n=1448 n=2151 n=2045 n=2039 Age, years 62 (11) 64 (12) 62 (12) 53 (8) 65 (9) 71 (9) Male 1461 (72%) 1653 (60%) 760 (52%) 1080 (50%) 1337 (65%) 1457 (71%) Current smoker 271 (13%) 336 (12%) 162 (11%) 219 (10%) 273 (13%) 277 (14%) Previous vascular disease 283 (14%) 430 (16%) 217 (15%) 43 (2%) 146 (7%) 741 (36%) Diabetes mellitus 469 (23%) 662 (24%) 293 (20%) 106 (5%) 345 (17%) 973 (48%) Treated hypertension 1701 (84%) 2389 (86%) 1261 (87%) 1841 (86%) 1751 (85%) 1767 (87%) Body-mass index, kg/m2 28 (5) 28 (6) 27 (5) 27 (5) 28 (5) 27 (6) Diastolic blood pressure, mm Hg 80 (13) 79 (13) 80 (12) 82 (12) 80 (12) 77 (13) Systolic blood pressure, mm Hg 139 (20) 139 (21) 141 (21) 132 (17) 139 (20) 147 (22) LDL cholesterol, mmol/L 2.9 (0.8) 2.9 (0.8) 2.7 (0.9) 2.9 (0.8) 2.9 (0.9) 2.8 (0.9) HDL cholesterol, mmol/L 1.1 (0.3) 1.1 (0.3) 1.1 (0.3) 1.2 (0.3) 1.1 (0.3) 1.1 (0.3) Estimated 5-year risk of cardiovascular disease, median (IQR) 10% (6%, 18%) 14% (9%, 24%) 20% (11%, 32%) 6% (5%, 8%) 14% (12%, 17%) 31% (24%, 42%) CKD stage at baseline CKD stage 3Ba 967 (45%) 649 (32%) 404 (20%) CKD stage 4 882 (41%) 968 (47%) 917 (45%) CKD stage 5, not on dialysis 302 (14%) 428 (21%) 718 (35%) CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; IQR, interquartile range; SHARP, Study of Heart and Renal Protection.

(24%, 42%) CKD stage at baseline CKD stage 3Ba 967 (45%) 649 (32%) 404 (20%) CKD stage 4 882 (41%) 968 (47%) 917 (45%) CKD stage 5, not on dialysis 302 (14%) 428 (21%) 718 (35%) CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; IQR, interquartile range; SHARP, Study of Heart and Renal Protection. Results are shown as mean (SD) or N (%), as appropriate, unless otherwise specified. Ten participants on kidney transplant at baseline were excluded. a 338 (17%) of participants with CKD stage 3A (eGFR 60-45 ml/min per 1.73 m2). Effects of statin/ezetimibe treatments Of the treatments considered, the least potent was ezetimibe 10 mg daily, which reduced LDL cholesterol by 18.5%; the most potent was atorvastatin 40 mg plus ezetimibe 10 mg daily, which reduced LDL cholesterol by 60.1%. Atorvastatin 40 mg daily reduced LDL cholesterol by 48% (Supplementary Table S2). The proportional reductions in risk of major vascular events with use of ezetimibe 10 mg daily were 8% (99% confidence interval [CI], 2%, 14%) in stage 3B, 8% (–1%, 17%) in stage 4, and 8% (–1%, 16%) in stage 5 not on dialysis. The proportional reductions in risk of major vascular events with use of atorvastatin 40 mg daily were 20% (6%, 33%) in stage 3B, 20% (–3%, 38%) in stage 4, and 19% (–3%, 36%) in stage 5 not on dialysis. The proportional reductions in risk of major vascular events with use of atorvastatin 40 mg plus ezetimibe 10 mg daily were 25% (7%, 40%) in stage 3B, 25% (–4%, 45%) in stage 4, and 23% (–3%, 43%) in stage 5 not on dialysis (Supplementary Table S3).

(–3%, 38%) in stage 4, and 19% (–3%, 36%) in stage 5 not on dialysis. The proportional reductions in risk of major vascular events with use of atorvastatin 40 mg plus ezetimibe 10 mg daily were 25% (7%, 40%) in stage 3B, 25% (–4%, 45%) in stage 4, and 23% (–3%, 43%) in stage 5 not on dialysis (Supplementary Table S3). US cost-effectiveness of statin/ezetimibe treatments In all categories of CKD patients, at current statin/ezetimibe prices (January 2019), treatment with ezetimibe 10 mg was both less effective and more expensive than treatment with atorvastatin 20 mg. Therefore, in health economic jargon, ezetimibe 10 mg was “dominated” by atorvastatin 20 mg; rosuvastatin 20 mg was dominated by similarly effective and slightly cheaper atorvastatin 40 mg; simvastatin 20 mg plus ezetimibe 10 mg was dominated by atorvastatin 40 mg; and atorvastatin 20 mg plus ezetimibe 10 mg was dominated by atorvastatin 40 mg (Supplementary Table S2). Additionally, atorvastatin 20 mg was projected to produce very similar health benefits at a similar additional cost per QALY to atorvastatin 40 mg (Supplementary Table S4). Therefore, we present results for atorvastatin 40 mg and atorvastatin 40 mg plus ezetimibe 10 mg only. However, since atorvastatin 20 mg could be considered a less intensive treatment option, and ezetimibe 10 mg could be used by patients who cannot tolerate or do not use a statin-based regimen, we also present results for these regimens.

nt results for atorvastatin 40 mg and atorvastatin 40 mg plus ezetimibe 10 mg only. However, since atorvastatin 20 mg could be considered a less intensive treatment option, and ezetimibe 10 mg could be used by patients who cannot tolerate or do not use a statin-based regimen, we also present results for these regimens. Lifetime use of atorvastatin 40 mg is projected to increase life expectancy by 0.26 years (0.23 QALYs) at a net cost of $20,300/QALY in patients with stage 3B disease; 0.37 years (0.31 QALYs) at a net cost of $44,200/QALY in patients with stage 4 disease; and 0.31 years (0.26 QALYs) at $78,200/QALY in patients with stage 5 disease not on dialysis. Similarly, it would increase life expectancy by 0.29 years (0.26 QALYs) at $38,100/QALY in those at low cardiovascular risk (<10%); by 0.32 years (0.27 QALYs) at $41,000/QALY in those at medium cardiovascular risk (10%–20%); and by 0.36 years (0.29 QALYs) at $55,000/QALY in those at high cardiovascular risk (≥20%) (Table 2). Within each cardiovascular risk group, the net cost per QALY was lowest for patients in stage 3B and highest for those in stage 5 not on dialysis, whereas within each CKD stage, net costs per QALY were similar at all levels of cardiovascular risk (Supplementary Figure S1). In almost all subgroups, patients who were younger at treatment initiation were projected to benefit the most but at the highest net cost per QALY. For example, patients <60 years old were projected to gain between 0.28 QALYs (if at low risk) and 0.51 QALYs (if at high risk), at a net cost, respectively, of $42,000 to $76,400/QALY; whereas the respective estimates for the patients ≥70 years old were 0.13 (low risk) to 0.22 (high risk) QALYs, at a net cost, respectively, of $10,700 to $42,300/QALY (Supplementary Table S5).Table 2 Health benefits and cost-effectiveness of statin-based treatments in moderate-to-advanced nondialysis CKD patients

ALY; whereas the respective estimates for the patients ≥70 years old were 0.13 (low risk) to 0.22 (high risk) QALYs, at a net cost, respectively, of $10,700 to $42,300/QALY (Supplementary Table S5).Table 2 Health benefits and cost-effectiveness of statin-based treatments in moderate-to-advanced nondialysis CKD patients Category of CKD patient Atorvastatin 40 mg dailya compared to no LDL-C lowering treatment Ezetimibe 10 mg plus atorvastatin 40 mg daily compared to atorvastatin 40 mg daily Life-years gained QALYs gained Additional cost per QALYb Life-years gained QALYs gained Additional cost per QALYb (A) US health care setting By CKD stage at baseline CKD stage 3Bc 0.26 0.23 $20,300 0.06 0.05 $43,600 CKD stage 4 0.37 0.31 $44,200 0.08 0.07 $58,400 CKD stage 5, not on dialysis 0.31 0.26 $78,200 0.07 0.06 $91,500 By 5-year risk of cardiovascular disease at baseline Low (<10%) 0.29 0.26 $38,100 0.06 0.06 $65,100 Medium (10%-20%) 0.32 0.27 $41,000 0.07 0.06 $56,700 High (≥20%) 0.36 0.29 $55,000 0.08 0.07 $64,400 (B) UK health care setting By CKD stage at baseline CKD stage 3Bc 0.28 0.25 £3800 0.07 0.06 £12,500 CKD stage 4 0.42 0.33 £10,500 0.09 0.07 £16,000 CKD stage 5, not on dialysis 0.37 0.29 £18,900 0.09 0.07 £23,900 By 5-year risk of cardiovascular disease at baseline Low (<10%) 0.33 0.29 £7,900 0.08 0.07 £17,800 Medium (10%-20%) 0.36 0.29 £9,400 0.08 0.07 £15,200 High (≥20%) 0.40 0.29 £14,200 0.09 0.07 £17,800 CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; LDL-C, low-density lipoprotein cholesterol; QALY, quality-adjusted life-year; UK, United Kingdom; US, United States.

0.29 £7,900 0.08 0.07 £17,800 Medium (10%-20%) 0.36 0.29 £9,400 0.08 0.07 £15,200 High (≥20%) 0.40 0.29 £14,200 0.09 0.07 £17,800 CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; LDL-C, low-density lipoprotein cholesterol; QALY, quality-adjusted life-year; UK, United Kingdom; US, United States. The CKD and cardiovascular risk categories are derived directly from the 6235 moderate-to-advanced non–dialysis-dependent CKD patients in the Study of Heart and Renal Protection (SHARP). a Atorvastatin 20 mg daily was projected to produce only slightly smaller health benefits at similar additional cost per QALY to atorvastatin 40 mg daily (see Supplementary Tables S4 and S7 for detailed results) and could be considered as an alternative less intensive treatment option. b Costs and outcomes discounted at 3% per annum (US) or 3.5% per annum (UK). c 338 (17%) of participants with CKD stage 3A (eGFR 60–45 ml/min per 1.73 m2).

a Atorvastatin 20 mg daily was projected to produce only slightly smaller health benefits at similar additional cost per QALY to atorvastatin 40 mg daily (see Supplementary Tables S4 and S7 for detailed results) and could be considered as an alternative less intensive treatment option. b Costs and outcomes discounted at 3% per annum (US) or 3.5% per annum (UK). c 338 (17%) of participants with CKD stage 3A (eGFR 60–45 ml/min per 1.73 m2). Adding ezetimibe 10 mg was estimated to provide further benefits: there were an additional 0.06 years (0.05 QALYs, net cost $43,600/QALY) in stage 3B; 0.08 years (0.07 QALYs, net cost $58,400/QALY) in stage 4; and 0.07 years (0.06 QALYs, net cost $91,500/QALY) in stage 5 not on dialysis. There were an additional 0.06 years (0.06 QALYs, net cost $65,100/QALY) in low-risk patients; 0.07 years (0.06 QALYs, net cost $56,700/QALY) in medium-risk patients; and 0.08 years (0.07 QALYs, net cost $64,400/QALY) in high-risk patients (Table 2). At the $100,000/QALY cost-effectiveness threshold, atorvastatin 40 mg plus ezetimibe 10 mg would be considered cost-effective with a >95% probability in all patients except those in stage 5 not on dialysis (76%) (Figure 1).Figure 1 Probability of a statin-based treatment to be cost-effective in moderate-to-advanced nondialysis chronic kidney disease (CKD) patients. Results shown for treatments on the cost-effectiveness frontier (i.e., the most cost-effective treatment for a given value of willingness to pay) within the range of willingness-to-pay values per quality-adjusted life-year (QALY). Typical cost-effectiveness thresholds are represented with dashed horizontal lines. Atorvastatin 20 mg daily was largely dominated by atorvastatin 40 mg daily and was omitted from the graph. LDL-C, low-density lipoprotein cholesterol; UK, United Kingdom; US, United States.

values per quality-adjusted life-year (QALY). Typical cost-effectiveness thresholds are represented with dashed horizontal lines. Atorvastatin 20 mg daily was largely dominated by atorvastatin 40 mg daily and was omitted from the graph. LDL-C, low-density lipoprotein cholesterol; UK, United Kingdom; US, United States. Similar to treatment with atorvastatin 40 mg alone, the net cost per QALY with the combination of atorvastatin 40 mg and ezetimibe 10 mg daily, compared to atorvastatin 40 mg daily, was lowest for patients in stage 3B and highest for those in stage 5 not on dialysis, at each level of risk (Supplementary Figure S1). Ezetimibe 10 mg daily, compared to no lipid-lowering treatment, was projected to result in 0.09 extra QALYs in stage 3B, 0.13 extra QALYs in stage 4, and 0.10 extra QALYs in stage 5 not on dialysis, at a net cost, respectively, of $31,000, $50,600, and $84,200/QALY, and 0.11 extra QALYs in low-risk patients, 0.11 extra QALYs in medium-risk patients, and 0.12 extra QALYs in high-risk patients at a net cost, respectively, of $50,900, $48,300, and $58,800/QALY (Supplementary Table S6).

LYs in stage 5 not on dialysis, at a net cost, respectively, of $31,000, $50,600, and $84,200/QALY, and 0.11 extra QALYs in low-risk patients, 0.11 extra QALYs in medium-risk patients, and 0.12 extra QALYs in high-risk patients at a net cost, respectively, of $50,900, $48,300, and $58,800/QALY (Supplementary Table S6). UK cost-effectiveness of statin/ezetimibe treatments In the UK setting, ezetimibe is now also available at low prices from generic treatment manufacturers (£0.074/d, January 2019). The treatment benefits, cost-effectiveness, and policy implications for different statin/ezetimibe treatments (i.e., atorvastatin 40 mg daily and ezetimibe 10 mg daily alone and in combination with a statin) were similar to those in the US (Table 2, Figure 1, Supplementary Tables S5–S7, Supplementary Figure S1).

2019). The treatment benefits, cost-effectiveness, and policy implications for different statin/ezetimibe treatments (i.e., atorvastatin 40 mg daily and ezetimibe 10 mg daily alone and in combination with a statin) were similar to those in the US (Table 2, Figure 1, Supplementary Tables S5–S7, Supplementary Figure S1). Sensitivity analyses The cost-effectiveness results were only minimally sensitive to further falls in the price of ezetimibe (Figure 2). If ezetimibe 10 mg were priced at less than $0.323/d (£0.019/d in the UK), its net cost per QALY would be under $100,000 (£20,000) for all nondialysis CKD patients.Figure 2 Cost-effectiveness of adding ezetimibe 10 mg to atorvastatin 40 mg daily for moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, at different ezetimibe cost. The CKD and cardiovascular risk categories are derived directly from the 6235 moderate-to-advanced non–dialysis-dependent CKD patients in the Study of Heart and Renal Protection (SHARP). Typical cost-effectiveness thresholds are represented with dashed horizontal lines. *A total of 338 (17%) participants with CKD stage 3A (estimated glomerular filtration rate [eGFR] 60–45 ml/min per 1.73 m2). At the $100,000/quality-adjusted life-year [QALY] threshold in the United States (US) (a), ezetimibe 10 mg daily becomes cost-effective in all categories of patients when its price reaches $0.323/d. At the £20,000/QALY threshold in the United Kingdom (UK) (b), ezetimibe 10 mg daily becomes cost-effective in all categories of patients when its price reaches £0.019/d.

