Browse the corpus

Retinal microvascular changes have been associated with inflammatory processes, which in turn have been shown to be involved in the pathogenesis of vascular disease (1–3). Serum amyloid A (SAA) is a sensitive indicator of inflammation with an expanded range and kinetics different from those associated with C-reactive protein (CRP) (4). Although levels of SAA and CRP have been shown to be associated with retinal vessel dimensions (2), it is currently unknown whether this association differs between individuals with and without diabetes.

inflammation with an expanded range and kinetics different from those associated with C-reactive protein (CRP) (4). Although levels of SAA and CRP have been shown to be associated with retinal vessel dimensions (2), it is currently unknown whether this association differs between individuals with and without diabetes. RESEARCH DESIGN AND METHODS This cross-sectional analysis was a prespecified substudy at two centers (London, U.K., and Dublin, Ireland) of the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT), a randomized controlled multicenter trial assessing the effect of two antihypertensive regimens on coronary heart disease end points (5–8). Ethics approval was obtained at both study sites, and all participants gave written informed consent. In addition to hypertension, individuals had at least three of the following risk factors: male sex, age >55 years, micro- or macroproteinuria, smoking history, dyslipidemia, family history of premature CHD, electrocardiogram abnormalities, left ventricular hypertrophy, type 2 diabetes, peripheral arterial disease, and previous stroke or transient ischemic attack. Retinal analyses were performed on digital 30-degree images of superior and inferior temporal fields as previously described (9). Arteriolar vessels were assessed up to third-generation branches as prespecified in the protocol. SAA and CRP concentrations were measured on a Dade Behring Nephelometer II (Dade Behring Diagnostic, Marburg, Germany). Coefficients of variation for intra- and interassay precision were <5.2 and <8.5%, respectively (10). Clinical and biochemical parameters of diabetic and nondiabetic patients were compared using Student's ttest; parameters were transformed if nonnormally distributed. Values are given as means ± SD if normally distributed and, otherwise, as median (interquartile range). Multiple linear regression analysis was used to compare retinal parameters between diabetic and nondiabetic individuals and to investigate the associations between SAA or CRP and retinal parameters. Prespecified explanatory variables for all models were age, sex, BMI, smoking status, and randomization to antihypertensive and lipid-lowering treatment in ASCOT. SAA and CRP were categorized by tertiles and analyzed as categorical and ordered factors (2). Statistical analyses were performed using Stata 10.0 (Stata Corporation, College Station, TX).

les for all models were age, sex, BMI, smoking status, and randomization to antihypertensive and lipid-lowering treatment in ASCOT. SAA and CRP were categorized by tertiles and analyzed as categorical and ordered factors (2). Statistical analyses were performed using Stata 10.0 (Stata Corporation, College Station, TX). RESULTS This study included 711 individuals (159 with and 552 without diabetes). Age was similar in diabetic and nondiabetic patients (61.4 ± 8.5 vs. 61.5 ± 7.7 years, respectively; P = 0.86), and the proportion of female participants was comparable (25.8 vs. 21.2%; P = 0.22). Diabetic patients had higher BMI (30.6 ± 5.4 vs. 28.8 ± 4.3 kg/m2; P < 0.001). Systolic blood pressure in diabetic and nondiabetic individuals was 159.1 ± 19.1 vs. 159.5 ± 16.9 mmHg, respectively, (P = 0.78); diastolic blood pressure was 90.4 ± 9.9 vs. 93.8 ± 9.7 mmHg (P < 0.001). Levels of CRP were similar in diabetic and nondiabetic individuals (median 1.69 mg/l [interquartile range 0.86–3.55] vs. 1.52 [0.77–3.39]; P = 0.44), but SAA was significantly higher in diabetic than in nondiabetic individuals (3.15 mg/l [2.05–4.90] vs. 2.65 [1.60–4.60]; P = 0.03). Diabetic individuals had shorter retinal arteriolar vessels than nondiabetic individuals (446.9 ± 103.7 vs. 466.4 ± 126.8 pixels; P = 0.03) with larger diameters (29.3 ± 3.1 vs. 28.3 ± 3.2 pixels; P = 0.001). This resulted in a significantly lower arteriolar length-to-diameter (L:D) ratio in diabetic individuals (12.8 [9.9–15.5] vs. 13.8 [11.2–17.0]; P = 0.001). Arteriolar tortuosity tended to be lower in diabetic than in nondiabetic individuals, but differences were not statistically significant (1.25 × 10−2 [0.63–2.27] vs. 1.48 × 10−2 [0.74–2.80]; P = 0.31).Figure 1A shows the association of SAA with arteriolar L:D ratio in diabetic and nondiabetic individuals. In diabetic patients, the association between SAA and arteriolar L:D ratio was negative (Ptrend = 0.005), whereas in nondiabetic patients, levels of SAA were positively associated with arteriolar L:D ratio (Ptrend = 0.028). Thedifferences between diabetic and nondiabetic patients were confirmed in an interaction test (P = 0.007). The association of SAA and arteriolar tortuosity showed similar findings (Fig. 1B; P = 0.05 for interaction by diabetes status). No consistent association was found for CRP and arteriolar L:D ratio (Fig. 1C), and there was a positive association between CRP and arteriolar tortuosity only for nondiabetic patients (Ptrend = 0.039).

