Browse the corpus

Malawi College of Medicine Research Ethics Committee (Ethics Committee reference no. P.11/12/1308) and the Liverpool School of Tropical Medicine Research Ethics Committee (Ethics Committee reference no. 12.40) approved the CAPS trial protocol that includes this work, a summary of which was published by The Lancet (12). Participants Following community engagement events that included village leaders and other community representatives, a list of all the adults living in each of the 50 villages participating in CAPS in Chikhwawa was obtained from local community health workers known as Health Surveillance Assistants. These lists were collated and used by an independent statistician at the BOLD (Burden of Obstructive Lung Disease) center in London to obtain a population-representative sample of adults >18 years of age with stratification by age and sex. All potential participants sampled in this way were then individually invited to participate with written informed consent (or witnesses thumbprint for those unable to read and write) obtained from those who agreed. People who were acutely unwell, not permanent residents, or pregnant were excluded.

tion by age and sex. All potential participants sampled in this way were then individually invited to participate with written informed consent (or witnesses thumbprint for those unable to read and write) obtained from those who agreed. People who were acutely unwell, not permanent residents, or pregnant were excluded. Procedures Fieldworkers who had undergone study-specific training and met the required quality standards did home visits according to standardized operating procedures. With the exception of the air pollution monitoring procedures that are not part of the BOLD study protocol, all procedures were conducted in accordance with the BOLD study protocol, which has been described previously (13). Minimal demographic information was collected from participants who declined to participate in the full study. Fieldworkers administered BOLD study questionnaires in the local language, Chichewa. Height and weight were measured using a portable stadiometer and scales. All eligible participants were asked to do before and after bronchodilator spirometry, which BOLD center–certified fieldworkers performed to European Respiratory Society/American Thoracic Society guidelines using the ndd EasyOne spirometer (ndd Medical Technologies) (14). Up to three repeat visits were arranged to achieve the required spirometry quality standards. Spirometry data were sent electronically to the BOLD center for quality control.

performed to European Respiratory Society/American Thoracic Society guidelines using the ndd EasyOne spirometer (ndd Medical Technologies) (14). Up to three repeat visits were arranged to achieve the required spirometry quality standards. Spirometry data were sent electronically to the BOLD center for quality control. Personal exposures to particulate matter <2.5 μm in aerodynamic diameter (PM2.5) and carbon monoxide (CO) were measured continuously for 48 hours using the indoor air pollution (IAP) 5000 series monitor (Aprovecho Research Center). The IAP 5000 sampled air from the breathing zone using a short tube and logged continuous PM2.5 and CO using a light-scattering photometer and an electrochemical cell CO sensor, respectively. All monitors were calibrated at the Aprovecho Research Center prior to use in the study. Monitors were worn in small backpacks apart from during sleep, when they were kept beside the sleeping mat or bed.

At a Glance Commentary Scientific Knowledge on the Subject Previous studies have suggested an association between household air pollution from solid fuel use and excess chronic obstructive pulmonary disease (COPD) risk, but the magnitude of the association varied greatly across different studies, and the evidence on respiratory diseases other than COPD in adults has been limited. Whether switching from solid to clean fuels or use of ventilation in adults may have any impact on respiratory hospitalization risk has not been examined in large-scale population-based cohort studies. What This Study Adds to the Field In this cohort study of 280,000 never-smoking Chinese adults, long-term solid fuel use for cooking was associated with significant excess risks of hospitalization and death from both acute and chronic respiratory diseases, including chronic lower respiratory disease and acute lower respiratory tract infection. The excess risk was greater among persistent wood than coal users, but smaller among those who switched from solid to clean fuels or used ventilated cookstoves. An association between solid fuel use and COPD admissions and death was found, but it was far weaker than estimates from meta-analysis of cross-sectional studies for airflow obstruction. This study also provides suggestive evidence that improved ventilation or switching to clean fuels may alleviate the excess respiratory risks associated with solid fuel use.

d COPD admissions and death was found, but it was far weaker than estimates from meta-analysis of cross-sectional studies for airflow obstruction. This study also provides suggestive evidence that improved ventilation or switching to clean fuels may alleviate the excess respiratory risks associated with solid fuel use. Household air pollution (HAP), arising mainly from domestic burning of solid fuels (e.g., coal and biomass) for cooking, is a leading cause of premature death and disease burden worldwide (1). Currently, >2.7 billion individuals, mainly those from rural areas in low- and middle-income countries (LMICs), are regularly exposed to high levels of HAP (2).

ising mainly from domestic burning of solid fuels (e.g., coal and biomass) for cooking, is a leading cause of premature death and disease burden worldwide (1). Currently, >2.7 billion individuals, mainly those from rural areas in low- and middle-income countries (LMICs), are regularly exposed to high levels of HAP (2). The biological plausibility (due to its resemblance to smoking) that solid fuel use is associated with higher risk of chronic obstructive pulmonary disease (COPD) in adults does not have a strong evidence base, as conclusions drawn from previous meta-analyses of studies with relatively small sample sizes were limited by high levels of heterogeneity and publication bias (3–6). In contrast, three out of the four more recent, larger studies have found no evidence of a significant association with airflow obstruction (7–10). There has also been little reliable evidence on the relationship between HAP and hospitalization or death from COPD, which is relevant to the understanding of the public health burden in LMICs such as China, where COPD is often diagnosed based on symptoms (chronic bronchitis) or radiological evidence (emphysema) rather than airflow obstruction, as spirometry is not routinely performed (8, 11). Few studies have investigated the effects of HAP on respiratory diseases other than COPD such as acute lower respiratory infection (ALRI) in adults (12, 13). We report findings on the use of solid fuels for cooking and its association with hospitalization and death from acute and chronic respiratory diseases in ∼280,000 never-smoking Chinese adults from the China Kadoorie Biobank (CKB) study.

her than COPD such as acute lower respiratory infection (ALRI) in adults (12, 13). We report findings on the use of solid fuels for cooking and its association with hospitalization and death from acute and chronic respiratory diseases in ∼280,000 never-smoking Chinese adults from the China Kadoorie Biobank (CKB) study. Methods Study Design Detailed methods of the CKB study have been described previously (14–16). Between 2004 and 2008, 512,000 adults aged 30–79 years were recruited from 10 areas across China (see Figure E1 in the online supplement) and undertook a computer-assisted interview and physical measurements (including spirometry) by trained health workers following standardized procedures (14, 15). The laptop-based questionnaire incorporated stringent logic and error checks to avoid coding errors, and the quality of data collection was closely monitored, with regular feedback and further training provided to health workers (14, 15). Spirometry was performed according to the American Thoracic Society guidelines as described previously (10), but no bronchodilator was administered. Approval was obtained from the Ethical Review Committee of the Chinese Center for Disease Control and Prevention (Beijing, China) and the Oxford Tropical Research Ethics Committee, University of Oxford (Oxford, United Kingdom). Written informed consent was obtained from all participants.

ronchodilator was administered. Approval was obtained from the Ethical Review Committee of the Chinese Center for Disease Control and Prevention (Beijing, China) and the Oxford Tropical Research Ethics Committee, University of Oxford (Oxford, United Kingdom). Written informed consent was obtained from all participants. Assessment of Solid Fuel Use At baseline, each participant was asked to recall, for up to their three most recent residences, how many years they had lived there, cooking frequency (no cooking facility/never/rarely, monthly, weekly, or daily), and ownership of ventilated cookstoves (17). Participants who cooked at least monthly, in each of their respective residences, were asked about the primary fuel type used (electricity, gas, coal, wood, charcoal, or other unspecified). If two or more fuel types were used at a residence, the one used most frequently and for the longest duration was recorded. Clean fuels included electricity or gas, whereas solid fuels comprised coal or wood (including charcoal because of their compositional and emission similarities) (12). Participants cooking weekly or daily were considered as cooking regularly (90% of whom cooked daily at baseline), and their HAP exposure at each residence was classified according to the primary fuel type. Long-term exposure was assessed by grouping participants who used the same primary fuel type throughout their three residences and those who had switched from solid to clean fuels before baseline separately. Long-term solid fuel users were further categorized into three groups (always coal, always wood, and a mixture of coal and wood), along with the estimated duration of continuous exposure to solid fuels for cooking during the recall period (<20, 20–39, and ≥40 yr). To explore the potential impact of ventilated cookstove use, a three-category composite exposure was derived (clean fuels, solid fuels with ventilated cookstoves, and solid fuels without ventilated cookstoves). Further details on exposure assessment are available online (Supplementary Methods section E1).

and ≥40 yr). To explore the potential impact of ventilated cookstove use, a three-category composite exposure was derived (clean fuels, solid fuels with ventilated cookstoves, and solid fuels without ventilated cookstoves). Further details on exposure assessment are available online (Supplementary Methods section E1). Follow-up for Mortality and Morbidity All participants were followed up through electronic linkage, using unique personal identification numbers, to established death and morbidity registries and to a nationwide health insurance system (∼99% coverage in the study areas), which provided coded fatal and nonfatal events (International Classification of Diseases, 10th revision [ICD-10]) (15). The endpoints investigated in this study include the first hospitalization event (during the follow-up period) or death from major respiratory diseases (including chronic lower respiratory disease [CLRD; ICD-10 J40–J47, where J41–J44 were considered as COPD], acute lower respiratory infection [ALRI; J12–J18 and J20–J22], acute upper respiratory infection [AURI; J00–J06], and other upper respiratory disease [J30–J39]) and death from any respiratory diseases (excluding those due to external agents: J00–J47, J80–J94, J96–J99). Participants without the above events were censored upon death, loss to follow-up, or January 1, 2016. To verify the validity of COPD diagnoses, a random sample of ∼1,000 COPD cases (∼10%) between 2004 and 2013 was adjudicated by respiratory physicians independently (18). Only 14% of the COPD cases had pre–bronchodilator spirometry performed. However, most (85%) COPD diagnoses were considered to be adequately supported by different sources of evidence based on clinical symptoms, risk exposure, radiological examinations, or spirometry in accordance with the existing clinical guidelines (18).

14% of the COPD cases had pre–bronchodilator spirometry performed. However, most (85%) COPD diagnoses were considered to be adequately supported by different sources of evidence based on clinical symptoms, risk exposure, radiological examinations, or spirometry in accordance with the existing clinical guidelines (18). Statistical Analysis Our analyses were restricted to never-smokers (n = 317,614), defined as those who had either never smoked or had smoked <100 cigarettes or equivalent during their lifetime. We excluded participants with unreliable recall information on residence duration (n = 1,573) and those with self-reported doctor-diagnosed major chronic diseases (chronic bronchitis, emphysema, tuberculosis, asthma, any cancer, stroke, transient ischemic attack, or coronary heart disease) prior to the baseline survey (n = 26,095). Participants who used other unspecified fuels at any residence (n = 2,527), those who switched from clean to solid fuels (n = 655), or those who had cooked previously but stopped at baseline (n = 8,926) were also excluded, leaving 277,838 participants in the final study population.

to the baseline survey (n = 26,095). Participants who used other unspecified fuels at any residence (n = 2,527), those who switched from clean to solid fuels (n = 655), or those who had cooked previously but stopped at baseline (n = 8,926) were also excluded, leaving 277,838 participants in the final study population. Direct standardization yielded age-, sex-, and study area–adjusted percentages or means of baseline characteristics for long-term cooking fuel exposure categories. We used Cox regression to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for first hospitalization or death from respiratory disease in association with long-term solid fuel use for cooking (referred to as risk of respiratory disease in the subsequent text), stratifying for age-at-risk (5-yr intervals), sex, and study area (10 areas), and adjusted for education (no formal school, primary school, middle school, or high school/college/university), household income (<10,000, 10,000–19,999, 20,000–34,999, or ≥35,000 yuan), occupation (agricultural worker, factory worker, non-manual worker, or others), alcohol consumption (never/rarely, occasional, ex-drinker or reduced intake, or weekly regular), body mass index (BMI; continuous), environmental tobacco smoke (ETS) exposure (<1 d/wk, 1–5 d/wk, or daily or almost every day), cookstove ventilation (all stoves, some stoves, or none), primary heating fuel exposure (always clean fuels, solid to clean fuel, always solid fuels, or others), and length of recall period (continuous), where appropriate. Fuller details of the selection process used for confounders for adjustment are provided online (Supplementary Methods section E2). The proportional hazard assumption was confirmed to be upheld by using standard methods (19). For exposure measures with more than two categories, a group-specific CI of the HR was calculated from the variance of the log hazard in each category (including the reference category) as described previously (16, 20), and more details are provided online (Supplementary Methods section E3). The cumulative probability of being hospitalized or dying from each specific cause during follow-up is presented using Kaplan-Meier plots.

ance of the log hazard in each category (including the reference category) as described previously (16, 20), and more details are provided online (Supplementary Methods section E3). The cumulative probability of being hospitalized or dying from each specific cause during follow-up is presented using Kaplan-Meier plots. We conducted subgroup analyses by baseline characteristics (birth year, age, sex, education, ETS, alcohol consumption, BMI, leg length, years of having a refrigerator at home [the latter two are proxies for the early life environment]). We performed further sensitivity analyses to reduce the potential impact of reverse causation and residual confounding by excluding 1) participants with <20 years of recall period (“frequent movers,” n = 26,742), 2) participants with poor self-reported health at baseline (n = 26,551), 3) participants who cooked weekly at baseline (n = 25,466), and 4) individuals with spirometry-defined airflow obstruction (n = 15,879) or chronic respiratory symptoms (n = 4,842) at baseline, respectively. Details of the assessment and definitions of airflow obstruction and chronic respiratory symptoms are available online (Supplementary Methods section E4). All analyses were conducted using SAS software version 9.3.

d airflow obstruction (n = 15,879) or chronic respiratory symptoms (n = 4,842) at baseline, respectively. Details of the assessment and definitions of airflow obstruction and chronic respiratory symptoms are available online (Supplementary Methods section E4). All analyses were conducted using SAS software version 9.3. Results Among the 277,838 never-smoking participants, the mean (SD) age was 50.3 (10.3) years and 91% were female. The mean total duration of the three most recent residences was 39.7 (14.5) years, with 91% participants having had at least 20 years of residence covered. Among 91% who reported regular cooking during the recall period, 52% used solid fuel throughout. Compared with long-term clean fuel users, solid fuel users were older, less educated, had lower income, were more likely to live in rural areas and to report poor general health status, and were less likely to use ventilated cookstoves. There was no major difference in BMI or exposure to ETS between the two groups (Table 1). Table 1. Baseline Characteristics of Never-Smoking Participants by Long-Term Primary Cooking Fuel Exposure

Results Among the 277,838 never-smoking participants, the mean (SD) age was 50.3 (10.3) years and 91% were female. The mean total duration of the three most recent residences was 39.7 (14.5) years, with 91% participants having had at least 20 years of residence covered. Among 91% who reported regular cooking during the recall period, 52% used solid fuel throughout. Compared with long-term clean fuel users, solid fuel users were older, less educated, had lower income, were more likely to live in rural areas and to report poor general health status, and were less likely to use ventilated cookstoves. There was no major difference in BMI or exposure to ETS between the two groups (Table 1). Table 1. Baseline Characteristics of Never-Smoking Participants by Long-Term Primary Cooking Fuel Exposure Characteristic Always Clean Solid to Clean Always Solid Never Cooked Regularly All Participants n 53,130 66,115 131,270 27,323 277,838 Age, yr, mean (SD) 45.3 (9.5) 50.9 (9.8) 53.0 (10.2) 45.6 (11.2) 50.3 (10.3) Female sex, % 86.8 97.0 95.5 40.7 90.9 Urban residence, % 88.0 79.2 8.5 49.8 44.3 No formal education, % 14.5 18.8 28.7 20.0 23.6 Household income <10,000 yuan/yr, % 18.3 20.4 37.8 22.6 28.6 Occupation, % Agricultural worker 19.7 26.6 48.4 31.0 41.3 Factory worker 13.1 12.1 11.1 15.9 12.0 Non–manual worker 17.9 13.9 6.6 16.2 9.9 Others* 49.3 47.4 34.0 36.8 36.9 Current drinker in males, % 21.3 21.5 18.5 19.6 19.1 Current drinker in females, % 2.0 1.7 1.5 2.0 1.6 Environmental tobacco smoke, % <1 d/wk 44.9 39.6 39.4 41.9 40.5 1–5 d/wk 17.8 19.1 18.8 17.3 19.0 Daily or almost every day 37.3 41.4 41.8 40.8 40.4 Cookstove ventilation, % All stoves 61.1 55.8 22.8 47.9 44.7 Some stoves 19.7 24.4 46.5 28.3 31.9 None 19.2 19.9 30.7 23.8 23.5 Body mass index, kg/m2, mean (SD) 23.8 (3.3) 24.2 (3.4) 23.6 (3.4) 23.7 (3.2) 23.8 (3.4) Systolic blood pressure, mm Hg, mean (SD) 127.9 (19.9) 128.7 (21.4) 130.2 (22.2) 128.4 (20.3) 129.7 (21.6) Self-reported poor health, % 8.3 8.2 10.4 9.7 9.1 Means and percentages were adjusted for age, sex, and study area when appropriate. Participants who switched from clean to solid fuels, used unspecified fuels, or cooked regularly but stopped were excluded from analysis (n = 12,108).

0.2 (22.2) 128.4 (20.3) 129.7 (21.6) Self-reported poor health, % 8.3 8.2 10.4 9.7 9.1 Means and percentages were adjusted for age, sex, and study area when appropriate. Participants who switched from clean to solid fuels, used unspecified fuels, or cooked regularly but stopped were excluded from analysis (n = 12,108). * “Others” in occupation include housewife/husband, retired, self-employed, unemployed, or other unspecified.

0.2 (22.2) 128.4 (20.3) 129.7 (21.6) Self-reported poor health, % 8.3 8.2 10.4 9.7 9.1 Means and percentages were adjusted for age, sex, and study area when appropriate. Participants who switched from clean to solid fuels, used unspecified fuels, or cooked regularly but stopped were excluded from analysis (n = 12,108). * “Others” in occupation include housewife/husband, retired, self-employed, unemployed, or other unspecified. During 2.6 million person-years of follow-up (mean, 9.1 [1.4] yr), 19,823 first hospitalization events and deaths from major respiratory diseases were recorded, including 10,553 CLRD, 4,398 COPD, 7,324 ALRI, and 3,011 AURI. Figure 1 presents the Kaplan-Meier probability of hospitalization or death from each cause-specific outcome across the three main exposure categories (always clean, solid to clean, or always solid). Compared with long-term clean fuel use, long-term solid fuel use for cooking was associated with higher risks of several major respiratory diseases, with adjusted HRs of 1.36 (group-specific 95% CI, 1.32–1.40) for all major respiratory diseases, 1.47 (1.41–1.52) for CLRD, 1.10 (1.03–1.18) for COPD, 1.16 (1.09–1.23) for ALRI, 1.59 (1.48–1.71) for AURI, 1.56 (1.40–1.73) for other upper respiratory disease, and 1.56 (1.28–1.89) for respiratory death. The HRs were significantly weaker in participants who switched from solid to clean fuels than those who used solid fuels persistently (for major respiratory disease, 1.14 [1.10–1.17] vs. 1.36 [1.32–1.40]) (Table 2). For major respiratory diseases, the corresponding HR was similar in men and women (1.46 [1.30–1.63] vs. 1.37 [1.32–1.41]) and across a range of baseline characteristics (Table E1).

clean fuels than those who used solid fuels persistently (for major respiratory disease, 1.14 [1.10–1.17] vs. 1.36 [1.32–1.40]) (Table 2). For major respiratory diseases, the corresponding HR was similar in men and women (1.46 [1.30–1.63] vs. 1.37 [1.32–1.41]) and across a range of baseline characteristics (Table E1). Figure 1. Kaplan-Meier probabilities of developing specific respiratory diseases during follow-up. Table 2. Incidence Rates and Adjusted Hazard Ratios for Hospitalization or Death from Major Respiratory Diseases by Long-Term Cooking Fuel Exposure

clean fuels than those who used solid fuels persistently (for major respiratory disease, 1.14 [1.10–1.17] vs. 1.36 [1.32–1.40]) (Table 2). For major respiratory diseases, the corresponding HR was similar in men and women (1.46 [1.30–1.63] vs. 1.37 [1.32–1.41]) and across a range of baseline characteristics (Table E1). Figure 1. Kaplan-Meier probabilities of developing specific respiratory diseases during follow-up. Table 2. Incidence Rates and Adjusted Hazard Ratios for Hospitalization or Death from Major Respiratory Diseases by Long-Term Cooking Fuel Exposure Number of Events Rate (per 100,000 Person-Years)* HR (95% CI)† Major respiratory diseases‡ Always clean 2,576 797 1.00 (0.96–1.04) Solid to clean 4,575 891 1.14 (1.10–1.17) Always solid 12,672 1,088 1.36 (1.32–1.40) Chronic lower respiratory disease§ Always clean 1,093 371 1.00 (0.94–1.07) Solid to clean 2,271 444 1.20 (1.15–1.26) Always solid 7,189 619 1.47 (1.41–1.52) Chronic obstructive pulmonary disease|| Always clean 357 192 1.00 (0.89–1.12) Solid to clean 778 167 0.96 (0.89–1.03) Always solid 3,263 222 1.10 (1.03–1.18) Acute lower respiratory infection¶ Always clean 1,037 344 1.00 (0.93–1.07) Solid to clean 1,871 308 1.08 (1.02–1.13) Always solid 4,416 328 1.16 (1.09–1.23) Acute upper respiratory infection** Always clean 444 108 1.00 (0.90–1.11) Solid to clean 584 149 1.13 (1.04–1.23) Always solid 1,983 194 1.59 (1.48–1.71) Other upper respiratory disease†† Always clean 327 75 1.00 (0.89–1.13) Solid to clean 424 70 1.10 (0.99–1.22) Always solid 984 113 1.56 (1.40–1.73) Respiratory death‡‡ Always clean 51 17 1.00 (0.75–1.33) Solid to clean 126 14 0.96 (0.78–1.19) Always solid 457 38 1.56 (1.28–1.89) Definition of abbreviations: CI = confidence interval; HR = hazard ratio; ICD-10 = International Classification of Diseases, 10th revision.

0.99–1.22) Always solid 984 113 1.56 (1.40–1.73) Respiratory death‡‡ Always clean 51 17 1.00 (0.75–1.33) Solid to clean 126 14 0.96 (0.78–1.19) Always solid 457 38 1.56 (1.28–1.89) Definition of abbreviations: CI = confidence interval; HR = hazard ratio; ICD-10 = International Classification of Diseases, 10th revision. * Event rates were adjusted for age, sex, and study area structure of the China Kadoorie Biobank study population. † Hazard ratios were stratified for age at risk, sex, and study area and adjusted for education, household income, occupation, alcohol consumption, body mass index, environmental tobacco smoke, cookstove ventilation, heating fuel, and length of recall period. ‡ ICD-10 codes J00–J06, J12–J18, J30–J22, J30–J39, and J40–J47. § ICD-10 codes J40–J47. || ICD-10 codes J41–J44. ¶ ICD-10 codes J12–J18 and J20–J22. ** ICD-10 codes J00–J06. †† ICD-10 codes J30–J39. ‡‡ ICD-10 codes J00–J47, J80–J94, and J96–J99.

† Hazard ratios were stratified for age at risk, sex, and study area and adjusted for education, household income, occupation, alcohol consumption, body mass index, environmental tobacco smoke, cookstove ventilation, heating fuel, and length of recall period. ‡ ICD-10 codes J00–J06, J12–J18, J30–J22, J30–J39, and J40–J47. § ICD-10 codes J40–J47. || ICD-10 codes J41–J44. ¶ ICD-10 codes J12–J18 and J20–J22. ** ICD-10 codes J00–J06. †† ICD-10 codes J30–J39. ‡‡ ICD-10 codes J00–J47, J80–J94, and J96–J99. Compared with participants who had always used clean fuels for cooking, the risk of major respiratory diseases increased with duration of persistent solid fuel use, with HRs of 1.32 (1.26–1.39), 1.41 (1.37–1.45), and 1.54 (1.48–1.60) in those who used solid fuels for <20, 20–39, and ≥40 years, respectively (Ptrend < 0.0001). Similar relationships were observed for each specific respiratory disease (Ptrend ≤ 0.003 for all comparisons) (Figure 2). Among long-term solid fuel users for cooking, those who used wood had higher HRs for major respiratory diseases than did those who used coal (1.37 [1.33–1.41] vs. 1.22 [1.15–1.29]), and those who switched between wood and coal had an intermediate risk (1.25 [1.19–1.31]). Similar patterns of association were observed for CLRD, COPD, ALRI, and respiratory death but not for other respiratory disease outcomes (Figure 3). Excess risk of major respiratory diseases among the solid fuel users with ventilated cookstoves were significantly lower compared with those who used unventilated cookstoves (1.22 [1.19–1.25] vs. 1.29 [1.24–1.35]). Similar associations were observed for CLRD, AURI, other upper respiratory disease, and respiratory death (Figure 4).

of major respiratory diseases among the solid fuel users with ventilated cookstoves were significantly lower compared with those who used unventilated cookstoves (1.22 [1.19–1.25] vs. 1.29 [1.24–1.35]). Similar associations were observed for CLRD, AURI, other upper respiratory disease, and respiratory death (Figure 4). Figure 2. Adjusted hazard ratios for major respiratory diseases by duration of continuous exposure to solid cooking fuel in never-smokers. Hazard ratios were stratified by age at risk (in 5-yr groups), sex, and study area and were adjusted for education, household income, occupation, alcohol consumption, body mass index, environmental tobacco smoke, cookstove ventilation, primary heating fuel exposure, and length of recall period. The black boxes represent hazard ratios, with the size inversely proportional to the variance of the logarithm of the hazard ratio, and the horizontal lines represent 95% confidence intervals. CI = confidence interval; COPD = chronic obstructive pulmonary disease; HR = hazard ratio. Figure 3. Adjusted hazard ratios for major respiratory diseases by type of primary cooking fuel used in never-smokers. Conventions are as in Figure 2. For definition of abbreviations, see Figure 2. Figure 4. Adjusted hazard ratios of major respiratory diseases associated with primary cooking fuel and use of ventilated cookstoves at baseline. Conventions are as in Figure 2 except that the hazard ratios were not adjusted for cookstove ventilation and length of recall period. For definition of abbreviations, see Figure 2.

4. Adjusted hazard ratios of major respiratory diseases associated with primary cooking fuel and use of ventilated cookstoves at baseline. Conventions are as in Figure 2 except that the hazard ratios were not adjusted for cookstove ventilation and length of recall period. For definition of abbreviations, see Figure 2. The strength of observed associations between solid fuel use for cooking and most respiratory diseases did not change substantially after excluding frequent movers, participants with poor self-reported health, those who cooked weekly, or those who had signs of airflow obstruction or chronic respiratory symptoms at baseline (Table E2). Discussion In this large study of 280,000 never-smoking Chinese adults who had no known prior history of major chronic diseases at baseline, long-term use of solid fuels for cooking was associated with significant elevated risks of hospitalization or death from both acute and chronic respiratory diseases, with consistent results in men and women and across a range of population subgroups. The excess risks appeared to be greater among those who used wood compared with coal. Switching from solid to clean fuels or use of ventilated cookstoves was associated with relatively smaller excess risks.

hronic respiratory diseases, with consistent results in men and women and across a range of population subgroups. The excess risks appeared to be greater among those who used wood compared with coal. Switching from solid to clean fuels or use of ventilated cookstoves was associated with relatively smaller excess risks. Most previous epidemiological studies on solid fuel use and respiratory diseases focused on COPD in adults, with most of them being cross-sectional or case-control studies examining airflow obstruction as the outcome (3–6, 9, 21, 22). Earlier pooled analyses of these studies, often with small sample sizes, reported large excess risks (summary odds ratios from 1.94 to 2.80) (3–6), but strong evidence of publication bias (P < 0.007) and high levels of heterogeneity (I2 = 85%) has been found (5). Four larger and more recent population-based cross-sectional studies involving 13,000 to 67,000 participants, including two conducted in China, reported much weaker associations (from no association to ∼40% excess risk) with airflow obstruction (7–9, 22). In contrast, the present study of 280,000 Chinese never-smokers found that long-term use of solid fuel for cooking was associated with ∼10% excess risk of COPD hospitalization or death. The cohort design of this study enabled us to take account of the influence of reverse causation by excluding those with a prior history of major respiratory diseases, signs of airflow obstruction, or chronic respiratory symptoms, and by examining prospectively recorded hospitalizations or deaths. Furthermore, our analyses were restricted to never-smokers, so the residual confounding from smoking, a leading cause of COPD, should be minimized.

or history of major respiratory diseases, signs of airflow obstruction, or chronic respiratory symptoms, and by examining prospectively recorded hospitalizations or deaths. Furthermore, our analyses were restricted to never-smokers, so the residual confounding from smoking, a leading cause of COPD, should be minimized. Many previous studies on COPD, including a previous cross-sectional analysis of CKB (10), examined spirometry-defined airflow obstruction, the hallmark of COPD, as the outcome. In the present study we focused on hospitalization and death, as there has been little information on the risk of respiratory hospitalizations and deaths associated with long-term HAP. Indeed, the low utility of spirometry for diagnosing COPD in China (7–10%) (8, 23) means many asymptomatic and mild airflow obstruction cases not requiring medical attention were less likely to have been identified, diagnosed, and captured in our records as COPD. Underdiagnosis of COPD is disproportionately higher in rural China (8), where solid fuel use is more prominent. The higher likelihood of undiagnosed cases in the exposed group means that the observed risks for COPD may well be diluted. In this regard, we observed a stronger association between long-term solid fuel use for cooking and CLRD (HR, 1.47 [95% CI, 1.41–1.52]), which included all COPD cases plus mostly unspecified bronchitis (ICD-10 J40; n = 7,471). It is possible that many of these unspecified bronchitis cases (but not acute bronchitis as included within ALRI) could be mild, early stages COPD or acute exacerbations of preexisting, but previously undetected, COPD, given that spirometry is rarely used for diagnosis in China. Nevertheless, this may also suggest that solid fuel use is more strongly associated with chronic bronchitis (or mucus hypersecretion in general) than with emphysema or other COPD phenotypes, which has been suggested in previous studies (6, 9, 24).

cted, COPD, given that spirometry is rarely used for diagnosis in China. Nevertheless, this may also suggest that solid fuel use is more strongly associated with chronic bronchitis (or mucus hypersecretion in general) than with emphysema or other COPD phenotypes, which has been suggested in previous studies (6, 9, 24). For non-COPD respiratory diseases, previous evidence has been more limited. Two small cohort studies on respiratory death (with 155 cases) and ALRI (with 229 participants, no case numbers were given) reported inconclusive findings (25, 26). A recent systematic review (13) of eight relevant studies on ALRI, most of which involved <1,000 disease events, found no consistent evidence. Our study included much larger numbers of events than all previous studies combined (∼7,300 ALRI, 3,000 AURI). We found strong evidence that long-term solid fuel use is associated with significantly elevated risk of hospitalizations or deaths from ALRI and AURI in adults. This highlights the potential need of considering adult ALRI when assessing the disease burden related to HAP exposure. It is worth noting that ALRI and AURI are acute recurring conditions. The observed associations reflect an overall shorter time to the first documented infection during the follow-up in solid fuel users, which may indirectly imply a higher rate of recurrent infection among them. Future analysis focusing on recurrent events (including acute exacerbations of COPD) should be able to clarify this.

served associations reflect an overall shorter time to the first documented infection during the follow-up in solid fuel users, which may indirectly imply a higher rate of recurrent infection among them. Future analysis focusing on recurrent events (including acute exacerbations of COPD) should be able to clarify this. Most previous studies on COPD have examined biomass (mostly wood) only, whereas we analyzed both coal and wood (combined as “solid fuels” and separately), the latter of which has been linked to higher levels of particulate pollution and possibly higher risk of COPD (6, 12). Consistently, the risks of CLRD, COPD, and ALRI in our study were higher among those that persistently used wood compared with those using coal. However, an earlier cross-sectional analysis of CKB on the prevalence of airflow obstruction found seemingly protective effects of wood burning (OR, 0.91 [95% CI, 0.86–0.98]) and a deleterious effect of coal use (1.10 [1.02–1.20]) at baseline in women (10). The two studies differ importantly by the disease outcome examined (prevalence of spirometry-detected airflow obstruction [10] vs. rate of clinical episodes of COPD), as well as inclusion criteria, exposure classification, and analysis strategy. In the current study participants with any prior chronic diseases were excluded. We classified individuals who cooked weekly or daily as regular users of fuels (clean or solid), whereas the previous analysis included also less frequent (monthly) cooks (who were more likely to be men, factory workers, and clean fuel users compared with the more frequent cooks). Furthermore, the current study has additionally adjusted for other important confounders that were not taken into account in the previous study (e.g., ETS, occupation, BMI). For upper respiratory disease, the excess risks appeared to be broadly similar in the long-term wood and coal users for reasons that are not fully understood. It is possible that the etiology or mechanisms between chronic respiratory disease and respiratory infections in relation to air pollutants generated by burning of different fuel types may differ. Further investigations including direct measurement of HAP and characterization of smoke constituents are planned and should help to clarify our findings.

y or mechanisms between chronic respiratory disease and respiratory infections in relation to air pollutants generated by burning of different fuel types may differ. Further investigations including direct measurement of HAP and characterization of smoke constituents are planned and should help to clarify our findings. It has been reported in both observational and intervention studies that HAP exposure and acute respiratory symptoms in adults may be reduced through adequately maintained cookstove ventilation (27). However, there has been no clear evidence on the long-term respiratory benefits of improved cookstove ventilation in adults (27). A retrospective cohort study involving 42,000 Chinese adults reported significantly lower risks of pneumonia mortality (225 cases) and self-reported physician diagnosis of COPD (1,487 cases) in lifelong coal users for cooking who adopted a ventilated cookstove compared with those who did not (28, 29). In contrast, another cohort study of 600 Chinese adults (74 cases) found no significant effect of improved ventilation on the risk of airflow obstruction (30). In our study, solid fuel users who used ventilated cookstoves had lower risks of CLRD and upper respiratory diseases, but not ALRI, COPD, or respiratory death, compared with those who used unventilated cookstoves. This is in agreement with existing evidence that improved ventilation generally may have more prominent benefits on mild, acute conditions but not on more severe diseases such as COPD or ALRI, possibly because the HAP levels after improvement remain substantially above the recommended threshold (27, 31). The discrepancy in the results on CLRD and COPD, as discussed above, may be related to the unspecified bronchitis (ICD-10: J40) that could be acute exacerbation of early stages of COPD. Future large-scale randomized controlled trials with long follow-up and appropriately designed interventions are needed to assess the effect of using ventilated cookstoves on major respiratory conditions such as ALRI or COPD in adults.

d bronchitis (ICD-10: J40) that could be acute exacerbation of early stages of COPD. Future large-scale randomized controlled trials with long follow-up and appropriately designed interventions are needed to assess the effect of using ventilated cookstoves on major respiratory conditions such as ALRI or COPD in adults. Compared with the long-term persistent solid fuel users, participants who had switched their primary cooking fuel from solid to clean fuels prior to the baseline survey had smaller excess risks of all respiratory diseases studied. Although limited, there is consistent trial evidence that switching from solid to clean fuels is associated with markedly greater HAP reduction than adopting improved ventilation (32). Our findings offer supportive evidence that clean fuel adoption may be beneficial for the prevention of acute and chronic respiratory conditions. Although this might seem intuitive, it highlights that the elevated risks associated with historical solid fuel use may still be attenuated by switching to clean fuels later in life, a phenomenon similar to that of smoking cessation (16). This should encourage greater efforts to facilitate universal access to clean energy especially in LMICs, as promoted in the United Nations Sustainable Development Goal 7 (33).

rical solid fuel use may still be attenuated by switching to clean fuels later in life, a phenomenon similar to that of smoking cessation (16). This should encourage greater efforts to facilitate universal access to clean energy especially in LMICs, as promoted in the United Nations Sustainable Development Goal 7 (33). The key strengths of this study lie in the large number of never-smokers, comprehensive investigation of prospectively documented hospitalization and death of a range of respiratory diseases, and the high consistency of exposure–outcome relationships across these diseases and across different population subgroups. Moreover, two common limitations of previous research on this topic, namely reverse causality and residual confounding from smoking, were carefully dealt with in this study. However, our study has several limitations that need to be taken into consideration. First, our outcome was based on linkages to hospitalization records and death certificates. Misclassification due to misdiagnosis is possible, especially for COPD owing to the low utility of spirometry in China. Although we have excluded participants with preexisting chronic diseases, admissions for COPD were unlikely to represent new onset “incident” cases, as COPD has a prolonged development period with risk factors that could trace back to preconception, meaning that it is difficult to establish temporality accurately. Nevertheless, the aim of this study was to investigate whether HAP may be associated with respiratory admissions and deaths, rather than the development of incident cases. We have also excluded those with signs of airflow obstruction at baseline or poor self-reported health in the sensitivity analyses, and the results persisted. Second, HAP exposure was estimated by self-reports of the main type of fuel used as in many other previous studies. It is possible that historical or concurrent exposure to solid fuel emission from secondary or neighborhood fuels could have elevated the background risks of clean fuel users, but we lack data on these, or from direct exposure measurement to more accurately assess exposure–response relationships. Third, instead of prospectively monitoring lifetime exposure, we were only able to estimate long-term exposure based on recall information on the three most recent residences of our participants. This might have resulted in misclassification, especially among clean fuel users who might have used solid fuels in their early life.

stead of prospectively monitoring lifetime exposure, we were only able to estimate long-term exposure based on recall information on the three most recent residences of our participants. This might have resulted in misclassification, especially among clean fuel users who might have used solid fuels in their early life. However, the recall period covered was on average 40 years (≥70% of the adulthood in 80% of participants), and the exclusion of participants with <20 years of recall information provided gave similar findings with all participants included. Fourth, residual confounding from early-life exposure and ETS is possible owing to the lack of direct early-life exposure data and the relatively crude adjustment on ETS (based on self-reported frequency of exposure). Nonetheless, the associations observed were consistent across subgroups defined by proxies of early-life exposures (leg length, education level, years of having a refrigerator at home), and additional adjustment for duration of exposure to ETS did not alter the relationship of interest (data not shown). Finally, our study sample has an imbalanced sex ratio (9:1), and one may argue that the findings may not be generalizable to men. However, in the sex-specific analyses (with >26,000 men), we found no evidence of heterogeneity.

for duration of exposure to ETS did not alter the relationship of interest (data not shown). Finally, our study sample has an imbalanced sex ratio (9:1), and one may argue that the findings may not be generalizable to men. However, in the sex-specific analyses (with >26,000 men), we found no evidence of heterogeneity. In conclusion, in Chinese adults, solid fuel use for cooking was associated with higher risks of admissions and death for both acute and chronic respiratory diseases, with the excess risk seemingly greater for wood than coal users, especially for CLRD, and in those with more prolonged use. A much weaker association with COPD was observed as compared with the earlier meta-analysis estimates used in global disease burden estimation. Moreover, use of ventilated cookstoves and switching to clean fuels were associated with smaller excess risks of some respiratory diseases associated with solid fuel use, reinforcing the need for strengthening the existing global initiatives to improve access to clean energy and to distribute improved cookstoves in communities where a complete switch to cleaner fuels is not yet feasible.

re associated with smaller excess risks of some respiratory diseases associated with solid fuel use, reinforcing the need for strengthening the existing global initiatives to improve access to clean energy and to distribute improved cookstoves in communities where a complete switch to cleaner fuels is not yet feasible. Acknowledgment The authors thank, as our chief acknowledgment, the participants, the project staff, and the China National Center for Disease Control and Prevention (CDC) and its regional offices for assisting with the fieldwork. We thank Judith Mackay in Hong Kong; Yu Wang, Gonghuan Yang, Zhengfu Qiang, Lin Feng, Maigeng Zhou, Wenhua Zhao, and Yan Zhang in China CDC; Lingzhi Kong, Xiucheng Yu, and Kun Li in the Chinese Ministry of Health; and Sarah Clark, Martin Radley, Mike Hill, Hongchao Pan in the Clinical Trial Service Unit and Epidemiological Studies Unit, Oxford, for assisting with the design, planning, organization, and conduct of the study. All figures of this manuscript were created using the Jasper package developed by Matthew Arnold in the CTSU.

Sarah Clark, Martin Radley, Mike Hill, Hongchao Pan in the Clinical Trial Service Unit and Epidemiological Studies Unit, Oxford, for assisting with the design, planning, organization, and conduct of the study. All figures of this manuscript were created using the Jasper package developed by Matthew Arnold in the CTSU. Members of the China Kadoorie Biobank Collaborative Group: International Steering Committee: Junshi Chen, Zhengming Chen (principal investigator), Robert Clarke, Rory Collins, Yu Guo, Liming Li (principal investigator), Jun Lv, Richard Peto, and Robin Walters. International Coordinating Centre, Oxford: Daniel Avery, Derrick Bennett, Ruth Boxall, Yumei Chang, Yiping Chen, Zhengming Chen, Robert Clarke, Huaidong Du, Simon Gilbert, Alex Hacker, Michael Holmes, Christiana Kartsonaki, Rene Kerosi, Garry Lancaster, Kuang Lin, John McDonnell, Iona Millwood, Qunhua Nie, Jayakrishnan Radhakrishnan, Paul Ryder, Sam Sansome, Dan Schmidt, Rajani Sohoni, Becky Stevens, Iain Turnbull, Robin Walters, Jenny Wang, Lin Wang, Neil Wright, Ling Yang, and Xiaoming Yang. National Coordinating Center, Beijing: Zheng Bian, Yu Guo, Xiao Han, Can Hou, Jun Lv, Pei Pei, Chao Liu, Biao Jing, Yunlong Tan, and Canqing Yu. Ten Regional Coordinating Centers: Qingdao China National Center for Disease Control and Prevention (CDC)—Zengchang Pang, Ruqin Gao, Shanpeng Li, Shaojie Wang, Yongmei Liu, Ranran Du, Yajing Zang, Liang Cheng, Xiaocao Tian, Hua Zhang, Yaoming Zhai, Feng Ning, Xiaohui Sun, and Feifei Li. Licang CDC—Silu Lv, Junzheng Wang, and Wei Hou. Heilongjiang Provincial CDC—Mingyuan Zeng, Ge Jiang, Xue Zhou. Nangang CDC: Liqiu Yang, Hui He, Bo Yu, Yanjie Li, Qinai Xu,Quan Kang, and Ziyan Guo. Hainan Provincial CDC—Dan Wang, Ximin Hu, Hongmei Wang, Jinyan Chen, Yan Fu, Zhenwang Fu, and Xiaohuan Wang. Meilan CDC—Min Weng, Zhendong Guo, Shukuan Wu, Yilei Li, Huimei Li, and Zhifang Fu. Jiangsu Provincial CDC—Ming Wu, Yonglin Zhou, Jinyi Zhou, Ran Tao, Jie Yang, and Jian Su. Suzhou CDC—Fang Liu, Jun Zhang, Yihe Hu, Yan Lu, Liangcai Ma, Aiyu Tang, Shuo Zhang, Jianrong Jin, and Jingchao Liu. Guangxi Provincial CDC—Zhenzhu Tang, Naying Chen, and Ying Huang. Liuzhou CDC—Mingqiang Li, Jinhuai Meng, Rong Pan, Qilian Jiang, Jian Lan, Yun Liu, Liuping Wei, Liyuan Zhou, Ningyu Chen, Ping Wang, Fanwen Meng, Yulu Qin, Sisi Wang. Sichuan Provincial CDC—Xianping Wu, Ningmei Zhang, Xiaofang Chen, and Weiwei Zhou. Pengzhou CDC—Guojin Luo, Jianguo Li, Xiaofang Chen, Xunfu Zhong, Jiaqiu Liu, and Qiang Sun.

Li, Jinhuai Meng, Rong Pan, Qilian Jiang, Jian Lan, Yun Liu, Liuping Wei, Liyuan Zhou, Ningyu Chen, Ping Wang, Fanwen Meng, Yulu Qin, Sisi Wang. Sichuan Provincial CDC—Xianping Wu, Ningmei Zhang, Xiaofang Chen, and Weiwei Zhou. Pengzhou CDC—Guojin Luo, Jianguo Li, Xiaofang Chen, Xunfu Zhong, Jiaqiu Liu, and Qiang Sun. Gansu Provincial CDC—Pengfei Ge, Xiaolan Ren, and Caixia Dong. Maiji CDC—Hui Zhang, Enke Mao, Xiaoping Wang, Tao Wang, and Xi Zhang. Henan Provincial CDC—Ding Zhang, Gang Zhou, Shixian Feng, Liang Chang, and Lei Fan. Huixian CDC—Yulian Gao, Tianyou He, Huarong Sun, Chen Hu, Xukui Zhang, Huifang Wu, and Pan He. Zhejiang Provincial CDC—Min Yu, Ruying Hu, and Hao Wang. Tongxiang CDC—Yijian Qian, Chunmei Wang, Kaixu Xie, Lingli Chen, Yidan Zhang, Dongxia Pan, and Qijun Gu. Hunan Provincial CDC—Yuelong Huang, Biyun Chen, Li Yin, Huilin Liu, Zhongxi Fu, Qiaohua Xu. Liuyang CDC—Xin Xu, Hao Zhang, Huajun Long, Xianzhi Li, Libo Zhang, and Zhe Qiu.

ncial CDC—Min Yu, Ruying Hu, and Hao Wang. Tongxiang CDC—Yijian Qian, Chunmei Wang, Kaixu Xie, Lingli Chen, Yidan Zhang, Dongxia Pan, and Qijun Gu. Hunan Provincial CDC—Yuelong Huang, Biyun Chen, Li Yin, Huilin Liu, Zhongxi Fu, Qiaohua Xu. Liuyang CDC—Xin Xu, Hao Zhang, Huajun Long, Xianzhi Li, Libo Zhang, and Zhe Qiu. Supported by the UK Medical Research Council: Global Challenges Research Fund (Foundation Award MR/P025080/1). K.H.C. is a recipient of the D.Phil. Scholarship from the Nuffield Department of Population Health and St. Anne’s College, University of Oxford. The CKB baseline survey and the first resurvey were supported by the Kadoorie Charitable Foundation in Hong Kong. Long-term follow-up has been supported by the UK Wellcome Trust (202922/Z/16/Z, 104085/Z/14/Z, and 088158/Z/09/Z) and grants from the National Natural Science Foundation of China (81390540, 81390541, and 81390544) and from the National Key Research and Development Program of China (2016YFC0900500, 2016YFC0900501, 2016YFC0900504, and 2016YFC1303904). The British Heart Foundation, UK Medical Research Council, and Cancer Research provide core funding to the Clinical Trial Service Unit and Epidemiological Studies Unit at Oxford University for the project. A list of members of the China Kadoorie Biobank Collaborative Group may be found before the beginning of the References.

Supported by the UK Medical Research Council: Global Challenges Research Fund (Foundation Award MR/P025080/1). K.H.C. is a recipient of the D.Phil. Scholarship from the Nuffield Department of Population Health and St. Anne’s College, University of Oxford. The CKB baseline survey and the first resurvey were supported by the Kadoorie Charitable Foundation in Hong Kong. Long-term follow-up has been supported by the UK Wellcome Trust (202922/Z/16/Z, 104085/Z/14/Z, and 088158/Z/09/Z) and grants from the National Natural Science Foundation of China (81390540, 81390541, and 81390544) and from the National Key Research and Development Program of China (2016YFC0900500, 2016YFC0900501, 2016YFC0900504, and 2016YFC1303904). The British Heart Foundation, UK Medical Research Council, and Cancer Research provide core funding to the Clinical Trial Service Unit and Epidemiological Studies Unit at Oxford University for the project. A list of members of the China Kadoorie Biobank Collaborative Group may be found before the beginning of the References. Author Contributions: Z.C., R.P., L.Y., and Y.C. contributed to the overall design and oversaw the conduct and long-term follow-up of the China Kadoorie Biobank study. K.H.C., O.P.K., D.A.B., K.B.H.L., and Z.C. conceived the present study. K.H.C. reviewed the literature, analyzed the data, and wrote the first draft of the report, supervised by O.P.K., D.A.B., K.B.H.L., and Z.C. All authors contributed to the interpretation and development of the report and approved the final version.

K.H.C., O.P.K., D.A.B., K.B.H.L., and Z.C. conceived the present study. K.H.C. reviewed the literature, analyzed the data, and wrote the first draft of the report, supervised by O.P.K., D.A.B., K.B.H.L., and Z.C. All authors contributed to the interpretation and development of the report and approved the final version. This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org. Originally Published in Press as DOI: 10.1164/rccm.201803-0432OC on September 21, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

At a Glance Commentary Scientific Knowledge on the Subject Noncommunicable respiratory diseases and exposure to air pollution are thought to be important causes of morbidity and mortality in sub-Saharan African adults. Recent BOLD (Burden of Obstructive Lung Disease) studies found a high burden of spirometric restriction but little spirometric obstruction in several sub-Saharan African countries and no association between spirometric obstruction and use of dirty-burning fuels. It is not known whether an association between spirometric obstruction and solid fuel use would be seen if personal exposure to air pollution were measured in addition to self-reported exposure. CAPS (Cooking and Pneumonia Study)—a trial of cleaner burning biomass-fueled cookstoves on pneumonia in children <5 years of age in rural Malawi—offered the opportunity to explore this and other secondary trial outcomes in adults. What This Study Adds to the Field In adults living in Chikhwawa, rural Malawi: 13.6% of participants had chronic respiratory symptoms (mainly cough); >40% had abnormal spirometry (mainly spirometric restriction); day-to-day air pollution exposures were approximately three times the World Health Organization upper safety limit; air pollution exposures were not associated with demographic, clinical, or spirometric characteristics; and there was no association between CAPS trial arm and any of the secondary trial outcomes.

ction); day-to-day air pollution exposures were approximately three times the World Health Organization upper safety limit; air pollution exposures were not associated with demographic, clinical, or spirometric characteristics; and there was no association between CAPS trial arm and any of the secondary trial outcomes. Highly polluting fuels, including animal dung, crop residues, wood, charcoal, and kerosene, are used by almost half the world’s population to provide energy for cooking, heating, and lighting (1–3). These fuels are typically burned in and around the home environment in inefficient ways (e.g., open fires), which leads to high levels of air pollution in and immediately outside of homes. The World Health Organization (WHO) has estimated that exposure to household air pollution leads to >4 million deaths each year (3). The latest Global Burden of Disease Study estimates this number is closer to 2.5 million, but even these lower estimates represent a substantial burden of morbidity and mortality that falls particularly heavily on the world’s poor (4). Household air pollution has been considered to increase the risk of pneumonia in children and of chronic obstructive pulmonary disease (COPD) and cardiovascular disease in adults (1–3).

ese lower estimates represent a substantial burden of morbidity and mortality that falls particularly heavily on the world’s poor (4). Household air pollution has been considered to increase the risk of pneumonia in children and of chronic obstructive pulmonary disease (COPD) and cardiovascular disease in adults (1–3). In 2017, we published the findings of a cluster-randomized controlled trial of introducing a cleaner-burning biomass-fueled cookstove to prevent pneumonia in children <5 years of age in rural Malawi (CAPS [Cooking and Pneumonia Study]) (5). CAPS is one of a small number of trials done to date to evaluate the effects of reducing biomass smoke exposure on health outcomes and is the largest trial of a cookstove intervention on health outcomes conducted anywhere in the world (n = 10,750 children from 8,626 households across 150 clusters). The major finding of this trial was that there was no difference between the intervention and control groups among children in pneumonia incidence defined using the criteria of the Integrated Management of Childhood Illness program. This unexpected finding has cast some doubt on the assumptions made by the Global Alliance for Clean Cookstoves that cleaner cookstoves and fuels save lives (6–11).

intervention and control groups among children in pneumonia incidence defined using the criteria of the Integrated Management of Childhood Illness program. This unexpected finding has cast some doubt on the assumptions made by the Global Alliance for Clean Cookstoves that cleaner cookstoves and fuels save lives (6–11). Herein we report the findings of a cross-sectional study of the prevalence and determinants of noncommunicable respiratory disease among adults living in communities that participated in CAPS, which addresses the prespecified secondary trial objective of determining prevalence and determinants of obstructive lung disease in adults in rural Malawi (5). In this setting, use of highly polluting fuels for day-to-day household energy requirements is the norm, and therefore a high burden of COPD associated with household air pollution was expected. Methods Study Design We performed a cross-sectional study of the prevalence and determinants of noncommunicable respiratory disease among adults living in Chikhwawa District, Malawi. Setting Chikhwawa is ∼50 km from the nearest city, Blantyre, on the southern Shire River Valley, and it consists primarily of subsistence farmers living in rural village communities. The Malawi College of Medicine Research Ethics Committee (Ethics Committee reference no. P.11/12/1308) and the Liverpool School of Tropical Medicine Research Ethics Committee (Ethics Committee reference no. 12.40) approved the CAPS trial protocol that includes this work, a summary of which was published by The Lancet (12).

ged continuous PM2.5 and CO using a light-scattering photometer and an electrochemical cell CO sensor, respectively. All monitors were calibrated at the Aprovecho Research Center prior to use in the study. Monitors were worn in small backpacks apart from during sleep, when they were kept beside the sleeping mat or bed. Variables Clinical outcomes were presence or absence of specific symptoms as assessed by a questionnaire. The questions (outcomes) asked were as follows: Do you usually have a cough when you don’t have a cold (cough outcome)? Do you usually bring up phlegm from your chest (phlegm outcome)? Have you had wheezing/whistling in your chest at any point in the last 12 months, in the absence of a cold (wheeze outcome)? Do you have shortness of breath when hurrying on the level or walking up a slight hill (dyspnea outcome)? And have your breathing problems interfered with your daily activities (functional limitation outcome)? A composite variable for any symptoms was created by defining as positive if an individual reported any of the above symptoms (any symptoms outcome). Continuous FEV1 and FVC spirometry values were used in the primary analysis. Spirometric obstruction and restriction were defined according to the NHANES III white reference range lower limits of normal (15).

Variables Clinical outcomes were presence or absence of specific symptoms as assessed by a questionnaire. The questions (outcomes) asked were as follows: Do you usually have a cough when you don’t have a cold (cough outcome)? Do you usually bring up phlegm from your chest (phlegm outcome)? Have you had wheezing/whistling in your chest at any point in the last 12 months, in the absence of a cold (wheeze outcome)? Do you have shortness of breath when hurrying on the level or walking up a slight hill (dyspnea outcome)? And have your breathing problems interfered with your daily activities (functional limitation outcome)? A composite variable for any symptoms was created by defining as positive if an individual reported any of the above symptoms (any symptoms outcome). Continuous FEV1 and FVC spirometry values were used in the primary analysis. Spirometric obstruction and restriction were defined according to the NHANES III white reference range lower limits of normal (15). Exposures of interest included personal exposure to PM2.5 or CO as measured by the personal monitoring device, and two exposures assessed by questionnaire: smoking status and previous episode of tuberculosis (TB). A questionnaire-assessed variable asking for any biomass exposure was considered, but as most (>99%) indicated yes, it was not included in any modeling.

re to PM2.5 or CO as measured by the personal monitoring device, and two exposures assessed by questionnaire: smoking status and previous episode of tuberculosis (TB). A questionnaire-assessed variable asking for any biomass exposure was considered, but as most (>99%) indicated yes, it was not included in any modeling. Raw PM2.5 and CO exposures were corrected for background levels using calibration values for each monitoring period. In cases where calibration data were missing or corrupted (<5%), aggregated mean calibration values were used. Observations where <2,000 minutes of time were recorded were excluded, as were monitoring periods affected by device malfunction. Both PM2.5 and CO were log10 transformed for presentation and inclusion in models due to large positive skew. Potential confounders/effect modifiers included were body mass index (BMI) and/or height (cm) and weight (kg) variables, as well as age, years of education, and sex. Study Size We initially invited 2,000 people to participate but increased this to 3,000 to achieve the required sample size. Participants were stratified into two age groups: 18–39 years and ≥40 years. We estimated that, after allowing for unequal age and sex distributions, refusals, and inability to provide spirometry measurements of acceptable quality, a sample of just 300 participants in any one sex/age stratum (1,200 total) would provide an estimate of chronic airflow limitation prevalence in this stratum with a precision (95% confidence interval [CI]) of ±3.3% to ±5.0% assuming a prevalence of 10–25%.

ility to provide spirometry measurements of acceptable quality, a sample of just 300 participants in any one sex/age stratum (1,200 total) would provide an estimate of chronic airflow limitation prevalence in this stratum with a precision (95% confidence interval [CI]) of ±3.3% to ±5.0% assuming a prevalence of 10–25%. Statistical Analyses Univariate analysis was completed using descriptive statistics to explore the characteristics of the study population. Descriptive analysis is presented for clarity using categorical versions of BMI (underweight, normal, overweight, or obese) and categorical versions of age; however, age, weight, and height were entered into models as continuous variables. Participants who completed the study in full or in part were assessed for selection bias using the χ2 and Student’s t tests. Multivariable logistic regression was used to estimate the strength of the association between measured exposure variables and dichotomous clinical outcomes, adjusting for potential confounders. All models were adjusted for age, sex, weight, and height. Linear multivariable regression was used to estimate the association between exposures and continuous lung function values (FEV1, FVC, and FEV1/FVC). Secondary exploratory trial efficacy analyses were by intention to treat. Statistical significance was nominally set at α = 0.05. Stata version 14.2 and R version 3.4 statistical software was used for data analysis (Stata statistical software: R.14; StataCorp, LLC).

nuous lung function values (FEV1, FVC, and FEV1/FVC). Secondary exploratory trial efficacy analyses were by intention to treat. Statistical significance was nominally set at α = 0.05. Stata version 14.2 and R version 3.4 statistical software was used for data analysis (Stata statistical software: R.14; StataCorp, LLC). Role of the Funding Source The funders had no role in the study design, data collection, analysis, interpretation, or writing of the report. The corresponding author had full access to all the study data and had final responsibility for the decision to submit for publication. Results Between August 2014 and July 2015, we attempted to contact the 3,000 adults sampled to invite them to participate, of whom 1,481 (49.4%) consented and completed BOLD study questionnaires. Of these, 950 (64.6%) went on to do spirometry; the remaining 520 (35.3%) were unable to do spirometry because they could not physically cooperate with the procedure (n = 258; 48.9%), had a fieldworker-determined contraindication (n = 193; 37%), or refused (n = 69; 13.3%). Of the 1,481 participants, 1,144 (77.2%) underwent personal air pollution exposure monitoring. There were 424 (28.6%) participants from CAPS intervention or control households (Figure 1). Figure 1. Participant recruitment flow diagram. TB = tuberculosis.

Results Between August 2014 and July 2015, we attempted to contact the 3,000 adults sampled to invite them to participate, of whom 1,481 (49.4%) consented and completed BOLD study questionnaires. Of these, 950 (64.6%) went on to do spirometry; the remaining 520 (35.3%) were unable to do spirometry because they could not physically cooperate with the procedure (n = 258; 48.9%), had a fieldworker-determined contraindication (n = 193; 37%), or refused (n = 69; 13.3%). Of the 1,481 participants, 1,144 (77.2%) underwent personal air pollution exposure monitoring. There were 424 (28.6%) participants from CAPS intervention or control households (Figure 1). Figure 1. Participant recruitment flow diagram. TB = tuberculosis. The mean age (SD) of participants was 43.8 (17.8) years and 57% were female (Table 1). Just more than half of the participants had been educated only to primary school level, with a third having had no formal school education. The use of biomass fuels for cooking was almost universal (99.8%). Table 1. Demographic and Clinical Characteristics (N = 1,481) Characteristic Level n (%) or Mean (SD) Age group, yr <39 686 (46.32) 40–49 259 (17.49) 50–59 216 (14.58) 60–69 160 (10.80) >70 160 (10.80) Sex Male 637 (43.01) Female 844 (56.99) Education None 485 (32.79) Primary 758 (51.25) Middle 205 (13.86) High school or college 31 (2.10) Missing 2 (0.0) Years of education, mean (SD) 4.20 (4.09) Years of education if any, mean (SD) 6.31 (3.44) Smoking Never smoker 1,152 (77.8) Current or ever smoker 328 (22.2)

Characteristic Level n (%) or Mean (SD) Age group, yr <39 686 (46.32) 40–49 259 (17.49) 50–59 216 (14.58) 60–69 160 (10.80) >70 160 (10.80) Sex Male 637 (43.01) Female 844 (56.99) Education None 485 (32.79) Primary 758 (51.25) Middle 205 (13.86) High school or college 31 (2.10) Missing 2 (0.0) Years of education, mean (SD) 4.20 (4.09) Years of education if any, mean (SD) 6.31 (3.44) Smoking Never smoker 1,152 (77.8) Current or ever smoker 328 (22.2) Pack-years of smoking 0 1,152 (77.8) Up to 10 pack-years 263 (17.8) >10 pack-years 63 (4.3) Missing 3 (0.0) BMI group, kg/m2 Underweight (BMI, <18.5) 188 (14.4) Normal (BMI, 18.5–25) 945 (72.5) Overweight (BMI, 25–30) 130 (10.0) Obese (BMI, >30) 40 (3.1) Missing 178 (12.0) Previous TB No 1,434 (92.3) Yes 47 (3.2) Symptoms Cough (Do you usually cough when you don’t have a cold?) 165 (11.1) Sputum (Do you usually bring up phlegm from your chest?) 38 (2.6) Wheeze (Have you had wheezing/whistling in your chest at any point in past 12 months in the absence of a cold?) 23 (1.6) MRC dyspnea II (Do you have shortness of breath when hurrying on the level or walking up a slight hill?) 23 (1.6) Any respiratory symptoms (Any of cough, sputum, wheeze without cold, exertional breathlessness as above?) 201 (13.6) Functional limitation (Have breathing problems interfered with your usual daily activities?) 43 (2.9) Definition of abbreviations: BMI = body mass index; MRC = Medical Research Council; TB = tuberculosis. Data are n (%) unless otherwise indicated.

Symptoms Cough (Do you usually cough when you don’t have a cold?) 165 (11.1) Sputum (Do you usually bring up phlegm from your chest?) 38 (2.6) Wheeze (Have you had wheezing/whistling in your chest at any point in past 12 months in the absence of a cold?) 23 (1.6) MRC dyspnea II (Do you have shortness of breath when hurrying on the level or walking up a slight hill?) 23 (1.6) Any respiratory symptoms (Any of cough, sputum, wheeze without cold, exertional breathlessness as above?) 201 (13.6) Functional limitation (Have breathing problems interfered with your usual daily activities?) 43 (2.9) Definition of abbreviations: BMI = body mass index; MRC = Medical Research Council; TB = tuberculosis. Data are n (%) unless otherwise indicated. One or more chronic respiratory symptom was reported by 201 (13.6%; 95% CI, 11.9–15.4) participants (Table 1 and Figure 2). Respiratory symptoms were more commonly reported with increasing age. Regular cough was reported by 11.1% (95% CI, 9.6–12.8) while 2.6% (95% CI, 1.9–3.5) reported usually coughing up phlegm. Breathlessness and wheeze were less commonly reported: 1.6% (95% CI, 1.0–2.3) and 1.6% (95% CI, 1.0–2.3), respectively. Respiratory symptoms that limited functional ability were reported by 2.9% (95% CI, 2.2–3.9). A previous diagnosis of TB was reported by 3.2% (95% CI, 2.4–4.2), which was more common with increasing age. Current or former smoking was reported by 22.1% (95% CI, 20.1–24.3), although only 4.3% had a >10 pack-year history. Many participants (14.4%) had a low BMI. Figure 2. Age-stratified prevalence of respiratory symptoms.

One or more chronic respiratory symptom was reported by 201 (13.6%; 95% CI, 11.9–15.4) participants (Table 1 and Figure 2). Respiratory symptoms were more commonly reported with increasing age. Regular cough was reported by 11.1% (95% CI, 9.6–12.8) while 2.6% (95% CI, 1.9–3.5) reported usually coughing up phlegm. Breathlessness and wheeze were less commonly reported: 1.6% (95% CI, 1.0–2.3) and 1.6% (95% CI, 1.0–2.3), respectively. Respiratory symptoms that limited functional ability were reported by 2.9% (95% CI, 2.2–3.9). A previous diagnosis of TB was reported by 3.2% (95% CI, 2.4–4.2), which was more common with increasing age. Current or former smoking was reported by 22.1% (95% CI, 20.1–24.3), although only 4.3% had a >10 pack-year history. Many participants (14.4%) had a low BMI. Figure 2. Age-stratified prevalence of respiratory symptoms. Of the 950 participants who did spirometry, 886 (93.2%) achieved BOLD study quality standards and were included in the analyses. Factors associated with declining or not completing spirometry to European Respiratory Society/American Thoracic Society standards were older mean age (48 vs. 39 yr; P < 0.001), being female (65% vs. 51%; P < 0.001), lower mean years of education (2 vs. 5 yr; P < 0.001), and lower mean BMI (20.7 vs. 21.3; P < 0.001). As shown in Table E1 in the online supplement, participants who completed spirometry were less likely to have cough, wheeze, and dyspnea compared with those who did not complete spirometry and were slightly more likely to have phlegm and functional limitation, although none of these differences was statistically significant. Spirometric obstruction and restriction were present in 8.7% (95% CI, 7.0–10.7) and 34.8% (95% CI, 31.7–38.0) of the 886 participants who met the required quality standards, respectively.

ere slightly more likely to have phlegm and functional limitation, although none of these differences was statistically significant. Spirometric obstruction and restriction were present in 8.7% (95% CI, 7.0–10.7) and 34.8% (95% CI, 31.7–38.0) of the 886 participants who met the required quality standards, respectively. Of the 1,144 participants (mean age [SD], 43.9 [17.9] yr; 57% female) who underwent personal exposure monitoring, 1,117 (97.6%) had valid exposure monitoring records. The 48-hour median personal PM2.5 and CO exposures were 71.0 μg/m3 (interquartile range [IQR], 44.6–119.2) and 1.23 ppm (IQR, 0.79–1.93), respectively. There was weak correlation between these two air pollution exposure measures (Figure 3). Figure 3. Scatter plot between personal exposure to particulate matter <2.5 μm in aerodynamic diameter (PM2.5) and carbon monoxide (CO).

Of the 1,144 participants (mean age [SD], 43.9 [17.9] yr; 57% female) who underwent personal exposure monitoring, 1,117 (97.6%) had valid exposure monitoring records. The 48-hour median personal PM2.5 and CO exposures were 71.0 μg/m3 (interquartile range [IQR], 44.6–119.2) and 1.23 ppm (IQR, 0.79–1.93), respectively. There was weak correlation between these two air pollution exposure measures (Figure 3). Figure 3. Scatter plot between personal exposure to particulate matter <2.5 μm in aerodynamic diameter (PM2.5) and carbon monoxide (CO). In logistic multivariable analysis, smoking (odds ratio [OR], 1.56; 95% CI, 1.01–2.41) and previous TB (OR, 2.81; 95% CI, 1.19–6.08) were associated with cough (Table 2 and Table E1). In continuous multivariable analysis, both FEV1 and FVC had a negative association with increasing age and were higher for men compared with women (Table 3). Smoking (coefficient estimate, −0.09; 95% CI, −0.16 to −0.01) and previous TB (coefficient estimate, −0.46; 95% CI, −0.64 to −0.28) were associated with FEV1, and previous TB was associated with FVC (coefficient estimate, −0.35; 95% CI, −0.56 to −0.15). There was no association between personal exposure to PM2.5 and any of the demographic and clinical characteristics or spirometric indices (Tables 4 and 5). The only statistically significant association was between exposure to CO and reporting “any chronic respiratory symptoms” (OR, 1.46; 95% CI, 1.04–2.05). There were no statistically significant associations between personal exposure to CO and any other demographic and clinical characteristics or to any spirometric indices. There were 424 (227 intervention; 197 control) participants in the CAPS intention-to-treat population; however, not all of them had complete spirometry (133 without) or exposure measures (87 without). There were no differences in respiratory symptoms, spirometric indices, or exposure to CO or PM2.5 between the intervention and control groups (Table 6).

197 control) participants in the CAPS intention-to-treat population; however, not all of them had complete spirometry (133 without) or exposure measures (87 without). There were no differences in respiratory symptoms, spirometric indices, or exposure to CO or PM2.5 between the intervention and control groups (Table 6). Table 2. Odds Ratios (95% Confidence Interval) for Chronic Respiratory Symptom Outcomes Estimated by Multivariable Logistic Regression Variable Cough Phlegm* Wheeze* Dyspnea* Functional Limitation Any Symptoms Age, yr 1.01 (1.00–1.02) 1.00 (0.97–1.02) 1.02 (0.99–1.05) 1.01 (0.98–1.04) 1.00 (0.95–1.02) 1.00 (0.97–1.02) Male Ref Ref Ref Ref Ref Ref Female 0.78 (0.49–1.25) 1.02 (0.42–2.51) 0.97 (0.30–3.28) 3.08 (0.88–11.65) 1.17 (0.28–2.37) 1.08 (0.70–1.67) Never smoked Ref Ref Ref Ref Ref Ref Ever smoked 1.56 (1.01–2.41) 1.37 (0.58–3.15) 0.77 (0.20–2.47) 1.85 (0.51–6.07) 0.65 (0.18–1.93) 1.59 (1.05–2.39) Previous TB: no Ref — — — Ref Ref Previous TB: yes 2.81 (1.19–6.08) — — — 2.64 (0.40–9.95) 2.50 (1.04–15.58) Years of education 0.97 (0.92–1.02) 0.90 (0.81–1.00) 0.99 (0.86–1.13) 0.96 (0.83–1.10) 1.06 (0.96–1.16) 0.98 (0.93–1.03) Definition of abbreviations: Ref = reference; TB = tuberculosis. All models were also adjusted for weight (kg) and height (cm); total n = 1,303 owing to missing weight data. * Only one person had both TB and wheeze, one person had both TB and phlegm, and one person had both TB and dyspnea; TB was excluded from these models.

Variable Cough Phlegm* Wheeze* Dyspnea* Functional Limitation Any Symptoms Age, yr 1.01 (1.00–1.02) 1.00 (0.97–1.02) 1.02 (0.99–1.05) 1.01 (0.98–1.04) 1.00 (0.95–1.02) 1.00 (0.97–1.02) Male Ref Ref Ref Ref Ref Ref Female 0.78 (0.49–1.25) 1.02 (0.42–2.51) 0.97 (0.30–3.28) 3.08 (0.88–11.65) 1.17 (0.28–2.37) 1.08 (0.70–1.67) Never smoked Ref Ref Ref Ref Ref Ref Ever smoked 1.56 (1.01–2.41) 1.37 (0.58–3.15) 0.77 (0.20–2.47) 1.85 (0.51–6.07) 0.65 (0.18–1.93) 1.59 (1.05–2.39) Previous TB: no Ref — — — Ref Ref Previous TB: yes 2.81 (1.19–6.08) — — — 2.64 (0.40–9.95) 2.50 (1.04–15.58) Years of education 0.97 (0.92–1.02) 0.90 (0.81–1.00) 0.99 (0.86–1.13) 0.96 (0.83–1.10) 1.06 (0.96–1.16) 0.98 (0.93–1.03) Definition of abbreviations: Ref = reference; TB = tuberculosis. All models were also adjusted for weight (kg) and height (cm); total n = 1,303 owing to missing weight data. * Only one person had both TB and wheeze, one person had both TB and phlegm, and one person had both TB and dyspnea; TB was excluded from these models. Table 3. Coefficient Estimates (95% Confidence Interval) for Continuous Spirometry Outcomes FEV1, FVC, and FEV1/FVC Ratio (n = 886)

All models were also adjusted for weight (kg) and height (cm); total n = 1,303 owing to missing weight data. * Only one person had both TB and wheeze, one person had both TB and phlegm, and one person had both TB and dyspnea; TB was excluded from these models. Table 3. Coefficient Estimates (95% Confidence Interval) for Continuous Spirometry Outcomes FEV1, FVC, and FEV1/FVC Ratio (n = 886) Variable FEV1 FVC FEV1/FVC Age, yr −0.02 (−0.02 to −0.02) −0.01 (−0.01 to −0.01) −0.28 (−0.31 to −0.24) Male Ref Ref Ref Female −0.53 (−0.60 to −0.45) −0.70 (−0.78 to −0.62) 1.37 (0.18 to 2.56) Never smoked Ref Ref Ref Ever smoked −0.09 (−0.16 to −0.01) −0.05 (−0.14 to 0.04) −1.76 (−2.99 to −0.53) Previous TB: no Ref Ref Ref Previous TB: yes −0.46 (−0.64 to −0.28) −0.36 (−0.56 to −0.15) −7.83 (−10.74 to −4.91) Years of education 0 (0 to 0.01) 0 (−0.01 to 0.01) 0.18 (0.05 to 0.3) Definition of abbreviations: Ref = reference; TB = tuberculosis. All models were also adjusted for weight (kg) and height (cm). Table 4. Odds Ratios (95% Confidence Interval) for Symptom Outcomes Estimated by Multivariable Logistic Regression in Participants with Exposure Measurements (n = 985)

Variable FEV1 FVC FEV1/FVC Age, yr −0.02 (−0.02 to −0.02) −0.01 (−0.01 to −0.01) −0.28 (−0.31 to −0.24) Male Ref Ref Ref Female −0.53 (−0.60 to −0.45) −0.70 (−0.78 to −0.62) 1.37 (0.18 to 2.56) Never smoked Ref Ref Ref Ever smoked −0.09 (−0.16 to −0.01) −0.05 (−0.14 to 0.04) −1.76 (−2.99 to −0.53) Previous TB: no Ref Ref Ref Previous TB: yes −0.46 (−0.64 to −0.28) −0.36 (−0.56 to −0.15) −7.83 (−10.74 to −4.91) Years of education 0 (0 to 0.01) 0 (−0.01 to 0.01) 0.18 (0.05 to 0.3) Definition of abbreviations: Ref = reference; TB = tuberculosis. All models were also adjusted for weight (kg) and height (cm). Table 4. Odds Ratios (95% Confidence Interval) for Symptom Outcomes Estimated by Multivariable Logistic Regression in Participants with Exposure Measurements (n = 985) Cough Phlegm* Wheeze* Dyspnea* Functional Limitation Any Symptoms Ever smoked (ref: never smoked) 1.72 (1.02–2.91) 0.99 (0.37–2.53) 0.35 (0.02–2.32) 2.62 (0.56–11.24) 0.78 (0.20–2.47) 1.67 (1.02–2.71) Previous TB (ref: no previous TB) 2.87 (1.07–6.87) — — — 3.00 (0.5–11.74) 2.47 (0.91–6.07) CO (log10 ppm) 1.29 (0.93–1.77) 1.50 (0.83–2.54) 2.12 (0.96–4.16) 1.27 (0.48–2.88) 1.45 (0.81–2.43) 1.46 (1.04–2.05) PM2.5 log10 μg/m3) 1.02 (0.95–1.13) 0.96 (0.88–1.11) 1.00 (0.87–1.38) 1.11 (0.89–1.67) 0.99 (0.90–1.16) 1.02 (0.95–1.11) Definition of abbreviations: CO = carbon monoxide; PM2.5 = particulate matter <2.5 μm in aerodynamic diameter; ref = reference; TB = tuberculosis. All models were adjusted for weight (kg), height (cm), age (yr), sex (male, female), and years of formal education.