] threshold in the United States (US) (a), ezetimibe 10 mg daily becomes cost-effective in all categories of patients when its price reaches $0.323/d. At the £20,000/QALY threshold in the United Kingdom (UK) (b), ezetimibe 10 mg daily becomes cost-effective in all categories of patients when its price reaches £0.019/d. When analyses were repeated with the annual treatment costs for RRT assumed similar to those for CKD stage 5 not on dialysis, the net cost per QALY with statin/ezetimibe treatments decreased substantially. In the US setting, the net cost per QALY with atorvastatin 40 mg daily decreased from $20,300 to $5000 in stage 3B, from $44,200 to $6900 in stage 4, and from $78,200 to $7100 in CKD stage 5 not on dialysis. The net cost per QALY for atorvastatin 40 mg plus ezetimibe 10 mg daily decreased from $43,600 to $27,500 in stage 3B, from $58,400 to $20,600 in stage 4, and from $91,500 to $19,900 in stage 5 not on dialysis. The effect was similar in the UK setting (Supplementary Table S8). The net costs per QALY only minimally increased when potential adverse effects of statin/ezetimibe treatments and their costs were projected (Supplementary Table S9). Reduced compliance with treatment was projected to result in lower health benefits but also lower incremental hospital costs and did not materially affect the results (Supplementary Table S10).

creased when potential adverse effects of statin/ezetimibe treatments and their costs were projected (Supplementary Table S9). Reduced compliance with treatment was projected to result in lower health benefits but also lower incremental hospital costs and did not materially affect the results (Supplementary Table S10). Discussion Lowering LDL cholesterol with statin-based treatments safely reduces cardiovascular risk in patients with moderate-to-advanced CKD who are not receiving maintenance dialysis, but there is no evidence that such treatment is effective in dialysis patients.7 We report that under the standard cost-effectiveness assumptions (i.e., a threshold of $100,000/QALY [£20,000 to £30,000/QALY in the UK]), low-cost statin treatments (e.g., atorvastatin 40 mg daily) are cost-effective in nondialysis CKD patients. Ezetimibe has recently come off patent in both countries and, at current prices ($0.203/d and £0.074/d, respectively), adding ezetimibe 10 mg to atorvastatin 40 mg daily is a cost-effective option in nondialysis CKD patients and would thus be the treatment of choice. The results remain robust across a range of sensitivity analyses.

recently come off patent in both countries and, at current prices ($0.203/d and £0.074/d, respectively), adding ezetimibe 10 mg to atorvastatin 40 mg daily is a cost-effective option in nondialysis CKD patients and would thus be the treatment of choice. The results remain robust across a range of sensitivity analyses. Despite the higher net cost per QALY, the finding that low-cost generic statins/ezetimibe are cost-effective for primary prevention of CVD in CKD is consistent with results for the general population at an increased cardiovascular risk.19, 20 Unlike general population estimates, in non–dialysis-dependent CKD, the net cost per QALY was generally higher in patients at more advanced CKD stage and/or higher predicted cardiovascular risk, especially for very low-cost treatments (e.g., atorvastatin 40 mg or ezetimibe 10 mg at a lower price). This is driven by reduced life expectancy, higher end-stage kidney disease risks, and RRT costs, which are incurred during the gained life expectancy and are substantial in more advanced CKD. Statin-based treatments would also be more cost-effective in patients receiving a kidney transplant, with lower RRT costs in the years after the transplantation.

pectancy, higher end-stage kidney disease risks, and RRT costs, which are incurred during the gained life expectancy and are substantial in more advanced CKD. Statin-based treatments would also be more cost-effective in patients receiving a kidney transplant, with lower RRT costs in the years after the transplantation. The present study builds upon a study by Erickson et al.21 and our findings support its conclusion that cheaper generic statins are likely cost-effective in CKD. However, our study takes a substantially more detailed look into the long-term effects of individual treatments across categories of CKD patients. We used the rich individual data from the SHARP study, which enabled us to present results generalizable to CKD patients at similar cardiovascular risk and/or CKD stage. We also used results from the CTT individual participant meta-analysis of 28 large trials, which provide the best available estimates for cardiovascular risk reductions with statin-based treatments at different levels of renal function.7 Linking these with the potency of individual statin/ezetimibe regimens enabled evaluation of the cost-effectiveness of different regimens reliably.

lysis of 28 large trials, which provide the best available estimates for cardiovascular risk reductions with statin-based treatments at different levels of renal function.7 Linking these with the potency of individual statin/ezetimibe regimens enabled evaluation of the cost-effectiveness of different regimens reliably. In the present study, treatment with rosuvastatin 20 mg was similarly effective but slightly more expensive than atorvastatin 40 mg (atorvastatin 40 mg is available at $0.103/d in the US and £0.034/d in the UK, compared to $0.119/d and £0.077/d, respectively, for rosuvastatin 20 mg). Small further fluctuations in prices would make rosuvastatin a cost-effective option. The use of less potent treatments (e.g., atorvastatin 20 mg or ezetimibe 10 mg alone) achieves smaller cardiovascular benefits. In general, statin/ezetimibe treatments that achieve larger reductions in LDL cholesterol are expected to achieve greater health benefits and, if available at low cost, be cost-effective. In the present analyses, we did not consider more potent statin treatments such as atorvastatin 80 mg and rosuvastatin 40 mg, since they are not routinely used in CKD patients due to safety concerns. However, if future evidence indicates safety of these regimens in CKD, our analysis suggests that, if available at low cost, they will be cost-effective.

ot consider more potent statin treatments such as atorvastatin 80 mg and rosuvastatin 40 mg, since they are not routinely used in CKD patients due to safety concerns. However, if future evidence indicates safety of these regimens in CKD, our analysis suggests that, if available at low cost, they will be cost-effective. Several limitations of the present analyses should be acknowledged. First, SHARP included only CKD patients without a prior history of myocardial infarction or coronary revascularization, whereas, in routine practice, coronary heart disease is highly prevalent in people with moderate-to-advanced CKD. Therefore, patients outside SHARP at similar CKD stages are likely to be at higher cardiovascular risk. However, the cost-effectiveness estimates corresponding to categories of risk are likely to be generalizable to such patients. Second, since SHARP did not directly assess the effects of ezetimibe monotherapy in patients with CKD, we incorporated an assumption that its effects on clinical outcomes are equivalent, per mmol/L reduction in LDL cholesterol, to those of statins. This assumption was derived from the results of the CTT meta-analysis of the effects of statin-based regimens at different levels of eGFR7 and by the IMPROVE-IT trial, which was conducted in people with an acute coronary syndrome treated with statin8 and demonstrated further cardiovascular risk reductions with ezetimibe similar to those with statin-only regimens achieving similar LDL cholesterol reductions.9 Third, the effect of treatments on non–health care costs, such as productivity or long-term care costs, were not included in the present analysis, which was conducted exclusively from a health services perspective. Finally, safety of statin-based regimens has been a subject of debate despite the acknowledgement that any adverse effects are rare and benefits strongly outweigh any harm.22, 23 We did not include adverse effects in our primary analysis due to the paucity of data for different treatments. However, sensitivity analyses incorporating estimated rates of potential adverse effects on muscle (e.g., myopathy and rhabdomyolysis) and diabetes yielded similar results.

s strongly outweigh any harm.22, 23 We did not include adverse effects in our primary analysis due to the paucity of data for different treatments. However, sensitivity analyses incorporating estimated rates of potential adverse effects on muscle (e.g., myopathy and rhabdomyolysis) and diabetes yielded similar results. In conclusion, statin-based treatments effectively reduce cardiovascular risk in nondialysis patients with CKD and, at current prices and cost-effectiveness thresholds, the available evidence suggests that low-cost statin/ezetimibe combination therapy is cost-effective. The most cost-effective regimen is one that maximizes the dose of statin chosen without compromising safety. Methods The SHARP CKD-CVD policy model The SHARP CKD-CVD policy model, which is a Markov state-transition model developed using the SHARP study6, 18 data and validated in 3 external CKD cohorts,17 was used to project cardiovascular events, CKD progression, health care costs, and health-related quality of life (QoL). A full description of the model has been published elsewhere.17 Briefly, it is based on parametric risk equations (3 survival equations for cardiovascular outcomes, a multinomial regression and a logistic regression for CKD progression, and 2 linear regressions predicting hospital costs and QoL). Each equation includes a range of clinically and/or statistically important covariates, including the patient’s sociodemographic status, comorbidities, and risk factors (eg age, most recent CKD stage, and detailed cardiovascular disease history.

progression, and 2 linear regressions predicting hospital costs and QoL). Each equation includes a range of clinically and/or statistically important covariates, including the patient’s sociodemographic status, comorbidities, and risk factors (eg age, most recent CKD stage, and detailed cardiovascular disease history. The model simulates annual risks of dying from vascular and nonvascular causes; experiencing a major atherosclerotic event or a hemorrhagic stroke; and progressing through CKD stages 3B (30 ≤ eGFR <45 ml/min per 1.73 m2), 4 (15 ≤ eGFR <30 ml/min per 1.73 m2), and 5 (eGFR <15 ml/min per 1.73 m2), or having dialysis or kidney transplant (Supplementary Figure S2). CKD patients Results are presented only for the 6235 SHARP nondialysis patients (since previous analyses have shown that lowering LDL cholesterol is not clinically effective in dialysis patients7); all such patients had moderate-to-advanced CKD (Table 1, Supplementary Table S1). At baseline, the study participants were categorized according to their CKD stage: (i) CKD stage 3B; (ii) CKD stage 4; or (iii) CKD stage 5 not on dialysis. They were also categorized by their 5-year risk of major vascular event16: (i) low (<10%); (ii) medium (≥10%, <20%); and (iii) high (≥20%) risk. eGFR was calculated using the CKD-EPI equation.24 The rates of nonvascular mortality were obtained from relevant population data (see Supplementary Table S11 for US rates and Schlackow et al.17 for UK rates).

ir 5-year risk of major vascular event16: (i) low (<10%); (ii) medium (≥10%, <20%); and (iii) high (≥20%) risk. eGFR was calculated using the CKD-EPI equation.24 The rates of nonvascular mortality were obtained from relevant population data (see Supplementary Table S11 for US rates and Schlackow et al.17 for UK rates). Effects of statin/ezetimibe treatments We considered a range of statin/ezetimibe regimens that are believed to be safe in CKD, including ezetimibe 10 mg, atorvastatin 20 mg, atorvastatin 40 mg, rosuvastatin 20 mg, simvastatin 20 mg plus ezetimibe 10 mg, atorvastatin 20 mg plus ezetimibe 10 mg, and atorvastatin 40 mg plus ezetimibe 10 mg daily. Treatment effects were projected on the risks of vascular death and major vascular events but not CKD progression25 or nonvascular mortality.6, 7 For each statin/ezetimibe regimen, the effects on vascular endpoints were expressed as relative risk reductions and evaluated separately for each CKD stage in 2 steps. First, the absolute reductions in LDL cholesterol were calculated using the expected proportional reductions in LDL cholesterol (Supplementary Table S2) and the mean LDL cholesterol of the SHARP participants in the respective CKD stage (Table 1, Supplementary Table S3). Second, these absolute reductions were combined with the rate ratios for vascular events per 1 mmol/L reduction in LDL cholesterol reported in the individual participant data meta-analysis by the CTT Collaboration (Supplementary Table S3).

HARP participants in the respective CKD stage (Table 1, Supplementary Table S3). Second, these absolute reductions were combined with the rate ratios for vascular events per 1 mmol/L reduction in LDL cholesterol reported in the individual participant data meta-analysis by the CTT Collaboration (Supplementary Table S3). Costs and QoL Annual hospital costs of managing a patient with CKD and cardiovascular complications were based on published data (see Supplementary Table S12 for the US costs and Kent et al.26 for the UK costs) and inflated to year 2015 using the Consumer Price Index27 (US) or Hospital & Community Health Services Index28 (UK). The costs of statin and ezetimibe treatments (January 2019) were obtained from the National Average Drug Acquisition Cost (NADAC) reports29 (US) and NHS Electronic Drug Tariff30 (UK) (Supplementary Table S2). Patients’ health-related QoL was derived from responses to the EuroQoL 5-dimensions 3-level (EQ-5D-3L) questionnaire31 completed by participants at the final SHARP follow-up visit and using the US32 or UK33 EQ-5D-3L utility tariffs (see Supplementary Table S13 for US estimates and Schlackow et al.17 for UK estimates) and stratified by CKD stage, cardiovascular morbidity, and other characteristics.

ons 3-level (EQ-5D-3L) questionnaire31 completed by participants at the final SHARP follow-up visit and using the US32 or UK33 EQ-5D-3L utility tariffs (see Supplementary Table S13 for US estimates and Schlackow et al.17 for UK estimates) and stratified by CKD stage, cardiovascular morbidity, and other characteristics. Cost-effectiveness analyses The cost-effectiveness analyses were performed from the perspectives of the US and UK health care systems. Health outcomes and costs were projected with lifelong treatment with each statin/ezetimibe regimen as well as no treatment, until patients reached 95 years of age or died. Costs and QALYs were discounted at an annual rate of 3% (US34) or 3.5% (UK35). For each treatment, the incremental cost-effectiveness ratios were calculated as the incremental cost per QALY gained with the treatment against the next less effective (and not dominated) treatment.36 Results are presented for categories of participants by CKD stage and, separately, by cardiovascular risk at baseline. Uncertainty in the cost-effectiveness was assessed using the nonparametric bootstrap,37 with the analysis replicated on 1000 sets of risk, cost, and QoL parameter estimates derived from refitting the original SHARP CKD-CVD risk equations17 on bootstrapped SHARP data or, for US hospital costs, using sampled values from the parametric distribution of costs (Supplementary Table S12).38 Uncertainty in the treatment effects was incorporated using values sampled from the respective lognormal distribution corresponding to the relative risk (99% CI) reported by the CTT collaboration.7 Cost-effectiveness acceptability curves were derived to summarize the probability of each treatment being cost-effective at different levels of willingness-to-pay thresholds.36

using values sampled from the respective lognormal distribution corresponding to the relative risk (99% CI) reported by the CTT collaboration.7 Cost-effectiveness acceptability curves were derived to summarize the probability of each treatment being cost-effective at different levels of willingness-to-pay thresholds.36 A schematic of our approach is presented in Supplementary Figure S3.

using values sampled from the respective lognormal distribution corresponding to the relative risk (99% CI) reported by the CTT collaboration.7 Cost-effectiveness acceptability curves were derived to summarize the probability of each treatment being cost-effective at different levels of willingness-to-pay thresholds.36 A schematic of our approach is presented in Supplementary Figure S3. Sensitivity analyses First, sensitivity analyses were performed to assess robustness of the results. The price of ezetimibe was varied from the current prices of $0.203/d (US) and £0.074/d (UK) to the price of atorvastatin 40 mg of $0.103/d (US) and £0.034/d (UK). The price at which ezetimibe becomes cost-effective for the commonly used thresholds of $100,000 (US) and £20,000 (UK) per QALY was calculated. Second, to estimate the likely effect of dialysis costs on cost-effectiveness results, the analyses were repeated with RRT costs replaced with those for CKD stage 5 not on dialysis. Third, an analysis incorporating potential rare adverse effects of atorvastatin 40 mg alone or in combination with ezetimibe 10 mg daily was performed. Specifically, we assumed that during each year in patients taking atorvastatin 40 mg daily (with or without ezetimibe), 0.011% will experience myopathy at a cost of $33 (£19; derived as the cost of 3 creatine kinase tests) and 0.001 QoL decrement (i.e., 0.017 decrement over 30 days), and 0.0042% will experience rhabdomyolysis at a cost of $13,600 (£8000), of whom 10% will die, with the rest experiencing a 3% QoL decrement (i.e., 50% decrement over 7.5 days of hospitalization followed by 20% decrement over 30 days of recovering) in the year of the rhabdomyolysis,23, 39 and 0.2% will develop diabetes.40 Finally, the effect of nonadherence to treatment was explored in scenarios where, respectively, 40%, 60%, and 80% of the patients were taking the medication.

over 7.5 days of hospitalization followed by 20% decrement over 30 days of recovering) in the year of the rhabdomyolysis,23, 39 and 0.2% will develop diabetes.40 Finally, the effect of nonadherence to treatment was explored in scenarios where, respectively, 40%, 60%, and 80% of the patients were taking the medication. All analyses were performed with R 3.4.141; the graphs were produced with the ggplot2 plotting system.42 Disclosure The SHARP study, including the analyses presented here, was funded by Merck & Co., Inc., Kenilworth, NJ USA, with additional support from the British Heart Foundation (CH/1996001/9454), and the UK Medical Research Council (A310). SHARP was initiated, conducted, and interpreted independently of the principal study funder (Merck & Co.). WH is supported by a Medical Research Council and Kidney Research UK Professor David Kerr Clinician Scientist Award. BM and MJL are supported by the National Institute for Health Research (NIHR) Oxford Biomedical Research Centre (BRC). The study funders/sponsors did not have any role in study design; collection, analysis, and interpretation of data; writing the report; and the decision to submit the report for publication. The Clinical Trial Service Unit of the University of Oxford (Oxford, UK) has a staff policy of not accepting honoraria or other payments from the pharmaceutical industry, except for the reimbursement of costs to participate in scientific meetings. WH, JE, RH, RC, MJL, CB, and BM report other grants for unrelated work.