tion of SAA and arteriolar tortuosity showed similar findings (Fig. 1B; P = 0.05 for interaction by diabetes status). No consistent association was found for CRP and arteriolar L:D ratio (Fig. 1C), and there was a positive association between CRP and arteriolar tortuosity only for nondiabetic patients (Ptrend = 0.039). There were no significant associations between venular parameters and either SAA or CRP. Figure 1 Association of SAA and CRP (categorized in tertiles) with arteriolar L:D ratio (A and C) and with arteriolar tortuosity (B and D). Ranges for tertiles 1, 2, and 3 (t1, t2, and t3, respectively) for SAA were 0.6–2.4, 2.5–3.9, and 4.0–92.6 mg/l, respectively, for diabetic individuals and 0.6–2.0, 2.1–3.6, and 3.7–162.0 mg/l for nondiabetic individuals. The corresponding values for CRP were 0.1–1.0, 1.0–2.7, and 2.8–49.8 mg/l for diabetic individuals and 0.1–1.0, 1.0–2.4, and 2.4–65.2 mg/l for nondiabetic individuals. P values for trend are derived from multiple linear regression analysis models adjusted for age, sex, BMI, smoking status, and antihypertensive and lipid-lowering treatment in the ASCOT and represent the association between inflammatory markers and retinal parameters over the entire range of SAA or CRP. *P < 0.05 for comparison with t1.

are derived from multiple linear regression analysis models adjusted for age, sex, BMI, smoking status, and antihypertensive and lipid-lowering treatment in the ASCOT and represent the association between inflammatory markers and retinal parameters over the entire range of SAA or CRP. *P < 0.05 for comparison with t1. CONCLUSIONS Diabetes status has a modifying effect on the association of SAA with retinal arteriolar architecture. Whereas increased levels of SAA were associated with higher L:D ratio and tortuosity in nondiabetic patients, inverse findings were observed in diabetic patients. Interaction tests confirmed that the modifying effect of diabetes status was unlikely to be a chance finding. CRP measurements showed less consistent associations with arteriolar measures according to diabetes status. Previous studies have consistently shown an association of inflammatory markers with retinal microvascular changes but have not reported results according to diabetes status (1–3). In the Beaver Dam Eye Study, increased levels of SAA were associated with smaller arteriolar diameters (2). Because a smaller arteriolar diameter results in an increased L:D ratio, these findings are compatible with those for nondiabetic individuals in the present analysis. It is noteworthy that participants in the Beaver Dam Eye Study mainly consisted of nondiabetic individuals (7.1% with diabetes) and exhibited a lower frequency of cardiovascular risk factors than participants in the ASCOT (2).