Cough Phlegm* Wheeze* Dyspnea* Functional Limitation Any Symptoms Ever smoked (ref: never smoked) 1.72 (1.02–2.91) 0.99 (0.37–2.53) 0.35 (0.02–2.32) 2.62 (0.56–11.24) 0.78 (0.20–2.47) 1.67 (1.02–2.71) Previous TB (ref: no previous TB) 2.87 (1.07–6.87) — — — 3.00 (0.5–11.74) 2.47 (0.91–6.07) CO (log10 ppm) 1.29 (0.93–1.77) 1.50 (0.83–2.54) 2.12 (0.96–4.16) 1.27 (0.48–2.88) 1.45 (0.81–2.43) 1.46 (1.04–2.05) PM2.5 log10 μg/m3) 1.02 (0.95–1.13) 0.96 (0.88–1.11) 1.00 (0.87–1.38) 1.11 (0.89–1.67) 0.99 (0.90–1.16) 1.02 (0.95–1.11) Definition of abbreviations: CO = carbon monoxide; PM2.5 = particulate matter <2.5 μm in aerodynamic diameter; ref = reference; TB = tuberculosis. All models were adjusted for weight (kg), height (cm), age (yr), sex (male, female), and years of formal education. * Only one person had both TB and wheeze, one person had both TB and phlegm, and one person had both TB and dyspnea; TB was excluded from these models. Table 5. Coefficient Estimates (95% Confidence Interval) for Continuous Spirometry Outcomes FEV1, FVC, and FEV1/FVC Ratio in Participants with Personal Air Pollution Exposure Measurements (n = 886)

* Only one person had both TB and wheeze, one person had both TB and phlegm, and one person had both TB and dyspnea; TB was excluded from these models. Table 5. Coefficient Estimates (95% Confidence Interval) for Continuous Spirometry Outcomes FEV1, FVC, and FEV1/FVC Ratio in Participants with Personal Air Pollution Exposure Measurements (n = 886) Variable FEV1 FVC FEV1/FVC Age, yr −0.02 (−0.02 to −0.01) −0.01 (−0.02 to −0.01) −0.28 (−0.35 to −0.20) Male Ref Ref Ref Female −0.58 (−0.61 to −0.44) −0.70 (−0.79 to −0.60) 1.25 (−0.09 to 2.56) Never smoked Ref Ref Ref Ever smoked −0.1 (−0.19 to −0.02) −0.07 (−0.16 to 0.03) −1.83 (−3.20 to −0.45) Previous TB: no Ref Ref Ref Previous TB: yes −0.32 (−0.52 to −0.11) −0.26 (−0.49 to −0.02) −6.16 (−9.48 to −2.85) Years of education 0 (0 to 0.01) 0 (−0.01 to 0.01) 0.15 (0.01 to 0.29) CO (log10 ppm) 0.01 (−0.04 to 0.06) 0.01 (−0.04 to 0.07) 0.13 (−0.68 to 0.94) PM2.5 (log10 μg/m3) 0 (−0.02 to 0.01) 0 (−0.01 to 0.01) −0.11 (−0.29 to 0.08) For definition of abbreviations, see Table 4. All models were also adjusted for weight (kg) and height (cm). Table 6. CAPS Intention-to-Treat Secondary Trial Analyses (n = 424)

Variable FEV1 FVC FEV1/FVC Age, yr −0.02 (−0.02 to −0.01) −0.01 (−0.02 to −0.01) −0.28 (−0.35 to −0.20) Male Ref Ref Ref Female −0.58 (−0.61 to −0.44) −0.70 (−0.79 to −0.60) 1.25 (−0.09 to 2.56) Never smoked Ref Ref Ref Ever smoked −0.1 (−0.19 to −0.02) −0.07 (−0.16 to 0.03) −1.83 (−3.20 to −0.45) Previous TB: no Ref Ref Ref Previous TB: yes −0.32 (−0.52 to −0.11) −0.26 (−0.49 to −0.02) −6.16 (−9.48 to −2.85) Years of education 0 (0 to 0.01) 0 (−0.01 to 0.01) 0.15 (0.01 to 0.29) CO (log10 ppm) 0.01 (−0.04 to 0.06) 0.01 (−0.04 to 0.07) 0.13 (−0.68 to 0.94) PM2.5 (log10 μg/m3) 0 (−0.02 to 0.01) 0 (−0.01 to 0.01) −0.11 (−0.29 to 0.08) For definition of abbreviations, see Table 4. All models were also adjusted for weight (kg) and height (cm). Table 6. CAPS Intention-to-Treat Secondary Trial Analyses (n = 424) Outcome Intervention (n = 227) Control (n = 197) Intervention vs. Control Coefficient Estimate (95% CI) P Value Symptoms, n (%) 22 (9.7) 26 (13.2) 0.90 (0.45 to 1.82)* 0.87 FEV1, median (IQR) 2.81 (2.39 to 3.26) 2.77 (2.40 to 3.10) 0.08 (−0.06 to 0.22) 0.26 FVC, median (IQR) 3.37 (2.88 to 3.91) 3.31 (2.83 to 3.86) 0.04 (−0.13 to 0.21) 0.62 Mean CO, median (IQR) 1.13 (0.79 to 1.90) 1.28 (0.82 to 1.79) 0.67 (−0.60 to 1.96) 0.30 Mean PM2.5, median (IQR) 67.90 (44.72 to 112.95) 64.47 (40.73 to 101.80) −931.6 (−2,073.6 to 209.7) 0.11 Definition of abbreviations: CAPS = Cooking and Pneumonia Study; CI = confidence interval; CO = carbon monoxide; IQR = interquartile range; PM2.5 = particulate matter <2.5 μm in diameter.

) 0.67 (−0.60 to 1.96) 0.30 Mean PM2.5, median (IQR) 67.90 (44.72 to 112.95) 64.47 (40.73 to 101.80) −931.6 (−2,073.6 to 209.7) 0.11 Definition of abbreviations: CAPS = Cooking and Pneumonia Study; CI = confidence interval; CO = carbon monoxide; IQR = interquartile range; PM2.5 = particulate matter <2.5 μm in diameter. Mean exposure per individual is calculated, and the median (IQR) of those values is reported. * Odds ratio (95% CI). Discussion The main findings of this cross-sectional study of the burden and determinants of noncommunicable respiratory disease in adults living in Chikhwawa, rural Malawi, were that: 13.6% of participants had chronic respiratory symptoms (mainly cough); >40% had abnormal spirometry (mainly spirometric restriction); day-to-day air pollution exposures were approximately three times the WHO upper safety limit; and there was no association between CAPS trial arm and any of the secondary trial outcomes in the subset of adults included both in this study and the trial.

gh); >40% had abnormal spirometry (mainly spirometric restriction); day-to-day air pollution exposures were approximately three times the WHO upper safety limit; and there was no association between CAPS trial arm and any of the secondary trial outcomes in the subset of adults included both in this study and the trial. The finding of a low prevalence of spirometric obstruction in this setting—where highly polluting fuels are almost universally used for household energy needs and where exposure to household air pollution is high—is surprising given that household air pollution–associated COPD has been suggested to be a major global health problem and as such would be expected to be highly prevalent in our study setting (16–19). This finding is consistent with an emerging body of evidence challenging the dogma that exposure to household air pollution is a major cause of COPD, including a recent pooled analysis of BOLD study data from low-, middle-, and high-income countries (20). This analysis found no association between spirometric obstruction and self-reported use of solid fuels for cooking or heating. This is, however, an area of controversy, with investigators disagreeing about the interpretation of the available data (21, 22).

study data from low-, middle-, and high-income countries (20). This analysis found no association between spirometric obstruction and self-reported use of solid fuels for cooking or heating. This is, however, an area of controversy, with investigators disagreeing about the interpretation of the available data (21, 22). Many of the studies conducted to date looking at the association between COPD and exposure to household air pollution have had important methodological limitations, including case definition and exposure assessments. So far, studies of the long-term effects of air pollution have had to use a self-reported history of exposure with all the limitations that this may imply. To improve exposure assessment, we included 48 hours of personal air pollution exposure measurements in study participants in addition to questionnaire-based exposure assessments. Although this approach and the particular devices used have their limitations, by doing this we were able to deliver the first study of the burden of noncommunicable lung disease anywhere in the world to incorporate BOLD study methodology and measurements of personal air pollution exposure and to do so in almost 1,000 participants.

approach and the particular devices used have their limitations, by doing this we were able to deliver the first study of the burden of noncommunicable lung disease anywhere in the world to incorporate BOLD study methodology and measurements of personal air pollution exposure and to do so in almost 1,000 participants. Although personal air pollution exposure levels were undoubtedly high and at levels at which adverse health effects would be expected, and although widely considered a risk factor for noncommunicable respiratory diseases and COPD in particular, the only respiratory outcome associated with measured exposure to PM2.5 or CO was “any chronic respiratory symptoms” with increased CO exposure. Interestingly, whether questionnaire-based or directly measured personal air pollution exposure assessments were used, there was no significant association between air pollution exposure and an increased risk of spirometric abnormalities. However, we acknowledge that 48-hour measurements of air pollutants may not be an adequate surrogate for cumulative exposure to household air pollution that has been associated, albeit by self-report, with COPD. Our observations, taken together with the findings of other recent studies, bring into question the extent to which household air pollution and other sources of air pollution play in the development of abnormal lung function in rural African settings similar to this one in rural Malawi. It is plausible that the levels of personal exposure to air pollution seen are not high enough to accelerate lung function decline and the development of airflow obstruction in the way that tobacco smoke does; a prospective cohort study of the rate of decline in lung function in relationship to air pollution exposures is needed in adults in sub-Saharan Africa to explore this further.

ion seen are not high enough to accelerate lung function decline and the development of airflow obstruction in the way that tobacco smoke does; a prospective cohort study of the rate of decline in lung function in relationship to air pollution exposures is needed in adults in sub-Saharan Africa to explore this further. This study benefited from being conducted at the same time and in the same villages as CAPS, which presented the opportunity to look for an effect of the CAPS intervention on respiratory symptoms, spirometric indices, and personal air pollution exposures in a subsample of adults. Consistent with the main trial findings of no effect of the intervention on pneumonia in children <5 years of age (5), we found no evidence that the intervention was associated with beneficial effects on any of these trial secondary outcome measures among adults. However, these analyses were exploratory secondary analyses limited by a relatively small number of participants and therefore statistical power to detect effects and, although sufficient to see an effect on symptoms and air pollution exposures, there was limited time between intervention and outcome assessment for potential effects on spirometric indices to be seen. Other possible explanations for the lack of effect of the CAPS intervention on these outcomes include insufficient levels of intervention adoption, insufficient reductions in emissions and exposures, and other sources of air pollution exposure overwhelming any potential effect of the intervention (5).

dices to be seen. Other possible explanations for the lack of effect of the CAPS intervention on these outcomes include insufficient levels of intervention adoption, insufficient reductions in emissions and exposures, and other sources of air pollution exposure overwhelming any potential effect of the intervention (5). A notable observation of this study was that 35% of participants had spirometric restriction when benchmarked against NHANES III white reference range values. We consider the approach we have taken of benchmarking against the NHANES III white reference ranges as the best we can do at this time while accepting that this and all other currently available alternatives are not ideal. That includes locally derived reference ranges that might be helpful in defining what is ‘usual’ lung function in asymptomatic nonsmoking Malawian adults but that may be far from “optimal potential normal lung function.” Because there is evidence that the prognostic significance of spirometric restriction holds irrespective of racial/ethnic group when benchmarked in this way (23, 24), the finding of such a high burden of spirometric restriction in the rural Malawian population, and elsewhere in sub-Saharan Africa (25, 26), is of considerable concern; observational cohort studies are needed to understand the clinical characteristics and prognostic significance of these findings. The underlying drivers of spirometric restriction in sub-Saharan African populations are not yet understood, but we hypothesize that these are primarily environmental insults experienced in early life (e.g., malnutrition, infections and air pollution exposures before conception, in utero, and during childhood) such that adulthood is reached without maximal potential lung function having been achieved. Cross-sectional studies of the burden and determinants of noncommunicable lung disease in children in sub-Saharan Africa are needed to explore whether the same patterns of abnormality are seen in early life and, if so, studies even earlier in the life course to identify potential windows of opportunity to intervene to maximize lung health.

es of the burden and determinants of noncommunicable lung disease in children in sub-Saharan Africa are needed to explore whether the same patterns of abnormality are seen in early life and, if so, studies even earlier in the life course to identify potential windows of opportunity to intervene to maximize lung health. Strengths of this study include that it was conducted in a highly challenging research setting in one of the world’s poorest rural communities as part of the CAPS protocol; it is the first of the global BOLD studies to be conducted in a rural sub-Saharan African setting; and it is also the first BOLD study to incorporate personal air pollution exposure measurements and to do so at scale. Limitations include questionnaire assessments for most variables with potential for recall bias; the potential bias (e.g., underestimation of the burden of spirometric abnormalities) caused by participants who did not do spirometry, although the quality of those that did spirometry was generally high; and air pollution exposure assessments that provided only a 48-hour snapshot of exposure and were based on a light-scattering method alone for PM2.5.

mation of the burden of spirometric abnormalities) caused by participants who did not do spirometry, although the quality of those that did spirometry was generally high; and air pollution exposure assessments that provided only a 48-hour snapshot of exposure and were based on a light-scattering method alone for PM2.5. In conclusion, we found that exposures to air pollution among Malawian adults living in communities participating in CAPS were at levels well beyond those considered safe by the WHO. In keeping with the primary outcome of the CAPS trial, we found no effect of the intervention on any of the secondary trial outcomes (i.e., respiratory symptoms, spirometric indices, or air pollution exposures) in the subsample of adults participating in both this study and the trial. The prevalence of chronic respiratory symptoms and abnormal spirometry suggests that there may be an important burden of noncommunicable respiratory disease in these communities. The characteristics of noncommunicable respiratory disease in sub-Saharan Africa may be different to those previously expected, with more spirometric restriction and less obstruction (and household air pollution–associated COPD) than has been thought to exist. There is a need to explore other plausible explanations for the poor lung function observed in these and other low- and middle-income country populations, including further exploration of the role of TB, recurrent pneumonia, and nutrition. Clinically effective and cost-effective approaches for the prevention and control of noncommunicable respiratory diseases are very much needed in sub-Saharan Africa.

served in these and other low- and middle-income country populations, including further exploration of the role of TB, recurrent pneumonia, and nutrition. Clinically effective and cost-effective approaches for the prevention and control of noncommunicable respiratory diseases are very much needed in sub-Saharan Africa. Acknowledgment The authors thank the trial participants, village leaders, and CAPS representatives, the study team in Chikhwawa, Malawi–Liverpool–Wellcome Trust Clinical Research Programme and Liverpool School of Tropical Medicine, the CAPS trial steering committee and data monitoring committee, the Malawi Ministry of Health, the Aprovecho Research Centre, the African Clean Energy company, and the BOLD Centre for their valued contributions to making this work a success. We thank Stephen Gordon for comments on the paper.

hool of Tropical Medicine, the CAPS trial steering committee and data monitoring committee, the Malawi Ministry of Health, the Aprovecho Research Centre, the African Clean Energy company, and the BOLD Centre for their valued contributions to making this work a success. We thank Stephen Gordon for comments on the paper. Supported by a New Investigator Research Grant from the Medical Research Council (MR/L002515/1), a Joint Global Health Trials Grant from the Medical Research Council, the UK Department for International Development and Wellcome Trust (MR/K006533/1), a U.S. NIH R56 grant (R56ES023566), and a Wellcome Trust Grant (085790/Z/08/Z). Additional support was provided by the National Institute for Health Research Global Health Research Unit on Lung Health and TB in Africa at the Liverpool School of Tropical Medicine (“IMPALA”). With regard to IMPALA (grant 16/136/35) specifically: IMPALA was commissioned by the National Institute of Health Research using Official Development Assistance funding. The views expressed in this publication are those of the authors and not necessarily those of the National Institute for Health Research or the Department of Health. Author Contributions: Design: K.M., P.B., and J.B. Acquisition of data: R.N., K.M., P.B., and J.B. Analysis of data: R.N., M.L., G.F., and S.J.R. Interpretation of data: R.N., M.L., G.F., S.J.R., J.M., P.B., J.B., and K.M. Writing the manuscript, approval of the version to be published, and agreement to be accountable for all aspects of the work: all authors.

and J.B. Acquisition of data: R.N., K.M., P.B., and J.B. Analysis of data: R.N., M.L., G.F., and S.J.R. Interpretation of data: R.N., M.L., G.F., S.J.R., J.M., P.B., J.B., and K.M. Writing the manuscript, approval of the version to be published, and agreement to be accountable for all aspects of the work: all authors. This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org. Originally Published in Press as DOI: 10.1164/rccm.201805-0936OC on August 24, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

At a Glance Commentary Scientific Knowledge on the Subject Several studies investigating corticosteroids in septic shock have found beneficial effects on shock duration. However, effects on survival are varied, with some studies reporting a survival advantage but others finding no benefit. The role of corticosteroids in septic shock remains uncertain, and the reasons for the variation in study outcomes remain unclear. Recently, two transcriptomic sepsis response signatures (SRSs) have been associated with immune function and outcome in sepsis. What This Study Adds to the Field This is the first study to examine, in septic shock, the interaction between SRS endotypes, and the response to norepinephrine or vasopressin, or to corticosteroids, in the context of a randomized trial. Although SRS endotype had no influence on mortality from sepsis based on vasopressor choice, there was a significant interaction between SRS endotype and treatment with hydrocortisone with regard to mortality. Patients with the immunocompetent SRS2 endotype who were treated with corticosteroids had poorer survival than those given placebo. SRS endotype at the onset of septic shock appears to influence response to corticosteroids. This finding may account for the variation in survival benefit attributed to corticosteroids if varying proportions of the SRS endoypes were recruited into previous trials.

osteroids had poorer survival than those given placebo. SRS endotype at the onset of septic shock appears to influence response to corticosteroids. This finding may account for the variation in survival benefit attributed to corticosteroids if varying proportions of the SRS endoypes were recruited into previous trials. Sepsis is defined as life-threatening organ dysfunction due to a dysregulated host response to infection (1), and is a major global health problem. Current treatment of septic shock relies on antibiotics, fluids, and vasopressors. No new or specific treatments for sepsis are in routine clinical practice. Corticosteroids have been proposed as an adjunctive treatment for septic shock. However, results of clinical trials have been contradictory regarding their impact on outcomes. Recently, two large clinical trials have been published examining the effect of corticosteroids on mortality in septic shock. The ADRENAL (Adjunctive Glucocorticoid Therapy in Patients with Septic Shock) study (2) compared a hydrocortisone infusion to placebo, whereas APROCCHSS (Hydrocortisone Plus Fludrocortisone for Adults with Septic Shock) (3) used a combination of hydrocortisone and fludrocortisone. Both showed clear benefits of corticosteroids on cardiovascular outcomes, demonstrated by shorter durations of shock; however, the effects on survival were inconsistent, with APROCCHSS reporting improved survival with corticosteroid treatment and ADRENAL reporting no difference. We hypothesize that variation in underlying patient phenotypes may account for these differences in outcome.

omes, demonstrated by shorter durations of shock; however, the effects on survival were inconsistent, with APROCCHSS reporting improved survival with corticosteroid treatment and ADRENAL reporting no difference. We hypothesize that variation in underlying patient phenotypes may account for these differences in outcome. We have previously identified two transcriptomic sepsis response signatures (SRSs) based on genome-wide expression profiling in patients with sepsis with community acquired pneumonia (4) and fecal peritonitis (5). SRS1 is a relatively immunosuppressed phenotype that is associated with increased mortality, whereas SRS2 is relatively immunocompetent. In this report, we describe the stratification of patients enrolled into the VANISH (Vasopressin vs. Norepinephrine as Initial Therapy in Septic Shock) clinical trial (6), comparing norepinephrine to vasopressin with or without hydrocortisone for the treatment of septic shock, based on their SRS endotypes. We aimed to determine if transcriptomic phenotype was associated with response to either norepinephrine or vasopressin, or to corticosteroids.

rapy in Septic Shock) clinical trial (6), comparing norepinephrine to vasopressin with or without hydrocortisone for the treatment of septic shock, based on their SRS endotypes. We aimed to determine if transcriptomic phenotype was associated with response to either norepinephrine or vasopressin, or to corticosteroids. Methods Study Design and Sample Collection Full details of patient selection and treatment allocation can be found in the online supplement. Patients were recruited into the VANISH trial as previously described (6, 7). The VANISH trial was a factorial (2 × 2), multicenter, double-blind, randomized clinical trial conducted in 18 intensive care units in the United Kingdom between February 2013 and May 2015, with a primary outcome of kidney failure–free days up to Day 28. The trial was approved by the Oxford A research ethics committee, and written consent was obtained from patients or their legal representatives. Adults with septic shock and who required vasopressors were eligible for the trial and were recruited within 6 hours of the onset of septic shock. Patients were randomized to receive either a blinded infusion of vasopressin or norepinephrine (study drug 1), this was titrated to maintain the target mean arterial pressure. Only if the maximum infusion rate of study drug 1 was reached did patients receive the blinded study drug 2, either hydrocortisone or placebo, as previously reported (8). Blood samples for RNA analysis were collected on the day of enrolment into the trial in 10 centers when research staff members were available, and RNA was extracted as described in the online supplement.

reached did patients receive the blinded study drug 2, either hydrocortisone or placebo, as previously reported (8). Blood samples for RNA analysis were collected on the day of enrolment into the trial in 10 centers when research staff members were available, and RNA was extracted as described in the online supplement. Outcomes The primary outcome for this analysis was survival at 28 days. Secondary outcomes were kidney failure–free days up to Day 28, intensive care unit and hospital mortality, rates of kidney failure, weaning from vasopressors for greater than 24 hours, time to shock reversal, duration of mechanical ventilation, and mean total Sequential Organ Failure Assessment score (SOFA).

t 28 days. Secondary outcomes were kidney failure–free days up to Day 28, intensive care unit and hospital mortality, rates of kidney failure, weaning from vasopressors for greater than 24 hours, time to shock reversal, duration of mechanical ventilation, and mean total Sequential Organ Failure Assessment score (SOFA). Statistical Analysis Patients were allocated to either SRS1 or SRS2 using a generalized linear model based on the set of seven genes (DYRK2, CCNB1IP1, TDRD9, ZAP70, ARL14EP, MDC1, and ADGRE3) derived from the previous study of patients with sepsis due to community acquired pneumonia (4, 5). Differential expression analysis was performed using the limma R package (9). For the primary outcome, SRS endotype and drug interaction was explored using binary logistic regression with an interaction term, and differences in survival were displayed using Kaplan-Meier curves using log-rank tests for significance, and Renyi tests when survival curves crossed. As numbers were low in some treatment subgroups, the analysis was repeated using exact logistic regression analysis as a sensitivity analysis (10). As randomization was not stratified by SRS endotype and RNA was only analyzed in a sample of patients, there may be imbalances of potential confounders (age, sex, acute illness score [APACHE (Acute Physiology and Chronic Health Evaluation) II score]), and comorbidities (ischemic heart disease, severe chronic obstructive pulmonary disease, chronic renal failure, cirrhosis, cancer, immunosuppression, and diabetes), so multivariable logistic regression was performed as a sensitivity analysis (5). For both comparisons, vasopressin versus norepinephrine and hydrocortisone versus placebo, only patients who received the study drug as allocated were included, as described in the per-protocol analysis in the primary analysis (6). Further details of the statistical analysis can be found in the online supplement.

(5). For both comparisons, vasopressin versus norepinephrine and hydrocortisone versus placebo, only patients who received the study drug as allocated were included, as described in the per-protocol analysis in the primary analysis (6). Further details of the statistical analysis can be found in the online supplement. Results Samples were available from 177 patients, Figure 1, but 1 patient was excluded, as the timing of the blood sample was recorded as 9 days after inclusion. The baseline characteristics of these patients were similar to the total trial population, and to those who did not have RNA sampling (see Table E1 in the online supplement). The 28-day mortality was also similar in those who did (27%) and did not (31%) have RNA samples taken (P = 0.43). Among these 176 patients, 83 (47%) were classified as SRS1 and 93 (53%) as SRS2. We compared global gene expression differences between the SRS endotypes in the VANISH patients to those observed in the derivation study of sepsis due to community-acquired pneumonia (4). We found that SRS, rather than study cohort, was the major driver of the observed variation in gene expression (Figure E1A), and that the differential gene expression results were strongly correlated (Pearson’s r = 0.858, P < 2.2 × 10−16; Figure E1B). Patients with SRS1 and SRS2 endotypes were similar with regard to demographics and baseline characteristics, with only a small difference in rates of ischemic heart disease (higher in SRS2) and serum lactate (higher in SRS1) (Table 1). Baseline characteristics were also similar when patients were stratified by SRS and treatment allocation (Table E2). The effect of vasopressor treatment on mortality at 28 days did not differ statistically between SRS groups (vasopressin vs. norepinephrine in SRS1, odds ratio [OR] = 1.50, 95% confidence interval [CI] = 0.58–3.88; SRS2, OR = 0.94, 95% CI = 0.36–2.46; interaction P = 0.50). However, in those patients who received the second study drug, either hydrocortisone or placebo, there was a statistically significant interaction between treatment and SRS endotype (hydrocortisone vs. placebo in SRS1, OR = 0.85, 95% CI = 0.30–2.43; SRS2, OR = 7.9, 95% CI = 1.6–39.9; interaction P = 0.02). Kaplan-Meier survival curves are shown in Figure 2. Similar results were obtained using exact logistic regression (hydrocortisone vs. placebo in SRS1, OR = 0.85, 95% CI = 0.26–2.73; SRS2, OR = 7.67, 95% CI = 1.45–78.8; interaction P = 0.046).

95% CI = 0.30–2.43; SRS2, OR = 7.9, 95% CI = 1.6–39.9; interaction P = 0.02). Kaplan-Meier survival curves are shown in Figure 2. Similar results were obtained using exact logistic regression (hydrocortisone vs. placebo in SRS1, OR = 0.85, 95% CI = 0.26–2.73; SRS2, OR = 7.67, 95% CI = 1.45–78.8; interaction P = 0.046). After adjustment for age, sex, disease severity (APACHE II), and comorbidities in multiple logistic regression, hydrocortisone use continued to be associated with increased mortality in those with an SRS2 phenotype (adjusted OR = 8.3, 95% CI = 1.4–47.8), and the treatment by SRS endotype interaction remained significant (interaction P = 0.03). In patients who received placebo, mortality was lower in those with the SRS2 compared with SRS1 endotype (unadjusted OR = 0.15, 95% CI = 0.03–0.76, P = 0.02; adjusted OR = 0.13, 95% CI = 0.02–0.74, P = 0.02; Figure E2), consistent with mortality differences associated with SRS endotypes reported previously (4, 5). Rates and duration of renal failure, and proportions of patients successfully weaned off vasopressors were similar based on SRS and study drug 2 combination (Table 2). Within both SRS1 and SRS2, those patients given hydrocortisone tended to be weaned more quickly from vasopressors (SRS1, HR = 1.3, 95% CI = 0.8–2.3; SRS2, HR = 1.1, 95% CI = 0.6–1.9), although the CIs clearly include 1. Rates of all serious adverse events in the study were the same between SRS endotypes (6 [7%] SRS1 vs. 6 [6%] SRS2, P = 0.84).

, those patients given hydrocortisone tended to be weaned more quickly from vasopressors (SRS1, HR = 1.3, 95% CI = 0.8–2.3; SRS2, HR = 1.1, 95% CI = 0.6–1.9), although the CIs clearly include 1. Rates of all serious adverse events in the study were the same between SRS endotypes (6 [7%] SRS1 vs. 6 [6%] SRS2, P = 0.84). Figure 1. Recruitment, randomization, treatment allocation, and RNA sampling in the VANISH (Vasopressin vs. Norepinephrine as Initial Therapy in Septic Shock) clinical trial. SRS = sepsis response signature. Table 1. Comparison of Baseline Characteristics of Patients with Sepsis Response Signature 1 and 2 Phenotypes

, those patients given hydrocortisone tended to be weaned more quickly from vasopressors (SRS1, HR = 1.3, 95% CI = 0.8–2.3; SRS2, HR = 1.1, 95% CI = 0.6–1.9), although the CIs clearly include 1. Rates of all serious adverse events in the study were the same between SRS endotypes (6 [7%] SRS1 vs. 6 [6%] SRS2, P = 0.84). Figure 1. Recruitment, randomization, treatment allocation, and RNA sampling in the VANISH (Vasopressin vs. Norepinephrine as Initial Therapy in Septic Shock) clinical trial. SRS = sepsis response signature. Table 1. Comparison of Baseline Characteristics of Patients with Sepsis Response Signature 1 and 2 Phenotypes Characteristics SRS1 SRS2 P Value n 83 93 — Age, median (IQR), yr 66 (53–78) 63 (53–75) 0.40 Men, n/total (%) 55/83 (66) 54/93 (58) 0.26 Weight, median (IQR), kg 75 (65–88) 74 (61–92) 0.72 BMI, median (IQR) 26 (23–31) 27 (22–32) 0.68 White race, n/total (%) 70/83 (84) 74/93 (80) 0.41 Recent surgical history, n/total (%) 15/83 (18) 12/93 (13) 0.34 APACHE II score, median (IQR) 23 (20–30) 24 (19–31) 0.70 Preexisting conditions, n/total (%) Ischemic heart disease 8/83 (10) 21/93 (23) 0.02 Severe COPD 5/83 (6) 5/93 (5) 0.85 Chronic kidney failure 4/83 (5) 4/93 (4) 0.87 Cirrhosis 3/83 (4) 8/93 (9) 0.17 Cancer 12/83 (14) 10/93 (11) 0.46 Immunocompromised 7/83 (8) 3/93 (3) 0.14 Diabetes 16/83 (19) 24/93 (26) 0.30 Organ failure, n/total (%) Respiratory 33/83 (40) 31/91 (34) 0.44 Kidney 18/83 (22) 22/93 (24) 0.76 Liver 4/73 (5) 8/82 (10) 0.32 Hematological 4/79 (5) 5/92 (5) 0.91 Neurological 27/79 (34) 29/90 (32) 0.79 Physiological variables, median (IQR) Mean arterial pressure, mm Hg 70.0 (64.0–76.0) 67.0 (60.5–75.0) 0.16 Heart rate, beats/min 96.0 (85.0–112.0) 92.0 (80.5–104.0) 0.10 Central venous pressure, mm Hg 14 (10–19) 13 (9–18) 0.09 Lactate, mmol/L 2.8 (1.8–4.9) 1.9 (1.3–3.3) 0.001 PaO2/FIO2, mm Hg 197 (122–322) 195 (137–299) 0.96 Creatinine, mg/dl 1.3 (1.0–2.1) 1.4 (0.8–2.3) 0.64 Bilirubin, mg/dl 1.0 (0.5–2.1) 0.7 (0.5–1.3) 0.10 Platelets, ×103/μl 192 (121–267) 187 (120–291) 0.98 GCS 14.0 (6.0–15.0) 13.5 (3.0–15.0) 0.70 Mechanical ventilation, n/total (%) 42/83 (51) 54/93 (58) 0.32 Renal replacement therapy, n/total (%) 3/83 (4) 2/93 (2) 0.56 Volume of i.v.

g/dl 1.3 (1.0–2.1) 1.4 (0.8–2.3) 0.64 Bilirubin, mg/dl 1.0 (0.5–2.1) 0.7 (0.5–1.3) 0.10 Platelets, ×103/μl 192 (121–267) 187 (120–291) 0.98 GCS 14.0 (6.0–15.0) 13.5 (3.0–15.0) 0.70 Mechanical ventilation, n/total (%) 42/83 (51) 54/93 (58) 0.32 Renal replacement therapy, n/total (%) 3/83 (4) 2/93 (2) 0.56 Volume of i.v. fluid in previous 4 h, median (IQR), ml 1,255 (547–2,054) 1,003 (557–1,665) 0.09 Patients receiving open-label vasopressor at randomization, n/total (%) 72/83 (87) 81/93 (87) 0.95 Time from onset of shock to receiving first study drug, median (IQR), h 4.0 (1.8–5.5) 3.4 (2.0–4.9) 0.44 Norepinephrine dose at randomization, median (IQR), μg/kg/min 0.16 (0.10–0.28) 0.14 (0.08–0.25) 0.25 Source of infection, n/total (%) Lung 32/82 (39) 45/91 (49) 0.17 Abdomen 21/82 (26) 15/91 (16) 0.14 Soft tissue or line 1/82 (1) 4/91 (4) 0.21 Other 28/82 (34) 27/91 (30) 0.53 Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; BMI = body mass index; COPD = chronic obstructive pulmonary disease; GCS = Glasgow Coma Score; IQR = interquartile range; i.v. = intravenous; SRS = sepsis response signature.

ine 1/82 (1) 4/91 (4) 0.21 Other 28/82 (34) 27/91 (30) 0.53 Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; BMI = body mass index; COPD = chronic obstructive pulmonary disease; GCS = Glasgow Coma Score; IQR = interquartile range; i.v. = intravenous; SRS = sepsis response signature. P values are from Mann-Whitney U tests for continuous variables and Pearson’s χ2 tests for binary variables. Missing data are shown in Table E3 in the online supplement. For the APACHE score, range 0–72, a higher score corresponds to more severe illness and a higher risk of death; for GCS, range 3–15, a lower score corresponds to a greater depression of consciousness; BMI was calculated as weight in kilograms divided by height in meters squared. Bold type indicates P < 0.05. Figure 2. Kaplan-Meier survival curves comparing survival with (A) norepinephrine (red line) and vasopressin (green line) and (B) hydrocortisone (red line) and placebo (green line) in sepsis response signature (SRS) 1 and SRS2. Crosses represent censored patients (n = 2 for SRS1 vasopressin, n = 1 for SRS1 placebo, and n = 1 for SRS1 hydrocortisone; all other patients were censored at death or Day 29). Table 2. Comparison of Outcomes of Patients with Sepsis Response Signature 1 and 2 Phenotypes Given Either Hydrocortisone or Placebo as Study Drug 2

Figure 2. Kaplan-Meier survival curves comparing survival with (A) norepinephrine (red line) and vasopressin (green line) and (B) hydrocortisone (red line) and placebo (green line) in sepsis response signature (SRS) 1 and SRS2. Crosses represent censored patients (n = 2 for SRS1 vasopressin, n = 1 for SRS1 placebo, and n = 1 for SRS1 hydrocortisone; all other patients were censored at death or Day 29). Table 2. Comparison of Outcomes of Patients with Sepsis Response Signature 1 and 2 Phenotypes Given Either Hydrocortisone or Placebo as Study Drug 2 SRS1 SRS2 P Value for Interaction Hydrocortisone Placebo Hydrocortisone Placebo n 27 35 31 24 Kidney failure–free days, median (IQR), d 25 (1–28) 25 (9–28) 25 (4–28) 28 (25–28) 0.43* 28-d mortality, n/total (%) 9/27 (33) 13/35 (37) 13/31 (42) 2/24 (8) 0.02† ICU mortality, n/total (%) 7/27 (26) 9/35 (26) 11/31 (35) 2/24 (8) 0.08† Hospital mortality, n/total (%) 8/27 (30) 12/35 (34) 13/31 (42) 2/24 (8) 0.02† Kidney failure, n/total (%) 13/27 (48) 19/35 (54) 17/31 (55) 10/24 (42) 0.30† No. weaned from vasopressors for >24 h, n/total (%) 25/27 (93) 31/35 (89) 28/31 (90) 23/24 (96) 0.36† Time to shock reversal, median (IQR), h 30.6 (18.1–77.7) 43.8 (21.5–91.5) 58.9 (36.1–82.3) 89.5 (31.5–122.0) 0.60‡ Duration of mechanical ventilation, median (IQR), d 3.0 (2.0–12.0) 6.0 (2.0–11.5) 6.0 (2.0–14.5) 9.0 (6.0–20.0) 0.67* Mean total SOFA score, median (IQR) 5.7 (3.6–9.0) 4.9 (3.6–7.2) 5.6 (3.7–8.3) 4.7 (3.5–6.3) 0.72§ Definition of abbreviations: IQR = interquartile range; SOFA = Sequential Organ Failure Assessment score; SRS = sepsis response signature.

, median (IQR), d 3.0 (2.0–12.0) 6.0 (2.0–11.5) 6.0 (2.0–14.5) 9.0 (6.0–20.0) 0.67* Mean total SOFA score, median (IQR) 5.7 (3.6–9.0) 4.9 (3.6–7.2) 5.6 (3.7–8.3) 4.7 (3.5–6.3) 0.72§ Definition of abbreviations: IQR = interquartile range; SOFA = Sequential Organ Failure Assessment score; SRS = sepsis response signature. Bold type indicates P < 0.05. * From the aligned rank transform test. † From logistic regression. ‡ From Cox regression, treating deaths as never having the event of interest. Results were similar treating death as a competing risk. § From linear regression, applying a square root transform to the outcome.

, median (IQR), d 3.0 (2.0–12.0) 6.0 (2.0–11.5) 6.0 (2.0–14.5) 9.0 (6.0–20.0) 0.67* Mean total SOFA score, median (IQR) 5.7 (3.6–9.0) 4.9 (3.6–7.2) 5.6 (3.7–8.3) 4.7 (3.5–6.3) 0.72§ Definition of abbreviations: IQR = interquartile range; SOFA = Sequential Organ Failure Assessment score; SRS = sepsis response signature. Bold type indicates P < 0.05. * From the aligned rank transform test. † From logistic regression. ‡ From Cox regression, treating deaths as never having the event of interest. Results were similar treating death as a competing risk. § From linear regression, applying a square root transform to the outcome. Discussion We were able to identify the two previously identified SRS endotypes within this septic shock population due to diverse etiologies in the VANISH clinical trial. In this study, a higher proportion of patients had the SRS1 endotype (47%) than in either the derivation (41%) or validation (35%) cohorts described in the original study (4). However, the original data were derived from a sepsis population where only a portion had septic shock, with under half requiring vasopressors. Vasopressor use and SOFA score were higher in SRS1 patients in the derivation study, suggesting more severe disease. It is therefore unsurprising that, in a sicker population of patients, all of whom had septic shock, the SRS1 endotype is more commonly represented. Importantly, although the two SRS endotypes have previously been described in both community-acquired pneumonia (4) and fecal peritonitis (5), this is the first time the endotypes have been demonstrated in patients with sepsis due to multiple different sources of infection.

e SRS1 endotype is more commonly represented. Importantly, although the two SRS endotypes have previously been described in both community-acquired pneumonia (4) and fecal peritonitis (5), this is the first time the endotypes have been demonstrated in patients with sepsis due to multiple different sources of infection. Transcriptomic profile at the onset of septic shock was associated with response to corticosteroids, but not vasopressin or norepinephrine. Those patients with the SRS2 endotype had significantly higher mortality when given corticosteroids compared with placebo. However, this effect on mortality was not seen in those with the SRS1 endotype. Previous work (4, 5) demonstrated that the SRS2 endotype was associated with a significantly lower mortality rate than SRS1. In patients with sepsis with pneumonia, 28-day mortality was 17% in SRS2 compared with 27% in SRS1 (4), and, in fecal peritonitis, 28-day mortality was 7.2% versus 20.8% for SRS2 and SRS1, respectively (5). This pattern was again seen in the current study when only those patients randomized to placebo were considered, where 28-day mortality was lower in SRS2 (8%) compared with SRS1 (37%). As inclusion/exclusion criteria and illness severity vary between different studies, actual mortality rates will inevitably vary.

(5). This pattern was again seen in the current study when only those patients randomized to placebo were considered, where 28-day mortality was lower in SRS2 (8%) compared with SRS1 (37%). As inclusion/exclusion criteria and illness severity vary between different studies, actual mortality rates will inevitably vary. Steroids have a clear benefit on time-to-shock resolution, reported in multiple clinical trials (2, 3, 8, 11). Despite this improvement in an important physiological measure, the overall effect on patient survival has been inconsistent between trials. Differences in the mortality effects of steroids in the recent ADRENAL (2) and APROCCHSS (3) trials may be explained by the current findings. If these trials recruited different proportions of patients with the two SRS endotypes, a trial with a greater proportion of SRS2 patients may find no survival advantage or may find harm due to steroids in septic shock. In observational studies, SRS1 has been associated with higher mortality than SRS2, and similar effects were seen in placebo patients in this trial. Overall, the mortality in the ADRENAL trial was lower than that seen in the APROCCHSS trial (28% vs. 46%, respectively, at Day 90), perhaps suggesting that a higher proportion of SRS2 patients may have been recruited. If the SRS2 patients are harmed with steroid treatment, it may explain why, overall, no mortality benefit was seen in the ADRENAL trial, despite improvement in shock resolution. Interestingly, the duration of shock tended to be shorter among both SRS1 and SRS2 patients successfully weaned from vasopressors, although the CIs are wide, possibly due to the small numbers in each subgroup.

ain why, overall, no mortality benefit was seen in the ADRENAL trial, despite improvement in shock resolution. Interestingly, the duration of shock tended to be shorter among both SRS1 and SRS2 patients successfully weaned from vasopressors, although the CIs are wide, possibly due to the small numbers in each subgroup. Because of the many mechanisms of action of corticosteroids, we can only speculate as to why the effect of steroids on mortality should vary between SRS endotypes. The SRS1 endotype has been shown to be a relatively immunosuppressed phenotype with features of endotoxin tolerance, T cell exhaustion, and downregulation of major histocompatibility class (MHC) II antigens, and is associated with higher mortality rates. The SRS2 endotype in contrast is relatively more immunocompetent and associated with lower mortality rates. Of particular interest is the upregulation of MHC II in SRS2 (4). Reduction in HLA-DR expression in sepsis has been associated with higher rates of nosocomial infection and worse survival (12), so it is plausible that improvement in antigen presentation improves immune function and bacterial clearance in the SRS2 group, improving survival compared with SRS1. However, corticosteroids are recognized to downregulate MHC II (12–14), which could provide a mechanism by which this protective advantage is removed. Corticosteroids also have actions affecting NF-κB, T cells, and apoptosis (15, 16), all of which showed evidence of differential expression between the SRS endotypes. Altered modulation of these pathways could account for different degrees of immunosuppression caused by corticosteroids between the SRS endotypes. It is therefore possible that corticosteroids may have beneficial cardiovascular effects in all patients with septic shock, but that the well-known immunosuppressive adverse effects of corticosteroids are only realized in the SRS2 patients. Immune dysfunction is recognized to increase the risk of nosocomial infection and to be associated with higher rates of mortality (17), yet clinical scoring systems, such as SOFA, do not include the immune system. This may account for why no difference in total SOFA score was seen despite mortality differences between treatment groups in our study.

ed to increase the risk of nosocomial infection and to be associated with higher rates of mortality (17), yet clinical scoring systems, such as SOFA, do not include the immune system. This may account for why no difference in total SOFA score was seen despite mortality differences between treatment groups in our study. Transcriptomic profiles in sepsis and response to corticosteroid therapy have been studied in children (18). In this previous study, a subgroup of patients was identified using RNA expression that was also associated with worse outcomes when patients were treated with corticosteroids, although this subgroup was the more immunosuppressed phenotype. However, as previously described (5, 19), the gene expression profiles appear to be different in adult and pediatric populations, with those in children being based, in part, on genes linked to glucocorticoid receptors. In the pediatric study corticosteroid treatment was based on physician choice rather than randomized allocation as in this clinical trial.

9), the gene expression profiles appear to be different in adult and pediatric populations, with those in children being based, in part, on genes linked to glucocorticoid receptors. In the pediatric study corticosteroid treatment was based on physician choice rather than randomized allocation as in this clinical trial. This study does have limitations. It is a post hoc analysis of samples collected as part of a clinical trial. Research blood sampling was not available in all centers, and, due to the emergency nature of the trial and the short recruitment window (maximum 6 h), it was not logistically possible to collect samples from all patients, thus limiting the sample size. However, the subset of patients in this analysis had similar baseline characteristics to the overall trial population, and the result was robust to adjustment for potential confounders and the use of statistical methods to account for small numbers. Although the analysis was post hoc, we used predefined endotype definitions based on previously published work and derived and validated in independent cohorts (4, 5). Importantly, treatment allocation was randomized and double blinded. Although the SRS endotypes are described according to their presumed immunological effects, this is based on gene expression data, and the absolute functional implications of the endotypes are still to be established. It is plausible that corticosteroids interact with SRS endotypes in ways that cannot be appreciated from transcriptomic data alone.

e described according to their presumed immunological effects, this is based on gene expression data, and the absolute functional implications of the endotypes are still to be established. It is plausible that corticosteroids interact with SRS endotypes in ways that cannot be appreciated from transcriptomic data alone. Although further work is required to validate these findings and to better understand the utility of endotype assignment based on transcriptomic profiles in sepsis, our findings suggest that SRS endotypes should be used in future biomarker-guided trials of corticosteroids in septic shock. Supported by the UK National Institute for Health Research (NIHR) under Research for Patient Benefit program grant PB-PG-0610-22350, NIHR Clinician Scientist Award NIHR/CS/009/007, and NIHR Research Professor award RP-2015-06-018 (A.C.G.); also supported by the NIHR Imperial Biomedical Research Centre, the UK Intensive Care Foundation, Wellcome Trust grant 090532/Z/09/Z to core facilities at the Wellcome Centre for Human Genetics, Wellcome Trust Investigator Award 204969/Z/16/Z (J.C.K.), and by the NIHR Oxford Biomedical Research Centre. The views expressed in this article 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. The funders of the study had no role in design or conduct of the study, collection, management, analysis, or interpretation of the data, or preparation, review, or approval of the manuscript, or the decision to submit for publication.

lth Service, the National Institute for Health Research, or the Department of Health. The funders of the study had no role in design or conduct of the study, collection, management, analysis, or interpretation of the data, or preparation, review, or approval of the manuscript, or the decision to submit for publication. Author Contributions: A.C.G. had full access to all of the data in the study, takes responsibility for the integrity of the data and the accuracy of the data analysis, and had final responsibility for the final decision to submit for publication. Study concept and design—J.C.K. and A.C.G.; acquisition, analysis, or interpretation of data—all authors; drafting of the manuscript—D.B.A. and A.C.G.; critical revision of the manuscript for important intellectual content—all authors; statistical analysis—D.B.A., K.L.B., S.S., D.A., J.C.K., and A.C.G.; obtained funding—A.C.G.; administrative, technical, or material support—D.B.A., K.L.B., F.A.-B., J.C.K., and A.C.G.; study supervision—S.J.B., D.A., and A.C.G. The gene expression data are available on ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) as of the time of publication, and the coefficients for the general linear model used to assign sepsis response signature class are also available. Individual participant data that underlie the results in this article, after deidentification (text, table, and figures), are available on request to anthony.gordon@imperial.ac.uk (data requesters will need to sign a data transfer agreement with Imperial College London).

sis response signature class are also available. Individual participant data that underlie the results in this article, after deidentification (text, table, and figures), are available on request to anthony.gordon@imperial.ac.uk (data requesters will need to sign a data transfer agreement with Imperial College London). This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org. Originally Published in Press as DOI: 10.1164/rccm.201807-1419OC on October 26, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

To the Editor: The human nasopharynx is frequently colonized by Streptococcus pneumoniae (the pneumococcus), serving as the reservoir for transmission, a state that necessarily precedes invasive pneumococcal infection. Influenza infection increases pneumococcal colonization density and dysregulates host immune responses, increasing the risk of secondary bacterial pneumonia and death (1–3). Live attenuated influenza vaccine (LAIV) nasal spray has been used in the United States since 2003, and it has reduced severe influenza disease in the United Kingdom since its introduction in 2013 into the national pediatric immunization program. In mice, LAIV vaccination increases the density and duration of pneumococcal colonization (2) and rates of otitis media. In children, LAIV is associated with increased rates and density of bacterial colonization (4). Although LAIV is safe and not associated with increases in pneumococcal disease, these data suggest that it could increase pneumococcal transmission to susceptible individuals (5).

colonization (2) and rates of otitis media. In children, LAIV is associated with increased rates and density of bacterial colonization (4). Although LAIV is safe and not associated with increases in pneumococcal disease, these data suggest that it could increase pneumococcal transmission to susceptible individuals (5). We therefore undertook two trials (EudraCT 2014-004634-26) using an established human challenge model to evaluate the effects of LAIV on the dynamics of pneumococcal colonization. Some of the results of these studies have been previously reported in the form of a preprint (https://doi.org/10.1101/343319). An extensive immunological investigation to accompany these clinical data has been published (6). Healthy nonsmoking volunteers, 18–50 years old, consented to participate in double-blinded, randomized, placebo-controlled trials reflecting alternative scenarios: 1) immunization first (LAIV precedes nasopharyngeal inoculation with pneumococcus by 3 days) and 2) colonization first (LAIV is administered 3 days after colonization with pneumococcus). The participants, who were uncolonized at baseline, randomly received either intervention (nasal LAIV paired with intramuscular placebo of normal saline; AstraZeneca) or control (nasal placebo of normal saline paired with intramuscular influenza vaccination [Fluarix Tetra; GlaxoSmithKline]) with concealment by blindfolding. All of the participants gave written informed consent, with approval from the North West NHS Research Ethics Committee (14/NW/1460).

normal saline; AstraZeneca) or control (nasal placebo of normal saline paired with intramuscular influenza vaccination [Fluarix Tetra; GlaxoSmithKline]) with concealment by blindfolding. All of the participants gave written informed consent, with approval from the North West NHS Research Ethics Committee (14/NW/1460). All of the participants were inoculated with S. pneumoniae serotype 6B strain BHN418 (80,000 cfu per nostril) in 0.1 ml solution (7). (The S. pneumoniae BHN418 sequence [GI:557376079] is available from https://www.ncbi.nlm.nih.gov/nuccore/557376079). “Colonization positivity” was determined by serial nasal washes and defined by detection of serotype 6B by culture at a programmed time point from 2 to 29 days (7, 8). In parallel, PCR detection of pneumococcal lytA was performed. In the “immunization first” study, LAIV vaccination preceded pneumococcal inoculation by 3 days (primary endpoint: colonization rate). This order was reversed for the “colonization first” study (primary endpoint: area under the curve [AUC] of bacterial density between Days 2 and 14). Results are presented as modified intention to treat, excluding those who did not receive immunization or inoculation per protocol, or did not complete follow-up. Generalized linear models were used to compare colonization positivity, duration of colonization, and AUC bacterial density, with generalized estimating equations used for comparison at multiple time points. Full methodological and other details are available online in the form of a preprint (https://doi.org/10.1101/343319).

lized linear models were used to compare colonization positivity, duration of colonization, and AUC bacterial density, with generalized estimating equations used for comparison at multiple time points. Full methodological and other details are available online in the form of a preprint (https://doi.org/10.1101/343319). In the “immunization first” study (Figure 1), we enrolled 202 participants; 130 of these subjects were inoculated and 117 were analyzed (n = 55 LAIV, n = 62 control; overall mean age, 20 yr [range, 18–48 yr]; 58% female). Pneumococcal colonization rates were similar in LAIV participants and control subjects (25/55 [45.5%] vs. 24/62 [38.7%]; odds ratio [OR], 1.32; P = 0.46), although the LAIV-treated group had consistently yet nonsignificantly higher rates at each time point. PCR detection rates were significantly higher in the LAIV group than in the control group at Day 2 (33/55 [60.0%] vs. 25/62 [40.3%]; OR, 2.22; P = 0.03). The median duration of colonization was not different between the groups by conventional microbiology (22 d [interquartile range (IQR), 22–29] and 22 d [IQR, 14–29] in the LAIV and control groups, respectively; P = 0.09) or PCR (median, 22 d [IQR, 7–29] LAIV vs. 14 d [IQR, 7–22] control; P = 0.45). Mean colonization densities were consistently increased in the LAIV group, with statistical significance at Day 9 representing a 10-fold (1 log10) increase in colonization density in the LAIV group (2.82 ± 1.78 vs. 1.81 ± 1.39 log10 titers, P = 0.03; Figure 1). PCR results showed the same pattern, with significantly higher densities in the LAIV group at Day 2 (P = 0.03).

ed in the LAIV group, with statistical significance at Day 9 representing a 10-fold (1 log10) increase in colonization density in the LAIV group (2.82 ± 1.78 vs. 1.81 ± 1.39 log10 titers, P = 0.03; Figure 1). PCR results showed the same pattern, with significantly higher densities in the LAIV group at Day 2 (P = 0.03). Figure 1. Immunization first: live attenuated influenza vaccine (LAIV) precedes nasopharyngeal inoculation with pneumococcus—effect on colonization dynamics of LAIV vaccination given at Day −3. Density dynamics after pneumococcal inoculation (on Day 0) are calculated from classical microbiology [log10(cfu/ml + 1)]. Mean density of Streptococcus pneumoniae for each nasal wash time point among participants in whom serotype 6B was detectable at any point. Bars represent SE. Inset: area under the curve (AUC) of density–time from Day 2 to Day 14 (box plot of median with interquartile range, with whiskers at 1.5× the interquartile range). *Statistically significant difference (P < 0.05).

al wash time point among participants in whom serotype 6B was detectable at any point. Bars represent SE. Inset: area under the curve (AUC) of density–time from Day 2 to Day 14 (box plot of median with interquartile range, with whiskers at 1.5× the interquartile range). *Statistically significant difference (P < 0.05). Four participants with laboratory-confirmed other viral infections (three influenza B in the control arm, one rhinovirus in the LAIV arm) had among the highest bacterial densities of their cohorts. Among pneumococcal-colonized individuals, the AUC of colonization density was higher in the LAIV group than in the control group, with borderline statistical significance at Days 2–14 (P = 0.05), and reached statistical significance after exclusion of participants who had nasal-swab PCR evidence of concurrent wild-type viral illness (three influenza B in the control arm, one rhinovirus in the LAIV arm; data not shown; P = 0.03) after presenting with symptoms of illness.

cal significance at Days 2–14 (P = 0.05), and reached statistical significance after exclusion of participants who had nasal-swab PCR evidence of concurrent wild-type viral illness (three influenza B in the control arm, one rhinovirus in the LAIV arm; data not shown; P = 0.03) after presenting with symptoms of illness. In the “colonization first” study (Figure 2), 316 participants consented, 206 were screened, and 163 participants were included in the modified intention-to-treat analysis (n = 73 LAIV, n = 90 control; overall mean age, 20 yr [range, 18–46 yr]; 55% female). Data from 17 participants (10%) were excluded owing to non-study-serotype S. pneumoniae colonization. AUC colonization densities for each time period were consistently lower in the LAIV group, although the difference was not statistically significant (P = 0.11 for Days 2–14 primary endpoint; Figure 2). By PCR, a significantly lower AUC was evident in the LAIV group compared with the control group on Days 2–27 (P = 0.03).

C colonization densities for each time period were consistently lower in the LAIV group, although the difference was not statistically significant (P = 0.11 for Days 2–14 primary endpoint; Figure 2). By PCR, a significantly lower AUC was evident in the LAIV group compared with the control group on Days 2–27 (P = 0.03). Figure 2. Colonization first: live attenuated influenza vaccine (LAIV) is administered to subjects already inoculated with pneumococcus—effect on colonization dynamics. Experimental inoculation to pneumococcus was performed on Day −3. Density dynamics after LAIV vaccination or control (on Day 0) are calculated from classical microbiology [log10(cfu/ml + 1)]. Mean density of Streptococcus pneumoniae for each nasal wash time point among participants in whom serotype 6B was detectable at any point. Bars represent SE. Inset: area under the curve (AUC) of density–time from Day 2 to Day 14 (the primary endpoint, box plot of median with interquartile range, with whiskers at 1.5× the interquartile range). Rates of colonization did not differ between the LAIV and control groups by conventional microbiology (36/73 [49.3%] vs. 45/90 [50.0%] respectively; OR, 0.97; P = 0.93). The median colonization duration did not differ between the two groups (21 vs. 27 d, P = 0.17) by conventional microbiology, although it was lower in the LAIV group by PCR (14 vs. 27 d, P = 0.001). There were no serious adverse events related to the intervention in either study.

Rates of colonization did not differ between the LAIV and control groups by conventional microbiology (36/73 [49.3%] vs. 45/90 [50.0%] respectively; OR, 0.97; P = 0.93). The median colonization duration did not differ between the two groups (21 vs. 27 d, P = 0.17) by conventional microbiology, although it was lower in the LAIV group by PCR (14 vs. 27 d, P = 0.001). There were no serious adverse events related to the intervention in either study. In the largest trial to date involving a controlled human coinfection model, we have studied for the first time the impact of coinfection of a live viral vaccine and a bacterial pathogen. Immunological parameters have been reported separately (6). Antecedent LAIV administration caused modest but significant transient effects on pneumococcal colonization, in keeping with a pediatric randomized controlled trial that showed an increased pneumococcal density after LAIV (2). In our study, the inverse scenario (LAIV after pneumococcal colonization) was associated with reduced colonization density and colonization rates at Day 27, decreased AUC, and earlier bacterial clearance.

keeping with a pediatric randomized controlled trial that showed an increased pneumococcal density after LAIV (2). In our study, the inverse scenario (LAIV after pneumococcal colonization) was associated with reduced colonization density and colonization rates at Day 27, decreased AUC, and earlier bacterial clearance. Our model, consistent with murine coinfection disease models, reinforces the notion that the precedence of pathogen exposure might determine disease outcome: pneumococcal infection after influenza might exacerbate disease, whereas pneumococcus infection preceding influenza might reduce mortality (9). We used complementary methods for bacterial detection: although PCR is more sensitive and could detect DNA in the absence of viable pathogen, the persistence beyond 2 days suggests lower-density colonization, which is unmeasurable by culture. These studies were limited by size and the evaluation of a single pneumococcal serotype in healthy adults likely to have neutralizing influenza antibodies. Any effect of LAIV in children may therefore be more pronounced owing to lower antibody titers, increased viral shedding, and higher natural rates of pneumococcal colonization acquisition. Future vaccine studies should evaluate the effect on pathogens not directly targeted by the vaccine, including their onward transmission. Acknowledgment The authors thank the Data Monitoring and Safety Committee (Brian Faragher, Christopher Green, and Robert C. Read).

These studies were limited by size and the evaluation of a single pneumococcal serotype in healthy adults likely to have neutralizing influenza antibodies. Any effect of LAIV in children may therefore be more pronounced owing to lower antibody titers, increased viral shedding, and higher natural rates of pneumococcal colonization acquisition. Future vaccine studies should evaluate the effect on pathogens not directly targeted by the vaccine, including their onward transmission. Acknowledgment The authors thank the Data Monitoring and Safety Committee (Brian Faragher, Christopher Green, and Robert C. Read). EHPC-LAIV Study Group: Jamie Rylance, Wouter A. A. de Steenhuijsen Piters, Sherin Pojar, Elissavet Nikolaou, Esther German, Elena Mitsi, Simon P. Jochems, Beatriz Carniel, Carla Solórzano, Jesús Reiné, Jenna F. Gritzfeld, Mei Ling J. N. Chu, Kayleigh Arp, Angela D. Hyder-Wright, Helen Hill, Caz Hales, Rachel Robinson, Cath Lowe, Hugh Adler, Seher Zaidi, Victoria Connor, Lepa Lazarova, Katherine Piddock, India Wheeler, Emma L. Smith, Ben Morton, John Blakey, Hassan Burhan, Artemis Koukounari, Duolao Wang, Michael J. Mina, Stephen B. Gordon, Debby Bogaert, Neil French, and Daniela Ferreira. Supported by the Bill and Melinda Gates Foundation and the UK Medical Research Council.

EHPC-LAIV Study Group: Jamie Rylance, Wouter A. A. de Steenhuijsen Piters, Sherin Pojar, Elissavet Nikolaou, Esther German, Elena Mitsi, Simon P. Jochems, Beatriz Carniel, Carla Solórzano, Jesús Reiné, Jenna F. Gritzfeld, Mei Ling J. N. Chu, Kayleigh Arp, Angela D. Hyder-Wright, Helen Hill, Caz Hales, Rachel Robinson, Cath Lowe, Hugh Adler, Seher Zaidi, Victoria Connor, Lepa Lazarova, Katherine Piddock, India Wheeler, Emma L. Smith, Ben Morton, John Blakey, Hassan Burhan, Artemis Koukounari, Duolao Wang, Michael J. Mina, Stephen B. Gordon, Debby Bogaert, Neil French, and Daniela Ferreira. Supported by the Bill and Melinda Gates Foundation and the UK Medical Research Council. Author Contributions: J.R., N.F., and D.M.F. designed the trial. J.R. conducted the trial according to the study protocol. J.R., W.A.A.d.S.P., M.J.M., D.B., N.F., and D.M.F. contributed to laboratory analysis, data interpretation, statistical analysis, and literature search. J.R., W.A.A.d.S.P., M.J.M., and D.M.F. drafted the report. All authors contributed to a critical review of the report. Originally Published in Press as DOI: 10.1164/rccm.201811-2081LE on February 13, 2019 Author disclosures are available with the text of this letter at www.atsjournals.org.

At a Glance Commentary Scientific Knowledge on the Subject Host immunity to Mycobacterium tuberculosis and the pathogenesis of cavitation, the key propagation mechanism of tuberculosis (TB), are poorly understood. There are hardly any data about the TB cavity because most studies have focused on the granuloma. What This Study Adds to the Field To our knowledge, this is the first study to interrogate the host transcriptome and characterize pathophysiological mechanisms at anatomically distinct locations within TB cavities. TB cavities were characterized by a centralized “sink” with profound downregulation of numerous immune pathways toward the cavity center, including a newly described neuroendocrine pathway, which correlated with poor bacterial containment.

ophysiological mechanisms at anatomically distinct locations within TB cavities. TB cavities were characterized by a centralized “sink” with profound downregulation of numerous immune pathways toward the cavity center, including a newly described neuroendocrine pathway, which correlated with poor bacterial containment. Tuberculosis (TB), first described in the Rigveda in India 3,500 years ago, has killed more than 1 billion people over the past two centuries, and remains the commonest infectious cause of death in many countries (1). Moreover, the advent of multidrug-resistant (MDR) TB threatens to wipe out recent gains in TB control (2, 3). An effective vaccine and/or immunotherapeutic intervention combined with interruption of transmission offer the only tangible hope of eliminating the disease. However, several recent TB vaccine candidates have failed to show clinical efficacy, or at best were only partially effective, despite promising data from animal studies (4–7). One reason may be lack of detailed knowledge about protective host immunity and the pathophysiology of cavitation. Previous work on granulomas emphasized the role of T-helper cell type 1 (Th1) immunity, CD4 T cells, IFN-γ signaling, IL-12, tumor necrosis factor (TNF)-α, eicosanoid signaling, and lipid dysregulation (8–12). One recent proteomics study found that the granuloma center was dominated by proinflammatory responses, whereas the area outside had an antiinflammatory signature (12). It is unclear if the same pathways drive cavitation or Mycobacterium tuberculosis (Mtb) replication in TB cavities. Cavitation underpins liquefactive necrosis, bacterial aerosolization, and hence disease transmission.

minated by proinflammatory responses, whereas the area outside had an antiinflammatory signature (12). It is unclear if the same pathways drive cavitation or Mycobacterium tuberculosis (Mtb) replication in TB cavities. Cavitation underpins liquefactive necrosis, bacterial aerosolization, and hence disease transmission. Although often arising from it, the pulmonary cavity is not the same as the granuloma. Over the last four centuries, the histology of TB cavities has been described based on autopsy work from Europe and the United States (13–15). Recent histological MDR-TB case reports from South Africa suggest a picture of failed immunity at the luminal edge of the TB lung cavity (16). The cause is unclear, but likely Mtb itself could have an immunosuppressive role via either release of mediators, as we have explored elsewhere, or by directly killing immune cells (17). But why, how, and exactly where does host immunity fail to control Mtb growth in this lesion? To answer these questions, we conducted a case-control clinical study where we performed RNA sequencing (RNA-Seq) on biopsies from anatomically distinct points within lung cavities of patients with MDR-TB and compared them with healthy lung tissue from those without TB (control subjects). Data were analyzed using standard modular and unsupervised approaches and subsequently modeled using our recently derived dynamical sink model (18).

iopsies from anatomically distinct points within lung cavities of patients with MDR-TB and compared them with healthy lung tissue from those without TB (control subjects). Data were analyzed using standard modular and unsupervised approaches and subsequently modeled using our recently derived dynamical sink model (18). Methods Patient Recruitment and Dissection Procedures Between 2012 and 2013, we recruited patients referred to Groote Schuur hospital for therapeutic surgical resection after failed MDR-TB chemotherapy (18). Control subjects were patients undergoing lung surgery for non-TB reasons, with no clinical or radiological features of TB. After surgery, the resected lung was immediately placed on ice, and transported to the BSL3 laboratory (∼200 m from the operating room).

c surgical resection after failed MDR-TB chemotherapy (18). Control subjects were patients undergoing lung surgery for non-TB reasons, with no clinical or radiological features of TB. After surgery, the resected lung was immediately placed on ice, and transported to the BSL3 laboratory (∼200 m from the operating room). Surgical dissection procedures were performed to avoid cross-contamination between biopsy positions, as previously described (18). Multiple 2-mm biopsies were taken at seven anatomically distinct sites within the cavity: 1) normal-appearing lung tissue 2–5 cm from the fibrotic cavity edge, 2) perifibrotic cavity edge, 3) center of the cavity wall, 4) luminal edge of the cavity wall, 5) air–caseum interface at cavity lumen center, 6) airways greater than or equal to 2.5 cm distal, and 7) proximal to cavity mouth; and 8) sputum. In control subjects, similarly sized biopsies were obtained from a single lung position. Each biopsy, including sputum, was 1) immediately placed in RNAlater for RNA-seq, 2) cultured for Mtb using mycobacterial growth indicator tube system, and 3) fixed in formalin for histopathology (hematoxylin and eosin staining) and immunohistochemistry. RNA extraction, sequencing and quality control were performed accordingly. Mycobacterial growth indicator tubes were monitored for time-to-positivity, and expressed as Mtb cfu/g of tissue. Immunohistochemistry was performed to corroborate RNA-Seq analysis findings. Full methodology is provided in the online supplement.

RNA extraction, sequencing and quality control were performed accordingly. Mycobacterial growth indicator tubes were monitored for time-to-positivity, and expressed as Mtb cfu/g of tissue. Immunohistochemistry was performed to corroborate RNA-Seq analysis findings. Full methodology is provided in the online supplement. Dual Positron Emission Tomography/Computed Tomography Scanning and Reading Positron emission tomography/computed tomography (PET-CT) imaging was performed in accordance with the globally accepted fluorodeoxyglucose PET/CT EANM procedure guidelines for tumor imaging (19). Measurements of cavity volume are detailed in the online supplement. Statistical and Bioinformatics Analyses Differentially expressed genes (DEGs), compared with non-TB control tissue, were defined as greater than log2 change and a Benjamini-Hochberg adjusted P less than 0.01. Modular analyses were performed using prevalidated and annotation-specific modules, as previously described (20–24). Reads were expressed per kilobase per million mapped reads (RPKM) for constituent genes of each module. Established immune pathways associated with DEGs were identified using ingenuity pathway analysis (IPA). We interrogated the whole transcriptome in an agnostic and unbiased fashion by identifying the most extensively changed physiological processes in the TB cavity.

ed reads (RPKM) for constituent genes of each module. Established immune pathways associated with DEGs were identified using ingenuity pathway analysis (IPA). We interrogated the whole transcriptome in an agnostic and unbiased fashion by identifying the most extensively changed physiological processes in the TB cavity. Mathematical Model RNA-Seq data were modeled to explore the relationship with bacterial burden and to integrate immune pathway interactions. Given that PET-CT scans and histological and RNA-Seq analyses revealed pathophysiological patterns consistent with our recently derived dynamical “sink” model (18), we modified standard linear models to incorporate these dynamical “sinks.” In a dynamical system, a point or state (e.g., cavity position) evolves over time according to specified rules (25). We modified the state to evolve over distance in the cavity rather than time. This model was then used to map the interaction of different pathways along the cavity positions, as detailed in the online supplement.

ical system, a point or state (e.g., cavity position) evolves over time according to specified rules (25). We modified the state to evolve over distance in the cavity rather than time. This model was then used to map the interaction of different pathways along the cavity positions, as detailed in the online supplement. Results Clinical and Radiological Characteristics of Patients and Control Subjects Clinical features describing the 14 patients with MDR-TB (11 MDR plus resistance to other drugs) are shown in Table E1 in the online supplement. The median (range) age was 33 (14–50) years. Two patients were HIV coinfected, but were on effective antiretroviral therapy. Typical preoperative PET-CT scans showed lesions consisting of consolidation with central cavitation and a median (range) lung cavity volume of 50 cm3 (15–389 cm3) (see Figure E1 and Table E1). Among the 10 control patients without TB, the median age was 30 (23–74) years. One control subject had HIV-infection and was on antiretroviral therapy.