The Clinical Trial Service Unit of the University of Oxford (Oxford, UK) has a staff policy of not accepting honoraria or other payments from the pharmaceutical industry, except for the reimbursement of costs to participate in scientific meetings. WH, JE, RH, RC, MJL, CB, and BM report other grants for unrelated work. Appendix The SHARP Collaborative Group steering committee R. Collins (chair), C. Baigent (study coordinator and chief investigator), M.J. Landray (clinical coordinator and co-principal investigator), C. Bray, Y. Chen (administrative coordinators), A. Baxter, A. Young (computing coordinators), M. Hill (director, central laboratory), C. Knott (nursing coordinator), A. Cass, B. Feldt-Rasmussen, B. Fellström, D.E. Grobbee, C. Grönhagen-Riska, M. Haas, H. Holdaas, L.S. Hooi, L. Jiang, B. Kasiske, U. Krairittichai, A. Levin, Z.A. Massy, V. Tesar, R. Walker, C. Wanner, D.C. Wheeler, A. Wiecek (national coordinators), T. Dasgupta, W. Herrington, D. Lewis, M. Mafham, W. Majoni, C. Reith (clinical support), J. Emberson, S. Parish (statistics), D. Simpson (lay member), J. Strony, T. Musliner (Merck Schering Plough, nonvoting), L. Agodoa, J. Armitage, Z. Chen, J. Craig, D. de Zeeuw, J.M. Gaziano, R. Grimm, V. Krane, B. Neal, V. Ophascharoensuk, T. Pedersen, P. Sleight, J. Tobert, and C. Tomson. The full list of the SHARP investigators is available elsewhere.6

D. Simpson (lay member), J. Strony, T. Musliner (Merck Schering Plough, nonvoting), L. Agodoa, J. Armitage, Z. Chen, J. Craig, D. de Zeeuw, J.M. Gaziano, R. Grimm, V. Krane, B. Neal, V. Ophascharoensuk, T. Pedersen, P. Sleight, J. Tobert, and C. Tomson. The full list of the SHARP investigators is available elsewhere.6 The procedure for access requests to the SHARP (Trial registration: NCT00125593) data is available at http://www.ndph.ox.ac.uk/about/data-access-policy. The SHARP CKD-CVD model interface and user guide are available at http://dismod.ndph.ox.ac.uk/kidneymodel/app/. Supplementary Material Table S1 Characteristics of Study of Heart and Renal Protection (SHARP) nondialysis participants by chronic kidney disease (CKD) stage and cardiovascular disease risk at baseline. Table S2 Reductions in low-density lipoprotein (LDL) cholesterol with statin-based treatments and daily drug treatment cost. Table S3 Average low-density lipoprotein (LDL) cholesterol, relative risk with statin-based treatment per 1-mmol/L reduction in LDL cholesterol, and relative risk with specific statin-based treatments, by chronic kidney disease (CKD) stage. Table S4 Health outcomes, United States (US) hospital care costs, and US additional cost per quality-adjusted life year with lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced nondialysis chronic kidney disease (CKD). Table S5 Cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in categories of moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, by age at treatment initiation.

Table S4 Health outcomes, United States (US) hospital care costs, and US additional cost per quality-adjusted life year with lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced nondialysis chronic kidney disease (CKD). Table S5 Cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in categories of moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, by age at treatment initiation. Table S6 Added (quality-adjusted) life years, extra hospital care costs, and additional cost per quality-adjusted life year with lifetime use of ezetimibe 10 mg daily compared to no low-density lipoprotein cholesterol (LDL-C)-lowering treatment in moderate-to-advanced nondialysis chronic kidney disease (CKD). Table S7 Health outcomes, United Kingdom (UK) hospital care costs, and UK additional cost per quality-adjusted life year with lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced nondialysis chronic kidney disease (CKD). Table S8 Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients with future renal replacement therapy costs excluded. Table S9 Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients, incorporating potential adverse effects on myopathy, rhabdomyolysis, and diabetes.

Table S8 Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients with future renal replacement therapy costs excluded. Table S9 Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients, incorporating potential adverse effects on myopathy, rhabdomyolysis, and diabetes. Table S10 Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients under different scenarios of compliance with treatment. Table S11 Annual rates of nonvascular death in moderate-to-advanced chronic kidney disease (CKD) patients (United States). Table S12 United States (US) annual hospital care costs in moderate-to-advanced chronic kidney disease (CKD). Table S13 Health-related quality of life in moderate-to-advanced chronic kidney disease (CKD): a linear regression model derived from Study of Heart and Renal Protection (SHARP) participant data using United States (US) EQ-5D value set. Figure S1 Additional quality-adjusted life-years (QALYs), hospital care costs, and cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in categories of moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, by CKD stage and cardiovascular risk at baseline.

Table S13 Health-related quality of life in moderate-to-advanced chronic kidney disease (CKD): a linear regression model derived from Study of Heart and Renal Protection (SHARP) participant data using United States (US) EQ-5D value set. Figure S1 Additional quality-adjusted life-years (QALYs), hospital care costs, and cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in categories of moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, by CKD stage and cardiovascular risk at baseline. Figure S2 Schematic of the Study of Heart and Renal Protection (SHARP) CKD-CVD lifetimes health outcomes model. Figure S3 Schematic of the information sources of the Study of Heart and Renal Protection (SHARP) CKD-CVD lifetime health outcomes model. Acknowledgements We thank the SHARP participants and the local clinical center staff, regional and national coordinators, steering committee, and data monitoring committee. Author Contributions Research concept and study design: BM, IS, WH, JE, RH, AG, CB; data acquisition: SK, WH, JE, CR, RC, ML, CB; data analysis/interpretation: BM, IS, SK, WH, JE, RH, RC, AG, ML, CB; statistical analysis: IS, BM; supervision: BM, CB. Each author contributed intellectual content during manuscript drafting/revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved. BM and IS had full access to all study data and take responsibility for data integrity and data analysis accuracy.

drafting/revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved. BM and IS had full access to all study data and take responsibility for data integrity and data analysis accuracy. Table S1. Characteristics of Study of Heart and Renal Protection (SHARP) nondialysis participants by chronic kidney disease (CKD) stage and cardiovascular disease risk at baseline. Table S2. Reductions in low-density lipoprotein (LDL) cholesterol with statin-based treatments and daily drug treatment cost. Table S3. Average low-density lipoprotein (LDL) cholesterol, relative risk with statin-based treatment per 1-mmol/L reduction in LDL cholesterol, and relative risk with specific statin-based treatments, by chronic kidney disease (CKD) stage. Table S4. Health outcomes, United States (US) hospital care costs, and US additional cost per quality-adjusted life year with lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced nondialysis chronic kidney disease (CKD). Table S5. Cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in categories of moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, by age at treatment initiation.

Table S4. Health outcomes, United States (US) hospital care costs, and US additional cost per quality-adjusted life year with lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced nondialysis chronic kidney disease (CKD). Table S5. Cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in categories of moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, by age at treatment initiation. Table S6. Added (quality-adjusted) life years, extra hospital care costs, and additional cost per quality-adjusted life year with lifetime use of ezetimibe 10 mg daily compared to no low-density lipoprotein cholesterol (LDL-C)-lowering treatment in moderate-to-advanced nondialysis chronic kidney disease (CKD). Table S7. Health outcomes, United Kingdom (UK) hospital care costs, and UK additional cost per quality-adjusted life year with lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced nondialysis chronic kidney disease (CKD). Table S8. Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients with future renal replacement therapy costs excluded. Table S9. Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients, incorporating potential adverse effects on myopathy, rhabdomyolysis, and diabetes.

Table S8. Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients with future renal replacement therapy costs excluded. Table S9. Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients, incorporating potential adverse effects on myopathy, rhabdomyolysis, and diabetes. Table S10. Sensitivity analysis of cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in moderate-to-advanced non-dialysis chronic kidney disease (CKD) patients under different scenarios of compliance with treatment. Table S11. Annual rates of nonvascular death in moderate-to-advanced chronic kidney disease (CKD) patients (United States). Table S12. United States (US) annual hospital care costs in moderate-to-advanced chronic kidney disease (CKD). Table S13. Health-related quality of life in moderate-to-advanced chronic kidney disease (CKD): a linear regression model derived from Study of Heart and Renal Protection (SHARP) participant data using United States (US) EQ-5D value set. Figure S1. Additional quality-adjusted life-years (QALYs), hospital care costs, and cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in categories of moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, by CKD stage and cardiovascular risk at baseline.

Table S13. Health-related quality of life in moderate-to-advanced chronic kidney disease (CKD): a linear regression model derived from Study of Heart and Renal Protection (SHARP) participant data using United States (US) EQ-5D value set. Figure S1. Additional quality-adjusted life-years (QALYs), hospital care costs, and cost-effectiveness of lifetime use of atorvastatin and ezetimibe treatments in categories of moderate-to-advanced nondialysis chronic kidney disease (CKD) patients, by CKD stage and cardiovascular risk at baseline. Figure S2. Schematic of the Study of Heart and Renal Protection (SHARP) CKD-CVD lifetimes health outcomes model. Figure S3. Schematic of the information sources of the Study of Heart and Renal Protection (SHARP) CKD-CVD lifetime health outcomes model. Supplementary material is linked to the online version of the paper at www.kidney-international.org.

Low birth weight, caused either by preterm birth or by intrauterine growth restriction (IUGR), has recently been shown to be associated with increased rates of renal and cardiovascular disease in adult life.1, 2, 3 The fetal origins hypothesis proposes that these diseases originate through metabolic or endocrine adaptations when the fetus is undernourished and result in permanent changes of the structure and function of the body.4, 5

n to be associated with increased rates of renal and cardiovascular disease in adult life.1, 2, 3 The fetal origins hypothesis proposes that these diseases originate through metabolic or endocrine adaptations when the fetus is undernourished and result in permanent changes of the structure and function of the body.4, 5 Early endothelial dysfunction, impaired arterial vasodilation, and aortic wall intima–media thickening (aIMT) occurring in utero may have an important role in premature in-utero stiffening of the aortic vessels and may predispose these individuals to hypertension, nephropathies, and the metabolic syndrome.6 Moreover, the developing kidneys appear to be extremely susceptible to IUGR and are often found to be small in proportion to body weight.7, 8 Several studies in animals and humans have described a reduced number of nephrons after IUGR.9, 10, 11 The reduced number of nephrons results in a decreased glomerular filtration surface area, while renal blood flow per glomerulus is increased in an attempt to maintain a normal overall glomerular filtration rate. According to Brenner's hyperfiltration hypothesis, this leads to glomerular hypertension and hypertrophy, which ultimately results in glomerulosclerosis and albuminuria.12, 13, 14 These findings clearly indicate that IUGR may have an adverse affect on the integrity and function of blood vessels and on glomerulogenesis during fetal life. Whether the finding of fetal aortic wall thickening predicts glomerular damage and microalbuminuria during infancy is not known.

rosis and albuminuria.12, 13, 14 These findings clearly indicate that IUGR may have an adverse affect on the integrity and function of blood vessels and on glomerulogenesis during fetal life. Whether the finding of fetal aortic wall thickening predicts glomerular damage and microalbuminuria during infancy is not known. During a recent longitudinal prospective study to evaluate aortic wall thickness in intrauterine growth-restricted fetuses and infants,15 we planned additional analyses to determine if aortic thickening was predictive of precocious renal damage during infancy. Therefore, the aims of the present study were to (1) compare abdominal aortic intima–media thickness (aIMT) among IUGR and appropriate for gestational age (AGA) fetuses in utero and at 18 months of age, and (2) to assess the relationship between IUGR, fetal aortic thickening, and glomerular function during infancy. RESULTS Fifty fetuses initially met our inclusion criteria for the study, but only data on 44 subjects (n=23 IUGR, n=21 AGA) were included in our final statistical analysis. Six subjects were excluded (two IUGR and four AGA) because they did not provide urine samples. Anthropometric and clinical characteristics of the study population are shown in Table 1.

r inclusion criteria for the study, but only data on 44 subjects (n=23 IUGR, n=21 AGA) were included in our final statistical analysis. Six subjects were excluded (two IUGR and four AGA) because they did not provide urine samples. Anthropometric and clinical characteristics of the study population are shown in Table 1. The median estimated fetal weight (EFW) at the time of the initial aIMT measurement was 1750 g (<5th percentile) in the IUGR group, and 2200 g (50th percentile) in the AGA group.16, 17 The median maternal age at the time of delivery was 30.5 years in the IUGR group and 31.2 years in the AGA group. Nine (39%) women in the IUGR group had a vaginal delivery, while 14 (61%) underwent cesarean section. In the AGA group, there were 14 (77%) vaginal deliveries and 7 (33%) cesarean sections. The median birth weight of the 23 children born IUGR was 1850 g (<5th percentile) at a median gestational age of 33 weeks. Neonates that were AGA had a median gestational duration of 38 weeks, with a median birth weight of 2975 g (50th percentile). In addition, no gender disparity was encountered on comparing IUGR and AGA children. The median fetal aIMT was significantly different between IUGR and AGA fetuses (IUGR 2.0 mm vs AGA 1.05 mm; P<0.001).

that were AGA had a median gestational duration of 38 weeks, with a median birth weight of 2975 g (50th percentile). In addition, no gender disparity was encountered on comparing IUGR and AGA children. The median fetal aIMT was significantly different between IUGR and AGA fetuses (IUGR 2.0 mm vs AGA 1.05 mm; P<0.001). At the 18-month follow-up evaluation, there was no statistically significant difference in median weight in the IUGR and AGA groups (12.3 vs 12.5 kg, P=0.23). The aIMT, however, remained significantly larger among infants in the IUGR group compared with those in the AGA group (2.1 vs 1.05 mm; P<0.001). IUGR infants had a higher systolic blood pressure (SBP) compared with AGA infants (P<0.001). In addition, median urinary microalbumin (11.1 vs 4.4 mg/l; P=0.001) and the albumin/creatinine ratio (A/CR) (26 vs 14.6 mg/l; P=0.001) (Figure 1) were higher in the IUGR infants compared with AGA infants. Univariate analysis defined the following variables as being related to aIMT in fetuses: IUGR (P<0.001) and estimated fetal weight (P<0.001), and the following in infants: IUGR (P<0.001), microalbuminuria (P<0.001), systolic blood pressure (P<0.001), and albumin/creatinine ratio (P<0.001) (Table 2).