these findings are compatible with those for nondiabetic individuals in the present analysis. It is noteworthy that participants in the Beaver Dam Eye Study mainly consisted of nondiabetic individuals (7.1% with diabetes) and exhibited a lower frequency of cardiovascular risk factors than participants in the ASCOT (2). CRP and SAA are classic acute-phase proteins, and their levels are often correlated (11). However, their concentration may differ due to diverse regulation by the cytokine network and differences in clearance rates (12). It is, therefore, not surprising that these two markers may differ in individuals with multiple and different underlying pathologies, such as hypertensive patients with and without type 2 diabetes. In particular, recent studies have suggested that SAA may be a more sensitive indicator of inflammation in cardiovascular and noncardiovascular disease than CRP (13–15), although other studies have not confirmed this (12). The cross-sectional design of this study limits conclusions regarding cause and effect. Although analyses were adjusted for relevant cardiovascular risk factors, this does not exclude the possibility that the analyses did not account for all potential confounders. In conclusion, the present findings suggest the involvement of diverse inflammatory mechanisms for the development of retinal microvascular disease in diabetic individuals compared with that for nondiabetic individuals—a concept that may need further investigation.

nalyses did not account for all potential confounders. In conclusion, the present findings suggest the involvement of diverse inflammatory mechanisms for the development of retinal microvascular disease in diabetic individuals compared with that for nondiabetic individuals—a concept that may need further investigation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Acknowledgments This study was supported by the Swiss National Science Foundation (grant 3233B0-115212 to C.S.). N.C., A.D.H., N.P., and S.T. are grateful for support from the National Institute for Health Research, Biomedical Research Centre funding scheme. R.J.T. is supported by a Sidney Sax fellowship from the National Health and Medical Research Council of Australia (grant 334173). This study was supported by an unrestricted grant from Pfizer. No other potential conflicts of interest relevant to this article were reported. We thank the staff and participants of the ASCOT for their important contribution to the study. We are grateful to M. Pepys from the Centre for Amyloidosis and Acute Phase Proteins, Royal Free and University College Medical School, London, U.K., for important support regarding the biochemical analyses.

ng logarithmic function. Multiple linear regression models were used to assess the relationship of 25(OH)D3 with Si, AIR, disposition index, HOMA-β, and HOMA-IR, respectively. Statistical analyses were performed using SPSS for Windows, version 16.0. P values <0.05 (two-sided) were regarded as statistically significant. RESULTS Serum concentration of 25(OH)D3 was 57.1 ± 26.0 nmol/l (mean ± SD), range 13.7–170.4 nmol/l. Only 91 (20%) subjects had levels ≥75 nmol/l, and a majority (n = 227) had biochemical vitamin D deficiency (<50 nmol/l) (13). Subject characteristics are presented across tertiles of serum 25(OH)D3 concentration (supplemental Table 1, available in an online-only appendix at http://care.diabetesjournals.org/cgi/content/full/dc09-1692/DC1).

s had levels ≥75 nmol/l, and a majority (n = 227) had biochemical vitamin D deficiency (<50 nmol/l) (13). Subject characteristics are presented across tertiles of serum 25(OH)D3 concentration (supplemental Table 1, available in an online-only appendix at http://care.diabetesjournals.org/cgi/content/full/dc09-1692/DC1). In unadjusted analyses, IVGTT-derived parameters did not differ across tertiles of 25(OH)D3, whereas fasting insulin, HOMA-IR, and HOMA-β were significantly different (all P < 0.015), with higher values among subjects in the lower tertile of 25(OH)D3 concentration (supplemental Table 2). Serum levels of 25(OH)D3 correlated negatively with BMI (r = −0.28, P < 0.001), AIR (r = −0.11, P = 0.033), fasting insulin (r = −0.14, P = 0.002), HOMA-IR (r = −0.14, P = 0.003), and HOMA-β (r = −0.15, P = 0.001), but not with Si (r = 0.062, P = 0.21) or disposition index (r = −0.059, P = 0.24). In a multivariate regression analysis including potential covariates (Table 1), serum 25(OH)D3 concentration was a statistically significant predictor of HOMA-IR, HOMA-β, and AIR (P < 0.05) but not of Si or disposition index when adjusting for sex, age, and geographic location. After adding BMI to the regression model, neither HOMA indexes nor AIR were significantly associated with 25(OH)D3 (Table 1). Table 1 Adjusted regression coefficients of 25(OH) vitamin D3 (nmol/l) with parameters of insulin action and secretion