CT scans showed lesions consisting of consolidation with central cavitation and a median (range) lung cavity volume of 50 cm3 (15–389 cm3) (see Figure E1 and Table E1). Among the 10 control patients without TB, the median age was 30 (23–74) years. One control subject had HIV-infection and was on antiretroviral therapy. Pathological and Microbiological Characteristics Gross pathological examination of resected TB-infected lungs revealed a median of two (one to three) cavities per resected specimen, with a diameter of 4 cm (2–8 cm) per cavity. Hematoxylin and eosin stains revealed site-specific histology (see Figure E2 and Table E1). The cavity wall was characterized by fibrosis and chronic inflammation (20 ± 7.7% and 35 ± 11% of the biopsied area, respectively; P = 0.014). Cell populations in the cavity wall included 14% (4.2–40%) neutrophils and 20% (8.9–40%) histiocytes (P = 0.34). Mtb could be visualized at all cavity positions, including normal-appearing tissue, both extracellularly and within neutrophils or macrophages. Lung tissue in control patients demonstrated normal histology. Mtb was culturable at all cavity positions, including normal-appearing lung tissue. However, the highest bacterial burden (i.e., lowest time-to-positivity) was at the air–caseum interface. No Mtb growth occurred in lung tissue from control patients (see Figure E3).

sue in control patients demonstrated normal histology. Mtb was culturable at all cavity positions, including normal-appearing lung tissue. However, the highest bacterial burden (i.e., lowest time-to-positivity) was at the air–caseum interface. No Mtb growth occurred in lung tissue from control patients (see Figure E3). RNA Sequencing Results and Principal Component Analyses Sixty-nine samples passed the stringent RNA-Seq quality criteria detailed in the online supplement. Reads were aligned to human genome, and were mapped to the 19,049 genes; 31.07% were splice variants (see Figure E4). All RNA-Seq reads were examined in toto using principal component analysis (see Figure E5). Transcripts clustered into four main groups: control subjects (i.e., no TB), normal-appearing tissue, the combination of three cavity wall positions, and a distinct subcluster of cavity center samples within the normal-appearing tissue cluster. Therefore, we assigned all samples to one of six groups for further analyses: 1) control subjects (non-TB), 2) normal-appearing tissue, 3) cavity wall, 4) air–caseum interface, 5) sputum, and 6) airways.

avity wall positions, and a distinct subcluster of cavity center samples within the normal-appearing tissue cluster. Therefore, we assigned all samples to one of six groups for further analyses: 1) control subjects (non-TB), 2) normal-appearing tissue, 3) cavity wall, 4) air–caseum interface, 5) sputum, and 6) airways. Modular Analyses to Map Cell Types and Effectors to Cavity Positions We used our validated modular analysis approach on the RNA-seq data to map modules to cavity positions (20–24). Figure E6 shows multiple upregulated and downregulated genes within each module, indicating that modules were not influenced by a single highly expressed gene. Figure 1 shows modules for six cell types. In most cases cells were most abundant in the cavity wall compared with control subjects, with the notable exception the module for natural killer cells. Conversely, the air–caseum interface at cavity center had the lowest cellular abundance. A similar pattern was observed in modules of effector functions, with the exception of Th2 and LPS responses (Figure 1). Overall, the modular analysis indicates that each cavity position had a remarkably consistent gene expression profile between different patients, with small error bars between patients, and was independent of cavity size and disease extent.

of effector functions, with the exception of Th2 and LPS responses (Figure 1). Overall, the modular analysis indicates that each cavity position had a remarkably consistent gene expression profile between different patients, with small error bars between patients, and was independent of cavity size and disease extent. Figure 1. Cell type and effector responses distributions in and around cavity. The graphs show the mean ± SEM of per kilobase per million mapped reads (RPKM) of all genes in the module at each position; “Normal” is normal-appearing lung in tuberculosis (TB), whereas “Non-TB lung” refers to the control subjects. Cavity wall positions were combined. The Kruskal-Wallis test P value (corrected for multiple comparisons) in each graph is shown for the comparison of RPKM values in control subjects without TB versus RPKM values in the different cavity positions. For RPKM comparison between control subjects without TB and normal-appearing lung tissue in patients with multidrug-resistant TB, P value was >0.2 for all comparisons. The RPKM values were higher in macroscopically normal-appearing tissue in patients with multidrug-resistant TB compared with non-TB control lung, but were highest in the cavity wall, with a precipitous decline below non-TB lung in the cavity center. However, expression of Th2 cytokine–inducible genes and LPS response were not higher in patients with multidrug-resistant TB than in control subjects. The highest RPKM values relative to control subjects were with the type II IFN (IFN-γ)–inducible genes in the cavity wall. Also notable are the increases in B cells in the cavity wall and airways. Mtb = Mycobacterium tuberculosis; NK = natural killer; Th2 = T-helper cell type 2; TNF = tumor necrosis factor.

l subjects. The highest RPKM values relative to control subjects were with the type II IFN (IFN-γ)–inducible genes in the cavity wall. Also notable are the increases in B cells in the cavity wall and airways. Mtb = Mycobacterium tuberculosis; NK = natural killer; Th2 = T-helper cell type 2; TNF = tumor necrosis factor. Unsupervised Analyses Using the Highest Differentially Expressed Pathways DEGs were mapped to different physiological pathways using IPA. To avoid bias when choosing the important pathways, we defined “important” as the highest differentially expressed pathways (up or down), regardless of whether they had known immunological function or not. IPA of the top 500 DEGs identified 60 physiological pathways that are shown in Figure 2. Different expressed pathways were confined to specific spatial locations within each cavity. Importantly, IPA results were consistent with our modular analyses findings in Figure 1, indicating that we reached the same conclusions using two different analytic approaches. These data in Figure 2 illustrate that the topology-constrained organization of interlinked pathways remained consistent regardless of cavity size and volume (which showed significant variation between patients). Thus, the TB cavity is mathematically an “attractor,” defined as the condition toward which trajectories of systems converge despite different starting conditions, in nonlinear dynamical systems.

ked pathways remained consistent regardless of cavity size and volume (which showed significant variation between patients). Thus, the TB cavity is mathematically an “attractor,” defined as the condition toward which trajectories of systems converge despite different starting conditions, in nonlinear dynamical systems. Figure 2. A cavity map of the 60 highest differentially expressed pathways. The fold change scale for the pathways is shown using a log2 scale (e.g., 6 on this scale is 64-fold change). The fold change is relative to nontuberculosis lung from control subjects, not normal-appearing tissue from subjects with tuberculosis. There is a functional “hole” in the cavity center, as can be seen by amount of pathways with blue color. However, not all pathways were downregulated in the cavity center; it can be seen that up to a quarter were actually upregulated. In contradistinction, the cavity wall had the most pathways upregulated and had the most intensely upregulated pathways of all. There are several prominent “nonimmunological” pathways that were among the top 60 expressed pathways, including, for example, colorectal cancer metastasis signaling via WNT signaling, and p38 MAPK signaling, downregulated in the cavity center but upregulated in the cavity wall.

t intensely upregulated pathways of all. There are several prominent “nonimmunological” pathways that were among the top 60 expressed pathways, including, for example, colorectal cancer metastasis signaling via WNT signaling, and p38 MAPK signaling, downregulated in the cavity center but upregulated in the cavity wall. Complex Pathway Expression Is Constrained by Cavity Topology In Figure 2, normal-appearing lung tissue had the least number of significantly regulated pathways of any spatial position (14/60 [23%]). Of note, serine/threonine-specific protein kinase C-θ (PKCθ) signaling was observed to be upregulated in normal-appearing lung tissue, suggesting ongoing antigen presentation at the immunological synapse and thus T-cell proliferation (Figure 3A). However, peroxisome proliferator–activated receptor-α signaling, and the related liver X receptor and retinoid acid receptor signaling, were the most downregulated, as shown in Figure 3B.

pearing lung tissue, suggesting ongoing antigen presentation at the immunological synapse and thus T-cell proliferation (Figure 3A). However, peroxisome proliferator–activated receptor-α signaling, and the related liver X receptor and retinoid acid receptor signaling, were the most downregulated, as shown in Figure 3B. Figure 3. Highly expressed pathways in normal-appearing tissue. Purple denotes proteins encoded by upregulated genes, and green denotes downregulated genes. (A) Serine/threonine-specific protein kinase C-θ signaling. Shown are upregulated MHCII, TCR, and downstream RAC genes; also shown is upregulation of the linked NFκB and c-JUN/c-FOS transcription signaling pathways. This suggests ongoing antigen presentation and processing and ongoing antiapoptotic signaling. (B) PPAR-α (peroxisome proliferator–activated receptor-α) signaling genes are shown downregulated in normal-appearing tissue. The consequences of the downregulation of PPAR-α are unclear; however, it is a selective negative regulator of T-helper cell type 17 differentiation and positive regulator for regulatory T cells. Thus, consequences could be increased T-helper cell type 17 expression and decreased regulatory T-cell expression.

g tissue. The consequences of the downregulation of PPAR-α are unclear; however, it is a selective negative regulator of T-helper cell type 17 differentiation and positive regulator for regulatory T cells. Thus, consequences could be increased T-helper cell type 17 expression and decreased regulatory T-cell expression. In contrast, 47/60 (78%) of pathways in the cavity wall demonstrated increased expression in Figure 2. Only peroxisome proliferator–activated receptor-α and liver X receptor and retinoid acid receptor, vitamin C antioxidant activation, and endothelial nitric oxide synthase signaling were downregulated in the cavity wall. The cavity wall had the highest expression of all pathways, including nitric oxide production and reactive oxygen species by macrophages, IL-1, IL-6, IFN-γ, and nuclear factor-κB (NF-κB) activation. Of note, triggering receptor expressed on myeloid cells-1 (TREM-1) signaling was one of the most highly expressed pathways. In addition, 15/60 (25%) highly expressed pathways changed in parallel with TREM-1 across cavity positions. Many of these, such as PI3K/AKT, ERK/MAPK, p38 MAPK, IL-1, IL-6, chemokine, NF-κB, IFN-γ, TNFR1, TNFR2, IL-8, IL-10, monocyte chemotactic protein 1, and cell adhesion via CD54, are downstream to, and are upregulated by, TREM-1 signaling (25–28). Thus, TREM-1 and the 15 related signaling pathways formed dynamic networks of interactions that changed in parallel at this cavity position, and are by definition a complex system.

-8, IL-10, monocyte chemotactic protein 1, and cell adhesion via CD54, are downstream to, and are upregulated by, TREM-1 signaling (25–28). Thus, TREM-1 and the 15 related signaling pathways formed dynamic networks of interactions that changed in parallel at this cavity position, and are by definition a complex system. The air–caseum interface (cavity center) exhibited the highest number (32/60 [53%]) of downregulated pathways, including RIG-I–like receptors, retinoic acid-mediated apoptosis, and hitherto described TREM-1–linked pathways. The most distinct feature of Figure 2 was that 5/32 (16%) of the downregulated pathways at the air–caseum interface were neuroendocrine-related (dopamine-, metabotropic glutamate receptor (mGluR)-, mGluR-dependent synaptic long-term depression formation; neuronal nitric oxide synthase, and prolactin-signaling), several of which are shown in Figure 4. Altogether there were more than 30 downregulated DEGs mapping to this neuroendocrine system, which were the most intensively downregulated genes in the entire transcriptome. Closely related was the decrease in Ca2+ signaling, including of receptor-operated Ca2+ channels that are activated by binding of neurotransmitters (see Figure E7). We propose these pathways as constituting the neuroendocrine system. In contrast, complement, which has been linked to control of Mtb burden, was one of the few pathways upregulated at the air–caseum interface (29, 30).

receptor-operated Ca2+ channels that are activated by binding of neurotransmitters (see Figure E7). We propose these pathways as constituting the neuroendocrine system. In contrast, complement, which has been linked to control of Mtb burden, was one of the few pathways upregulated at the air–caseum interface (29, 30). Figure 4. Neuroendocrine signaling pathways in the cavity center. (A) Dopamine receptor signaling in the cavity center showed decreased expression of genes encoding three of the five dopamine and uptake receptors, and the downstream parathyroid hormone and prolactin pathways. (B) Glutamate receptor signaling showed decrease in the expression of the two varieties of receptors: ionotropic (iGluR) and metabotropic receptors. Genes for each iGluR pharmacological class were downregulated: group I genes (GRM1 and GRM5) activate phospholipase C, and both group II (GRM2) and group III (GRM4, GRM7, and GRM8) genes inhibit the cyclic AMP cascade. (C) In synaptic long-term depression, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (gene AMPAR), which works via PLA2 (phospholipase A2) and PKC (protein kinase C), was downregulated, as were PLA2 and PKC. iGluR AMPAR genes, such as GRIA and GRIN, were also downregulated, as were genes encoding the vesicular glutamate transporters SLC17A and SLC1A6/7. The related calcium signaling is shown in Figure E7. GluR = glutamate receptor; mGluR = metabotropic glutamate receptor; PRL = prolactin; PTH = parathyroid hormone.

nd PKC. iGluR AMPAR genes, such as GRIA and GRIN, were also downregulated, as were genes encoding the vesicular glutamate transporters SLC17A and SLC1A6/7. The related calcium signaling is shown in Figure E7. GluR = glutamate receptor; mGluR = metabotropic glutamate receptor; PRL = prolactin; PTH = parathyroid hormone. In the airways, 18/60 (30%) pathways were extensively downregulated, whereas 36/60 (60%) were upregulated (Figure 2). Thus, the airways had mixed picture that only partially reflected the rest of the TB cavity. Ontological Integration of Multiple Networks and Mtb Burden by Spatial Location Given the multiplicity (60) of pathways, we sought to limit complexity and to organize the information using mathematical modeling. Our nonlinear dynamical sink model mapped along the TB cavity space resulted in an excellent fit for the multiple pathways and the associated Mtb burden as shown in Figure 5 and Figure E8, whereas standard linear models did not (see Figures E9 and E10) (see online supplement for a detailed explanation) (18, 25). The model parameter estimates for the dynamical sink model are shown in Table E2. Thus, the multiple pathways formed complex systems that had nonlinear and spatially constrained interactions, which fulfills the definition of complex adaptive systems (31–33).

(see online supplement for a detailed explanation) (18, 25). The model parameter estimates for the dynamical sink model are shown in Table E2. Thus, the multiple pathways formed complex systems that had nonlinear and spatially constrained interactions, which fulfills the definition of complex adaptive systems (31–33). Figure 5. Model fit for dynamical sink with coupled interaction network. Model fits are shown with mean per kilobase per million mapped read (RPKM) values for each module (consisting of all genes in the module), for (A) macrophages, (B) neutrophils, (C) complement, (D) Th1, (E) Th2, (F) IL-10, (G) TREM-1, (H) protein kinase C-θ, (I) neuroendocrine, and (J) Mycobacterium tuberculosis burden. There were deep wells for the sink encountered for macrophages (A), complement (C), Th1 (D), IL-10 (F), and TREM-1 (G) at 6–8 cm. The neuroendocrine system/pathway looks flat because of the scale used and the relatively low RPKM reads (mostly downregulated compared with nontubersulosis lung). The scaled-up figure is shown in Figure E8. Table E3 shows the dynamical sink parameters of the neuroendocrine module. CI = confidence interval; Mtb = Mycobacterium tuberculosis; PKCθ = protein kinase C-θ; Th1 = T-helper cell type 1; TREM-1 = triggering receptor expressed on myeloid cells-1.

with nontubersulosis lung). The scaled-up figure is shown in Figure E8. Table E3 shows the dynamical sink parameters of the neuroendocrine module. CI = confidence interval; Mtb = Mycobacterium tuberculosis; PKCθ = protein kinase C-θ; Th1 = T-helper cell type 1; TREM-1 = triggering receptor expressed on myeloid cells-1. Next, we calculated partial rank correlation coefficients between both observed and Latin hypercube sampling–based RPKM values of the different effector modules, with results shown in Figure E11. Latin hypercube sampling magnified the resolution of the observed RPKM values and allowed us to examine dynamical sink model sensitivity. Figures E11A and E11B shows strong correlation (i.e., r > 0.6 or < −0.6) between macrophage and neutrophil counts and both complement stimulation and expansion, but negative correlation with macrophage and neutrophil infection and bursting at different cavity positions. Infected macrophages and neutrophils burst after intracellular bacteria replication and are assumed to release more than 25 Mtb per cell (termed “burst size”) into the extracellular environment (hence increasing extracellular bacterial burden and infecting more cells), as part of necrosis, based on prior studies (34–36). The assumption of necrosis as opposed to apoptosis was based on the finding that apoptosis signaling was not upregulated in cavities; indeed, retinoic acid–mediated apoptosis was significantly downregulated. Extracellular Mtb burden varied by cavity position (see Figure E3 and Table E1), which differentially influenced immune cell abundance by cavity position. The pattern of cell abundance could be explained either via influx of macrophages and neutrophils to the respective locations, or increased expression, or differential location dependent immune cell death rates (which were all captured in the model): the results of the modeling suggest that depletion of different immune cells, such as macrophages and neutrophils, and other cell types in the “sink” is caused by Mtb killing of the immune cells.

or increased expression, or differential location dependent immune cell death rates (which were all captured in the model): the results of the modeling suggest that depletion of different immune cells, such as macrophages and neutrophils, and other cell types in the “sink” is caused by Mtb killing of the immune cells. Neuroendocrine and PKCθ Expression Correlation with Mtb Burden The mathematical model allowed us to explore the correlation between Mtb burden and different physiological pathways, based on in silico simulations plus observed data (see Figure E12). There was poor correlation between bacterial burden and either Th1 or Th2 system expression, whereas TREM-1 expression showed good correlation only in the cavity wall (see Figures E12A–E12F). Uniquely, PKCθ demonstrated negative correlation with Mtb burden that was highest in the cavity wall (see Figure E12G). Neuroendocrine expression had high negative correlation with bacterial burden: bacterial burden increased with increased neuroendocrine expression (see Figure E12H).

vity wall (see Figures E12A–E12F). Uniquely, PKCθ demonstrated negative correlation with Mtb burden that was highest in the cavity wall (see Figure E12G). Neuroendocrine expression had high negative correlation with bacterial burden: bacterial burden increased with increased neuroendocrine expression (see Figure E12H). Model Simulation-Perturbation Experiments What happens to Mtb burden and each physiological module at each cavity position when a particular pathway is either stimulated or inhibited to a specified degree in silico? An example is shown for the neuroendocrine system in Figure E13. The simulations predict profound neuroendocrine system “dose-dependent” changes in bacterial burden in Figure E13A, with increase in Mtb burden as neuroendocrine signaling increases. Figures E13B–E13I show negative relationships between neuroendocrine expression and expression of Th1, IL-10, complement, and infection of macrophages and neutrophils. These relationships were constrained by spatial position.

rial burden in Figure E13A, with increase in Mtb burden as neuroendocrine signaling increases. Figures E13B–E13I show negative relationships between neuroendocrine expression and expression of Th1, IL-10, complement, and infection of macrophages and neutrophils. These relationships were constrained by spatial position. Immunohistochemistry Confirmation of RNA-Seq and Model Findings We performed immunohistochemistry on the cavity wall, airways, and control lung tissue, to confirm the RNA-Seq and mathematical modeling findings. Figure E14 shows that the abundance of CD4+ and CD8+ was similar between the cavity wall, airways, and noninfected control subjects. For the macrophage lineage, Figure E14C demonstrated a 1.49-fold higher CD68+ cell population (i.e., macrophage lineage) in the cavity wall compared with noninfected control subjects, with similar levels in the airways. FOXP3+ cell stains revealed a 15-fold higher abundance of FOXP3+ cells (regulatory T cells) in the cavity wall compared with airways (see Figure E14D). For the proposed neuroendocrine system, we stained for chromogranin A (parathyroid hormone secretory system specific) in airways, the cavity wall, and in control subjects, with results shown in Figure E14E. The airways had lower chromogranin ratios than the cavity wall, and were lower than in control subjects without TB, consistent with RNA-seq findings and our dynamical sink model. We also used a neuroglia-specific S100B stain, and the airways showed significantly lower intensity for S100B stain than the cavity wall (see Figure E14F). For another confirmation, cavity wall biopsy neurofibrin staining was positive in four of eight patient cavities tested.

-seq findings and our dynamical sink model. We also used a neuroglia-specific S100B stain, and the airways showed significantly lower intensity for S100B stain than the cavity wall (see Figure E14F). For another confirmation, cavity wall biopsy neurofibrin staining was positive in four of eight patient cavities tested. Discussion Our main goal was to identify how and where host immunity failed to control Mtb burden in MDR-TB cavities, and to identify the specific host immunity pathways that failed. We identified at least 60 different highly expressed pathways that formed complex adaptive system networks, which we integrated using a dynamical sink model onto approximately 600 cm3 histopathological space/tissue volume. In the cavity wall, including the luminal edge of the cavity wall, approximately 80% of the pathways were upregulated, and were proinflammatory. The high negative correlation (r < −0.6) between pathway expression and Mtb burden in the cavity wall was encountered with macrophages, neutrophils, TREM-1, and PKCθ, whereas the high positive correlation (r > 0.6) was neuroendocrine (see Figure E12); the negative correlation means increased expression of the pathway was associated with improved bacterial containment (lower Mtb burden) and thus protective. In contrast, at the air–caseum interface, only 7% of the 60 pathways were upregulated, more than 50% were downregulated (including TREM-1–linked pathways, neuroendocrine, macrophage, and neutrophil pathways), and had the highest Mtb burden of all, marking this as the location of immune failure. The highest negative correlation between pathway expression and Mtb burden at air–caseum interface were macrophages and neutrophils, whereas high positive correlation was with neuroendocrine pathway, suggesting that these specific pathways play the major role in bacterial containment and, thus, host immunity failure.

. The highest negative correlation between pathway expression and Mtb burden at air–caseum interface were macrophages and neutrophils, whereas high positive correlation was with neuroendocrine pathway, suggesting that these specific pathways play the major role in bacterial containment and, thus, host immunity failure. Modeling and simulations to explore several possible explanations for this immune failure, including location-dependent reduced influx of immune cells or proliferation or cell death rates, revealed that the most likely mechanism was of Mtb killing the immune cells via cell bursting, which was maximal in the cavity center. Other possible mechanisms of cell kill, such as cytotoxic agents released by Mtb, will be investigated in separate modeling. The pathways identified here may also be used as biomarkers of the failed bacterial containment in the TB cavity. In addition, molecules targeting these networks could potentially offer new targets for therapy and immunomodulation and to limit transmission (37).

leased by Mtb, will be investigated in separate modeling. The pathways identified here may also be used as biomarkers of the failed bacterial containment in the TB cavity. In addition, molecules targeting these networks could potentially offer new targets for therapy and immunomodulation and to limit transmission (37). In 2009, Anyanful and colleagues reported that a brief exposure of Caenorhabditis elegans to toxigenic Escherichia coli conditioned the worms to survive subsequent exposure, because of dopamine signaling linked to innate immune responses (38, 39). In the lung, dopamine receptors are known to directly control alveolar cell inflammatory processes, and indirectly via parathyroid hormone and prolactin release (40, 41). Prolactin promotes proinflammatory innate immune responses via NF-κB and IRF-1 (42). We found decreased expression of dopamine receptor signaling and the downstream pathways (parathyroid hormone, prolactin, and IRIF-1) in TB cavity wall. In addition to dopamine, mGluR signaling, mGluR-dependent synaptic long-term depression formation, and calcium signaling were decreased in parallel. Thus, the neuroendocrine response pathways in the TB cavity that we describe expand the system beyond the dopamine pathway (39). Colonic inflammation in a rat model reduced hippocampal mGluR-dependent synaptic long-term depression formation, which was reversed by chronic administration of minocycline, a drug that also has direct anti-TB effect (43, 44). Moreover, Sarm1 expression decreases mGluR-dependent synaptic long-term depression formation: Sarm1 is a negative regulator of Toll-like receptor signaling, TNF-α, and antiviral cytokines production in mice (45–47).

reversed by chronic administration of minocycline, a drug that also has direct anti-TB effect (43, 44). Moreover, Sarm1 expression decreases mGluR-dependent synaptic long-term depression formation: Sarm1 is a negative regulator of Toll-like receptor signaling, TNF-α, and antiviral cytokines production in mice (45–47). Our modeling and simulations revealed that neuroendocrine system expression had a high positive correlation with neutrophil and macrophage burst sizes and associated cell-specific death from the Mtb infection in TB cavities. Moreover, our simulations demonstrate dose-dependent changes to the Th1 system with perturbation in the neuroendocrine system (see Figure E13), a major adaptive immune mechanism effecting Mtb control. Furthermore, high correlation coefficients between observed neuroendocrine RPKMs and Mtb burden, and the dose-dependent results of the in silico perturbation exercises, all suggest that these findings are likely biologically meaningful. Pharmacological manipulation of this neuroendocrine system, once better characterized, may offer a novel approach to reverse failed immunity in the TB cavity, limit Mtb burden, and reduce transmission.

t results of the in silico perturbation exercises, all suggest that these findings are likely biologically meaningful. Pharmacological manipulation of this neuroendocrine system, once better characterized, may offer a novel approach to reverse failed immunity in the TB cavity, limit Mtb burden, and reduce transmission. There were several potential limitations of our study. First, transcriptomic data were normalized to control subjects without TB rather than pericavitory normal-appearing tissue from the same patient (interpatient vs. intrapatient control subjects). However, viable Mtb was commonly present in pericavitory normal-appearing tissue, thus it was far from normal and unsuitable for normalization, making this approach biologically unsound. Second, five South African control subjects had evidence of prior TB, which could introduce biasing toward the null. Therefore, we added a pooled sample of five Americans with no history of TB; the transcriptomic signatures in the control subjects were remarkably similar when compared by nationality (see online supplement). Third, our findings pertain to chronic MDR-TB (and MDR-TB plus resistance to quinolones and other drugs) that failed therapy, and may not be generalizable to drug-sensitive TB. Finally, biopsies were only performed from a single cavity rather than from multiple cavities, limited immunohistochemical analysis was undertaken, and functional experiments to interrogate the role of neuroendocrine system were not performed. However, our express goal was to characterize the cavity transcriptome, and we now plan to address the discoveries in a more granular manner.

than from multiple cavities, limited immunohistochemical analysis was undertaken, and functional experiments to interrogate the role of neuroendocrine system were not performed. However, our express goal was to characterize the cavity transcriptome, and we now plan to address the discoveries in a more granular manner. In summary, construction of a spatially compartmentalized transcriptomic map of MDR-TB cavities identified several hitherto unrecognized pathways involved in the host immune response to TB. Further studies are warranted to investigate the functional characteristics of these pathways and their effect on controlling Mtb proliferation. Supported by the Baylor Research Institute (T.G.), NIH (DP2 OD001886, R01AI079497, and R56 AI111985; T.G.), South African MRC (K.D.), European Developing Clinical Trials Partnership (K.D.), National Research Foundation (K.D.), Oppenheimer Foundation (K.D.), Wellcome Trust (207511/Z/17/Z [M.N.] and WT101766/Z/13/Z [G.P.]), Medical Research Council (MR/M003833/1; C.T.T.), and NIHR Biomedical Research Funding to University College London and University College London Hospitals.

cal Trials Partnership (K.D.), National Research Foundation (K.D.), Oppenheimer Foundation (K.D.), Wellcome Trust (207511/Z/17/Z [M.N.] and WT101766/Z/13/Z [G.P.]), Medical Research Council (MR/M003833/1; C.T.T.), and NIHR Biomedical Research Funding to University College London and University College London Hospitals. Author Contributions: K.D., L.L., and T.G.: conceived of the research, designed the study, interpreted data, and wrote the draft of the manuscript. G.M. and T.G.: developed and implemented mathematical model. H.W.: performed the dissection, hematoxylin and eosin stains, and immunohistochemistry. K.D., L.L., M.D., A.P., and R.M.: recruited the patients. R.S.: performed positron emission tomography/computed tomography scans, interpreted positron emission tomography/computed tomography scans, and scored the positron emission tomography/computed tomography scans. T.P., A.L., and L.M.: recruited the patients and performed the surgical resection. L.L.: performed the microbiology work under R.M.W.’s supervision. L.L. and S.S.: performed the RNA extraction, curated the data, and interpreted the data. P.R.: performed the RNA sequencing under E.W.’s supervision. P.R. and E.W.: performed RNA sequencing quality control. S.J.B.: performed further RNA sequencing control, realignment, and differential expression calculations and established the RNA sequencing pipeline. G.P., C.T.T., and M.N.: performed modular analyses. J.G.P.: performed statistical analyses. K.D. and T.G.: directed the research. All authors: wrote sections of the manuscript on the work they did and saw and edited the final manuscript.

differential expression calculations and established the RNA sequencing pipeline. G.P., C.T.T., and M.N.: performed modular analyses. J.G.P.: performed statistical analyses. K.D. and T.G.: directed the research. All authors: wrote sections of the manuscript on the work they did and saw and edited the final manuscript. This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org. Originally Published in Press as DOI: 10.1164/rccm.201807-1361OC on January 29, 2019 Author disclosures are available with the text of this article at www.atsjournals.org.

The extracellular matrix (ECM) is of fundamental importance for the functional capabilities of the lung because it serves to maintain tissue tensile strength, elasticity, and barrier function. Furthermore, it is increasingly apparent that the ECM not only acts as a scaffold for tissue resident cells but also directly orchestrates essential cellular processes, including morphogenesis, signal transduction, migration, proliferation, and wound repair (1). Accordingly, an aberrant ECM, as is frequently seen in chronic lung diseases, is not only an end-stage pathological manifestation that compromises tissue functionality but also likely dictates the development and progression of the disease. Collagen accumulation around the airways of patients with asthma is a hallmark pathological feature of ECM remodeling that is associated with irreversible changes in airway obstruction (2). However, our understanding of these pathological changes is limited to a macroscopic assessment of the gross quantities of collagen as adjudged by historical histological stains, and this tells us very little about how the structural or biochemical properties of the collagen may be aberrant. This oversimplification likely neglects potentially important differences in the organization of the collagen, which could impact its capacity to modulate cellular behaviors and define tissue architecture and functionality.

very little about how the structural or biochemical properties of the collagen may be aberrant. This oversimplification likely neglects potentially important differences in the organization of the collagen, which could impact its capacity to modulate cellular behaviors and define tissue architecture and functionality. In a study presented in this issue of the Journal, Mostaço-Guidolin and colleagues (pp. 431–443) used nonlinear optical microscopy (NLOM) to assess the biochemical and structural features of collagen and elastin in nontransplantable donor lungs from individuals with and without asthma (3). The nonlinear interactions that are the basis of the images generated by NLOM occur naturally between light and specific biological molecules owing to their intrinsic natural properties, thus allowing these macromolecules to be imaged without the application of exogenous stains. In their elegant study, the authors comprehensively demonstrate that fibrillar collagen was not only increased in mass within the lamina propria of small and large airways of patients with asthma, but also disorganized and fragmented. These changes were apparent in patients with asthma regardless of age or sex, and were comparable between patients with fatal disease and those with nonfatal disease. Rationalizing these changes, the authors demonstrated that fibroblasts isolated from the airways of patients with asthma were defective in collagen I fiber formation relative to those derived from healthy control subjects. They attributed this failing to a reduced expression of decorin (a proteoglycan that is required for normal spacing of collagen fibrils within fibers) by airway fibroblasts from subjects with asthma. In support of this assertion, the authors demonstrated that packaging of collagen fibrils was indeed disorganized in the airways of patients with asthma.

reduced expression of decorin (a proteoglycan that is required for normal spacing of collagen fibrils within fibers) by airway fibroblasts from subjects with asthma. In support of this assertion, the authors demonstrated that packaging of collagen fibrils was indeed disorganized in the airways of patients with asthma. The study by Mostaço-Guidolin and colleagues represents a step change in the manner in which we should now view the ECM in defining a disease state. It is no longer enough to consider the relative levels of ECM macromolecules—we should also consider the potential implications of more subtle differences in morphology and physiology. However, these studies inevitably raise further questions. At present, the postulated mechanism whereby reduced fibroblast decorin expression drives disorganized airway collagen in patients with asthma remains primarily correlative and needs to be proved experimentally. The authors’ assertion that changes in decorin are causal is supported by observations that decorin knockout mice present with collagen fibril disorganization in the skin and tendons (4). Future studies should seek to manipulate decorin levels in murine models of allergic airways disease and primary patient cells to validate that they impact collagen organization and ultimately asthma pathophysiology. Furthermore, it is important to confirm that decorin expression is indeed reduced in fibroblasts proximal to the disorganized airway collagen of patients with asthma, as well as to gain a fuller understanding of the pathways that define its expression and determine whether they can be therapeutically manipulated.

logy. Furthermore, it is important to confirm that decorin expression is indeed reduced in fibroblasts proximal to the disorganized airway collagen of patients with asthma, as well as to gain a fuller understanding of the pathways that define its expression and determine whether they can be therapeutically manipulated. It is interesting that disorganized collagen is observed in the airways of patients with asthma regardless of sex and age, and whether they have fatal or nonfatal disease, but this ultimately raises questions about the role that changes in collagen organization play in driving disease pathology and severity. Analysis of cadavers, although clearly advantageous for whole-lung analysis of ECM by NLOM, has limitations with regard to the availability of paired patient histories. In the future, it would be of interest to correlate these observed changes in collagen organization with underlying inflammation, lung function, disease severity, and, more broadly, patient endotype. This ultimately raises a more generic question as to the physiological significance of disorganized collagen. The abundance of collagen fibers correlates with the resistance and elasticity of parenchymal tissues (5), and the organized assembly of collagen fibrils is of central importance in defining the tensile strength of the collagen (6). However, there is clearly a need to more fully understand the effects of disorganized collagen on airway mechanics and airway hyperresponsiveness. More broadly, given the capacity of ECM macromolecules to regulate pleiotropic cellular activities, it will be intriguing to determine how structural changes in collagen impact its signaling capacity. A more in-depth understanding of the processes that drive disorganized airway collagen in individuals with asthma, and ensuing implications of these changes in regulating airway mechanics and behavior of proximal cells, could ultimately lead to identification of new therapeutic targets.

n impact its signaling capacity. A more in-depth understanding of the processes that drive disorganized airway collagen in individuals with asthma, and ensuing implications of these changes in regulating airway mechanics and behavior of proximal cells, could ultimately lead to identification of new therapeutic targets. The chronic inflammation observed in asthmatic airways was classically believed to be the driver of airway remodeling, but it is now acknowledged that both processes can occur in parallel and independently, and are of comparable importance in defining the clinical course of disease (7). Given the importance of both airway inflammation and remodeling in defining asthma pathogenesis, it is noteworthy that there is significant disparity in the assessment of these dual processes. The vast majority of clinical studies have focused on assessment of the inflammatory rather than airway remodeling profile of patients owing to the challenges of assessing airway remodeling noninvasively, with analysis generally necessitating bronchial biopsies. It is intriguing, therefore, that NLOM has successfully been coupled with endoscopy techniques for the assessment of airways in vivo (8), and given that visualization of collagens and elastin by NLOM does not necessitate the addition of exogenous stains, it is clear that this represents a potentially exciting, less invasive technology with clear clinical applications (Figure 1). Ideally, unbiased statistical approaches to define asthma endotypes should incorporate a remodeling phenotype together with routinely assessed clinical and inflammatory parameters, as the contributions and interactions of each factor will ultimately dictate disease outcomes. It is tantalizing to consider that the combination of endoscopy with NLOM technology could lead to a more routine assessment of airway collagen and elastin deposition in individuals with asthma, which could be tracked longitudinally with inflammation and clinical parameters to delineate the evolution of the disease. Therapeutic strategies for asthma are entering an age of personalized medicine (9, 10), and incorporating remodeling phenotypes to stratify patients will be fundamentally important for identifying optimal treatment regimens.

itudinally with inflammation and clinical parameters to delineate the evolution of the disease. Therapeutic strategies for asthma are entering an age of personalized medicine (9, 10), and incorporating remodeling phenotypes to stratify patients will be fundamentally important for identifying optimal treatment regimens. Furthermore, assessment of airway remodeling is rarely included as an outcome measure in clinical trials owing to the invasive nature of tissue biopsies, but the use of NLOM could facilitate a longitudinal interrogation of the effects of therapeutic interventions on changes to the ECM and ensuing disease control or remission. Figure 1. Future clinical applications for nonlinear optical microscopy. The ability to combine nonlinear optical microscopy with bronchoscopy to generate images of extracellular matrix (ECM) macromolecular structure without the use of exogenous stains makes this approach an exciting, less invasive way to longitudinally assess airway remodeling in patients. R.J.S. is a Wellcome Trust–funded (209458/Z/17/Z) Senior Research Fellow in Basic Biomedical Sciences. Originally Published in Press as DOI: 10.1164/rccm.201904-0722ED on April 15, 2019 Author disclosures are available with the text of this article at www.atsjournals.org.

For over half a century, clinicians and researchers have endeavored to understand the relationship between oxygen delivery and lactic acidosis (1, 2) (or, as discussed below, perhaps more readily considered as “hyperlactatemia with or without acidemia”). In health, pyruvate (the generally acknowledged end product of glycolysis) is metabolized by mitochondria to acetyl coenzyme A to feed the tricarboxylic acid cycle. Excess pyruvate is reduced by lactate dehydrogenase to l-lactate. Notably, this reduction consumes a proton: pyruvate + NADH + H+ ↔ lactate + NAD+. Lactate is subsequently oxidized back to pyruvate, either locally or after transfer to organs that use lactate as a fuel source (e.g., liver, kidney, and brain) or that convert it back to glucose (the Cori cycle in the liver). In concert, these processes maintain normal blood lactate levels. During sepsis, lactate levels frequently rise. Indeed, hyperlactatemia (a measurable surrogate for cellular/metabolic perturbations) is closely associated with sepsis prognosis and is now one of the criteria for septic shock (3). However, it remains challenging to determine clinically when a persistently elevated serum lactate level indicates ongoing inadequacy of oxygen delivery, or when the problem lies elsewhere. The brainstem response to give yet more fluid is often inappropriate and potentially injurious.

the criteria for septic shock (3). However, it remains challenging to determine clinically when a persistently elevated serum lactate level indicates ongoing inadequacy of oxygen delivery, or when the problem lies elsewhere. The brainstem response to give yet more fluid is often inappropriate and potentially injurious. Hyperlactatemia during sepsis may result from anaerobic glycolysis. When whole-body oxygen delivery fails to meet cellular demands, tissues transition from predominant mitochondrial aerobic respiration to less efficient ATP generation by glycolysis. This is most commonly observed at the time of initial patient presentation, and in many cases can be resolved by administration of intravenous fluids with or without vasoactive agents. However, other factors may also increase serum lactate levels in sepsis, including β2-receptor stimulation from endogenous/exogenous catecholamines, impaired tissue oxygen extraction (mitochondrial dysfunction with or without microcirculatory dysfunction), liver dysfunction, and thiamine deficiency.

vasoactive agents. However, other factors may also increase serum lactate levels in sepsis, including β2-receptor stimulation from endogenous/exogenous catecholamines, impaired tissue oxygen extraction (mitochondrial dysfunction with or without microcirculatory dysfunction), liver dysfunction, and thiamine deficiency. To aid the clinician in his/her decision-making, Gattinoni and colleagues (pp. 582–589) in this issue of the Journal propose a conceptual model relating oxygen delivery and utilization, serum lactate concentration, and acidemia (4). They analyzed data from 1,741 ICU patients who were enrolled in the ALBIOS (Albumin Italian Outcome Sepsis) trial, using serum lactate, central venous oxygen saturation (ScvO2), and blood gas measurements taken at study enrollment (5). Fundamentally, their proposed model frames two clinical questions:1. Is an elevated lactate level due to inadequate oxygen delivery and therefore potentially responsive to interventions that increase oxygen delivery? 2. How does an elevated serum lactate level affect arterial pH and base excess?

To aid the clinician in his/her decision-making, Gattinoni and colleagues (pp. 582–589) in this issue of the Journal propose a conceptual model relating oxygen delivery and utilization, serum lactate concentration, and acidemia (4). They analyzed data from 1,741 ICU patients who were enrolled in the ALBIOS (Albumin Italian Outcome Sepsis) trial, using serum lactate, central venous oxygen saturation (ScvO2), and blood gas measurements taken at study enrollment (5). Fundamentally, their proposed model frames two clinical questions:1. Is an elevated lactate level due to inadequate oxygen delivery and therefore potentially responsive to interventions that increase oxygen delivery? 2. How does an elevated serum lactate level affect arterial pH and base excess? Hyperlactatemia and ScvO2 High values of ScvO2 suggest systemic oxygen delivery in excess of oxygen demands, impaired cellular (mitochondrial) oxygen use, and/or microcirculatory shunting. Low ScvO2 values imply inadequate oxygen delivery that fails to meet metabolic demands. Gattinoni and colleagues propose the use of ScvO2 to personalize sepsis management, reserving interventions to increase oxygen delivery to only those patients with low ScvO2 values. Of note, only 35% of patients in the ALBIOS trial had ScvO2 values <70%. Other recent sepsis trials reported similar ScvO2 values after initial resuscitation (6).

propose the use of ScvO2 to personalize sepsis management, reserving interventions to increase oxygen delivery to only those patients with low ScvO2 values. Of note, only 35% of patients in the ALBIOS trial had ScvO2 values <70%. Other recent sepsis trials reported similar ScvO2 values after initial resuscitation (6). This proposal is not inherently novel. Both the concept of early goal-directed therapy (EGDT) (7) and the Surviving Sepsis Campaign recommendations (8) suggest that a low ScvO2 should trigger interventions (e.g., fluid, inotropes, and blood) to increase oxygen delivery. This concept has a strong physiologic rationale, but the devil is in the details. First, the patients in the ALBIOS study and the three recent EGDT trials (6) were all enrolled after initial resuscitation. On first presentation, many patients will have impaired oxygen delivery and thus lower ScvO2 values, and a higher likelihood of responding positively to empiric fluid administration. An important caveat is that a low ScvO2 in sepsis does not automatically equate to hypovolemia. Cardiomyopathy can also contribute, and may be worsened by excessive fluid administration. Second, many patients with sepsis-associated hyperlactatemia have ScvO2 values that fall within an indeterminate range, and even patients with an elevated ScvO2 may respond physiologically to fluid administration (9). Moreover, ScvO2 is a “global” (or rather an “upper-body”) measure of the oxygen supply/demand balance, and may miss imbalances in specific tissue beds (10).

atemia have ScvO2 values that fall within an indeterminate range, and even patients with an elevated ScvO2 may respond physiologically to fluid administration (9). Moreover, ScvO2 is a “global” (or rather an “upper-body”) measure of the oxygen supply/demand balance, and may miss imbalances in specific tissue beds (10). Finally, the history of sepsis research is paved with physiologically rational interventions that nonetheless failed to improve patient outcomes (11). The recent EGDT trials showed no benefit in targeting ScvO2 even among a subset of patients with baseline values <70% (6). Interventions to increase oxygen delivery may have unintended consequences outside the mechanistic pathway assessed by ScvO2 measurement (12, 13). Therefore, an ScvO2-based strategy to personalize interventions for patients with sepsis-associated hyperlactatemia requires careful evaluation in clinical trials before any recommendation regarding standard-of-care implementation in clinical practice can be made.

anistic pathway assessed by ScvO2 measurement (12, 13). Therefore, an ScvO2-based strategy to personalize interventions for patients with sepsis-associated hyperlactatemia requires careful evaluation in clinical trials before any recommendation regarding standard-of-care implementation in clinical practice can be made. Hyperlactatemia and Arterial pH According to the “strong ion” theory, lactate is a strong anion and thus should be completely dissociated from hydrogen in plasma, generating an acidosis. However, some patients with sepsis and hyperlactatemia have a concurrently decreased pH (acidemia), whereas others maintain a normal pH. This suggests mechanisms that enable relatively rapid respiratory or metabolic compensation. Gattinoni and colleagues found that the ability to maintain a normal pH despite elevated lactate levels was more closely correlated with renal function than with respiratory compensation. They propose that an indirect measure of the accumulation of renally excreted fixed acids in plasma—the “alactic base excess”—could be used to assess the kidneys’ ability to compensate for acid-base disturbances.

elevated lactate levels was more closely correlated with renal function than with respiratory compensation. They propose that an indirect measure of the accumulation of renally excreted fixed acids in plasma—the “alactic base excess”—could be used to assess the kidneys’ ability to compensate for acid-base disturbances. The standard base excess, defined as the amount of strong acid that must be added to each liter of oxygenated blood to return the pH to 7.40 at a PaCO2 of 40 mm Hg, quantifies the degree of metabolic acidosis or alkalosis independently of respiratory compensation. Contributors to base excess include lactate, strong ions such as sodium and chloride, albumin, and ions that accumulate in renal failure, such as phosphate and sulfate (14). By adding lactate to the standard base excess, the authors arrive at the alactic base excess, which they assert quantifies “the role of renal function in the acid–base balance in sepsis.”

te, strong ions such as sodium and chloride, albumin, and ions that accumulate in renal failure, such as phosphate and sulfate (14). By adding lactate to the standard base excess, the authors arrive at the alactic base excess, which they assert quantifies “the role of renal function in the acid–base balance in sepsis.” This suggestion is certainly interesting but requires further thought and investigation. Renal compensation for acid–base disturbances has traditionally been considered to be slower than respiratory compensation. Detailed data on urine output, stage of acute kidney injury (15), minute ventilation, and other physiologic measures would be required before the relative causal effects of kidney injury in compensating for acidosis could be fully understood. The alactic base excess is not necessarily an explicit measure of renal function. For example, administration of 0.9% sodium chloride decreases the base excess, even in the presence of stable renal function and lactate concentrations (16). The impact of concurrent liver dysfunction requires consideration, and only a few such cases were included in the ALBIOS database. Nonetheless, the concept of alactic base excess and the role of renal function in modifying acidemia warrant evaluation in future physiologic studies.

ction and lactate concentrations (16). The impact of concurrent liver dysfunction requires consideration, and only a few such cases were included in the ALBIOS database. Nonetheless, the concept of alactic base excess and the role of renal function in modifying acidemia warrant evaluation in future physiologic studies. In summary, Gattinoni and colleagues are to be congratulated for advancing an ambitious conceptual model relating oxygen delivery, lactate generation, renal function, and acidemia in sepsis. We are eager to see future research to confirm and refine this model, and move us closer to the authors’ vision of a more personalized approach to early hemodynamic management for sepsis. M.W.S. was supported in part by the NHLBI (K23HL143053). M.S. was supported in part by the Medical Research Council, Wellcome Trust, European Union, and National Institute for Health Research. Originally Published in Press as DOI: 10.1164/rccm.201904-0899ED on May 19, 2019 Author disclosures are available with the text of this article at www.atsjournals.org.

At a Glance Commentary Scientific Knowledge on the Subject Bronchiectasis is associated with chronic bacterial infection and neutrophilic inflammation. The underlying pathogenesis of the disease is poorly understood. Patients with chronic neutrophil inflammation have impaired innate and adaptive immunity, but the mechanisms by which neutrophilic inflammation links to impaired responses to infection are poorly characterized. In a proteomic study, we identified PZP (pregnancy zone protein), previously identified in the serum of pregnant women, in the airway of patients with severe bronchiectasis and Pseudomonas aeruginosa infection. PZP is a powerful T-cell immunosuppressant believed to prevent rejection of the fetal allograft during pregnancy. We hypothesized that PZP may be associated with airway infection susceptibility in bronchiectasis. In this study, we aimed to determine the source of PZP release in the airway and its association with chronic infection, airway inflammation, and disease severity.

revent rejection of the fetal allograft during pregnancy. We hypothesized that PZP may be associated with airway infection susceptibility in bronchiectasis. In this study, we aimed to determine the source of PZP release in the airway and its association with chronic infection, airway inflammation, and disease severity. What This Study Adds to the Field We demonstrate, for the first time to our knowledge, that elevated airway concentrations of PZP are associated with disease severity, frequent exacerbations, and airway infection in patients with bronchiectasis. PZP was found in the cytoplasm of neutrophils and was released during acute and chronic pulmonary inflammation. PZP was found to be associated with neutrophil extracellular traps in vitro and correlated with neutrophil extracellular traps detected in bronchiectasis patient sputum in vivo. Our findings implicate NETosis in the pathophysiology of bronchiectasis, and given its known immunosuppressive effects, PZP may therefore provide a novel link between chronic neutrophilic inflammation and impaired host immunity to infection.

trophil extracellular traps detected in bronchiectasis patient sputum in vivo. Our findings implicate NETosis in the pathophysiology of bronchiectasis, and given its known immunosuppressive effects, PZP may therefore provide a novel link between chronic neutrophilic inflammation and impaired host immunity to infection. Bronchiectasis is a chronic respiratory disease characterized by lung inflammation, impaired mucociliary clearance, and recurrent airway infection leading to permanent tissue destruction and airway dilation. There is no licensed treatment for bronchiectasis, and therapeutic development has been severely limited by poor understanding of the pathogenesis of the disease (1). Bronchiectasis is a heterogeneous disease in terms of etiology, inflammatory profile, patient characteristics, comorbidities, and background treatments. Bronchiectasis is a global health problem, and further heterogeneity is added by differences in the aforementioned factors between different geographical regions (2–4). An apparent paradox in bronchiectasis is the persistence of pathogens in the airway despite the presence of a robust inflammatory response. During acute inflammation with gram-negative pathogens such as Pseudomonas aeruginosa or gram-positive pathogens such as Staphylococcus aureus, neutrophil recruitment is followed by phagocytosis of pathogens and clearance of both bacteria and neutrophils through apoptosis and efferocytosis by macrophages. During chronic inflammation, this process may be impaired with reduced phagocytosis, reduced neutrophil apoptosis, and a switch to tolerance and containment of infection through neutrophil extracellular trap (NET) formation. The mechanisms leading to this immunological tolerance remain largely unexplored in bronchiectasis (5).

hronic inflammation, this process may be impaired with reduced phagocytosis, reduced neutrophil apoptosis, and a switch to tolerance and containment of infection through neutrophil extracellular trap (NET) formation. The mechanisms leading to this immunological tolerance remain largely unexplored in bronchiectasis (5). PZP (pregnancy zone protein) is a high–molecular-weight glycoprotein that was originally described as being elevated in the serum of women during pregnancy (6). The synthesis of PZP is estrogen dependent, and it is detectable in serum a few weeks after conception and is reported to return to nearly undetectable concentrations immediately postpartum (7). PZP is a broad-spectrum immunosuppressant (8) with antiproteinase activity. Its role in pregnancy is believed to be suppression of cell-mediated immunoreactivity (9, 10) to prevent rejection of the fetus. PZP has been shown to depress T-lymphocyte immunoreactivity, T-cell recruitment, migration, proliferation, and IL-2 production (9). These immunosuppressive effects are profound, illustrated by a study showing that intravenous infusion of PZP was sufficient to prevent rejection of heart allografts in mice (11). Recent experiments using PZP knockout mice have shown that PZP also increases susceptibility to viral infection (12). PZP has never previously been found in the lung or studied in the context of chronic respiratory disease, however.

s infusion of PZP was sufficient to prevent rejection of heart allografts in mice (11). Recent experiments using PZP knockout mice have shown that PZP also increases susceptibility to viral infection (12). PZP has never previously been found in the lung or studied in the context of chronic respiratory disease, however. Chronic infection with P. aeruginosa is consistently associated with disease severity and poor outcomes in bronchiectasis. We therefore performed a proteomic study to profile sputum from patients with bronchiectasis and P. aeruginosa infection compared with those without chronic P. aeruginosa infection. Unexpectedly, PZP was identified as being elevated in patients with P. aeruginosa infection. We therefore hypothesized that, given the established immunosuppressive role of PZP, elevated sputum PZP would be associated with increased susceptibility to chronic airway infection and more severe disease. Our results demonstrate, for the first time, to our knowledge, that PZP is released from neutrophils during degranulation and NET formation. PZP is associated with airway infection in patients with bronchiectasis and may be an unexpected mechanism through which NETs modulate T-cell function leading to increased susceptibility to respiratory infection. Methods For detailed methods, please refer to the online supplement.

Our results demonstrate, for the first time, to our knowledge, that PZP is released from neutrophils during degranulation and NET formation. PZP is associated with airway infection in patients with bronchiectasis and may be an unexpected mechanism through which NETs modulate T-cell function leading to increased susceptibility to respiratory infection. Methods For detailed methods, please refer to the online supplement. Patients and Clinical Assessments Patients were recruited at a specialist bronchiectasis clinic at Ninewells Hospital, Dundee, United Kingdom. Inclusion criteria were age 18 years or older, bronchiectasis confirmed by high-resolution computed tomographic scan, chronic expectoration with ability to provide a sputum sample at the study visit, and provision of written informed consent. Exclusion criteria were bronchiectasis due to cystic fibrosis, active allergic bronchopulmonary aspergillosis, active nontuberculous mycobacterial infection, chronic use of oral corticosteroids, a primary clinical diagnosis of another respiratory disease (chronic obstructive pulmonary disease [COPD] or asthma), and inability to provide informed consent. Ethical approval for the study was given by the East of Scotland Research Ethics Committee (approval number 12/ES/0059).

, chronic use of oral corticosteroids, a primary clinical diagnosis of another respiratory disease (chronic obstructive pulmonary disease [COPD] or asthma), and inability to provide informed consent. Ethical approval for the study was given by the East of Scotland Research Ethics Committee (approval number 12/ES/0059). Severity of disease was evaluated using the Bronchiectasis Severity Index (BSI), as previously described (13). Exacerbations were defined as administration of antibiotics for increasing respiratory symptoms as defined by the British Thoracic Society (14). Sputum was obtained from all subjects during a period of clinical stability and split into whole (unprocessed) sputum for microbiology and sputum that was diluted 1:8 with phosphate-buffered saline and then centrifuged at 3,000 g for 15 minutes at 4°C. All sputum processing took place within 2 hours of expectoration, and freeze–thaw cycles were avoided.

ts during a period of clinical stability and split into whole (unprocessed) sputum for microbiology and sputum that was diluted 1:8 with phosphate-buffered saline and then centrifuged at 3,000 g for 15 minutes at 4°C. All sputum processing took place within 2 hours of expectoration, and freeze–thaw cycles were avoided. Sputum Protein Profiling Sputum protein profiling was performed using nanoflow liquid chromatography/tandem mass spectrometry as previously described (15). Protein identification and label-free quantification were performed using MaxQuant software (https://www.maxquant.org/) (version 1.4.1.2) against the UniProt human database (version 2014-07-09; www.uniprot.org). The false discovery rate for protein identification was set to 1% at the protein level. Data visualization was performed using SIMCA-P software (version 13.0.3; Umetrics). For statistical analysis, the dataset was log2 transformed before being subjected to Student’s t test using Perseus software (https://maxquant.net/perseus/) (version 1.5.4.1). The Benjamini-Hochberg false discovery rate method was used, and corrected P values less than 0.05 were considered significant. PZP ELISA PZP was measured using a commercial ELISA kit for human and mouse (Cloud-Clone Corp.). Validation was performed according to published recommendations (16). Quantification of Sputum NETs Measurement of histone–elastase and DNA–elastase complexes provides a semiquantitative assessment of NETs in sputum, and assays were performed as previously described (17).

PZP ELISA PZP was measured using a commercial ELISA kit for human and mouse (Cloud-Clone Corp.). Validation was performed according to published recommendations (16). Quantification of Sputum NETs Measurement of histone–elastase and DNA–elastase complexes provides a semiquantitative assessment of NETs in sputum, and assays were performed as previously described (17). Leukocyte Studies in Healthy Volunteers Neutrophils and peripheral blood monocytes were isolated from healthy volunteers using Percoll gradient density centrifugation as previously described (18). Immunofluorescence was used to confirm and localize PZP within neutrophils and NETs and to identify colocalization with other neutrophil proteins. NET formation was induced by treatment for 4 hours with 600 nM phorbol myristate acetate (PMA). Colocalization of PZP with neutrophil granule proteins was quantified using the Manders overlap coefficient, which calculates the proportion of overlap of each channel with the other, with a value of 1 indicating perfect colocalization and 0 indicating no colocalization. Electron microscopy was used to identify the cellular location of PZP after staining with anti-PZP antibody and Nanogold secondary antibody (Nanoprobes Inc.). Appropriate negative controls were included.

Colocalization of PZP with neutrophil granule proteins was quantified using the Manders overlap coefficient, which calculates the proportion of overlap of each channel with the other, with a value of 1 indicating perfect colocalization and 0 indicating no colocalization. Electron microscopy was used to identify the cellular location of PZP after staining with anti-PZP antibody and Nanogold secondary antibody (Nanoprobes Inc.). Appropriate negative controls were included. Murine Model of Acute Inflammation Female 10–12-week-old C57BL/6 mice were infected with S. aureus strain RN6390 at an infecting dose of 3 × 108 cfu. At 24 hours after infection, the trachea was carefully dissected and intubated, and BAL was performed with three instillations of 0.4 ml of phosphate-buffered saline. BAL supernatant was used for PZP quantification and cells for cytospins to quantify neutrophils. Sputum Bacteriology Quantitative bacterial culture was performed as described in the online supplement. Microbiota Sequencing PCR and sequencing of the 16S ribosomal (r)RNA gene were performed on an Illumina MiSeq system as previously described (17). Analysis was performed in Quantitative Insights Into Microbial Ecology (detailed methods in online supplement). α Diversity was evaluated using the Shannon-Wiener diversity index and the Berger-Parker index. To compare groups, patients were split into those with predominant (>50% operational taxonomic units [OTUs]) Proteobacteria and those with predominant Firmicutes at the phylum level, as previously described.

ine supplement). α Diversity was evaluated using the Shannon-Wiener diversity index and the Berger-Parker index. To compare groups, patients were split into those with predominant (>50% operational taxonomic units [OTUs]) Proteobacteria and those with predominant Firmicutes at the phylum level, as previously described. Antibiotic Response Study Patients were asked to attend the research center if they developed symptoms of an exacerbation. Patients were reviewed by a physician, and those who were prescribed antibiotics for a protocol-defined exacerbation were included in the study. Patients received treatment for 14 days on the basis of their previous sputum microbiology. Spontaneous sputum samples obtained as baseline and after 14 days were used for PZP measurement. COPD Cohort Study To compare sputum PZP concentrations in bronchiectasis with those from patients with COPD, 40 patients with COPD without underlying bronchiectasis were enrolled while clinically stable (4 wk free from antibiotic or corticosteroid therapy). Spontaneous sputum samples were obtained and processed in the same way as the bronchiectasis samples with sputum PZP and sputum NETs measured by ELISA.

patients with COPD, 40 patients with COPD without underlying bronchiectasis were enrolled while clinically stable (4 wk free from antibiotic or corticosteroid therapy). Spontaneous sputum samples were obtained and processed in the same way as the bronchiectasis samples with sputum PZP and sputum NETs measured by ELISA. Statistical Analysis Statistical analysis was performed using Prism version 8 software (GraphPad Software). Categorical variables are presented as frequencies and percentages, and statistical differences were analyzed using the χ2 test or Fisher’s exact test when required. Continuous variables are presented as mean and SD or median and interquartile range (IQR) when data are not normally distributed. Sputum PZP was not normally distributed, so data were logarithmically transformed and then analyzed using Student’s t test for comparisons of two groups and one-way ANOVA for more than two groups. A paired Student’s t test was used to compare changes in sputum PZP with antibiotic therapy. Linear variables were correlated by Spearman correlation. Principal component analysis (with the dataset logarithmically transformed, mean centered, and unit variance scaled) was performed using SIMCA-P version 13.0.3 software. We defined statistical significance as a two-tailed P < 0.05 for all analyses.

otic therapy. Linear variables were correlated by Spearman correlation. Principal component analysis (with the dataset logarithmically transformed, mean centered, and unit variance scaled) was performed using SIMCA-P version 13.0.3 software. We defined statistical significance as a two-tailed P < 0.05 for all analyses. Results Sputum protein profiling was performed in 20 patients with bronchiectasis (9 with P. aeruginosa infection and 11 without) to explore potential biomarkers relevant to disease severity in bronchiectasis. Characteristics of the patients included are shown in Table E1 in the online supplement. Principal component analysis of sputum protein profiles revealed two distinct clusters in which sputum profiles of patients with P. aeruginosa are well separated from those without (Figure 1A). A total of 80 proteins were found to be significantly associated with P. aeruginosa in sputum. Among differentially expressed proteins, many were previously identified biomarkers of bronchiectasis lung disease, including neutrophil elastase, myeloperoxidase, S100-A8, and S100-A9, as shown in the loadings plot in Figure 1B. We identified PZP, not previously described as present in sputum or as a neutrophil-associated protein, to be differentially expressed between samples with P. aeruginosa infection and those without. Direct comparison of PZP in the P. aeruginosa infected and uninfected groups is shown in Figure E1. The raw proteomics dataset has been uploaded as Table E4. We therefore explored whether PZP was associated with neutrophilic inflammation and bronchiectasis disease severity.

th P. aeruginosa infection and those without. Direct comparison of PZP in the P. aeruginosa infected and uninfected groups is shown in Figure E1. The raw proteomics dataset has been uploaded as Table E4. We therefore explored whether PZP was associated with neutrophilic inflammation and bronchiectasis disease severity. Figure 1. Principal component analysis of sputum protein profiles in bronchiectasis. Twenty patients with bronchiectasis, including nine with Pseudomonas aeruginosa (PA) infection (labeled in green) and eleven without (labeled in black), were included. (A and B) The scores plot based on the first two components is shown in A, and the loadings plot is shown in B. The cumulative R2X = 0.31 and Q2 = 0.21. R2X represents the fraction of the variation of the X variables explained by the first two components, and Q2 indicates the fraction of the variation of the X variables predicted by the model. Sputum PZP (pregnancy zone protein) concentrations are associated with samples with P. aeruginosa infection. Patient Cohort One hundred twenty-four patients were included in the study. There was a slight female predominance and a mean age of 67 years, typical of European bronchiectasis cohorts (19–21). The clinical characteristics of our cohort are shown in Table 1. Table 1. Patient Characteristics

Figure 1. Principal component analysis of sputum protein profiles in bronchiectasis. Twenty patients with bronchiectasis, including nine with Pseudomonas aeruginosa (PA) infection (labeled in green) and eleven without (labeled in black), were included. (A and B) The scores plot based on the first two components is shown in A, and the loadings plot is shown in B. The cumulative R2X = 0.31 and Q2 = 0.21. R2X represents the fraction of the variation of the X variables explained by the first two components, and Q2 indicates the fraction of the variation of the X variables predicted by the model. Sputum PZP (pregnancy zone protein) concentrations are associated with samples with P. aeruginosa infection. Patient Cohort One hundred twenty-four patients were included in the study. There was a slight female predominance and a mean age of 67 years, typical of European bronchiectasis cohorts (19–21). The clinical characteristics of our cohort are shown in Table 1. Table 1. Patient Characteristics Characteristics Data N 124 Age, yr, mean (SD) 69.1 (10.7) Sex, F, % 53.2 BSI Mean (SD) 7.8 (4.2) Mild, n (%) 30 (24.2) Moderate, n (%) 56 (45.2) Severe, n (%) 38 (30.6) Exacerbation, frequency/yr, mean (SD) 2.6 (2.1) FEV1% predicted, mean (SD) 78.1 (25.5) Etiology, n (%) Idiopathic 82 (66.1) COPD 16 (12.9) Postinfective 8 (6.5) Inflammatory bowel disease 4 (3.2) Immunodeficiency 3 (2.4) ABPA* 2 (1.6) Rheumatoid arthritis 2 (1.6) PCD 1 (0.8) Aspiration 1 (0.8) Young syndrome 1 (0.8) Asthma 1 (0.8) GERD 1 (0.8) Hematological malignancy 1 (0.8) Other 1 (0.8) Smoking, n (%) Current smoker 12 (9.7) Ex-smoker 52 (41.9) Never smoker 60 (48.4) Antibiotics, n (%) Long-term macrolides 27 (21.8) Inhaled 5 (4) Definition of abbreviations: ABPA = allergic bronchopulmonary aspergillosis; BSI = Bronchiectasis Severity Index; COPD = chronic obstructive pulmonary disease; GERD = gastroesophageal reflux disease; PCD = primary ciliary dyskinesia.

Never smoker 60 (48.4) Antibiotics, n (%) Long-term macrolides 27 (21.8) Inhaled 5 (4) Definition of abbreviations: ABPA = allergic bronchopulmonary aspergillosis; BSI = Bronchiectasis Severity Index; COPD = chronic obstructive pulmonary disease; GERD = gastroesophageal reflux disease; PCD = primary ciliary dyskinesia. * These patients had previously treated ABPA rather than active disease. Association of Sputum PZP with Disease Severity in Bronchiectasis The median serum concentration of PZP was 4.1 μg/ml (IQR, 2.2–9.9), consistent with published data describing concentrations expected in healthy men and women (<10 μg/ml in men and postmenopausal women, 10–30 μg/ml in premenopausal women) (22, 23). The median sputum concentration was 65.9 μg/ml (IQR, 36.9–205.8). There was no significant difference in sputum PZP between male and female patients (median, 65 μg/ml [IQR, 39.1–215] vs. 66.8 [IQR, 34.7–198.5] μg/ml; P = 0.8). There was no significant correlation between serum and sputum PZP concentrations (see Figure E2). Using the validated BSI, we observed a clear relationship between sputum PZP and bronchiectasis severity. There was a significant elevated median sputum PZP in patients with severe disease of 163 μg/ml (IQR, 64.6–854.1) compared with patients with mild disease (58.6 μg/ml; IQR, 25.3–163.8) or those with moderate disease (52.6 μg/ml; IQR, 24.1–97.3) (P < 0.001) (Figure 2A). It was observed that sputum PZP concentrations were not normally distributed, so data are displayed as log10 values.

h severe disease of 163 μg/ml (IQR, 64.6–854.1) compared with patients with mild disease (58.6 μg/ml; IQR, 25.3–163.8) or those with moderate disease (52.6 μg/ml; IQR, 24.1–97.3) (P < 0.001) (Figure 2A). It was observed that sputum PZP concentrations were not normally distributed, so data are displayed as log10 values. Figure 2. Association between sputum PZP (pregnancy zone protein) and clinical outcomes. (A) Severity of disease as stratified by Bronchiectasis Severity Index (BSI). (B) Exacerbation frequency in previous year. Sputum PZP was also higher in patients with frequent exacerbations (≥3/yr) than in patients with less frequent exacerbations, as shown in Figure 2B. Relationships were also observed with quality of life using the Quality of Life Questionnaire–Bronchiectasis respiratory symptom score (P < 0.0001), and a weaker association was observed with FEV1 percent predicted (r = −0.21; P = 0.02). Patients with higher PZP sputum concentrations also had a higher daily sputum volume (r = 0.2; P = 0.02), whereas hospitalization for a severe exacerbation was also predicted by higher sputum PZP (P < 0.0001). No significant associations were observed between serum PZP and severity of disease (Figure E3).

1; P = 0.02). Patients with higher PZP sputum concentrations also had a higher daily sputum volume (r = 0.2; P = 0.02), whereas hospitalization for a severe exacerbation was also predicted by higher sputum PZP (P < 0.0001). No significant associations were observed between serum PZP and severity of disease (Figure E3). PZP Is a Marker of Airway Chronic Infection Airway infection is linked to disease severity, and we next validated the previous observation that PZP concentrations are higher in patients with chronic airway infection. Using standard microbial culture, we found that the most frequently isolated pathogens were Haemophilus influenzae (n = 29) and P. aeruginosa (n = 16). Overall, 75 patients had chronic infection with pathogens, and 49 did not. As shown in Figure 3A, patients with P. aeruginosa infection had significantly higher concentrations of sputum PZP than patients with no growth of pathogens. H. influenzae, Moraxella catarrhalis, and Enterobacteriaceae were also associated with higher concentrations of sputum PZP.

ection with pathogens, and 49 did not. As shown in Figure 3A, patients with P. aeruginosa infection had significantly higher concentrations of sputum PZP than patients with no growth of pathogens. H. influenzae, Moraxella catarrhalis, and Enterobacteriaceae were also associated with higher concentrations of sputum PZP. Figure 3. Association between sputum PZP (pregnancy zone protein) and microbiological outcomes. (A) Sputum PZP by microorganism growth on standard culture. (B) Sputum PZP by dysbiosis (defined as >50% reads of single phylum on 16S sequencing). (C) Patients were divided into those with PZP above and below the median value for the population, and the percentage operational taxonomic units (OTUs) were compared. The 15 genera most strongly associated with lower PZP concentrations and the 15 genera most strongly associated with higher PZP concentrations are shown. *P < 0.05, **P < 0.001, and ***P < 0.0001. When we assessed bacterial diversity by 16S rRNA sequencing, we observed no relationship between sputum PZP and α diversity using either the Shannon-Wiener diversity index or the Berger-Parker index. PZP concentrations were significantly different in patients with different microbiota profiles at the phylum level, however, with those with Proteobacteria dysbiosis (defined as >50% reads) having significantly higher sputum PZP (P = 0.01) (Figure 3B). Patients with PZP concentrations above the median of the population had a higher average percentage of OTUs classified as Pseudomonas and a lower percentage of OTUs classified as Streptococcus and Veillonella (Figure 3C).

a dysbiosis (defined as >50% reads) having significantly higher sputum PZP (P = 0.01) (Figure 3B). Patients with PZP concentrations above the median of the population had a higher average percentage of OTUs classified as Pseudomonas and a lower percentage of OTUs classified as Streptococcus and Veillonella (Figure 3C). Sputum PZP Is Related to Airway Bacterial Load and Is Reduced by Antibiotic Therapy Bacterial load was quantified in 60 patients with bronchiectasis. The characteristics of this subgroup are shown in the online supplement (Table E2). PZP concentrations were significantly associated with increased bacterial load (P < 0.0001 by ANOVA (Figure 4A). Significant differences were observed between those with bacterial loads above 10 (7) and all other subgroups (P < 0.05 for all pairwise comparisons). Excluding patients infected with P. aeruginosa produced similar results (Figure E4). Figure 4. Association between bacterial load in sputum and PZP (pregnancy zone protein). (A) Sputum bacterial load is associated with sputum PZP (P value refers to comparison by ANOVA). (B) Changes in PZP at the start (Day 0) and end (Day 14) of antibiotic therapy for an acute exacerbation of bronchiectasis. The P value refers to comparison by paired Student’s t test. (C) Comparison of patients in B with positive or negative cultures at the end of antibiotic treatment for an exacerbation. P value refers to unpaired Student’s t test.

(Day 0) and end (Day 14) of antibiotic therapy for an acute exacerbation of bronchiectasis. The P value refers to comparison by paired Student’s t test. (C) Comparison of patients in B with positive or negative cultures at the end of antibiotic treatment for an exacerbation. P value refers to unpaired Student’s t test. Reduction in bacterial load with antibiotic treatment of exacerbations was associated with reduced sputum PZP concentrations. Twenty patients were treated, 18 of whom had detectable bacterial loads at baseline. Figure 4B shows the changes in sputum PZP from baseline to the end of treatment after 14 days. PZP was significantly reduced (P = 0.0014 by paired t test). Six patients, five of whom had P. aeruginosa infection, had detectable bacterial loads despite 14 days of antibiotics. Those patients who remained culture positive after 14 days had higher PZP at the end of treatment than those who achieved bacterial clearance (P = 0.01) (Figure 4C). Taken together, these data show that PZP increases with increasing bacterial load, regardless of the infecting pathogen, and that reducing bacterial load with antibiotic therapy reduces sputum PZP.

e after 14 days had higher PZP at the end of treatment than those who achieved bacterial clearance (P = 0.01) (Figure 4C). Taken together, these data show that PZP increases with increasing bacterial load, regardless of the infecting pathogen, and that reducing bacterial load with antibiotic therapy reduces sputum PZP. PZP Is Released from Neutrophils during Activation and Acute Infection We next examined the potential source of PZP detected in the bronchiectasis airway. There was no detectable PZP in the supernatants or cell lysates of primary human bronchial epithelial cells. Mononuclear cells released small amounts of PZP into the supernatant (mean, 46.7 ng/ml; SD, 18.2; n = 4), with higher concentrations seen upon stimulation with 2.5 ng/ml PMA (mean, 177.8 ng/ml; SD, 6.4; n = 4). Neutrophils secreted large amounts of PZP when stimulated with formylmethionylleucylphenylalanine and PMA (Figure 5A). A dose-dependent relationship was noted with all stimulants. PZP was also released after incubation with bacteria (Escherichia coli and P. aeruginosa) in a dose- and time-dependent manner (Figures 5A and E5). Immunofluorescence by confocal microscopy confirmed the presence of PZP in neutrophils (Figures 5B, E6, and E7 and Videos E1 and E2). Staining was seen in a diffuse, punctate pattern throughout the cytoplasm, possibly suggestive of granules; however, significant colocalization with other granule proteins (neutrophil elastase, lactoferrin, myeloperoxidase, and matrix metalloproteinase 9) was not observed (Manders overlap coefficient, <0.3).

1 and E2). Staining was seen in a diffuse, punctate pattern throughout the cytoplasm, possibly suggestive of granules; however, significant colocalization with other granule proteins (neutrophil elastase, lactoferrin, myeloperoxidase, and matrix metalloproteinase 9) was not observed (Manders overlap coefficient, <0.3). Figure 5. PZP (pregnancy zone protein) is released from neutrophils in response to formylmethionylleucylphenylalanine (fMLP), phorbol myristol acetate (PMA), and bacterial infection in mice and humans. (A) Neutrophils (n = 107) were treated with the indicated stimuli for 30 minutes, and PZP was measured in cell-free supernatants. The no-stimulation control was phosphate-buffered saline (PBS) only. Statistically significant differences were obtained using Student’s t test (n = 4 biological replicates per condition). (B) C57BL/6 mice aged 10–12 weeks were infected with Staphylococcus aureus strain RN6390 (n = 18) or PBS control (n = 4). BAL PZP was measured by ELISA. Significant differences were determined using Student’s t test. An association was found between BAL neutrophils and BAL PZP (n = 10 mice undergoing BAL, all infected with S. aureus). (C) Microscopic image of neutrophils showing DNA. Blue = DAPI; green = neutrophil elastase; red = PZP.