At the 18-month follow-up evaluation, there was no statistically significant difference in median weight in the IUGR and AGA groups (12.3 vs 12.5 kg, P=0.23). The aIMT, however, remained significantly larger among infants in the IUGR group compared with those in the AGA group (2.1 vs 1.05 mm; P<0.001). IUGR infants had a higher systolic blood pressure (SBP) compared with AGA infants (P<0.001). In addition, median urinary microalbumin (11.1 vs 4.4 mg/l; P=0.001) and the albumin/creatinine ratio (A/CR) (26 vs 14.6 mg/l; P=0.001) (Figure 1) were higher in the IUGR infants compared with AGA infants. Univariate analysis defined the following variables as being related to aIMT in fetuses: IUGR (P<0.001) and estimated fetal weight (P<0.001), and the following in infants: IUGR (P<0.001), microalbuminuria (P<0.001), systolic blood pressure (P<0.001), and albumin/creatinine ratio (P<0.001) (Table 2). In addition, a positive association was observed between the prenatal aIMT values and 18-month postnatal aIMT (regression coefficient 0.86, standard error 0.24; P=0.001 and regression coefficient 0.46, standard error 0.22; P=0.04, respectively) within the children born with IUGR, a phenomenon not observed among AGA children. Among IUGR infants, SBP was also positively associated with the post-natal aIMT (regression coefficient 0.02, standard error 0.009; P=0.04) (Table 3). DISCUSSION The present study highlights that aortic wall thickness and microalbuminuria are significantly higher in IUGR compared with AGA infants at a mean age of 18 months.

In addition, a positive association was observed between the prenatal aIMT values and 18-month postnatal aIMT (regression coefficient 0.86, standard error 0.24; P=0.001 and regression coefficient 0.46, standard error 0.22; P=0.04, respectively) within the children born with IUGR, a phenomenon not observed among AGA children. Among IUGR infants, SBP was also positively associated with the post-natal aIMT (regression coefficient 0.02, standard error 0.009; P=0.04) (Table 3). DISCUSSION The present study highlights that aortic wall thickness and microalbuminuria are significantly higher in IUGR compared with AGA infants at a mean age of 18 months. Although the exact mechanisms that underlie these associations remain unclear, these data, indicating early glomerular damage, extend the findings of our previous study on the natural course of aIMT among IUGR fetuses15 and other studies in intrauterine fetal growth-restricted children, adolescents, and young adults at risk for premature cardiovascular and kidney diseases.11, 12, 18, 19, 20, 21

, these data, indicating early glomerular damage, extend the findings of our previous study on the natural course of aIMT among IUGR fetuses15 and other studies in intrauterine fetal growth-restricted children, adolescents, and young adults at risk for premature cardiovascular and kidney diseases.11, 12, 18, 19, 20, 21 Aortic intima–media thickening is the best currently available non-invasive marker of preclinical atherosclerosis. Evidence from non-invasive ultrasound studies of the neonatal aorta, combined with fetal and early childhood postmortem studies, indicates that impaired fetal growth, in utero exposure to maternal hypercholesterolemia, and diabetic macrosomia might all be important risk factors for vascular changes consistent with the earliest physical signs of atherosclerosis.21, 22, 23 These changes first develop in the intima of the aorta by both native and oxidized LDL and macrophage infiltration.24 IUGR has also been associated with oligonephropathy. According to the hyperfiltration hypothesis, the decreased glomerular filtration surface leads to glomerular hypertension and hypertrophy, which causes systemic hypertension and glomerular damage, and ultimately results in glomerulosclerosis and albuminuria later in life.12, 13, 14 However, whether fetal intima–media thickening and early endothelial dysfunction among IUGR infants are significant contributors to later atherosclerosis and glomerulosclerosis is not yet known.25, 26

pertension and glomerular damage, and ultimately results in glomerulosclerosis and albuminuria later in life.12, 13, 14 However, whether fetal intima–media thickening and early endothelial dysfunction among IUGR infants are significant contributors to later atherosclerosis and glomerulosclerosis is not yet known.25, 26 It is now possible to measure aortic wall thickness accurately and reproducibly in vivo during fetal and postnatal life with external ultrasonography. Ultrasound-based measurement of aIMT in IUGR children was found to be a sensitive marker of hypertension in young IUGR children and of atherosclerosis risk in adult life,26, 27 supporting the epidemiological link between impaired fetal growth and later cardiovascular disease. In addition, Singh and Hoy28 described an association between low birth weight, kidney size, and albuminuria in aboriginal subjects between 4 to 72 years of age. Recently, Keijzer-Veen's18 prospective follow-up study showed an association between the severity of IUGR and renal function in young adults born very prematurely. On average, their low birth weight subjects had lower glomerular filtration rates (GFRs), higher serum creatinine concentrations, and higher microalbumin secretion at the age of 19 years. More recently, Puddu et al.29 reported high levels of microalbuminuria in a group of very low birth weight infants.

ery prematurely. On average, their low birth weight subjects had lower glomerular filtration rates (GFRs), higher serum creatinine concentrations, and higher microalbumin secretion at the age of 19 years. More recently, Puddu et al.29 reported high levels of microalbuminuria in a group of very low birth weight infants. Microalbuminuria is one of the first symptoms of developing renal disease and precedes a decrease in GFR. The higher risk of aortic intima–media thickening and microalbuminuria may have clinical implications for IUGR infants at 18 months of age. Early endothelial dysfunction and intima–media thickening could be significant contributors to premature stiffening of the arterial tree, which ultimately might predispose these individuals to systemic hypertension and increased cardiovascular risk.22, 23, 24, 27, 28

l implications for IUGR infants at 18 months of age. Early endothelial dysfunction and intima–media thickening could be significant contributors to premature stiffening of the arterial tree, which ultimately might predispose these individuals to systemic hypertension and increased cardiovascular risk.22, 23, 24, 27, 28 Our results are in agreement with previous studies that correlate low birth weight with endothelial damage that may influence arterial function and the overall incidence of cardiovascular events.6, 9, 22 Unlike these studies, which focused on aortic wall thickness in term IUGR infants and in high-risk children and young adults, we evaluated the natural course of this marker of endothelial dysfunction in fetuses with IUGR and severe Doppler abnormalities, both in utero and at 18 months of age. These results suggest that, in addition to other pathogenic mechanisms, higher arterial thickness is already present in IUGR fetuses during intrauterine life and this, together with factors such as impaired nephrogenesis and glomerulosclerosis, could have a role in programming adult disease, as previously suggested.12, 13, 14, 24, 26

uggest that, in addition to other pathogenic mechanisms, higher arterial thickness is already present in IUGR fetuses during intrauterine life and this, together with factors such as impaired nephrogenesis and glomerulosclerosis, could have a role in programming adult disease, as previously suggested.12, 13, 14, 24, 26 There are limitations to our study that must be considered in the interpretation of these results. First of all, the markedly greater range in ultrasound aIMT measurements and the wide pattern of Doppler abnormalities of IUGR fetuses limited the possibility of making comparisons with other studies, performed with different equipment (7–12 MHz probe), in both term growth-restricted neonates and in high-risk children and young adults. Second, the follow-up is limited to the second year of life. Third, we are unable to establish whether renal function will decrease more rapidly with age compared with aortic wall thickening. It is also not known whether screening for microalbuminuria and aIMT will help to prevent the loss of renal function, atherosclerosis, and hypertension at an early stage. In conclusion, these findings indicate that IUGR children have a higher risk of both aortic wall thickening and glomerular proteinuria, which may contribute to cardiovascular and renal disease in later life. Follow-up studies are needed to confirm the prognostic role of these markers of atherosclerosis and renal disease.

In conclusion, these findings indicate that IUGR children have a higher risk of both aortic wall thickening and glomerular proteinuria, which may contribute to cardiovascular and renal disease in later life. Follow-up studies are needed to confirm the prognostic role of these markers of atherosclerosis and renal disease. PATIENTS AND METHODS Subjects were recruited from the Obstetrics and Gynaecology clinics at the University Hospital of Padua (Italy) between January 2006 and August 2008. Written informed consent was obtained from each woman before enrolment and the project was approved by the University Hospital Committee for Research on Human Subjects. Some of the subjects had been recruited as part of a separate multicenter study.16 Data concerning women and their pregnancies were recorded according to the routine practice of the Department of Obstetrics and Gynaecology. Inclusion criteria at admission were single pregnancy, gestational age determined from known last maternal menstrual period and/or ultrasound dating before 20 weeks of gestation, and women originating from the Veneto region (Italy). Exclusion criteria were twin pregnancy, major congenital anomalies, pregnancies complicated by maternal history of cardiovascular disease or endocrine disorders such as diabetes, hypercholesterolemia, pre-eclampsia, thyroid or adrenal problems, and clinical chorioamnionitis. Women who received alcohol, nicotine, or medications such as ritodrine and corticosteroids (except for fetal lung maturation) were excluded.

aternal history of cardiovascular disease or endocrine disorders such as diabetes, hypercholesterolemia, pre-eclampsia, thyroid or adrenal problems, and clinical chorioamnionitis. Women who received alcohol, nicotine, or medications such as ritodrine and corticosteroids (except for fetal lung maturation) were excluded. Fetuses were classified as IUGR if the estimated fetal weight (EFW) was <10th percentile and the umbilical artery pulsatility index (PI) >2 standard deviations (s.d.) above the mean, and AGA if the EFW was between 10th and 90th percentiles.16, 17 All fetuses had at least three ultrasound and Doppler examinations during pregnancy. In each enrolled IUGR and AGA fetus, EFW and antenatal testing were available, despite no indications or guidelines recommending the use of Doppler ultrasonography in uncomplicated pregnancies. Fetal aIMT was measured in each IUGR and AGA subject at a median gestational age of 32 weeks (interquartile range (IQR) 30–34 weeks) by high-resolution ultrasound scan (Antares, Siemens Medical Solutions, Mountain View, CA) using a 3.5–5 MHz linear array transducer, as previously reported.15 Fetal examination required no more than 20 min, with about 40 min required for the children. A single, skilled practitioner performed all ultrasound studies, in both fetuses and children, using the same equipment, unaware of their clinical course and outcomes. The follow-up examination was performed at a median corrected postnatal age of 18 months (IQR 15–21 months) and included aIMT, growth parameters, blood pressure, and a urine sample.

Fetal aIMT was measured in each IUGR and AGA subject at a median gestational age of 32 weeks (interquartile range (IQR) 30–34 weeks) by high-resolution ultrasound scan (Antares, Siemens Medical Solutions, Mountain View, CA) using a 3.5–5 MHz linear array transducer, as previously reported.15 Fetal examination required no more than 20 min, with about 40 min required for the children. A single, skilled practitioner performed all ultrasound studies, in both fetuses and children, using the same equipment, unaware of their clinical course and outcomes. The follow-up examination was performed at a median corrected postnatal age of 18 months (IQR 15–21 months) and included aIMT, growth parameters, blood pressure, and a urine sample. Blood pressure measurements were performed using a standard Doppler sphygmomanometer (Philip Medical System Monitor, Agilent, M3046A model, M4, Boebingen, Germany) using a cuff size appropriate for the subject's right arm circumference. Three independent measurements for each child were taken and the arithmetic mean was used for the study, as recommended by the current guidelines.30 Urine microalbumin was assayed using an immunonephelometric method (BN™II, Siemens Medical Solutions). Urine creatinine was tested by a colorimetric Jaffè method (Roche Diagnostics, Monza, Italy).

Blood pressure measurements were performed using a standard Doppler sphygmomanometer (Philip Medical System Monitor, Agilent, M3046A model, M4, Boebingen, Germany) using a cuff size appropriate for the subject's right arm circumference. Three independent measurements for each child were taken and the arithmetic mean was used for the study, as recommended by the current guidelines.30 Urine microalbumin was assayed using an immunonephelometric method (BN™II, Siemens Medical Solutions). Urine creatinine was tested by a colorimetric Jaffè method (Roche Diagnostics, Monza, Italy). Data are presented as median (IQR). Differences between groups were tested using the Fisher's exact test for categorical variables and the Mann–Whitney U-test for continuous variables. As previously reported, intra-observer and inter-observer aIMT correlation coefficients were 0.876 and 0.856, respectively.15 The association between aIMT and continuous variables was tested with Spearman's rank correlation. Multivariate analysis by median regression was performed to identify the independent effect of IUGR/AGA on fetal and infant aIMT after controlling for potential confounders (sex, gestational and corrected postnatal age, estimated fetal weight, body weight, systolic and diastolic blood pressure, microalbuminuria, and A/CR). A P<0.05 was considered statistically significant. Statistical analysis was performed using the SPSS 17 software package (SPSS, Chicago, IL) and R 2.5 statistical language. All the authors declared no competing interests.

Data are presented as median (IQR). Differences between groups were tested using the Fisher's exact test for categorical variables and the Mann–Whitney U-test for continuous variables. As previously reported, intra-observer and inter-observer aIMT correlation coefficients were 0.876 and 0.856, respectively.15 The association between aIMT and continuous variables was tested with Spearman's rank correlation. Multivariate analysis by median regression was performed to identify the independent effect of IUGR/AGA on fetal and infant aIMT after controlling for potential confounders (sex, gestational and corrected postnatal age, estimated fetal weight, body weight, systolic and diastolic blood pressure, microalbuminuria, and A/CR). A P<0.05 was considered statistically significant. Statistical analysis was performed using the SPSS 17 software package (SPSS, Chicago, IL) and R 2.5 statistical language. All the authors declared no competing interests. Figure 1 Urine microalbumin in intrauterine growth restricted (IUGR) and appropriate for gestational age (AGA) infants. Median (interquartile range). Urinary microalbumin was significantly higher in the IUGR infants compared with the AGA group (11.1 vs 4.4 mg/l; P<0.001). Table 1 Anthropometric, sonographic, and clinical measurements among IUGR and AGA fetuses and infants IUGR (n=23) AGA (n=21) P Prenatal measurements Gestational age (weeks) 32.1 (29.9–33.7) 32 (30–34) 0.99 Estimated fetal weight (g) 1750 (1450–2050) 2200 (1930–2470) <0.001 Fetal aIMT (mm) 2.00 (1.78–2.23) 1.05 (0.95–1.15) 0.001

nthropometric, sonographic, and clinical measurements among IUGR and AGA fetuses and infants IUGR (n=23) AGA (n=21) P Prenatal measurements Gestational age (weeks) 32.1 (29.9–33.7) 32 (30–34) 0.99 Estimated fetal weight (g) 1750 (1450–2050) 2200 (1930–2470) <0.001 Fetal aIMT (mm) 2.00 (1.78–2.23) 1.05 (0.95–1.15) 0.001 Neonatal measurements Gestational age (weeks) 33 (31–36) 38 (35–41) 0.001 Birth weight (g) 1850 (1850–2200) 2975 (2700–3250) 0.001 Gender: male 12 (52.1) 10 (47.6) 0.87 Postnatal measurements Corrected postnatal age (months) 18 (12.8–25.2) 19 (13.3–26.7) 0.48 Body weight (kg) 12.3 (11.2–13.4) 12.5 (11.3–14) 0.23 Length (cm) 87 (80–90) 86,5 (79–92) 0.34 Infant aIMT (mm) 2.1 (1.4–3.0) 1.05 (0.95–1.25) <0.001 Systolic BP (mm Hg) 123 (107–139) 103 (95.5–112.5) <0.001 Diastolic BP (mm Hg) 65 (57.6–72.4) 64 (59–71) 0.99 Urine microalbumin (mg/l) 11.1 (1.9–19.5) 4.4 (0.0–8.8) 0.001 A/CR (mg/g) 26 (9.8–41.4) 14.6 (8.2–21.2) 0.001 Abbreviations: A/CR, albumin/creatinine ratio; AGA, appropriate for gestational age; aIMT, aortic intima–media thickness; BP, blood pressure; IUGR, intrauterine growth restriction. Values are shown as number (%) or as median (IQR). Table 2 Aortic intima–media thickness and associated anthropometric and clinical characteristics of fetuses and infants (univariate analysis) aIMT: median (IQR) P-value Fetuses Group   <0.001 AGA 1.05 (0.95–1.15) IUGR 2.00 (1.78–2.23) Sex   0.71 F 1.63 (1.09–2.06) M 1.35 (1.04–2.00) Gestational age 0.07a 0.65 EFW −0.72a <0.001