Low serum concentrations of 25-hydroxyvitamin D [25(OH)D] have been linked to disturbances in glucose metabolism (1–3), development of type 2 diabetes (4), and increased risk of the metabolic syndrome (5–7). To explore the associations between serum concentrations of 25(OH)D and glucose metabolism, we evaluated the relationship between 25(OH)D status and insulin secretion and action estimated both by the homeostatic model assessment (HOMA) and intravenous glucose tolerance test (IVGTT) in a large sample of European subjects with the metabolic syndrome. RESEARCH DESIGN AND METHODS Cross-sectional data were obtained from baseline assessment of 446 Caucasian subjects, aged 35–70 years, BMI 20–40 kg/m2, recruited for the LIPGENE study (NCT00429195) performed in eight European countries in 2005 and 2006. All subjects had the metabolic syndrome defined by three or more slightly modified National Cholesterol Education Program (NCEP) Adult Treatment Panel (ATP)-III criteria (8): levels of fasting plasma glucose >5.5 mmol/l, triglycerides ≥1.5 mmol/l, HDL cholesterol <1.0 mmol/l (males) or <1.3 mmol/l (females), blood pressure ≥130/85 mmHg or on blood pressure–lowering medication, and waist circumference >102 cm (males) or >88 cm (females). The study was approved by local ethics committees at each center (Dublin, Reading, Oslo, Marseille, Maastricht, Cordoba, Krakow, and Uppsala) and confirmed to the Declaration of Helsinki. All participants gave written informed consent.

re–lowering medication, and waist circumference >102 cm (males) or >88 cm (females). The study was approved by local ethics committees at each center (Dublin, Reading, Oslo, Marseille, Maastricht, Cordoba, Krakow, and Uppsala) and confirmed to the Declaration of Helsinki. All participants gave written informed consent. A questionnaire was used to assess the level of physical activity (9), smoking habits, alcohol consumption, and demographic data. Anthropometric and blood pressure measurements were recorded according to standard protocols. An insulin-modified IVGTT was performed as described earlier (10). Measures of insulin sensitivity (Si) were obtained using the MINMOD Millennium Program (version 6.02, Richard N. Bergman) (11). The acute insulin response to glucose (AIR) was defined as the incremental area under the curve from 0 to 8 min. Disposition index was calculated as AIR × Si. HOMA indexes (HOMA2, version 2.2.2 http://www.dtu.ox.ac.uk/index.php?maindoc=/homa) were used to assess insulin resistance (HOMA-IR) and β-cell function (HOMA-β) from fasting blood samples (12). Vitamin 25(OH)D2 and 25(OH)D3 were analyzed with high-performance liquid chromatography/mass spectrometry. Only 15 subjects (3%) had measurable concentrations of 25(OH)D2, mean 10.1 nmol/l, range 6.5–24.6 nmol/l. Including 25(OH)D2 in the analyses did not influence the result. All examinations were performed in January/February to avoid seasonal variation. Correlations between parameters were calculated with Pearson's or Spearman's correlation coefficient as appropriate. Non–normally distributed data were transformed using logarithmic function. Multiple linear regression models were used to assess the relationship of 25(OH)D3 with Si, AIR, disposition index, HOMA-β, and HOMA-IR, respectively. Statistical analyses were performed using SPSS for Windows, version 16.0. P values <0.05 (two-sided) were regarded as statistically significant.