). BAL PZP was measured by ELISA. Significant differences were determined using Student’s t test. An association was found between BAL neutrophils and BAL PZP (n = 10 mice undergoing BAL, all infected with S. aureus). (C) Microscopic image of neutrophils showing DNA. Blue = DAPI; green = neutrophil elastase; red = PZP. We studied whether acute pulmonary infection, which provokes a neutrophil-mediated inflammatory response, would result in an increase in PZP in BAL. S. aureus was used as a common bronchiectasis pathogen that provokes a robust neutrophil response, including the formation of NETs. At 24 hours after infection, S. aureus was detected in lung homogenates of infected mice, and this was associated with BAL neutrophilia. As shown in Figure 5C, infection resulted in an increase in PZP in BAL that was directly correlated to BAL neutrophil count.

bust neutrophil response, including the formation of NETs. At 24 hours after infection, S. aureus was detected in lung homogenates of infected mice, and this was associated with BAL neutrophilia. As shown in Figure 5C, infection resulted in an increase in PZP in BAL that was directly correlated to BAL neutrophil count. PZP Is Found in the Nucleus and Cytoplasm of Neutrophils and Released into NETs On the basis of the observation that PMA, an inducer of NET formation, was a strong stimulus for PZP release from neutrophils, we hypothesized that PZP may be a marker of NETosis. In bronchiectasis sputum in vivo, we identified a strong correlation between PZP in sputum and NETs measured by histone–elastase or DNA–elastase complexes (Figure 6A). Experimentally induced NETs from healthy control neutrophils treated with PMA showed staining for PZP in association with DNA and neutrophil elastase in a “studded” pattern typical of NETs (Figure 6B). To further investigate the localization of PZP in neutrophils, we used immune electron microscopy. This revealed staining within the euchromatin of the nuclei and a diffuse cytoplasmic pattern of PZP within the neutrophil. Our granulocyte extraction method produces neutrophil preparations that are >95% pure but may contain small numbers of eosinophils. We unexpectedly observed positive staining of eosinophils in a similar pattern to neutrophils with cytoplasmic and nuclear staining.

ffuse cytoplasmic pattern of PZP within the neutrophil. Our granulocyte extraction method produces neutrophil preparations that are >95% pure but may contain small numbers of eosinophils. We unexpectedly observed positive staining of eosinophils in a similar pattern to neutrophils with cytoplasmic and nuclear staining. Figure 6. PZP (pregnancy zone protein) is a cytoplasmic and nuclear protein that is released in neutrophil extracellular traps (NETs). (A) In patients with bronchiectasis, extracellular PZP correlates with markers of NET formation. (B) Peripheral blood neutrophils were induced to undergo NET formation with 600 nM phorbol myristol acetate for 4 hours. The NETs contain PZP. (C and D) Electron microscopy of neutrophils (C) and neutrophils plus eosinophils (D) showing diffuse cytoplasmic and nuclear staining for PZP (black dots). Bronchiectasis is not alone in causing chronic neutrophilic inflammation of the airway. COPD is characterized by variable degrees of both neutrophilic and eosinophilic inflammation, and NETs have been reported in the COPD airway. We therefore measured PZP and NETs in sputum from patients with COPD during a period of stability and compared these with samples obtained from our bronchiectasis cohort. Characteristics of the subjects with COPD are shown in Table E3. Both PZP and NETs were significantly lower in subjects with COPD than in those with bronchiectasis (Figure 7).

and NETs in sputum from patients with COPD during a period of stability and compared these with samples obtained from our bronchiectasis cohort. Characteristics of the subjects with COPD are shown in Table E3. Both PZP and NETs were significantly lower in subjects with COPD than in those with bronchiectasis (Figure 7). Figure 7. (A) PZP (pregnancy zone protein) concentrations in sputum in patients with bronchiectasis (n = 124) or chronic obstructive pulmonary disease (COPD) (n = 40). (B) Neutrophil extracellular traps measured using the histone–elastase complex assay in bronchiectasis and COPD. For both assays, comparisons are by Student’s t test. Discussion This study has identified PZP, which was previously described as a serum protein elevated in the blood of pregnant women, as an unexpected component of NETS. It was found to be released into the bronchiectasis airway during chronic infection, and it is elevated in patients with the most severe disease and frequent exacerbations. We found concentrations of PZP in sputum at least 10 times higher than concentrations in serum, suggestive of local production by inflammatory cells in the airway. It is known that patients with more severe bronchiectasis have higher degrees of airway neutrophilic inflammation (24), including markers such as neutrophil elastase (25), matrix metalloproteinases (26, 27), and cathelicidin (28). PZP followed a similar pattern with clear associations with the multidimensional BSI, a higher concentration in the “frequent exacerbator phenotype,” and an association with respiratory symptoms.

mmation (24), including markers such as neutrophil elastase (25), matrix metalloproteinases (26, 27), and cathelicidin (28). PZP followed a similar pattern with clear associations with the multidimensional BSI, a higher concentration in the “frequent exacerbator phenotype,” and an association with respiratory symptoms. The major driver of airway neutrophilic inflammation in bronchiectasis is bacterial infection according to the vicious vortex concept of bronchiectasis pathophysiology. We demonstrate that PZP is increased in patients with chronic infection, particularly with P. aeruginosa and H. influenzae, both frequent pathogens in this disease. Molecular “microbiome”–based analysis of sputum confirmed elevated PZP in patients with Proteobacteria dysbiosis and airway dominance of Pseudomonas. This correlates with the clinically observed phenotypes in which patients who regularly culture Proteobacteria (predominantly Pseudomonas and Haemophilus) display clinically more severe phenotypes (29). We demonstrate that neutrophils are the likely source of PZP, although we found that, in common with many immune proteins, PZP was present in a number of cell types, including monocytes and eosinophils. PZP appears to be present throughout the cytoplasm of neutrophils and is not concentrated within primary or secondary granules; it is also consistently visible within the nuclei of both neutrophils and eosinophils. We demonstrate in a mouse model of S. aureus pneumonia that PZP is released into BAL after infection in parallel with neutrophil influx.

ghout the cytoplasm of neutrophils and is not concentrated within primary or secondary granules; it is also consistently visible within the nuclei of both neutrophils and eosinophils. We demonstrate in a mouse model of S. aureus pneumonia that PZP is released into BAL after infection in parallel with neutrophil influx. Neutrophils eliminate pathogens through phagocytosis, a relatively noninflammatory intracellular process, and extracellularly through degranulation. NET formation is a distinct antimicrobial pathway in which neutrophils can extrude extracellular DNA, histones, and bactericidal proteins intended to trap and neutralize pathogens. Components released in NETs include antimicrobial peptides (lactoferrin, defensins, LL-37, and bacterial permeability–increasing protein), proteases (neutrophil elastase, proteinase 3, and gelatinase), and enzymes responsible for reactive oxygen species generation (myeloperoxidase). NETs have been described in many chronic diseases, including chronic respiratory diseases such as COPD and cystic fibrosis. It is believed that they serve a beneficial role in preventing spread of bacterial infection but also contribute to tissue damage.

onsible for reactive oxygen species generation (myeloperoxidase). NETs have been described in many chronic diseases, including chronic respiratory diseases such as COPD and cystic fibrosis. It is believed that they serve a beneficial role in preventing spread of bacterial infection but also contribute to tissue damage. The identification of PZP in NETs in bronchiectasis is intriguing because it has no known antimicrobial effects. The known effects of PZP are as an antiproteinase and as a powerful T-cell immunosuppressant. We speculate PZP may be involved in immunological tolerance in patients with bronchiectasis by interacting with NETs to modulate T-cell functions. The persistence of bacterial infection despite an apparently intact cell-mediated immune system is a feature of bronchiectasis that remains unexplained. It is well established that both gram-positive and gram-negative organisms can induce NET formation, and it has been shown that neutrophils undergo NET formation in circumstances in which phagocytosis is prevented, such as with physical barriers preventing phagocytosis, as may occur with biofilms or with high bacterial loads. We demonstrated a strong association between PZP and elevated airway bacterial load above 107 cfu/g, suggesting that NET formation may be the dominant neutrophil phenotype in patients with high bacterial loads. Antibiotic therapy was able to reduce PZP concentrations consistent with a cause-and-effect relationship between bacterial infection and elevated PZP.

een PZP and elevated airway bacterial load above 107 cfu/g, suggesting that NET formation may be the dominant neutrophil phenotype in patients with high bacterial loads. Antibiotic therapy was able to reduce PZP concentrations consistent with a cause-and-effect relationship between bacterial infection and elevated PZP. The positive staining of eosinophils in a similar pattern to neutrophils is of interest. The inflammation in bronchiectasis is a predominantly neutrophilic process, but eosinophilic phenotypes do exist. Multiple previous reports indicate that PZP is regulated by estrogens and other female reproductive hormones, but we found no evidence of a sex-based difference in serum or sputum PZP concentrations. Bronchiectasis has a female predominance, but our evidence suggests that PZP is unlikely to play any role in this sex disparity. Both the male and female patient groups in our cohort are elderly. It has been shown that the physiological fluctuations in estrogen concentration affect exacerbations and P. aeruginosa mucoid transformation in cystic fibrosis (30), but such fluctuations will not occur in a predominantly postmenopausal cohort. Testosterone has also been shown to affect the antimicrobial susceptibility of P. aeruginosa and has established roles in host defense (31). Future studies are required to define how sex hormones may be implicated in the pathogenesis of bronchiectasis, but we found no evidence in established disease that PZP was associated with sex differences.

shown to affect the antimicrobial susceptibility of P. aeruginosa and has established roles in host defense (31). Future studies are required to define how sex hormones may be implicated in the pathogenesis of bronchiectasis, but we found no evidence in established disease that PZP was associated with sex differences. In summary, this study has identified a novel neutrophil protein that is released predominantly from neutrophil cytoplasm into NETs during chronic inflammation. We demonstrate high concentrations of PZP in the airway in severe disease and in patients with high bacterial loads. These data suggest, for the first time, to our knowledge, that NETs are present in the bronchiectasis airway, are associated with disease severity, and contain proteins with the potential to have profound effects on the innate and adaptive immune system. The validation of this observation in multiple datasets using proteomics and ELISA suggests that this finding is robust. Bronchiectasis is a heterogeneous disease, and there is increasing interest in identifying phenotypes and endotypes with distinct clinical outcomes and treatment responses. The finding that PZP is a marker of neutrophil-mediated inflammation may be important for other diseases in which neutrophils play a crucial role. Further mechanistic work is required to determine whether PZP is simply a marker of chronic neutrophilic inflammation or if it has a direct role in the pathogenesis of chronic infection in bronchiectasis.

r of neutrophil-mediated inflammation may be important for other diseases in which neutrophils play a crucial role. Further mechanistic work is required to determine whether PZP is simply a marker of chronic neutrophilic inflammation or if it has a direct role in the pathogenesis of chronic infection in bronchiectasis. The strength of this study is the use of multiple methods of clinical assessment and two different cohorts to validate our findings, as well as the confirmation of the finding of PZP in neutrophils using multiple methods, including ELISA, immunofluorescence, and electron microscopy. Nevertheless, the study has limitations. There are no animal models of bronchiectasis, so there is no direct method of testing whether PZP is directly involved in the pathogenesis of bronchiectasis. We used an acute model of pulmonary inflammation with S. aureus because this organism promotes a robust neutrophil-mediated response, but we acknowledge that alternative models, such as models of chronic P. aeruginosa, may provide complementary information. In conclusion, this study has identified a novel protein in the bronchiectasis airway associated with NETs, chronic infection, and disease severity. Supported by the Scottish Government Chief Scientist Office (clinical academic fellowship [S.F.] and senior clinical fellowship [J.D.C.]) and by the British Lung Foundation through the GSK/British Lung Foundation Chair of Respiratory Research.

In conclusion, this study has identified a novel protein in the bronchiectasis airway associated with NETs, chronic infection, and disease severity. Supported by the Scottish Government Chief Scientist Office (clinical academic fellowship [S.F.] and senior clinical fellowship [J.D.C.]) and by the British Lung Foundation through the GSK/British Lung Foundation Chair of Respiratory Research. Author Contributions: Conception and design: S.F., A. Shoemark, and J.D.C. Patient recruitment and data collection: S.F., A. Shoemark, T.C.F., and J.D.C. Laboratory experiments: S.F., A. Shoemark, A.J.D., H.R.K., A. Smith, S.O., B.T., J.-Y.C., D.C., and J.T.J.H. Analysis and interpretation: S.F., A. Shoemark, J.T.J.H., and J.D.C. Drafting of the manuscript: S.F., A. Shoemark, and J.D.C. Revision of the manuscript and final approval: all authors. This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org. Originally Published in Press as DOI: 10.1164/rccm.201812-2351OC on July 2, 2019 Author disclosures are available with the text of this article at www.atsjournals.org.

At a Glance Commentary Scientific Knowledge on the Subject The preponderance of data associate complement with pulmonary arterial hypertension (PAH) via analysis of circulating complement components without studying the lung directly. In this study, we attempted to resolve this by analyzing both the local lung-specific processes in experimental animal pulmonary hypertension models and human lung specimens and the correlation of dysregulated complement to clinical outcome in patients with PAH. What This Study Adds to the Field We believe our study demonstrates, for the first time, that the immunoglobulin-driven activation of the complement cascade, and specifically its alternative pathway, in the pulmonary vascular adventitia is a critical mechanism initiating proinflammatory responses in pulmonary hypertension and PAH.

udy Adds to the Field We believe our study demonstrates, for the first time, that the immunoglobulin-driven activation of the complement cascade, and specifically its alternative pathway, in the pulmonary vascular adventitia is a critical mechanism initiating proinflammatory responses in pulmonary hypertension and PAH. Pulmonary hypertension (PH) is a debilitating cardiopulmonary disorder with an average life expectancy <5 years from the time of diagnosis. Inflammation and immunity have emerged as critical early pathogenic elements of pulmonary arterial hypertension (PAH) (1–3). Perivascular and adventitial accumulation of monocytes and macrophages and augmented expression of proinflammatory cytokines and chemokines (GM-CSF, CCL2, CX3CL1, CXCL12, and IL6), have been consistently reported in patients with PAH and experimental preclinical PH models (4–8). Moreover, immune dysregulation has been suggested to underlie certain forms of PAH (2, 9). Nevertheless, the mechanisms triggering initial activation of the immune system and inflammation in noninfectious forms of PAH remain elusive.

nsistently reported in patients with PAH and experimental preclinical PH models (4–8). Moreover, immune dysregulation has been suggested to underlie certain forms of PAH (2, 9). Nevertheless, the mechanisms triggering initial activation of the immune system and inflammation in noninfectious forms of PAH remain elusive. The complement system is an essential component of innate immunity; however, it exerts functions beyond those of targeting pathogenic threats. Importantly, this versatile pathway of immune defense can be triggered to become a potent mediator driving various inflammatory diseases (10). Little is known about the role of complement in PH and PAH pathogenesis, particularly its regulation in hypoxic signaling—“sterile inflammation” (11)—or whether complement is important in PAH clinically. Although several studies of PAH have focused on complement activation in the circulation (12, 13), emerging evidence suggests an important role for local, tissue- or cell-specific production of complement components and activation of the complement cascade (14). Complement is activated through three major interconnected pathways: classical, alternative, and lectin (10, 15). The alternative pathway has received particular attention in diseases characterized by sterile inflammation (11, 16). A traditional view that activation of the complement cascade by antigen–antibody immune complexes involves only the classical pathway has been recently challenged by studies demonstrating that the lectin and alternative pathways can be triggered by antibodies or immune complexes (17–21). An important role of alternative pathway–driven antibody-mediated complement activation has been demonstrated in the pathophysiology of aortic aneurysm, rheumatoid arthritis, and ischemia/reperfusion (14, 16, 18–22). Importantly, even when complement activation is initiated through classical or lectin pathways, 90% of downstream complement degradation fragments can be generated through the alternative pathway amplification loop (23).

pathophysiology of aortic aneurysm, rheumatoid arthritis, and ischemia/reperfusion (14, 16, 18–22). Importantly, even when complement activation is initiated through classical or lectin pathways, 90% of downstream complement degradation fragments can be generated through the alternative pathway amplification loop (23). In this study, we tested the hypothesis that activation of the complement cascade, specifically involving the alternative pathway, is a critical pathobiological mechanism regulating the early proinflammatory and pro-proliferative processes that characterize experimental hypoxic PH. Furthermore, we sought to determine whether there was evidence of complement activation in rodent models of more severe PH (sugen-hypoxia, monocrotaline) and in late-stage human PAH. Finally, we elucidated whether circulating complement components can serve as biomarkers of disease outcome in human PAH. Some of the results of these studies have been previously reported in the form of abstracts (24, 25).

In this study, we tested the hypothesis that activation of the complement cascade, specifically involving the alternative pathway, is a critical pathobiological mechanism regulating the early proinflammatory and pro-proliferative processes that characterize experimental hypoxic PH. Furthermore, we sought to determine whether there was evidence of complement activation in rodent models of more severe PH (sugen-hypoxia, monocrotaline) and in late-stage human PAH. Finally, we elucidated whether circulating complement components can serve as biomarkers of disease outcome in human PAH. Some of the results of these studies have been previously reported in the form of abstracts (24, 25). Methods Animal Models Mice (male) were purchased from Jackson Laboratories: C57BL/6J, B10.D2-Hc0 (C5 [complement component 5]-deficient [C5−/−]), C3 (complement component 3)-deficient (C3−/−), and B6.129S2-Ighmtm1Cgn/J (μMT− mice lacking mature B lymphocytes and thus lacking all circulating immunoglobulins) (26). Cfb (complement factor B)-deficient (Cfb−/−) mice were bred in-house (27). Wistar-Kyoto male rats were from Charles Rivers Laboratories. On delivery from the vendor, all animals were acclimatized for at least a week in a sea-level (SL) chamber (barometric pressure [PB] = 760 mm Hg) because PB is 640 mm Hg at Denver altitude. In-house–bred Cfb−/− mice were placed into SL chambers on weaning. Thereafter, control groups remained in SL chambers, whereas experimental groups were placed for 3 days into hypobaric (PB = 380 mm Hg) hypoxic chambers (with oxygen levels approximately 12%; sample size for each SL or hypoxic group was 6–8 rats or 8–12 mice) (4, 28, 29). Six IgG-injected hypoxic μMT− mice were used. Standard veterinary care was provided in compliance with institutional animal care and use committee–approved protocols at the University Colorado Denver. Specimens of bovine lung tissues were obtained from Holstein neonatal (15-d-old) male calves; the experimental hypoxic group (n = 7) was exposed from Day 1 after birth for 2 weeks to hypobaric hypoxia (PB = 445 mm Hg), whereas age-matched controls (n = 6) were kept at ambient altitude (PB = 640 mm Hg), as described previously (4).

bovine lung tissues were obtained from Holstein neonatal (15-d-old) male calves; the experimental hypoxic group (n = 7) was exposed from Day 1 after birth for 2 weeks to hypobaric hypoxia (PB = 445 mm Hg), whereas age-matched controls (n = 6) were kept at ambient altitude (PB = 640 mm Hg), as described previously (4). Additional experimental animal information can be found in the online supplement (ANIMAL MODELS). Human Tissues Human lung specimens from normal (rejected) donors and patients with idiopathic PAH (IPAH; n = 6, each group, Table E3 in the online supplement) were provided by the Pulmonary Hypertension Breakthrough Initiative, funded through an NHLBI R24 grant (No. R24HL123767) and the Cardiovascular Medical Research Education Fund. Immunofluorescent and immunohistochemistry (IHC) staining and quantification, qRT-PCR, in vitro experiments, GM-CSF ELISA, RNAscope in situ hybridization, IgG injections of μMT− mice, and right ventricular systolic pressure (RVSP) assessment were performed as described in the online supplement. Statistical Analysis Data are presented as mean ± SEM. GraphPad Prism 6.0 (GraphPad Software Inc) was used to determine significance. Unpaired, two-tailed Student t test was used to compare two groups. One-way ANOVA and Sidak correction for multiple comparisons were used to compare more than two groups. The Kolmogorov-Smirnov, Shapiro-Wilk, and D’Agostino tests were used to assess for normality before applying parametric statistical tests. P value significance was set at 0.05.

t t test was used to compare two groups. One-way ANOVA and Sidak correction for multiple comparisons were used to compare more than two groups. The Kolmogorov-Smirnov, Shapiro-Wilk, and D’Agostino tests were used to assess for normality before applying parametric statistical tests. P value significance was set at 0.05. Developing the Complement–PAH Network Patient cohorts Patients with IPAH or heritable PAH (n = 218) were recruited at the National Pulmonary Hypertension service at Hammersmith Hospital, London, United Kingdom. The diagnostic criteria for IPAH or heritable PAH over the course of this study were stable: raised mean pulmonary artery pressure of more than 25 mm Hg, with pulmonary capillary wedge pressure less than 15 mm Hg (and pulmonary vascular resistance >3 Wood units) at rest with exclusion of known associated diseases. The guidelines quoted were internationally agreed. All samples and data were obtained with informed consent and local research ethics committee approval. We assessed patients for eligibility between October 24, 2002, and August 13, 2013. Patients were censored on May 15, 2014, using National Health Service records. Median follow-up was 2.2 years. At the end of the follow-up period, 55 patients had died; no patients were lost to follow-up. The mortality of this cohort was previously presented (12). Nine patients underwent lung or heart and lung transplantation and were censored at this date. Patients were not fasting and were sampled at their routine clinical appointment visits. Peripheral venous blood samples were collected using EDTA. Vacutainer tubes (BD Biosciences) were immediately put on ice, centrifuged (1,300 × g, 15 min) within 30 minutes of collection, and plasma aliquots were stored at −80°C until required.

e not fasting and were sampled at their routine clinical appointment visits. Peripheral venous blood samples were collected using EDTA. Vacutainer tubes (BD Biosciences) were immediately put on ice, centrifuged (1,300 × g, 15 min) within 30 minutes of collection, and plasma aliquots were stored at −80°C until required. Complement–PAH network development Forty prognostic plasma proteins identified in a published proteomic analysis of patients with IPAH or heritable PAH (12) were evaluated, using the (incomplete) consolidated human interactome, which contains information on 170,303 protein–protein interactions (30) (Table E4). Nodes in the consolidated human interactome represent proteins, and edges represent functional associations rather than physical protein–protein binding (31). Expression data from functionally associated proteins in the complement–PAH network were used to determine whether PAH subgroups could be identified from the original PAH study population (12). We used the “elbow” method to determine the best number of clusters K. The elbow method ran K-means clustering on the data set for a range of values for K (e.g., 1–8) and, for each value of K, calculated the sum of squared errors (SSE). The optimal number is the elbow position. The underlying assumption with this method is that increasing the number of clusters beyond the elbow position will not further reduce SSE substantially. Using the elbow chart, the greatest decrease in slope for SSE across sequentially clusters was at K 2–3 from K 1–2. Therefore, we selected K = 2 and then used K-means clustering algorithm to cluster the data into two clusters. The network-based patient cluster assignment was tested for association with all-cause mortality from time of sampling by Kaplan–Meier analysis, as described previously (12). Kaplan–Meier plots illustrating events (deaths) in relation to biomarker levels were assessed by the log-rank test.

r the data into two clusters. The network-based patient cluster assignment was tested for association with all-cause mortality from time of sampling by Kaplan–Meier analysis, as described previously (12). Kaplan–Meier plots illustrating events (deaths) in relation to biomarker levels were assessed by the log-rank test. Results Hypoxia Activates the Alternative Pathway of Complement Little is known about the mechanisms involved in the early initiating stages of PH, especially those shaping pulmonary vascular microenvironments that would further perpetuate vascular remodeling. We sought to examine whether the early responses to sterile environmental stimuli (in the absence of infection or extensive injury) would be associated with complement activation, specifically via the alternative pathway that has been reported in diseases characterized by sterile inflammation (11, 16). We chose to use a commonly accepted rodent model of hypoxia-induced PH in its early stage (3-d hypoxic challenge), when the cellular processes characteristic of PH pathogenesis (accumulation of monocytes and macrophages, increased cell proliferation) are observed (4, 7, 28, 29).

erized by sterile inflammation (11, 16). We chose to use a commonly accepted rodent model of hypoxia-induced PH in its early stage (3-d hypoxic challenge), when the cellular processes characteristic of PH pathogenesis (accumulation of monocytes and macrophages, increased cell proliferation) are observed (4, 7, 28, 29). Lungs of 3-day hypoxic wild-type (WT) mice and rats demonstrated robust vascular-specific deposition of complement C3, whereas no C3 deposition was detected in SL animals (Figure 1A). Furthermore, markedly increased numbers of cells expressing receptors for anaphylatoxins C5a and C3a (C5aR1 and C3aR1, respectively) were observed in perivascular and periairway areas of elastic pulmonary arteries (Figures 1Ba–1Bc, quantified in Figure 1Bd, and confirmed in whole-lung lysates via RT-PCR in Figure 1Be).

Furthermore, markedly increased numbers of cells expressing receptors for anaphylatoxins C5a and C3a (C5aR1 and C3aR1, respectively) were observed in perivascular and periairway areas of elastic pulmonary arteries (Figures 1Ba–1Bc, quantified in Figure 1Bd, and confirmed in whole-lung lysates via RT-PCR in Figure 1Be). Figure 1. Lungs of hypoxic rodents demonstrate prominent vascular-specific deposition of complement C3 (component 3) and robust accumulation of cells expressing complement anaphylatoxin receptors. (A) Hypoxia-induced deposition of complement C3 in mice (a, green fluorochrome) and rats (b, red fluorochrome) is prominently observed in perivascular areas and also encompasses medial and luminal areas. (Aa and Ab) Lung cryosections were labeled with species-specific anti–complement C3 antibodies conjugated with green (a, for mouse) or red (b, for rat) fluorochrome. Cell nuclei are labeled with DAPI (blue). Scale bars, 100 μm. (B) Expression of receptors for complement anaphylatoxins C5a and C3a (C5aR1 and C3aR1, respectively) is markedly upregulated in the lungs of animals with experimental pulmonary hypertension. In rodents exposed to hypoxia (HX) for three days (a, mice; b and c, rats), cells expressing C5aR1 and C3aR1 (red fluorochrome) are markedly increased in numbers and localized to perivascular areas, whereas only a few pulmonary adventitial cells express these receptors in the lungs of sea-level (SL) animals. Lung cryosections were labeled with species-specific monoclonal antibodies against C5aR1 (mice and rats) and C3aR1 (available for rats only). Autofluorescence of elastic lamellae (green) defines tunica media, and cell nuclei are labeled with DAPI (blue). Scale bars, 100 μm. (Bd) Quantification of red fluorescence was performed as described in the online supplement and is presented in arbitrary units (AU). (Be) qRT-PCR analysis demonstrates that expression levels of C5ar1 and C3ar1 mRNA in the whole lungs of mice and rats are significantly upregulated by 3-day exposure to HX compared with SL controls. Unpaired/two-tailed test was performed for comparing two groups. **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001. AW = airway; PA = pulmonary artery.

onstrates that expression levels of C5ar1 and C3ar1 mRNA in the whole lungs of mice and rats are significantly upregulated by 3-day exposure to HX compared with SL controls. Unpaired/two-tailed test was performed for comparing two groups. **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001. AW = airway; PA = pulmonary artery. RT-PCR analysis of whole lungs of 3-day hypoxic mice demonstrated robust augmentation of the key activator of the complement alternative pathway, Cfb, and modest increases in complement C3 mRNA but no change in expression of regulators or inhibitors of the alternative pathway, Cfh (complement factor H), and Cd55/Daf (Figure 2A). Moreover, no hypoxia-induced changes were detected for complement C4b mRNA (defining the classical and lectin pathways; Figure E1). These findings in whole mouse lungs were corroborated by RNA-sequencing analysis of flow-sorted murine lung interstitial macrophages (IMs) in our previously published article (29), in which hypoxic IMs showed 3.2-fold increases in Cfb and concurrent decreases in Cfh (−1.43-fold) and Cd55/Daf (−2.47-fold) over SL controls (Table E1). Because these results suggested hypoxia-induced activation or amplification of the alternative pathway locally within the lung, we sought to identify the cellular source of Cfb. RNAscope in situ hybridization demonstrated minimal Cfb expression in SL mice, whereas robust Cfb upregulation was observed in pulmonary adventitia and airways of hypoxic mice (Figures 2B and E2). Thus, the alternative pathway and its activator Cfb emerged as critical hypoxia-induced constituents in the lung vasculature.

Ascope in situ hybridization demonstrated minimal Cfb expression in SL mice, whereas robust Cfb upregulation was observed in pulmonary adventitia and airways of hypoxic mice (Figures 2B and E2). Thus, the alternative pathway and its activator Cfb emerged as critical hypoxia-induced constituents in the lung vasculature. Figure 2. The alternative and terminal C5 (component 5) complement pathways are essential in driving hypoxia-induced proinflammatory processes in pulmonary perivascular areas. (A and B) The alternative pathway of complement is activated by hypoxic exposure. (A) qRT-PCR analysis of whole lung extracts from wild-type (WT) mice exposed to hypoxia (HX) for 3 days demonstrates robust augmentation of Cfb (complement factor B) and modest upregulation of complement C3 mRNA expression levels compared with sea-level (SL) controls. Unpaired/two-tailed test was performed for comparing two groups. (B) Lungs of 3-day HX WT mice demonstrate robust augmentation of Cfb mRNA expression, as detected via RNAscope in situ hybridization, in cells localized to pulmonary artery (PA) perivascular areas and airways (AWs). Fast Red chromogen (red), used for message detection in in situ hybridization, can be visualized by both light (upper panels) and fluorescent (bottom panels) microscopy. Gill’s hematoxylin (blue) was used for nuclear counterstaining in light microscopy imaging (upper panels). (C–E) Attenuated accumulation of CD68+ macrophages is observed in the lungs of complement (Cfb and C5)-deficient HX mice compared with WT HX counterparts, as validated by CD68 immunostaining (C, red fluorochrome) and its quantification (D) performed as described in the online supplement and presented in arbitrary units (AU), as well as by qRT-PCR of whole lung tissues (E). (Unpaired/two-tailed test was performed for comparing two groups in each cohort). Scale bars, 100 μm. *P < 0.05, **P ≤ 0.01, and ****P ≤ 0.0001; ns = not significant. Cfh = complement factor H; CT = cycle threshold.

nline supplement and presented in arbitrary units (AU), as well as by qRT-PCR of whole lung tissues (E). (Unpaired/two-tailed test was performed for comparing two groups in each cohort). Scale bars, 100 μm. *P < 0.05, **P ≤ 0.01, and ****P ≤ 0.0001; ns = not significant. Cfh = complement factor H; CT = cycle threshold. Hypoxia-induced Perivascular Inflammation Is Complement Dependent A distinctive feature of PH is early and persistent perivascular inflammation (3, 4, 32), as confirmed in 3-day hypoxic WT mice (Figure E3). Using mice genetically deficient in specific complement components, we tested the hypothesis that hypoxia-induced lung inflammation is complement dependent. Surprisingly, and in contrast to a previous publication (33), complement C3−/− mice demonstrated robust accumulation of CD68+ macrophages, augmented Il6 and Ccl2 expression, and cell proliferation (Figure E4), potentially due to previously described generation of anaphylatoxin C5a in C3−/− mice (15). Consequently, C3−/− mice were not further analyzed in this study. We next tested the Cfb−/− strain (to determine the role of the alternative pathway), and the C5−/− strain (to define the role of the terminal C5 pathway). Although hypoxic WT mice displayed increases in perivascular CD68+ macrophages, the lungs of hypoxic Cfb−/− and C5−/− mice demonstrated decreased accumulation of CD68+ macrophages (Figures 2C–2E).

n (to determine the role of the alternative pathway), and the C5−/− strain (to define the role of the terminal C5 pathway). Although hypoxic WT mice displayed increases in perivascular CD68+ macrophages, the lungs of hypoxic Cfb−/− and C5−/− mice demonstrated decreased accumulation of CD68+ macrophages (Figures 2C–2E). Next we examined expression and localization of Csf2/GM-CSF, a potent cytokine implicated in mobilization and proinflammatory activation of monocytes and macrophages in PAH (34). In situ hybridization demonstrated that, in SL-WT mice, Csf2 was localized mainly to airways but not to pulmonary vasculature, whereas robust upregulation of Csf2 expression was detected in pulmonary arteries of hypoxic WT mice (Figure 3A). Remarkably, Cfb−/− and C5−/− mice demonstrated abrogation of hypoxia-induced Csf2 upregulation in lung vasculature. Airways of all mouse strains, SL or hypoxic, maintained Csf2 expression without any significant visual change. This was verified by qRT-PCR demonstrating abrogation of hypoxia-induced Csf2 upregulation in mice deficient in complement (Cfb and C5; Figure 3B). In addition, we analyzed expression of a potent monocyte chemoattractant, Ccl2/MCP1. In situ hybridization demonstrated significant hypoxia-induced Ccl2/MCP1 upregulation in vasculature and airways (Figure 3C), whereas MCP1 protein was detected only in perivascular and adventitial areas (Figure 3D). RT-PCR analysis showed potent upregulation of Ccl2 in hypoxic WT lungs but significantly attenuated Ccl2 expression in the lungs of mice deficient in complement (Cfb and C5; Figure 3E).

tion in vasculature and airways (Figure 3C), whereas MCP1 protein was detected only in perivascular and adventitial areas (Figure 3D). RT-PCR analysis showed potent upregulation of Ccl2 in hypoxic WT lungs but significantly attenuated Ccl2 expression in the lungs of mice deficient in complement (Cfb and C5; Figure 3E). Figure 3. Hypoxia-induced Csf2 and Ccl2 mRNA expression is markedly augmented in the lungs of wild-type (WT) mice but is abrogated (for Csf2) or partially decreased (for Ccl2) in the lungs of complement (Cfb [complement factor B] and C5 [component 5])-deficient mice. (A) RNAscope in situ hybridization demonstrates that Csf2 expression is markedly augmented in the pulmonary arteries (PAs) of WT mice exposed to 3-day hypoxia (HX) but is abrogated in the PAs of complement (Cfb and C5)-deficient mice. Airways (AWs) of all mouse strains, both sea level (SL) and HX, maintain Csf2 expression without any visually significant change. (B) qRT-PCR analysis of whole murine lung tissues demonstrates that hypoxia-induced upregulation of Csf2 mRNA expression, detected in WT mice, is abrogated in complement (Cfb and C5)-deficient mice. One-way ANOVA with a Sidak multiple-comparison test with single pooled variance was performed for multiple group comparison. (C–E) HX-induced Ccl2 expression is complement (alternative and C5 pathways) dependent: levels of Ccl2 mRNA expression (C and E) and CCL2 protein expression (D) are markedly augmented in the lungs of 3-day HX WT mice; in contrast, the lungs of complement (Cfb and C5)-deficient HX mice demonstrate significantly decreased mRNA expression levels (E). One-way ANOVA with a Sidak multiple-comparison test with a single pooled variance was performed for multiple-group comparison. Scale bars, 100 μm. *P ≤ 0.05 and ****P ≤ 0.0001; ns = not significant. CT = cycle threshold.

ment (Cfb and C5)-deficient HX mice demonstrate significantly decreased mRNA expression levels (E). One-way ANOVA with a Sidak multiple-comparison test with a single pooled variance was performed for multiple-group comparison. Scale bars, 100 μm. *P ≤ 0.05 and ****P ≤ 0.0001; ns = not significant. CT = cycle threshold. Hypoxia-induced Perivascular Cell Proliferation Is Complement Dependent Augmented lung-cell proliferation is a distinctive characteristic of the early response to hypoxic insult (7), confirmed by increased numbers of replicating Ki67+ perivascular cells in hypoxic WT lungs (Figures 4A and 4B) and markedly augmented mRNA expression of Cdk1 (cyclin-dependent kinase 1) directly involved in cell-cycle progression (35) (Figure 4C). In contrast, lungs of hypoxic Cfb−/− and C5−/− mice displayed significantly fewer Ki67+cells and attenuated Cdk1 expression (Figures 4A–4C), implicating the complement signaling in partially regulating hypoxia-induced lung-cell proliferation.

ndent kinase 1) directly involved in cell-cycle progression (35) (Figure 4C). In contrast, lungs of hypoxic Cfb−/− and C5−/− mice displayed significantly fewer Ki67+cells and attenuated Cdk1 expression (Figures 4A–4C), implicating the complement signaling in partially regulating hypoxia-induced lung-cell proliferation. Figure 4. Hypoxia (HX)-induced perivascular cell proliferation is complement (alternative and C5 [component 5] pathways) dependent. (A) Profound increases in perivascular cell proliferative responses are observed in the lungs of wild-type (WT) mice exposed to 3-day HX, as compared with sea-level (SL) controls, but are significantly attenuated in Cfb (complement factor B)- and C5-deficient HX mice, as evaluated by immunostaining for nuclear proliferation-associated Ki67 antigen (red). Scale bars, 100 μm. (B) Quantification of red fluorescence was performed as described in the online supplement and is presented in arbitrary units (AU). (C) qRT-PCR analysis of whole lung tissues for cell cycle–progression marker Cdk1 (cyclin-dependent kinase 1) expression. One-way ANOVA with a Sidak multiple-comparison test with single pooled variance was performed for multiple-group comparison. *P < 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001. AW = airway; PA = pulmonary artery. Notably, RVSPs of 3-day hypoxic WT and complement-deficient mice were elevated likely because of vasoconstriction at this early time point in the disease process (Figures E5A–E5C).

Figure 4. Hypoxia (HX)-induced perivascular cell proliferation is complement (alternative and C5 [component 5] pathways) dependent. (A) Profound increases in perivascular cell proliferative responses are observed in the lungs of wild-type (WT) mice exposed to 3-day HX, as compared with sea-level (SL) controls, but are significantly attenuated in Cfb (complement factor B)- and C5-deficient HX mice, as evaluated by immunostaining for nuclear proliferation-associated Ki67 antigen (red). Scale bars, 100 μm. (B) Quantification of red fluorescence was performed as described in the online supplement and is presented in arbitrary units (AU). (C) qRT-PCR analysis of whole lung tissues for cell cycle–progression marker Cdk1 (cyclin-dependent kinase 1) expression. One-way ANOVA with a Sidak multiple-comparison test with single pooled variance was performed for multiple-group comparison. *P < 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001. AW = airway; PA = pulmonary artery. Notably, RVSPs of 3-day hypoxic WT and complement-deficient mice were elevated likely because of vasoconstriction at this early time point in the disease process (Figures E5A–E5C). Csf2/GM-CSF Expression and Production by Cultured Pulmonary Fibroblasts Is Complement Dependent Because we have previously shown that pulmonary perivascular fibroblasts from patients with PAH (termed PH-Fibs; Table E2) exhibit a CSF2/GM-CSF–expressing phenotype (5), we used these cells to evaluate whether the complement signaling directly regulates CSF2/GM-CSF expression and production. Exposure of serum-starved PH-Fibs to complement-sufficient human serum for 2 to 6 hours resulted in profound CSF2 mRNA increases under normoxia (21% O2) and, at 2 hours, under hypoxic conditions, whereas incubation in serum with a selectively inhibited alternative pathway (depleted in CFB) consistently resulted in significant attenuation of CSF2 mRNA (Figures 5Aa and 5Ab) and protein expression (Figure 5Ac). Similar findings were obtained with serum depleted of another essential activator of the alternative pathway, CFD (complement factor D; Figure 5B). On the contrary, CSF2 expression in PH-Fibs was complement C4 (classical pathway) independent (Figure 5C). Furthermore, PH-Fibs exposed to serum with inhibited terminal C5 pathway (C5-depleted sera) or with inhibited assembly of membrane-attack complex (MAC; C6-depleted sera) also demonstrated significantly decreased CSF2 expression under hypoxia (3%O2, 2 h) but no significant attenuation under normoxia (Figure 5D), suggesting a potential role for C5 and sublytical MAC assembly specifically in hypoxia-induced increases in CSF2 expression.

y of membrane-attack complex (MAC; C6-depleted sera) also demonstrated significantly decreased CSF2 expression under hypoxia (3%O2, 2 h) but no significant attenuation under normoxia (Figure 5D), suggesting a potential role for C5 and sublytical MAC assembly specifically in hypoxia-induced increases in CSF2 expression. Figure 5. Human pulmonary adventitial fibroblasts regulate Csf2/GM-CSF expression in a complement (alternative and C5 [component 5] pathways)-dependent manner. Fibroblasts, isolated from pulmonary arteries of patients with idiopathic pulmonary arterial hypertension (PH-Fibs; n = 5; Table E2), were serum-starved in serum-free medium (SFM) for 72 hours and incubated with complement-sufficient normal human serum (HS) or HS depleted in specific complement components. (A and B) PH-Fibs substantially upregulate CSF2 mRNA (Aa, Ab, and B) and protein (Ac) expression in response to complement-sufficient HS, whereas their responses to serum with inhibited alternative pathway (CFBdpl [complement factor B–depleted], Aa and Ab; CFDdpl [complement factor D–depleted], B) are significantly attenuated. This pattern was observed under both normoxia (NX; 21% O2) and hypoxia (HX; 3% O2) conditions; however, at a 2-hour time point (three different cell populations shown, Aa), the levels of HX-induced CSF2 mRNA expression were higher than those in NX, which was in contrast to a 6-hour time point (n = 5 cell populations shown, Ab). The qRT-PCR findings were confirmed at a protein level via ELISA (Ac) using conditioned medium generated at a 6-hour time point. (C) Expression of CSF2 by PH-Fibs was observed to be independent of complement C4 (main component of the classical and lectin pathways). (D) Two-hour exposure of PH-Fibs to HS depleted in complement C5 (C5dpl) or C6 (C6dpl) resulted in substantial attenuation of CSF2 expression compared with HS under HX (3% O2) but not under NX (21% O2) conditions. *P < 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001; ns = not significant. C4dpl = component 4–depleted; CT = cycle threshold.

PH-Fibs to HS depleted in complement C5 (C5dpl) or C6 (C6dpl) resulted in substantial attenuation of CSF2 expression compared with HS under HX (3% O2) but not under NX (21% O2) conditions. *P < 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001; ns = not significant. C4dpl = component 4–depleted; CT = cycle threshold. Hypoxia-induced IgG Deposition Contributes to Complement Activation and Perivascular Proinflammatory, Pro-proliferative Processes Having determined an essential role of complement signaling in hypoxia-induced perivascular inflammation and cell proliferation, we next sought to define its upstream triggers. Because previous studies have suggested an important role for (auto)immune mechanisms in the pathophysiology of PH (2, 9), we evaluated the contribution of immunoglobulins to hypoxia-induced complement activation and proinflammatory lung remodeling. Pulmonary vasculature of 3-day hypoxic mice and rats demonstrated profound deposition of both IgM (Figures 6A and 6B) and IgG (Figures 6C and 6D) but each in a specific compartmentalized pattern: IgM deposition was restricted to luminal/medial areas (Figures 6A and 6B), whereas IgG deposition encompassed all vascular layers but was especially prominent perivascularly (Figures 6C and 6D). Furthermore, luminal/medial IgM deposition pattern clearly correlated with that of deposited C4 (complement component of classical and lectin pathways; Figure 6E), whereas perivascular plus luminal/medial IgG deposition pattern strongly correlated with that of complement C3 (component of the alternative pathway; Figures 6F and 6G). No IgM, IgG, C4, or C3 deposition was detected in SL animals. Interestingly, although the final degradation fragments of complement components C4 and C3 (C4d and C3d, respectively) were not yet detectable at the early, 3-day hypoxic time point, they were readily observed at 3 weeks of sustained hypoxic exposure in a compartmentalized pattern that clearly correlated with the pattern of their respective native C4 and C3 components: luminal/medial for C4d and perivascular for C3d (Figure E6).

ively) were not yet detectable at the early, 3-day hypoxic time point, they were readily observed at 3 weeks of sustained hypoxic exposure in a compartmentalized pattern that clearly correlated with the pattern of their respective native C4 and C3 components: luminal/medial for C4d and perivascular for C3d (Figure E6). Figure 6. Hypoxia (HX)-induced vascular deposition patterns of IgM and IgG are highly compartmentalized and correlate with deposition patterns of complement components C4 and C3, respectively. (A–D) Highly compartmentalized patterns of IgM versus IgG deposition and complement components C4 versus C3 are observed in experimental rodents (mice and rats) exposed to 3-day HX: IgM deposition (red) in mice (A) and rats (B) is observed in the luminal and medial areas, whereas deposition of IgG (red) in mice (C) and rats (D) is detected mainly in perivascular/adventitial areas and additionally encompasses luminal/medial areas. (E–G) In 3-day HX mice, luminal/medial deposition of complement C4 (red, E) correlates with that of IgM (red, A and B), whereas strong perivascular (in addition to luminal/medial) deposition of complement C3 (green, F and G) clearly correlates with that of IgG (red, C, D, and G). No deposition of IgM, IgG, C4, or C3 was detected in the lungs of sea-level (SL) rodents. Autofluorescence of elastic lamellae (green in A, B, and E–G) or α-smooth muscle actin (green in C and D) defines tunica media. Cell nuclei are labeled with DAPI (blue). Scale bars, 100 μm. AW = airway; PA = pulmonary artery.

G). No deposition of IgM, IgG, C4, or C3 was detected in the lungs of sea-level (SL) rodents. Autofluorescence of elastic lamellae (green in A, B, and E–G) or α-smooth muscle actin (green in C and D) defines tunica media. Cell nuclei are labeled with DAPI (blue). Scale bars, 100 μm. AW = airway; PA = pulmonary artery. Having observed consistent deposition of immunoglobulins in hypoxic rodents, we proceeded with evaluating hypoxic responses in mice lacking all circulating immunoglobulins (μMT− mice) (26). Although lungs of 3-day hypoxic WT mice, juxtaposed to SL-WT counterparts, demonstrated robust perivascular deposition of complement C3, profound inflammation (augmented accumulation of CD68+ macrophages, Cd68 and Csf2, Ccl2 mRNA expression), and increased cell proliferation (Ki67+ cells, Cdk1 expression; Figures 7A–7I), remarkably, μMT− mice were protected from hypoxia-induced vascular changes (no detectable complement C3 deposition, reduced perivascular accumulation of CD68+macrophages, near-complete abrogation of Csf2, significant attenuation of Ccl2 expression, and decreases in cell proliferation; Figures 7A–7I). Interestingly, although RVSPs of 3-day hypoxic WT mice, compared with SL-WT, were augmented likely because of vasoconstriction, RVSPs of hypoxic μMT− mice were almost unchanged compared with SL (Figure E7).

ogation of Csf2, significant attenuation of Ccl2 expression, and decreases in cell proliferation; Figures 7A–7I). Interestingly, although RVSPs of 3-day hypoxic WT mice, compared with SL-WT, were augmented likely because of vasoconstriction, RVSPs of hypoxic μMT− mice were almost unchanged compared with SL (Figure E7). Figure 7. μMT mice deficient in all circulating immunoglobulins exhibit a “hypoxia-protective” phenotype. (A) Exposure to 3-day hypoxia (HX) fails to activate complement C3 (component 3) cascade in the lungs of μMT mice. Although the lungs of 3-day HX wild-type (WT) mice demonstrate activation of the complement cascade, as defined by strong perivascular deposition of complement C3 (green, lower left), no C3 deposition is observed in the lungs of 3-day HX μMT mice (lower right). Control sea-level (SL) mice of both strains do not display any C3 deposition. Cell nuclei are labeled with DAPI (blue). (B–I) Hypoxia-induced proinflammatory and proliferative responses are attenuated in the lungs of μMT mice. Accumulation of CD68+ macrophages (B–D) and Csf2 and Ccl2 cytokine/chemokine expression (E and F) are strongly (for Cd68 and Csf2) or moderately (for Ccl2) attenuated in the lungs of 3-day HX μMT− mice compared with HX WT counterparts. (B) Immunofluorescent staining for CD68 macrophage marker (red fluorescence) and α-smooth muscle actin (green fluorescence). DAPI staining (blue) defines cell nuclei. (C) Quantification of red fluorescence was performed as described in the online supplement and is presented in arbitrary units (AU). (D–F) qRT-PCR analysis of whole lung tissues for mRNA expression of macrophage marker Cd68 (D), Csf2 (E), and Ccl2 (F). One-way ANOVA with a Sidak multiple-comparison test with single pooled variance was performed for multiple-group comparison. (G–I) Hypoxia-induced perivascular cell proliferative responses, defined by immunostaining for nuclear proliferation-associated Ki67 antigen (red, G, quantified in H) and qRT-PCR analysis of whole lung tissues for cell cycle–progression marker Cdk1 (cyclin-dependent kinase 1) expression (I) are significantly attenuated in the lungs of HX μMT mice compared with HX WT counterparts. (G) Autofluorescence of pulmonary artery elastic lamellae (green) defines tunica media; cell nuclei are labeled with DAPI (blue). *P ≤ 0.05 and ****P ≤ 0.0001; ns = not significant. Scale bars, 100 μm. AW = airway; CT = cycle threshold; PA = pulmonary artery.

n the lungs of HX μMT mice compared with HX WT counterparts. (G) Autofluorescence of pulmonary artery elastic lamellae (green) defines tunica media; cell nuclei are labeled with DAPI (blue). *P ≤ 0.05 and ****P ≤ 0.0001; ns = not significant. Scale bars, 100 μm. AW = airway; CT = cycle threshold; PA = pulmonary artery. Because of consistently observed perivascular colocalization of IgG, C3, and CD68+, C5aR1+, and C3aR+ cells, we sought to determine whether hypoxic μMT− mice reconstituted with IgG to its normal circulating levels (36) using an injection protocol similar to that in human immunodeficient patients (37) (online supplement) would recapitulate the WT hypoxic phenotype. Strikingly, most features of the pathologic hypoxia-induced WT phenotype were restored (perivascular deposition of IgG and C3, augmented accumulation of CD68+, C5aR1+macrophages, increased expression of Csf2, Ccl2, and Cdk1; Figures 8A–8I). These data indicated an essential contribution of IgG to proinflammatory changes in the lung vasculature.

of the pathologic hypoxia-induced WT phenotype were restored (perivascular deposition of IgG and C3, augmented accumulation of CD68+, C5aR1+macrophages, increased expression of Csf2, Ccl2, and Cdk1; Figures 8A–8I). These data indicated an essential contribution of IgG to proinflammatory changes in the lung vasculature. Figure 8. Reconstitution of circulating IgG in hypoxic μMT− mice restores a proinflammatory phenotype. Five consecutive daily injections of normoxic μMT− mice with normal mouse IgG (2 mg/mouse) (36), equivalent to the “loading/initiation” phase of IgG reconstitution in immune-deficient human individuals (37), was followed by 3 “blank” days and subsequently by exposure to 3-day hypobaric hypoxia (HX). (A) Robust deposition of IgG (red) is observed mainly in pulmonary arteries (PAs) in a perivascular-specific manner but also encompasses vascular luminal and medial compartments. Some periairway deposition is also detected. (B) Deposition of complement C3 (component 3; green) is prominently observed in PA perivascular areas, as well as in vascular medial and luminal compartments and partially in periairway areas. (C and D) Robust accumulation of CD68+and C5aR1-expressing macrophages (red) is observed in perivascular areas. Cell nuclei are labeled with DAPI (blue). Scale bars, 100 μm. (E) Quantification of increased accumulation of CD68+and C5aR1+cells (red fluorescence) was performed as described in the online supplement and is presented in arbitrary units (AU). Quantification of IgG and C3 deposition was not performed because the sea-level lung samples were completely negative for the staining. (F–I) qRT-PCR analysis demonstrates augmented expression of Cd68 (macrophage marker), Csf2 and Ccl2 (cytokine and chemokine, respectively), and Cdk1 (cyclin-dependent kinase 1; cell cycle–progression marker) in the whole lungs of HX immunoglobulin-deficient μMT mice that were reconstituted with circulating IgG (HX + IgG) compared with PBS-injected HX μMT mice.*P < 0.05, ***P ≤ 0.001, and ****P ≤ 0.0001. AW = airway; C5aR1 = receptor for anaphylatoxin C5a; CT = cycle threshold.

ndent kinase 1; cell cycle–progression marker) in the whole lungs of HX immunoglobulin-deficient μMT mice that were reconstituted with circulating IgG (HX + IgG) compared with PBS-injected HX μMT mice.*P < 0.05, ***P ≤ 0.001, and ****P ≤ 0.0001. AW = airway; C5aR1 = receptor for anaphylatoxin C5a; CT = cycle threshold. Complement Cascade Is Activated in a Perivascular-Specific Manner in Various Experimental PH Models and Human PAH To test whether complement activation is sustained at later time points of hypoxic exposure and in other experimental PH models, as well as in human PAH, we evaluated deposition of C3d (the final activation/degradation fragment of complement C3), which was not detectable at an early, 3-day hypoxic exposure time point but constitutes a commonly accepted marker of complement cascade activation (38). C3d deposition was consistently observed in a perivascular-specific pattern in the lungs of experimental mice, in rats and calves with chronic (2–3 wk) hypoxic PH, and in rats with sugen-hypoxia PH and monocrotaline PH (Figures 9A and E8A). In lungs of human patients with IPAH, C3d was detected immunohistochemically in pulmonary perivascular areas, with evidence of less prominent expression within the neointima of arteries with intima thickening (Figures 9Ba and E8B). The IHC signal of C3d deposition in normal (rejected) lung donors was minimal to absent.

In lungs of human patients with IPAH, C3d was detected immunohistochemically in pulmonary perivascular areas, with evidence of less prominent expression within the neointima of arteries with intima thickening (Figures 9Ba and E8B). The IHC signal of C3d deposition in normal (rejected) lung donors was minimal to absent. Figure 9. The complement cascade is activated in the lungs of experimental animal models of pulmonary hypertension (PH) and patients with idiopathic pulmonary arterial hypertension (IPAH). (A and B) Activation of the complement cascade, as defined by deposition of C3d fragment (terminal activation product of complement component 3 [C3]) is most prominently observed in a perivascular-specific manner in the lungs of experimental animal models of hypoxic PH (mouse, rat, and calf), sugen-hypoxia (SU-HX) PH and monocrotaline‐PH (MCT‐PH) rat models, and humans with IPAH. OCT-embedded lung cryosections of experimental animal specimens (A) and formalin-fixed paraffin-embedded sections of human lung specimens (B) were labeled with a biotinylated C3d-specific monoclonal antibody (mAb; red in A, brown in B), developed by Thurman and colleagues (38). This mAb distinguishes tissue-bound C3d from the intact C3 or C3b, allowing assessment of tissue-specific activation of the complement cascade. (A) Exposure of experimental animals to sustained hypoxia was performed for 3 weeks for mice and rats (n = 5, each group) and 2 weeks for calves (n = 5, each group); control (CO) animals were maintained at sea-level (SL) or ambient (Denver, Colorado) altitude. An image of the SU-HX rat model is shown at 2 weeks of hypoxic exposure (the most proinflammatory time point), and an image of the MCT-PH rat lung sample is shown at a 2-week time point. Autofluorescence of pulmonary artery (PA) elastic lamellae (green) defines tunica media in hypoxic experimental animal models and in human specimens, whereas α-smooth muscle actin defines tunica media in MCT-PH and SU-HX lung specimens. Images of C3d immunofluorescent staining for all analyzed human specimens are shown in Figure E8A. IPAH and normal (rejected for lung transplant) donor cohorts are described in Table E3A. Cell nuclei are labeled with DAPI (blue). Scale bars, 100 μm. (Ba) Immunohistochemistry (IHC) staining demonstrates C3d deposition in a medium-sized PA of a patient with IPAH. The C3d IHC signal is largely restricted to perivascular areas.

normal (rejected for lung transplant) donor cohorts are described in Table E3A. Cell nuclei are labeled with DAPI (blue). Scale bars, 100 μm. (Ba) Immunohistochemistry (IHC) staining demonstrates C3d deposition in a medium-sized PA of a patient with IPAH. The C3d IHC signal is largely restricted to perivascular areas. Cohorts of patients with IPAH and CO (rejected for lung transplant) donors (n = 6, each) are described in Table E3. Images of C3d IHC staining of all analyzed human specimens (n = 6, each cohort) are shown in Figure E8B. (Bb) Quantification of IHC staining was performed via MetaMorph software, in which the dynamic range for gray intensity levels ranges between 0 (white) and 256 (black). Background levels of gray intensity are thus determined largely by the counterstain for the IHC (methyl green) and potentially represent baseline of nonspecific IHC signal to the detection system. The quantification of the IHC signal (presented in arbitrary units [AU]) for C3d revealed a 2.382-fold increase in IPAH lungs over CO donor lungs. (C–E) Plasma complement is a critical determinant of clinical outcomes in patients with PAH. (C) Previously published data (12) identifying circulating proteins (Table E4) with prognostic importance to patients with PAH (n = 218) were analyzed using a network medicine approach. Differentially expressed proteins were mapped to the consolidated human interactome resulting in a network that was enriched with complement pathway intermediaries, referred to in this article as the complement–PAH network (13 proteins and 18 protein–protein interactions). (D) Two distinct PAH patient clusters were identified based on biological information derived solely from the complement–PAH network. Oval represents the estimated cluster boundaries determined by the patient data in each cluster. (E) Previously published data (12) identifying circulating proteins (Table E4) with prognostic importance to patients with PAH (n = 218) were analyzed using a network medicine approach. A total of 37 differentially expressed proteins were mapped to the consolidated human interactome resulting in the complement–PAH network. Kaplan–Meier survival estimates in patients with PAH divided into two clusters (as shown in D) based on the plasma levels of proteins in the complement–PAH network are presented; thin vertical marks indicate where patients were censored during the time course. ***P ≤ 0.001. AW = airway; OCT = optimal cutting temperature compound; PC = principal component.

tients with PAH divided into two clusters (as shown in D) based on the plasma levels of proteins in the complement–PAH network are presented; thin vertical marks indicate where patients were censored during the time course. ***P ≤ 0.001. AW = airway; OCT = optimal cutting temperature compound; PC = principal component. These data decisively demonstrated that complement activation is a longitudinal marker of PH and PAH in different experimental PH models, across species, and at different stages of the disease process. Furthermore, complement activation is initiated and maintained in the pulmonary perivascular areas.

tients with PAH divided into two clusters (as shown in D) based on the plasma levels of proteins in the complement–PAH network are presented; thin vertical marks indicate where patients were censored during the time course. ***P ≤ 0.001. AW = airway; OCT = optimal cutting temperature compound; PC = principal component. These data decisively demonstrated that complement activation is a longitudinal marker of PH and PAH in different experimental PH models, across species, and at different stages of the disease process. Furthermore, complement activation is initiated and maintained in the pulmonary perivascular areas. Plasma Complement Is a Critical Determinant of Clinical Outcomes in Patients with PAH Next we tested whether complement signaling can also serve as a biomarker in the circulation of patients with PAH. Using unbiased analytic methods, we first aimed to determine which complement signaling proteins are clinically relevant to PAH. A biomarker panel of prognostic proteins in PAH (12) (Table E4) was mapped to the consolidated human interactome (30), which includes information on functional associations between proteins that is based on protein–protein interactions. From this approach, we observed that 13 PAH proteins were connected to at least one other PAH protein, corresponding to 18 protein–protein interactions. This network was heavily populated by complement intermediaries, particularly those of the alternative pathway (C3, CFB, CFD, CFH, and CFP) and C7 of the terminal MAC pathway. As such, this was termed the “complement–PAH network” (Figure 9C). Next we hypothesized that focusing on functionally related proteins could be informative for determining differences in outcome among patients with PAH. Therefore, a K-means clustering algorithm was performed on the original cohort of patients with PAH (12) using the expression profile of only proteins in the complement–PAH network (39). The elbow method was used to determine the optimal number of patient clusters (n = 2). The summarized expression profile of proteins in the complement–PAH network alone was sufficient to identify two distinct subgroups of patients with PAH (Figure 9D) that corresponded to significant and meaningful differences in the rate of all-cause mortality (Figure 9E).

the optimal number of patient clusters (n = 2). The summarized expression profile of proteins in the complement–PAH network alone was sufficient to identify two distinct subgroups of patients with PAH (Figure 9D) that corresponded to significant and meaningful differences in the rate of all-cause mortality (Figure 9E). Discussion This study establishes 1) essential roles for the alternative and C5 complement pathways in initiating lung inflammation through generation of Csf2/GM-CSF and Ccl2/MCP1, perivascular monocyte and macrophage accumulation, and augmentation of cell proliferation; 2) the critical contribution of immunoglobulins, specifically IgG, to complement-driven lung inflammation; 3) perpetuation of perivascular-specific activation of the complement cascade in the lungs of various experimental animal PH models and humans with PAH; and 4) complement signaling (the alternative pathway, in particular) as an important correlative determinant of clinical outcomes in patients with PAH.

g inflammation; 3) perpetuation of perivascular-specific activation of the complement cascade in the lungs of various experimental animal PH models and humans with PAH; and 4) complement signaling (the alternative pathway, in particular) as an important correlative determinant of clinical outcomes in patients with PAH. The complement system plays an essential role in defensive immune processes. However, dysregulated complement activation may turn this beneficial protective system into a destructive tissue-damaging villain, as shown in various inflammatory, autoimmune, and ischemic disorders (10). Mounting experimental data unveil the relative involvement of each of the three activation pathways (classical, lectin, and alternative) in complement-mediated tissue injury and remodeling. In the present study, we identified the activated alternative pathway as an essential regulator of hypoxia-induced proinflammatory and pro-proliferative changes in the lung. Similar findings of alternative pathway activation were recently reported by high-throughput analysis of the plasma proteome in patients with PAH, in which increases in the activator of the alternative pathway, CFD, and decreases in the inhibitor CFH identified patients with PAH with high risk of mortality (12).

lung. Similar findings of alternative pathway activation were recently reported by high-throughput analysis of the plasma proteome in patients with PAH, in which increases in the activator of the alternative pathway, CFD, and decreases in the inhibitor CFH identified patients with PAH with high risk of mortality (12). An important and novel finding of this study demonstrates, both in vivo and in vitro, complement (alternative and C5 pathways)-dependent vascular-specific regulation of Csf2/GM-CSF expression and production. GM-CSF has emerged as a potent key mediator of tissue inflammation (40). Locally produced at sites of inflammation, GM-CSF promotes recruitment, prosurvival reprograming, and proinflammatory activation of monocytes and macrophages and participates in induction of inflammasome (34, 41). Interestingly, among various myeloid cell types, only the inflammatory Ly6Chi/CCR2+ monocytes are vitally dependent on GM-CSF to instigate tissue inflammation and damage (41). A recent study reported increased numbers of inflammatory Ly6Chi/CCR2+ monocytes in blood and lungs of hypoxic mice (42). Consequently, it is plausible that the hypoxia-induced Ly6Chi/CCR2+ monocyte expansion is dependent on GM-CSF signaling. An elegant recent study by the Rabinovitch group demonstrated that GM-CSF is a critical mediator of monocyte and macrophage recruitment and PH and PAH development (34). Our findings corroborate this exciting report by showing that attenuated Csf2 expression corresponds with reduced monocyte and macrophage recruitment; however, the novelty of our findings is in delineating the molecular mechanisms underlying Csf2/GM-CSF regulation (i.e., complement dependence) and demonstrating that GM-CSF is an essential downstream mediator of complement-induced injury in PH. Our in vivo data also demonstrate complement (alternative and C5 pathways)-dependent regulation of hypoxia-induced Ccl2/MCP1 expression and cell proliferation. Several reports have suggested Csf2/GM-CSF–dependent mechanisms of Ccl2/MCP1 induction (43, 44) and cell proliferation (45). It is also plausible that complement-dependent cell-proliferative responses may be a result of terminal C5 pathway activation, which generates both pro-proliferative anaphylatoxin C5a (10) and MAC, of which the latter has been shown to induce smooth muscle cell replication at sublytical concentrations (46).

ation (45). It is also plausible that complement-dependent cell-proliferative responses may be a result of terminal C5 pathway activation, which generates both pro-proliferative anaphylatoxin C5a (10) and MAC, of which the latter has been shown to induce smooth muscle cell replication at sublytical concentrations (46). An intriguing finding of the present study is in establishing the critical contribution of immunoglobulins, specifically the IgG class, to complement activation and downstream proinflammatory changes in experimental PH and human PAH. PH and PAH have long been considered to exhibit an immune-mediated component, as demonstrated by the detection of circulating autoantibodies and activated bronchus-associated lymphoid tissue (2, 47). However, the mechanisms by which the observed autoimmune dysregulation is initiated in PH, and specifically in hypoxic forms of sterile inflammation, are unknown. Conventionally, the classical pathway is held responsible for inducing antibody-mediated complement activation, in which pentameric IgM is more effective in activating complement than monomeric IgG. However, certain IgG-immune complexes have been shown to activate the lectin pathway (20) and suggested to drive the alternative pathway (48). In vivo, the alternative pathway exhibits spontaneous low-grade hydrolysis of complement C3 (“tick-over” mechanism) to generate C3a and C3b fragments. C3b is short-lived but can covalently bind to certain IgG molecules immobilized on the cell surface, with the docking site within the IgG CH1 domain, and can form C3b2–IgG complexes that are more stable than the C3b itself (49). Subsequent bivalent binding of properdin results in up to 11-fold potentiation of C3 cleavage, rendering significant enhancement of alternative pathway amplification in the presence of adherent C3b–IgG complexes (48). Furthermore, tissue deposition of IgG may render this surface advantageous for alternative pathway propagation by diminishing binding of the inhibitory CFH (48). Intriguingly, the alternative pathway alone can promote in vivo inflammation and tissue damage and, in vitro, is capable of activating complement C3 (21). However, when all three pathways (classical, lectin, and alternative) are intact, the alternative pathway drives only the subsequent amplification and not the initial activation step (21).

native pathway alone can promote in vivo inflammation and tissue damage and, in vitro, is capable of activating complement C3 (21). However, when all three pathways (classical, lectin, and alternative) are intact, the alternative pathway drives only the subsequent amplification and not the initial activation step (21). We suggest that, at least in some immune complex–mediated PH and PAH cases, the alternative pathway serves as a critical facilitator of inflammatory dysregulation in its role as a potent amplification loop following initiation by the classical or lectin pathway (as shown in our study by luminal/medial deposition of complement component C4 and supported by studies in other immune complex–mediated diseases) (20, 21, 23). However, delineation of specific activation pathways initiated by hypoxia upstream of the alternative amplification loop requires further investigation.

athway (as shown in our study by luminal/medial deposition of complement component C4 and supported by studies in other immune complex–mediated diseases) (20, 21, 23). However, delineation of specific activation pathways initiated by hypoxia upstream of the alternative amplification loop requires further investigation. Our data, which show a “hypoxia-protected” phenotype of immunoglobulin-deficient μMT− mice and demonstrate that IgG-reconstituted μMT mice partially restored the hypoxia-induced pathological phenotype, suggest immunoglobulins (IgG in particular) as important facilitators of the proinflammatory hypoxic phenotype. Because μMT− mice were injected with normal IgG, we propose that naturally occurring, so-called “natural antibodies” (N-Abs) bind to hypoxic injury–generated neoepitopes on self-cells. N-Abs are normally present in circulation of healthy persons in the absence of exogenous antigen stimulation (“preexisting” antibodies) and exert first-response protective and regulatory functions (50). Binding of N-Abs to damaged self-cells and further activation of complement may propagate a downstream cascade of adaptive immune responses leading to generation of specific autoantibodies (51). The neoepitopes recognized by N-Abs in response to hypoxic insult are the subject of ongoing investigation by our group. In cases of more advanced PH or PAH (Figure E9) or more injurious stimuli (monocrotaline PH or elastase-induced aortic aneurism models), autoimmune reactions were suggested to be initiated or propagated by antigen-specific IgG rather than N-Abs, and specific autoantibodies have been identified (2, 20, 47). However, we speculate that in the early setting of 3-day experimental hypoxia, preexisting N-Abs are likely at play.

induced aortic aneurism models), autoimmune reactions were suggested to be initiated or propagated by antigen-specific IgG rather than N-Abs, and specific autoantibodies have been identified (2, 20, 47). However, we speculate that in the early setting of 3-day experimental hypoxia, preexisting N-Abs are likely at play. The preponderance of data, with the exception of a sole report (33), associate complement with PAH via analysis of circulating complement components (13, 52) without studying the lung directly. This knowledge gap casts uncertainty on the translational importance of complement to the pathogenesis of PAH. In the present study, we attempted to resolve this uncertainty by analyzing both the local lung-specific processes and the correlation of dysregulated complement to clinical outcome in patients with PAH. With regard to the latter, network medicine was used to explore integrated biological pathways that are important in PAH clinically. Our findings suggest that clinical outcomes in patients with PAH could be determined by the complement–PAH network, generating unbiased, validated, and clinically important results. This provides further context toward clarifying the molecular mechanisms underpinning the role of complement in PH and PAH. The relevance of experimental animal data to the human condition was validated by our findings of similar perivascular-specific patterns of complement activation (C3d deposition) in the lungs of PH animals and patients with IPAH. Importantly, a perivascular-specific pattern of complement activation, localization of C3ar1/C5ar1-expressing cells, and IgG deposition provides further support for the essential role of the adventitia in PH-associated inflammatory processes (7, 8).

on (C3d deposition) in the lungs of PH animals and patients with IPAH. Importantly, a perivascular-specific pattern of complement activation, localization of C3ar1/C5ar1-expressing cells, and IgG deposition provides further support for the essential role of the adventitia in PH-associated inflammatory processes (7, 8). In conclusion, the data of this study establish a causal role for complement activation and immunoglobulins in the early, initiating, proinflammatory stage of experimental hypoxic PH and propose the alternative pathway of complement signaling as a prognostic factor in clinical PAH. Dysregulated complement signaling thus emerges as a persistent longitudinal determinant of PH and PAH pathobiology. Because the etiology of PH and PAH remains elusive, further delineation of diagnostic and predictor targets in the initial asymptomatic preclinical period would facilitate development of more effective prevention and treatment strategies. Acknowledgment The authors would like to express gratitude to Andrzej M. Poczobutt, Marcia McGowan, and Aya Laux for the invaluable technical help and assistance with manuscript preparation, to Dr. Suzette R. Riddle for discussing experimental design and technical issues, to Amanda R. Flockton for help with cultured cells, to Dr. R. M. Torres for valuable discussions and suggestions, and to Dr. Lan Zhao (Imperial College London) for providing MCT-PH rat lung sections.

ssistance with manuscript preparation, to Dr. Suzette R. Riddle for discussing experimental design and technical issues, to Amanda R. Flockton for help with cultured cells, to Dr. R. M. Torres for valuable discussions and suggestions, and to Dr. Lan Zhao (Imperial College London) for providing MCT-PH rat lung sections. Supported by NIH grants P01 HL014985-44 (K.R.S.); R01DK076690 and R01DK113586 (J.M.T.); R56HL131787, R01HL139613-01, R21HL145420, National Scleroderma Foundation, and Cardiovascular Medical Research Education Foundation (B.A.M.); U01 125215 (J.A.L.); R01DK076690 and R01DK113586 (J.M.T); P01 HL014985-44 and R24HL123767 (R.M.T.); and by the British Heart Foundation RE/18/4/34215 (M.R.W.).

DK113586 (J.M.T.); R56HL131787, R01HL139613-01, R21HL145420, National Scleroderma Foundation, and Cardiovascular Medical Research Education Foundation (B.A.M.); U01 125215 (J.A.L.); R01DK076690 and R01DK113586 (J.M.T); P01 HL014985-44 and R24HL123767 (R.M.T.); and by the British Heart Foundation RE/18/4/34215 (M.R.W.). Author Contributions: M.G.F. conceived the project, designed and performed experiments, analyzed and interpreted data, made the figures, and wrote the manuscript. B. A. McKeon designed and performed experiments, provided statistical analysis and interpretation of the data, and made the figures. J.M.T. provided intellectual input, materials, and resources; interpreted data; and edited the manuscript. M.L., H.Z., S.K., S.H., and A.G. performed experiments and analyzed the data. B. A. Maron, M.R.W., and J.A.L. generated and analyzed the network data, provided intellectual input, and edited the manuscript. T.S. and M.A.F. performed animal experiments and analyzed the data. J.L. provided materials, resources, and methodology. C.J.R. generated and analyzed the network data and edited the manuscript. P.G. and R.-S.W. generated and analyzed the network data. R.M.T. provided intellectual input and materials and resources. V.M.H. provided intellectual input and edited the manuscript. K.R.S. provided intellectual oversight of the project, interpreted the data, and edited the manuscript. This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.

Author Contributions: M.G.F. conceived the project, designed and performed experiments, analyzed and interpreted data, made the figures, and wrote the manuscript. B. A. McKeon designed and performed experiments, provided statistical analysis and interpretation of the data, and made the figures. J.M.T. provided intellectual input, materials, and resources; interpreted data; and edited the manuscript. M.L., H.Z., S.K., S.H., and A.G. performed experiments and analyzed the data. B. A. Maron, M.R.W., and J.A.L. generated and analyzed the network data, provided intellectual input, and edited the manuscript. T.S. and M.A.F. performed animal experiments and analyzed the data. J.L. provided materials, resources, and methodology. C.J.R. generated and analyzed the network data and edited the manuscript. P.G. and R.-S.W. generated and analyzed the network data. R.M.T. provided intellectual input and materials and resources. V.M.H. provided intellectual input and edited the manuscript. K.R.S. provided intellectual oversight of the project, interpreted the data, and edited the manuscript. This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org. Originally Published in Press as DOI: 10.1164/rccm.201903-0591OC on September 23, 2019 Author disclosures are available with the text of this article at www.atsjournals.org.