Aortic intima–media thickness and associated anthropometric and clinical characteristics of fetuses and infants (univariate analysis) aIMT: median (IQR) P-value Fetuses Group   <0.001 AGA 1.05 (0.95–1.15) IUGR 2.00 (1.78–2.23) Sex   0.71 F 1.63 (1.09–2.06) M 1.35 (1.04–2.00) Gestational age 0.07a 0.65 EFW −0.72a <0.001 Infants Group   <0.001 AGA 1.05 (1.00–1.10) IUGR 2.10 (1.65–2.55) Sex   0.50 F 1.50 (1.10–2.13) M 1.34 (1.00–2.03) Corrected postnatal age −0.11a 0.46 Weight −0.02a 0.89 Microalbuminuria 0.53a 0.001 SBP 0.81a <0.001 DBP 0.09a 0.58 A/CR 0.65a <0.001 Abbreviations: A/CR, albumin/creatinine ratio; AGA, appropriate for gestational age; aIMT, aortic intima–media thickness; DBP, diastolic blood pressure; EFW, estimated fetal weight; F, female; IQR, interquartile range; IUGR, intrauterine growth restriction; M, male; SBP, systolic blood pressure. a Spearman's rank correlation. Table 3 Aortic intima–media thickness and associated anthropometric and clinical characteristics of fetuses and infants (multivariate analysis) aIMT Regression coefficient Standard error P-value Fetuses Group: IUGR 0.86 0.24 0.001 Sex: male −0.003 0.04 0.95 Gestational age 0.006 0.08 0.94 EFW −0.001 0.0004 0.49

a Spearman's rank correlation. Table 3 Aortic intima–media thickness and associated anthropometric and clinical characteristics of fetuses and infants (multivariate analysis) aIMT Regression coefficient Standard error P-value Fetuses Group: IUGR 0.86 0.24 0.001 Sex: male −0.003 0.04 0.95 Gestational age 0.006 0.08 0.94 EFW −0.001 0.0004 0.49 Infants Group: IUGR 0.46 0.22 0.04 Sex: male −0.02 0.11 0.86 Corrected postnatal age 0.005 0.004 0.24 Weight 0 0.0001 0.99 Microalbuminuria 0.03 0.02 0.24 SBP 0.02 0.009 0.04 DBP 0.01 0.01 0.33 A/CR 0.009 0.01 0.35 Abbreviations: A/CR, albumin/creatinine ratio; aIMT, aortic intima–media thickness; DBP, diastolic blood pressure; EFW, estimated fetal weight; IUGR, intrauterine growth restriction; SBP, systolic blood pressure.

Crescentic glomerulonephritis is mediated by inappropriate humoral and cellular immune responses toward not well-characterized (self) antigens that might result from defects in central and peripheral tolerance. CD4+CD25+FoxP3+ regulatory T cells (Tregs) are generally thought to be of critical importance for the maintenance of peripheral tolerance.1 Until now, little is known about the suppressive mechanisms mediated by CD4+CD25+FoxP3+ Tregs and their behavior during proliferative and crescentic glomerulonephritis.2, 3 With respect to human disease, one report indicates that CD25+ T cells emerge following acute episodes of Goodpasture's disease suppressing the response to the Goodpasture autoantigen.4 Furthermore, the potential of Tregs to suppress the pathogenic immune responses in kidney disease has recently been proven in a model of crescentic glomerulonephritis in mice. The adoptive transfer of CD4+CD25+ Tregs showed that Tregs downregulate the nephritogenic immune response directly at the systemic site of antigen-specific T-cell priming, namely the renal lymph node.5 In consecutive studies, Eller et al.6 demonstrated that the chemokine receptor CCR7 (chemokine (C-C motif) receptor 7) regulates the trafficking of Tregs to the lymph node in experimental glomerulonephritis, which is essential for their anti-inflammatory properties. Our group recently provided evidence that Treg cells migrate in a CCR6–CCL20-dependent manner into the inflamed kidney in crescentic glomerulonephritis and might locally suppress immune-mediated renal inflammation.7

ymph node in experimental glomerulonephritis, which is essential for their anti-inflammatory properties. Our group recently provided evidence that Treg cells migrate in a CCR6–CCL20-dependent manner into the inflamed kidney in crescentic glomerulonephritis and might locally suppress immune-mediated renal inflammation.7 This study was designed to examine the potential role of endogenous Tregs in crescentic glomerulonephritis. We therefore induced the T cell-dependent model of nephrotoxic nephritis (NTN) in ‘depletion of regulatory T cell' (DEREG) mice that express the diphtheria toxin (DTx) receptor and green fluorescent protein (GFP) under control of the FoxP3 (forkhead box P3) promoter, allowing selective and efficient depletion of FoxP3+ Treg cells by DTx injection.8 Hence, we intended to address whether Treg depletion influences the clinical course of experimental glomerulonephritis, and if so, to study their potential mechanisms of action. RESULTS Functional involvement of endogenous Tregs in NTN To investigate the role of endogenous Tregs, we determined the frequency of CD4+CD25+FoxP3+ Tregs upon induction of NTN in a time course experiment. Interestingly, the number of Tregs was increased from day 5 onward, as shown by immunohistological FoxP3 staining (Figure 1a and b).

involvement of endogenous Tregs in NTN To investigate the role of endogenous Tregs, we determined the frequency of CD4+CD25+FoxP3+ Tregs upon induction of NTN in a time course experiment. Interestingly, the number of Tregs was increased from day 5 onward, as shown by immunohistological FoxP3 staining (Figure 1a and b). To evaluate the immunosuppressive capacity of Tregs from nephritic mice compared with their counterparts from healthy control mice, we isolated splenic CD4+CD25+ Tregs from both nephritic and healthy control mice and performed T-cell co-culture experiments. Enzyme-linked immunosorbent assay (ELISA) analysis of supernatants from co-cultures of Tregs and CD4+CD25− responder T cells indicated that Tregs from nephritic mice maintained their in vitro suppressive capacity and might be even more potent than Tregs from healthy controls to inhibit the production of interleukin (IL)-2 and interferon-γ (IFNγ) by responder T cells (Figure 1c, IL-2 (pg/ml): single culture of responder T cells: 334.3±22.0; responder T cells+Tregs (control): 79.3±10.5; responder T cells+Tregs (NTN): 43.7±13.5; ***P<0.0001; IFNγ (pg/ml): single culture of responder T cells: 8490±772, responder T cells+Tregs (control): 6084±2949, responder T cells+Tregs (NTN): 4137±845.2; *P<0.05).

2 (pg/ml): single culture of responder T cells: 334.3±22.0; responder T cells+Tregs (control): 79.3±10.5; responder T cells+Tregs (NTN): 43.7±13.5; ***P<0.0001; IFNγ (pg/ml): single culture of responder T cells: 8490±772, responder T cells+Tregs (control): 6084±2949, responder T cells+Tregs (NTN): 4137±845.2; *P<0.05). Depletion of Tregs in DEREG mice by injection of DTx To examine the involvement of endogenous Tregs in the model of crescentic glomerulonephritis, we used C57BL/6 DEREG mice in which specific depletion of FoxP3+ Tregs by injection of DTx and tracking of green fluorescent Tregs by fluorescent-activated cell sorting (FACS) analysis can be performed. To assess the efficiency of Treg depletion after repetitive injection of 1 μg DTx per mouse (day −1 and day 3 upon NTN induction), FoxP3 expression was quantified in spleen and kidney at 7 days after NTN induction. Indeed, frequency of CD4+FoxP3+ Tregs was significantly reduced in both spleen (Figure 2a; 7.8 vs <1%) and kidney tissue (Figure 2b; 20.1 vs 2.6%) of DTx-treated mice in contrast to control nephritic mice as shown by intracellular FoxP3 staining with subsequent FACS analysis. GFP expression was also reduced in accordance to FoxP3 expression (data not shown). Successful renal Treg depletion was confirmed by histological staining of FoxP3 (Figure 2c). The frequency of Tregs was significantly reduced by ∼80% at 4 days after the second DTx challenge (Figure 2d: FoxP3+ Tregs (cells per low power field): 24.7±2.7 vs 5.3±1.5; **P<0.01).

dance to FoxP3 expression (data not shown). Successful renal Treg depletion was confirmed by histological staining of FoxP3 (Figure 2c). The frequency of Tregs was significantly reduced by ∼80% at 4 days after the second DTx challenge (Figure 2d: FoxP3+ Tregs (cells per low power field): 24.7±2.7 vs 5.3±1.5; **P<0.01). Aggravated NTN in Treg-depleted nephritic mice Next, we induced NTN in C57BL/6 wild-type (wt) and DEREG mice with and without Treg depletion. Examination of periodic acid-Schiff (PAS)-stained kidney sections in the T cell-mediated autologous phase at day 7 after injection of the nephritogenic sheep serum revealed glomerular and tubulointerstitial damage. First, PAS-positive material was deposited intraglomerularly. Moreover, formation of cellular crescents and hypercellularity account for glomerular alterations (Figure 3a). In the tubulointerstitial compartment, aberrant leukocyte infiltration and focal destruction of the regular tissue structure were detectable indicated by tubular dilatation and intratubular protein casts (data not shown).

rmation of cellular crescents and hypercellularity account for glomerular alterations (Figure 3a). In the tubulointerstitial compartment, aberrant leukocyte infiltration and focal destruction of the regular tissue structure were detectable indicated by tubular dilatation and intratubular protein casts (data not shown). To quantify glomerular and tubular tissue damage, PAS-stained kidney sections were evaluated as previously described.7 The frequency of glomerular crescent formation on day 7 of NTN was significantly increased in Treg-depleted DEREG mice compared with nephritic DEREG mice without DTx injection (Figure 3b (%): wt+DTx: 1.7±1.7, DEREG+DTx: 1.7±1.0, wt NTN: 16.1±3.2, DEREG NTN: 16.1±3.2, wt NTN+DTx: 15.0± 3.3, DEREG NTN+DTx: 29.5±3.6; *P<0.05). As a measure of tubulointerstitial injury, an increase of the interstitial area estimated by point counting was observed in all nephritic groups. It did, however, not reveal differences with respect to Treg depletion (data not shown). The same results were obtained for urinary albumin/creatinine ratio (DEREG NTN vs DEREG NTN+DTx: 136.4±15.8 vs 133.7±19.6). Furthermore, blood urea nitrogen was elevated in all nephritic groups compared with controls, with a tendency to a further increase in Treg-depleted mice (Figure 3c (mg/dl): DEREG NTN vs DEREG NTN+DTx: 47±5.6 vs 57.9±10.4; P=0.33).

urinary albumin/creatinine ratio (DEREG NTN vs DEREG NTN+DTx: 136.4±15.8 vs 133.7±19.6). Furthermore, blood urea nitrogen was elevated in all nephritic groups compared with controls, with a tendency to a further increase in Treg-depleted mice (Figure 3c (mg/dl): DEREG NTN vs DEREG NTN+DTx: 47±5.6 vs 57.9±10.4; P=0.33). Increased renal T-cell and macrophage/monocyte/dendritic cell (DC) recruitment in Treg-depleted nephritic mice To investigate the role of Tregs regarding recruitment of T effector cells into the inflamed kidney, we performed immunohistological stainings of kidney sections for the pan T-cell marker CD3 in Treg-depleted and nondepleted nephritic DEREG and wt mice (Figure 4a). At 7 days after induction of NTN, nephritic mice showed an increased renal infiltration of CD3+ T cells, predominantly in the interstitial and periglomerular area. Moreover, the frequency of intraglomerular T cells in nephritic mice was increased compared with nonnephritic controls. Interestingly, T-cell infiltrates in Treg-depleted nephritic mice were significantly increased in contrast to nondepleted nephritic mice (Figure 4b (cells per high power field): wt+DTx: 13.0±2.0, DEREG+DTx: 22.7±1.3, wt NTN: 22.5±6.0, DEREG NTN: 32.5±5.4, wt NTN+DTx: 23.5±3.6, DEREG NTN+DTx: 62.3±10.4; *P<0.05). This suggests an important role of Tregs regarding inhibition of T effector cell recruitment into the inflamed kidney. We also analyzed the number of F4/80+ macrophages/monocytes and DCs in kidney sections of healthy control and nephritic mice with and without Treg depletion (Figure 4c). Similarly, we detected a general increase of macrophage/monocyte/DC infiltrates in nephritic mice in contrast to the healthy control group. Treg depletion resulted in further enhancement of macrophage/monocyte/DC recruitment into the kidney (Figure 4d (cells per high power field): wt+DTx: 9.0±1.5, DEREG+DTx: 10.7±1.7, wt NTN: 14.0±2.6, DEREG NTN: 26.3±3.6, wt NTN+DTx: 17.3±1.9, DEREG NTN+DTx: 42.3±5.3; *P<0.05).

ce in contrast to the healthy control group. Treg depletion resulted in further enhancement of macrophage/monocyte/DC recruitment into the kidney (Figure 4d (cells per high power field): wt+DTx: 9.0±1.5, DEREG+DTx: 10.7±1.7, wt NTN: 14.0±2.6, DEREG NTN: 26.3±3.6, wt NTN+DTx: 17.3±1.9, DEREG NTN+DTx: 42.3±5.3; *P<0.05). Upregulation of renal IFNγ response upon Treg depletion and NTN induction To assess the renal T helper cell response on day 7 upon NTN induction, we first isolated RNA from renal cortex of nephritic DEREG and wt mice as well as the corresponding controls and measured the expression of Th1 and Th17 effector cytokines. IFNγ mRNA expression was significantly upregulated in kidney tissue of Treg-depleted nephritic DEREG mice in contrast to nondepleted nephritic mice (Figure 5a: IFNγ: wt+DTx: 0.8±0.1, DEREG+DTx: 2.0±0.8, wt NTN: 1.0±0.2, DEREG NTN: 1.1±0.3, wt NTN+DTx: 1.4±0.2, DEREG NTN+DTx: 7.0±1.6; ***P<0.0001). IL-17 was generally upregulated in nephritic mice. However, in contrast to IFNγ, renal expression of IL-17 was unchanged in Treg-depleted and nondepleted nephritic mice (Figure 5a). The increased IFNγ expression induced by ablation of Tregs correlated with elevated CXCL10 (chemokine (C-X-C motif) ligand 10) expression (data not shown), one of the ligands for CXCR3 (chemokine (C-X-C motif) receptor 3) receptors that mainly mediate the migration of Th1 cells into the kidney.9

ritic mice (Figure 5a). The increased IFNγ expression induced by ablation of Tregs correlated with elevated CXCL10 (chemokine (C-X-C motif) ligand 10) expression (data not shown), one of the ligands for CXCR3 (chemokine (C-X-C motif) receptor 3) receptors that mainly mediate the migration of Th1 cells into the kidney.9 To verify the Treg-mediated suppression of IFNγ responses, we performed intracellular cytokine staining in renal CD4+ T cells from control and nephritic mice after in vitro stimulation with phorbol 12-myristate 13-acetate/ionomycin. Corresponding to the elevated IFNγ mRNA expression in kidney tissue, the IFNγ protein expression was strongly increased in renal T cells isolated from Treg-depleted nephritic mice compared with nondepleted nephritic mice (Figure 5b: 43.6 vs 17.4%). Again, protein expression of IL-17 remained nearly unchanged between 2 and 3% in all groups of nephritic mice (Figure 5b). Quantification of three independent FACS analyses displayed a significant increase of IFNγ- but not IL-17-producing CD4+ T cells in Treg-depleted nephritic DEREG mice in contrast to nondepleted nephritic DEREG or wt mice (Figure 5c: IFNγ+CD4+ T cells (%): wt NTN: 17±1.4, DEREG NTN: 15.8±1.4, wt NTN+DTx: 18.5±0.7, DEREG NTN+DTx: 35.5±4.3; ***P<0.0001).