In unadjusted analyses, IVGTT-derived parameters did not differ across tertiles of 25(OH)D3, whereas fasting insulin, HOMA-IR, and HOMA-β were significantly different (all P < 0.015), with higher values among subjects in the lower tertile of 25(OH)D3 concentration (supplemental Table 2). Serum levels of 25(OH)D3 correlated negatively with BMI (r = −0.28, P < 0.001), AIR (r = −0.11, P = 0.033), fasting insulin (r = −0.14, P = 0.002), HOMA-IR (r = −0.14, P = 0.003), and HOMA-β (r = −0.15, P = 0.001), but not with Si (r = 0.062, P = 0.21) or disposition index (r = −0.059, P = 0.24). In a multivariate regression analysis including potential covariates (Table 1), serum 25(OH)D3 concentration was a statistically significant predictor of HOMA-IR, HOMA-β, and AIR (P < 0.05) but not of Si or disposition index when adjusting for sex, age, and geographic location. After adding BMI to the regression model, neither HOMA indexes nor AIR were significantly associated with 25(OH)D3 (Table 1). Table 1 Adjusted regression coefficients of 25(OH) vitamin D3 (nmol/l) with parameters of insulin action and secretion Model 1* Model 2† Model 3‡ β SE P β SE P β SE P Si (mU · l−1 · min−1) 0.005 0.003 0.17 0.003 0.003 0.60 0.002 0.003 0.69 AIR (mU · l−1 · min−1) −1.47 0.60 0.041 −1.26 0.60 0.078 −1.20 0.63 0.079 Disposition index −3.23 1.44 0.30 −3.65 1.45 0.17 −3.36 1.53 0.20 HOMA-IR −0.004 0.002 0.016 −0.002 0.002 0.19 −0.002 0.002 0.24 HOMA-β (%) −0.185 0.067 0.007 −0.128 0.066 0.063 −0.113 0.068 0.070 *Model 1: adjusted for age, sex, and geographic location. †Model 2: further adjusted for BMI.

Model 1* Model 2† Model 3‡ β SE P β SE P β SE P Si (mU · l−1 · min−1) 0.005 0.003 0.17 0.003 0.003 0.60 0.002 0.003 0.69 AIR (mU · l−1 · min−1) −1.47 0.60 0.041 −1.26 0.60 0.078 −1.20 0.63 0.079 Disposition index −3.23 1.44 0.30 −3.65 1.45 0.17 −3.36 1.53 0.20 HOMA-IR −0.004 0.002 0.016 −0.002 0.002 0.19 −0.002 0.002 0.24 HOMA-β (%) −0.185 0.067 0.007 −0.128 0.066 0.063 −0.113 0.068 0.070 *Model 1: adjusted for age, sex, and geographic location. †Model 2: further adjusted for BMI. ‡Model 3: further adjusted for education, smoking, alcohol consumption, and use of vitamin supplements. To further explore these relationships, we compared subjects with a severe biochemical vitamin D deficiency (<25 nmol/l, n = 20) to subjects with sufficient vitamin D status (≥75 nmol/l, n = 91). Only BMI was significantly different between groups (P = 0.001), whereas HOMA and IVGTT parameters were not.