In this issue of the Journal, Martinez-Zayas and colleagues (pp. 212–223) report and validate a novel prediction model (HOMER) to calculate the probability of patients with non-small cell lung cancer (NSCLC) having mediastinal lymph node involvement (1). Determining a patient’s likelihood of lymph node metastasis is paramount in determining the stage of lung cancer and therefore appropriate treatment options. Clinical staging, including imaging modalities and biopsy techniques, remains a challenge and frequently falls short of surgical staging, depending on how aggressive the preoperative evaluation is (2). Accurate staging has been associated with improved survival and remains a huge emphasis in the care of patients with lung cancer (3). The study by Martinez-Zayas and colleagues is the first to derive and validate a risk model aimed at discriminating between the most clinically useful forms of nodal disease in patients who were both surgical and nonsurgical candidates: N0, N1, and N2/3 disease.

ge emphasis in the care of patients with lung cancer (3). The study by Martinez-Zayas and colleagues is the first to derive and validate a risk model aimed at discriminating between the most clinically useful forms of nodal disease in patients who were both surgical and nonsurgical candidates: N0, N1, and N2/3 disease. The authors should be commended for the statistical rigor used to derive and validate their model. Covariates used to develop the model were pragmatic, clinically relevant, and appropriately limited by the last common outcome. By externally validating their prediction model at other medical centers, the authors offer a model with the possibility of geographic stability for patients with NSCLC without T4 tumors or distant metastasis, after adjusting for the local institution’s population. The authors further supported their model with temporal validation to show stability over time (4). HOMER therefore has the potential to be generalizable in both the short term and the long term for patients with NSCLC seeking treatment at well-practiced thoracic oncology centers that use systematic endobronchial ultrasound-guided transbronchial needle aspiration (EBUS TBNA) lymph node staging. To carry out the systematic EBUS lymph node staging that the output of HOMER applies to, an examination of the intrathoracic nodes is required by EBUS, beginning with contralateral N3 nodes, followed by N2 and then N1 lymph nodes. Any lymph node measuring ≥5 mm in short axis is sampled, aiming for a minimum of three N2/3 lymph node stations sampled per procedure (5).

ng that the output of HOMER applies to, an examination of the intrathoracic nodes is required by EBUS, beginning with contralateral N3 nodes, followed by N2 and then N1 lymph nodes. Any lymph node measuring ≥5 mm in short axis is sampled, aiming for a minimum of three N2/3 lymph node stations sampled per procedure (5). There are several clinically useful applications of HOMER. Assuming a patient is not a surgical candidate, the preferred treatment for N0 disease is definitive stereotactic ablative radiotherapy (SABR) and will not be confirmed surgically. Therefore, making an accurate clinical prediction of this disease state is crucial (6). Previous work was not able to discriminate between patients with N0 and N1 disease (7). HOMER presents two exciting ways to bring evidence-based decision making to these patients. First, the model can help predict the pretest probability that an EBUS TBNA will detect NSCLC, based on widely available clinical and radiographic data. As pointed out elegantly in the discussion, this can allow a more objective discussion about the risk and benefit of requiring an EBUS TBNA before SABR. As the low risk of complication is approached by the predicted probability of lymph node metastasis detected by EBUS TBNA, one can more confidently consider avoiding invasive mediastinal staging. This is especially relevant for patients at increased risk for complications during bronchoscopy. This may also be useful for patients with confirmed NSCLC from a transthoracic needle biopsy with radiographic N2/N3 disease who are at extremely high risk for bronchoscopy.

consider avoiding invasive mediastinal staging. This is especially relevant for patients at increased risk for complications during bronchoscopy. This may also be useful for patients with confirmed NSCLC from a transthoracic needle biopsy with radiographic N2/N3 disease who are at extremely high risk for bronchoscopy. The other way HOMER can be used for these patients is to calculate a posttest probability of N1 disease in a patient being considered for SABR who has a negative EBUS TBNA. Current guidelines appropriately lean toward cytologic or pathologic confirmation for mediastinal staging. They suggest preoperative invasive mediastinal staging in patients with NSCLC unless the tumor is T1 (<3 cm) and peripheral, and the mediastinal lymph nodes are radiographically negative by computed tomography and positron emission tomography (7). This recommendation is based on a low false-negative rate (i.e., lymph node metastasis) in this patient population, as determined by older descriptive studies (8). Importantly, there are more recent data to support occult lymph node metastasis and a limited sensitivity for EBUS TBNA in similar patient populations (9). After making an assumption about the sensitivity of EBUS TBNA, a clinician can calculate a posttest probability, using HOMER, as is also demonstrated in the discussion. The ability to have an objective probability of N1 disease after a negative EBUS TBNA can assist the multidisciplinary lung cancer team when weighing the harm of SABR with occult N1 disease versus the harm of a larger radiation field and the addition of chemotherapy for presumed N1 disease.

onstrated in the discussion. The ability to have an objective probability of N1 disease after a negative EBUS TBNA can assist the multidisciplinary lung cancer team when weighing the harm of SABR with occult N1 disease versus the harm of a larger radiation field and the addition of chemotherapy for presumed N1 disease. As the authors warn in the discussion, HOMER should not be used to calculate the sensitivity of EBUS TBNA or the pretest probability of nodal disease, as there was no gold standard (i.e., surgical lymph node dissection) to compare with EBUS TBNA. Therefore, one must often put the model in the context of an assumed EBUS TBNA sensitivity, which is probably dependent on technique, lymph node size, necrosis, and tumor cellularity of each nodal metastasis. As mentioned here, EBUS TBNA may not be highly sensitive for NSCLC in radiographically normal lymph nodes (8). Clinicians may need to integrate HOMER with other observational studies associating standardized uptake value (SUVmax) of the primary tumor, adenocarcinoma histology, non-lower lobe tumors, and tumor size with occult lymph node metastasis after negative preoperative positron emission tomography/computed tomography (10–13).

icians may need to integrate HOMER with other observational studies associating standardized uptake value (SUVmax) of the primary tumor, adenocarcinoma histology, non-lower lobe tumors, and tumor size with occult lymph node metastasis after negative preoperative positron emission tomography/computed tomography (10–13). HOMER affords the opportunity to integrate data-driven decision making into our NSCLC staging and treatment decisions, much in the way we use probability to guide the management of lung nodules (14). The model’s performance could even be further refined as more data become available for patients with N1 disease. There are exciting ways to imagine an extended data set and similar methods being employed to predict other clinically meaningful outcomes in NSCLC. For nonsurgical patients, can we model the probability of long-term clinical response, using SABR, after a negative systematic EBUS? For surgical patients, can a model to predict occult lymph node metastasis after a negative systematic EBUS be similarly derived and validated? HOMER is an excellent example of using evidence collected from current practice to rigorously create a novel prediction tool to aid future clinical decisions. It is an important guide in practice and in principle, as we continue to strive for more evidence-based and data-driven care for patients with lung cancer.

ated? HOMER is an excellent example of using evidence collected from current practice to rigorously create a novel prediction tool to aid future clinical decisions. It is an important guide in practice and in principle, as we continue to strive for more evidence-based and data-driven care for patients with lung cancer. N.N. is supported by a UK Medical Research Council Clinical Academic Research Partnership. This work was in part undertaken at University College London, University College London Hospital, which received a proportion of funding from the UK Department of Health’s NIHR Biomedical Research Centre’s funding scheme (N.N.). Originally Published in Press as DOI: 10.1164/rccm.201910-1933ED on October 28, 2019 Author disclosures are available with the text of this article at www.atsjournals.org.

At a Glance Commentary Scientific Knowledge on the Subject Idiopathic pulmonary fibrosis (IPF) is a devastating disease where the lungs become scarred. It is not known what causes the scarring, but there have been 17 regions of the genome that have been reported as associated with increased susceptibility to IPF from previous genome-wide association studies. These identify host defense (particularly mucus production), cell–cell adhesion, signaling, and telomere maintenance as important processes in the development of lung fibrosis. What This Study Adds to the Field By combining all previous IPF genome-wide association studies, we have identified three novel regions of the genome identified with IPF risk and confirmed 11 of the 17 previously reported regions. The three novel regions implicate the genes DEPTOR, KIF15, and MAD1L1. These findings support recent research that shows mTOR signaling promotes lung fibrogenesis and also implicate spindle-assembly genes in the development of IPF. Idiopathic pulmonary fibrosis (IPF) is a devastating lung disease characterized by the buildup of scar tissue. It is believed that damage to the alveolar epithelium is followed by an aberrant wound-healing response leading to the deposition of dense fibrotic tissue, reducing the lungs’ flexibility and inhibiting gas transfer (1). Treatment options are limited, and half of individuals diagnosed with IPF die within 3 to 5 years (1, 2). Two drugs (pirfenidone and nintedanib) have been approved for the treatment of IPF, but neither offer a cure, and they only slow disease progression.

reducing the lungs’ flexibility and inhibiting gas transfer (1). Treatment options are limited, and half of individuals diagnosed with IPF die within 3 to 5 years (1, 2). Two drugs (pirfenidone and nintedanib) have been approved for the treatment of IPF, but neither offer a cure, and they only slow disease progression. IPF is associated with a number of environmental and genetic factors. Identifying regions of the genome contributing to disease risk improves our understanding of the biological processes underlying IPF and helps in the development of new treatments (3). To date, genome-wide association studies (4–8) (GWAS) have reported 17 common variant (minor allele frequency [MAF] >5%) signals associated with IPF, stressing the importance of host defense, telomere maintenance, cell–cell adhesion, and signaling with respect to disease susceptibility. The sentinel (most strongly associated) variant, rs35705950, in one of these signals that maps to the promoter region of the MUC5B gene has a much larger effect on disease susceptibility than other reported risk variants with each copy of the risk allele associated with a fivefold increase in odds of disease (9). Despite this, the variant rs35705950 has a risk allele frequency of only 35% in cases (compared with 11% in the general population) and so does not explain all IPF risk. Rare variants (MAF < 1%) in telomere-related and surfactant genes have also been implicated in familial pulmonary fibrosis and sporadic IPF (10, 11).

Despite this, the variant rs35705950 has a risk allele frequency of only 35% in cases (compared with 11% in the general population) and so does not explain all IPF risk. Rare variants (MAF < 1%) in telomere-related and surfactant genes have also been implicated in familial pulmonary fibrosis and sporadic IPF (10, 11). In this study, we aimed to identify previously unreported genetic associations with IPF to improve our understanding of disease susceptibility and generate new hypotheses about disease pathogenesis. We conducted a large GWAS of IPF susceptibility by utilizing all European cases and controls recruited to any previously reported IPF GWAS (5–8) and meta-analyzing the results. This was followed by replication in individuals not previously included in IPF GWAS and bioinformatic analysis of gene expression data to identify the genes underlying the identified association signals. As specific IPF-associated variants have also been shown to overlap with other related respiratory traits including lung function in the general population, chronic obstructive pulmonary disease (COPD) (with genetic effects in opposite directions between COPD and IPF) (12–14), and interstitial lung abnormalities (ILAs) (which might be a precursor lesion for IPF) (15), we tested for association of the IPF susceptibility variants with these respiratory phenotypes in independent datasets. Finally, using polygenic risk scores, we tested whether there was still a substantial contribution to IPF risk from genetic variants with as yet unconfirmed associations with IPF susceptibility.

(15), we tested for association of the IPF susceptibility variants with these respiratory phenotypes in independent datasets. Finally, using polygenic risk scores, we tested whether there was still a substantial contribution to IPF risk from genetic variants with as yet unconfirmed associations with IPF susceptibility. Some of the results of these studies have been previously reported in the form of two abstracts and a preprint (16–18).

(15), we tested for association of the IPF susceptibility variants with these respiratory phenotypes in independent datasets. Finally, using polygenic risk scores, we tested whether there was still a substantial contribution to IPF risk from genetic variants with as yet unconfirmed associations with IPF susceptibility. Some of the results of these studies have been previously reported in the form of two abstracts and a preprint (16–18). Methods Study Cohorts We analyzed genome-wide data from three previously described independent IPF case–control collections (named here as the Chicago [5], Colorado [6], and UK [8] studies; please refer to the online supplement for summaries of these collections). Two more independent case–control collections (named here as the UUS [United States, United Kingdom, and Spain] and Genentech studies) were included as replication datasets. The new UUS study recruited cases from the United States, United Kingdom, and Spain and selected controls from UK Biobank (19) (full details on the recruitment, genotyping, and quality control of UUS cases and controls can be found in the online supplement). The previously described (20) Genentech study consisted of cases from three IPF clinical trials and controls from four non-IPF clinical trials (see the online supplement). All studies were restricted to unrelated individuals of European ancestry, and we applied stringent quality control measures (full details of the quality control measures of each study can be found in the online supplement and Figure E1 in the online supplement). All studies diagnosed cases using American Thoracic Society and European Respiratory Society guidelines (21–23) and had appropriate institutional review board or ethics approval.

es (full details of the quality control measures of each study can be found in the online supplement and Figure E1 in the online supplement). All studies diagnosed cases using American Thoracic Society and European Respiratory Society guidelines (21–23) and had appropriate institutional review board or ethics approval. Genotype data for the Colorado, Chicago, UK, and UUS studies were imputed separately using the Haplotype Reference Consortium r1.1 panel (24) (see the online supplement). For individuals in the Genentech study, genotypes were derived from whole-genome sequencing data. Duplicated individuals between studies were removed (see the online supplement). Identification of IPF Susceptibility Signals In each of the Chicago, Colorado, and UK studies separately, a genome-wide analysis of IPF susceptibility, using SNPTEST (25) v2.5.2, was conducted adjusting for the first 10 principal components to account for fine-scale population structure. Only biallelic autosomal variants that had a minor allele count ≥10 were in the Hardy–Weinberg Equilibrium (P > 1 × 10−6) and were well-imputed (imputation quality R2 > 0.5) in at least two studies were included. A genome-wide meta-analysis of the association summary statistics was performed across the Chicago, Colorado, and UK studies using R v3.5.1 (discovery stage). Conditional analyses were performed to identify independent association signals in each locus (see the online supplement).

in at least two studies were included. A genome-wide meta-analysis of the association summary statistics was performed across the Chicago, Colorado, and UK studies using R v3.5.1 (discovery stage). Conditional analyses were performed to identify independent association signals in each locus (see the online supplement). Sentinel variants (defined as the variant in an association signal where no other variants within 1 Mb showed a stronger association) of the novel signals reaching genome-wide significance in the meta-analysis (P < 5 × 10−8), and nominally significant (P < 0.05) with consistent direction of effect in each study, were further tested in the replication samples. We considered novel signals to be associated with IPF susceptibility if they reached a Bonferroni-corrected threshold (P < 0.05/number of signals followed up) in a meta-analysis of the UUS and Genentech studies (replication stage; see the online supplement). Previously reported signals with P < 5 × 10−8 in the discovery meta-analysis were deemed a confirmed association.

F susceptibility if they reached a Bonferroni-corrected threshold (P < 0.05/number of signals followed up) in a meta-analysis of the UUS and Genentech studies (replication stage; see the online supplement). Previously reported signals with P < 5 × 10−8 in the discovery meta-analysis were deemed a confirmed association. Characterization of Signals and Functional Effects To further refine our association signals to include only variants with the highest probabilities of being causal, Bayesian fine-mapping was undertaken. This approach takes all variants within the associated locus and, using the GWAS association results, calculates the probability of each variant being the true causal variant (under the assumptions that there is one causal variant and that the causal variant has been measured). The probabilities are then combined across variants to define the smallest set of variants that is 95% likely to contain the causal variant (i.e., the 95% credible set) for each IPF susceptibility signal (see the online supplement).

ptions that there is one causal variant and that the causal variant has been measured). The probabilities are then combined across variants to define the smallest set of variants that is 95% likely to contain the causal variant (i.e., the 95% credible set) for each IPF susceptibility signal (see the online supplement). To identify which genes might be implicated by the IPF susceptibility signals, we identified whether any variants in the credible sets were genic coding variants and defined as deleterious (using Variant Effect Predictor [VEP] [26]). In addition, we tested to see if any of the credible set variants were associated with gene expression using three expression quantitative trait loci (eQTL) resources (the Lung eQTL study [n = 1,111] [27–29], the NESDA-NTR [Netherlands Study of Depression and Anxiety-Netherlands Twin Register] blood eQTL database [n = 4,896] [30], and 48 tissues in GTEx [31] [n between 80 and 491]; see the online supplement). Where IPF susceptibility variants were found to be associated with expression levels of a gene, we tested whether the same variant was likely to be causal both for differences in gene expression and IPF susceptibility. We only report associations with gene expression where the probability of the same variant driving both the IPF susceptibility signal and gene expression signal exceeded 80% (see the online supplement).

sted whether the same variant was likely to be causal both for differences in gene expression and IPF susceptibility. We only report associations with gene expression where the probability of the same variant driving both the IPF susceptibility signal and gene expression signal exceeded 80% (see the online supplement). To investigate whether the IPF susceptibility variants that were in noncoding regions of the genome might be in regions with regulatory functions (for example, in regions of open chromatin), we investigated the likely functional impact of those variants using DeepSEA (deep learning-based sequence analyzer) (32). Taking all of the IPF susceptibility variants together, we tested for overall enrichment in regulatory regions specific to particular cell and tissue types using FORGE (functional element overlap analysis of the results of GWAS experiments) (33) and GARFIELD (GWAS analysis of regulatory or functional information enrichment with LD correction) (34). Finally, we investigated whether the genes that were near to the IPF susceptibility variants were more likely to be differentially expressed between IPF cases and controls in four lung epithelial cell types, using SNPsea (35). More details are provided in the online supplement.

tion enrichment with LD correction) (34). Finally, we investigated whether the genes that were near to the IPF susceptibility variants were more likely to be differentially expressed between IPF cases and controls in four lung epithelial cell types, using SNPsea (35). More details are provided in the online supplement. Shared Genetic Susceptibility with Other Respiratory Traits As previous studies have reported shared genetic susceptibility for IPF and other lung traits (12, 13, 15), we investigated whether the new and previously reported IPF susceptibility signals were associated with quantitative lung function measures in a GWAS of 400,102 individuals (36) or with ILAs in a GWAS comparing 1,699 individuals with an ILA and 10,247 controls (37). Lung function measures investigated were FEV1, FVC, the ratio FEV1/FVC (used in the diagnosis of COPD), and peak expiratory flow. We applied a Bonferroni corrected P value threshold to define variants also associated with ILAs or lung function.

in a GWAS comparing 1,699 individuals with an ILA and 10,247 controls (37). Lung function measures investigated were FEV1, FVC, the ratio FEV1/FVC (used in the diagnosis of COPD), and peak expiratory flow. We applied a Bonferroni corrected P value threshold to define variants also associated with ILAs or lung function. Polygenic Risk Scores The contribution of as yet unreported variants to IPF susceptibility was assessed using polygenic risk scores. For each individual in the UUS study, the weighted score was calculated as the number of risk alleles, multiplied by the effect size of the variant (as a weighting), summed across all variants included in the score. Effect sizes were taken from the discovery GWAS and independent variants selected using a linkage disequilibrium r2 ≤ 0.1. As we wanted to explore the contribution from as yet unreported variants, we excluded variants within 1 Mb of each IPF susceptibility locus from the risk score calculation (see the online supplement).

ect sizes were taken from the discovery GWAS and independent variants selected using a linkage disequilibrium r2 ≤ 0.1. As we wanted to explore the contribution from as yet unreported variants, we excluded variants within 1 Mb of each IPF susceptibility locus from the risk score calculation (see the online supplement). The score was tested to identify whether it was associated with IPF susceptibility, adjusting for 10 principal components to account for fine-scale population structure, using PRSice v1.25 (38). We altered the number of variants included in the risk score calculation using a sliding P threshold (PT) such that the variant had to have a P value <PT in the genome-wide meta-analysis to be included in the score. This allows us to explore whether variants that do not reach statistical significance in GWAS of current size contribute to disease susceptibility. We used the recommended significance threshold of P < 0.001 for determining significantly associated risk scores (38). Results Following quality control, 541 cases and 542 controls from the Chicago study, 1,515 cases and 4,683 controls from the Colorado study, and 612 cases and 3,366 controls from the UK study were available (Table 1 and Figure E1) to contribute to the discovery stage of the genome-wide susceptibility analysis (Figure 1). For the replication stage of the GWAS, after quality control, there were 792 cases and 10,000 controls available in the UUS study and 664 cases and 1,874 controls available in the Genentech study (see the online supplement). Table 1. Demographics of Study Cohorts

Results Following quality control, 541 cases and 542 controls from the Chicago study, 1,515 cases and 4,683 controls from the Colorado study, and 612 cases and 3,366 controls from the UK study were available (Table 1 and Figure E1) to contribute to the discovery stage of the genome-wide susceptibility analysis (Figure 1). For the replication stage of the GWAS, after quality control, there were 792 cases and 10,000 controls available in the UUS study and 664 cases and 1,874 controls available in the Genentech study (see the online supplement). Table 1. Demographics of Study Cohorts Chicago Colorado UK UUS Genentech Cases Controls Cases Controls Cases Controls Cases Controls Cases Controls n 541 542 1,515 4,683 612 3,366 792 10,000 664 1,874 Genotyping array/sequencing Affymetrix 6.0 SNP array Illumina Human 660W Quad BeadChip Affymetrix UK BiLEVE array Affymetrix UK BiLEVE and UK Biobank arrays Affymetrix UK Biobank and Spain Biobank arrays Affymetrix UK BiLEVE and UK Biobank arrays HiSeq X Ten platform (Illumina) Imputation panel HRC HRC HRC HRC — Age, yr, mean 68 63* 66 — 70† 65 69 58 68 — Sex, M, % 71‡ 47§ 68 49 70.8 70.0 75.2 72.1 73.5 27.1 Ever smokers, % 72 42 — — 72.9‖ 70.0 68.7¶ 68.0 67.3 18.1** Definition of abbreviations: HRC = Haplotype Reference Consortium; UUS = United States, United Kingdom, and Spain. * Age only available for 103 Chicago controls. † Age available for 602 UK cases. ‡ Sex only available for 500 Chicago cases. § Sex only available for 510 Chicago controls. ‖ Smoking status only recorded for 236 UK cases.

Chicago Colorado UK UUS Genentech Cases Controls Cases Controls Cases Controls Cases Controls Cases Controls n 541 542 1,515 4,683 612 3,366 792 10,000 664 1,874 Genotyping array/sequencing Affymetrix 6.0 SNP array Illumina Human 660W Quad BeadChip Affymetrix UK BiLEVE array Affymetrix UK BiLEVE and UK Biobank arrays Affymetrix UK Biobank and Spain Biobank arrays Affymetrix UK BiLEVE and UK Biobank arrays HiSeq X Ten platform (Illumina) Imputation panel HRC HRC HRC HRC — Age, yr, mean 68 63* 66 — 70† 65 69 58 68 — Sex, M, % 71‡ 47§ 68 49 70.8 70.0 75.2 72.1 73.5 27.1 Ever smokers, % 72 42 — — 72.9‖ 70.0 68.7¶ 68.0 67.3 18.1** Definition of abbreviations: HRC = Haplotype Reference Consortium; UUS = United States, United Kingdom, and Spain. * Age only available for 103 Chicago controls. † Age available for 602 UK cases. ‡ Sex only available for 500 Chicago cases. § Sex only available for 510 Chicago controls. ‖ Smoking status only recorded for 236 UK cases. ¶ Smoking status only recorded for 753 idiopathic pulmonary fibrosis cases in UUS. ** Smoking status only recorded for 481 of the Genentech controls.

* Age only available for 103 Chicago controls. † Age available for 602 UK cases. ‡ Sex only available for 500 Chicago cases. § Sex only available for 510 Chicago controls. ‖ Smoking status only recorded for 236 UK cases. ¶ Smoking status only recorded for 753 idiopathic pulmonary fibrosis cases in UUS. ** Smoking status only recorded for 481 of the Genentech controls. Figure 1. Manhattan plot of discovery analysis results. The x axis shows chromosomal position, and the y axis shows the −log(P value) for each variant in the discovery genome-wide analysis. The red line shows genome-wide significance (P < 5 × 10−8), and variants in green met the criteria for further study in the replication analysis (i.e., reached genome-wide significance in the discovery meta-analysis and had P < 0.05 and consistent direction of effects in each study). Genes in gray are previously reported signals that reach significance in the discovery genome-wide meta-analysis. Genes in black are the novel signals identified in the discovery analysis that reach genome-wide significance when meta-analyzing discovery and replication samples. The signals that did not replicate are shown in red. For ease of visualization the y axis has been truncated at 25.

he discovery genome-wide meta-analysis. Genes in black are the novel signals identified in the discovery analysis that reach genome-wide significance when meta-analyzing discovery and replication samples. The signals that did not replicate are shown in red. For ease of visualization the y axis has been truncated at 25. To identify new signals of association, we meta-analyzed the genome-wide association results for IPF susceptibility for the Chicago, Colorado, and UK discovery studies. This gave a maximum sample size of up to 2,668 cases and 8,591 controls for 10,790,934 well-imputed (R2 > 0.5) variants with minor allele count ≥10 in each study and which were available in two or more of the studies (Figure E2).

iation results for IPF susceptibility for the Chicago, Colorado, and UK discovery studies. This gave a maximum sample size of up to 2,668 cases and 8,591 controls for 10,790,934 well-imputed (R2 > 0.5) variants with minor allele count ≥10 in each study and which were available in two or more of the studies (Figure E2). Three novel signals (in 3p21.31 [near KIF15, Figure 2A], 7p22.3 [near MAD1L1, Figure 2B], and 8q24.12 [near DEPTOR, Figure 2C]) showed a genome-wide significant (P < 5 × 10−8) association with IPF susceptibility in the discovery meta-analysis and were also significant after adjusting for multiple testing (P < 0.01) in the replication stage comprising 1,467 IPF cases and 11,874 controls (Tables 2 and E1). Two additional loci were genome-wide significant in the genome-wide discovery analysis but did not reach significance in the replication studies. The sentinel variants of these two signals were a low-frequency intronic variant in RTEL1 (MAF = 2.1%, replication P = 0.012) and a rare intronic variant in HECTD2 (MAF = 0.3%, replication P = 0.155). Conditional analyses did not identify any additional independent association signals at the new or previously reported IPF susceptibility loci (Figure E5).

ignals were a low-frequency intronic variant in RTEL1 (MAF = 2.1%, replication P = 0.012) and a rare intronic variant in HECTD2 (MAF = 0.3%, replication P = 0.155). Conditional analyses did not identify any additional independent association signals at the new or previously reported IPF susceptibility loci (Figure E5). Figure 2. Region plots of three novel idiopathic pulmonary fibrosis susceptibility loci from discovery genome-wide meta-analysis. Each point represents a variant with chromosomal position on the x axis and the −log(P value) on the y axis. Variants are colored in by linkage disequilibrium with the sentinel variant. Blue lines show the recombination rate, and gene locations are shown at the bottom of the plot. Region plots are shown for the three replicated novel idiopathic pulmonary fibrosis susceptibility loci, i.e., (A) the susceptibility signal on chromosome 3 near KIF15, (B) the susceptibility signal on chromosome 7 near MAD1L1, and (C) the susceptibility signal on chromosome 8 near DEPTOR. Table 2. Discovery and Replication Association Analysis Results for the Five Signals Reaching Significance in the Discovery Genome-Wide Association Studies that Have Not Previously Been Reported as Associated with Idiopathic Pulmonary Fibrosis

Figure 2. Region plots of three novel idiopathic pulmonary fibrosis susceptibility loci from discovery genome-wide meta-analysis. Each point represents a variant with chromosomal position on the x axis and the −log(P value) on the y axis. Variants are colored in by linkage disequilibrium with the sentinel variant. Blue lines show the recombination rate, and gene locations are shown at the bottom of the plot. Region plots are shown for the three replicated novel idiopathic pulmonary fibrosis susceptibility loci, i.e., (A) the susceptibility signal on chromosome 3 near KIF15, (B) the susceptibility signal on chromosome 7 near MAD1L1, and (C) the susceptibility signal on chromosome 8 near DEPTOR. Table 2. Discovery and Replication Association Analysis Results for the Five Signals Reaching Significance in the Discovery Genome-Wide Association Studies that Have Not Previously Been Reported as Associated with Idiopathic Pulmonary Fibrosis Chr Pos rsid Locus Major Allele Minor Allele MAF (%) Discovery Meta-Analysis Replication Meta-Analysis Meta-Analysis of Discovery and Replication OR [95% CI] P Value OR [95% CI] P Value OR [95% CI] P Value 3 44902386 rs78238620 KIF15 T A 5.3 1.58 [1.37–1.83] 5.12 × 10−10 1.48 [1.24–1.77] 1.43 × 10−5 1.54 [1.38–1.73] 4.05 × 10−14 7 1909479 rs12699415 MAD1L1 G A 42.0 1.28 [1.19–1.37] 7.15 × 10−13 1.29 [1.18–1.41] 2.27 × 10−8 1.28 [1.21–1.35] 9.38 × 10−20 8 120934126 rs28513081 DEPTOR A G 42.8 0.82 [0.76–0.87] 1.20 × 10−9 0.87 [0.80–0.95] 0.002 0.83 [0.79–0.88] 1.93 × 10−11 10 93271016 rs537322302 HECTD2 C G 0.3 7.82 [3.77–16.2] 3.43 × 10−8 1.75 [0.81–3.78] 0.155 3.85 [2.27–6.54] 6.25 × 10−7 20 62324391 rs41308092 RTEL1 G A 2.1 2.12 [1.67–2.69] 7.65 × 10−10 1.45 [1.08–1.94] 0.012 1.82 [1.51–2.19] 2.24 × 10−10 Definition of abbreviations: Chr = chromosome; CI = confidence interval; MAF = minor allele frequency; OR = odds ratio; Pos = position; rsid = reference SNP cluster ID.

55 3.85 [2.27–6.54] 6.25 × 10−7 20 62324391 rs41308092 RTEL1 G A 2.1 2.12 [1.67–2.69] 7.65 × 10−10 1.45 [1.08–1.94] 0.012 1.82 [1.51–2.19] 2.24 × 10−10 Definition of abbreviations: Chr = chromosome; CI = confidence interval; MAF = minor allele frequency; OR = odds ratio; Pos = position; rsid = reference SNP cluster ID. The minor allele is the effect allele, and the MAF is taken from across the studies used in the discovery meta-analysis.

55 3.85 [2.27–6.54] 6.25 × 10−7 20 62324391 rs41308092 RTEL1 G A 2.1 2.12 [1.67–2.69] 7.65 × 10−10 1.45 [1.08–1.94] 0.012 1.82 [1.51–2.19] 2.24 × 10−10 Definition of abbreviations: Chr = chromosome; CI = confidence interval; MAF = minor allele frequency; OR = odds ratio; Pos = position; rsid = reference SNP cluster ID. The minor allele is the effect allele, and the MAF is taken from across the studies used in the discovery meta-analysis. To identify the likely causal genes for each new signal, we investigated whether any of the variants were also associated with changes in gene expression (Table 3). The sentinel variant (rs78238620) of the novel signal on chromosome 3 was a low-frequency variant (MAF = 5%) in an intron of KIF15 with the minor allele being associated with increased susceptibility to IPF and decreased expression of KIF15 in brain tissue and the nearby gene TMEM42 in thyroid (31) (Figure E7 and Tables E2 and E3i). The IPF risk allele for the novel chromosome 7 signal (rs12699415, MAF = 42%) was associated with decreased expression of MAD1L1 in heart tissue (31) (Figure E8 and Tables E2 and E3ii). For the signal on chromosome 8, the sentinel variant (rs28513081) was located in an intron of DEPTOR, and the IPF risk allele was associated with decreased expression of DEPTOR (in colon, lung, and skin [27–29, 31]) and RP11-760H22.2 (in colon and lung [31]). The risk allele was also associated with increased expression of DEPTOR (in whole blood [30]), TAF2 (in colon [31]), RP11-760H22.2 (in adipose [31]), and KB-1471A8.1 (in adipose and skin [31], Figure E9 and Tables E2 and E3iii). There were no variants predicted to be highly deleterious within the fine-mapped signals for any of the loci.

s also associated with increased expression of DEPTOR (in whole blood [30]), TAF2 (in colon [31]), RP11-760H22.2 (in adipose [31]), and KB-1471A8.1 (in adipose and skin [31], Figure E9 and Tables E2 and E3iii). There were no variants predicted to be highly deleterious within the fine-mapped signals for any of the loci. Table 3. Gene Expression and Spirometric Results for the Three Novel IPF Susceptibility Loci Chr rsid of Sentinel Variant Annotation eQTL FEV1 FVC FEV1/FVC Lung Tissue Nonlung Tissue β [95% CI] P Value β [95% CI] P Value β [95% CI] P Value 3 rs78238620 Intron (KIF15) — ↓ KIF15 ↓ TMEM42 −0.011 [−0.022 to 0.000] 0.069 −0.022 [−0.033 to 0.011] 2.92 × 10−4 0.017 [0.006 to 0.028] 0.005 7 rs12699415 Intron (MAD1L1) — ↓ MAD1L1 −0.007 [−0.012 to −0.002] 0.011 −0.011 [−0.016 to −0.007] 1.41 × 10−5 0.008 [0.003 to 0.012] 0.005 8 rs28513081 Intron (DEPTOR) ↓ DEPTOR ↓ RP11-760H22.2 ↕ DEPTOR ↕ RP11-760H22.2 ↑ KB-1471A8.1 ↑ TAF2 0.001 [−0.004 to 0.006] 0.822 −0.005 [−0.010 to −0.001] 0.045 0.011 [0.006 to 0.016] 4.22 × 10−5 Definition of abbreviations: Chr = chromosome; CI = confidence interval; eQTL = expression quantitative trait loci; IPF = idiopathic pulmonary fibrosis; rsid = reference SNP cluster ID.

↕ RP11-760H22.2 ↑ KB-1471A8.1 ↑ TAF2 0.001 [−0.004 to 0.006] 0.822 −0.005 [−0.010 to −0.001] 0.045 0.011 [0.006 to 0.016] 4.22 × 10−5 Definition of abbreviations: Chr = chromosome; CI = confidence interval; eQTL = expression quantitative trait loci; IPF = idiopathic pulmonary fibrosis; rsid = reference SNP cluster ID. Annotation of the variant was taken from Variant Effect Predictor (VEP). A list of all variants included in the credible sets with their annotations and eQTL results can be found in Table E3. For colocalization, only genes where there was a greater than 80% probability of colocalization between the IPF risk signal and gene expression of that gene are reported in this table. In the colocalization column, ↑ denotes that the allele that increases IPF risk was associated with increased expression of the gene, ↓ denotes that the IPF risk allele was associated with decreased expression of the gene, and ↕ denotes that the IPF risk allele was associated with increased expression in some tissues and decreased expression in others. Full results from the eQTL and colocalization analyses can be found in Table E2. The spirometric results for the three novel IPF risk loci are taken from Shrine and colleagues (36) using the allele associated with increased IPF risk as the effect allele, with β being the change in z-score units. Results for all IPF risk variants can be found in Table E6.

alization analyses can be found in Table E2. The spirometric results for the three novel IPF risk loci are taken from Shrine and colleagues (36) using the allele associated with increased IPF risk as the effect allele, with β being the change in z-score units. Results for all IPF risk variants can be found in Table E6. We confirmed genome-wide significant associations with IPF susceptibility for 11 of the 17 previously reported signals (in or near TERC, TERT, DSP, 7q22.1, MUC5B, ATP11A, IVD, AKAP13, KANSL1, FAM13A, and DPP9; Table E1 and Figure E4). The signal at FAM13A, while genome-wide significant in the discovery meta-analysis, was not significant in the Chicago study. This was the only signal reaching genome-wide significance in the discovery genome-wide meta-analysis that did not reach at least nominal significance in each study in the discovery analysis. Three further previously reported signals at 11p15.5 (near MUC5B) were no longer genome-wide significant after conditioning on the MUC5B promoter variant (Table E1), consistent with previous reports (6, 39).

ome-wide meta-analysis that did not reach at least nominal significance in each study in the discovery analysis. Three further previously reported signals at 11p15.5 (near MUC5B) were no longer genome-wide significant after conditioning on the MUC5B promoter variant (Table E1), consistent with previous reports (6, 39). Of the 14 IPF susceptibility signals (i.e., the 11 previously reported signals we confirmed and three novel signals), the only variant predicted to have a potential functional effect on gene regulation through disruption of chromatin structure or transcription factor binding motifs (using DeepSEA) was rs2013701 (in an intron of FAM13A), which was associated with a change in DNase I hypersensitivity in 18 cell types and FOXA1 in the T-47D cell line (a breast cancer cell line derived from a pleural effusion, Table E4). The 14 IPF susceptibility signals were found to be enriched in DNase I hypersensitivity site regions in multiple tissues including fetal lung tissue (Figures E10 and E11). No enrichment in differential expression in airway epithelial cells between IPF cases and healthy controls was observed for the 14 IPF susceptibility signals when using SNPsea (Table E5).

found to be enriched in DNase I hypersensitivity site regions in multiple tissues including fetal lung tissue (Figures E10 and E11). No enrichment in differential expression in airway epithelial cells between IPF cases and healthy controls was observed for the 14 IPF susceptibility signals when using SNPsea (Table E5). Previous studies have reported an overlap of genetic association loci between lung function and IPF (12). We undertook a lookup of the 14 IPF susceptibility loci in the largest GWAS of lung function in the general population published to date (36). The sentinel variants of 12 of the 14 IPF susceptibility loci were at least nominally associated (P < 0.05) with one or more lung function trait in general population studies (Tables 3 and E6). After adjustments for multiple testing (P < 5.2 × 10−4), the previously reported variants at FAM13A, DSP, and IVD were associated with decreased FVC, and variants at FAM13A, DSP, 7q22.1 (ZKSCAN1), and ATP11A were associated with increased FEV1/FVC. Similarly, for the three novel susceptibility variants, all showed at least a nominal association with decreased FVC and increased FEV1/FVC. We observed a nominally significant association of the MUC5B IPF risk allele with decreased FVC and increased FEV1/FVC. The IPF risk alleles at MAPT were significantly associated with both increased FEV1 and FVC. To determine how the variants identified for IPF susceptibility are related to differences in lung function between cases and controls, we investigated whether variants known to be associated with lung function show an association in our IPF GWAS. Of the 279 variants reported (36) as associated with lung function (Table E7), 8 showed an association with lung function after corrections for multiple testing (located in or near MCL1, DSP, ZKSCAN1, OBFC1, IVD, MAPT, and two signals in FAM13A).

to be associated with lung function show an association in our IPF GWAS. Of the 279 variants reported (36) as associated with lung function (Table E7), 8 showed an association with lung function after corrections for multiple testing (located in or near MCL1, DSP, ZKSCAN1, OBFC1, IVD, MAPT, and two signals in FAM13A). As interstitial lung abnormalities may be a precursor to IPF in a subset of patients, and there have been previous reports of shared genetic etiology between IPF and ILAs (37, 40, 41), we investigated whether our three new signals and the 11 previously reported signals were associated with ILAs in the largest ILA GWAS reported to date (37). Eight of the IPF susceptibility loci were at least nominally significantly associated with either ILAs or subpleural ILAs with consistent direction of effects (i.e., the allele associated with increased IPF risk was also associated with increased ILA risk). The new KIF15, MAD1L1, and DEPTOR signals were not associated with ILAs (although the rare risk allele at HECTD2 that did not replicate in our study showed some association with an increased risk of subpleural ILAs [P = 0.003] with a large effect size similar to that observed in the IPF discovery meta-analysis).

risk). The new KIF15, MAD1L1, and DEPTOR signals were not associated with ILAs (although the rare risk allele at HECTD2 that did not replicate in our study showed some association with an increased risk of subpleural ILAs [P = 0.003] with a large effect size similar to that observed in the IPF discovery meta-analysis). To quantify the impact of as yet unreported variants on IPF susceptibility, polygenic risk scores were calculated excluding the 14 IPF susceptibility variants (as well as all variants within 1 Mb). The polygenic risk score was significantly associated with increased IPF susceptibility despite exclusion of the known genetic association signals (including MUC5B). As the PT for inclusion of variants in the score was increased, the risk score became more significant reaching a plateau at around PT = 0.2 with risk score P < 3.08 × 10−23 and explaining around 2% of the phenotypic variation (Figure E12), suggesting that there is a modest but statistically significant contribution of additional as yet undetected variants to IPF susceptibility. Further increasing PT beyond 0.2 did not improve the predictive accuracy of the risk score. Discussion We undertook the largest GWAS of IPF susceptibility to date and identified three novel signals of association that implicated genes not previously known to be important in IPF.

To quantify the impact of as yet unreported variants on IPF susceptibility, polygenic risk scores were calculated excluding the 14 IPF susceptibility variants (as well as all variants within 1 Mb). The polygenic risk score was significantly associated with increased IPF susceptibility despite exclusion of the known genetic association signals (including MUC5B). As the PT for inclusion of variants in the score was increased, the risk score became more significant reaching a plateau at around PT = 0.2 with risk score P < 3.08 × 10−23 and explaining around 2% of the phenotypic variation (Figure E12), suggesting that there is a modest but statistically significant contribution of additional as yet undetected variants to IPF susceptibility. Further increasing PT beyond 0.2 did not improve the predictive accuracy of the risk score. Discussion We undertook the largest GWAS of IPF susceptibility to date and identified three novel signals of association that implicated genes not previously known to be important in IPF. The strongest evidence for the new signal on chromosome 8 implicates DEPTOR, which encodes the dishevelled, Egl-10 and Pleckstrin domain–containing mTOR-interacting protein. DEPTOR inhibits mTOR (mammalian target of rapamycin) kinase activity as part of both the mTORC1 and mTORC2 protein complexes. The IPF risk allele at this locus was associated with decreased gene expression of DEPTOR in lung tissue (Table E2). TGFβ-induced DEPTOR suppression can stimulate collagen synthesis (42), and the importance of mTORC1 signaling via 4E-BP1 for TGFβ-induced collagen synthesis has recently been demonstrated in fibrogenesis (43). MAD1L1, implicated by a new signal on chromosome 7 and eQTL analyses of nonlung tissue, is a mitotic checkpoint gene, mutations in which have been associated with multiple cancers including lung cancer (44, 45). Studies have shown that MAD1, a homolog of MAD1L1, can inhibit TERT activity (or possibly enforce expression of TERT when the promoter E-box is mutated) (45, 46). This could suggest that MAD1L1 may increase IPF susceptibility through reduced telomerase activity. Another spindle-assembly–related gene (47), KIF15, was implicated by the new signal on chromosome 3 (along with TMEM42).

TERT activity (or possibly enforce expression of TERT when the promoter E-box is mutated) (45, 46). This could suggest that MAD1L1 may increase IPF susceptibility through reduced telomerase activity. Another spindle-assembly–related gene (47), KIF15, was implicated by the new signal on chromosome 3 (along with TMEM42). The genome-wide study also identified two signals that were not replicated after multiple testing adjustments. RTEL1, a gene involved in telomere elongation regulation, has not previously been identified in an IPF GWAS; however, the collective effect of rare variants in RTEL1 has been reported as associated with IPF susceptibility (48–54). The ubiquitin E3 ligase encoded by HECTD2 has been shown to have a proinflammatory role in the lung, and other HECTD2 variants may be protective against acute respiratory distress syndrome (55). However, the lack of replication for these signals in our data suggests that further exploration of their relationship to interstitial lung diseases is warranted.

CTD2 has been shown to have a proinflammatory role in the lung, and other HECTD2 variants may be protective against acute respiratory distress syndrome (55). However, the lack of replication for these signals in our data suggests that further exploration of their relationship to interstitial lung diseases is warranted. By combining the largest available GWAS datasets for IPF, we were able to confirm 11 of 17 previously reported signals. Conditional analysis at the 11p15.5 region indicated that previously reported signals at MUC2 and TOLLIP were not independent of the association with the MUC5B promoter variant. Previously reported signals at EHMT2, OBFC1, and MDGA2 were only found to be associated in one of the discovery studies and showed no evidence of an association with IPF susceptibility in the other two discovery studies. Only the 11 signals that we confirmed in our data were included in subsequent analyses.

er variant. Previously reported signals at EHMT2, OBFC1, and MDGA2 were only found to be associated in one of the discovery studies and showed no evidence of an association with IPF susceptibility in the other two discovery studies. Only the 11 signals that we confirmed in our data were included in subsequent analyses. The IPF susceptibility signals at DSP, FAM13A, 7q22.1 (ZKSCAN1), and 17q21.31 (MAPT) have also been reported as associated with COPD, although with opposite effects (i.e., the allele associated with increased risk of IPF being associated with decreased risk of COPD). Spirometric diagnosis of COPD was based on a reduced FEV1/FVC ratio. In an independent dataset of 400,102 individuals, eight of the IPF signals were associated with decreased FVC and with a comparatively weaker effect on FEV1. This is consistent with the lung function abnormalities associated with IPF, as well as the decreased risk of COPD. Of note, only around 3% of previously reported lung function signals (36) also showed association with IPF susceptibility in our study. This suggests that while some IPF susceptibility variants might represent genes and pathways that are important in general lung health, others are likely to represent more disease-specific processes. Using polygenic risk scores, we demonstrated that, despite the relatively large proportion of disease susceptibility explained by the known genetic signals of association reported here, IPF is highly polygenic with potentially hundreds (or thousands) of as yet unidentified variants associated with disease susceptibility.

Using polygenic risk scores, we demonstrated that, despite the relatively large proportion of disease susceptibility explained by the known genetic signals of association reported here, IPF is highly polygenic with potentially hundreds (or thousands) of as yet unidentified variants associated with disease susceptibility. A strength of our study was the large sample size compared with previous GWAS and the availability of an independent replication dataset. A limitation of our study was that the controls used were generally younger in all studies included, and there were differences in sex and smoking distributions in some of the studies. As age, sex, and smoking status were not available for all individuals in four of our datasets, we were unable to adjust for these variables without substantially reducing our sample size. However, cases and controls in the UUS and UK datasets were matched for age, sex, and smoking. The three novel signals replicated in all of the discovery and replication datasets, providing reassurance that the signals we report are robust despite differences between the datasets. As we had limited information beyond IPF diagnosis status for a large proportion of the individuals included in the studies, we cannot rule out some association with other age-related conditions that are comorbid with IPF. However, other age-related conditions were not excluded from either the cases or controls. For the signals near KIF15 and MAD1L1, there was substantial evidence for an association with gene expression in nonlung tissues but not in either of the two (nonfibrotic) lung tissue eQTL datasets. This could reflect cell type-specific effects that are missed when studying whole tissue or effects that are disease-dependent. Finally, our study was not designed to identify rare functional variant associations. As both common and rare variants are known to be important in IPF susceptibility (39), this is a limitation of our study.

ect cell type-specific effects that are missed when studying whole tissue or effects that are disease-dependent. Finally, our study was not designed to identify rare functional variant associations. As both common and rare variants are known to be important in IPF susceptibility (39), this is a limitation of our study. In summary, we report new biological insights into IPF susceptibility and demonstrate that further studies to identify the genetic determinants of IPF susceptibility are needed. Our new signals of association with IPF susceptibility provide increased support for the importance of mTOR signaling in pulmonary fibrosis as well as the possible implication of mitotic spindle-assembly genes. Acknowledgment This research has been conducted using the UK Biobank Resource under application 8389. This research used the ALICE and SPECTRE High Performance Computing Facilities at the University of Leicester.

In summary, we report new biological insights into IPF susceptibility and demonstrate that further studies to identify the genetic determinants of IPF susceptibility are needed. Our new signals of association with IPF susceptibility provide increased support for the importance of mTOR signaling in pulmonary fibrosis as well as the possible implication of mitotic spindle-assembly genes. Acknowledgment This research has been conducted using the UK Biobank Resource under application 8389. This research used the ALICE and SPECTRE High Performance Computing Facilities at the University of Leicester. R.J.A. is an Action for Pulmonary Fibrosis Research Fellow. L.V.W. holds a GSK/British Lung Foundation Chair in Respiratory Research. R.G.J. is supported by a National Institute for Health Research (NIHR) Research Professorship (NIHR reference RP-2017-08-ST2-014). I.N. is supported by the NHLBI (R01HL130796). B.G.-G. is funded by Agencia Canaria de Investigación, Innovación y Sociedad de la Información (TESIS2015010057) cofunded by European Social Fund. J.M.O. is supported by the NHLBI (K23HL138190). C.F. is supported by the Spanish Ministry of Science, Innovation and Universities (grant RTC-2017-6471-1; Ministerio de Ciencia e Innovacion/Agencia Estatal de Investigación/Fondo Europeo de Desarrollo Regional, Unión Europea) cofinanced by the European Regional Development Funds “A way of making Europe” from the European Union and by agreement OA17/008 with Instituto Tecnológico y de Energías Renovables to strengthen scientific and technological education, training, research, development and innovation in Genomics, Personalized Medicine and Biotechnology. The Spain Biobank array genotyping service was performed at CEGEN-PRB3-ISCIII, which is supported by PT17/0019, of the PE I+D+i 2013–2016, funded by Instituto de Salud Carlos III, and cofinanced by the European Regional Development Funds. P.L.M. is an Action for Pulmonary Fibrosis Research Fellow. M.O. is a fellow of the Parker B. Francis Foundation and a Scholar of the Michael Smith Foundation for Health Research. B.D.H. is supported by NIH K08 HL136928, Parker B. Francis Research Opportunity Award. M.H.C. and G.M.H. are supported by NHLBI grants R01HL113264 (M.H.C.), R01HL137927 (M.H.C.), R01HL135142 (M.H.C. and G.M.H.), R01111024 (G.M.H.), and R01130974 (G.M.H.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funding body has no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript. T.M.M. is supported by an NIHR Clinician Scientist Fellowship (NIHR Ref: CS-2013-13-017) and a British Lung Foundation Chair in Respiratory Research (C17-3). M.D.T.

ews of the NIH. The funding body has no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript. T.M.M. is supported by an NIHR Clinician Scientist Fellowship (NIHR Ref: CS-2013-13-017) and a British Lung Foundation Chair in Respiratory Research (C17-3). M.D.T. is supported by a Wellcome Trust Investigator Award (WT202849/Z/16/Z). The research was partially supported by the NIHR Leicester Biomedical Research Centre; the views expressed are those of the author(s) and not necessarily those of the National Health Service (NHS), the NIHR, or the Department of Health. I.P.H. was partially supported by the NIHR Nottingham Biomedical Research Centre; the views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR, or the Department of Health. I.S. is supported by Medical Research Council (G1000861) and Asthma UK (AUK-PG-2013-188). D.F. was supported by an Intermediate Fellowship from the Wellcome Trust (097152/Z/11/Z). This work was partially supported by the National Institute for Health Research (NIHR) Oxford Biomedical Research Centre. V.N. is funded by an NIHR Clinical Lectureship. G.G. is supported by project grant 141513-051 from the Icelandic Research Fund and Landspitali Scientific Fund A-2016-023, A-2017-029, and A-2018-025. D.J.L. and A.M. are supported by Multi-Ethnic Study of Atherosclerosis (MESA) and the MESA SNP Health Association Resource (SHARe) project are conducted and supported by the NHLBI in collaboration with MESA investigators. Support for MESA is provided by contracts HHSN268201500003I, N01-HC-95159, N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168, N01-HC-95169, UL1-TR-000040, UL1-TR-001079, UL1-TR-001420, UL1-TR-001881, and DK063491. Funding for SHARe genotyping was provided by NHLBI Contract N02-HL-64278. Genotyping was performed at Affymetrix (Santa Clara, California) and the Broad Institute of Harvard and Massachusetts Institute of Technology (Boston, Massachusetts) using the Affymetrix Genome-Wide Human SNP Array 6.0. This work was supported by NIH grants R01 HL131565 (A.M.), R01 HL103676 (D.J.L.), and R01 HL137234 (D.J.L.).

ng was performed at Affymetrix (Santa Clara, California) and the Broad Institute of Harvard and Massachusetts Institute of Technology (Boston, Massachusetts) using the Affymetrix Genome-Wide Human SNP Array 6.0. This work was supported by NIH grants R01 HL131565 (A.M.), R01 HL103676 (D.J.L.), and R01 HL137234 (D.J.L.). Data availability statement: Full summary statistics for the genome-wide meta-analysis can be accessed from https://github.com/genomicsITER/PFgenetics. Author Contributions: R.J.A., J.M.O., C.F., I.N., R.G.J., and L.V.W. designed the study. R.J.A., B.G.-G., S.-F.M., A.D., M.L.P., L.M.K., M.O., X.L., M. Ng, B.D.H., R.K.P., P.S., D.F., A.P.M., K.T.Z., and B.L.Y. analyzed the data. J.M.O., S.-F.M., M.O., R.B., M.M.-M., R.K.P., P.S., H.L.B., W.A.F., S.P.H., M.R.H., N.H., R.B.H., R.J.M., A.B.M., V.N., E.O., H.P., G.S., M.K.B.W., Y.Z., N.K., A.A., M.E.S., M. Neighbors, X.R.S., G.G., V.G., H.H., D.J.L., A.M., J.D.N., G.T.O’C., V.E.O., H.X., T.E.F., Y.B., K.H., P.J., D.C.N., D.D.S., W.T., I.P.H., I.S., M.D.T., T.M.M., M.H.C., G.M.H., D.A.S., B.L.Y., P.L.M., C.F., I.N., R.G.J., and L.V.W. were responsible for recruitment, screening and genotyping of cases and controls for idiopathic pulmonary fibrosis, interstitial lung abnormalities, and gene expression analyses. J.M.O., D.A.S., C.F., I.N., R.G.J., and L.V.W. supervised and coordinated the study. R.J.A., R.G.J., and L.V.W. led the writing of the manuscript. All authors contributed to drafting and providing critical feedback on the manuscript. This article has a related editorial.

Author Contributions: R.J.A., J.M.O., C.F., I.N., R.G.J., and L.V.W. designed the study. R.J.A., B.G.-G., S.-F.M., A.D., M.L.P., L.M.K., M.O., X.L., M. Ng, B.D.H., R.K.P., P.S., D.F., A.P.M., K.T.Z., and B.L.Y. analyzed the data. J.M.O., S.-F.M., M.O., R.B., M.M.-M., R.K.P., P.S., H.L.B., W.A.F., S.P.H., M.R.H., N.H., R.B.H., R.J.M., A.B.M., V.N., E.O., H.P., G.S., M.K.B.W., Y.Z., N.K., A.A., M.E.S., M. Neighbors, X.R.S., G.G., V.G., H.H., D.J.L., A.M., J.D.N., G.T.O’C., V.E.O., H.X., T.E.F., Y.B., K.H., P.J., D.C.N., D.D.S., W.T., I.P.H., I.S., M.D.T., T.M.M., M.H.C., G.M.H., D.A.S., B.L.Y., P.L.M., C.F., I.N., R.G.J., and L.V.W. were responsible for recruitment, screening and genotyping of cases and controls for idiopathic pulmonary fibrosis, interstitial lung abnormalities, and gene expression analyses. J.M.O., D.A.S., C.F., I.N., R.G.J., and L.V.W. supervised and coordinated the study. R.J.A., R.G.J., and L.V.W. led the writing of the manuscript. All authors contributed to drafting and providing critical feedback on the manuscript. This article has a related editorial. This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org. Originally Published in Press as DOI: 10.1164/rccm.201905-1017OC on November 11, 2019 Author disclosures are available with the text of this article at www.atsjournals.org.

, it is significant that Ladjemi and colleagues found so few IgG-secreting B cells in COPD. A final implication of this study is that the answer to the question of whether LLF B cells in COPD are bad or beneficial (26) is that many appear to be trying to help. Such a wealth of insights from catching B cells in the act. Supported by Merit Review award I01 CX000911 from the Clinical Laboratory Research and Development Service, Department of Veterans Affairs, and grant U01 HL137880 from the NHLBI, NIH. Originally Published in Press as DOI: 10.1164/rccm.201810-1907ED on October 23, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

To the Editor: The hazardous effects of environmental tobacco smoke (ETS) on the lung health of small children are generally recognized. Governmental actions need improvement in decreasing worldwide smoking prevalence (1). The global average of young men starting smoking is 40%; it is 9% for women (1). As children are born to young families, they are often exposed to ETS. We have previously shown that maternal smoking in particular adversely affects the lung function of preschoolers with asthma (2), and now we wanted to objectively examine whether early exposure to ETS has a long-term effect on lung function. We enrolled 105 children (median age, 5 yr) referred to Helsinki University Hospital with multitrigger wheeze and evidence of bronchodilator response (≥35% decrease in respiratory resistance at 5 Hz [R5]) or exercise-induced bronchoconstriction (≥35% increase in R5), or both (3). Exclusion criteria included the use of corticosteroids in the previous 6 months and respiratory tract infection in the preceding 2 weeks. After 10 years, 64 children participated in a follow-up visit. We assessed the child’s exposure to ETS at preschool age with urine cotinine levels, using gas chromatography (2), as well as with a questionnaire. Cotinine is a stable metabolite of nicotine and can be used to objectively measure exposure to ETS. It is found in saliva, urine, hair, and blood a few days after exposure to ETS. We have previously presented the gas chromatographic method and the correlation between urinary cotinine levels and ETS in this particular population (2).

stable metabolite of nicotine and can be used to objectively measure exposure to ETS. It is found in saliva, urine, hair, and blood a few days after exposure to ETS. We have previously presented the gas chromatographic method and the correlation between urinary cotinine levels and ETS in this particular population (2). Fractional exhaled nitric oxide (FeNO) and impulse oscillometry were performed following current guidelines both at enrollment and at the 10-year follow-up visit (2–5). Impulse oscillometry indices of interest included the total respiratory system resistance (R5) and reactance (X5) at 5 Hz, and area under reactance (AX), reflecting the capacitive component of reactance that is affected by small airway function. At the follow-up visit, lung function was also assessed by spirometry (6), and the children underwent dosimetric bronchial provocation with methacholine including five cumulative dose steps (7). Increased airway hyperresponsiveness was considered to be present if a ≤400-μg dose provoked at least a 20% decrease in the FEV1 (PD20FEV1). A blood eosinophil percentage of ≥4% was considered eosinophilia. A wheal diameter of ≥3 mm was considered positive for skin prick tests to local aeroallergens (birch, timothy grass, meadow fescue, mugwort, Cladosporium herbarum, dog, cat, horse, cow, and Dermatophagoides pteronyssinus).

Fractional exhaled nitric oxide (FeNO) and impulse oscillometry were performed following current guidelines both at enrollment and at the 10-year follow-up visit (2–5). Impulse oscillometry indices of interest included the total respiratory system resistance (R5) and reactance (X5) at 5 Hz, and area under reactance (AX), reflecting the capacitive component of reactance that is affected by small airway function. At the follow-up visit, lung function was also assessed by spirometry (6), and the children underwent dosimetric bronchial provocation with methacholine including five cumulative dose steps (7). Increased airway hyperresponsiveness was considered to be present if a ≤400-μg dose provoked at least a 20% decrease in the FEV1 (PD20FEV1). A blood eosinophil percentage of ≥4% was considered eosinophilia. A wheal diameter of ≥3 mm was considered positive for skin prick tests to local aeroallergens (birch, timothy grass, meadow fescue, mugwort, Cladosporium herbarum, dog, cat, horse, cow, and Dermatophagoides pteronyssinus). Nonnormally distributed data were analyzed using the Mann-Whitney U test (dichotomous comparisons), and normally distributed data with Pearson’s correlation test. We analyzed data with four categories, using one-way ANOVA. Because of the lack of impulse oscillometry reference values for adolescents, the upper quartile of R5 and lower quartile of X5 were considered suggestive of obstruction. A P value ≤0.05 was considered statistically significant. Data were analyzed using IBM SPSS 25.0.

zed data with four categories, using one-way ANOVA. Because of the lack of impulse oscillometry reference values for adolescents, the upper quartile of R5 and lower quartile of X5 were considered suggestive of obstruction. A P value ≤0.05 was considered statistically significant. Data were analyzed using IBM SPSS 25.0. Parents provided written informed consent at both study visits; children provided written informed consent only at the follow-up visit. The Research Ethics Committee of the regional university hospital approved the research protocol. Statistical approaches were verified by the biostatistics department of the university. Some of the results of this study have been previously reported in the form of an abstract (8). In this population, current wheezing decreased during the follow-up time, whereas atopy increased (Table 1). A general decrease was observed in blood eosinophilia, FeNO, and R5 levels. A significant proportion of parents ceased smoking during the follow-up time (43%). Only one parent started smoking after enrollment. Recent asthma symptoms (during the last 2 months) were reported by 22 (34%) teenagers, and a recent need for any asthma medication (during the last 2 months) by 23 (36%). None of the teenagers reported smoking. Table 1. Demographics and Test Results of Study Patients (n = 64)

In this population, current wheezing decreased during the follow-up time, whereas atopy increased (Table 1). A general decrease was observed in blood eosinophilia, FeNO, and R5 levels. A significant proportion of parents ceased smoking during the follow-up time (43%). Only one parent started smoking after enrollment. Recent asthma symptoms (during the last 2 months) were reported by 22 (34%) teenagers, and a recent need for any asthma medication (during the last 2 months) by 23 (36%). None of the teenagers reported smoking. Table 1. Demographics and Test Results of Study Patients (n = 64) Patient Characteristics and Measurements Preschool Teenage Age, yr, mean (range) 5.62 (3–7) 14.22 (12–16) ISO-BMI, kg/m2, mean (SD) 22.92 (4.16) 23.1 (3.8) ISO-BMI > 25 kg/m2, n (%) 16 (25) 15 (23) Wheezing during previous yr, n (%)* 53 (83) 7 (11) Parental smoking, n (%)* 21 (33) 13 (20) Parental asthma, n (%)* 15 (24) 18 (28) Positive SPT, n (%) 45 (70) 59 (92) Urine cotinine, μg/L, mean (SD) 2.06 (3.56) — FeNOz-score, mean (SD) 2.54 (1.57) 2.08 (2.61) Abnormal FeNO (z-score ≥ 2 SD), n (%) 36 (60) 31 (49) Blood eosinophils, %, mean (SD) 6.92 (4.76) 5.16 (4.19) Blood eosinophilia (≥4%), n (%) 53 (83) 34 (54) FEV1z-score, mean (SD) — −0.45 (1.16) Abnormal FEV1 (z-score ≤ −1.645 SD), n (%) — 9 (14) R5, kPa/L/s, mean (SD) 0.89 (0.22) 0.36 (0.09) Upper quartile R5 (R5 < 0.451 kPa/L/s), n (%) — 12 (8) X5, kPa/L/s, mean (SD) −0.27 (0.10) −0.09 (0.03) Lower quartile X5 (X5 > −0.121 kPa/L/s), n (%) — 7 (5) Methacholine challenge, μg, median (IQR) — 1,709 (283–2,600) AHR to methacholine (PD20FEV1 ≤ 400 μg), n (%) — 17 (27) Definition of abbreviations: AHR = airway hyperresponsiveness; FeNO = fractional exhaled nitric oxide; ISO-BMI = sex- and age-adjusted body mass index for children; IQR = interquartile range; PD20FEV1 = provocative dose causing 20% decrease in FEV1; R5 = respiratory resistance at 5 Hz; SPT = skin prick test; X5 = respiratory reactance at 5 Hz.

ns: AHR = airway hyperresponsiveness; FeNO = fractional exhaled nitric oxide; ISO-BMI = sex- and age-adjusted body mass index for children; IQR = interquartile range; PD20FEV1 = provocative dose causing 20% decrease in FEV1; R5 = respiratory resistance at 5 Hz; SPT = skin prick test; X5 = respiratory reactance at 5 Hz. * Self-reported. Preschool cotinine levels were not associated with recent symptoms or need for medication in adolescence. Children with high urinary cotinine levels had parents with asthma (P = 0.046) more often than those with low levels. However, we observed that parental asthma was not connected with symptoms, need for medication, or lung function results. Overweight (body mass index > 25 kg/m2) in adolescence correlated with baseline FEV1/FVC (P = 0.025), R5–20 (P = 0.007), resonant frequency (P = 0.002), and AX (P = 0.014) at the same age, but not with preschool cotinine levels. Preschool urinary cotinine results correlated with FeNO z-scores (R = 0.227; P = 0.009), blood eosinophils (%) (R = 0.496; P < 0.001), baseline FEV1 z-scores (R = −0.219; P = 0.041), baseline R5 (kPa/L/s; R = 0.272; P = 0.015), baseline X5 (kPa/L/s; R = −0.297; P = 0.009), and baseline AX (kPa/L; R = 0.316; P = 0.012) in adolescence. Nevertheless, no correlation with methacholine responsiveness (PD20FEV1) or baseline FEV1/FVC was observed.

P < 0.001), baseline FEV1 z-scores (R = −0.219; P = 0.041), baseline R5 (kPa/L/s; R = 0.272; P = 0.015), baseline X5 (kPa/L/s; R = −0.297; P = 0.009), and baseline AX (kPa/L; R = 0.316; P = 0.012) in adolescence. Nevertheless, no correlation with methacholine responsiveness (PD20FEV1) or baseline FEV1/FVC was observed. In the dichotomous analysis of the data, a significant association with cotinine remained with FeNO, blood eosinophilia, increased baseline R5, and decreased baseline X5. A significant association was not observed with decreased baseline FEV1 or decreased PD20FEV1 (Figure 1). Only the children’s baseline X5 remained low (P = 0.020) if the parents had quit smoking before the follow-up visit compared with nonsmoking families. The adverse effects of maternal smoking only could be detected with impulse oscillometry indices, X5 (P = 0.003) and AX (P = 0.048), as well as with blood eosinophilia (P = 0.045). Other measures showed a similar pattern. Figure 1. Preschool urinary cotinine levels compared with multiple pulmonary function tests and blood eosinophilia in teenage years. AHR = airway hyperresponsiveness measured with methacholine challenge; FeNO = fractional exhaled nitric oxide; R5 = respiratory resistance at 5 Hz; X5 = respiratory reactance at 5 Hz.

In the dichotomous analysis of the data, a significant association with cotinine remained with FeNO, blood eosinophilia, increased baseline R5, and decreased baseline X5. A significant association was not observed with decreased baseline FEV1 or decreased PD20FEV1 (Figure 1). Only the children’s baseline X5 remained low (P = 0.020) if the parents had quit smoking before the follow-up visit compared with nonsmoking families. The adverse effects of maternal smoking only could be detected with impulse oscillometry indices, X5 (P = 0.003) and AX (P = 0.048), as well as with blood eosinophilia (P = 0.045). Other measures showed a similar pattern. Figure 1. Preschool urinary cotinine levels compared with multiple pulmonary function tests and blood eosinophilia in teenage years. AHR = airway hyperresponsiveness measured with methacholine challenge; FeNO = fractional exhaled nitric oxide; R5 = respiratory resistance at 5 Hz; X5 = respiratory reactance at 5 Hz. The influence of overweight and preschool ETS on lung function and inflammation markers were significant but essentially different. In our study, children exposed to ETS were generally not overweight, and thus the mechanisms of lung function deficits are probably distinct. Both factors are known to be associated with lower economic status. As a result of questionnaire limitations, the confounding effect cannot be excluded.

ut essentially different. In our study, children exposed to ETS were generally not overweight, and thus the mechanisms of lung function deficits are probably distinct. Both factors are known to be associated with lower economic status. As a result of questionnaire limitations, the confounding effect cannot be excluded. We quantified the child’s exposure to ETS at preschool age and compared it with objective measures of lung function in adolescence. Although the small sample size restricted the power of the analyses, we found long-lasting multifactorial effects of ETS on children with asthma. It is notable that the defects measurable with impulse oscillometry suggest small airway dysfunction similar to chronic obstructive pulmonary disease (9) and persisted even when parents quit smoking during the follow-up. It is possible that the actual airway remodeling might occur in the child’s epigenetic environment (10). In conclusion, early-life exposure to tobacco smoke causes chronic airway inflammation and defects in lung function in children with asthma. These defects are measurable even a decade later as changes in FeNO, blood eosinophil levels, impulse oscillometry, and spirometry. Smoking cessation is good; prevention is even better. Supported by the Finnish Cultural Foundation, Väinö and Laina Kivi Foundation, Pediatric Research Foundation, Sigrid Juselius Foundation, and Helsinki University Central Hospital Research Funds.

In conclusion, early-life exposure to tobacco smoke causes chronic airway inflammation and defects in lung function in children with asthma. These defects are measurable even a decade later as changes in FeNO, blood eosinophil levels, impulse oscillometry, and spirometry. Smoking cessation is good; prevention is even better. Supported by the Finnish Cultural Foundation, Väinö and Laina Kivi Foundation, Pediatric Research Foundation, Sigrid Juselius Foundation, and Helsinki University Central Hospital Research Funds. Author Contributions: A.S.P. and M.J.M. designed the study; S.K. was responsible of the acquisition of data; L.P.M. supervised and interpreted all pulmonary testing; K.L. analyzed the data and drafted the manuscript; A.K.-S., L.P.M., A.S.P., and M.J.M. revised the writing and checked the accuracy of statistical analyses; and all authors have read the work and approved the submission to this journal. Originally Published in Press as DOI: 10.1164/rccm.201809-1729LE on December 18, 2018 Author disclosures are available with the text of this letter at www.atsjournals.org.

To the Editor: Yoon and colleagues (1) recently showed that using a C-reactive protein point-of-care test (CRP-POC) as the initial screening tool compared with using the World Health Organization symptom screener could lead to a substantial reduction in the total number of Xpert tests used to diagnose tuberculosis (TB) in a population with only a slight loss in the number of cases identified. Using two sputum samples as the “gold standard” for a pulmonary TB diagnosis, they found that there were seven false-positive cases using the conventional approach (World Health Organization screen followed by Xpert) and four using a CRP-POC screen followed by Xpert.

nly a slight loss in the number of cases identified. Using two sputum samples as the “gold standard” for a pulmonary TB diagnosis, they found that there were seven false-positive cases using the conventional approach (World Health Organization screen followed by Xpert) and four using a CRP-POC screen followed by Xpert. Although not a focus of their article, their results (Table E2 in their online supplement) illustrate an important problem with using TB sputum culture as a gold standard. A single culture had a sensitivity of only 79% (160 detected) out of the 203 individuals identified by two cultures as positive TB cases. As there is no reason to expect that the sensitivity of a second sputum sample would be better than the first sample, these results imply that some true TB cases will be missed even using two samples. In fact, if the sensitivity of each test were 79%, one would expect that 4.41% (21% × 21%) of true-positive cases would be missed using two cultures, equal to about 9 [203/(1 − 0.0441) − 203] additional cases in this study. Taking account of these additional cases of pulmonary TB, however, reduces the estimated sensitivity to 75% (160 detected out of 212 [203 identified and 9 missed] cases). Thus, an even higher fraction of pulmonary TB cases would be missed by the gold standard of two sputum cultures. Continuing the calculations above, the sensitivity is about 73% (160 detected out of 219 [203 detected and 16 missed] cases), implying that about 7.29% (27% × 27%) of cases are missed, which is consistent with 16 missed cases among 219 true TB cases (16/219 = 7.31%).

missed by the gold standard of two sputum cultures. Continuing the calculations above, the sensitivity is about 73% (160 detected out of 219 [203 detected and 16 missed] cases), implying that about 7.29% (27% × 27%) of cases are missed, which is consistent with 16 missed cases among 219 true TB cases (16/219 = 7.31%). Moreover, this assumes that the second test, performed in a population for which the first test was negative, has the same sensitivity as the first test. However, results for Xpert (2) suggest that this is an overly optimistic assumption. For example, for sputum-negative, culture-positive specimens, sensitivity for a single sample is reported as 72.5%, whereas for three samples it is only 90.2% (estimated 95% confidence interval, 84.9–93.8%). If sensitivity for each separate test was 72.5%, however, then the expected sensitivity for three specimens would be 97.9% (1 − 0.2753). Importantly, if the sensitivity of a single Xpert test is about 60% (as reported in Table E2A of Yoon’s paper), then one would expect Xpert to diagnose about 60% of the 16 cases missed by the gold standard, or about 10 individuals. Thus, it is possible that all 7 individuals positive by Xpert had pulmonary TB that was missed by the two sputum cultures, and incorrectly considered false positives. It seems advisable that such patients be followed clinically to determine whether they actually had TB (a true positive) or whether the clinical course suggests that they did not have TB (a false positive).

by Xpert had pulmonary TB that was missed by the two sputum cultures, and incorrectly considered false positives. It seems advisable that such patients be followed clinically to determine whether they actually had TB (a true positive) or whether the clinical course suggests that they did not have TB (a false positive). In the absence of such follow-up, it might be appropriate to consider any diagnostic test positive for pulmonary TB as diagnostic for the presence of TB, whether confirmed by the gold standard or not, especially if the false-positive rate of the diagnostic test is low. This publication was made possible with help from the Harvard University Center for AIDS Research (CFAR), an NIH-funded program (P30 AI060354), which is supported by the following NIH Co-Funding and Participating Institutes and Centers: National Institute of Allergy and Infectious Diseases, National Cancer Institute, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NHLBI, National Institute on Drug Abuse, National Institute of Mental Health, National Institute on Aging, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of General Medical Sciences, National Institute on Minority Health and Health Disparities, National Institute of Dental and Craniofacial Research, Office of AIDS Research, and Fogarty International Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. Originally Published in Press as DOI: 10.1164/rccm.201811-2112LE on December 11, 2018

This publication was made possible with help from the Harvard University Center for AIDS Research (CFAR), an NIH-funded program (P30 AI060354), which is supported by the following NIH Co-Funding and Participating Institutes and Centers: National Institute of Allergy and Infectious Diseases, National Cancer Institute, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NHLBI, National Institute on Drug Abuse, National Institute of Mental Health, National Institute on Aging, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of General Medical Sciences, National Institute on Minority Health and Health Disparities, National Institute of Dental and Craniofacial Research, Office of AIDS Research, and Fogarty International Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. Originally Published in Press as DOI: 10.1164/rccm.201811-2112LE on December 11, 2018 Author disclosures are available with the text of this letter at www.atsjournals.org.