nt FACS analyses displayed a significant increase of IFNγ- but not IL-17-producing CD4+ T cells in Treg-depleted nephritic DEREG mice in contrast to nondepleted nephritic DEREG or wt mice (Figure 5c: IFNγ+CD4+ T cells (%): wt NTN: 17±1.4, DEREG NTN: 15.8±1.4, wt NTN+DTx: 18.5±0.7, DEREG NTN+DTx: 35.5±4.3; ***P<0.0001). To evaluate whether the absence of Tregs during the induction phase of nephritis is sufficient for the aggravated phenotype in Treg-depleted mice, we directly compared a single injection of DTx on day –1 with the continuous Treg depletion (days –1 and 3; Supplementary Figure S1 online). Interestingly, early depletion of Tregs resulted in a sustained reduction of splenic Tregs from 16.7 to 5.9% up to day 7, whereas a double DTx injection was necessary for complete depletion of splenic Tregs (Supplementary Figure S1A online: CD4+FoxP3+GFP+ T cells (%): 16.7±0.75 vs 2.9±0.5; ***P<0.0001). Accordingly, glomerular damage was only slightly elevated in mice with a single early depletion but was considerably aggravated in continuously depleted mice (Supplementary Figure S1B online).

te depletion of splenic Tregs (Supplementary Figure S1A online: CD4+FoxP3+GFP+ T cells (%): 16.7±0.75 vs 2.9±0.5; ***P<0.0001). Accordingly, glomerular damage was only slightly elevated in mice with a single early depletion but was considerably aggravated in continuously depleted mice (Supplementary Figure S1B online). Augmented systemic immune response in Treg-depleted nephritic mice To investigate the role of endogenous CD4+CD25+FoxP3+ Tregs regarding the systemic immune response in the mouse model of NTN, splenocytes were isolated from wt and DEREG mice 7 days after induction of NTN and stimulated in vitro with sheep immunoglobulin G (IgG), the nephritogenic antigen. Consistent with the results obtained in the inflamed target organ, namely in the kidney, ELISA analysis of splenocyte supernatants indicated that Treg depletion in nephritic mice resulted in an increased systemic IFNγ secretion compared with NTN mice without Treg depletion (Figure 6a, IFNγ (pg/ml): wt+DTx: 1.0±0.02, DEREG+DTx: 1.0±0.02, wt NTN: 421.0±54.5, DEREG NTN: 608.7±37.2, wt NTN+DTx: 307.0±37.9, DEREG NTN+DTx: 3457.0±175.8; ***P<0.0001). IL-17 secretion by splenocytes from mice with NTN was not elevated in Treg-depleted nephritic mice (Figure 6a, IL-17 (pg/ml): wt+DTx: 1.0±0.02, DEREG+DTx: 1.0±0.02, wt NTN: 39.3±3.7, DEREG NTN: 45.7±3.8, wt NTN+DTx: 60.8±3.0, DEREG NTN+DTx: 25.0±2.1; ***P<0.0001).

307.0±37.9, DEREG NTN+DTx: 3457.0±175.8; ***P<0.0001). IL-17 secretion by splenocytes from mice with NTN was not elevated in Treg-depleted nephritic mice (Figure 6a, IL-17 (pg/ml): wt+DTx: 1.0±0.02, DEREG+DTx: 1.0±0.02, wt NTN: 39.3±3.7, DEREG NTN: 45.7±3.8, wt NTN+DTx: 60.8±3.0, DEREG NTN+DTx: 25.0±2.1; ***P<0.0001). Moreover, systemic humoral immune responses to sheep IgG are important for initiation of the autologous phase of NTN. Hence, we determined the IgG antibody response directed against the nephritogenic antigen in serum samples by ELISA for sheep IgG-specific mouse IgG subclasses. Depletion of Tregs resulted in significantly increased levels of total IgG and the isotypes of IgG1, IgG2a/c,10 and IgG2b (Figure 6b). Furthermore, we performed immunohistochemical staining of kidney sections for sheep IgG (Supplementary Figure S2A online) and mouse IgG (Supplementary Figure S2B online) to evaluate the amount of antibody deposition in injured glomeruli. Semiquantitative assessment of mouse IgG-stained sections revealed that discrete granular mouse IgG immune deposit within the glomeruli were increased in nephritic mice but were not further elevated in nephritic Treg-depleted mice compared with nephritic mice without depletion (Supplementary Figure S2B online).

meruli. Semiquantitative assessment of mouse IgG-stained sections revealed that discrete granular mouse IgG immune deposit within the glomeruli were increased in nephritic mice but were not further elevated in nephritic Treg-depleted mice compared with nephritic mice without depletion (Supplementary Figure S2B online). Reduced disease severity in Treg-depleted nephritic mice upon anti-IFNγ treatment To investigate the role of endogenous Tregs regarding Treg-mediated suppression of Th1 immune responses, we inhibited the Th1-specific IFNγ response by injection of an anti-IFNγ antibody together with the first DTx dose at 1 day before NTN induction in DEREG mice. Indeed, aggravation of glomerular damage as measured by a significant increase of crescent formation in Treg-depleted nephritic controls was significantly inhibited in Treg-depleted, anti-IFNγ-treated nephritic mice (Figure 7a (%): wt: 1.1±0.6, DEREG NTN: 27±1.3, DEREG NTN+DTx: 39.5±4.2, DEREG NTN+αIFNγ: 19.5±2.3, DEREG NTN+DTx+αIFNγ: 26.3±1.4; **P<0.01). However, blood urea nitrogen was elevated in all nephritic groups compared with controls, but did not show a significant reduction in anti-IFNγ-treated mice (Figure 7b). FACS analysis of renal T cells indicated that anti-IFNγ treatment reversed the enhanced frequency of IFNγ-producing CD4+ Th1 cells in Treg-depleted nephritic mice on day 7 upon NTN induction (Figure 7c; 28.6 vs 48%). Concordantly, renal expression of T-bet, a Th1-specific transcription factor, as well as of the IFNγ-inducible chemokine CXCL10, which recruits CXCR3+ Th1 cells, were significantly inhibited upon anti-IFNγ treatment (Figure 7d: CXCL10: wt: 1±0.08, DEREG+NTN: 7.5±0.8, DEREG NTN+DTx: 19.5±2.8, DEREG NTN+αIFNγ: 8±0.9, DEREG NTN+DTx +αIFNγ: 11.3±0.4; T-bet: wt: 1±0.24, DEREG+NTN: 2.7±0.4, DEREG NTN+DTx: 7.9±0.9, DEREG NTN+αIFNγ: 2.1±0.35, DEREG NTN+DTx +αIFNγ: 4.1±1.3; ***P<0.0001; *P<0.05). Moreover, measurement of the systemic humoral anti-sheep immune response in mouse serum indicated that the IFNγ-related IgG2a/c response was significantly inhibited in anti-IFNγ-treated Treg-depleted nephritic DEREG mice compared with the nontreated ones (Figure 7e (optical density at 450 nm): wt: 0.17±0.01, DEREG+NTN: 0.39±0.03, DEREG NTN+DTx: 1.36±0.18, DEREG NTN+αIFNγ: 0.37±0.03, DEREG NTN+DTx +αIFNγ: 0.87±0.16; ***P<0.0001; *P<0.05). In contrast, concentrations of the ‘non-Th1-related' antibodies IgG1 and IgG2b remained unaffected (Figure 7e).

h the nontreated ones (Figure 7e (optical density at 450 nm): wt: 0.17±0.01, DEREG+NTN: 0.39±0.03, DEREG NTN+DTx: 1.36±0.18, DEREG NTN+αIFNγ: 0.37±0.03, DEREG NTN+DTx +αIFNγ: 0.87±0.16; ***P<0.0001; *P<0.05). In contrast, concentrations of the ‘non-Th1-related' antibodies IgG1 and IgG2b remained unaffected (Figure 7e). DISCUSSION Tregs develop in the thymus in order to regulate autoimmunity in the periphery. Upon inflammation, they become activated and increase in number, either by proliferation of thymus-derived naturally occurring natural Tregs and/or by peripheral conversion of CD4+CD25− responder T cells to CD4+CD25+FoxP3+ Tregs.11 Here we describe that upon induction of experimental crescentic glomerulonephritis, frequency as well as absolute number of FoxP3+ Tregs increased substantially within renal tissue and more weakly within spleen.

ing natural Tregs and/or by peripheral conversion of CD4+CD25− responder T cells to CD4+CD25+FoxP3+ Tregs.11 Here we describe that upon induction of experimental crescentic glomerulonephritis, frequency as well as absolute number of FoxP3+ Tregs increased substantially within renal tissue and more weakly within spleen. Depletion of Tregs by repetitive injections of DTx on day –1 and day 3 upon NTN induction resulted in exacerbation of glomerular crescent formation and mouse anti-sheep IgG antibody production as well as in enhanced systemic and renal IFNγ production. However, depletion of Tregs only during the heterologous phase of NTN nephritis, that is, when antigen-specific priming is likely to occur in the lymphoid tissue, failed to increase crescent formation (Supplementary Figure S1B online). We observed similar results by starting Treg depletion at the beginning of the autologous phase (data not shown). Hence, it seems likely that both Tregs that home in the lymphoid tissue and Tregs that migrate into the inflamed kidney control glomerulonephritis. The enhanced IFNγ response observed in the aggravated phenotype correlated with an increase of macrophage/monocyte/DC and T-cell infiltration into periglomerular and interstitial areas of the renal cortex. As Tregs were depleted in this experimental setting, the infiltrating CD3+ T cells most likely consisted of activated effector T lymphocytes of which the CD4+ cells indeed produced substantially increased amounts of IFNγ whereas IL-17 production remained unaffected, although glomerular injury in our model is mediated by a Th1 as well as by a Th17 response.9, 12 However, it seems noteworthy that the systemic release of IL-17 detected in supernatants of splenocyte cultures was significantly suppressed in the Treg-depleted group. In view of the fact that Tregs also suppress maturation and function of DCs,13 this can be explained by the observation that IL-12, the major inducer of IFNγ, decreases IL-17 production.14 Intriguingly, a recent publication also showed that Tregs from patients with active pulmonary tuberculosis suppressed IFNγ production, whereas the Th17 response remained unaffected.15 Taken together, the most dramatic effect of Treg depletion observed in our study was aggravation of the IFNγ T-cell response. These results suggest that the observed increase of frequency and absolute number of immunosuppressive Tregs during acute NTN regulates the systemic and local inflammatory Th1 response.

ected.15 Taken together, the most dramatic effect of Treg depletion observed in our study was aggravation of the IFNγ T-cell response. These results suggest that the observed increase of frequency and absolute number of immunosuppressive Tregs during acute NTN regulates the systemic and local inflammatory Th1 response. Noteworthy, in our experiments DEREG mice followed the equivalent characteristic features of NTN compared with C57BL/6 wt mice. Furthermore, DTx neither caused renal toxicity in wt or DEREG mice nor nonresponsiveness toward itself,16 as PAS staining failed to show histopathological alterations and DTx efficiently depleted Tregs as measured 4 days after the second injection, respectively.

acteristic features of NTN compared with C57BL/6 wt mice. Furthermore, DTx neither caused renal toxicity in wt or DEREG mice nor nonresponsiveness toward itself,16 as PAS staining failed to show histopathological alterations and DTx efficiently depleted Tregs as measured 4 days after the second injection, respectively. Up to now, three reports dealt with suppressive effects of Tregs in the planted antigen model of nephrotoxic serum nephritis. Adoptive transfer of CD4+CD25+ Tregs protected mice from nephritis induced by anti-glomerular basement membrane rabbit serum. Measuring tissue levels of FoxP3 mRNA, the authors concluded that most of the adoptively transferred Tregs migrated to secondary lymphoid organs but not into the inflamed kidney.5 However, using FACS analysis and quantitative immunohistochemistry, we could recently demonstrate that adoptively transferred Tregs migrate into the kidney as long as they express the chemokine receptor CCR6,7 and CCR6 expression by adoptively transferred Tregs was absolutely necessary for their protective effect. Hence, together with the finding that Treg depletion only in the heterologous phase failed to induce the aggravated phenotype, it seems that CD4+FoxP3+ Tregs have to accumulate within the inflamed kidney, at least in dependence of CCR6, in order to sufficiently suppress glomerulonephritis. Results of the third report suggest that lymphoid homing of Tregs is a prerequisite for downregulation of anti-glomerular basement membrane nephritis.6 Our results presented here show that in contrast to spleen, CD4+FoxP3+ Tregs substantially increased within kidney during NTN nephritis in comparison with normal frequency of ∼5% in healthy mice, and depletion of these cells significantly correlated with aggravation of glomerulonephritis. The necessity of inflammatory tissue-homing Tregs for suppression of inflammatory tissue damage has also been observed in models of intestinal inflammation17 and inflammatory liver injury.18 Taken together, these data indicate that the anti-inflammatory functions of Tregs in experimental glomerulonephritis take place in the inflamed kidney itself and in the secondary lymphatic organs such as the renal lymph nodes.

as also been observed in models of intestinal inflammation17 and inflammatory liver injury.18 Taken together, these data indicate that the anti-inflammatory functions of Tregs in experimental glomerulonephritis take place in the inflamed kidney itself and in the secondary lymphatic organs such as the renal lymph nodes. In addition to anti-glomerular basement membrane nephritis,5 adoptively transferred Tregs have been shown to prevent murine lupus nephritis.19 However, Treg cells did not only suppress ‘immune-mediated' glomerulonephritis, but also ‘non-immune' toxin-mediated kidney injury in the adriamycin nephropathy model. In this model, polyclonal CD4+CD25+ Tregs20 and, moreover, FoxP3-transduced T cells,21 γδ T cells, or alternatively activated macrophages3 were approved as efficient immunotherapy against renal injury. As anti-inflammatory cytokines represent one critical mechanism of regulatory immune cells including Tregs, these findings support the concept that Tregs not only suppress adaptive T-cell responses but also pathology driven by innate immune mechanisms.22 Hence, the immunosuppressive mechanisms by which Tregs suppress glomerulonephritis resulting from different etiologies remain to be elucidated.

immune cells including Tregs, these findings support the concept that Tregs not only suppress adaptive T-cell responses but also pathology driven by innate immune mechanisms.22 Hence, the immunosuppressive mechanisms by which Tregs suppress glomerulonephritis resulting from different etiologies remain to be elucidated. In patients with autoimmune kidney disease such as lupus nephritis, decreased proportions of Tregs have been reported to inversely correlate with clinical disease activity.23, 24 However, several studies also reported unaltered or even increased numbers of Tregs in lupus patients and positive correlation with disease activity. Similar results were obtained when the functional immunosuppressive activity of Tregs from lupus patients was assessed. These divergent results can be explained by the fact that in contrast to the murine system, in humans FoxP3 is rather an activity marker for T cells than a specific marker for Tregs. Although one can imagine that Treg numbers and function are impaired during chronic disease, our results shown here demonstrate that during acute glomerulonephritis, Treg numbers initially increased and their immunosuppressive function was at least as effective as the function of Tregs from healthy controls.