‡Model 3: further adjusted for education, smoking, alcohol consumption, and use of vitamin supplements. To further explore these relationships, we compared subjects with a severe biochemical vitamin D deficiency (<25 nmol/l, n = 20) to subjects with sufficient vitamin D status (≥75 nmol/l, n = 91). Only BMI was significantly different between groups (P = 0.001), whereas HOMA and IVGTT parameters were not. CONCLUSIONS We found no significant associations between IVGTT-derived parameters of insulin secretion and action and serum 25(OH)D3 concentrations. At variance with our findings, Chiu et al. (2) observed a positive association between vitamin D status and insulin sensitivity in 126 glucose-tolerant students investigated by hyperglycemic clamp, remaining significant also after adjustment for BMI. The reason for the different results between this study and ours might be the differences in populations or methods used to assess insulin sensitivity. In the former study, there were also inverse relationships between first- and second-phase insulin secretion and serum 25(OH)D concentrations that were not significant after adjusting for covariates, in accordance with our results.

s might be the differences in populations or methods used to assess insulin sensitivity. In the former study, there were also inverse relationships between first- and second-phase insulin secretion and serum 25(OH)D concentrations that were not significant after adjusting for covariates, in accordance with our results. A significant relationship between 25(OH)D and fasting insulin and HOMA-IR has been reported by others (1,14,15). The reason for the differences between these and our results may be that we investigated a more homogeneous group of subjects that all had the metabolic syndrome and hence some degree of insulin resistance. We speculate that vitamin D status may be more closely associated with hepatic insulin sensitivity reflected by fasting glucose and insulin levels than with peripheral insulin sensitivity, as measured by IVGTT. Thus, the link between vitamin D status and tissue-specific insulin action requires further investigation.

e. We speculate that vitamin D status may be more closely associated with hepatic insulin sensitivity reflected by fasting glucose and insulin levels than with peripheral insulin sensitivity, as measured by IVGTT. Thus, the link between vitamin D status and tissue-specific insulin action requires further investigation. Strengths of our study included the use of IVGTT with minimal modeling to assess insulin secretion and insulin action. This extends the knowledge from previous investigations that mostly were based on fasting blood samples. Furthermore, the inclusion of subjects from eight different centers across Europe and limiting the data sampling to 2 months of the year also are advantageous. Limitations of the study were that we only investigated one ethnic group of individuals and that rather few had severe vitamin D deficiency. Also, since the presence of metabolic syndrome was an inclusion criterion for participation in the study, cross-sectional relationships may be attenuated in our population. In conclusion, we found no correlations between vitamin 25(OH)D3 and IVGTT-based estimates of insulin action and secretion in this large sample of subjects with the metabolic syndrome. Prospective and interventional studies using reliable techniques are needed to further elucidate the relation between 25(OH)D and insulin action and secretion.

no correlations between vitamin 25(OH)D3 and IVGTT-based estimates of insulin action and secretion in this large sample of subjects with the metabolic syndrome. Prospective and interventional studies using reliable techniques are needed to further elucidate the relation between 25(OH)D and insulin action and secretion. Supplementary Material Online-Only Appendix The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Acknowledgments The study was supported by LIPGENE—a European Union 6th Framework Program Integrated Project (FOOD-CT-2003-505944); the Norwegian Foundation for Health and Rehabilitation; South-Eastern Norway Regional Health Authority; and Johan Throne Holst Foundation for Nutrition Research. Ciber Physiopathology of Obesity and Nutrition is an initiative of Instituto de Salud Carlos III Government of Spain. No potential conflicts of interest relevant to this article were reported. Parts of this study were presented in abstract form at the 69th Annual Meeting of the American Diabetes Association, New Orleans, Louisiana, 5–9 June 2009; at the 3rd International Congress on Pre-Diabetes and the Metabolic Syndrome, Nice, France, 1–4 April 2009; and at the 45th Annual Meeting of the European Association for the Study of Diabetes, Vienna, Austria, 27 September to 1 October 2009.