From the Authors: We welcome the interest shown by Dr. Parker in our recent publication that described the yield and efficiency of novel intensified tuberculosis (TB) case-finding algorithms for people living with HIV (1). We agree that sputum mycobacterial culture has imperfect sensitivity for active pulmonary TB. Nonetheless, culture is the best available microbiological reference standard for evaluation of novel TB diagnostics. The addition of clinical follow-up can be helpful but also has limitations. For example, in the context of our study, initiation of antiretroviral therapy complicates assessment of whether clinical and radiological improvements are a result of TB treatment. Furthermore, in most TB-endemic areas, culture is not routinely available for TB diagnosis, and Xpert MTB/RIF (Xpert) is used as the confirmatory TB test. In such settings, all patients with a positive Xpert result are in fact regarded as a TB case and initiated on TB treatment.

provements are a result of TB treatment. Furthermore, in most TB-endemic areas, culture is not routinely available for TB diagnosis, and Xpert MTB/RIF (Xpert) is used as the confirmatory TB test. In such settings, all patients with a positive Xpert result are in fact regarded as a TB case and initiated on TB treatment. If we consider all Xpert-positive patients to be true positives, as suggested by Dr. Parker and as occurs in routine clinical practice, our study conclusions would remain unchanged: 1) CRP (C-reactive protein)-based TB screening followed by Xpert confirmatory testing would identify a similar proportion of TB cases as the current intensified case-finding (ICF) algorithm (54% [95% confidence interval (CI), 45–63] vs. 58% [95% CI, 49–66]; difference in yield 4% [95% CI, −9% to +16%]; P = 0.57); 2) CRP-based ICF would use less than half as many Xpert assays (9 vs. 4); and 3) the addition of a single culture would substantially increase ICF yield, detecting ≥77% of all TB cases, regardless of the screening strategy used. Thus, we believe HIV programs should first consider implementation of CRP-based TB screening, which may then enable the routine use of more sensitive confirmatory tests, such as culture, to improve ICF yield.

ld substantially increase ICF yield, detecting ≥77% of all TB cases, regardless of the screening strategy used. Thus, we believe HIV programs should first consider implementation of CRP-based TB screening, which may then enable the routine use of more sensitive confirmatory tests, such as culture, to improve ICF yield. Supported by grants from the National Institute of Allergy and Infectious Diseases/NIH (K23 AI114363 to C.Y.); NIH and University of California, San Francisco–Gladstone Institute of Virology and Immunology Center for AIDS Research (P30 AI027763 to C.Y.); the Nina Ireland Program for Lung Health (C.Y.); National Institute of Allergy and Infectious Diseases/NIH Presidential Emergency Plan for AIDS Relief, Center for AIDS Research Administrative Supplement (P30 A120163 to A.C.). The funding organizations had no role in the study design; collection, analysis, and interpretation of data; or writing of the manuscript. Originally Published in Press as DOI: 10.1164/rccm.201811-2218LE on December 11, 2018 Author disclosures are available with the text of this letter at www.atsjournals.org.

Searching unceasingly throughout the body, antibodies pursue their targets like relentless wraiths. Antibodies of the IgG subclasses can initiate target lysis either directly by activating the classical complement cascade or by alerting cellular effectors such as natural killer cells via activating Fcγ receptors to inflict the lethal hit by antibody-dependent cellular cytotoxicity (1). The Fc portions of IgG and IgM induce myeloid cell phagocytosis, and the Fc fragment of IgE launches mast cell degranulation. Distributing throughout total body water, antibodies ensure that the immune system is constantly and everywhere vigilant. Antibody class switching relates directly to chronic obstructive pulmonary disease (COPD) pathogenesis. Ever since Cosio and Guerassimov proposed an autoimmune etiology of COPD (2), and lung lymphoid follicles (LLFs) (3) and elastin-specific antibodies (4) were demonstrated in advanced emphysema, the question of how autoantibodies might contribute to COPD progression has engendered intense investigation (5). Indeed, unbiased analyses of gene expression strongly link lung B cells to emphysema (6, 7).

d lung lymphoid follicles (LLFs) (3) and elastin-specific antibodies (4) were demonstrated in advanced emphysema, the question of how autoantibodies might contribute to COPD progression has engendered intense investigation (5). Indeed, unbiased analyses of gene expression strongly link lung B cells to emphysema (6, 7). However, humoral immunity includes a gentler component, secretory immunoglobulin A (sIgA), which is crucial to maintain mucosal barriers against bacteria transgression (8) and, when focally absent, is also intimately involved in COPD pathology (9). sIgA possesses two superpowers: it promotes immune exclusion by chaining respiratory microbes to mucus, and it neutralizes proinflammatory factors such as LPS, typically without inducing inflammation. sIgA activates neither the classical complement cascade nor phagocytes, with the exception of eosinophils (reviewed in Reference 10), via its several receptors (11). sIgA’s importance is illustrated by the resources expended on its production: ∼3 g daily, mostly excreted into the gut to maintain symbiosis with commensal bacteria (12).

either the classical complement cascade nor phagocytes, with the exception of eosinophils (reviewed in Reference 10), via its several receptors (11). sIgA’s importance is illustrated by the resources expended on its production: ∼3 g daily, mostly excreted into the gut to maintain symbiosis with commensal bacteria (12). Previous key observations about IgA in lung host defense and pathology were made by the group at the Université Catholique de Louvain (13–15). It is only fitting that Ladjemi and colleagues (pp. 592–602) contribute another in this issue of the Journal (16). Using lung tissues removed for clinical indications (subjects with COPD, n = 37; control subjects, n = 34) plus murine models of chronic Pseudomonas aeruginosa and of smoking, they assessed Ig class expression by B cells in LLFs in COPD and during chronic lung infection. The study has several technical strengths, including rigorous quantification of immunohistochemical staining results using color deconvolution and a melting-curve analysis of the PCR reactions that independently confirmed IgA production.

Ig class expression by B cells in LLFs in COPD and during chronic lung infection. The study has several technical strengths, including rigorous quantification of immunohistochemical staining results using color deconvolution and a melting-curve analysis of the PCR reactions that independently confirmed IgA production. There are multiple novel and interesting results. The first is that IgA+ B cell numbers were increased in LLFs in distal lung parenchyma in subjects with COPD relative to smokers without COPD, and correlated with spirometrically defined severity (16). That was not true in proximal airways, which do not depend on sIgA transcytosis, extending previous studies (3, 9). IgG+ B cells were not similarly increased, a crucial finding that is considered further below. Interestingly, LLF IgA+ B cells were also increased in their murine models by infection, but not by cigarette smoke exposure. The survival of human peripheral blood B cells in vitro was unexpectedly prolonged by cigarette smoke extract, but not by LPS—a finding that merits mechanistic investigation in future studies.

. Interestingly, LLF IgA+ B cells were also increased in their murine models by infection, but not by cigarette smoke exposure. The survival of human peripheral blood B cells in vitro was unexpectedly prolonged by cigarette smoke extract, but not by LPS—a finding that merits mechanistic investigation in future studies. The central results provide clues to the control mechanisms within LLFs of Ig class switching, the quintessential example of T-cell help. In lymph node germinal centers, Ig class switching depends largely on a specialized CD4+ T-cell subset, T follicular helper (Tfh) cells. This independent lineage is identified by expression of the transcription factor B cell lymphoma 6, which the authors examined. LPS can also induce human IgM+ memory B cells to switch directly to IgA secretion, an intriguing possibility given the observation by Ladjemi and colleagues that most LLF B cells (70–80%) were IgM+. Nevertheless, another key finding is the expression of IL-21 within LLFs in COPD by T cells, including IL-17–secreting T (T17), but not Tfh, cells. These results support a seminal murine study that showed that LLF development depends on T17 cells and CD11bhigh conventional dendritic cells, unlike the formation of lymph nodes, which requires lymphoid inducers (17). Along with the relative paucity of follicular dendritic cells in LLFs, these findings provide novel insights into the rules governing LLF formation in COPD.

d that LLF development depends on T17 cells and CD11bhigh conventional dendritic cells, unlike the formation of lymph nodes, which requires lymphoid inducers (17). Along with the relative paucity of follicular dendritic cells in LLFs, these findings provide novel insights into the rules governing LLF formation in COPD. IL-21 is a four-α-helical bundle cytokine that signals via the common receptor γ chain, as do IL-2, -4, -7, -9, and -15 (18). IL-21 promotes B-cell maturation outside the bone marrow. It drives division of naive human B cells, accelerates Ig affinity maturation and differentiation into plasma cells, and, with CD40L, increases IL-10 secretion by class-switched memory B cells (19). Without appropriate costimulation, however, B cells exposed to IL-21 undergo apoptosis, a check on bystander activation. Similarly, in the absence of granulocyte-macrophage colony-stimulating factor, IL-21 induces apoptosis of conventional dendritic cells, as another means to maintain self-tolerance (20). IL-21 has opposite effects on two types of T regulatory cells (TReg), favoring expansion of T effectors over Foxp3+ TReg (21) while supporting the differentiation of Foxp3− IL-10–producing Tr1 cells (22). Thus, IL-21’s actions are complex, and although it has been reported to be overproduced in several autoimmune diseases (18), its ultimate role in COPD pathogenesis requires further study.

voring expansion of T effectors over Foxp3+ TReg (21) while supporting the differentiation of Foxp3− IL-10–producing Tr1 cells (22). Thus, IL-21’s actions are complex, and although it has been reported to be overproduced in several autoimmune diseases (18), its ultimate role in COPD pathogenesis requires further study. Because LLFs are not unique to COPD, as the authors point out, this study has broader importance. LLFs also occur in cystic fibrosis and bronchiectasis, which are clearly linked with chronic bacterial overgrowth, but also in idiopathic pulmonary fibrosis, pulmonary hypertension, and lung cancer, which are generally not considered to be. Hence, understanding LLFs could help explain how adaptive immunity is involved in a wide range of lung diseases. In other organs, lymphoid neogenesis (the more general term for such ectopic lymphoid tissue) is implicated as an antigen-driven process associated with autoimmunity (23). Whether the same is true during the entire decades-long evolution of heterogeneous conditions such as COPD remains an unsettled question.

diseases. In other organs, lymphoid neogenesis (the more general term for such ectopic lymphoid tissue) is implicated as an antigen-driven process associated with autoimmunity (23). Whether the same is true during the entire decades-long evolution of heterogeneous conditions such as COPD remains an unsettled question. Regarding the source of the antigens that drive IgA production in COPD, Ladjemi and colleagues suggest both pathogens and altered self. This prudently impartial hedge acknowledges the current limits of our understanding. Still, IgA’s chiefly noninflammatory properties suggest that regardless of the stimulus, the B cells that make it in LLFs in COPD are unlikely to contribute to tissue destruction. IgA is not entirely devoid of pathological potential, as shown by IgA nephropathy, the most common glomerular disease outside of sub-Saharan Africa (24), and its involvement in several uncommon forms of bullous skin disease (25). At least in the kidney, IgA appears to be capable of activating complement via the lectin pathway. However, with these exceptions, IgA antibodies are not implicated in autoimmunity. Hence, it is significant that Ladjemi and colleagues found so few IgG-secreting B cells in COPD. A final implication of this study is that the answer to the question of whether LLF B cells in COPD are bad or beneficial (26) is that many appear to be trying to help. Such a wealth of insights from catching B cells in the act.

Idiopathic pulmonary fibrosis (IPF) is a chronic lung disease of unknown origin characterized by progressive scarring of lung parenchyma due to aberrant activation of stromal cells producing excessive amounts of extracellular matrix. Growing evidence suggests that injury to the alveolar epithelium may represent a primary event that initiates and promotes fibrosis in IPF (1). In advanced disease, alveolar derangement is often accompanied by bronchiolization, with ectopic emergence of mucociliary epithelium in the dilated alveolar spaces, known as “honeycomb cysts” (2). Specific mechanisms underlying these structural changes are unknown, and no currently available therapies halt disease progression. The prognosis for patients with IPF remains poor, with a median survival of 2–4 years after diagnosis, comparable to aggressive types of cancer (3). Whereas some progress has been made in identifying systemic biomarkers of IPF (4), no biomarkers reflecting potentially targetable mechanistic aspects of alveolar derangement in IPF that correlate with disease aggressiveness have so far been described.

ears after diagnosis, comparable to aggressive types of cancer (3). Whereas some progress has been made in identifying systemic biomarkers of IPF (4), no biomarkers reflecting potentially targetable mechanistic aspects of alveolar derangement in IPF that correlate with disease aggressiveness have so far been described. In this issue of the Journal, Prasse and colleagues (pp. 622–630) address this problem by transcriptional profiling of samples obtained by BAL of 212 patients with IPF from three independent cohorts (Freiburg, Siena, and Leuven) and correlating the obtained gene expression data with patients’ survival (5). BAL is a minimally invasive procedure that allows access to cells from the most distal regions of the respiratory tree, including the alveoli, the epicenter of IPF pathology. Because predominant cell types obtained by BAL are macrophages (normally >80% of BAL cells) followed by lymphocytes and other leukocytes, BAL has largely been used to evaluate the immune microenvironment in IPF lungs (6). In contrast to this approach, Prasse and colleagues (5) hypothesized that BAL samples from patients with IPF may harbor prognostically relevant information about structural changes in the alveoli, potentially concealed in the cumulative gene expression reflecting the status of diverse, including nonimmune, cells recovered by BAL.

to this approach, Prasse and colleagues (5) hypothesized that BAL samples from patients with IPF may harbor prognostically relevant information about structural changes in the alveoli, potentially concealed in the cumulative gene expression reflecting the status of diverse, including nonimmune, cells recovered by BAL. By following this logic, the authors first identified 1,582 genes whose expression in BAL samples from patients with IPF correlated with shorter survival. Using biostatistical methods, this gene set was reduced to fewer than 10 genes, which, when combined with the Gender-Age-Physiology (GAP) index, known to predict IPF mortality on the basis of basic patient characteristics (7), estimated poor IPF outcomes more robustly than the GAP index alone. Notably, one-tenth of identified 1,582 IPF mortality-related genes were previously found upregulated in airway basal cells (ABCs) cultured from human airway epithelial samples obtained by bronchoscopy (8). Accordingly, cells expressing KRT5 (keratin 5), an ABC marker, and “related” protein KRT6 were detected in BAL samples from patients with IPF, but not healthy individuals, or patients with sarcoidosis or chronic obstructive pulmonary disease. Thus, the presence of ABCs or ABC-like cells in BAL samples may represent a novel, prognostically relevant feature of IPF.

), an ABC marker, and “related” protein KRT6 were detected in BAL samples from patients with IPF, but not healthy individuals, or patients with sarcoidosis or chronic obstructive pulmonary disease. Thus, the presence of ABCs or ABC-like cells in BAL samples may represent a novel, prognostically relevant feature of IPF. ABCs are stem cells of the airway epithelium, normally absent in the alveolar region that is maintained by its local progenitors, that is, type II alveolar epithelial cells, which in IPF become the primary target of injury (1). In earlier studies, cells having ABC features have been found in the remodeled alveolar epithelium of IPF lungs (9, 10). Aberrant reepithelialization of the damaged alveolar epithelium by epithelial progenitors mobilized from adjacent bronchioles may lead to bronchiolization (9, 11), a characteristic feature of alveolar remodeling in IPF. In mice, KRT5/6-expressing ABC-like cells emerge in the lung parenchyma after severe injury caused by influenza virus infection (12), and their long-term persistence results in cysts with histologic features of bronchiolization (13), resembling IPF honeycomb lesions. Because the latter are most frequently found in patients with late-stage IPF, it is logical that the emergence of ABC-like cells in BAL samples, if it reflects alveolar bronchiolization driven by these cells as a stereotypic response to severe alveolar injury (14), correlates with shorter survival of patients.

lesions. Because the latter are most frequently found in patients with late-stage IPF, it is logical that the emergence of ABC-like cells in BAL samples, if it reflects alveolar bronchiolization driven by these cells as a stereotypic response to severe alveolar injury (14), correlates with shorter survival of patients. The fact that an ABC signature can be detected in BAL samples from subjects with IPF with poorer prognosis implies a possibility that mobilization of these cells to sites of alveolar injury, if the above theory is correct, demarcates the transition to a more aggressive disease. Recruitment of ABCs from the bronchioles to injured alveoli and capturing these cells by BAL would require disassembly of hemidesmosomes, which normally keep ABCs firmly attached to the basement membrane (9). A similar process occurs in skin basal cells, when they migrate to cover the injured epidermis during the wound-healing process (15). Whereas in the skin such migration occurs within the same tissue compartment and leads to physiological repair, alveolar “colonization” by ABCs would lead to an aberrant, airway-like regeneration (11), making the resulting “alveolar” epithelium incapable of performing its respiratory function.

und-healing process (15). Whereas in the skin such migration occurs within the same tissue compartment and leads to physiological repair, alveolar “colonization” by ABCs would lead to an aberrant, airway-like regeneration (11), making the resulting “alveolar” epithelium incapable of performing its respiratory function. It should be noted, however, that many genes in the ABC signature detected by Prasse and colleagues (5) in the IPF BAL samples are not bona fide ABC markers; rather, they are molecular features of squamous metaplasia, an injury-associated histologic pattern, which can be produced by ABCs (16). Examples include calcium-binding protein S100A14, KRT6, stratifin, and neuregulin (16–18). The latter gene has been found earlier to be expressed in squamous cells in the remodeled alveolar epithelium in IPF (18). ABC-derived squamous metaplasia can promote a fibrotic response in subjacent fibroblasts (16), potentially relevant to IPF pathogenesis. Both disassembly of hemidesmosomes required for basal cell migration and squamous metaplasia are dependent on epidermal growth factor receptor signaling (15, 19), which could represent a candidate pathway of ABC-mediated alveolar remodeling in IPF.

n subjacent fibroblasts (16), potentially relevant to IPF pathogenesis. Both disassembly of hemidesmosomes required for basal cell migration and squamous metaplasia are dependent on epidermal growth factor receptor signaling (15, 19), which could represent a candidate pathway of ABC-mediated alveolar remodeling in IPF. Thus, the ABC signature in the IPF BAL samples observed by Prasse and colleagues (5) may represent an “echo” of regenerative crisis in the diseased lung, potentially mediated by ABCs or ABC-like cells mobilized in response to alveolar injury (Figure 1). Further studies, including those employing evaluation of BAL samples at single-cell resolution, as performed for IPF lung tissues (20), are needed to identify the cellular origin and targetable pathways of alveolar remodeling in IPF. The advantage of BAL as a discovery tool is that evaluation of cells sampled by this method from the primary site of IPF pathology can be performed within the lifetime of patients, so that the knowledge about patient-specific disease pathways learned using this approach can be translated into effective personalized therapies.

antage of BAL as a discovery tool is that evaluation of cells sampled by this method from the primary site of IPF pathology can be performed within the lifetime of patients, so that the knowledge about patient-specific disease pathways learned using this approach can be translated into effective personalized therapies. Figure 1. The normal bronchoalveolar region of the human lung (top left) is composed of terminal bronchioles lined by the distal airway epithelium, which continue into the respiratory alveolar region lined by type 1 and type 2 alveolar epithelial cells. The bronchoalveolar duct junction (BADJ) demarcates a boundary between the most distal conducting airways and alveoli. The airway epithelium is maintained by basal stem cells, normally attached to the basement membrane via hemidesmosome (HD) integrins. The alveolar epithelium is maintained by type 2 alveolar epithelial cells capable of self-renewal and generating type 1 cells. The latter form a gas exchange unit with the pulmonary capillaries in the alveolar interstitium. Alveolar macrophages are the resident immune cells located on the alveolar surface. BAL enables the sampling of cells from the bronchoalveolar region (top right). Under normal conditions, the major components of the BAL fluid are alveolar macrophages followed by other leukocytes (Leu). In idiopathic pulmonary fibrosis (IPF), the alveolar epithelium becomes damaged (bottom left) with preferential injury to and loss of type 2 cells and acquisition of remodeling phenotypes, including bronchiolization, that is, emergence of pseudostratified mucociliary epithelium, and squamous metaplasia, paralleled by fibrotic changes in the alveolar interstitium. It is possible that basal cells from adjacent bronchioles can migrate to areas of alveolar injury, colonize the damaged alveolar epithelium in IPF, and contribute to disease progression by generating some of these lesions. Consistent with this theory, Prasse and colleagues (5) have found that in patients with IPF with particularly short survival, the BAL fluid contains basal-like cells (bottom right), and upregulation of the gene expression program of these cells in BAL samples is indicative of IPF mortality. Thus, the emergence of airway basal or basal-like cells in the alveolar region in IPF may represent a biomarker of regenerative crisis in the lung that determines disease aggressiveness. ECM = extracellular matrix.

upregulation of the gene expression program of these cells in BAL samples is indicative of IPF mortality. Thus, the emergence of airway basal or basal-like cells in the alveolar region in IPF may represent a biomarker of regenerative crisis in the lung that determines disease aggressiveness. ECM = extracellular matrix. Supported by NIH grants R01HL123544 and R01HL127393. Originally Published in Press as DOI: 10.1164/rccm.201808-1557ED on September 5, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

After three decades, the coepidemic of HIV and tuberculosis remains a serious and challenging public health problem. In sub-Saharan Africa, undiagnosed tuberculosis remains an acute and common cause of HIV-related death (1). As a response to this, in 2008 the World Health Organization launched a three-pronged initiative that includes isoniazid preventive therapy, intensified case finding, and tuberculosis infection control partnered with a scale-up of antiretroviral therapy (2). Although improvements have been made in the last 10 years, in 2014, approximately 30 million of the 37 million people living with HIV globally were not screened for tuberculosis (3).

preventive therapy, intensified case finding, and tuberculosis infection control partnered with a scale-up of antiretroviral therapy (2). Although improvements have been made in the last 10 years, in 2014, approximately 30 million of the 37 million people living with HIV globally were not screened for tuberculosis (3). Reliable detection of tuberculosis in persons with HIV is a challenge in resource-constrained settings. In these areas, tuberculosis diagnosis relies heavily on sputum smear microscopy, chest radiography, and symptom screening. However, persons living with HIV often have reduced lung immunopathology and paucibacillary disease (4, 5). Studies of individuals initiating antiretroviral therapy in tuberculosis-endemic settings revealed that up to 15–30% had sputum culture–positive disease (6, 7); however, testing all individuals by sputum culture is resource intensive and may additionally lead to diagnostic delays. To address these challenges, the World Health Organization currently recommends that intensified case finding should include a preliminary symptom screen (i.e., weight loss, current cough, fever, and night sweats) followed by a confirmatory Xpert MTB/RIF test for individuals who screen positive. This algorithm may reliably exclude tuberculosis through the symptom screen (8), but at the programmatic level it is heavily resource intensive and difficult to implement fully (9).

., weight loss, current cough, fever, and night sweats) followed by a confirmatory Xpert MTB/RIF test for individuals who screen positive. This algorithm may reliably exclude tuberculosis through the symptom screen (8), but at the programmatic level it is heavily resource intensive and difficult to implement fully (9). New diagnostic methods for tuberculosis—preferably nonsputum-based, rapid, point-of-care tests—are urgently needed. Toward this end, the World Health Organization issued a target product profile for triage tests that recommends a minimum of 90% sensitivity and 70% specificity. Previous work by Yoon and colleagues demonstrated that point-of-care CRP (C-reactive protein) testing met these benchmarks in a cohort of individuals initiating antiretroviral therapy (10). Although much attention in the tuberculosis diagnostic field has been focused on novel “omics” approaches (11), including transcriptional, proteomic, and metabolic signatures, a simple, rapid test with equivalent or better accuracy is available now for $2. However, questions remain about how optimally to integrate CRP and other diagnostics into systematic screening algorithms for HIV-infected individuals.

omics” approaches (11), including transcriptional, proteomic, and metabolic signatures, a simple, rapid test with equivalent or better accuracy is available now for $2. However, questions remain about how optimally to integrate CRP and other diagnostics into systematic screening algorithms for HIV-infected individuals. In this issue of the Journal, Yoon and colleagues (pp. 643–650) report findings from a large, prospectively followed cohort of HIV-infected patients and contribute two important advances (12). First, they were able to evaluate and compare the accuracy of several novel diagnostic algorithms that include CRP, Determine TB-LAM, Xpert MTB/RIF, and culture with the current global guideline for intensified case finding. Second, the authors assessed costs associated with the use of these novel algorithms, which is critical when evaluating novel diagnostics in resource-constrained settings.

diagnostic algorithms that include CRP, Determine TB-LAM, Xpert MTB/RIF, and culture with the current global guideline for intensified case finding. Second, the authors assessed costs associated with the use of these novel algorithms, which is critical when evaluating novel diagnostics in resource-constrained settings. In this cohort, all participants were antiretroviral therapy naive and the majority had advanced HIV infection (median CD4 count, 153 cells/μl). All were screened by CRP, urine LAM, sputum Xpert, and sputum liquid culture, and the accuracy, yield, and cost per tuberculosis case detected were compared for various algorithms. This approach generated several important insights into screening in this population. First, although CRP has lower sensitivity than symptom screening (88% vs. 97%), its substantially higher specificity (69% vs. 13%) means that far fewer individuals will require confirmatory testing (40% vs. 88%). Other recent prospective studies of patients with severe HIV and low CD4 counts from Malawi, Ghana, Cameroon, and South Africa have similarly shown that >90% of the patients had a positive symptom screen (13–16). Although symptom screening is simple to implement and low-cost, its low specificity results in a large number of patients needing follow-up diagnostic testing. Findings from this study provide support for considering the replacement of symptom-based screening with CRP as a preliminary “screen-in” test for persons with HIV entering care. Point-of-care CRP testing has the advantages of being inexpensive, nonsputum based, objective, and relatively easy to implement in resource-constrained clinical settings.

ide support for considering the replacement of symptom-based screening with CRP as a preliminary “screen-in” test for persons with HIV entering care. Point-of-care CRP testing has the advantages of being inexpensive, nonsputum based, objective, and relatively easy to implement in resource-constrained clinical settings. A key challenge is that both algorithms—symptom screening and CRP testing, followed by sputum Xpert—have inadequate sensitivity, estimated in this study at 59% and 56%, respectively. Yoon and colleagues (12) demonstrate that the screening resources saved by using CRP tests rather than symptom screening could be used for confirmatory testing by TB-LAM (for those with CD4 < 100) and sputum culture in addition to Xpert. This approach improves the overall diagnostic yield to 78% while containing costs, resulting in a cost per tuberculosis case diagnosed of $92 (compared with $102 for symptom screening followed by Xpert). This algorithm may represent the best balance of yield and costs for clinics in countries with high HIV and tuberculosis burdens, achieving 92% of the yield of the highest-sensitivity algorithm (symptom screening followed by TB-LAM, Xpert, and culture) and better specificity at nearly half the cost per case diagnosed ($92 vs. $172).

may represent the best balance of yield and costs for clinics in countries with high HIV and tuberculosis burdens, achieving 92% of the yield of the highest-sensitivity algorithm (symptom screening followed by TB-LAM, Xpert, and culture) and better specificity at nearly half the cost per case diagnosed ($92 vs. $172). The past decade has seen major advances in tools for diagnosing tuberculosis in individuals with advanced HIV disease; however, major questions remain about how to effectively integrate these tools into pragmatic screening algorithms in high-burden settings. This study provides evidence that simple algorithms using these new diagnostics can improve case detection while controlling costs. Replication of these findings should be expeditiously conducted in other settings, and, if they are confirmed, global screening guidelines should be revised. Although there is a robust pipeline for new tuberculosis diagnostics, we should not wait to capitalize on the extraordinary progress in diagnostics that has been made over the past 10 years to decrease the 400,000 tuberculosis-related deaths among persons living with HIV every year (17). L.M. is supported by a Ruth L. Kirschstein National Research Service Award and NIH T32 Training Grant T32 AI 052073. Originally Published in Press as DOI: 10.1164/rccm.201809-1702ED on October 1, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

To the Editor: The ability to link data from sources such as the U.S. Census is now enabling researchers to direct their focus toward reporting neighborhood and contextual characteristics that increase the risk for adverse health outcomes and are independent of patient-level attributes. This is all the more important because disparities in health outcomes likely arise as a result of both individual exposures and contextual factors (1). Research regarding disparities have until recently been challenging because of the high response bias associated with collecting individual-level socioeconomic measures (2). However, area-based measures from the U.S. Census’s American Community Survey and the National Center for Health Statistics Urban-Rural Classification Scheme can be used to gain insight into the role of area-based measures as independent risk factors for diseases, as demonstrated in the work by Raju and colleagues (3). Understanding area-based risk factors could help researchers design, target, monitor, and assess public health programs, including prevention interventions. First, some limitations of Raju and colleagues’ analysis need to be emphasized. Although the authors used census tract–based determinants as area-based measures, it is important to acknowledge the possibility of ecological fallacy, and that these determinants provide information regarding the neighborhood that is not reducible to the individual level (4). Although the authors have defined neighborhoods as census tracts, nearby neighborhoods may also influence health outcomes and disparities.

t to acknowledge the possibility of ecological fallacy, and that these determinants provide information regarding the neighborhood that is not reducible to the individual level (4). Although the authors have defined neighborhoods as census tracts, nearby neighborhoods may also influence health outcomes and disparities. Second, data structures arising from both individual and neighborhood levels are inherently hierarchical and correlated. To account for geographical correlation, it is important to analyze such data using multilevel models. Multilevel models can account for a lack of independence, evaluate multivariate associations, incorporate covariates at both individual and geographic levels, and model interactions between variables (5). Multilevel models have been used to evaluate health disparities and to describe the relationship between geographic exposures for a wide variety of health outcomes. They can also help researchers quantify the proportion of variability associated with being in a specific neighborhood. The authors are to be applauded for taking the research on chronic obstructive pulmonary disease risk factors a step further by investigating area-based measures. Although the availability of public datasets such as that provided by the U.S. Census make such investigations possible, it must be emphasized that their implementation is challenging. Several resources, such as the Public Health Disparities Geocoding Project (6), are available to provide guidance on techniques to conduct research on socioeconomic gradients in health.

as that provided by the U.S. Census make such investigations possible, it must be emphasized that their implementation is challenging. Several resources, such as the Public Health Disparities Geocoding Project (6), are available to provide guidance on techniques to conduct research on socioeconomic gradients in health. Originally Published in Press as DOI: 10.1164/rccm.201811-2073LE on December 22, 2018 Author disclosures are available with the text of this letter at www.atsjournals.org.

From the Authors: We thank Dr. Chandrasekhar for her letter and comments in response to our recent publication (1). A major goal of our study was to understand the community-level characteristics, such as rural residence and poverty, that contribute to chronic obstructive pulmonary disease (COPD), in addition to describing neighborhoods that face the highest burden of disease. We additionally hope, as Dr. Chandrasekhar has highlighted, that our study brings attention to how researchers can link publicly available datasets to answer questions of significant clinical relevance and develop targeted public health interventions. Dr. Chandrasekhar has raised a number of methodological points that are worthy of consideration in performing such an analysis, regarding 1) the need to use multilevel modeling and 2) the risk of the ecological fallacy.

answer questions of significant clinical relevance and develop targeted public health interventions. Dr. Chandrasekhar has raised a number of methodological points that are worthy of consideration in performing such an analysis, regarding 1) the need to use multilevel modeling and 2) the risk of the ecological fallacy. Dr. Chandrasekhar makes note of the use of multilevel modeling to account for geographical correlation in data structures that take into account neighborhood- and individual-level factors. Although multilevel modeling is frequently used in analyses of geographic factors and disease, this approach requires a significant number of units within each group (census tract in our case) to ensure the stability of the fitting algorithms and to estimate SE. The literature recommends including at least 25 individuals per group for such an analysis, but there were on average only four subjects per census tract (2). As a result, we used standard survey methods in our study, with sample weights and strata provided in the National Health Interview Survey. This accounts for the complex survey design and adjusts the variances for clustering within the sampling unit.

s, but there were on average only four subjects per census tract (2). As a result, we used standard survey methods in our study, with sample weights and strata provided in the National Health Interview Survey. This accounts for the complex survey design and adjusts the variances for clustering within the sampling unit. Regarding the risk of the ecologic fallacy, we agree that this represents a potential limitation when census-level characteristics cannot be reduced to the individual level. Our primary goal, however, was to understand the geographic areas with the highest burden of COPD. We ultimately believe that future research should focus on studying COPD in poor, rural areas to better elucidate individual-level effects and characteristics. We thank Dr. Chandrasekhar for her thoughtful observations. She has raised a number of important points that are worth considering in future analyses that aim to describe the combined impact of neighborhood- and individual-level factors on disease. We hope that our findings pave the way for future investigations into the risk factors for COPD that contribute to health disparities among individuals residing in rural and poor communities. Supported by grants from the National Institute on Minority Health and Health Disparities/NIH (P50MD010431), National Institute of Environmental Health Sciences/NIH (F32 ES029786-01 and R21ES025840), and the Environmental Protection Agency (R836150). Originally Published in Press as DOI: 10.1164/rccm.201811-2185LE on December 22, 2018

Supported by grants from the National Institute on Minority Health and Health Disparities/NIH (P50MD010431), National Institute of Environmental Health Sciences/NIH (F32 ES029786-01 and R21ES025840), and the Environmental Protection Agency (R836150). Originally Published in Press as DOI: 10.1164/rccm.201811-2185LE on December 22, 2018 Author disclosures are available with the text of this letter at www.atsjournals.org.

From the Authors: In our recent research letter focusing on the acute effects of smoking on the serum levels of sRAGE (soluble receptor for advanced glycation end products), we showed that smoking three cigarettes within 1 hour significantly decreases serum sRAGE levels within 2 hours (1). In addition, we also determined the effect of chronic cigarette smoke exposure on serum sRAGE levels by comparing smokers with never smokers (originally reported as “data not shown”). Here, we did not find any difference in serum sRAGE levels, which is in line with previous studies (2). In contrast, as rightfully mentioned in the response to our research letter, Biswas and colleagues previously reported that serum sRAGE levels were increased in smokers compared with nonsmokers (3). Biswas explains the discrepancies between their study and other studies by noting that most of the studies were not specifically designed to explore the effect of smoking on sRAGE in healthy individuals, whereas their study was (4). Further, in his original paper, he states that the differences may also be explained by the fact that his study population was of overall younger age compared with those in the other studies (3). Indeed, characteristics of the study population may affect the outcomes of sRAGE measurements; however, other factors, including the method and timing of serum preparation, the method of sRAGE quantification, and, most importantly, the timing of the last smoked cigarette before blood sampling may also drive the observed differences in serum sRAGE levels. Although our initial research letter only showed data of serum sRAGE levels in healthy control subjects versus patients with chronic obstructive pulmonary disease (COPD) (1), our study was also designed to investigate the chronic effects of smoking in healthy individuals. Specifically, to investigate the effect of chronic smoke exposure on the serum levels of sRAGE, we used a well-controlled cohort (ClinicalTrials.gov Identifier: NCT00848406) of young (18–40 yr old) and old (>40 yr old) smokers and never smokers without airway obstruction (Figure 1A) (5).

smoking in healthy individuals. Specifically, to investigate the effect of chronic smoke exposure on the serum levels of sRAGE, we used a well-controlled cohort (ClinicalTrials.gov Identifier: NCT00848406) of young (18–40 yr old) and old (>40 yr old) smokers and never smokers without airway obstruction (Figure 1A) (5). To adequately measure serum sRAGE levels, we used the highly sensitive and selective simplified immunoprecipitation in 96-well ELISA format–coupled liquid chromatography–mass spectrometry assay, which we recently demonstrated to be superior to the commonly used sRAGE ELISA (6). When we focused on the healthy control subjects, we found that there were no differences in serum sRAGE levels between smokers and never smokers, whether old or young (Figure 1B). Of note, the definition of “nonsmokers” in our study and the one used by Biswas are not exactly the same. Whereas our nonsmokers had never smoked, the nonsmokers in Biswas’s study included subjects who had not smoked during the last 5 consecutive years. Moreover, the age of the study subjects does not influence the serum sRAGE levels. Indeed, our data show no differences in serum sRAGE levels between young (average 23.5 yr) and old (average 54.7 yr) subjects in either the never-smokers group or the smokers group (Figure 1B). In addition, our young subjects were even younger than the study population of Biswas, which had an average age of 34.1 years. Our data therefore indicate that neither age nor chronic smoke exposure affects serum sRAGE levels. The discrepancy in study results when comparing the serum sRAGE levels in smokers and nonsmokers may be explained by our finding that smoking before blood sampling acutely decreases serum sRAGE levels (1). Therefore, controlling or monitoring smoking behavior before blood sampling may be used as a precautionary measure to decrease the variability between measurements and increase the value of sRAGE as a biomarker for COPD. Lastly, we agree with Biswas that more research is needed regarding the effect of smoking on serum sRAGE levels and the underlying mechanisms before sRAGE can be clinically used as a biomarker for COPD.

ary measure to decrease the variability between measurements and increase the value of sRAGE as a biomarker for COPD. Lastly, we agree with Biswas that more research is needed regarding the effect of smoking on serum sRAGE levels and the underlying mechanisms before sRAGE can be clinically used as a biomarker for COPD. Figure 1. Serum sRAGE (soluble receptor for advanced glycation end products) levels in never smokers and smokers. (A) Patient characteristics. BMI = body mass index (kg/m2); N = number of group participants; sex (M) = number of males in a group. Data are shown as mean ± SEM. (B) Levels of sRAGE were measured in serum of young (18–40 yr old) smokers (n = 26) and nonsmokers (n = 28), and age-matched old (>40 yr old) smokers (n = 28) and nonsmokers (n = 28) without airway obstruction using immunoprecipitation in 96-well ELISA format–coupled liquid chromatography–mass spectrometry. Data are shown as individual measurements and mean ± SEM. Supported by the Netherlands Organization for Scientific Research NWO (Domain Applied and Engineering Sciences; Perspectief program P12-04; projects 13541 and 13544), Lung Foundation Netherlands (project 6.2.15.044JO), and Noordelijke CARA Stichting (project: 2016/01). Author Contributions: Conception and design of the study: S.D.P., F.K., M.K., P.H., R.B., and N.H.T.t.H. Acquisition of data: S.D.P., F.K., M.K., V.R.W., and A.F. Analysis of data: S.D.P., F.K., M.K., V.R.W., and A.F. Drafting of the manuscript: S.D.P. and F.K. Manuscript revision: S.D.P., F.K., M.K., V.R.W., A.F., M.v.d.B., P.H., R.B., and N.H.T.t.H.

and design of the study: S.D.P., F.K., M.K., P.H., R.B., and N.H.T.t.H. Acquisition of data: S.D.P., F.K., M.K., V.R.W., and A.F. Analysis of data: S.D.P., F.K., M.K., V.R.W., and A.F. Drafting of the manuscript: S.D.P. and F.K. Manuscript revision: S.D.P., F.K., M.K., V.R.W., A.F., M.v.d.B., P.H., R.B., and N.H.T.t.H. Originally Published in Press as DOI: 10.1164/rccm.201812-2257LE on December 27, 2018 Author disclosures are available with the text of this letter at www.atsjournals.org.

The authors of Mayer-Hamblett and colleagues (1), which was published in the November 1, 2018, issue of the Journal, have alerted us to an errant data point that was included in the article. The authors explained that the value for one of the placebo participants’ height at visit 8 (18 mo after randomization) was discovered to be too small when compared with additional data they obtained for this subject. The calculation for FEV1% predicted uses height; therefore, this participant’s FEV1% predicted was calculated incorrectly, giving a falsely high FEV1% predicted at the final study visit (18-mo visit). Therefore, the third sentence in Secondary Clinical Outcomes in the Results section (p. 1181) should be corrected to read as follows:No significant difference between treatment groups was seen with respect to the average 18-month change in FEV1% predicted among participants who were old enough to perform spirometry (n = 69 azithromycin and n = 63 placebo; 0.47% difference; 95% CI, −5.45 to 6.40; P = 0.863; Figure E6), or any other spirometry parameters (data not shown). In addition, corrected graphs have been included in the online supplement for this article (Figures E5 and E6). The authors state that the corrected data and calculations bring the estimate of the treatment effect closer to the null hypothesis and do not alter their original interpretation of the result of no significant difference between treatment groups with respect to lung function. They apologize for the error.

To the Editor: I read the article by Pouwels and colleagues with great interest (1). To the best of my knowledge, the authors for the first time have explored the acute effect of cigarette smoking on the serum levels of sRAGE (soluble receptor for advanced glycation end products). They elegantly showed a significant reduction of serum sRAGE levels in subjects who smoked three cigarettes within 1 hour. This effect was shown in patients with chronic obstructive pulmonary disease as well as in young and old healthy control subjects without airway obstruction. Based on a time-course study using three healthy subjects, the authors claimed that the maximum decline of serum sRAGE levels occurred after approximately 8 hours of cigarette smoking, which was not fully restored after 48 hours. In fact, the data presented in Figure 2B in Reference 1 demonstrate that the serum sRAGE values remained persistently low after 48 hours and were almost similar to the maximum decline values observed after 8 hours of cigarette smoking. The latter finding suggests that active smokers who regularly smoke several cigarettes per day should have lower serum levels of sRAGE than never smokers. However, Pouwels and colleagues did not observe any difference in sRAGE values between active smokers and never smokers (data not shown). To support this finding, the authors cited previous studies that also found no difference in sRAGE levels between smokers and nonsmokers, and stated that recent smoking within the smokers group may be the reason why some studies found decreased serum sRAGE levels in smokers (1). Unfortunately, Pouwels and colleagues did not cite our study in which we found elevated serum sRAGE levels in otherwise healthy, nondiabetic cigarette smokers (2).

nd nonsmokers, and stated that recent smoking within the smokers group may be the reason why some studies found decreased serum sRAGE levels in smokers (1). Unfortunately, Pouwels and colleagues did not cite our study in which we found elevated serum sRAGE levels in otherwise healthy, nondiabetic cigarette smokers (2). Cigarette smoke is known to increase the formation of AGEs and the expression of RAGE (3, 4). However, the effect of cigarette smoking on sRAGE is inconsistent across the literature. Decreased, elevated, and unchanged levels of sRAGE were found in different studies, as reviewed by Prasad and colleagues (5). However, most of those studies, as I explained previously (6), were not specifically designed to explore the effect of smoking on sRAGE and thus were confounded by the presence of other diseases or conditions that affect sRAGE levels. Therefore, in our study, we specifically aimed to compare sRAGE levels between cigarette smokers and nonsmokers, controlling for the majority of confounding variables (2). In that study, we showed for the first time a significant elevation of sRAGE in cigarette smokers, a strong correlation between sRAGE and the number of cigarettes smoked per day, and an independent association of sRAGE with smoking habit (2). Although the exact mechanism of this apparently surprising finding is not yet known, we proposed a number of scientifically valid explanations (2, 6). Now, Pouwels and colleagues have identified the acute effect of cigarette smoking on sRAGE, which is the opposite of the chronic effect of smoking previously identified by our group (2). Therefore, further studies are required to explore the true effect of cigarette smoking on serum sRAGE levels and to explain the discrepancy among these studies. These issues need to be resolved before we can consider sRAGE as a biomarker for inflammatory conditions or as a protective factor against AGEs and other RAGE ligands.

, further studies are required to explore the true effect of cigarette smoking on serum sRAGE levels and to explain the discrepancy among these studies. These issues need to be resolved before we can consider sRAGE as a biomarker for inflammatory conditions or as a protective factor against AGEs and other RAGE ligands. Originally Published in Press as DOI: 10.1164/rccm.201811-2169LE on December 27, 2018 Author disclosures are available with the text of this letter at www.atsjournals.org.

When Robert Koch reported his discovery of the tubercle bacillus in 1882 in a lecture and in the scientific paper he published just a few weeks later in Berliner Medicinische Wochenschrift, he described the staining techniques that allowed him to see the rod-shaped bacteria that he had successfully isolated and grown in pure culture (1). Paul Erlich had attended Koch’s lecture and quickly refined the staining technique, making it easier and quicker. Shortly thereafter, Ziehl and Neelsen further modified the technique and developed the method basically still used today. By 1883, Koch recognized that the development of a relatively simple and rapid staining method had important implications for patient care. He wrote, “It was soon found that with Ehrlich’s method of staining, the recognition of tubercle bacilli could readily be made use of in diagnosis. We owe it to this circumstance alone that it has become a general custom to search for the bacilli in the sputum” (2).

g method had important implications for patient care. He wrote, “It was soon found that with Ehrlich’s method of staining, the recognition of tubercle bacilli could readily be made use of in diagnosis. We owe it to this circumstance alone that it has become a general custom to search for the bacilli in the sputum” (2). The acid-fast bacilli (AFB) smear remains the main mode of diagnosis of tuberculosis in most of the places in the world where tuberculosis is common. If tuberculosis were something like beer brewing, or cheesemaking, this kind of artisanal approach to diagnosis might seem authentic and appealing. But tuberculosis is not beer brewing or cheesemaking, and the persistence of a 19th-century technique for diagnosing the world’s leading cause of death resulting from a single infectious agent in the 21st century is a disgrace. By now, it is well-appreciated that smears detect only about half of all cases of culture-positive tuberculosis, and quality control is notoriously difficult, especially in places where it is relied on most heavily (3, 4). The article in this issue of the Journal by Lee and colleagues (pp. 784–794) provides further evidence that it is time for the AFB smear to find its place in medical museums and history books, rather than in modern diagnostic labs (5).

toriously difficult, especially in places where it is relied on most heavily (3, 4). The article in this issue of the Journal by Lee and colleagues (pp. 784–794) provides further evidence that it is time for the AFB smear to find its place in medical museums and history books, rather than in modern diagnostic labs (5). Sputum samples were collected from each of nearly 3,000 consecutive patients being evaluated for possible tuberculosis. One aliquot was analyzed using semiquantitative nucleic acid amplification with GeneXpert MTB/RIF, and one was analyzed by conventional AFB smear microscopy and culture. Culture results were considered the gold standard for a diagnosis of tuberculosis.

000 consecutive patients being evaluated for possible tuberculosis. One aliquot was analyzed using semiquantitative nucleic acid amplification with GeneXpert MTB/RIF, and one was analyzed by conventional AFB smear microscopy and culture. Culture results were considered the gold standard for a diagnosis of tuberculosis. The results were clear and convincing. Overall, 8.9% of patients provided samples that were culture positive for Mycobacterium tuberculosis. Of those, 102 had AFB smear-positive sputum and 161 were smear negative. In addition, another 9% (265) of patients were culture positive for nontuberculous mycobacteria, and 82 of those patients were AFB smear positive. Overall, then, the sensitivity of AFB smear was 38.8%, and the specificity was 96.7%. This compares with a sensitivity of 74.1% and a specificity of 97.5% for Xpert. Notably, AFB smear sensitivity varied by time of collection (morning samples had greater yield than spot samples), but this was not true for Xpert. Results from Xpert were reported back to clinicians on average about 16 hours faster than results from AFB smears. Thus, Xpert results overall were more accurate, available more quickly, and less affected by several operational issues than were AFB smears.

ater yield than spot samples), but this was not true for Xpert. Results from Xpert were reported back to clinicians on average about 16 hours faster than results from AFB smears. Thus, Xpert results overall were more accurate, available more quickly, and less affected by several operational issues than were AFB smears. This article amplifies results of many earlier, smaller, or laboratory-based studies that showed the promise of nucleic acid amplification-based tuberculosis diagnostics (6–9). Indeed, uptake of Xpert has been advancing around the world in both resource-rich and resource-limited settings and in countries with both high and low burdens of tuberculosis (10). Technological advances that will make it easier to use this test at the point of care will likely accelerate this trend. Still, there has been reluctance and even opposition to making this test the standard initial means of diagnosis for suspected pulmonary tuberculosis (11). Objections have been raised that the test is too costly, requires too much maintenance, does not provide information regarding infectiousness, and does not allow a clinician to assess response to therapy in the way that a decreasing AFB smear grade does. In addition, an early paper noted that introduction of Xpert in South Africa had not resulted in a decrease in TB mortality in the communities in which it was being used, although there was realization that this was mostly a systems issue (12).

sess response to therapy in the way that a decreasing AFB smear grade does. In addition, an early paper noted that introduction of Xpert in South Africa had not resulted in a decrease in TB mortality in the communities in which it was being used, although there was realization that this was mostly a systems issue (12). These concerns are real, but we should also not overestimate the performance of AFB smears, especially in many high-burden countries, where quality control is chronically terrible. Cost is a serious issue, and national tuberculosis control programs, ministries of health, advocacy groups, and others should work hard to negotiate reasonable prices. Still, we should accept the fact that newer tools (diagnostics, drugs, and vaccines) are likely to have some additional costs associated with them under any circumstances. This is the cost of progress, and improving the lives of patients by allowing them access to the best diagnostics and drugs should be a priority that competes with other budgetary demands. In addition, the costs of delayed diagnosis (and the prolonged infectious period that accompanies it) and of incorrect diagnosis are also considerable, both in real dollars and in the under- and overtreatment of individual patients that result. As the paper by Lee demonstrates, Xpert is semiquantitative enough to allow it to probably replace AFB smear grading for use in determining infectiousness and response to treatment (5). In addition, in a world in which multidrug-resistant tuberculosis is a still unchecked threat, and in which, by the best estimate of the World Health Organization, only a small minority of patients are even diagnosed, Xpert provides nearly immediate information about whether a strain of M. tuberculosis is susceptible to rifampin or not. An AFB smear cannot do this.

stant tuberculosis is a still unchecked threat, and in which, by the best estimate of the World Health Organization, only a small minority of patients are even diagnosed, Xpert provides nearly immediate information about whether a strain of M. tuberculosis is susceptible to rifampin or not. An AFB smear cannot do this. The concern that introduction of Xpert in many regions has not led to decreases in tuberculosis mortality is both serious and a bit misleading. Ultimately, the goal of introducing new diagnostics, drugs, and vaccines is of course to reduce TB incidence and mortality. But this is often as much of a systems issue as it is an issue of the tools themselves. It behooves TB control programs to work diligently to adopt these new tools in a way that takes advantage of their potential (12, 13). The operational and economic considerations are not trivial. However, it would not have made sense to tell the Wright brothers not to bother inventing airplanes because there were no airports at which to land them. A recent letter proposed that clinicians say the following to any patient who is offered only an AFB smear for diagnosis of possible tuberculosis: “I apologize for only being able to offer you a century-old test that will miss the diagnosis half the time and may cause you to take toxic medications that won’t work because the test can’t detect resistance. We have not been successful in bringing modern diagnostic tests into use” (14). Precisely. Originally Published in Press as DOI: 10.1164/rccm.201809-1772ED on October 12, 2018

The concern that introduction of Xpert in many regions has not led to decreases in tuberculosis mortality is both serious and a bit misleading. Ultimately, the goal of introducing new diagnostics, drugs, and vaccines is of course to reduce TB incidence and mortality. But this is often as much of a systems issue as it is an issue of the tools themselves. It behooves TB control programs to work diligently to adopt these new tools in a way that takes advantage of their potential (12, 13). The operational and economic considerations are not trivial. However, it would not have made sense to tell the Wright brothers not to bother inventing airplanes because there were no airports at which to land them. A recent letter proposed that clinicians say the following to any patient who is offered only an AFB smear for diagnosis of possible tuberculosis: “I apologize for only being able to offer you a century-old test that will miss the diagnosis half the time and may cause you to take toxic medications that won’t work because the test can’t detect resistance. We have not been successful in bringing modern diagnostic tests into use” (14). Precisely. Originally Published in Press as DOI: 10.1164/rccm.201809-1772ED on October 12, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

Before the introduction of mechanical ventilation in the 1950s, preterm infants often died after birth from severe respiratory failure due to lung immaturity, surfactant deficiency, and the lack of suitable ventilators to support small infants. Survival of preterm infants dramatically improved with the introduction of continuous positive airway pressure (1), recognition of the importance of surfactant insufficiency in the pathobiology and treatment of neonatal respiratory distress syndrome (2), and the use of antenatal steroids (3). Although mechanical ventilation allowed an increasing number of preterm infants to survive, it was soon recognized that lung injury due in part to ventilator-induced lung injury led to persistent mortality and late morbidities including chronic lung disease, or bronchopulmonary dysplasia (BPD), as originally described by Northway and colleagues (4). Overall, advances in respiratory care over the past several decades, including improved strategies for mechanical ventilation, have led to dramatic advances in neonatal care and have been lifesaving for countless babies.

sease, or bronchopulmonary dysplasia (BPD), as originally described by Northway and colleagues (4). Overall, advances in respiratory care over the past several decades, including improved strategies for mechanical ventilation, have led to dramatic advances in neonatal care and have been lifesaving for countless babies. Despite these achievements, the adverse effects of mechanical ventilation on short- and long-term respiratory outcomes after preterm birth persist. Data from the Neonatal Research Network in 2015 showed that 87% of extremely preterm infants (gestational age, 22–28 wk) who survived more than 12 hours were treated with some form of mechanical ventilation (5). Although mechanical ventilation is lifesaving and often unavoidable, its use is associated with the development of BPD (6). Animal models show that even a short duration of mechanical ventilation to a preterm lung injures the lung and reduces the response to surfactant therapy (7). The many efforts to limit lung injury include improving ventilator strategies (8–10) such as volume-controlled ventilation (11, 12), antenatal steroids (13), and surfactant (14). These strategies have allowed many babies to avoid aggressive mechanical ventilation and to even decrease the use of early mechanical ventilation. In fact, rates of mechanical ventilation have dropped in the past 15 years (5). However, these advances have led to the survival of infants at lower gestational ages and birthweights with continued need for mechanical ventilation to survive. Hence, the rates of BPD remain fixed at 40% (5). In addition, despite substantial increases in the use of less-invasive ventilation after birth, there was no significant decline in oxygen dependence at 36 weeks and no significant improvement in lung function in childhood over time (15). Thus, a better understanding of the pathophysiology of mechanical ventilation–induced newborn lung injury and potential therapeutic interventions is desperately needed.

r birth, there was no significant decline in oxygen dependence at 36 weeks and no significant improvement in lung function in childhood over time (15). Thus, a better understanding of the pathophysiology of mechanical ventilation–induced newborn lung injury and potential therapeutic interventions is desperately needed. The “new BPD” seen today, 50 years after Northway and colleagues’ study, consists of an arrest of lung development (16), with defective alveolar septation and capillary formation (17, 18). This arrest of alveolar septation is associated with lung cell apoptosis in mechanically ventilated rodent models (19–21). Past studies have suggested that the mechanism of alveolar type II epithelial cell (AT-II) injury involves membrane lipid peroxidation, suppressed proliferation, and excessive apoptosis (19, 22). These studies also demonstrated that caspase 3–dependent AT-II cell apoptosis, initiated by the apoptosis-associated gene Fas/FasL and transduced by Bcl-2/Bax, is a key event in abnormal alveolar development seen in BPD (19, 22). Previous work from Yeganeh and colleagues and others demonstrated that the lung epithelial cell death seen with mechanical ventilation is via the extrinsic and not the intrinsic apoptotic pathway (23) and that autophagy may be an upstream regulator of apoptosis in cell fate decisions (24, 25). Furthermore, a recent review highlighted the observation that abnormal autophagy leading to increased cell apoptosis can be triggered by oxidative stress from reactive oxygen species (26). From this body of work, it is clear that more targeted studies on key pathways promoting autophagy will reveal new therapeutic targets to maintain autophagy balance and limit apoptosis and cell death.

l autophagy leading to increased cell apoptosis can be triggered by oxidative stress from reactive oxygen species (26). From this body of work, it is clear that more targeted studies on key pathways promoting autophagy will reveal new therapeutic targets to maintain autophagy balance and limit apoptosis and cell death. Sphingolipids, essential constituents of plasma membranes, are associated with lung inflammation, apoptosis, inhibition of surfactant protein B expression, as well as remodeling of the airway epithelium and vascular endothelium (27). Together with vascular growth factors, sphingolipids have emerged as vital components of lung alveolarization during development and are important determinants of lung responses to damage and repair. The lungs of infants who have died with BPD show evidence of reduced lung expression of vascular endothelial growth factor (17). Sphingolipids coupled with vascular endothelial growth factor signaling are important mediators of alveolar-capillary development, which may be impacted in infants with BPD (28, 29).

repair. The lungs of infants who have died with BPD show evidence of reduced lung expression of vascular endothelial growth factor (17). Sphingolipids coupled with vascular endothelial growth factor signaling are important mediators of alveolar-capillary development, which may be impacted in infants with BPD (28, 29). In this issue of the Journal, Yeganeh and colleagues (pp. 760–772) provide exciting new mechanistic information about ventilation-induced autophagy/apoptosis-related newborn lung injury (30). They explored the relationship between ceramide production, autophagy, and apoptosis in mechanical ventilation–induced epithelial cell death using in vivo mechanical ventilation rat model and in vitro cell stretch experiments. They report that mechanical ventilation increased pulmonary ceramide production that triggered autophagy and subsequent extrinsic apoptosis of lung epithelial cells. In addition, they show that prevention of ceramide generation by SMPD1, a sphingomyelinase, prevented autophagy-mediated cell death in mechanically ventilated newborn rats. This work is exciting, as it reveals a potential novel therapeutic target for the treatment of ventilation-induced lung injury in newborn infants with respiratory failure. The current findings extend the current mechanistic understanding of mechanical ventilation–induced cell death in newborn lungs by exposing ceramides as an upstream regulator of autophagy leading to increased apoptosis.

et for the treatment of ventilation-induced lung injury in newborn infants with respiratory failure. The current findings extend the current mechanistic understanding of mechanical ventilation–induced cell death in newborn lungs by exposing ceramides as an upstream regulator of autophagy leading to increased apoptosis. Altered ceramide content contributes to the pathology of abnormal lung development and respiratory diseases. Previous work by this group demonstrated elevated levels of ceramide in the tracheal aspirates of preterm infants who required mechanical ventilation in the first week of life, suggesting that ceramide concentration may be a useful early biomarker for the development of BPD (31). Although therapeutic targeting of ceramides for the prevention of oxidative and mechanical stretch injury is very exciting, much work will be needed to find the appropriate balance of pathologic and physiologic apoptosis in postnatal lung development and disease. In addition, the development of selective agents to control autophagy will take much investigation, but efforts by Yeganeh and colleagues and other investigative groups have some promising leads (30). Such mechanistic insights into the pathogenesis of mechanical ventilation–related lung injury are of the upmost importance, given the increasing numbers of babies at the edge of viability who are at the highest risk of respiratory failure needing mechanical ventilation during their newborn course. Originally Published in Press as DOI: 10.1164/rccm.201810-1857ED on October 29, 2018

Altered ceramide content contributes to the pathology of abnormal lung development and respiratory diseases. Previous work by this group demonstrated elevated levels of ceramide in the tracheal aspirates of preterm infants who required mechanical ventilation in the first week of life, suggesting that ceramide concentration may be a useful early biomarker for the development of BPD (31). Although therapeutic targeting of ceramides for the prevention of oxidative and mechanical stretch injury is very exciting, much work will be needed to find the appropriate balance of pathologic and physiologic apoptosis in postnatal lung development and disease. In addition, the development of selective agents to control autophagy will take much investigation, but efforts by Yeganeh and colleagues and other investigative groups have some promising leads (30). Such mechanistic insights into the pathogenesis of mechanical ventilation–related lung injury are of the upmost importance, given the increasing numbers of babies at the edge of viability who are at the highest risk of respiratory failure needing mechanical ventilation during their newborn course. Originally Published in Press as DOI: 10.1164/rccm.201810-1857ED on October 29, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

Some have questioned why the Journal has published so many papers on the health effects of household air pollution (HAP) from domestic cooking in recent years. There are compelling reasons to do so. Approximately 3 billion people, or 40% of the world’s population, still cook using open fires or simple stoves fueled by kerosene, biomass (wood, animal dung, and crop waste), or coal that generate considerable HAP (1). The Global Burden of Disease project attributes multiple noncommunicable diseases, including chronic obstructive pulmonary disease (COPD) and lung cancer, to HAP (2). In low- and middle-income countries (LMICs) where most women do not smoke tobacco, HAP may be the major cause of COPD and lung cancer in women. The World Health Organization estimates that HAP causes close to 3 million deaths per year, including 500,000 deaths of young children due to pneumonia or nearly half of all deaths due to pneumonia in children under 5 (1). Simply put, HAP has a huge public health impact that is theoretically preventable.

ncer in women. The World Health Organization estimates that HAP causes close to 3 million deaths per year, including 500,000 deaths of young children due to pneumonia or nearly half of all deaths due to pneumonia in children under 5 (1). Simply put, HAP has a huge public health impact that is theoretically preventable. A major criticism of the evidence base used to estimate the health impacts of HAP is the relative paucity of both measured exposure data and objectively measured outcome data, evidence gaps that are understandable given the difficulty of obtaining such data in the low-resource environments of countries where cooking with dirty fuels is common (3). The paper by Lee and colleagues (pp. 738–746) published in this edition of the Journal provides data that help fill both of these gaps (4). These investigators used 48-hour personal monitoring of carbon monoxide (CO) exposures four times over the course of pregnancy to assess prenatal exposure to HAP among rural participants in a cluster-randomized intervention trial of cleaner-burning cookstoves (GRAPHS [Ghana Randomized Air Pollution and Health Study]), an impressive exposure assessment effort. They then measured lung function parameters (i.e., the ratio of the time to peak tidal expiratory flow to expiratory time, tidal volume, respiratory rate, and minute ventilation with flow–volume loops, as well as passive respiratory system compliance with the single-occlusion technique) on the infant offspring of the monitored mothers at 1 month of age. This again is an impressive effort, and the first to obtain high-quality infant lung function measurements in such a low-resource setting.

ion with flow–volume loops, as well as passive respiratory system compliance with the single-occlusion technique) on the infant offspring of the monitored mothers at 1 month of age. This again is an impressive effort, and the first to obtain high-quality infant lung function measurements in such a low-resource setting. The primary finding of this elegant study is that maternal CO exposure during gestation was associated with lower infant lung function. This effect of HAP exposure was greater in girls than in boys. Moreover, infant lung function measured at 1 month of age was associated with an increased risk of pneumonia before age 1 as assessed by active weekly surveillance by fieldworkers, followed by physician evaluation of suspected cases using the World Health Organization’s Integrated Management of Childhood Illness guidelines. Physician-assessed severe pneumonia was also associated with infant lung function.

ed risk of pneumonia before age 1 as assessed by active weekly surveillance by fieldworkers, followed by physician evaluation of suspected cases using the World Health Organization’s Integrated Management of Childhood Illness guidelines. Physician-assessed severe pneumonia was also associated with infant lung function. Although the authors acknowledge some limitations of their study, including the relatively small sample size, lack of dietary data, and inability to confirm a diagnosis of pneumonia with chest imaging, the implications of their findings are profound. The fetal programming hypothesis, which proposes that prenatal environmental conditions are important determinants of disease in adulthood, is supported by both epidemiological and animal experimental data (5). Advances in epigenetic research also provide a plausible mechanism for such programming. Considerable evidence exists to suggest that maternal exposure to ambient air pollution and environmental tobacco smoke during gestation can lead to reduced lung function in offspring, which in turn can lead to a low lung function trajectory in adulthood associated with an increased risk of COPD (6–8). In addition to the lower infant lung function associated with maternal prenatal CO exposure, Lee and colleagues found that lower infant lung function increased the risk of early-childhood pneumonia. These findings now add exposure to HAP as another prenatal environmental risk factor for poor respiratory health later in life. The greater effect of prenatal exposure to HAP on girls in the current study is notable given that the greatest burden of HAP-related respiratory disease in LMICs is for adult women.

neumonia. These findings now add exposure to HAP as another prenatal environmental risk factor for poor respiratory health later in life. The greater effect of prenatal exposure to HAP on girls in the current study is notable given that the greatest burden of HAP-related respiratory disease in LMICs is for adult women. What can be done to prevent the harmful effects of in utero exposure to HAP on respiratory health during the subsequent life course? The published results from randomized control trials of “cleaner” biomass cookstoves are mixed with regard to the efficacy of such interventions for the prevention of early-childhood pneumonia (9, 10), and the results of a multicountry trial of liquefied-petroleum gas stoves are not likely to be published for several years. Moreover, widespread distribution of liquefied-petroleum gas stoves may not be available in many low-income countries for a number of years. That said, reduction of prenatal and infant exposures to HAP through behavioral changes, such as not burning rubbish near the home and keeping young children away from open biomass fires, may be of some benefit. The results of one study conducted in rural Guatemala suggest that reduced exposure to HAP through the use of a chimney stove intervention could improve lung function as measured by spirometry later in childhood (11).

rubbish near the home and keeping young children away from open biomass fires, may be of some benefit. The results of one study conducted in rural Guatemala suggest that reduced exposure to HAP through the use of a chimney stove intervention could improve lung function as measured by spirometry later in childhood (11). Poverty is inextricably intertwined with exposure to HAP as drivers of early-childhood respiratory illnesses that put children on a lower lung function growth trajectory and at increased risk of developing an adult respiratory illness (12). As the economies of LMICs develop, increased emissions from traffic and power generation will contribute to the cumulative exposure to air pollution. A great challenge for public health officials in these countries will be to prevent increased exposures of children to air pollution while the necessary economic development is being pursued. Distributed energy generation from solar-power microgrids is one potential solution, but the search for low-cost, feasible, clean cooking solutions must continue so that this public health problem can be addressed sooner rather than later. Originally Published in Press as DOI: 10.1164/rccm.201809-1698ED on October 1, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

Airway mucus has a central role in the pathogenesis of many major lung diseases (1). Excessive mucus production is the hallmark of chronic bronchitis in chronic obstructive pulmonary disease and is associated with reduced lung function (2) and poor clinical outcomes. Mucus plugging is a major contributor to airway obstruction in fatal asthma. Mucoviscidosis, a synonym for cystic fibrosis, describes a change in the properties of mucus that results in increased susceptibility to infection and progressive loss of lung function in this disease. Understanding the regulation and functional properties of mucus promises to open new avenues to treatment of many individuals with lung disease (3).

cystic fibrosis, describes a change in the properties of mucus that results in increased susceptibility to infection and progressive loss of lung function in this disease. Understanding the regulation and functional properties of mucus promises to open new avenues to treatment of many individuals with lung disease (3). Airway mucus is a heterogeneous mixture of water, ions, and organic macromolecules. Mucins are large, heavily glycosylated, gel-forming proteins that represent approximately 1.5% of the dry weight of normal airway mucus. Two homologous mucin genes, MUC5AC and MUC5B, play prominent roles in the lung. These genes arose from a common ancestor but have evolved to have different expression patterns and distinct functions (4). Much of our knowledge about mucins comes from mouse studies. In mice, Muc5b expression begins during embryonic development and continues postnatally (5). MUC5B protein is found in secretory cells within large (cartilaginous) airways and, at lower concentrations, in intrapulmonary axial airways, but not in bronchioles. MUC5B is also produced in submucosal glands, which in mice are limited to the proximal trachea. Deletion of Muc5b resulted in impaired mucociliary clearance and airway infections (6). Muc5ac is transiently expressed during embryonic lung development, but little if any MUC5AC is found in postnatal lungs unless mice are infected or challenged with allergen. In response to these stimuli, proximal airway secretory cells produce and secrete large amounts of MUC5AC. Deleting Muc5ac reduced airway reactivity in an asthma model (7), whereas transgenic overexpression of Muc5ac provided protection against respiratory viral infection (8).

are infected or challenged with allergen. In response to these stimuli, proximal airway secretory cells produce and secrete large amounts of MUC5AC. Deleting Muc5ac reduced airway reactivity in an asthma model (7), whereas transgenic overexpression of Muc5ac provided protection against respiratory viral infection (8). Mouse and human airways differ in several respects, and it is important to understand how insights from mouse studies apply to humans. In contrast to mice, humans have increased airway diameters, submucosal glands in more distal airways, and easily detectable MUC5AC expression, even in the absence of apparent infection or allergy. In this issue of the Journal, Okuda and colleagues (pp. 715–727) systematically map the regional distribution of MUC5AC and MUC5B along the proximal–distal axis of the normal human lung (9). A strength of this work is that multiple methods were used to measure mucins: mucin mRNAs were localized and quantified using in situ hybridization and digital droplet polymerase chain reaction, and quantitative immunohistochemistry was used to localize and quantify mucin proteins within epithelial cells. As reported previously, MUC5B was expressed in gland cells, and MUC5AC was not. In the superficial epithelium of the trachea and proximal and segmental bronchi, both mucins were expressed at roughly similar levels, as determined by measurements of MUC5B and MUC5AC mRNAs and the volume of intracellular MUC5B and MUC5AC protein staining. In distal bronchi and bronchioles, MUC5B and especially MUC5AC expression decreased, resulting in an increased MUC5B:MUC5AC ratio. Neither mucin was detected in terminal bronchioles. Because distal bronchioles have a much larger surface area than more proximal airways, the authors estimate that most airway MUC5B expression is accounted for by distal bronchioles. In contrast, most MUC5AC expression was attributed to larger airways (bronchi).

ratio. Neither mucin was detected in terminal bronchioles. Because distal bronchioles have a much larger surface area than more proximal airways, the authors estimate that most airway MUC5B expression is accounted for by distal bronchioles. In contrast, most MUC5AC expression was attributed to larger airways (bronchi). Okuda and colleagues also highlight secretory cell heterogeneity by using in situ hybridization to show that mucin-producing gland cells are MUC5B+ MUC5AC− CCSP−, whereas superficial airway cells are typically MUC5B+ MUC5AC− CCSP+ (distal bronchioles) or MUC5B+ MUC5AC+ CCSP+ (proximal bronchi). As noted by the authors, the nomenclature used for airway secretory cells, which includes club cells, goblet cells, mucous cells, serous cells, and indeterminate cells, is based principally on histologic criteria that fail to capture the heterogeneity revealed by recent studies. In the absence of a clear consensus about what these names signify, we suggest that investigators use molecular markers for classifying secretory cells. The application of single-cell RNA sequencing to airway epithelial cells (10, 11) is revealing airway epithelial cell transcriptomes at unprecedented resolution and will help us to better understand heterogeneity in secretory cells. These methods will need to be combined with other methods that provide spatial information to gain a deeper understanding of how mucin-producing cells and other cells that secrete macromolecules, ions, and water vary regionally in the lung.

olution and will help us to better understand heterogeneity in secretory cells. These methods will need to be combined with other methods that provide spatial information to gain a deeper understanding of how mucin-producing cells and other cells that secrete macromolecules, ions, and water vary regionally in the lung. Two limitations of the study deserve mention. First, many of the observations were based on a subset of histologically normal-appearing lungs from only 5 subjects, a sample size too small to define the range of normal in human populations. Nonetheless, the methods used here and the resulting insights into typical patterns of mucin expression are a useful basis for designing future studies dependent on measuring mucins in normal and diseased lungs. A second limitation is that the methods used here measure mucin RNA and stored protein-stained volumes, and not mucin protein secreted from different portions of the airway. The results of this study suggest that the MUC5B:MUC5AC protein ratio within airway mucus is higher in the distal lung (where very little MUC5AC is produced). However, this may be offset by secretion of substantial amounts of MUC5B-rich mucus from glands in the proximal airway.

ed from different portions of the airway. The results of this study suggest that the MUC5B:MUC5AC protein ratio within airway mucus is higher in the distal lung (where very little MUC5AC is produced). However, this may be offset by secretion of substantial amounts of MUC5B-rich mucus from glands in the proximal airway. Many important questions remain to be addressed. Understanding how expression of mucins and other genes are spatially regulated within the normal airway epithelium, and how genetic and environmental factors that affect gene regulation affect disease risk, remains a major challenge. The discovery that a regulatory variant affecting MUC5B expression in distal airways is associated with a very large increase in risk of developing pulmonary fibrosis is a compelling early step toward this goal (12). It will also be critical to identify disease-associated changes in mucin gene expression in different regions of the lung and to understand how these affect mucus function. Recent studies show that differences in mucus composition are associated with dramatic differences in mucus organization and function. For example, MUC5B and MUC5AC are found within distinct domains of mucus plugs in fatal asthma, and the MUC5AC-rich domains play a unique role in mucostasis by tethering to the epithelium (13). In pigs, MUC5B from submucosal gland ducts formed strands composed of multiple MUC5B filaments, whereas MUC5AC emerged from superficial secretory cells as wispy threads or sheets, and it seems likely that these distinct structures contribute differently to mucociliary transport (14). Understanding how regional and disease-associated differences in mucins and other mucus components affect host defense and lung function is likely to be a long but rewarding journey. Okuda and colleagues have provided a map that will help us find our way.