for Tregs. Although one can imagine that Treg numbers and function are impaired during chronic disease, our results shown here demonstrate that during acute glomerulonephritis, Treg numbers initially increased and their immunosuppressive function was at least as effective as the function of Tregs from healthy controls. In conclusion, depletion of Tregs exacerbated acute kidney inflammation and glomerular injury, an effect that was most likely mediated by IFNγ, as IFNγ production by intrarenal and splenic CD4+ T cells substantially increased and anti-IFNγ antibodies significantly suppressed this effect. Hence, this study might be useful to evaluate novel therapeutic approaches by modulating Tregs in order to suppress Th1 immune responses in kidney diseases. MATERIALS AND METHODS Animals Male DEREG mice expressing the DTx receptor and GFP under control of the FoxP3 promoter8 and sex/age-matched (8–10 weeks old) C57BL/6 wt controls were obtained from animal facilities of the Universitätsklinikum Hamburg-Eppendorf. Animals received humane care according to guidelines of the National Institute of Health in Germany. Experiments were approved by the institutional review board ‘Behörde für Soziales, Familie, Gesundheit und Verbraucherschutz' (Hamburg).

ere obtained from animal facilities of the Universitätsklinikum Hamburg-Eppendorf. Animals received humane care according to guidelines of the National Institute of Health in Germany. Experiments were approved by the institutional review board ‘Behörde für Soziales, Familie, Gesundheit und Verbraucherschutz' (Hamburg). Animal treatment and functional studies Nephrotoxic serum nephritis was induced in C57BL/6 wt and DEREG mice by intraperitoneal injection of 500 μl of nephrotoxic sheep serum per mouse. Controls were injected intraperitoneally with an equal amount of nonspecific sheep IgG. To deplete Tregs in DEREG mice, DTx (1 μg per mouse) was injected intraperitoneally either twice 24 h before NTN induction and 3 days upon NTN induction or only once 24 h before NTN induction. Rat anti-mouse IFNγ monoclonal antibody (XGM1.2; see Mumberg et al.25); 500 μg/mouse) or rat IgG was injected intraperitoneally 24 h before NTN induction. Blood samples for measurement of blood urea nitrogen and assessment of systemic antibody response were obtained at the time of killing. For urine sample collection, mice were housed in metabolic cages for 6 h. Urinary creatinine and blood urea nitrogen were measured by standard laboratory methods. Albuminuria was determined by standard ELISA analysis (mice-albumin kit; Bethyl Laboratories, Montgomery, TX).

response were obtained at the time of killing. For urine sample collection, mice were housed in metabolic cages for 6 h. Urinary creatinine and blood urea nitrogen were measured by standard laboratory methods. Albuminuria was determined by standard ELISA analysis (mice-albumin kit; Bethyl Laboratories, Montgomery, TX). Real-time quantitative reverse transcription-PCR analysis Total RNA was reversely transcribed followed by quantitative reverse transcription-PCR using BIORAD CFX96 real-time system and ABsoluteQPCR SYBR mix (Fisher Scientific GmbH, Schwerte, Germany). Primer pairs were used as described previously.7, 8, 9 Relative mRNA levels were calculated after normalization to 18S rRNA using the CFX96 Manager software (Bio-Rad Laboratories, Munich, Germany).

anscription-PCR using BIORAD CFX96 real-time system and ABsoluteQPCR SYBR mix (Fisher Scientific GmbH, Schwerte, Germany). Primer pairs were used as described previously.7, 8, 9 Relative mRNA levels were calculated after normalization to 18S rRNA using the CFX96 Manager software (Bio-Rad Laboratories, Munich, Germany). Morphological examinations Light microscopy and immunohistochemistry were performed by routine procedures. Crescent formation was assessed in 30 glomeruli per mouse in a blinded fashion in PAS-stained paraffin sections. Tubulointerstitial injury was estimated by point counting in four independent areas of renal cortex per mouse in 200-fold magnification.7 Paraffin-embedded sections (2 μm) were either stained with an antibody directed against the pan-T-cell marker CD3 (A0452; Dako, Hamburg, Germany), the Treg transcription factor FoxP3 (FJK-16s; eBioscience, San Diego, CA), the macrophage/monocyte/DC marker F4/80 (BM8; BMA, Augst, Switzerland), and sheep or mouse IgG (Jackson ImmunoResearch Europe, Newmarket, UK). Tissue sections were developed with the Vectastain ABC-AP kit (Vector Laboratories, Burlingame, CA). For FoxP3 staining, tissue sections were incubated with a polyclonal rabbit anti-rat secondary antibody (Dako) and developed with the ZytoChemPlus (AP) Polymer Kit (Zytomed Systems, Berlin, Germany). CD3+ cells in 20 glomerular cross-sections and F4/80+ and CD3+ cells in 20 tubulointerstitial high power fields ( × 400) per kidney were counted in a blinded fashion. For quantification of FoxP3+ Tregs, at least seven low power fields ( × 100) were counted. Glomerular mouse IgG deposition was scored from 0 to 3 in 20 glomeruli per mouse as previously described.7

nd F4/80+ and CD3+ cells in 20 tubulointerstitial high power fields ( × 400) per kidney were counted in a blinded fashion. For quantification of FoxP3+ Tregs, at least seven low power fields ( × 100) were counted. Glomerular mouse IgG deposition was scored from 0 to 3 in 20 glomeruli per mouse as previously described.7 Antigen-specific humoral immune response Mouse anti-sheep IgG antibody titers were measured by ELISA using sera collected 7 days after induction of nephritis.7, 9, 26 In brief, ELISA microtiter plates were coated with 100 μl sheep IgG (100 μg/ml; Sigma, St Louis, MO) in carbonate–bicarbonate buffer overnight at 4 °C. After blocking with 1% bovine serum albumin in Tris-buffered saline (Sigma), plates were incubated with serial dilutions of mouse serum for 1 h at room temperature. Bound mouse IgG was detected using peroxidase-conjugated goat anti-mouse IgG (Biozol, Eching, Germany), TMB peroxidase substrate, and absorbance readings (450 nm). Lack of crossreactivity of the secondary antibody with sheep IgG was demonstrated by omitting the primary antibody. The bound mouse immunoglobulin isotypes were detected using peroxidase-conjugated rabbit anti-mouse IgG, IgG1, IgG2a/c, and IgG2b antibodies (Zymed-Invitrogen, Karlsruhe, Germany).

ance readings (450 nm). Lack of crossreactivity of the secondary antibody with sheep IgG was demonstrated by omitting the primary antibody. The bound mouse immunoglobulin isotypes were detected using peroxidase-conjugated rabbit anti-mouse IgG, IgG1, IgG2a/c, and IgG2b antibodies (Zymed-Invitrogen, Karlsruhe, Germany). Leukocyte isolation from various tissues Previously described methods for leukocyte isolation from murine kidneys were used.7, 26 In brief, kidneys were finely minced and digested for 45 min at 37 °C with 0.4 mg/ml collagenase D (Roche, Mannheim, Germany) and 0.01 mg/ml DNAse I in Dulbecco's modied Eagle's medium (Roche) supplemented with 10% heat-inactivated fetal calf serum (Invitrogen, Darmstadt, Germany). Cell suspensions were sequentially filtered through 70 and 40 μm nylon meshes and washed with Hank's balanced salt solution without Ca2+ and Mg2+ (Invitrogen). Single-cell suspensions were separated using Percoll density gradient (70 and 40%) centrifugation. Single-cell suspensions of spleens were prepared according to standard laboratory procedures. In brief, tissue was passed through 70 and 40 μm nylon meshes before erythrocyte lysis. Subsequently, cells were washed several times with Hank's balanced salt solution and resuspended in RPMI-1640 with 10% fetal calf serum. Viability of the cells was assessed by Trypan blue staining before flow cytometry.

procedures. In brief, tissue was passed through 70 and 40 μm nylon meshes before erythrocyte lysis. Subsequently, cells were washed several times with Hank's balanced salt solution and resuspended in RPMI-1640 with 10% fetal calf serum. Viability of the cells was assessed by Trypan blue staining before flow cytometry. Flow cytometry Leukocytes were stained using a standard protocol. For T-cell differentiation, isolated cells were stained with anti-CD3 (APC; eBioscience), anti-CD4 (APC-AlexaFluor750), and anti-CD45 (PerCP; both from BD Biosciences, Franklin Lakes, NJ) upon a blocking step. Staining of intracellular IFNγ and IL-17 was performed as described previously.12 In brief, splenocytes or renal leukocytes were stimulated with phorbol 12-myristate 13-acetate (50 ng/ml; Sigma) and ionomycin (1 μg/ml; Calbiochem-Merck, Darmstadt, Germany) for 5 h. After 30 min of incubation, Brefeldin A (10 μg/ml; Sigma) was added. After several washing steps and staining of cell surface markers, cells were permeabilized using Cytofix/Cytoperm (BD Biosciences). Subsequently, intracellular staining was performed using rat anti-mouse IFNγ and IL-17 antibodies (V450 or PE; BD Biosciences) and an anti-FoxP3 antibody (eBioscience). Data were recorded using BD LSRII Flow Cytometry system and BD FACSDiva software.

rs, cells were permeabilized using Cytofix/Cytoperm (BD Biosciences). Subsequently, intracellular staining was performed using rat anti-mouse IFNγ and IL-17 antibodies (V450 or PE; BD Biosciences) and an anti-FoxP3 antibody (eBioscience). Data were recorded using BD LSRII Flow Cytometry system and BD FACSDiva software. Isolation and culture of splenic CD4+CD25+ Tregs and responder T cells Spleens were excised from C57BL/6 wt 8 days after induction of NTN, and from healthy controls and passed through 100 μm nylon meshes. Sorting procedures were carried out by MACS according to the manufacturers' instructions (MACS CD4+ T-Cell-Isolation Kit; Miltenyi Biotec, Bergisch-Gladbach, Germany). Briefly, CD4+ T cells were enriched using a biotinylated antibody cocktail depleting all other blood cell types and anti-biotin microbeads. CD4+CD25+ T cells were isolated by positive selection using phycoerythrin-labeled anti-CD25 monoclonal antibody and anti-phycoerythrin microbeads. Purity and intracellular FoxP3 expression was controlled by flow cytometry. Next, 1 × 105 wt responder T cells (CD4+CD25−) isolated from healthy mice were cultured alone or with 1 × 105 CD4+CD25+ Tregs from nephritic wt or healthy controls for 72 h in 96-well plates precoated with anti-CD3 monoclonal antibody (5 μg/ml; clone 145–2C11; BD Biosciences). Cytokine concentrations were measured in supernatants by ELISA.

onder T cells (CD4+CD25−) isolated from healthy mice were cultured alone or with 1 × 105 CD4+CD25+ Tregs from nephritic wt or healthy controls for 72 h in 96-well plates precoated with anti-CD3 monoclonal antibody (5 μg/ml; clone 145–2C11; BD Biosciences). Cytokine concentrations were measured in supernatants by ELISA. Statistical analysis Results are expressed as mean±s.e.m. Differences between individual experimental groups were compared by Student's t-test. In case of multiple comparisons, one-way analysis of variance with post analysis by Tukey–Kramer test was used. Experiments that did not yield enough independent data for statistical analysis due to the experimental setup were repeated at least three times. We thank Anett Peters, Sabrina Bennstein, and Elena Tasika for the perfect technical assistance. This work was supported by grants from the Deutsche Forschungsgemeinschaft (Klinische Forschergruppe 228: PA 754/7-1 to UP and J-ET, MI 476/4-1 to H-WM, and TI169/9-1 to GT). SUPPLEMENTARY MATERIAL Figure S1 Correlation of efficiency of Treg depletion and crescent formation upon single or double injection of DTx to DEREG mice. Figure S2. Semiquantitative scoring of glomerular mouse IgG deposition. Supplementary material is linked to the online version of the paper at http://www.nature.com/ki All the authors declared no competing interests. Supplementary Material Supplementary Figure1 Click here for additional data file. Supplementary Figure2 Click here for additional data file.

Figure S2. Semiquantitative scoring of glomerular mouse IgG deposition. Supplementary material is linked to the online version of the paper at http://www.nature.com/ki All the authors declared no competing interests. Supplementary Material Supplementary Figure1 Click here for additional data file. Supplementary Figure2 Click here for additional data file. Figure 1 Endogenous regulatory T cells (Tregs) in the murine model of nephrotoxic nephritis (NTN). (a) Representative immunohistochemistry for Tregs (forkhead box P3 (FoxP3)) in kidney sections from nephritic mice and nonnephritic controls (original magnification × 400). (b) Quantification of tubulointerstitial FoxP3+ T-cell infiltration per low power field (lpf; original magnification × 100) during the time course of NTN (n⩾4). (c) 1 × 105 Splenic CD4+CD25− responder T cells were cultured without or with 1 × 105 splenic CD4+CD25+ Tregs from nonnephritic wild-type (wt) controls or nephritic wt mice and stimulated with plate-bound anti-CD3 monoclonal antibody (mAb) for 72 h. Secretion of the cytokines interleukin-2 (IL-2) and interferon-γ (IFNγ) was assessed in supernatants by enzyme-linked immunosorbent assay (ELISA; *P<0.05, ***P<0.0001).

CD25+ Tregs from nonnephritic wild-type (wt) controls or nephritic wt mice and stimulated with plate-bound anti-CD3 monoclonal antibody (mAb) for 72 h. Secretion of the cytokines interleukin-2 (IL-2) and interferon-γ (IFNγ) was assessed in supernatants by enzyme-linked immunosorbent assay (ELISA; *P<0.05, ***P<0.0001). Figure 2 Depletion of regulatory T cells (Tregs) in DEREG (depletion of regulatory T cell) mice by diphtheria toxin (DTx). Systemic depletion of Tregs was monitored by fluorescent-activated cell sorting (FACS) analysis of (a) splenic and (b) renal CD4+ T cells isolated from DEREG mice 7 days after induction of nephrotoxic nephritis (NTN) and 4 days after second application of DTx. (c) Representative kidney sections from nephritic DEREG mice without and with DTx treatments show FoxP3+ staining. (d) Quantification of renal and splenic FoxP3+ T cells indicates a strong reduction in numbers of FoxP3+ T cells after DTx treatments (**P<0.01). Figure 3 Aggravated nephrotoxic nephritis (NTN) in regulatory T cell (Treg)-depleted DEREG (depletion of regulatory T cell) mice. (a) Representative photographs of periodic acid-Schiff (PAS)-stained kidney sections of control mice (wild-type (wt)+diphtheria toxin (DTx)) and nephritic wt or DEREG mice without or with DTx treatment (original magnification × 400) on day 7 after induction of NTN. (b) Quantification of glomerular crescents (*P<0.05) was performed. (c) Blood urea nitrogen (BUN) was determined.