Adipokines, inflammatory cytokines secreted from adiopose tissue, have been suggested to play a key role in the development of insulin resistance and diabetes (1). Cathepsin S is a potent cysteine protease that is highly expressed and secreted in adipose tissue of obese individuals (2) and has been suggested to be an important regulator of inflammatory activity (3). We thus hypothesized that cathepsin S levels would be involved in the early dysregulation of glucose and insulin metabolism before development of diabetes. Accordingly, we investigated the association between serum cathepsin S and the two major underlying causes of diabetes—impaired insulin sensitivity and impaired insulin secretion—in a community-based sample of elderly men without diabetes. In secondary analyses, we also investigated the longitudinal association between serum cathepsin S and the incidence of diabetes. RESEARCH DESIGN AND METHODS The design and selection criteria of the Uppsala Longitudinal Study of Adult Men (ULSAM) have been described previously (4), and further details can be found on the Internet (www.pubcare.uu.se/ULSAM/). The present analyses are based on the third examination cycle (baseline 1991–1995; n = 1,221, mean age 71 years) where 1,161 men were free from diabetes. Of these, 905 men had valid measurements of cathepsin S and covariates. Follow-up data on diabetes status at the fourth examination cycle (1998–2002) were available for 597 participants.

are based on the third examination cycle (baseline 1991–1995; n = 1,221, mean age 71 years) where 1,161 men were free from diabetes. Of these, 905 men had valid measurements of cathepsin S and covariates. Follow-up data on diabetes status at the fourth examination cycle (1998–2002) were available for 597 participants. Venous blood samples were drawn at baseline and stored at –70°C until analysis. Serum levels of cathepin S was measured by enzyme-linked immunosorbant assay (human cathepsin S [Total], DY1183, R&D Systems) in frozen samples (mean freezer time 14.6 years [range 12.9–16.7]) (5). Serum levels of high-sensitivity C-reactive protein, interleukin (IL)-6, adiponectin, cystatin C, and triglycerides were performed as previously described (5). Diabetes was diagnosed as fasting plasma glucose ≥7.0 mmol/L (≥126 mg/dL) or use of oral hypoglycemic agents or insulin. The euglycemic-hyperinsulinemic clamp technique according to DeFronzo (6) was used, with a slight modification to suppress hepatic glucose production (7), for estimation of in vivo sensitivity to insulin. The glucose infusion rate during the last hour (M value) was used as the measure of insulin sensitivity. An oral glucose tolerance test (OGTT) was performed, and β-cell function was estimated by the early insulin response: [(insulin30min − insulin0min)/(glucose30min − glucose0min)].

n of in vivo sensitivity to insulin. The glucose infusion rate during the last hour (M value) was used as the measure of insulin sensitivity. An oral glucose tolerance test (OGTT) was performed, and β-cell function was estimated by the early insulin response: [(insulin30min − insulin0min)/(glucose30min − glucose0min)]. Statistical analysis Linear regression analyses were used in separate multivariable models to assess cross-sectional associations between cathepsin S (independent variable) and insulin sensitivity (dependent variable) or insulin secretion (dependent variable) (Table 1). Logistic regression was used to investigate the longitudinal association between cathepsin S and the development of diabetes. Table 1 Cross-sectional associations between cathepsin S, insulin sensitivity, and early insulin response (n = 905) RESULTS Baseline characteristics of the study population are presented in Supplementary Table 1. Higher serum cathepsin S was significantly associated with decreased insulin sensitivity (glucose disposal rate, M) in all multivariable (models A–E, Table 1), but no association was found between cathepsin S and early insulin response. The results were similar in participants with BMI <30 kg/m2 (Table 1). At the follow-up after 6 years (median follow-up 6.5 years [range 4.5–9.2]), 41 participants had developed diabetes. One SD increase in cathepsin S at baseline was associated with a 41–48% risk of developing diabetes in all multivariable models (model D: odds ratio per SD increase 1.48 [95% CI 1.08–2.01], P = 0.01, Supplementary Table 2).