structures contribute differently to mucociliary transport (14). Understanding how regional and disease-associated differences in mucins and other mucus components affect host defense and lung function is likely to be a long but rewarding journey. Okuda and colleagues have provided a map that will help us find our way. Originally Published in Press as DOI: 10.1164/rccm.201809-1818ED on October 30, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

The history of digital biomarker research based on computed tomography (CT) in interstitial lung disease (ILD) is long, spanning more than 20 years. Early studies involved radiologists visually quantifying the extent of parenchymal disease and investigating its prognostic effect, mainly in the setting of fibrotic lung disease. This research has mostly provided consistent results: that with increasing fibrosis, honeycombing or severity of traction bronchiectasis comes with an increased risk for mortality (1, 2). Attempts have also been made to construct multidimensional staging systems for different ILDs designed to provide an objective score that maps to an evidence-based management strategy, much in the same way that lung cancer is staged (3, 4).

y of traction bronchiectasis comes with an increased risk for mortality (1, 2). Attempts have also been made to construct multidimensional staging systems for different ILDs designed to provide an objective score that maps to an evidence-based management strategy, much in the same way that lung cancer is staged (3, 4). Despite these efforts, CT-based biomarkers and staging tools have largely failed to translate from research to routine clinical practice for a number of reasons. First, visual quantification of ILD on CT is a matter of fine judgement, liable to significant interobserver variability and, as a continuous variable, is not easily applied to management decisions in an individual patient (5). Perhaps most important, concerns remain regarding the reproducibility of visual CT scoring, even in the hands of expert radiologists (6). These shortcomings have spurred a groundswell of interest in objective computer-based quantitative CT (QCT), beginning with simple measures of lung density, followed by more sophisticated textural analysis capable of quantifying the extent and distribution of specific parenchymal patterns such as honeycombing and ground glass opacification (7, 8). Recent innovations in computer-based ILD evaluation include the discovery of a new computer-derived CT parameter, so-called vascular-related structures, which provides sharper prognostic discrimination than traditional CT markers of disease severity in several fibrotic lung diseases, as well as the application of deep learning to QCT in idiopathic pulmonary fibrosis (9, 10).

de the discovery of a new computer-derived CT parameter, so-called vascular-related structures, which provides sharper prognostic discrimination than traditional CT markers of disease severity in several fibrotic lung diseases, as well as the application of deep learning to QCT in idiopathic pulmonary fibrosis (9, 10). Common to most of this research is that it has focused on the lung parenchyma, whereas in contrast, the mediastinum has been relatively ignored. This shortcoming is surprising, given that ILD radiologists all over the world know very well that mediastinal node enlargement frequently occurs in patients with ILD. It is also remarkable when one considers the importance of evaluating mediastinal lymph nodes in other pulmonary diseases such as lung cancer and sarcoidosis. And yet the importance of mediastinal lymph node enlargement in ILD is a poorly understood and underrated phenomenon. Too often I have myself been guilty of reporting “enlarged mediastinal lymph nodes consistent with the presence of ILD” merely to highlight that I have not missed this finding, rather than to convey its significance.

portance of mediastinal lymph node enlargement in ILD is a poorly understood and underrated phenomenon. Too often I have myself been guilty of reporting “enlarged mediastinal lymph nodes consistent with the presence of ILD” merely to highlight that I have not missed this finding, rather than to convey its significance. In this issue of the Journal, Adegunsoye and colleagues (pp. 747–759) make a strong argument for systematic radiologic evaluation of mediastinal lymph node (MLN) enlargement, as well as the distribution of mediastinal lymphadenopathy in patients with ILD (11). The purpose of their study was to test whether outcome distinctions exist between patients with ILD with and without enlarged MLNs. Their primary outcome measure was all transplant-free survival with all-cause and respiratory hospitalizations, lung function, and plasma cytokine concentrations as secondary outcomes. The authors left very little room for error in their study design, which followed a robust discovery-validation protocol using three cohorts from separate institutions with different referral patterns to verify the generalizability of their findings. MLN measurements were made at stations 1–9 (as designated by the International Association for the Study of Lung Cancer) by two thoracic radiologists who underwent hands-on training before the study to ensure consistency in their analyses. It is noteworthy (and of practical relevance) that the agreement between the two radiologists on lymph node enlargement, the total number of enlarged nodes, and the site of largest lymph nodes was remarkably good (k = 0.64–0.69). This result refreshingly contrasts the reported reproducibility of semiquantitative scoring of CT patterns such as honeycombing, which, as mentioned earlier, is inconsistent at best (5, 12).

ph node enlargement, the total number of enlarged nodes, and the site of largest lymph nodes was remarkably good (k = 0.64–0.69). This result refreshingly contrasts the reported reproducibility of semiquantitative scoring of CT patterns such as honeycombing, which, as mentioned earlier, is inconsistent at best (5, 12). MLN enlargement analyses of the paratracheal and lower mediastinal nodes (stations 1–7 and 8–9, respectively) were made using a binary ≥10 mm or <10 mm in short axis dimension categorization. Compared with those without enlarged MLNs, older, male patients with increasing smoking exposure and, therefore, not surprisingly, patients with IPF, more commonly had MLN enlargement. Interestingly, although MLN enlargement was more common than not in patients with interstitial pneumonia with autoimmune features it was proportionately less frequent in patients with connective tissue disease-related ILD and chronic hypersensitivity pneumonitis. The primary finding comes from the survival analysis, which demonstrated an increased risk for mortality associated with MLN enlargement in the all-comers ILD discovery cohort, which was then replicated in all three validation cohorts, with lower mediastinal lymphadenopathy consistently conferring a higher mortality risk than paratracheal lymphadenopathy. These survival differences were independent of radiologic honeycombing or an enlarged pulmonary artery (which may be a marker of pulmonary hypertension), and on subgroup analysis, the greatest prognostic separation between those with and without enlarged MLNs was in patients with IPAF and unclassifiable ILD. Because these ILD subtypes incorporate disorders that range from being intrinsically stable to being inexorably progressive, these results suggest that MLN enlargement may be a novel CT biomarker of progressive disease behavior. Given the growing research focus on the progressive fibrotic phenotype, this finding may be of great practical importance for several reasons (13). First, in a proportion of patients with IPF, baseline investigations may be insufficient to guide initial management decisions, but knowledge of likely short-term disease behavior may increase confidence sufficiently to allow a working diagnosis of IPF to be made (14). Second, if the eagerly anticipated results of the INBUILD study are positive, then accurate prediction of progressive fibrotic ILD at presentation would allow initiation of antifibrotic therapy without delay (15).

disease behavior may increase confidence sufficiently to allow a working diagnosis of IPF to be made (14). Second, if the eagerly anticipated results of the INBUILD study are positive, then accurate prediction of progressive fibrotic ILD at presentation would allow initiation of antifibrotic therapy without delay (15). Last, early identification of patients who will develop progressive fibrotic ILD would enable early initiation of quality of life–improving measures and facilitate planning for the future. The authors also evaluated relationships between more than 40 circulating cytokines and MLN enlargement, as well as their associations with outcome. Specifically, elevated levels of the anti-inflammatory cytokine IL-10 (an essential regulator of proinflammatory responses in pulmonary fibrosis) were associated with increased mortality in patients with and without enlarged MLNs. Taking these findings together, it is likely that the most predictive models of future progressive fibrotic ILD will come from a combination of CT and serum biomarkers of disease progression.

flammatory responses in pulmonary fibrosis) were associated with increased mortality in patients with and without enlarged MLNs. Taking these findings together, it is likely that the most predictive models of future progressive fibrotic ILD will come from a combination of CT and serum biomarkers of disease progression. Being able to identify how ILD will progress in a specific patient will allow clinicians to initiate patients on appropriate treatment at the earliest opportunity, as well as disease progression. This remains one of the most urgent challenges for effective management for patients with progressive ILD. Until now, CT biomarker research in ILD has focused on baseline and longitudinal changes based on the extent of disease in the lung parenchyma. In this issue of the Journal, Adegunsoye and colleagues provided compelling evidence that the mediastinum can no longer be overlooked. I will adjust my CT reports accordingly. Originally Published in Press as DOI: 10.1164/rccm.201810-1892ED on October 18, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

Anyone who has ever performed successful cardiopulmonary resuscitation (CPR) knows instantly that they have done something truly incredible. The significance of the act and the elemental feeling of usefulness is the same for all members of the extended healthcare team or the public who have just saved a life. CPR is also unique in that its core principles are well understood by many, or at least that’s what we think. It consists of clearing a patient’s airway, rhythmically pumping the thorax to circulate blood around the body, providing some ventilation to replace spontaneous breathing, and if a shockable arrhythmia is present, performing defibrillation. Initial research into CPR yielded major gains. Campaigns to convert bystanders into competent responders have been success stories of major proportions, leading to marked improvements in rates of survival with good neurological outcome (1–3). However, recent clinical research into CPR has not improved its effectiveness. Large randomized controlled trials have tested the optimal types and doses of inotropic agents, bicarbonate, or anti-arrhythmic agents without identifying beneficial treatments (4–6). Perhaps the most striking (and highly powered) recent randomized controlled trial was the CCC (Continuous Chest Compression) study, which roughly translates into: ventilation is not sufficiently important to warrant interruption of CPR (7).

ate, or anti-arrhythmic agents without identifying beneficial treatments (4–6). Perhaps the most striking (and highly powered) recent randomized controlled trial was the CCC (Continuous Chest Compression) study, which roughly translates into: ventilation is not sufficiently important to warrant interruption of CPR (7). A novel insight in this issue of the Journal by Grieco and colleagues (pp. 728–737) has the potential to move beyond this impasse (8). The investigators began with careful observations of end-tidal CO2 (ETCO2) tracings during CPR in victims of cardiac arrest. Because exhaled CO2 reflects delivery of venous blood to the lungs, ETCO2 tracings are recommended to monitor the effectiveness of CPR. But during CPR, the ETCO2 tracing is characterized by oscillations; this is assumed to represent variable ventilation and exhalation of CO2. However, in many patients, these CPR-related oscillations were inconsistent or absent, presumably reflecting obstruction to airflow that prevented passage of the CPR-related CO2 oscillations to the ETCO2 monitor at the end of the endotracheal tube. Because large airways were patent in each patient (endotracheal tubes were present in all), the investigators hypothesized that the airflow obstruction occurred in small airways. To quantify the degree of airway closure, they developed an Airway Opening Index, representing the changing ETCO2 values during CPR compared with the maximal ETCO2. Thus, a higher index reflects greater transmission of CO2 and greater airway patency.

hypothesized that the airflow obstruction occurred in small airways. To quantify the degree of airway closure, they developed an Airway Opening Index, representing the changing ETCO2 values during CPR compared with the maximal ETCO2. Thus, a higher index reflects greater transmission of CO2 and greater airway patency. Two sets of experiments confirmed that small airway closure could explain this phenomenon. Using a bench lung model, they reproduced the ETCO2 waveforms observed in the real patients by simulating airway closure, and demonstrated that incremental levels of positive end-expiratory pressure (PEEP) increased transmission of the oscillations, as well as elevating inspiratory flow and minute ventilation. In addition, the highest ETCO2 value during CPR represented the closest estimate of the actual alveolar CO2. Each of these findings was recapitulated in human cadavers (Thiel model), where the alveoli were loaded with CO2 before CPR; although PEEP improved transmission of oscillations, elevating the airway index, it did not compromise the effect of chest compressions on intrathoracic pressure.

mate of the actual alveolar CO2. Each of these findings was recapitulated in human cadavers (Thiel model), where the alveoli were loaded with CO2 before CPR; although PEEP improved transmission of oscillations, elevating the airway index, it did not compromise the effect of chest compressions on intrathoracic pressure. If upheld, these findings could have immense application to the conduct of CPR in several direct ways. First, although perfusion with poorly oxygenated blood is preferable to no perfusion, without at least some ventilation, the perfusing blood will ultimately contain very little oxygen. Given that oxygenating tissues is the primary function of circulating blood, how could it be that ventilation during CPR seems not to be important? The idea that small airways close during CPR and prevent alveolar delivery of the applied breath could explain this paradox, because if the airways could be opened by titrated PEEP, the importance of ventilation during CPR could be properly evaluated. Second, it is possible that many trials reporting failure of carefully considered therapies (e.g., inotropic agents, antiarrhythmic medications, and even advanced airway techniques [9]) may represent false-negative results because ventilation was inadequate in so many patients. This is because if return of circulation does not occur rapidly, then ongoing CPR with inadequate ventilation would render any cointervention ineffective. Third, measuring the Airway Opening Index (or a version of it) should be easy, given that ETCO2 monitors are now widely available, including in ambulances. Thus, small airway closure during CPR could be detected and eliminated with small levels of PEEP.

nadequate ventilation would render any cointervention ineffective. Third, measuring the Airway Opening Index (or a version of it) should be easy, given that ETCO2 monitors are now widely available, including in ambulances. Thus, small airway closure during CPR could be detected and eliminated with small levels of PEEP. Although this study, and another promising to increase perfusion during CPR using inhaled nitric oxide (10), are grounds for optimism, there are important reasons to be cautious. Several steps are required to corroborate this theory of small airway closure. The results need to be reproduced by others, and the nature and locus of the airway closure need to be better understood, perhaps using novel imaging. Although the cadaver studies were reassuring about the lack of adverse effects of PEEP, the optimal level of PEEP, balancing airway patency against perfusion, needs to be understood in individual patients. Finally, clinical trialists in the field will need all of their accumulated insight and experience to guard against a false-negative (or, far less likely, a false-positive) controlled trial. In conclusion, CPR is a very special gift. Research in the field made early gains, but more recently progress has been slow. The physiologic insight by Grieco and colleagues has the potential to make a major positive difference to the field: If treating small airway closure works, survival after CPR might increase, and more important, so too might the quality of life among survivors.

ade early gains, but more recently progress has been slow. The physiologic insight by Grieco and colleagues has the potential to make a major positive difference to the field: If treating small airway closure works, survival after CPR might increase, and more important, so too might the quality of life among survivors. D.C.S. and B.P.K. each receive research support from the Canadian Institutes of Health Research. B.P.K. holds the Dr. Geoffrey Barker Chair in Critical Care Research. Originally Published in Press as DOI: 10.1164/rccm.201810-1912ED on November 1, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

In addition to serving as a tissue for energy storage, adipose tissue has become a well-recognized endocrine organ that secretes a variety of adipokines with important pleiotropic functions. One of these adipokines is leptin, discovered in 1994 by Zhang and colleagues (1). Much of the research on leptin has focused on its role on metabolism, particularly in central nervous system regulation of energy homeostasis and obesity, as well as its peripheral effects on obesity-related cardiometabolic diseases. The excess adiposity in obese humans leads to high circulating levels of leptin. Paradoxically, despite leptin’s well-described effects on suppressing appetite and increasing energy expenditure, these individuals remain obese, reflecting a state of leptin resistance (2). A few years after its discovery, it became evident that leptin has a significant effect on ventilation and control of breathing (3, 4). At the central nervous system level, leptin increases the hypercapnic ventilatory response. Yet, severely obese patients afflicted with obesity hypoventilation syndrome (OHS) continue to hypoventilate despite having high circulating levels of leptin, in line with leptin resistance. Further evidence in support of leptin resistance at the central nervous system level comes from experiments in which parenterally administered recombinant leptin was shown to be largely ineffective in reducing weight in the vast majority of obese individuals (5). For leptin to affect the respiratory center and increase minute ventilation, it has to first cross the blood–brain barrier (BBB). One proposed mechanism for leptin resistance is impaired leptin transport across the BBB (6).

hown to be largely ineffective in reducing weight in the vast majority of obese individuals (5). For leptin to affect the respiratory center and increase minute ventilation, it has to first cross the blood–brain barrier (BBB). One proposed mechanism for leptin resistance is impaired leptin transport across the BBB (6). Against this background, in this issue of the Journal, Berger and colleagues (pp. 773–783) postulated that intranasal administration of leptin could bypass the BBB and thus promote its physiologic action on the control of respiration and the upper airway, consequently mitigating sleep-disordered breathing (SDB) (7). The finding that intranasal leptin can alleviate hypoventilation and upper-airway obstruction in an obese mice model is exciting because it provides a much-needed novel therapeutic approach for the management of SDB. Since its introduction in 1981, positive airway pressure (PAP) therapy has remained the gold standard treatment for SDB (8). However, despite its high effectiveness, PAP therapy has low clinical efficacy, primarily due to suboptimal adherence in a large proportion of patients (9). Moreover, approximately 25% of patients with OHS remain hypercapnic despite high levels of adherence to nocturnal PAP therapy (10). As such, it is not surprising that there has been much interest in discovering novel therapeutic targets and modalities (11–13). Any therapy that is able to effectively maintain upper-airway patency and normalize ventilation throughout the entire sleep period will go a long way toward improving long-term health outcomes in patients with SDB, and will be a welcome addition to our armamentarium to treat disordered breathing during sleep. One concern regarding ventilatory stimulants is exposing the upper airway to increasingly negative intrathoracic pressure, thereby promoting upper-airway collapse. However, Berger and colleagues demonstrated that intranasal leptin improves inspiratory flow limitation despite its ventilatory stimulant effect (7). This finding, in conjunction with transneuronal tracer experiments, suggests that at the central nervous system level, leptin can simultaneously improve ventilatory response and upper-airway tone through synaptic connections between leptin receptor–expressing cells and hypoglossal and phrenic motor neurons.

effect (7). This finding, in conjunction with transneuronal tracer experiments, suggests that at the central nervous system level, leptin can simultaneously improve ventilatory response and upper-airway tone through synaptic connections between leptin receptor–expressing cells and hypoglossal and phrenic motor neurons. Although prior work has explored the effect of intranasal leptin administration on reducing food intake in rats (14, 15), Berger and colleagues provide a very elegant, albeit preliminary, physiological demonstration of how leptin delivered intranasally can overcome “central leptin deficiency” and lead to demonstrable improvement in upper-airway resistance and ventilation in a murine model (7). However, as with any well-designed and novel animal experiment, there are many unanswered questions. First and foremost, the demonstration of an acute effect of intranasal leptin has less clinical relevance for managing a chronic disease such as SDB. Therefore, more research is needed to demonstrate the long-term efficacy of intranasal leptin in alleviating SDB. In theory, if this acute effect is sustained after repeated administrations without significant side effects, the long-term modulation of cerebral areas that control appetite as well as breathing may have further desirable effects on health beyond those of improving SDB and ventilation. Such effects cannot be assumed until the experiments are done and new evidence becomes available. It also remains unclear whether improvement in ventilation during sleep would be sustained during wakefulness to ameliorate daytime hypoventilation, a hallmark of OHS. Second, the exact mechanism by which intranasal leptin exerts its action of relieving SDB needs to be further elucidated. The use of the intranasal route stemmed from its ability to bypass leptin resistance, which is attributed, at least in part, to limited permeability of the BBB to leptin. However, recent work suggests that leptin transport into cerebrospinal fluid is intact in obese mice (16). Although this does not contradict the finding that intranasal leptin, and not intraperitoneal leptin, relieved SDB, it is clear that much more needs to be explored regarding the traffic of leptin into the central nervous system and its mechanisms of actions on various parts of the brain.

d is intact in obese mice (16). Although this does not contradict the finding that intranasal leptin, and not intraperitoneal leptin, relieved SDB, it is clear that much more needs to be explored regarding the traffic of leptin into the central nervous system and its mechanisms of actions on various parts of the brain. Prior studies exploring the effect of intranasal leptin on appetite and weight in mice and rats used substantially lower concentrations (0.1 or 0.2 mg/kg) (14, 15) than the current study, in which the delivered dose was 0.4 mg/kg. Whether the dose needs to be adjusted based on the therapeutic goal (i.e., appetite vs. respiratory modulation) also requires further investigation. Lastly, further research is needed to identify the patient population that will be most responsive to this therapeutic modality. Although we are excited by this novel finding of intranasal leptin in the murine model, the translation to humans cannot be taken on a mere leap of faith, as men are not mice. The sleep medicine community eagerly awaits additional experiments and clinical trials exploring intranasal leptin in the management of SDB. Originally Published in Press as DOI: 10.1164/rccm.201810-1925ED on October 26, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

Despite increased focus over the past decade, the management of patients with acute kidney injury (AKI) remains largely supportive, including dialysis for severe cases. Clinical trials in AKI examining timing of dialysis, intensity of dialysis, pharmacotherapy, and novel biologics have been consistently negative (1–6). One postulated reason for this dearth of positive trials is the inherent delay in intervention for patients with AKI due to a reliance on serum creatinine, and researchers have embarked on a decades-long journey to identify a biomarker of AKI that would identify patients while kidney damage was actively ongoing and before serum creatinine increases. Numerous biomarkers of tubular injury have now been identified (7, 8), and in the ICU these biomarkers have modest sensitivity, specificity, and association with outcomes (9, 10). However, these biomarkers have so far failed to recategorize the heterogeneous syndrome of AKI into more clinically useful subtypes or be incorporated into clinical practice. There has been some progress with biomarkers of cell cycle arrest, most notably TIMP2*IGFBP7 (tissue inhibitor of metalloproteinase-2*insulin growth factor binding protein-7), to identify patients at high risk of AKI (11, 12). In patients after cardiac bypass surgery at high risk for AKI (as denoted by elevated TIMP2*IGFBP7), there was a lower incidence and decreased severity of AKI in patients who were randomized to a “KDIGO (Kidney Disease: Improving Global Outcomes) bundle,” which included monitoring of hemodynamic parameters, avoidance of nephrotoxins, and holding angiotensin-converting-enzyme inhibitors (13). Although prevention may be possible, the role of biomarkers in guiding treatments or response to therapy remains unclear.

“KDIGO (Kidney Disease: Improving Global Outcomes) bundle,” which included monitoring of hemodynamic parameters, avoidance of nephrotoxins, and holding angiotensin-converting-enzyme inhibitors (13). Although prevention may be possible, the role of biomarkers in guiding treatments or response to therapy remains unclear. For this reason, the article by Bhatraju and colleagues (pp. 863–872) in this issue of the Journal represents meaningful progress (14). The authors applied latent class analysis to a discovery group of 794 patients admitted with systemic inflammatory response syndrome to the ICU and a replication cohort of 425 patients with acute respiratory distress syndrome (ARDS) and identified two subphenotypes of AKI (AKI-SP1 and AKI-SP2). The patients in AKI-SP2 were sicker and had worse renal function; higher rates of sepsis, ARDS, and mortality; and lower rates of renal recovery. The authors determined via least absolute shrinkage and selection operator method that the ratio of angiopoietin-1 and angiopoietin-2 (Ang1/Ang2) and sTNFR-1 (soluble tumor necrosis factor receptor-1) were sufficient to accurately distinguish between the two subphenotypes of AKI (c-statistic > 0.93).

of renal recovery. The authors determined via least absolute shrinkage and selection operator method that the ratio of angiopoietin-1 and angiopoietin-2 (Ang1/Ang2) and sTNFR-1 (soluble tumor necrosis factor receptor-1) were sufficient to accurately distinguish between the two subphenotypes of AKI (c-statistic > 0.93). Ang1 and Ang2 are endothelial growth factors, which both bind to the extracellular portion of the Tie-2 receptor. They have opposing actions: Ang-1 stabilizes the vascular endothelium, and Ang-2 destabilizes the vascular endothelium. Consequently, the ratio of these endothelial growth factors provides an assessment of endothelial dysfunction and is associated with prognosis in several cohorts of critically ill patients with and without AKI (15, 16).

actions: Ang-1 stabilizes the vascular endothelium, and Ang-2 destabilizes the vascular endothelium. Consequently, the ratio of these endothelial growth factors provides an assessment of endothelial dysfunction and is associated with prognosis in several cohorts of critically ill patients with and without AKI (15, 16). These sophisticated statistical techniques and biomarkers determined what clinicians intuitively understand: patients with more severe inflammation do worse. The authors then reidentified the subphenotypes in a random subset of 328 patients from the VASST (Vasopressin in Septic Shock Trial) who had measurements available for Ang1/Ang2 and IL-8 (17). (Soluble tumor necrosis factor receptor-1 was not available in the VASST cohort, but IL-8 was notably different between AKI-SP1 and AKI-SP2 in the discovery and replication cohorts.) This clinical trial was a randomized, double-blind study comparing vasopressin and norepinephrine infusions to norepinephrine alone in 776 patients with septic shock. The study had shown no differences in mortality or rates of renal failure between patients in either treatment group. Once patients were recategorized into the AKI subphenotypes, patients in AKI-SP1 (the less ill group) had improved 90-day mortality with early addition of vasopressin compared with norepinephrine alone. This association persisted after adjustment for Acute Physiology and Chronic Health Evaluation II score, suggesting discriminating ability of the AKI subphenotypes beyond simply severity of illness. If replicated, these findings hold the potential to guide clinicians in more accurately assessing prognosis of patients as well as expected response to therapy.

ment for Acute Physiology and Chronic Health Evaluation II score, suggesting discriminating ability of the AKI subphenotypes beyond simply severity of illness. If replicated, these findings hold the potential to guide clinicians in more accurately assessing prognosis of patients as well as expected response to therapy. Despite these findings, much work remains. There were no urine specimens available from these cohorts, and it is unclear if the particular model the authors identified is truly the best model to distinguish subphenotypes of AKI in patients with systemic inflammatory response syndrome. Moreover, given the heterogeneity of AKI and the innumerable settings in which it occurs, this particular model may not be helpful for all patients with AKI, such as those post cardiac surgery, post contrast exposure, or with heart failure. Other investigators will have to identify subphenotypes of patients with AKI in these settings, and these subphenotypes may be identified by a combination of biomarkers of tubular injury, cell cycle arrest, cardiac dysfunction, clinical variables, or a novel biomarker. Incorporation of phenotyping may therefore be an important component of future trials for AKI and could possibly be the key to breaking the trend of negative clinical trials in AKI.

tified by a combination of biomarkers of tubular injury, cell cycle arrest, cardiac dysfunction, clinical variables, or a novel biomarker. Incorporation of phenotyping may therefore be an important component of future trials for AKI and could possibly be the key to breaking the trend of negative clinical trials in AKI. It is important to note that our molecular understanding of acute tubular injury, which, if biopsies were available in these cohorts, probably would have been the most common histologic entity, remains suboptimal. Inflammation and endothelial dysfunction are typically considered a systemic response instead of intrinsic to the kidney. The KPMP (Kidney Precision Medicine Project), which will integrate molecular, structural, and clinical information from kidney biopsy specimens of patients with AKI, will be critically important to untangle the molecular underpinnings of acute tubular injury. Ideally, these data will improve our ability to identify disease subgroups that would respond to therapy.

which will integrate molecular, structural, and clinical information from kidney biopsy specimens of patients with AKI, will be critically important to untangle the molecular underpinnings of acute tubular injury. Ideally, these data will improve our ability to identify disease subgroups that would respond to therapy. Instead, this particular article can serve as a framework for other investigators to attempt to identify other subphenotypes of AKI. To move forward, we recommend the following next steps. First, investigators should attempt to identify novel subphenotypes of AKI in other common AKI settings using currently existing biomarkers. As our molecular understanding of AKI improves from studies like KPMP, these subphenotypes should ultimately be identified based on underlying molecular mechanisms. Second, investigators should demonstrate that these subphenotypes respond differently to therapies in previously completed clinical trials. This is an important step and, given the plethora of negative clinical trials in AKI, there is ample opportunity. Ultimately, we hope these findings would change the way patients are selected and enrolled into future clinical trials. Although biorepositories require time, money, and effort, we urge investigators to continue maintaining and creating them, as they can yield valuable information years later.

ere is ample opportunity. Ultimately, we hope these findings would change the way patients are selected and enrolled into future clinical trials. Although biorepositories require time, money, and effort, we urge investigators to continue maintaining and creating them, as they can yield valuable information years later. It remains to be proven if the subphenotypes of AKI in patients with critical illness identified by this study will be useful in clinical practice. Regardless, the findings are important because they suggest that it is possible to untangle the complex and heterogeneous clinical syndrome of AKI into groups of patients who respond to a particular therapy. In other words, we may be one step closer to personalized and precision medicine for AKI. Originally Published in Press as DOI: 10.1164/rccm.201810-2032ED on November 6, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

From the Authors: We appreciate the editorial by Fan (1) and the letter by Kredel and colleagues regarding our recent publication (2). Both compare the near-apneic ventilation strategy we applied, associated with high-flow veno-venous extracorporeal membrane oxygenation (ECMO), with near-apneic strategies applied in association with low-flow extracorporeal CO2 removal systems (ECCO2R). We think this comparison overlooks a fundamental difference. In the original experience in patients with acute respiratory distress syndrome (ARDS) reported by Gattinoni and coworkers (3), intermittent sighs to peak airway pressures of 35–45 cm H2O and high positive end-expiratory pressure (PEEP) levels from 15 to 25 cm H2O were applied. In Johannes and colleagues’ study (4), PEEP levels above 20 cm H2O were used after a recruitment maneuver in an experimental model of ARDS. In contrast, our near-apneic strategy kept PEEP at 10 cm H2O and maximal airway pressures at 20 cm H2O. Although the decreases in V˙e were of similar magnitude to our study, airway pressures differ markedly. As ECCO2R does not contribute to oxygenation, very high airway pressures have to be applied in severe ARDS to maintain oxygenation, so that static stress and strain remain high, and eventually right ventricular function and hemodynamics may be compromised. This potential risk has become more apparent after the negative results of the OSCILLATE (Oscillation for ARDS Treated Early) and ART (Alveolar Recruitment for Acute Respiratory Distress Syndrome Trial) trials (5, 6). In contrast, in our study, we could keep significantly lower mean and driving airway pressures, avoiding both static and dynamic stress and strain. We believe this is the fundamental reason why our results differ from those of Johannes and colleagues (4), who found no positive effect of decreasing Vt to 3 or even 0 ml/kg on lung tissue inflammation.

ould keep significantly lower mean and driving airway pressures, avoiding both static and dynamic stress and strain. We believe this is the fundamental reason why our results differ from those of Johannes and colleagues (4), who found no positive effect of decreasing Vt to 3 or even 0 ml/kg on lung tissue inflammation. In fact, we have recently presented in abstract form the results of a study evaluating 3 different airway pressures (low: PEEP 0 cm H2O–peak inspiratory pressure [PIP] 10 cm H2O; moderate: PEEP 10 cm H2O–PIP 20 cm H2O; high: PEEP 20 cm H2O–PIP 30 cm H2O) applied during a near-apneic protocol in the same model of ARDS supported by venovenous ECMO (7). We found that low and high airway pressures were associated with increased lung water and higher histologic scores, respectively, compared with a near-apneic protocol using moderate airway pressures (which is the same protocol used in the near-apneic group of the present study).

f ARDS supported by venovenous ECMO (7). We found that low and high airway pressures were associated with increased lung water and higher histologic scores, respectively, compared with a near-apneic protocol using moderate airway pressures (which is the same protocol used in the near-apneic group of the present study). The issue of mechanical ventilation during ECMO has been poorly studied. Most studies published up to now have been surveys (8), observational descriptive studies (9), and noncontrolled studies to assess feasibility or physiologic effects of certain interventions (10). Our study is one of the first efforts to compare different ventilatory strategies during ECMO in a controlled design. The study was planned as a proof of concept regarding the value of resting the lungs by minimizing the energy imposed. We believe the results provide significant evidence in favor of the lung rest concept. The fact that not all the measured variables were modified by the ventilator strategy is completely expected in a 24-hour experimental model comparing clinically relevant strategies. However, histologic lung injury, which is a major component of ARDS, was clearly improved by near-apneic ventilation.

lung rest concept. The fact that not all the measured variables were modified by the ventilator strategy is completely expected in a 24-hour experimental model comparing clinically relevant strategies. However, histologic lung injury, which is a major component of ARDS, was clearly improved by near-apneic ventilation. In the recently published EOLIA, the largest randomized clinical trial to date on venovenous ECMO for severe ARDS, patients assigned to the ECMO group had a reduction in their mechanical power by 2.5 times in relation to the control group (conventional protective protocol) (11). Although this is a significant reduction, if our experimental near-apneic protocol would have been in place, the reduction in mechanical power compared with the control group would have been in the order of 18 times. It is uncertain whether this would have resulted in better clinical outcomes; however, based on our data, we think this should be assessed in future trials. We fully agree with Fan (1) and Kredel and colleagues (2) that several uncertainties remain about the role of prone position, spontaneous breathing, or specific ventilatory variables to achieve the ideal lung rest. While we wait for clinical studies in this area, we will continue addressing these questions via an experimental approach.

th Fan (1) and Kredel and colleagues (2) that several uncertainties remain about the role of prone position, spontaneous breathing, or specific ventilatory variables to achieve the ideal lung rest. While we wait for clinical studies in this area, we will continue addressing these questions via an experimental approach. The story of prone position has taught us that we should not give up sound concepts only because they are old or we have not been able to find their place. Instead, we must learn from our mistakes, refresh the valuable old concepts with new perspectives, and challenge our current approaches. We think that our study, despite all the limitations of an experimental design, is a significant step in that direction. Originally Published in Press as DOI: 10.1164/rccm.201812-2258LE on January 4, 2019 Author disclosures are available with the text of this letter at www.atsjournals.org.

Obstructive sleep apnea (OSA) afflicts 3–9% of women and 10–17% of men in the United States (1) and is associated with a host of comorbid cardiovascular and metabolic conditions, including hypertension, diabetes, coronary artery disease, and stroke. Despite decades of research, it remains unclear which, if any, patients with OSA are at greatest risk for developing these possible complications and which might be safely left untreated. A number of previous epidemiologic and case–control studies identified the greatest risk to be in the group with “severe OSA,” which is to say the group with an apnea–hypopnea index (AHI) > 30 events/h (2, 3). However, recent randomized controlled trials that enrolled patients based on their AHI have not confirmed these findings, leaving investigators and clinicians wondering whether there might not be better ways to stratify risk in patients with OSA (4, 5).

the group with an apnea–hypopnea index (AHI) > 30 events/h (2, 3). However, recent randomized controlled trials that enrolled patients based on their AHI have not confirmed these findings, leaving investigators and clinicians wondering whether there might not be better ways to stratify risk in patients with OSA (4, 5). OSA is characterized by repeated episodes of upper-airway occlusion during sleep, which can be complete (apnea) or incomplete (hypopnea) and of varying duration. Obstructive episodes trigger a number of ensuing pathophysiologic disturbances, including hypoxemia, hypercapnia, sympathoexcitation, and intrathoracic pressure swings. Although there has been some debate on this point, the traditional model of OSA posits that airway occlusion is terminated when the subject experiences an arousal from sleep, thereby restoring pharyngeal dilator muscle tone and opening the airway. Viewed from this perspective, longer apneas or hypopneas must therefore be characterized by some relative failure of the arousal’s normal protective function, either due to inadequate chemostimulation leading to arousal or due to some defect of the arousal response itself. When obstructive episodes last longer, downstream effects such as hypoxemia and sympathoexcitation will be worse, not only because the subject is exposed to them for a longer period of time but also because of their increasing magnitude. Stated plainly, it seems that longer apneas must be physiologically worse than shorter apneas.

tive episodes last longer, downstream effects such as hypoxemia and sympathoexcitation will be worse, not only because the subject is exposed to them for a longer period of time but also because of their increasing magnitude. Stated plainly, it seems that longer apneas must be physiologically worse than shorter apneas. In this issue of the Journal, Butler and colleagues (pp. 903–912) analyze data from the Sleep Heart Health Study and show us that things are not so simple (6). Among 5,712 subjects, almost a quarter of whom died during the 11 years of follow-up, individuals with the shortest-duration obstructive events had a significantly increased risk of death (1.31; 95% confidence interval, 1.11–1.54) after adjusting for the AHI, demographic factors, and prevalent cardiovascular and metabolic disease. The relationship was observed in both men and women, and was strongest in those with moderate OSA.

test-duration obstructive events had a significantly increased risk of death (1.31; 95% confidence interval, 1.11–1.54) after adjusting for the AHI, demographic factors, and prevalent cardiovascular and metabolic disease. The relationship was observed in both men and women, and was strongest in those with moderate OSA. How can we reconcile this observation with our established knowledge of OSA pathophysiology? First, we should recognize that perhaps we have paid too much attention to the presence and morphology of OSA during polysomnography as it relates to downstream effects, and not enough to upstream effects. That is to say, OSA manifests uniquely in a given subject based on a large number of underlying factors, which are only incompletely understood. Obesity and anatomical crowding of the upper airway are obvious predispositions. Chemical control instability and loop gain are less obvious, but are supremely important in the pathogenesis of Cheyne-Stokes respiration and central sleep apnea, and increasingly recognized as contributing to OSA as well (7, 8). Cheyne-Stokes respiration may be a particularly instructive example: although its presence surely predicts a worse prognosis, its treatment does not seem to improve that prognosis (9), possibly because the upstream disturbances that cause the sleep apnea are the relevant ones, not the downstream disturbances that are caused by the sleep apnea.

may be a particularly instructive example: although its presence surely predicts a worse prognosis, its treatment does not seem to improve that prognosis (9), possibly because the upstream disturbances that cause the sleep apnea are the relevant ones, not the downstream disturbances that are caused by the sleep apnea. Similarly, although it is difficult to imagine why a shorter apnea should be more harmful than a longer one when viewed in the light of its lesser downstream effects, it is easy to envision some upstream predisposing host factor that prematurely causes apnea termination (the authors call it “arousability”). It is possible that the mortality risk demonstrated in this study stems not from the shorter apneas per se, but rather from this predisposing factor. If this is the case, then treatment of OSA in these individuals would not be expected to confer a mortality benefit, as treatment would decrease the frequency of apneas and apnea-related arousals, but would not be expected to improve arousability itself. On the other hand, it is possible that increased arousability exerts its effects through sleep fragmentation. However, that explanation is not borne out by the data, as sleep efficiency was not different between the groups. Why might one patient be more arousable than another? Genetic factors surely play a role, and the authors have previously described a genetic locus associated with apnea duration (10). Other predisposing factors, such as frailty, unrecognized physical or mental illness, and sympathetic overactivity, might also contribute.

might one patient be more arousable than another? Genetic factors surely play a role, and the authors have previously described a genetic locus associated with apnea duration (10). Other predisposing factors, such as frailty, unrecognized physical or mental illness, and sympathetic overactivity, might also contribute. The study has some limitations. Because the subjects were drawn from a community-based sample, there was a paucity of severe cases. Only 12% of the subjects had severe OSA, and the mean apnea length in the longest quartile was only 27.8 seconds, which is not long enough to result in the profound hypoxemia that is often seen in clinic populations. It is telling that measures of oxygen saturation (minimum SaO2 and total sleep time SaO2 < 90%) were either not significantly different or only minimally different across the apnea length quartiles. Therefore, these findings should not be generalized to patients with very severe OSA and lengthy apneas resulting in severe hypoxemia. It is always the unexpected result that drives new research and moves the field forward. Increased arousability causing shorter apneas has been believed for decades to be a protective mechanism against the harmful effects of OSA. If shorter apneas actually confer greater risk, then clearly our understanding of OSA is still in its infancy. We do not even know who is friend and who is foe. Originally Published in Press as DOI: 10.1164/rccm.201810-1935ED on October 29, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

The last 30 years of pulmonary hypertension research is a qualified translational success story. Patients with portopulmonary hypertension (PoPH) have benefitted from inclusion in the licensed indications for novel therapies, despite small numbers of patients enrolled in studies (1). There has, however, been very little movement in our parallel understanding of the pathophysiology. Perhaps as a consequence of this gap between evolving treatments aimed primarily at other disease causes, as well as our lack of mechanistic insight in PoPH, outcomes for patients remain poor. Modern registry data demonstrate 5-year survival stuck around 40% for patients with PoPH, in contrast to the improving survival in other disease forms (2). With a paucity of funded research, limited preclinical modeling, and no real external drivers for industry to engage in this question rather than focus resources on subsets of patients in phase 3 trials, there has been little in the way of new hypotheses to consider. An added complication is that patients have two disease processes, pulmonary hypertension and liver disease, and the relationship between the degree and nature of liver disease and splanchnic and pulmonary pressures has not been clearly resolved (3).

ials, there has been little in the way of new hypotheses to consider. An added complication is that patients have two disease processes, pulmonary hypertension and liver disease, and the relationship between the degree and nature of liver disease and splanchnic and pulmonary pressures has not been clearly resolved (3). In this edition of the Journal (pp. 891–902), Nikolic and colleagues report a potentially fundamental advance in our understanding of disease (4). BMP9 (bone morphogenetic protein 9), a ligand of the TGF-β (transforming growth factor-β) superfamily that has a selective binding affinity to the BMPR2 (bone morphogenetic protein receptor type 2)/ALK1 (activin receptor-like kinase 1) complex, is significantly reduced in PoPH but not in other forms of pulmonary arterial hypertension (PAH). BMP9 is emerging as an important and novel regulator of vascular homeostasis (5). The concept that BMP signaling may be important in the liver vasculature has clear precedent. Hereditary hemorrhagic telangiectasia (HHT) is characterized by arteriovenous malformations that affect organs heterogeneously. They are found commonly in the liver, and HHT is associated in around 80% of individuals with mutations in ALK1 and the circulating coreceptor endoglin (6). In addition to the established link between HHT, BMP signaling, and PAH, the genetics of PAH have been pointing for some time to the critical importance of this specific ligand and its receptor complex. Completing the tertiary receptor/ligand complex, mutations in BMPR2 and BMP9 cause PAH (7). Fitting beautifully with the human genetics, the BMPR2/ALK1/endoglin tertiary complex is highly expressed in the pulmonary endothelium, and BMP9 circulates at physiological levels (8). To complete the background story, BMP9 is produced predominantly by the liver (9).

complex, mutations in BMPR2 and BMP9 cause PAH (7). Fitting beautifully with the human genetics, the BMPR2/ALK1/endoglin tertiary complex is highly expressed in the pulmonary endothelium, and BMP9 circulates at physiological levels (8). To complete the background story, BMP9 is produced predominantly by the liver (9). We therefore have a ligand that is highly pulmonary endothelial specific and regulates vascular homeostasis, and PAH develops when there is a reduction in signaling downstream of BMP9 related to rare mutations in all of the ligand receptors and, moreover, the BMP9 ligand itself. We now know that in liver disease, the development of PAH is associated with reduced levels of BMP9, and this was not clearly demonstrated in the context of all patients with liver disease and no PoPH. Critically, BMP9 is not reduced in other forms of PAH, and therefore the reduction in BMP9 is unlikely to be simply a consequence of pulmonary vascular remodeling and secondary to dysregulated homeostasis. Though the number of patients in Nikolic and colleagues’ study (4) with PoPH was modest (n = 28), this was repeatable in two separate cohorts. The underlying causes of the liver disease were heterogeneous, though with a predominance for hepatitis C virus, and were reasonably matched with the liver disease without PoPH group. This seems likely to be a consequence of liver disease itself and therefore relevant to the spectrum of underlying diseases, though further work will be needed to confirm this. There is the tantalizing suggestion that the liver disease with no PH group, though not significantly reduced overall, may have a biphasic or nonnormal distribution, with a small number of patients with low BMP9 (significantly below the 99th percentile in the control group). Future work will have to clarify if the reduction of BMP9 precedes the development of PH and BMP9 is reduced on a spectrum related to extent of liver disease. The authors conclude that this is not the case on the basis of a lack of association with fibrosis scores and the utility of BMP9 in predicting PoPH, particularly in multivariate models over and above classical factors. As they acknowledge in their article, their numbers are small, and as such, these analyses need to be treated with caution.

at this is not the case on the basis of a lack of association with fibrosis scores and the utility of BMP9 in predicting PoPH, particularly in multivariate models over and above classical factors. As they acknowledge in their article, their numbers are small, and as such, these analyses need to be treated with caution. One interesting area not commented on in the report by Nikolic and colleagues (4) is hepatopulmonary syndrome (HPS). HPS is characterized by liver disease, intrapulmonary vasodilation, and hypoxemia. There were only two patients with HPS in this cohort, but further work can clarify if BMP9 is also altered in these patients, who may sit on a spectrum with PoPH. PoPH and HPS share similarities but with wildly different phenotypes. Both syndromes are thought to relate to an imbalance in vasoactive regulators, though in PoPH we see pulmonary vasoconstriction, and in HPS we see the opening up of intrapulmonary shunts. HPS will usually regress with liver transplant, but PoPH classically does not (10). It is not currently clear why patients with liver disease, portal hypertension, and hyperdynamic states can have two apparently diametrically opposed pathologies. Cirrhosis and liver disease are known not just to profoundly affect liver and splanchnic vascular beds but also to have significant systemic vascular effects (11). We know that the effects of mutations downstream of BMP9, notably ALK1 and endoglin in HHT, are not restricted to the pulmonary vasculature. The liver features prominently in arteriovenous malformation, and vascular dysplasia phenotypes, though heterogeneous, are partially influenced by mutation status (6). Hervé and colleagues suggested 20 years ago that a “hepatic factor contained in normal hepatic venous blood plays a role in the control of pulmonary angiogenesis” (12). BMP9 now looks like a strong candidate for this vascular homeostatic role. Looking from the perspective of PAH, if there was any doubt about the centrality of BMPR2 signaling across multiple forms of PAH, this work surely adds another significant brick in the wall.

a role in the control of pulmonary angiogenesis” (12). BMP9 now looks like a strong candidate for this vascular homeostatic role. Looking from the perspective of PAH, if there was any doubt about the centrality of BMPR2 signaling across multiple forms of PAH, this work surely adds another significant brick in the wall. The recent demonstration of a role for BMP10 in zebrafish vascular homeostasis and arteriovenous malformations (13) means that alternative BMP signaling may need to be reevaluated in the context of liver disease. In addition to clarifying an association between PoPH and BMP9 in human disease, Nikolic and colleagues (4) confirm previous work on an orphan model of liver disease–associated pulmonary hypertension: the carbon tetrachloride–induced cirrhotic murine model (14). This underused model will give the field an animal model for mechanistic and therapeutic studies focused on modulating the pathway in the future. The net result of this work is to open up a rich vein of possible research looking at the role of BMPs in pulmonary and systemic vascular homeostasis in liver disease and cementing a model for preclinical animal work to compliment patient studies. The recent demonstrations of efficacy in animal models of PAH using BMP9 therapy and ligand traps for transforming growth factor-β (15, 16) mean that the roadmap to experimental studies in liver disease is both plausible and possible. The authors are to be highly commended for reinvigorating preclinical studies in PoPH. Originally Published in Press as DOI: 10.1164/rccm.201810-1886ED on November 1, 2018

The recent demonstration of a role for BMP10 in zebrafish vascular homeostasis and arteriovenous malformations (13) means that alternative BMP signaling may need to be reevaluated in the context of liver disease. In addition to clarifying an association between PoPH and BMP9 in human disease, Nikolic and colleagues (4) confirm previous work on an orphan model of liver disease–associated pulmonary hypertension: the carbon tetrachloride–induced cirrhotic murine model (14). This underused model will give the field an animal model for mechanistic and therapeutic studies focused on modulating the pathway in the future. The net result of this work is to open up a rich vein of possible research looking at the role of BMPs in pulmonary and systemic vascular homeostasis in liver disease and cementing a model for preclinical animal work to compliment patient studies. The recent demonstrations of efficacy in animal models of PAH using BMP9 therapy and ligand traps for transforming growth factor-β (15, 16) mean that the roadmap to experimental studies in liver disease is both plausible and possible. The authors are to be highly commended for reinvigorating preclinical studies in PoPH. Originally Published in Press as DOI: 10.1164/rccm.201810-1886ED on November 1, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

To the Editor: We are very grateful to Dr. Chalkias and Dr. Xanthos for their thoughtful comments regarding the description of the phenomenon of intrathoracic airway closure reported in the Journal (1). These authors recently reported an impressive series of 300 out-of-hospital patients with cardiac arrest who were resuscitated with a strategy combining rapid intubation, continuous chest compression (CC), and positive pressure ventilation delivered via a ventilator (2). The unexpectedly high percentage of return of spontaneous circulation reported in this study was significantly associated with highest mean airway pressure (Paw) measured after 3 minutes of resuscitation via an external monitor. CO2 measured via a mainstream monitor was similar between survivors and nonsurvivors. The authors concluded that a mean Paw above 42.5 mbar was associated with a higher chance of return of spontaneous circulation.

with highest mean airway pressure (Paw) measured after 3 minutes of resuscitation via an external monitor. CO2 measured via a mainstream monitor was similar between survivors and nonsurvivors. The authors concluded that a mean Paw above 42.5 mbar was associated with a higher chance of return of spontaneous circulation. Interestingly, the apparent negative effect of a low mean Paw during CC could be related to (or associated with) the intrathoracic airway closure we recently reported (1). In fact, the transmission of pressure generated by CC at the airway opening is limited or absent in the case of intrathoracic airway closure. Conversely, the expected beneficial effect of positive pressure delivered by the ventilator (which refers to the thoracic pump effect) can be effective only if the positive airway pressure applied at the airway opening is transmitted to the intrathoracic compartment, although this transmission will be limited by intrathoracic airway closure. The methodological difficulty of capturing the highest value of CO2 that seems the best surrogate of alveolar CO2 during resuscitation limits the interpretation of the lack of difference reported in their study.

o the intrathoracic compartment, although this transmission will be limited by intrathoracic airway closure. The methodological difficulty of capturing the highest value of CO2 that seems the best surrogate of alveolar CO2 during resuscitation limits the interpretation of the lack of difference reported in their study. Therefore, if we accept that the association between a mean Paw below 42.5 mbar and a worse prognosis reported in the study of Chalkias and colleagues might be explained by intrathoracic airway closure, several different mechanisms could still be at play. First, intrathoracic airway closure could simply be a marker of poor prognosis that also limits transmission of pressure generated by CC at the airway opening, making the calculated mean airway pressure at the mouth lower. Second, intrathoracic airway closure may have impaired the transmission of positive pressure generated by ventilation to the intrathoracic compartment, thus limiting its expected beneficial effect on the thoracic pump effect. By overcoming intrathoracic airway closure, higher mean airway pressure could be beneficial on both circulation and ventilation. Finally, one cannot exclude that the lower mean Paw associated with the lower chance of return of spontaneous circulation in the abovementioned study could be simply the reflection of less effective CC, independent of intrathoracic airway closure. These fascinating physiological discussions deserve additional observations to better understand the mechanisms at play and the evolution of airway closure along the time of resuscitation. The use of the capnogram during CC, based on the description of Grieco and colleagues, may permit us to adapt ventilator settings according to intrathoracic airway closure to balance both the beneficial and potential harmful effects of positive airway pressure during resuscitation.

sure along the time of resuscitation. The use of the capnogram during CC, based on the description of Grieco and colleagues, may permit us to adapt ventilator settings according to intrathoracic airway closure to balance both the beneficial and potential harmful effects of positive airway pressure during resuscitation. Originally Published in Press as DOI: 10.1164/rccm.201811-2234LE on January 3, 2019 Author disclosures are available with the text of this letter at www.atsjournals.org.

Exposure to particulate matter and ozone has severe repercussions on public health in the United States and globally. Fine particulate matter (or PM2.5, particulate matter with aerodynamic diameter of 2.5 μm or less) has been associated with exacerbations of asthma and chronic obstructive pulmonary disease, cardiovascular disease mortality, and lung cancer (1–4). Ozone exposure has been linked to increased risks for asthma exacerbations and other respiratory illnesses, as well as to cardiovascular disease mortality (5, 6). Fossil fuel combustion–related emissions of particulate matter and precursors of ozone have been estimated to cause approximately 210,000 premature deaths annually in the United States (7). Further, combustion-related emissions that affect air quality also are the major drivers of climate change. There is new evidence that the effect of climate change on wildfires could double the numbers of premature deaths resulting from fine particulate matter exposure by 2100 (8). The National Ambient Air Quality Standards (NAAQS), established by the U.S. Environmental Protection Agency (EPA) under the Clean Air Act, are the federal standards for air quality levels that are designed to protect the health of even the most vulnerable populations with a margin of safety. As part of the process for establishing the NAAQS, the U.S. EPA reviews and evaluates the “most policy-relevant science, including key science judgments that are important to inform the development of the risk and exposure assessments,” in a process called an Integrated Science Assessment (9).

with a margin of safety. As part of the process for establishing the NAAQS, the U.S. EPA reviews and evaluates the “most policy-relevant science, including key science judgments that are important to inform the development of the risk and exposure assessments,” in a process called an Integrated Science Assessment (9). U.S. EPA review of the science has relied on investigations such as the Six Cities Study, which initially found that exposure to fine particulate matter was associated with premature mortality and later found that reductions in fine particulate matter concentrations with EPA enforcement were associated with a decrease in mortality (10). The U.S. EPA estimated in 2011 that control of particulate matter will result in 230,000 adult lives saved by 2020 (11).

culate matter was associated with premature mortality and later found that reductions in fine particulate matter concentrations with EPA enforcement were associated with a decrease in mortality (10). The U.S. EPA estimated in 2011 that control of particulate matter will result in 230,000 adult lives saved by 2020 (11). Multicity studies have provided evidence for the associations of fine particulate matter and ozone exposures with cardiopulmonary mortality, but not for morbidity. National data on hospitalizations are only available from Medicare, the U.S. system that provides health insurance for adults 65 years and older, and therefore may not be representative of other populations. As noted in this issue of the Journal by Strosnider and colleagues (pp. 882–890), evidence for morbidity effects of these pollutants on populations not covered by Medicare has only been found in single-city studies (12). Reliance on these single-city studies for this purpose has been problematic, however, as the characteristics of air pollution and population characteristics vary by city, along with each study’s methodology, which makes it difficult to pool these studies to form an evidence base that can be generalized to reflect the U.S. population, let alone populations in other countries.

s been problematic, however, as the characteristics of air pollution and population characteristics vary by city, along with each study’s methodology, which makes it difficult to pool these studies to form an evidence base that can be generalized to reflect the U.S. population, let alone populations in other countries. Strosnider and colleagues have accessed a unique database, maintained by the National Environmental Public Health Tracking Network (13), which contains emergency room visit data for conditions related to environmental exposures, such as respiratory illness. Using daily data at the county level from 17 states, representing 45% of the U.S. population, they were able to analyze associations among ozone, fine particulate matter, and respiratory emergency room visits for all ages in this sample. Ambient air pollution levels were modeled and downscaled to census tract centroids and population-weighted to the county level. Using a two-stage process, the authors fit time-series models and then distributed lag models to account for air pollution effects up to 1 week after exposure, adjusting for pollutants, temperature, dew point, and other variables.

led and downscaled to census tract centroids and population-weighted to the county level. Using a two-stage process, the authors fit time-series models and then distributed lag models to account for air pollution effects up to 1 week after exposure, adjusting for pollutants, temperature, dew point, and other variables. In general, the results of Strosnider and colleagues support the U.S. EPA’s determination of a likely causal link between exposures to either fine particulate matter or ozone and respiratory illness (12). However, for the first time, they found that fine particulate matter exposure was more strongly associated with respiratory emergency department visits among children, rather than adults. In contrast, effects were stronger among adults than children for ozone exposure–respiratory emergency department visit relationships. Other pollutant/illness associations they examined also yielded differences by age group. These findings provide evidence that reliance on Medicare data for estimating effects of ozone and fine particulate matter may not accurately estimate effects on populations younger than 65 years. As noted by the authors, age differences in the pollution effects of respiratory illness are supported by varying rates of emergency department visits for respiratory outcomes by age, disease pathology, age-specific patterns of exposure, and even pollutant characteristics.

effects on populations younger than 65 years. As noted by the authors, age differences in the pollution effects of respiratory illness are supported by varying rates of emergency department visits for respiratory outcomes by age, disease pathology, age-specific patterns of exposure, and even pollutant characteristics. There are limitations in any study. Although the authors used a robust approach in their analyses, model misspecification, that is, that their statistical model inadvertently missed important variables or proposed a wrong statistical functional form, could always be a possibility. However, the authors conducted sensitivity analyses that indicated that this is unlikely. Misclassification of the respiratory outcomes, if present, also does not appear to be a likely candidate to bias the study results in a particular direction. Exposure error in the pollution model (but likely random with respect to the outcome) and the national representativeness of the final sample are also concerns that may limit the usefulness of the study. However, the estimates provided here are the best evidence to date of age differences in the effects of ozone and fine particulate matter on respiratory illness, which will be very valuable in assessing whether the NAAQS need to be reevaluated.

final sample are also concerns that may limit the usefulness of the study. However, the estimates provided here are the best evidence to date of age differences in the effects of ozone and fine particulate matter on respiratory illness, which will be very valuable in assessing whether the NAAQS need to be reevaluated. Ideally, the results of this study would be used in U.S. EPA rulemaking, but it is important to note that these data are based on confidential emergency room visit records, which are maintained by state agencies and were aggregated to protect patient privacy. The currently proposed transparency (so-called “Secret Science”) rule by the U.S. EPA would prohibit regulators from considering studies for review in rulemaking that do not provide the underlying data (14). If this rule is approved, then it is possible that these study findings might not be taken into consideration for any changes in the NAAQS on fine particulate matter or ozone, as they are based on personal patient data. In fact, the seminal Six Cities Study results on fine particulate matter and mortality would probably also not be able to be used under the proposed rule. This would be unfortunate, as the best available science would not be included in U.S. EPA’s rulemaking, and thus would not provide the best NAAQS to protect the nation’s public health. Other recent actions by the U.S. EPA, including disbanding expert science panels that made recommendations on new standards for ozone and particulate matter, as well as moves to eliminate consideration of health cobenefits in calculating costs of pollution rule-making, also bring into question the agency’s commitment to defend the public from exposure to harmful levels of these contaminants.

panels that made recommendations on new standards for ozone and particulate matter, as well as moves to eliminate consideration of health cobenefits in calculating costs of pollution rule-making, also bring into question the agency’s commitment to defend the public from exposure to harmful levels of these contaminants. Originally Published in Press as DOI:10.1164/rccm.201811-2106ED on November 28, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

To the Editor: We would like to congratulate Scales and Kavanagh (1) for their insightful comments reported in the editorial accompanying the study by Grieco and colleagues (2). It is true that research on resuscitation made early gains, but recent progress has been slow because of the dispersion of researchers to aspects other than elucidating the physiology and pathophysiology of cardiac arrest and resuscitation. Although our understanding of the interaction between chest compression and mechanical ventilation remains limited, expert opinions will probably continue to rely on flawed studies that neither report nor take into consideration, when interpreting the results, the method of postintubation ventilation (self-inflating bag or ventilator), while suggesting simultaneously that early intubation during cardiopulmonary resuscitation (CPR) does not improve, or even decreases, survival (3). Ventilation with a self-inflating bag in intubated patients usually results in excessive ventilation volume and rate, thus aggravating oxygenation and hemodynamics, and surprisingly, it continues to be a major limitation in resuscitation studies.

ry resuscitation (CPR) does not improve, or even decreases, survival (3). Ventilation with a self-inflating bag in intubated patients usually results in excessive ventilation volume and rate, thus aggravating oxygenation and hemodynamics, and surprisingly, it continues to be a major limitation in resuscitation studies. Cordioli and colleagues (4) demonstrated that ventilation during CPR by using currently recommended chest compression rates takes place entirely below FRC and is associated with negative intrathoracic pressures during decompression. Although the thoracic pump theory is not widely accepted among the resuscitation community, the study of Cordioli and colleagues suggests that both cardiac pump and thoracic pump have a role in forward blood flow and tissue oxygenation. In this context, the study by Grieco and colleagues (2) strengthens the evidence-based notion that the harmony between circulation and ventilation during CPR is critical. Achieving the correct balance between too little and too much ventilation is of major importance for optimizing survival, and theoretically, there must be an intrathoracic pressure limit at which the effect of a thoracic pump should be maximal. Above this limit, intrathoracic pressure would be deleterious, and under this limit, ventilation may not provide adequate blood oxygenation because of small airway closure, increasing pulmonary vascular resistance and impairing pulmonary and systemic blood flow.

at which the effect of a thoracic pump should be maximal. Above this limit, intrathoracic pressure would be deleterious, and under this limit, ventilation may not provide adequate blood oxygenation because of small airway closure, increasing pulmonary vascular resistance and impairing pulmonary and systemic blood flow. Our group has recently shown an association between mean airway pressure and outcome of CPR in mechanically ventilated patients, with a value of 42.5 mbar being associated with return of spontaneous circulation (5). In our patients, simultaneous positive pressure ventilation in time with each chest compression prevented a loss of intrathoracic pressure via the airway, and probably kept the small airways open. In this study, we found no difference in end-tidal carbon dioxide between survivors and nonsurvivors, probably because of the maintenance of flow in small airways and the improvement in minute-volume ventilation during CPR (6). Of note, the rise in intrathoracic pressure in mechanically ventilated patients undergoing CPR is transmitted equally to all intrathoracic structures and squeezes out the pulmonary vessels, which increases forward blood flow, arterial oxygen partial pressure, and aortic pressure. Moreover, as hemodynamics may be aggravated in prolonged CPR because of vascular tone deterioration, the pressing effect of positive pressure ventilation and increased intrathoracic pressure on aortic wall may increase aortic resistance and retrograde volume loading, therefore enhancing the compression-related blood flow (5).

r, as hemodynamics may be aggravated in prolonged CPR because of vascular tone deterioration, the pressing effect of positive pressure ventilation and increased intrathoracic pressure on aortic wall may increase aortic resistance and retrograde volume loading, therefore enhancing the compression-related blood flow (5). Collectively, the study by Grieco and colleagues and our findings highlight the favorable effects of the thoracic pump and the importance of intubation and mechanical ventilation in patients with cardiac arrest, supporting our deduction that the interplay between ventilation and chest compression during CPR is a key point to optimize outcomes (6). As proper timing of compression and ventilation seems to be the key for improving the circulation, the focus of the resuscitation community must immediately return to the elucidation of the physiology and pathophysiology of cardiac arrest and resuscitation. Originally Published in Press as DOI: 10.1164/rccm.201811-2075LE on January 3, 2019 Author disclosures are available with the text of this letter at www.atsjournals.org.

The lack of a specific medical therapy for sepsis, a dysregulated response to infection that is responsible for up to half of all inpatient deaths (1), plagues critical care. Numerous clinical trials have failed to improve mortality (2, 3). Furthermore, the failure of many anticytokine therapies challenges the classic paradigm of observing an association between plasma levels of a purported marker and sepsis mortality, testing the marker’s causality and modifiability in animal models, and then moving toward clinical trials to test marker blockade. Correlation does not equate with causation, and strategies to modify a correlated but noncausal biomarker are unlikely to improve sepsis survival. Although controlled interventional trials provide strong evidence for causality, it is frequently unethical or impractical to randomize subjects to high- or low-biomarker infusions, leaving us to bridge this gap with observational designs. Our field needs smarter tools to dissect correlation from causality and identify the causal biomarkers of sepsis to speed the development of sepsis therapy.

ity, it is frequently unethical or impractical to randomize subjects to high- or low-biomarker infusions, leaving us to bridge this gap with observational designs. Our field needs smarter tools to dissect correlation from causality and identify the causal biomarkers of sepsis to speed the development of sepsis therapy. Fortunately, tools to infer causality from observational data do exist. Classically, such methods have been applied to avoid making policy decisions based on biased or inconsistent association estimates (4) due to measurement error, uncontrolled confounding, or reverse causation. One potential solution is to use an instrumental variable analysis. This approach is valid if the instrument, or reliably measured variable, has a strong association with a potential mediator variable, and there is no correlation between the instrument and the outcome being studied (5). For example, if distance from a grocery store reliably predicts intake of fresh produce, then the association between grocery store distance and lean weight can be used to infer whether produce intake has a causal relationship with weight. Genetic researchers extended this approach by using genotype(s) to predict biomarkers, testing the association between biomarker-predicting genotypes and disease, and inferring whether the biomarker has a causal relationship with disease. Called Mendelian randomization (MR) analysis, this genetic instrumental variable strategy is attractive because genotypes are assigned at random by gametogenesis and genotype assignment always precedes outcome. Such MR analyses have provided evidence that low-density lipoprotein cholesterol (LDL-C) plasma concentrations are causally related to cardiovascular disease and mortality (6), whereas markers such as C reactive protein are not (7), thus focusing therapeutic efforts on modifying plasma LDL-C. Furthermore, MR has enabled the identification of novel genetic regulators of LDL-C, which has translated to a new class of lipid-lowering agents (8).

related to cardiovascular disease and mortality (6), whereas markers such as C reactive protein are not (7), thus focusing therapeutic efforts on modifying plasma LDL-C. Furthermore, MR has enabled the identification of novel genetic regulators of LDL-C, which has translated to a new class of lipid-lowering agents (8). Could similar strategies be applied to identify key causal intermediates for sepsis death? In this issue of the Journal, Trinder and colleagues (pp. 854–862) implicate serum high-density lipoprotein cholesterol (HDL-C) as a potentially causal contributor to sepsis survival, and suggest that medications that boost HDL deserve investigation for sepsis (9). The foundation for this work was the group’s prior observation that low HDL-C was a strong predictor of organ dysfunction or death among patients presenting to an emergency room with suspected sepsis (10). Because HDL-C can bind and sequester pathogen lipids, including endotoxin, patients with lower HDL-C may have worse sepsis outcomes. The authors used an astute approach to identify genetic predictors of serum HDL-C and performed an MR analysis of the effect of HDL-C on sepsis survival. First, they performed targeted resequencing of 10 HDL-C–associated genes in 200 subjects with suspected sepsis, focusing on SNPs that influence coding sequence or splicing. For each candidate gene, they tested whether subjects with low HDL-C had an excess of coding SNPs compared with subjects with normal or high HDL-C, and the gene CETP—encoding CETP (cholesteryl ester transfer protein)—was the only one to demonstrate an HDL-C association during sepsis. Furthermore, one missense CETP SNP, rs1800777, drove the association between CETP, HDL-C, and increased sepsis-related organ failures. The SNP seems to be a CETP gain-of-function variant, with rs1800777 carriers exhibiting higher plasma CETP activity. Because the sequencing was performed in the same 200 subjects in whom the authors first reported an association between low HDL-C and sepsis death, raising concerns about selection bias and generalizability, the authors validated that rs1800777 was associated with decreased sepsis survival in two additional sepsis populations. Finally, the authors used rs1800777 as a genetic instrument to predict a portion of HDL-C variance.

ion between low HDL-C and sepsis death, raising concerns about selection bias and generalizability, the authors validated that rs1800777 was associated with decreased sepsis survival in two additional sepsis populations. Finally, the authors used rs1800777 as a genetic instrument to predict a portion of HDL-C variance. By MR analysis, each log decrease in genetically predicted HDL-C during early sepsis was associated with an increase in the adjusted hazard ratio for mortality, leading to the causal inference that lower serum HDL-C during sepsis has a causal effect on reduced sepsis survival. This study has several strengths, including its sophisticated design to test suspected functional genetic variants via a sequencing approach. By focusing on genome-wide association study–validated loci that influence HDL-C, the authors were more likely to discover a strong relationship between genotype and HDL-C, and they showed the SNP’s gain-of-function action by testing plasma levels of CETP activity. The consistency of the SNP’s association with both HDL-C and sepsis organ failure and survival in multiple populations lends confidence that rs1800077 is a valid genetic instrument for making a causal inference about HDL-C. Most importantly, by establishing serum HDL-C as a potential causal intermediate in sepsis survival, this study introduces HDL-C modification as a highly novel therapeutic strategy for sepsis, which is an exciting concept in that agents to inhibit CETP already exist.

nstrument for making a causal inference about HDL-C. Most importantly, by establishing serum HDL-C as a potential causal intermediate in sepsis survival, this study introduces HDL-C modification as a highly novel therapeutic strategy for sepsis, which is an exciting concept in that agents to inhibit CETP already exist. Trinder and colleagues acknowledge that although their genetic instrument meets validity criteria, it is rare: only 10 of the 200 subjects in the early infection cohort carried this SNP, and small sample sizes are at risk for unstable effect estimates. However, the validation of the SNP–mortality association in additional sepsis populations is reassuring. HDL-C is less established than other potential sepsis prognostic biomarkers, and thus it will be important to ensure the consistency of this association in much larger populations. Finally, the authors acknowledge some inconsistencies in the data supporting a strategy of CETP blockade in sepsis, including worrisome observations of increased infections and excess mortality in randomized trials of one CETP inhibitor, torcetrapib, for coronary arterial disease (11). In addition, the recent disappointing results of the EUPHRATES (Evaluating the Use of Polymyxin B Hemoperfusion in a Randomized controlled trial of Adults Treated for Endotoxemia and Septic shock) trial, which randomized subjects with septic shock and elevated endotoxin activity assays to a hemofiltration therapy targeted at reducing endotoxin activity (12), likewise dampens enthusiasm for the notion that modifying endotoxin availability is a helpful approach in sepsis.