)-stained kidney sections of control mice (wild-type (wt)+diphtheria toxin (DTx)) and nephritic wt or DEREG mice without or with DTx treatment (original magnification × 400) on day 7 after induction of NTN. (b) Quantification of glomerular crescents (*P<0.05) was performed. (c) Blood urea nitrogen (BUN) was determined. Figure 4 Increased renal T-cell and macrophage (Mφ)/monocyte/dendritic cell (DC) recruitment in regulatory T cell (Treg)-depleted nephritic mice. Representative photographs of kidney sections from nephritic DEREG (depletion of regulatory T cell) mice with or without diphtheria toxin (DTx) treatment were immunohistochemically stained for the (a) T-cell marker CD3 and the (c) monocyte/macrophage/DC marker F4/80, 7 days after nephrotoxic nephritis (NTN) induction (original magnification × 400). (b) Quantification of tubulointerstitial and glomerular CD3+ T cells in nephritic mice without or with DTx treatment and (d) tubulointerstitial F4/80+ cells per high power field (hpf) was assessed (*P<0.05).

e/macrophage/DC marker F4/80, 7 days after nephrotoxic nephritis (NTN) induction (original magnification × 400). (b) Quantification of tubulointerstitial and glomerular CD3+ T cells in nephritic mice without or with DTx treatment and (d) tubulointerstitial F4/80+ cells per high power field (hpf) was assessed (*P<0.05). Figure 5 Upregulation of renal Th1 response upon regulatory T cell (Treg) depletion in nephritic mice. (a) Total RNA was extracted from kidneys of wild-type (wt) and DEREG (depletion of regulatory T cell) mice with/without nephrotoxic nephritis (NTN)±diphtheria toxin (DTx) treatment. Subsequently, quantitative real-time reverse transcription (RT)-PCR was performed for interferon-γ (IFNγ) and interleukin-17 (IL-17) expression (***P<0.0001). The expression levels are indicated as x-fold of nonnephritic wt controls. (b) Renal single-cell suspensions from wt and DEREG mice without or with NTN±DTx treatment were stimulated in vitro with phorbol 12-myristate 13-acetate (PMA)/ionomycin. Intracellular cytokine production of IFNγ and IL-17 was analyzed by flow cytometry. Representative dot plots are depicted. Cells are gated on CD4+ T cells and numbers represent events in quadrants in percentage of all gated events. (c) The frequencies of IFNγ- and IL-17-producing renal CD4+ T cells were quantified (***P<0.0001). All analyses were performed at day 7 after induction of NTN.

w cytometry. Representative dot plots are depicted. Cells are gated on CD4+ T cells and numbers represent events in quadrants in percentage of all gated events. (c) The frequencies of IFNγ- and IL-17-producing renal CD4+ T cells were quantified (***P<0.0001). All analyses were performed at day 7 after induction of NTN. Figure 6 Enhanced systemic Th1 immune response in regulatory T cell (Treg)-depleted mice. (a) Cytokine secretion of interferon-γ (IFNγ) and interleukin-17 (IL-17) in supernatants of cultured splenocytes after treatment with sheep immunoglobulin G (IgG) was measured by enzyme-linked immunosorbent assay (ELISA; ***P<0.0001). (b) Circulating titers of mouse anti-sheep total IgG and isotypes of IgG1, IgG2a/c, and IgG2b at day 7 after induction of nephritis were measured by ELISA in wild-type (wt) and DEREG (depletion of regulatory T cell) mice without and with nephrotoxic nephritis (NTN) and diphtheria toxin (DTx) treatment. Serum was diluted as indicated (***P<0.0001, **P<0.01, *P<0.05).

IgG and isotypes of IgG1, IgG2a/c, and IgG2b at day 7 after induction of nephritis were measured by ELISA in wild-type (wt) and DEREG (depletion of regulatory T cell) mice without and with nephrotoxic nephritis (NTN) and diphtheria toxin (DTx) treatment. Serum was diluted as indicated (***P<0.0001, **P<0.01, *P<0.05). Figure 7 Attenuation of renal and systemic Th1 immune response upon neutralization of interferon-γ (IFNγ). (a) Quantification of glomerular crescent formation in depleted or nondepleted nephritic DEREG (depletion of regulatory T cell) mice without or with administration of anti-IFNγ antibody (**P<0.01). (b) Blood urea nitrogen (BUN) levels were determined in nonnephritic wild-type (wt) littermates and nephritic DEREG mice. (c) Renal single-cell suspensions from wt mice and nephritic DEREG mice±diphtheria toxin (DTx)±αIFNγ treatment were stimulated in vitro with phorbol 12-myristate 13-acetate (PMA)/ionomycin. Intracellular cytokine production of IFNγ was analyzed by flow cytometry. Representative dot plots are depicted. Cells are gated on CD4+ T cells and numbers represent events in quadrants in percentage of all gated events. (d) Quantitative real-time PCR analysis of renal mRNA expression of CXCL10 (chemokine (C-X-C motif) ligand 10) and T-bet was performed in nephritic DEREG mice (±DTx; ±αIFNγ) and wt littermates (***P<0.0001, *P<0.05). The mRNA levels are expressed as x-fold of nonnephritic wt controls. (e) Circulating titers of mouse anti-sheep total immunoglobulin G (IgG) and isotypes of IgG1, IgG2a/c, and IgG2b were measured by enzyme-linked immunosorbent assay (ELISA) in wt mice and DEREG nephrotoxic nephritis (NTN) mice with or without treatment of DTx and/or αIFNγ. Serum was diluted as indicated (***P<0.0001, **P<0.01, *P<0.05). All analyses were performed 7 days after NTN induction.

In incident and prevalent patients on dialysis, vascular calcifications are frequent. The results of the first seminal report showing that dialysis patients have remarkably higher incidence of coronary artery, mitral, and aortic valve calcification1 have been repeatedly confirmed in recent years; hence, latest guidelines have included the detection of extra osseous calcification as a test for diagnosing mineral and bone disorders in chronic kidney disease (CKD).2 Vascular calcifications, namely, coronary artery calcifications (CACs), are regarded as markers of severe vasculopathy1, 2, 3, 4, 5 and strong predictors of cardiovascular events.2, 4, 5, 6 Importantly, both in young7 and adult dialysis patients,1, 6, 8, 9 calcifications progress more rapidly than in controls; the progression is an additional independent factor responsible for cardiovascular events. CACs are present even in patients at early stages of CKD (CKD patients). Indeed, CACs were found in almost one half of non-diabetic predialysis patients (CKD stages 3–5);10 higher prevalence of CAC in CKD patients compared with controls has been subsequently confirmed by larger studies that have included diabetic CKD patients.11, 12, 13, 14

in patients at early stages of CKD (CKD patients). Indeed, CACs were found in almost one half of non-diabetic predialysis patients (CKD stages 3–5);10 higher prevalence of CAC in CKD patients compared with controls has been subsequently confirmed by larger studies that have included diabetic CKD patients.11, 12, 13, 14 Faster progression of calcifications has been observed in CKD patients;15, 16, 17, 18, 19 the underlying mechanisms are still not well determined. In fact, although in dialysis patients, the progression of calcification seems to be linked to deranged mineral metabolism,7, 8, 9 in CKD patients, this link has not been confirmed15, 16, 17, 19 despite faster progression of CAC occurred in patients with high-normal serum phosphorus15 and progression was reduced by binding the ion.16 Mild-to-moderate elevations in serum creatinine levels are associated with increased rates of all-cause20, 21 and cardiovascular mortality.20, 22, 23 Thus, CKD patients are more exposed to risk of mortality than to start renal replacement therapy.24, 25 Multiple possible explanations exist for the association between CKD and increased mortality. Reduced kidney function is associated with independent and strong risk factors, such as inflammation, abnormal apolipoprotein levels, elevated plasma homocysteine, enhanced coagulability, anemia, left ventricular hypertrophy, and endothelial dysfunction.22, 24, 25, 26

he association between CKD and increased mortality. Reduced kidney function is associated with independent and strong risk factors, such as inflammation, abnormal apolipoprotein levels, elevated plasma homocysteine, enhanced coagulability, anemia, left ventricular hypertrophy, and endothelial dysfunction.22, 24, 25, 26 The aim of the present study is to ascertain a potential association between the presence and progression of CAC with cardiac events in CKD-patients not on dialysis. This association has never been evaluated before. RESULTS Study population The initial cohort consisted of 188 consecutive asymptomatic CKD patients who fulfilled the inclusion criteria. Seven patients (3.7%) were lost at the follow-up: three moved to another clinic and four withdrew the consent; thus, the final cohort was represented by 181 patients. No racial/ethnic-based differences were present. In the whole-cohort study, kidney diseases were: glomerulonephritis 27% (20.5–33.6, confidence interval, CI), ischemic nephropathy 11% (6.4–15.7, CI), biopsy-proven diabetic nephropathy 8.3% (4.2–12.3, CI), and other diseases 28.8% (19.5–32.4, CI); renal disease was unknown in the remaining cases. Median duration of diabetes was 174 months (48–228, interquartile range, IQR); hypertension was present in 95.6% (92.6–98.6, CI), with a median duration of 72 months (32–130, IQR). Patients were followed-up for a median time of 745 days (403–1078, IQR). During the observation, time-measured creatinine clearance remained stable and no patient required dialysis treatment.

RESULTS Study population The initial cohort consisted of 188 consecutive asymptomatic CKD patients who fulfilled the inclusion criteria. Seven patients (3.7%) were lost at the follow-up: three moved to another clinic and four withdrew the consent; thus, the final cohort was represented by 181 patients. No racial/ethnic-based differences were present. In the whole-cohort study, kidney diseases were: glomerulonephritis 27% (20.5–33.6, confidence interval, CI), ischemic nephropathy 11% (6.4–15.7, CI), biopsy-proven diabetic nephropathy 8.3% (4.2–12.3, CI), and other diseases 28.8% (19.5–32.4, CI); renal disease was unknown in the remaining cases. Median duration of diabetes was 174 months (48–228, interquartile range, IQR); hypertension was present in 95.6% (92.6–98.6, CI), with a median duration of 72 months (32–130, IQR). Patients were followed-up for a median time of 745 days (403–1078, IQR). During the observation, time-measured creatinine clearance remained stable and no patient required dialysis treatment. At baseline, CACs were found in 54.7% (47.4–61–9, CI) of the whole study population. Among males and females, there was no significant difference in incidence of calcification (56.8% (40.8–72.7, CI) and 55.5% (47.4–63.7, CI), respectively) and median CAC score (107 Agatston unit (AU; 36–145.5, IQR) and 168 AU (72.7–341.2, IQR), respectively). Median CAC score was 109 AU (55–220, IQR), 153 AU (37–257, IQR), 177 AU (87–454, IQR), and 168 AU (84–311, IQR) in patients at stages 2, 3, 4, and 5, respectively.

–72.7, CI) and 55.5% (47.4–63.7, CI), respectively) and median CAC score (107 Agatston unit (AU; 36–145.5, IQR) and 168 AU (72.7–341.2, IQR), respectively). Median CAC score was 109 AU (55–220, IQR), 153 AU (37–257, IQR), 177 AU (87–454, IQR), and 168 AU (84–311, IQR) in patients at stages 2, 3, 4, and 5, respectively. Patients were divided in two groups (⩽100 and >100 AU) according to baseline CAC score. The median interval between first multislice computed tomography (MSCT) and cardiac event or end of the study was 689 days (410–922, IQR) and 820 days (380–1178, IQR) in patients with CAC score ⩽100 and >100 AU, respectively. Baseline characteristics of the two groups are reported in Table 1. Patients with CAC score ⩽100 AU (range: 0–98; mean: 14.6±27.3 (s.d.); and median: 0 (0–17, IQR)) were 119; patients with CAC score >100 AU (range: 105–2860; mean: 384.7±453.8 (s.d.); and median: 234 (163–430, IQR)) were 62. CAC score was >1000 AU in four patients of the latter group. Patients with CAC score >100 AU were more likely to be older (P=0.0001) and to have diabetes (28.8% (17.1–40.6, CI) versus 7.6% (2.8–12.4, CI; P<0.0002)), hypertension (100 versus 93.3% (88.8–97.8, CI; P=0.0368)), longer duration of hypertension (median duration: 120 months (48–170, IQR) versus 62 months (24–120, IQR); P<0.005).

ith CAC score >100 AU were more likely to be older (P=0.0001) and to have diabetes (28.8% (17.1–40.6, CI) versus 7.6% (2.8–12.4, CI; P<0.0002)), hypertension (100 versus 93.3% (88.8–97.8, CI; P=0.0368)), longer duration of hypertension (median duration: 120 months (48–170, IQR) versus 62 months (24–120, IQR); P<0.005). No significant differences were observed between other variables of the two groups. In particular, there was no difference in measured creatinine clearance, in variables of mineral metabolism (intact parathyroid hormone, serum calcium, and phosphorus), lipids (total cholesterol, triglycerides, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol), inflammation (homocysteine, fibrinogen, and high-sensitivity C-reactive protein), nutrition (total protein and serum albumin), and anemia (hematocrit). Furthermore, no significant difference was found in therapy with calcium channel blockers, statins, and phosphate binders. During the observation time, 29 patients (16%) had a cardiac event (cardiac death or myocardial infarction). The events occurred more frequently in patients with CAC score >100 AU (27.5% (16.2–38.7, CI)) than in those with CAC score ⩽100 AU (7.6% (2.7–12.4, CI); P=0.0003); in the former group, the hazard risk (HR) for cardiac events was 4.11 (1.77–9.57, CI; P<0.0006).

a cardiac event (cardiac death or myocardial infarction). The events occurred more frequently in patients with CAC score >100 AU (27.5% (16.2–38.7, CI)) than in those with CAC score ⩽100 AU (7.6% (2.7–12.4, CI); P=0.0003); in the former group, the hazard risk (HR) for cardiac events was 4.11 (1.77–9.57, CI; P<0.0006). Association between baseline CAC score and survival is shown in Figure 1. After adjustment for age, diabetes, glomerular filtration rate (GFR) (as 24-h-measured creatinine clearance), and hypertension, baseline CAC score >100 AU was a significant (P=0.0017) predictor of cardiac events. In the whole study population, survival was also analyzed on the basis of the annualized progression of CAC score (absent: 25th percentile; moderate: 25th–75th percentiles; and accelerated: >75th percentile); the results are shown in Figure 2. After adjustment for age, diabetes, GFR (as 24-h-measured creatinine clearance), hypertension, and baseline CAC score, survival was significantly (P<0.0068) worse in patients with accelerated progression; in patients with absent or moderate progression, survival curves overlapped.

e); the results are shown in Figure 2. After adjustment for age, diabetes, GFR (as 24-h-measured creatinine clearance), hypertension, and baseline CAC score, survival was significantly (P<0.0068) worse in patients with accelerated progression; in patients with absent or moderate progression, survival curves overlapped. Finally, survival was assessed in two groups taking in account both baseline value of CAC score and progression. With CAC score ⩽100 AU, HR for cardiac events was 0.41 and 3.26 in patients with absent and accelerated progression, respectively; in contrast, with CAC score >100 AU, HR was 4.72 and 2.97 in patients with absent and accelerated progression, respectively. Therefore, accelerated progression affected survival of patients with a low baseline CAC score by a greater extent compared with those with a high one. The final multiple Cox regression model is shown in Table 2. Among all potential confounders (listed in Table 1), only diabetes reached a P-value <0.2 and entered into multiple Cox regression analysis. In this model, baseline CAC score and progression independently predicted cardiac events. In addition, stratification by both factors (baseline CAC score and progression) pointed out a possible effect modification. Baseline CAC score >100 AU and accelerated progression (>75th percentile) significantly interacted and reduced the risk of outcomes (HR<1). Because of this interaction, HR was 4.2 (as a result of 8.4 × 6.3 × 0.08) in patients with CAC score >100 AU and accelerated progression; therefore, it was lower than that caused by each variable itself.

AU and accelerated progression (>75th percentile) significantly interacted and reduced the risk of outcomes (HR<1). Because of this interaction, HR was 4.2 (as a result of 8.4 × 6.3 × 0.08) in patients with CAC score >100 AU and accelerated progression; therefore, it was lower than that caused by each variable itself. DISCUSSION Large epidemiological studies have reported that CKD patients are more likely to die than to start renal replacement therapy.24, 25 CACs are regarded as markers of severe vasculopathy and strong predictor of cardiovascular mortality in patients on dialysis.3, 4, 5, 26, 27, 28 As CACs are present even at early stages of CKD,10, 11, 15, 16, 17 it is reasonable to hypothesize that CAC may be involved in the high mortality occurring in CKD-patients not on dialysis yet. On this regard, an associations between severity of CAC and all-cause mortality and between presence of CAC and combined outcomes have been evidenced in proteinuric diabetic29 and non-diabetic30 CKD patients, respectively. All-cause mortality and combined outcomes may mask the incidence of cardiac events that are more strictly dependent on CAC. Therefore, in the present study, cardiac events such as cardiac death or myocardial infarction were recorded, and a potential association with the presence and the progression of CAC was evaluated. The impact of CAC progression on mortality has never been evaluated in CKD patients.