er 6 years (median follow-up 6.5 years [range 4.5–9.2]), 41 participants had developed diabetes. One SD increase in cathepsin S at baseline was associated with a 41–48% risk of developing diabetes in all multivariable models (model D: odds ratio per SD increase 1.48 [95% CI 1.08–2.01], P = 0.01, Supplementary Table 2). CONCLUSIONS Our study suggests that increased cathepsin S levels are involved in the early dysregulation of glucose and insulin metabolism, before the development of diabetes. Comparison with the literature Previous data on the association between circulating cathepsin S and the underlying causes of diabetes are scarce. A small study in women observed no associations between serum cathepsin S and insulin sensitivity, as evaluated by the quantitative insulin-sensitivity check index (QUICKI) (8). However, QUICKI results have limitations as an indicator of insulin sensitivity (9) which may explain the discrepancy with the current study. One study reported increased cathepsin S levels in patients with type 2 diabetes (10). However, the longitudinal association of cathepsin S and diabetes incidence has not been reported previously.

sults have limitations as an indicator of insulin sensitivity (9) which may explain the discrepancy with the current study. One study reported increased cathepsin S levels in patients with type 2 diabetes (10). However, the longitudinal association of cathepsin S and diabetes incidence has not been reported previously. Potential mechanisms Our understanding of the importance of adipose tissue-induced inflammation in the development of insulin resistance and diabetes is increasing (1). Cathepsin S may play a part in this process. Cathepsin S is released by macrophages (11) and participates in the pathophysiologic remodeling of extracellular matrix (12), which leads to adipogenesis and/or adipose cell hyperthrophy (13). This adipose tissue expansion may trigger hypoxia, which in turn results in local low-grade inflammation that has been suggested to be a causal link to insulin resistance (14). Also, cystatin C, the endogenous inhibitor of cathepsin S, has been found to be elevated in obese subjects, both in the circulation and in adipose tissue expression, independently of reduced estimated glomerular filtration rate, which could be a reflection of adipose tissue growth control through cathepsin inhibitions (13). The associations between cathepsin S and insulin sensitivity were independent of adiposity measures, inflammatory markers, and cystatin C in the current study, which would argue against adipose tissue–derived inflammation as the sole mechanistic explanation of our results. Still, we cannot rule out that there may be substantial residual confounding because the adiposity measurements and circulating inflammatory markers used in the current study may both be poor proxies for specific inflammation in adipose tissue. Cathepsin S has also been shown to be associated with triglyceridemia (8) and an increased cardiovascular risk (5), but these mechanisms did not appear to mediate the present associations (model E, Table 1).

ammatory markers used in the current study may both be poor proxies for specific inflammation in adipose tissue. Cathepsin S has also been shown to be associated with triglyceridemia (8) and an increased cardiovascular risk (5), but these mechanisms did not appear to mediate the present associations (model E, Table 1). Clinical implications The development of selective inhibitors of cathepsin S is currently pursued by several pharmaceutical companies (15), but whether cathepsin S inhibitors improve insulin sensitivity or prevent diabetes remains to be established. Limitations Limitations of the study include the unknown generalizability to women and other age and ethnic groups, the large number of participants lost to follow-up, and the modest number of incident diabetes events during follow-up. Also, it is not possible to establish causality with a cross-sectional study design, and there is a risk of reverse causation. Further studies are needed for validation, for exploration of the underlying pathophysiology, and for evaluation of the clinical utility of measuring cathepsin S. This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc12-0494/-/DC1. Acknowledgments This study was supported by the Swedish Research Council (2006-6555), Swedish Heart-Lung Foundation, Thuréus Foundation, Dalarna University, and Uppsala University. The funding sources did not play any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

Council (2006-6555), Swedish Heart-Lung Foundation, Thuréus Foundation, Dalarna University, and Uppsala University. The funding sources did not play any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. No potential conflicts of interest relevant to this article were reported. E.J. wrote the manuscript, researched data, and contributed to discussion. U.R., E.I., and J.S. reviewed the manuscript and contributed to discussion. M.J., E.N., D.I., S.B., and L.L. reviewed the manuscript. A.L. contributed data and reviewed the manuscript. J.Ä. researched data, edited the manuscript, contributed to discussion, and provided funding. J.Ä. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of data analysis.