Treated for Endotoxemia and Septic shock) trial, which randomized subjects with septic shock and elevated endotoxin activity assays to a hemofiltration therapy targeted at reducing endotoxin activity (12), likewise dampens enthusiasm for the notion that modifying endotoxin availability is a helpful approach in sepsis. Although it remains to be seen whether CETP inhibition might be a viable therapeutic option in sepsis, the study by Trinder and colleagues is nonetheless a robust example of employing genomic and statistical tools in observational clinical cohorts to identify novel therapeutic targets in sepsis. Similar approaches should be embraced by investigators in our field, with dedicated attempts to replicate prior findings while generating new discoveries. The validation of causal intermediates should accelerate translation from observation to safe, testable interventions, ideally leading to improved sepsis therapy. Originally Published in Press as DOI: 10.1164/rccm.201811-2084ED on November 14, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

More than 30 years ago, Dr. Cole described the “vicious circle” of host-mediated, inflammatory tissue damage, a paradigm of bronchiectasis pathogenesis in which an amplified, poorly controlled, and chronic inflammatory response to an “incursion” occurs (1). Since then, substantial progress has been made in elucidating aspects of this inflammation, most notably the aberrant behavior of neutrophil inflammation in bronchiectasis (2, 3). In this issue of the Journal, Mac Aogáin and colleagues (pp. 842–853) shift the focus from neutrophilic inflammation to the connection between atopy and sensitization (4). This connection has been developed in respiratory diseases such as asthma, chronic obstructive pulmonary disease, and cystic fibrosis (CF). Outside of allergic bronchopulmonary aspergillosis, little attention has been given to its presence in bronchiectasis. Investigation of atopy in bronchiectasis (5) predates the publication of Cole’s “vicious circle,” but only sparse evidence to support its role has been obtained (6).

disease, and cystic fibrosis (CF). Outside of allergic bronchopulmonary aspergillosis, little attention has been given to its presence in bronchiectasis. Investigation of atopy in bronchiectasis (5) predates the publication of Cole’s “vicious circle,” but only sparse evidence to support its role has been obtained (6). The current study cohort is unique: subjects were derived from the CAMEB (Cohort of Asian and Matched European Bronchiectasis) study, a collaboration between bronchiectasis researchers in Asia and Dundee, Scotland. This is perhaps an unexpected collaboration, given the geographic disparity, but it is testament to the fact that the collegiality of the bronchiectasis community spans continents. Patients with stable bronchiectasis from Singapore and Kuala Lumpur, Malaysia (SG-KL), were matched by age, sex, and bronchiectasis severity score with patients with stable bronchiectasis from Dundee, Scotland (DD). After excluding confounding etiologies (allergic bronchopulmonary aspergillosis, asthma, chronic obstructive pulmonary disease, and CF), the investigators measured sensitization to multiple environmental allergens such as the house dust mite (HDM) and recombinant allergens of Aspergillus fumigatus (rAsp f) and Alternaria alternata (Alt a). A separate cohort of 149 patients with allergic rhinitis but no bronchiectasis served as a high-allergic-sensitization comparator group.

sured sensitization to multiple environmental allergens such as the house dust mite (HDM) and recombinant allergens of Aspergillus fumigatus (rAsp f) and Alternaria alternata (Alt a). A separate cohort of 149 patients with allergic rhinitis but no bronchiectasis served as a high-allergic-sensitization comparator group. There are three distinct results to be highlighted. First, a surprisingly high frequency of sensitization was identified in the combined Asian and European cohort of patients with bronchiectasis: 57.6% mounted class 3 or higher sensitization to at least one allergen (significantly higher than the 26.9% observed for the comparator allergic rhinitis group). HDMs were the most common allergens to incite sensitization, and the highest IgE titers were seen in patients with sensitivity to two or more allergens. Second, there appeared to be a geographical influence on sensitivity: subjects from the SG-KL contingent manifested higher responses to HDM allergens and a single Aspergillus allergen, rAsp 1, whereas the DD contingent manifested higher responses to the A. alternaria and Aspergillus allergens, with the exception of rAsp 1. These associations held up in the SG-KL group after matched analysis, whereas the DD group maintained a significantly higher sensitization to just one of the fungal allergens: rAsp17.

p 1, whereas the DD contingent manifested higher responses to the A. alternaria and Aspergillus allergens, with the exception of rAsp 1. These associations held up in the SG-KL group after matched analysis, whereas the DD group maintained a significantly higher sensitization to just one of the fungal allergens: rAsp17. Third, Mac Aogáin and colleagues correlated the level of sensitization patterns to clinical outcomes such as lung function and the bronchiectasis severity index. Looking at the CAMEB as a whole, patients with sensitization to three or more allergens had the poorest lung function and greatest disease severity. Taking into account geographic patterns of sensitization, poorer lung function was noted in SG-KL subjects with sensitivity to HDM and in DD subjects with sensitization to rAsp f 1. Additionally, SG-KL subjects who demonstrated sensitization to rAsp f 17, albeit at a lower frequency than the DD cohort, had more frequent exacerbations. This signal was not seen in the overall group when both geographic groups were analyzed on the basis of high sensitization levels, and higher sensitization to three or more allergens was not associated with exacerbation frequency.

sp f 17, albeit at a lower frequency than the DD cohort, had more frequent exacerbations. This signal was not seen in the overall group when both geographic groups were analyzed on the basis of high sensitization levels, and higher sensitization to three or more allergens was not associated with exacerbation frequency. Further analyzing sensitization patterns, Mac Aogáin and colleagues applied multiplex sputum cytokine and chemokine profiling to the sensitization and clinical phenotype patterns to yield “immunoallertypes,” demonstrating both fungalus- and HDM-driven patterns. Using hierarchical cluster analysis, they identified two distinct immunoallertypes that were independent of geography and instead were characterized by distinct sensitization patterns and immune profiles. First, patients with high sensitization to HDM allergens exhibited a chemokine-dominant airway profile with high GRO (CXCL1), MCP-1 (CCL2), and eotaxin-1 (CCL11) in addition to anti-inflammatory elements (IL-1RA, IL-10, and G-CSF). Meanwhile, the fungal-driven proinflammatory group with marked responses to the fungal allergens was characterized by elevated airway tumor necrosis factor-α, IL-1α, and IL-1β. The fungal proinflammatory group demonstrated worse disease as indicated by the bronchiectasis severity index and lower FEV1. Next, Mac Aogáin and colleagues used a Markov blanket approach to identify features shared among sensitization pattern, immune cytokine/chemokine profile, and geographic origin to reveal “intra-immunoallertypes.” By this route, they were able to identify subgroups of immune signatures unique to a geographic origin. This important finding provides the first data to suggest that the clinical heterogeneity of bronchiectasis may be due to specific immune profiles. The authors’ Table 1 presents sensitization patterns, immune profiles, and notable clinical outcomes in a consolidated fashion.

ne signatures unique to a geographic origin. This important finding provides the first data to suggest that the clinical heterogeneity of bronchiectasis may be due to specific immune profiles. The authors’ Table 1 presents sensitization patterns, immune profiles, and notable clinical outcomes in a consolidated fashion. Overall, this comprehensive study produces a multiplicity of results that are complex, if not somewhat heterogeneous themselves, but this may be unavoidable at this early stage of endotyping a disease that is more syndrome than disease and includes heterogeneity in its definition. Longitudinal data would be informative with regard to whether sensitization has a role in bronchiectasis pathogenesis or the bronchiectasis state predisposes the host to atopy. The study is notable on several levels. The international collaboration is, again, remarkable. The study moves the field of bronchiectasis forward in two significant ways: first, it describes atopy and sensitization in bronchiectasis on a large, multicenter scale that has not been done before. Second, it identifies sensitization as an “endophenotype” of bronchiectasis. The need for endophenotypes in bronchiectasis was born out of failed trials that sent a painful but clear message that non-CF bronchiectasis not only is not CF but also will not behave like CF in response to therapies with proven efficacy in CF. These data may in part explain the baffling failures of large-scale therapeutic trials to demonstrate improvement in exacerbation frequency from inhaled antibiotics despite a decrease in bacterial burden. With further study, the sensitization patterns and immune profiles identified in this work will mature as guides to categorization and therapy. No doubt Mac Aogáin and colleagues have gotten off to a productive start.

improvement in exacerbation frequency from inhaled antibiotics despite a decrease in bacterial burden. With further study, the sensitization patterns and immune profiles identified in this work will mature as guides to categorization and therapy. No doubt Mac Aogáin and colleagues have gotten off to a productive start. Originally Published in Press as DOI: 10.1164/rccm.201810-1949ED on October 25, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

To the Editor: In cases in which pulmonary gas exchange is mainly guaranteed by extracorporeal support, the optimal ventilation strategy to protect the lung remains unclear. It is generally accepted that the ventilator should be set to prevent further ventilator-associated lung injury. Nevertheless, even a lung-protective approach with low Vts may still aggravate lung injury. Thus, an ultraprotective approach with very low Vts (<6 ml/kg) is frequently used in patients undergoing extracorporeal support to facilitate the healing of the injured lung (1). A very interesting concept is the reduction of the Vts to near apneic oxygenation, as done by Araos and colleagues (2). These researchers examined three different ventilation strategies in a swine acute respiratory distress syndrome model over the course of 24 hours, using extracorporeal membrane oxygenation to examine nonprotective, conventional, and near-apneic ventilation. The researchers found that histopathologic lung injury was lower in the conventional and especially the near-apneic group. However, wet-dry lung weight ratio and expression of most genes indicating fibroproliferation were not different between the groups. As remarked in the editorial by Fan (3), there was no comparison of ultraprotective strategies, and the three strategies differed not only in their Vts but also in positive end-expiratory pressure (PEEP) level and respiratory rate. Fan raised the question whether ventilation is needed at all during extracorporeal lung support. This was primarily described by Kolobow in an animal study (4).

aprotective strategies, and the three strategies differed not only in their Vts but also in positive end-expiratory pressure (PEEP) level and respiratory rate. Fan raised the question whether ventilation is needed at all during extracorporeal lung support. This was primarily described by Kolobow in an animal study (4). Our group also conducted a study using a similar acute respiratory distress syndrome model (5). In the conventional group, protective mechanical ventilation with 6 ml/kg Vt was used. Unlike Araos and colleagues, we used arteriovenous extracorporeal lung assist to reduce Vts to 3 ml/kg body weight, and apneic oxygenation with Vts set to zero in further experimental groups. Moreover, an “open lung concept” was used in all groups by using PEEP levels above the lower inflection point of the lung. This strategy resulted in continuous airway pressure above 20 cm H2O, even in the apneic group. Mean respiratory rate was similar in the 6 ml/kg and the 3 ml/kg group, with 20 and 17–18 breaths/min, respectively. After 24 hours, a histopathologic examination of the dependent lung showed more inflammation, alveolar exudation, and atelectasis with 3 ml/kg or no Vts. In contrast, alveolar overdistension was reduced with apneic oxygenation in the nondependent lung areas (5).

ml/kg group, with 20 and 17–18 breaths/min, respectively. After 24 hours, a histopathologic examination of the dependent lung showed more inflammation, alveolar exudation, and atelectasis with 3 ml/kg or no Vts. In contrast, alveolar overdistension was reduced with apneic oxygenation in the nondependent lung areas (5). Hence, our study addressed several of the shortcomings of the data presented by Araos and colleagues and may help to answer the questions raised by Fan (3). Ventilation with protective Vts led to overdistension in the nondependent lung. Nevertheless, despite using high positive airway pressures, the dependent lung in the apneic group showed a worse lung injury score compared with protective Vts. Thus, the combination of both strategies as “near apneic ventilation with low respiratory rates” and higher PEEP levels might be very appealing. This strategy might prevent derecruitment of the dependent lung via repeated recruitment at a low rate set above higher PEEP levels. Overdistension of the nondependent lung may be prevented because of lower peak pressures and minimized shear stress resulting from a low respiratory rate. Another point is that using lower airway, and thus intrathoracic, pressures might reduce hemodynamic compromise. This is enabled by lower respiratory rates and lower Vts. Theoretically, a strategy with sufficient PEEP, low respiratory rates, and very low Vts individually adapted to the size of the residual nonconsolidated lung parts combined with prone positioning might be optimal to protect the lung during extracorporeal lung support.

s enabled by lower respiratory rates and lower Vts. Theoretically, a strategy with sufficient PEEP, low respiratory rates, and very low Vts individually adapted to the size of the residual nonconsolidated lung parts combined with prone positioning might be optimal to protect the lung during extracorporeal lung support. We strongly agree with Fan that the optimal ventilator strategy during extracorporeal gas exchange should now be addressed in clinical studies. Originally Published in Press as DOI: 10.1164/rccm.201810-1985LE on January 4, 2019 Author disclosures are available with the text of this letter at www.atsjournals.org.

Cystic fibrosis (CF) lung disease is characterized by a vicious cycle of mucus secretion, airway infection, and inflammation. Neutrophils are the primary inflammatory cell involved in this process and are recruited from the blood into the airway lumen early in the disease process, as demonstrated by BAL studies performed in infants with CF (1, 2). These neutrophils contain an array of inflammatory mediators, oxidants, and proteases that are critical for antimicrobial defense. Large amounts of these enzymes escape from neutrophils in cell death and during phagocytosis, and directly damage the airway epithelium. Another mechanism is the release of enzyme through exocytosis, but the mechanisms that control the degranulation and release of these enzymes are less well understood (2). One particular enzyme, neutrophil elastase (NE), is capable of digesting diverse proteins and contributes to the progression of structural lung disease (3, 4). Higher levels of free extracellular NE in sputum have been shown to predict subsequent lung function decline (5). The antiprotease defenses in the airways are designed to neutralize free proteases such as NE and prevent their damaging effects. However, these defenses are eventually overwhelmed and degraded by the protease burden in the lung (6).

cellular NE in sputum have been shown to predict subsequent lung function decline (5). The antiprotease defenses in the airways are designed to neutralize free proteases such as NE and prevent their damaging effects. However, these defenses are eventually overwhelmed and degraded by the protease burden in the lung (6). In this issue of the Journal, Margaroli and colleagues (pp. 873–881) advance our understanding of neutrophilic inflammation in the early CF lung (7). This cross-sectional study included children with CF under 3 years of age with evidence of early changes of CF lung disease as measured by the computed tomography (CT) Perth-Rotterdam annotated grid morphometric analysis for CF (PRAGMA-CF) scoring method, but virtually no bronchiectasis. As expected, BAL fluid (BALF) samples obtained the same day the CT was performed showed neutrophilic inflammation. Findings in the CF cohort were compared with disease control subjects recruited from the Aerodigestive Clinic who were undergoing bronchoscopy for a clinical indication and also showed evidence of neutrophilic inflammation, albeit less marked than that observed in the CF group. The airway neutrophils in individuals with CF demonstrated increased expression of surface markers reflecting hyperexocytosis of NE-rich granules into the airway lumen. This phenotype was seen in the airway neutrophils but not in blood neutrophils, and was not observed in control patients. This suggests that the inflammatory milieu of the CF airway recruits neutrophils from the blood and stimulates them to adopt an activated state. This neutrophil phenotype with hyperactive exocytosis of NE-rich granules correlated with early structural lung damage by CT coupled with the PRAGMA-CF scoring system. Therefore, this distinguishing feature of neutrophilic airway inflammation in CF could potentially be a key process in early CF lung disease.

ted state. This neutrophil phenotype with hyperactive exocytosis of NE-rich granules correlated with early structural lung damage by CT coupled with the PRAGMA-CF scoring system. Therefore, this distinguishing feature of neutrophilic airway inflammation in CF could potentially be a key process in early CF lung disease. By contrast, other biomarkers of neutrophilic inflammation, such as the BAL neutrophil percentage and free extracellular NE activity (measured by a sensitive Förster resonance energy transfer–based assay) did not correlate with structural changes on CT. This is contrary to previously reported work of AREST-CF (Australian Respiratory Early Surveillance Team for Cystic Fibrosis) group regarding the role of free NE, which predicted the subsequent development of bronchiectasis (4, 8). Potential reasons for this discrepancy may be the use of a more sensitive assay for free NE in the current study compared with the group’s previous studies (4, 8), and the comparison of free NE with the sensitive PRAGMA-CF score, which has been reported to be more sensitive for detecting early structural lung abnormalities (9). In addition, the cross-sectional design of the current study, with structural changes detected at a single time point, may not be reflective of the impact of released NE and the dynamics of airway inflammation over time.

which has been reported to be more sensitive for detecting early structural lung abnormalities (9). In addition, the cross-sectional design of the current study, with structural changes detected at a single time point, may not be reflective of the impact of released NE and the dynamics of airway inflammation over time. Although the data on neutrophil exocytosis are novel, this study has limitations. First, this was not a specifically designed prospective study; patients were drawn from different cohorts with different inclusion criteria. This may have resulted in a more heterogeneous study population and introduced variability into the data. Second, the control group was small and contained subjects with a mixture of underlying diagnoses. Ideally, individuals with CF would be compared with patients with a disease process also characterized by neutrophilic airway inflammation and structural lung damage over time, such as primary ciliary dyskinesia. Third, clinical data such as the temporal relationship between testing and episodes of increased respiratory symptoms, antibiotic use, respiratory microbiology, or functional measures of lung function (e.g., the lung clearance index) are not available. Fourth, BALF only reflects the inflammatory process the airway lumen and not in the airway wall, where structural remodeling takes place. Finally, surface expression assessed by flow cytometry may not be a direct representation of exocytosis, and a functional assay may be better suited to reflect the impact of exocytosis on the inflammatory process in the airways.

ocess the airway lumen and not in the airway wall, where structural remodeling takes place. Finally, surface expression assessed by flow cytometry may not be a direct representation of exocytosis, and a functional assay may be better suited to reflect the impact of exocytosis on the inflammatory process in the airways. Biomarkers to detect neutrophilic inflammation could be useful for tracking the progression of airway pathology over time and be included as outcome measures in interventional trials. The results of this study raise the question as to whether neutrophil exocytosis could serve as a biomarker for airway inflammation in CF. However, bronchoscopy for BALF is an invasive procedure and impractical for repeated measurements in the clinical setting. A noninvasive test of neutrophil exocytosis would be required before this biomarker could be translated to the clinical setting.

exocytosis could serve as a biomarker for airway inflammation in CF. However, bronchoscopy for BALF is an invasive procedure and impractical for repeated measurements in the clinical setting. A noninvasive test of neutrophil exocytosis would be required before this biomarker could be translated to the clinical setting. Neutrophil exocytosis–specific inhibitors with antiinflammatory activity have been developed and tested in animal models (10). Could these small-molecule drugs have therapeutic potential in CF lung disease? Although this is an enticing prospect, our understanding of the process of neutrophil degranulation is still in its infancy; therefore, it is difficult to predict whether the neutrophil activation process would be an appropriate therapeutic target. If the major driver for enzyme release is cell death rather than exocytosis from live neutrophils, then pursuing the antiprotease shield mechanism would likely be a more successful approach. Ultimately, longitudinal studies directly assessing neutrophil degranulation will be required to determine exactly how NE and other neutrophil-derived products damage the airways, to better define the best targets for antiinflammatory treatment in patients with CF. Originally Published in Press as DOI: 10.1164/rccm.201810-1951ED on November 5, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

There is a paradox in the field of pulmonary rehabilitation (PR). There is now vast literature showing evidence that PR is safe, effective, and cost-effective (1, 2). Furthermore, PR improves exercise tolerance, reduces dyspnea, and enhances quality of life likely better than any other available therapy (3), and has been shown to shorten hospital admissions in chronic obstructive pulmonary disease (COPD) (4). Despite the documented benefits of PR, the increasing prevalence of COPD, and the availability of Medicare and other insurance coverage, there is mounting concern that poor PR reimbursement in the United States may accelerate the decline in PR availability, further jeopardizing the limited availability of a key intervention in chronic lung disease. A recent analysis demonstrated that only ∼3% of Medicare beneficiaries with COPD receive PR (5). In the United States, it would seem as though PR itself requires rehabilitation. We believe that the core problem in PR in the United States has been insufficient funding. The full effect of low reimbursement is hard to know with precision, but it has the potential to influence availability of what has been repeatedly acknowledged as the standard of care in chronic lung disease. Fortunately, there is a mechanism that is available to ATS members to rectify the situation. What follows is a trip through the weeds of how Medicare determines reimbursement, and suggested actions for the ATS and its members.

ity of what has been repeatedly acknowledged as the standard of care in chronic lung disease. Fortunately, there is a mechanism that is available to ATS members to rectify the situation. What follows is a trip through the weeds of how Medicare determines reimbursement, and suggested actions for the ATS and its members. We believe that ATS members need to consider the information provided here and lend their efforts as advocates, much as the NHLBI Action Plan for COPD prescribes. There are only 220 ATS members who list PR as their primary assembly, only 72 of which are based in the United States. However, there are potentially thousands of ATS members who prescribe PR and whose patients will benefit from increased funding and availability of PR, but only if we all pull together, as you are about to learn.

nly 220 ATS members who list PR as their primary assembly, only 72 of which are based in the United States. However, there are potentially thousands of ATS members who prescribe PR and whose patients will benefit from increased funding and availability of PR, but only if we all pull together, as you are about to learn. The causes of the decline and stagnation of PR reimbursement in the United States are complex. The decline is at least in part tied to a change in Medicare PR reimbursement in 2010, when a new bundled payment code (healthcare common procedure coding system [HCPCS]), G0424, for COPD was introduced. This was accomplished through many years of lobbying by professional organizations, including the American College of Chest Physicians, the American Association of Cardiovascular and Respiratory Care, the National Association for Medical Directors of Respiratory Care, and the American Association for Respiratory Care. The growing evidence of clinical effectiveness of PR was highlighted by favorable outcomes for the patient group receiving PR in NETT (National Emphysema Therapy Trial) (6). These and a plethora of other data provided the scientific foundation for adoption of the National Coverage Determination Policy by the Centers for Medicare & Medicaid Services (7).

fectiveness of PR was highlighted by favorable outcomes for the patient group receiving PR in NETT (National Emphysema Therapy Trial) (6). These and a plethora of other data provided the scientific foundation for adoption of the National Coverage Determination Policy by the Centers for Medicare & Medicaid Services (7). The new code, G0424, applied nationally, pays for 1 hour of PR, including all costs of staff, medical director, gym, hospital overhead, and so on, and has a 72-visit lifetime cap. G0424 was implemented to cover patients with Global Initiative for Chronic Obstructive Lung Disease spirometry stages 2, 3, and 4 COPD. A separate set of codes may be used to cover patients with non-COPD lung disease. Other than cardiac rehabilitation, G0424 is the only therapy designation in which individuals can be treated in a group setting.

over patients with Global Initiative for Chronic Obstructive Lung Disease spirometry stages 2, 3, and 4 COPD. A separate set of codes may be used to cover patients with non-COPD lung disease. Other than cardiac rehabilitation, G0424 is the only therapy designation in which individuals can be treated in a group setting. In 2010, Medicare initially arbitrarily established a payment rate of approximately $50.46 for 1 unit (1 h) of G0424. Medicare acknowledged in 2011 that failure to carefully construct the charge for G0424 that reports a combination of services previously reported separately under-represents the cost of providing the service described by G0424 and can have significant adverse impact on future payments (8). In plain English, what this means is that if providers throughout the country do not establish a fair charge for G0424 that reflects the complexity and actual expense of all components of the service, Medicare reimbursement would continue to be low. An important context for setting the charge for G0424 is that most hospital charges are roughly fivefold greater than what is actually paid by Medicare and other insurance, a practice that cannot be ignored when considering fair reimbursement for PR (9).

of the service, Medicare reimbursement would continue to be low. An important context for setting the charge for G0424 is that most hospital charges are roughly fivefold greater than what is actually paid by Medicare and other insurance, a practice that cannot be ignored when considering fair reimbursement for PR (9). Historically, the majority of PR providers and hospitals have never adequately modified PR charges to reflect the increase in time and resources used for the bundled G0424 1-hour billing code from the original model of billing separately for both exercise and education in 15-minute increments. The effect on reimbursement is a result of Medicare’s use of PR charges (as well as information from the hospital cost report) to calculate annual changes in PR reimbursement. A recent review of charges for PR for patients with COPD submitted to Medicare in 2016 from claims billed by 1,350 US hospitals indicates that low charges for the PR bundled code continue to persist. This practice has likely contributed to the reality that cardiac rehabilitation reimbursement for 1 hour of treatment is now more than double that of PR ($116.65 vs. $55.96). It also might be relevant to consider what Medicare reimburses for other procedures: $229 for a pulmonary function test and $409 for an echocardiogram (9), which is four to seven times more than the cost of an hour of PR.

ion reimbursement for 1 hour of treatment is now more than double that of PR ($116.65 vs. $55.96). It also might be relevant to consider what Medicare reimburses for other procedures: $229 for a pulmonary function test and $409 for an echocardiogram (9), which is four to seven times more than the cost of an hour of PR. What can be done? ATS members need to meet with their hospital administrators and educate them about the benefits of PR for their patients and their institutions. The charge master, a comprehensive listing of items billable to a patient’s health insurance provider, needs to be updated to include all resources used in a PR program: space, oxygen, therapists, exercise machines, educational materials, medical director time, overhead charges, and so on. All items used in the PR program should be listed and an accurate total charge calculated. Hospital administrators can then set the appropriate charge rates for PR services. If this is done throughout the country, the national average charge for PR will increase and the Medicare payment for each unit of G0424 will increase. If we are successful, reimbursement for PR will increase over a 3–5-year period as annual data are accumulated by the Centers for Medicare & Medicaid Services. Fortunately, for those of us who do not understand how to read a medical bill, let alone navigate a charge master, there is an app. We refer you to the Pulmonary Rehabilitation Toolkit that details resources for PR billing (https://www.aacvpr.org/Advocacy/Pulmonary-Rehabilitation-Toolkit) (10).

What can be done? ATS members need to meet with their hospital administrators and educate them about the benefits of PR for their patients and their institutions. The charge master, a comprehensive listing of items billable to a patient’s health insurance provider, needs to be updated to include all resources used in a PR program: space, oxygen, therapists, exercise machines, educational materials, medical director time, overhead charges, and so on. All items used in the PR program should be listed and an accurate total charge calculated. Hospital administrators can then set the appropriate charge rates for PR services. If this is done throughout the country, the national average charge for PR will increase and the Medicare payment for each unit of G0424 will increase. If we are successful, reimbursement for PR will increase over a 3–5-year period as annual data are accumulated by the Centers for Medicare & Medicaid Services. Fortunately, for those of us who do not understand how to read a medical bill, let alone navigate a charge master, there is an app. We refer you to the Pulmonary Rehabilitation Toolkit that details resources for PR billing (https://www.aacvpr.org/Advocacy/Pulmonary-Rehabilitation-Toolkit) (10). It is time for the pulmonary medicine and lung disease scientific community to bring our concerns to hospital administrators. These administrators need to be made aware of the concerns regarding G0424 billing and the effect of undervalued charges on Medicare payment. It is also time for practitioners and scientists to partner with PR clinicians and administrators to determine whether charges for their PR program reasonably represent the complexity of the intervention, the acuity of the target population, and the value of this evidence based intervention.

harges on Medicare payment. It is also time for practitioners and scientists to partner with PR clinicians and administrators to determine whether charges for their PR program reasonably represent the complexity of the intervention, the acuity of the target population, and the value of this evidence based intervention. Originally Published in Press as DOI: 10.1164/rccm.201809-1711ED on January 22, 2019 Author disclosures are available with the text of this article at www.atsjournals.org.

Most new drug treatments fail because they lack efficacy (1). In sepsis research, new therapies must contend with an additional barrier: the intractable heterogeneity of the sepsis syndrome (2). Together, these challenges have so far proved insurmountable. Hundreds of clinical trials have been conducted, at a cost of hundreds of millions of dollars, to test new agents to modulate the host response to injury in sepsis. None have succeeded (2). The sepsis syndrome itself is simultaneously too broad and too narrow. Sepsis encompasses numerous different etiologies and pathophysiological processes, but—by definition (3)—excludes sterile injuries that lead to the same pathophysiology and organ failures, such as trauma, burns, hemorrhage, and pancreatitis. Some components of heterogeneity in sepsis are clinically apparent, such as variability in causal pathogens, comorbidities, environmental factors, and host genetics. But there is also evidence from recent studies (4–6) that important pathophysiological processes that are active in sepsis patients may vary in ways that are not directly observable at the bedside. If so, there is a chance that these processes may be amenable to different treatments (Figure 1). Figure 1. Deep phenotyping in practice. In a heterogeneous population, composite gene expression signals, either alone or in combination with clinical and other observations, may predict net benefit from a particular therapy. In reality, it is very likely that some patients will belong to multiple endotypes (indicated by colors on the image).

ng in practice. In a heterogeneous population, composite gene expression signals, either alone or in combination with clinical and other observations, may predict net benefit from a particular therapy. In reality, it is very likely that some patients will belong to multiple endotypes (indicated by colors on the image). Large observational studies of blood transcriptomics applied to sepsis populations have provided several models based on molecular classification of patients with sepsis. In particular, the Genomic Advances in Sepsis (GAinS) consortium in the United Kingdom (4, 6) and the Molecular Diagnosis and Risk Stratification of Sepsis (MARS) consortium in the Netherlands detected distinct molecular endotypes in leukocyte genome-wide expression profiles from samples collected on ICU admission. The MARS consortium identified four molecular endotypes in all-cause sepsis (designated MARS 1–4) (6), whereas the GAinS consortium identified two molecular endotypes in community-acquired pneumonia (designated sepsis response signature 1 [SRS1] and SRS2) (4). More recently, in an impressive demonstration of the power of open science and data sharing (7), Sweeney and colleagues (5) identified three clinical signatures—termed inflammopathic, coagulopathic, and adaptive—using pooled data from publicly available gene expression data from other studies of patients with sepsis. Both the MARS and SRS molecular endotypes were associated with different mortality rates.

ing (7), Sweeney and colleagues (5) identified three clinical signatures—termed inflammopathic, coagulopathic, and adaptive—using pooled data from publicly available gene expression data from other studies of patients with sepsis. Both the MARS and SRS molecular endotypes were associated with different mortality rates. This is a necessary first step. But after these observations, the question remains as to whether the MARS/SRS signatures relate to therapeutically targetable immunopathologies. Subgroups may reflect different disease severities, or other features of the patients that are irrelevant to their care. To detect a treatment effect in these subgroups, it is necessary to acquire gene expression data from patients enrolled in randomized clinical trials.

to therapeutically targetable immunopathologies. Subgroups may reflect different disease severities, or other features of the patients that are irrelevant to their care. To detect a treatment effect in these subgroups, it is necessary to acquire gene expression data from patients enrolled in randomized clinical trials. For the first time, direct evidence of such an effect is reported by Antcliffe and colleagues (pp. 980–986) in this issue of the Journal (8). Using data from the VANISH (Vasopressin versus Norepinephrine as Initial Therapy in Septic Shock) trial, a generalized linear model based on a previously identified seven-gene SRS classifier (DYRK2, CCNB1IP1, TDRD9, ZAP70, ARL14EP, MDC1, and ADGRE3) enabled the authors to stratify 176 patients as SRS1 (47%) or SRS2 (53%). Patients stratified in this fashion did not differ in demographics and most baseline clinical characteristics (except for rates of ischemic heart disease). However, in line with the group’s previous findings (4), 28-day mortality in the placebo group was higher in SRS1 (37%) than in SRS2 (8%) patients (8). Serum lactate at baseline was also higher in SRS1 patients. Together, these observations indicate that, to some extent, the SRS classification reflects disease severity.

owever, in line with the group’s previous findings (4), 28-day mortality in the placebo group was higher in SRS1 (37%) than in SRS2 (8%) patients (8). Serum lactate at baseline was also higher in SRS1 patients. Together, these observations indicate that, to some extent, the SRS classification reflects disease severity. If severity (rather than distinct pathophysiology) underlies the difference between these groups of patients, an interaction with steroid treatment might be anticipated. Large trials of steroids in sepsis and septic shock (9, 10) have reported trends toward a treatment benefit in patients with the highest risk of death. Whether these trends are real, and if so, whether they are simply a consequence of a higher event rate in this group (heterogeneity of treatment effect), are open questions at present. Based on these studies, we would have predicted a higher probability of detectable benefit from steroids among patients classified as SRS1. In fact, an interaction was detected between hydrocortisone use and SRS2-classified patients, resulting in increased mortality estimates with an adjusted odds ratio of 8.3 (95% confidence interval, 1.4–47.8), that is, a signal consistent with harm from steroid treatment in the less-severe SRS2-classified group. Collectively, these results and those from therapeutic trials using subclassifications of acute respiratory distress syndrome (11, 12) imply that there are divergent effects from a single intervention across and within different patient endotypes, bringing them closer to the definition of a true disease endotype (13).

If severity (rather than distinct pathophysiology) underlies the difference between these groups of patients, an interaction with steroid treatment might be anticipated. Large trials of steroids in sepsis and septic shock (9, 10) have reported trends toward a treatment benefit in patients with the highest risk of death. Whether these trends are real, and if so, whether they are simply a consequence of a higher event rate in this group (heterogeneity of treatment effect), are open questions at present. Based on these studies, we would have predicted a higher probability of detectable benefit from steroids among patients classified as SRS1. In fact, an interaction was detected between hydrocortisone use and SRS2-classified patients, resulting in increased mortality estimates with an adjusted odds ratio of 8.3 (95% confidence interval, 1.4–47.8), that is, a signal consistent with harm from steroid treatment in the less-severe SRS2-classified group. Collectively, these results and those from therapeutic trials using subclassifications of acute respiratory distress syndrome (11, 12) imply that there are divergent effects from a single intervention across and within different patient endotypes, bringing them closer to the definition of a true disease endotype (13). The investigators in the VANISH trial are to be congratulated for having the foresight to acquire transcriptomic data within a randomized controlled trial. Although confirmatory replication will be necessary, this work brings us a step closer to the primary aim of stratified medicine research: new phenotypes with direct therapeutic consequences. It is our view that future clinical trials in critical illness should consider from the outset the probability that any new therapy may have a differential effect in a subgroup of patients, and that subgroup may only be identifiable through deep phenotyping. Among the available methodologies, preservation of whole-blood RNA is the most pragmatic way to enable future deep phenotyping.

uld consider from the outset the probability that any new therapy may have a differential effect in a subgroup of patients, and that subgroup may only be identifiable through deep phenotyping. Among the available methodologies, preservation of whole-blood RNA is the most pragmatic way to enable future deep phenotyping. The dichotomous SRS1/2 classification simplifies analysis, but the groupings are drawn by bisecting what appears to be a unimodal distribution (4). This suggests that the SRS classification reflects two extremes of a continuously varying underlying biological process. This move from the identification of subgroups to the detection of continuous “treatable traits” within clinical populations has become a major focus of work in other fields (13); we, and many others, would argue that sepsis research is in particular need of these new approaches (2). Going further, it is very plausible that any physiological process that is active in a large proportion of patients with sepsis will also be active in some patients with severe sterile injury. As with other therapeutic approaches in critical care medicine, new treatable traits may be generalizable across critical illnesses.

further, it is very plausible that any physiological process that is active in a large proportion of patients with sepsis will also be active in some patients with severe sterile injury. As with other therapeutic approaches in critical care medicine, new treatable traits may be generalizable across critical illnesses. If the information necessary to predict response to a given therapy is present in measured clinical variables, or in the whole-blood transcriptome, then detecting it becomes entirely a matter of data analysis. With current techniques, huge numbers of patients will be needed to overcome signal/noise ratios. Integration of transcriptomic signatures with genetic associations (14) may enable more efficient detection of key underlying processes. Ultimately, these approaches may identify new, specific drug targets to modulate the host response to critical injury (15), and actionable estimates of individual treatment effect for critically ill patients. Originally Published in Press as DOI: 10.1164/rccm.201811-2148ED on December 12, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

To the Editor: We read with great interest the work of Adegunsoye and colleagues showing a significant association between enlarged mediastinal lymph nodes (MLNs) on chest computed tomography and survival in patients with interstitial lung diseases (ILDs) (1). They report a 66% prevalence of enlarged MLNs according the type of ILD, with various potential causes of development as previously pointed out. The authors raise the hypothesis that enlarged MLNs may be reflective of underlying immunologic phenomena in lung tissue, which in turn contribute to the pathophysiology of disease progression in pulmonary fibrosis. However, we suggest that the potential involvement of environmental exposures in ILDs, particularly anthracosis, should be discussed. Anthracosis caused by coal dust and other environmental factors such as air pollution, biomass fuels used extensively for cooking (“hut lung”), and cigarette smoking is also known to be a source of damage in MLNs (2).

tential involvement of environmental exposures in ILDs, particularly anthracosis, should be discussed. Anthracosis caused by coal dust and other environmental factors such as air pollution, biomass fuels used extensively for cooking (“hut lung”), and cigarette smoking is also known to be a source of damage in MLNs (2). A broad array of inhaled exposures are risk factors for developing an ILD, particularly idiopathic pulmonary fibrosis (IPF) in genetically susceptible patients (3). Inhaled agents are known to induce a series of lesions in alveolar epithelial cells, causing a biochemical oxidant injury and thereafter an immunological response when healing mechanisms (e.g., inflammation, coagulation, and epithelial repair) are put in place, resulting in pulmonary fibrosis. It is very likely that MLNs are involved in this process. Indeed, autopsy studies have revealed higher levels of inorganic particles, such as silicon and aluminum, in the MLNs of patients with IPF compared with controls (3).

mmation, coagulation, and epithelial repair) are put in place, resulting in pulmonary fibrosis. It is very likely that MLNs are involved in this process. Indeed, autopsy studies have revealed higher levels of inorganic particles, such as silicon and aluminum, in the MLNs of patients with IPF compared with controls (3). Interestingly, inhalation of occupational dusts may be an aggravating factor associated with a poor prognosis in several diseases, and particularly in IPF. In a large Korean cohort of patients with IPF, Lee and colleagues evaluated the prognosis of IPF according to the patients’ work and found that the wood or chemical dust–exposure group showed the worst outcomes (4). This group displayed a greater annual decline in FVC% and a higher mortality compared with nonexposed patients, with a hazard ratio of 1.813 (95% confidence interval [CI], 1.049–3.133, P = 0.033) (4). Based on U.S. death certificates from 1999 to 2003, Pinheiro and colleagues identified three industry categories with potential exposure to wood and metal dust that were associated with statistically significant risk estimates for IPF mortality: fabricated structural metal products (mortality odds ratio [MOR], 1.7 [95% CI, 1.0–3.1]), metal mining (MOR, 2.2 [95% CI, 1.1–4.4]), and wood buildings and mobile homes (MOR, 5.3 [95% CI, 1.2–23.8]) (5). Gold and colleagues examined potential associations of occupational exposures with the risk of mortality from systemic autoimmune diseases, using U.S. death certificates from 26 states (6). Farming occupation was associated with death from any systemic autoimmune disease (odds ratio [OR], 1.3 [95% CI, 1.2–1.4]), mining machine operators were at increased risk of death from systemic lupus erythematosus (OR, 1.8 [95% CI, 1.2–2.7]), and risk of death from systemic sclerosis was associated with usual occupation as industrial machinery repairers (OR, 2.3 [95% CI, 1.4–3.9]).

toimmune disease (odds ratio [OR], 1.3 [95% CI, 1.2–1.4]), mining machine operators were at increased risk of death from systemic lupus erythematosus (OR, 1.8 [95% CI, 1.2–2.7]), and risk of death from systemic sclerosis was associated with usual occupation as industrial machinery repairers (OR, 2.3 [95% CI, 1.4–3.9]). Enlarged MLNs in ILDs may be at least in part a marker of occupational or environmental exposure. Thus, we may hypothesize that the prognostic impact of MLNs observed in the study by Adegunsoye and colleagues could be related to the negative effects of unrecognized exposures. It would have been interesting to look at the patients’ occupational potential exposures, and ideally to perform a cytological analysis of MLNs to verify the presence or absence of lymph node anthracosis or anthracofibrosis in these patients. Originally Published in Press as DOI: 10.1164/rccm.201811-2209LE on January 17, 2019 Author disclosures are available with the text of this letter at www.atsjournals.org.

For patients with many incurable, life-shortening lung conditions, lung transplantation offers “curative” therapy and the only therapeutic option with a reliable chance to improve their quality of life (QOL). But what is really meant by the term “QOL,” and how do we measure this abstract construct? For individuals, QOL refers to their perceptions of how well their needs and wants are met across dimensions of life that matter most to them. Thus, accurate assessments of QOL require knowledge of patients’ perceptions, their needs and wants, and the dimensions of life they care about. Carefully crafted questionnaires, developed with systematically collected input from patients with the condition of interest, can capture all these things.

atter most to them. Thus, accurate assessments of QOL require knowledge of patients’ perceptions, their needs and wants, and the dimensions of life they care about. Carefully crafted questionnaires, developed with systematically collected input from patients with the condition of interest, can capture all these things. Until now, investigators and other stakeholders interested in examining the effect of lung transplantation on a person’s QOL have had to rely on existing questionnaires (1–3). However, none of those questionnaires adequately address all of the domains that are important to patients after lung transplantation, including, among other things, emotional well-being, pulmonary and extrapulmonary symptoms, effects of immunosuppression, and neurocognitive status. In fact, some experts say that many existing questionnaires—even ones that call themselves “QOL” or “health-related QOL” questionnaires—are really measures of health status. They measure certain domains (and do it well), but by confining themselves to assessing physical and/or emotional health, they fail to capture other domains that contribute to a person’s QOL, including physical and emotional well-being as well as material comforts; relationships with a spouse/partner, family, and friends; being able to help other people; learning; independence; working; understanding oneself; expressing oneself; and socializing (4, 5).

apture other domains that contribute to a person’s QOL, including physical and emotional well-being as well as material comforts; relationships with a spouse/partner, family, and friends; being able to help other people; learning; independence; working; understanding oneself; expressing oneself; and socializing (4, 5). In this issue of the Journal, Singer and colleagues (pp. 1008–1019) describe the development and initial testing of a questionnaire that aims to assess QOL after lung transplantation (6). Rather than developing their own items de novo, these lung transplant and questionnaire-development experts used clusters of items from existing questionnaires to capture aspects of life after lung transplant. They initially identified 126 items comprising clusters from eight questionnaires. Next, they conducted 43 cognitive interviews with lung transplant recipients who were asked whether the words and grammar used for individual items were clear, and whether there was redundancy in the items or clusters. For clusters that tapped the same construct (e.g., respiratory symptoms), interviewees were asked to select which item cluster more clearly reflected their definition of health-related QOL. Forty-two items were dropped after the interviews, leaving 84 items for field testing. Results from an exploratory factor analysis led the investigators to drop another 24 items. This left their questionnaire, which they call the Lung Transplant Health-Related Quality of Life (LT-QOL) survey, with 60 items comprising 10 scales (pulmonary symptoms, gastrointestinal symptoms, neuromuscular symptoms, treatment burden, worry about future health, cognitive limitations, sexual problems, anxiety/depression, health distress, and general QOL).

call the Lung Transplant Health-Related Quality of Life (LT-QOL) survey, with 60 items comprising 10 scales (pulmonary symptoms, gastrointestinal symptoms, neuromuscular symptoms, treatment burden, worry about future health, cognitive limitations, sexual problems, anxiety/depression, health distress, and general QOL). Analyses showed that the LT-QOL has acceptable-to-excellent psychometric properties, ranging from internal consistency to floor and ceiling effects. In support of its validity, its scores correlated in expected directions with relevant domains from other questionnaires completed in the same sitting. Its scores also correlated, as hypothesized, with concurrently collected spirometric and walk-test data, and LT-QOL scores were significantly worse for respondents with severe chronic lung allograft dysfunction than for those without. These analyses suggest that the LT-QOL is a reliable questionnaire, and it would seem to assess domains that are meaningful to patients after lung transplantation. In short, it has easily passed the first psychometric hurdle. In my opinion, the methods used to develop this questionnaire were sound, and the numerous cognitive interviews with patients in the target population ensure its relevance.

ld seem to assess domains that are meaningful to patients after lung transplantation. In short, it has easily passed the first psychometric hurdle. In my opinion, the methods used to develop this questionnaire were sound, and the numerous cognitive interviews with patients in the target population ensure its relevance. The LT-QOL has some limitations, and there is more work to be done on it. Although the authors describe how the LT-QOL is more comprehensive than existing questionnaires, its 10 scales cover only five broad constructs: 1) physical health, 2) emotional/mental health, 3) treatment burden, 4) sexual health, and 5) general QOL. Measurement experts may argue that the LT-QOL does not measure health-related QOL. As alluded to above, inquiring about the frequency or intensity of symptoms (physical or emotional/mental) gets at health status, but that line of inquiry is subtly different from specifically asking how much those symptoms—or the condition in all its aspects—affects the full spectrum of life domains that feed into QOL. I would call the LT-QOL a hybrid health-status/QOL questionnaire developed specifically for the post–lung transplant population. Its last two items—which I really like (“I am able to enjoy life” and “I am content with the quality of my life right now”)—tap general QOL, but even they do not reveal whether or how the transplant (and all it involves and impacts) has affected the respondent’s QOL or ability to enjoy life. It is important to emphasize what makes this questionnaire distinct, as I suspect many investigators who will use the LT-QOL will not have expertise in measurement and will not review its content in detail. Typically, they would, I suspect, see its name and assume it captures a comprehensive view of QOL after lung transplantation. They should know what they’re getting by using it.

inct, as I suspect many investigators who will use the LT-QOL will not have expertise in measurement and will not review its content in detail. Typically, they would, I suspect, see its name and assume it captures a comprehensive view of QOL after lung transplantation. They should know what they’re getting by using it. There may be some hidden redundancy in the LT-QOL: 10% (n = 6) of the items (from three different scales) include the term “worry,” “worried,” or “worrying” (e.g., “I worry that my lung transplant will not work . . . about getting infections . . . that my health will get worse . . . about not being able to stop or control worrying,” “worrying too much about different things,” and “is health a worry in your life?”). But redundancy only lengthens the questionnaire; it does not detract from its psychometric soundness.

l not work . . . about getting infections . . . that my health will get worse . . . about not being able to stop or control worrying,” “worrying too much about different things,” and “is health a worry in your life?”). But redundancy only lengthens the questionnaire; it does not detract from its psychometric soundness. This work suggests that the LT-QOL is a nice new tool that is capable of capturing patients’ perceptions of the effects of lung transplantation, and it provides a foundation upon which to conduct additional analyses to support the validity of this tool. Remember that establishing the validity of a questionnaire is an ongoing process, not a threshold phenomenon, and involves using well-formulated hypotheses to test that questionnaire in many studies, under multiple conditions, to understand what its scores are able to reveal about respondents. The LT-QOL is primed and ready for use in observational and interventional studies. The investigators have placed the LT-QOL in the public domain, so others are free to use it. They suggest administering a generic QOL questionnaire alongside it. The next phase of analyses should assess how the LT-QOL tracks the post–lung transplant course, establish its responsiveness to changes in health status as time passes after transplantation, and determine its sensitivity for detecting differences in the health-status/QOL trajectory between groups over time. Supported in part by grants from the Munn Foundation and the Clarence V. Laguardia Foundation. Originally Published in Press as DOI: 10.1164/rccm.201810-2001ED on November 20, 2018

This work suggests that the LT-QOL is a nice new tool that is capable of capturing patients’ perceptions of the effects of lung transplantation, and it provides a foundation upon which to conduct additional analyses to support the validity of this tool. Remember that establishing the validity of a questionnaire is an ongoing process, not a threshold phenomenon, and involves using well-formulated hypotheses to test that questionnaire in many studies, under multiple conditions, to understand what its scores are able to reveal about respondents. The LT-QOL is primed and ready for use in observational and interventional studies. The investigators have placed the LT-QOL in the public domain, so others are free to use it. They suggest administering a generic QOL questionnaire alongside it. The next phase of analyses should assess how the LT-QOL tracks the post–lung transplant course, establish its responsiveness to changes in health status as time passes after transplantation, and determine its sensitivity for detecting differences in the health-status/QOL trajectory between groups over time. Supported in part by grants from the Munn Foundation and the Clarence V. Laguardia Foundation. Originally Published in Press as DOI: 10.1164/rccm.201810-2001ED on November 20, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

Pulmonary vascular disease (PVD) and established pulmonary hypertension (PH) are common associations of bronchopulmonary dysplasia (BPD) (1). Although the reported incidence of PH is 14–44% in infants with recognized BPD (2, 3), recent evidence indicates that up to 20% of extremely low gestational age neonates without BPD will develop some degree of PVD during the neonatal period (2, 3). The mechanistic interrelation of both pathologies in more prematurely born infants is informed by the tandem development of the alveoli and microvasculature (4). BPD and PH share similar risk factors and overlapping symptoms, with some pointing to early PVD as an essential causative factor in the pathobiology of BPD (2) and others suggesting that PVD could be a distinct feature of prematurity, rather than a manifestation of BPD (5–7). Regardless of its association with BPD, it is more likely that PVD is just one of many factors leading to impaired respiratory function after preterm birth. It is not surprising that early identification of PVD, independent of the diagnosis BPD, may predict later pulmonary dysfunction, especially because preterm birth is associated with an increased risk of PH in childhood, adolescence, and adulthood (8).

actors leading to impaired respiratory function after preterm birth. It is not surprising that early identification of PVD, independent of the diagnosis BPD, may predict later pulmonary dysfunction, especially because preterm birth is associated with an increased risk of PH in childhood, adolescence, and adulthood (8). Improved echocardiographic assessment of PVD has led to increased recognition that disrupted pulmonary vascular growth in preterm infants may contribute to the pathogenesis of late respiratory disease (LRD) (2, 7). The diagnosis of PVD and the true prevalence of PH in preterm infants has been difficult to discern because of the paucity of reliable noninvasive measures to evaluate pulmonary hemodynamics (1) and an underappreciation for the practice of screening for PVD in extremely low gestational age neonates, with a focus primarily on infants with BPD (9). However, preterm infants without a diagnosis of BPD also remain at risk for respiratory morbidities and abnormal lung function into childhood, emphasizing the need to focus on alternate measures that explore differing mechanisms of disrupted pulmonary vascular and airway growth after prematurity (10).

nts with BPD (9). However, preterm infants without a diagnosis of BPD also remain at risk for respiratory morbidities and abnormal lung function into childhood, emphasizing the need to focus on alternate measures that explore differing mechanisms of disrupted pulmonary vascular and airway growth after prematurity (10). In this issue of the Journal, Mourani and colleagues (pp. 1020–1027) leveraged a multicenter cohort of preterm infants to demonstrate that echocardiographic evidence of PVD at 7 days of age was associated with a higher incidence of LRD in early childhood (11). This builds on their previous report demonstrating that early echocardiographic findings of PVD are strongly associated with the development and severity of BPD and late PH at 36 weeks (2). They also found that maternal diabetes and invasive mechanical ventilation support at 1 week of age were associated with LRD. Although BPD was predictive of LRD, there were 32 infants (14%) who did not have echocardiographic evidence of early PVD, late PH, or clinical BPD, but did have LRD. The article pursues the “vascular hypothesis,” which states that pulmonary vascular disturbances can contribute to later pulmonary dysfunction in former preterm infants. These data show that early identification of PVD, independent of later development of BPD, may contribute to the pathobiology of longer-term respiratory morbidity in former preterm infants.

hesis,” which states that pulmonary vascular disturbances can contribute to later pulmonary dysfunction in former preterm infants. These data show that early identification of PVD, independent of later development of BPD, may contribute to the pathobiology of longer-term respiratory morbidity in former preterm infants. The study is timely, as the ability of a BPD diagnosis to predict the impact of prematurity on respiratory disease beyond the neonatal period has been questioned. Similar to the large, prospective multicenter cohort study from the NIH PROP (Prematurity and Respiratory Outcomes Program) (12), these data identify those preterm infants at risk of developing late respiratory morbidity in the first week of life. Although prediction of late morbidity by perinatal risk factors and BPD alone in the PROP cohort exceeded that of the current study, it is likely that both perinatal and postnatal factors affect the airspaces and the pulmonary vasculature, driving the clinical trajectories of children born prematurely. The identification of high-risk infants earlier in their course may provide a critical window for applying established or emerging therapies to prevent progressive PVD and PH and to improve late respiratory morbidities (12).

and the pulmonary vasculature, driving the clinical trajectories of children born prematurely. The identification of high-risk infants earlier in their course may provide a critical window for applying established or emerging therapies to prevent progressive PVD and PH and to improve late respiratory morbidities (12). Significant challenges exist in the noninvasive assessment of pulmonary hemodynamics and thereby in the identification of key adaptive mechanistic underpinnings of PVD in premature infants (9). Traditional echocardiographic markers of PH have relied on a combination of qualitative assessments (interventricular septal wall motion and right ventricular [RV] morphological changes) and quantitative estimates based on the tricuspid regurgitant jet velocity. Echocardiographic evidence of PVD often precedes the onset of overt clinical signs, symptoms, and detection of PH. In the pulmonary circulation, the key components of RV afterload, resistance and compliance, evolve together, but in opposite directions in both health and disease (13). In early PVD, a small increase in PVR is accompanied by a significant reduction in vascular compliance, an initial response that may not result in an immediate change in pulmonary arterial pressure, limiting the applicability of many of the current screening modalities that only rely on detecting an increase in pressure. With more advanced stages of PVD (i.e., PH), vascular stiffness will reach its maximum limits, and any further increase in PVR is not associated with further reduction in compliance. The recognition of alterations in septal wall morphology and function at 7 days in some preterm infants is indicative of elevated pressures seen in PH (2, 14). Emerging, noninvasive indices of PVD (e.g., RV systolic time intervals [7], measures of RV function [e.g., strain parameters (14)]) that more broadly capture components of RV performances and afterload (pressure, resistance, and compliance) may prove to be more informative than pressure estimates alone.

in PH (2, 14). Emerging, noninvasive indices of PVD (e.g., RV systolic time intervals [7], measures of RV function [e.g., strain parameters (14)]) that more broadly capture components of RV performances and afterload (pressure, resistance, and compliance) may prove to be more informative than pressure estimates alone. Early evidence of PVD in preterm infants adds to the growing list of complications of being born premature that may increase susceptibility or be a marker for greater risk of late pulmonary disease beyond the neonatal period and into early childhood, adolescence, and adulthood (8, 15). Adding these results to their previous work, Mourani and colleagues have now shown that echocardiographic evidence of PVD at 1 week of age is an early predictor of BPD, late PH, and late respiratory disease (11). Newborns with PVD may be particularly susceptible to secondary insults; future studies should use these early risk factors as predictive biomarkers toward enrolling the highest-risk infants into clinical intervention trials (6). Originally Published in Press as DOI: 10.1164/rccm.201810-1983ED on November 5, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

To the Editor: We read with interest the article by Coates and colleagues (1). The authors elaborate on the recently published “ERS Technical Standard on Bronchial Challenge Testing: General Considerations and Performance of Methacholine Challenge Tests” (2) and suggest that PD20 (the provocative dose of methacholine [MCh] that results in a 20% fall in FEV1) should replace PC20 (the provocative concentration of MCh that results in a 20% fall in FEV1). We have the following comments:1. MCh dose versus concentration: Based on the suggested new standard for using dose instead of concentration of MCh to evaluate and report airway reactivity, the authors recommended using a mechanical breath simulator to collect MCh on a filter representing the mouth and calculate the deposited dose. Alternatively, they suggest that nebulizer manufacturers should provide those who perform MCh challenge tests with information regarding the aerosol output characteristics of their nebulizer. No mechanism is suggested regarding how to make this happen, and pulmonary function testing (PFT) laboratories are left with considerable confusion as to how to proceed.

r manufacturers should provide those who perform MCh challenge tests with information regarding the aerosol output characteristics of their nebulizer. No mechanism is suggested regarding how to make this happen, and pulmonary function testing (PFT) laboratories are left with considerable confusion as to how to proceed. 2. Nebulizer replication studies: The authors cite recent articles comparing the original Wright nebulizers with newer nebulizers (specifically the SOLO Aerogen vibrating mesh device and the Aero Eclipse). These papers showed that PD20, but not PC20, was independent of the delivery system used. However, these nebulizers are infrequently used in most PFT laboratories. We wonder if it is not essential to replicate their findings against other commonly used nebulizers. 3. Nebulizer cost and durability: Vibrating mesh nebulizers are expensive and have well-known problems involving mesh clogging and circuit issues. Also, their mass median aerodynamic diameter is considerably larger than that of the Wright small-volume nebulizer. We wonder how replacing concentration × 2 minutes at each doubling concentration with the hyperprecision “dose” would improve our ability to provide more clinically relevant answers. In particular, we have found no studies that showed this change to be of sufficiently increased benefit to offset the considerable increase in cost and difficulty of obtaining similar data in virtually all current PFT labs.

the hyperprecision “dose” would improve our ability to provide more clinically relevant answers. In particular, we have found no studies that showed this change to be of sufficiently increased benefit to offset the considerable increase in cost and difficulty of obtaining similar data in virtually all current PFT labs. If the authors’ recommendations are adopted before such evidence is available, a widely used important test for airway hyperreactivity would become a test that is exclusive to a relatively small number of academic PFT labs in large centers, thus inconveniencing patients in smaller communities who must travel, often many miles, to the centers that are able to implement the change! Originally Published in Press as DOI: 10.1164/rccm.201809-1765LE on January 14, 2019 Author disclosures are available with the text of this letter at www.atsjournals.org.

From the Authors: We welcome the interest shown by Lescoat and colleagues and Khamis and colleagues in our publication (1), and thank the authors for their letters. Although the clinical value of plasma biomarkers is well established in many chronic disease states, we recognize that limitations exist regarding their use in prognostication of outcomes. As alluded to by Lescoat and colleagues, the magnitude of the prognostic effect for individual plasma biomarkers will likely vary across diverse forms of interstitial lung disease (ILD) and at different thresholds. Indeed, circulating plasma biomarker levels may be lower relative to biomarker concentrations within specific organs that are directly involved in tissue repair and homeostasis. Also, the extent of disease activity that typically occurs across multiple extrapulmonary organs, such as those affected in connective tissue disease associated with ILD, may accentuate this variation. With regard to IL-6, it has been suggested that this cytokine has a bidirectional role in the pathogenesis of lung fibrosis. Whereas IL-6 blockade at an early inflammatory stage can accelerate lung fibrosis, blockade at an early fibrotic stage may ameliorate subsequent fibrogenesis (2). These factors could conceivably account for the potentially favorable results that are observed when IL-6 is therapeutically targeted in scleroderma-associated ILD (3). We did find that mean plasma IL-6 levels were nonsignificantly decreased in subjects with enlarged mediastinal lymph nodes (MLNs) in our study, but chose to report median plasma cytokine values in our comparative analyses because these values are less subject to the influence of outliers (1). In both our primary and replication cohorts, the median plasma IL-6 levels did not differ by MLN size (Figure 1) and did not predict mortality risk. We therefore reiterate that we cannot conclude from the data presented in our study that IL-6 might be protective in fibrotic ILD, and agree with Lescoat and colleagues that further study of the blockade of IL-6 in clinical trials is warranted (1).

d not differ by MLN size (Figure 1) and did not predict mortality risk. We therefore reiterate that we cannot conclude from the data presented in our study that IL-6 might be protective in fibrotic ILD, and agree with Lescoat and colleagues that further study of the blockade of IL-6 in clinical trials is warranted (1). Figure 1. (A) Box plots depicting IL-6 (pg/ml) levels stratified by MLN size (mm) in patients with ILD within the UCHICAGO (n = 116) and UCDAVIS (n = 118) cohorts. Comparison of cytokine concentrations in patients with MLN < 10 mm and MLN ≥ 10 mm, using the Wilcoxon signed-rank test for matched nonparametric data in 10,000 bootstrap replications to improve precision at the 95% confidence interval level. (B) Box plots depicting NT-proBNP (pg/ml) levels stratified by MLN size (mm) in patients with ILD within the UCHICAGO cohort (n = 628). For clarity, NT-proBNP data points for two subjects (16,116 and 22,812 pg/ml) are not depicted. Group comparisons for unmatched nonparametric data were conducted using the Pearson chi-squared test for equality of the medians between patients with MLN ≥ 10 mm (purple) and MLN < 10 mm (gray). ILD = interstitial lung disease; MLN = mediastinal lymph node; NTpro-BNP = N-terminal pro–B-type natriuretic peptide; UCDAVIS = University of California, Davis; UCHICAGO = University of Chicago.

g the Pearson chi-squared test for equality of the medians between patients with MLN ≥ 10 mm (purple) and MLN < 10 mm (gray). ILD = interstitial lung disease; MLN = mediastinal lymph node; NTpro-BNP = N-terminal pro–B-type natriuretic peptide; UCDAVIS = University of California, Davis; UCHICAGO = University of Chicago. Lescoat and colleagues also raise the important issue of chronic heart failure, which may be prevalent in ILD and should be carefully considered when evaluating the impact of novel indices on assessment of clinical outcomes such as hospitalization and mortality. Because it is possible that cardiac disease may at least in part causally mediate the link between MLN enlargement and ILD outcomes (4, 5), we explored the association of NT-proBNP (N-terminal pro–B-type natriuretic peptide) levels with MLN enlargement in our outcome analyses. We found that including this cardiac biomarker in our prognostic models did not significantly alter the predictive value of MLN for all-cause mortality or hospitalization. Furthermore, plasma NT-proBNP levels did not differ with MLN enlargement and lacked a strong correlation with MLN diameter (Figure 1).

ome analyses. We found that including this cardiac biomarker in our prognostic models did not significantly alter the predictive value of MLN for all-cause mortality or hospitalization. Furthermore, plasma NT-proBNP levels did not differ with MLN enlargement and lacked a strong correlation with MLN diameter (Figure 1). We also agree with Khamis and colleagues, who, along with Lescoat and colleagues, wondered about the potential role of inhalational exposures in the etiopathogenesis of MLN enlargement and ILD. Perhaps the lung’s activity as an immunologic organ and the associated enlarged MLNs reflect ongoing inflammation from active immunologic responses to anthracosis and other dust-related environmental exposures. Indeed, the average MLN diameters in smokers within our cohort were larger than those observed in never smokers. As astutely outlined by Khamis and colleagues, patients with various ILDs commonly have inorganic and organic environmental exposures that may impact outcomes. Although our prognostic models adjusted for duration of exposure to inhalational tobacco use, other unmeasured environmental exposures could potentially contribute to disease progression in pulmonary fibrosis. As such, it is certainly possible that MLN enlargement may be a biomarker of ongoing occupational or environmental exposure. Although these exposures may play a causal role, our study was not designed to assess causality, and thus we are unable to ascertain the potential etiologies for the observed MLN enlargement. In addition, it should be noted that many of our patients with ILD had MLNs of unremarkable size, which suggests that enlarged MLNs are not a required precursor to ILD development, and therefore the relationship between prior environmental exposures and enlarged MLN size is unclear. Future investigations should systematically assess the cytologic characteristics of enlarged MLNs and evaluate the enlargement in the context of dose and temporal relationships with occupational and environmental exposures.

erefore the relationship between prior environmental exposures and enlarged MLN size is unclear. Future investigations should systematically assess the cytologic characteristics of enlarged MLNs and evaluate the enlargement in the context of dose and temporal relationships with occupational and environmental exposures. Altogether, we concur with Lescoat and colleagues and Khamis and colleagues in the realization that beyond prognostication, our study constitutes a crucial first step toward elucidating the role of MLNs in disease etiopathogenesis and refining the current classification of ILD. Supported by NIH R21AI126031, NIH K23HL138190, NIH R01AI125644, and NIH R01HL130796. Originally Published in Press as DOI: 10.1164/rccm.201811-2208LE on January 17, 2019. Author disclosures are available with the text of this letter at www.atsjournals.org.

From the Authors: We thank Drs. Amirav and Newhouse for their letter and interest in our editorial on characterizing nebulizer performance for methacholine challenge tests (1). We respectfully disagree with the premise of their letter. We believe that the science and clinical relevance of the previous 1999 guidelines need to be updated. The main problem is that the English-Wright nebulizer is no longer widely available, and if pulmonary function labs were to use as a substitute currently available nebulizers that have much higher aerosol output than the English-Wright nebulizer, every concentration step would deliver a much higher stimulus dose than intended by the 1999 guidelines. Regarding the need to calculate a delivered methacholine dose, the authors state that we offered no mechanism for how to compel nebulizer manufacturers to characterize the performance of their nebulizer. This was, in fact, the main purpose of our letter: to call out to the manufacturers to provide this essential service. We acknowledged that this would be beyond the capabilities of most pulmonary function labs, but it should be very much achievable by nebulizer manufacturers and aerosol scientists. Our hope was that this letter would emphasize to manufacturers that the American Thoracic Society and European Respiratory Society are counting on them to help the pulmonary function lab community.

of most pulmonary function labs, but it should be very much achievable by nebulizer manufacturers and aerosol scientists. Our hope was that this letter would emphasize to manufacturers that the American Thoracic Society and European Respiratory Society are counting on them to help the pulmonary function lab community. The authors also suggest that the data cited regarding the comparison of the English-Wright nebulizer with other nebulizers should include information about other commonly used nebulizers. We certainly agree, and remain hopeful that such data will be forthcoming. The data we cited, including those obtained with a vibrating mesh nebulizer, were simply meant as examples of how dose, not concentration, should be the common unit of measurement across devices. Regarding the point made about how the current recommendations might not provide more clinically relevant information, we would like to emphasize that at present there is significant variability in the way methacholine challenge tests are performed, resulting in the potential for imprecision and diagnostic error. No other diagnostic test in modern medicine would allow such a lack of rigorous standards or interlaboratory variation. With better defined and updated methodology, physicians can now have more confidence in the results of testing. Originally Published in Press as DOI: 10.1164/rccm.201811-2216LE on January 14, 2019 Author disclosures are available with the text of this letter at www.atsjournals.org.

Survival for individuals with cystic fibrosis (CF) is improving over time, but progressive respiratory failure remains the number one cause of death for individuals with CF (1). Historically, FEV1 <30% of the predicted value has prompted discussions in CF clinics about the potential need for lung transplantation (LTx) (2, 3). However, survival with advanced lung disease is increasing over time, with a recent estimate of median survival of 6.6 years after FEV1 <30% in the United States (3–6). Despite the improved survival times for individuals with FEV1 <30%, rates of death in the United States are approximately 10% per year after this lung function threshold is reached (6). Although FEV1 has been shown to have a strong and consistent association with death or LTx in CF, there are other predictors as well, including malnutrition, hypoxemia, hypercarbia, pulmonary hypertension, increased frequency of exacerbations or hospitalizations, sputum culture positive for Burkholderia cepacia, massive hemoptysis, and reduced 6-minute-walk test distance (3, 4, 7–11). Despite these data, estimating the time until death or LTx in patients with CF is exceedingly difficult, and care teams need more and better tools to prognosticate in this patient population.

ons, sputum culture positive for Burkholderia cepacia, massive hemoptysis, and reduced 6-minute-walk test distance (3, 4, 7–11). Despite these data, estimating the time until death or LTx in patients with CF is exceedingly difficult, and care teams need more and better tools to prognosticate in this patient population. In this issue of the Journal, Hebestreit and colleagues (pp. 987–995) present a multicenter, international, retrospective study of clinically indicated cardiopulmonary exercise testing (CPET) for individuals with CF (12). Ten centers (in Europe, Australia, and North America) contributed CPET data from over 500 individuals with CF, age ≥10 years, between 2000 and 2007. Data from a valid maximal CPET were available for 433 individuals, with follow-up of the cohort through 2014. The subjects selected were relatively healthy despite having a clinical indication for CPET (mean FEV1, 73% predicted; 5-year survival rate, 93%). The investigators found that V˙o2peak, workpeak, V˙e/V˙o2, and V˙e/V˙co2 were all associated with the composite outcome after adjustment for other known predictors of death and/or LTx in multivariable models. Using Ward’s hierarchical clustering, the investigators identified four clusters, which included continuous and binary clinical and physiological parameters. This cluster analysis identified a group of individuals with low FEV1, low body mass index z-scores, and worse CPET performance with dismal outcomes over the course of 10 years (63% death or LTx). Although FEV1 was the most important variable for clustering, the CPET-derived parameters had a stronger influence on clustering than other traditional risk factors for death or LTx.

h low FEV1, low body mass index z-scores, and worse CPET performance with dismal outcomes over the course of 10 years (63% death or LTx). Although FEV1 was the most important variable for clustering, the CPET-derived parameters had a stronger influence on clustering than other traditional risk factors for death or LTx. This study has several strengths. First, this is the largest study of CPET in CF, and it confirms prior single-center findings regarding the prognostic value of CPET-derived parameters in adults and adolescents with CF (13–15). Because of the large sample size in this study, analyses could be adjusted for important potential confounders of the association between CPET performance and death or LTx. The investigators identified strong relationships between CPET variables and death or LTx in the entire cohort that were independent of FEV1, as well as among individuals with advanced lung disease (FEV1 < 40%) and in short-term (2 yr) sensitivity analyses. Second, this study had long-term follow-up of clinical outcomes, with very few individuals lost to follow-up or missing primary endpoint data 5 years or more after CPET (n = 58, excluded from analyses). Third, the use of cluster analysis highlights the importance of focusing prognostication on the highest-risk group (individuals with low FEV1, malnutrition, and poor CPET performance). One of the key benefits of CPET is that it represents a functional and dynamic assessment of the cardiopulmonary system. Such an evaluation provides important clinical variables that are unavailable during a static test of airflow, such as office spirometry.

uals with low FEV1, malnutrition, and poor CPET performance). One of the key benefits of CPET is that it represents a functional and dynamic assessment of the cardiopulmonary system. Such an evaluation provides important clinical variables that are unavailable during a static test of airflow, such as office spirometry. One of the fundamental challenges of prognosticating in CF is that the event rate (death or LTx) in the overall population is low. Prognostication is most relevant for individuals with an imminent risk of death, to avoid missing the opportunity for LTx in the appropriate individuals with CF. Incorporation of CPET could augment the complex decision-making that occurs around the timing of evaluation and listing for LTx. Interestingly, because of longstanding evidence of CPET-derived parameters (e.g., V˙o2peak and V˙e/V˙co2 slope) as predictors of death in patients with systolic heart failure, the selection of heart transplant candidates has incorporated CPET for more than a decade (16), and carefully collected prospective data support the prognostic value of CPET for these individuals (17–19). The identification of threshold values for CPET parameters to guide the timing of listing individuals with CF for LTx could be invaluable because of the documented prolonged survival with low lung function and the poor positive predictive value of FEV1 <30% predicted (10).

c value of CPET for these individuals (17–19). The identification of threshold values for CPET parameters to guide the timing of listing individuals with CF for LTx could be invaluable because of the documented prolonged survival with low lung function and the poor positive predictive value of FEV1 <30% predicted (10). Some key weaknesses of the study were acknowledged by the authors. One concern raised was the potential for confounding by CF center practices. They identified significant differences in outcomes at the CF center level. This in turn led the investigators to adjust for clinical site in their models. Although this analysis can take into account within-site correlation of participants, it cannot address potential differential indication bias. Individuals who underwent CPET at each site may have had different disease severities or clinical indications that could not be accounted for in the analysis. When indication bias occurs in an observational study, it remains a challenge to address analytically. The investigators would have needed a separate control population of individuals who had an equal probability (potentially via the propensity score) of undergoing CPET but did not receive the test. Differential outcome ascertainment is also a potential source of bias for this study, as the investigators attempted to minimize the risk of bias from loss to follow-up or informative censoring, but may have introduced ascertainment bias when the cohort was limited to individuals with a minimum of 5 years of follow-up at the testing CF center (e.g., healthier individuals may have moved away from the center). Thus, it remains challenging to generalize the results of this study to the greater CF population. Despite these limitations, the data presented provide strong observational evidence for the potential role of CPET in risk stratification for individuals with CF.

ealthier individuals may have moved away from the center). Thus, it remains challenging to generalize the results of this study to the greater CF population. Despite these limitations, the data presented provide strong observational evidence for the potential role of CPET in risk stratification for individuals with CF. In conclusion, CPET adds prognostic information beyond the FEV1 and could be a dynamic marker of disease progression in CF. The study by Hebestreit and colleagues is a call to action to perform a prospective study of CPET for individuals with CF—ideally, individuals with severe CF. CPET is another tool in the prognostication tool kit for CF and prospective research is imperative for individuals with advanced lung disease approaching LTx. Supported by grants from the NIH/NHLBI (K23 HL138154), Cystic Fibrosis Foundation (RAMOS17A0), and the Cystic Fibrosis Foundation Lung Transplant Consortium (LEASE16A3). Originally Published in Press as DOI: 10.1164/rccm.201810-2053ED on November 13, 2018 Author disclosures are available with the text of this article at www.atsjournals.org.

We all know influenza can be bad. Aside from the fevers, cough, miserable body aches, and severe fatigue, people can actually die from it. Pregnant women, young children, and the elderly are most at risk for mortality. Recent modeling estimates (1) suggest the global mortality from seasonal influenza has been previously underestimated. From 1999 to 2015, influenza accounted for as many as 645,000 annual excess respiratory deaths, with likely many additional circulatory deaths. The highest mortality rates occurred in sub-Saharan Africa and Southeast Asia in people older than 75 years. Air pollution can also be bad, especially in developing countries. According to the World Health Organization, ambient air pollution caused 4.2 million premature deaths in 2016, with 91% of these in low- and middle-income countries (2). This does not include the health risks for the approximately 3 billion people that cook or heat their homes with kerosene, coal, and biomass fuels, including wood.

the World Health Organization, ambient air pollution caused 4.2 million premature deaths in 2016, with 91% of these in low- and middle-income countries (2). This does not include the health risks for the approximately 3 billion people that cook or heat their homes with kerosene, coal, and biomass fuels, including wood. But what if air pollution takes influenza from bad to worse? This could mean that air pollution is increasing influenza mortality, in addition to its own significant mortality. In this issue of the Journal, Rebuli and colleagues (pp. 996–1007), in their clinical study (3), addressed the question of whether exposure to wood smoke worsens epithelial mucosal responses to influenza virus infection. Their experimental model is nasal inoculation with the influenza virus vaccine, which is a mixture of live attenuated influenza viruses (LAIV), followed by nasal lavage. Thirty-nine healthy men and women were randomly exposed for 2 hours at rest to filtered air or wood smoke, followed by LAIV inoculation. Nasal lavage was performed before and 1 and 2 days after exposure/viral challenge, with assessment of changes in the expression of 255 genes and 30 cytokine proteins involved in inflammation. The researchers also assessed expression of viral genes as markers of infection and replication.

, followed by LAIV inoculation. Nasal lavage was performed before and 1 and 2 days after exposure/viral challenge, with assessment of changes in the expression of 255 genes and 30 cytokine proteins involved in inflammation. The researchers also assessed expression of viral genes as markers of infection and replication. LAIV infection caused the expected changes in inflammatory gene expression, including the expression of viral genes, confirming infection and replication. Surprisingly, the primary analysis showed no significant wood smoke effects on any of the 255 inflammatory response genes. Only IP-10 (IFN-γ–induced protein 10 kDa) and IL-6 increased after LAIV; wood smoke partially suppressed the increase in IP-10.

the expression of viral genes, confirming infection and replication. Surprisingly, the primary analysis showed no significant wood smoke effects on any of the 255 inflammatory response genes. Only IP-10 (IFN-γ–induced protein 10 kDa) and IL-6 increased after LAIV; wood smoke partially suppressed the increase in IP-10. However, in a planned secondary analysis, sex interacted significantly with exposure for 25 genes. Subsequent sex-specific analyses confirmed sex differences in gene expression before exposure, and in response to wood smoke. Many more genes were upregulated in men than in women before exposure. In the subjects exposed to filtered air followed by LAIV, women showed a more robust response than men. In the 8 men exposed to wood smoke compared with the 9 men exposed to filtered air, 13 genes increased expression more than twofold. In the 12 women exposed to wood smoke compared with 10 exposed to filtered air, 18 genes were differentially expressed, all downregulated, mostly less than twofold. Thus, the men had more inflammatory gene expression than women at baseline, with some genes increasing further with wood smoke and LAIV. Women had reduced gene expression at baseline, increased responses to filtered air/LAIV, and slight suppression of responses after wood smoke/LAIV. These wood smoke changes in opposite directions explain the negative outcome in the primary aggregate analysis.

with some genes increasing further with wood smoke and LAIV. Women had reduced gene expression at baseline, increased responses to filtered air/LAIV, and slight suppression of responses after wood smoke/LAIV. These wood smoke changes in opposite directions explain the negative outcome in the primary aggregate analysis. Sometimes we fail to consider the possibility of sex differences in the design of clinical studies, including previous studies of wood smoke exposure (4), and Rebuli and colleagues make an important contribution in this regard. The biological differences between men and women may affect their responses to a variety of environmental insults, including air pollutants and influenza virus. Men and women differ in their ability to control a long list of viral infections, including influenza virus, and mortality from viral infections is generally greater in men than women (5). Humoral and cellular antiviral immune responses are stronger in women. However, this could also increase aberrant responses such as autoimmunity in women relative to men. Ghosh, and colleagues (5) have reviewed the mechanisms involved for these sex-related differences. Of course there are the hormonal influences. And it turns out that many immune response genes are encoded on the X chromosome. One of the two X chromosomes is inactivated in female cells. Thus, a loss-of-function mutation in one of these X-linked immune response genes would be expressed in one-half of the cells in women, but all the cells in men, resulting in increased X-linked immunodeficiency in men.

se genes are encoded on the X chromosome. One of the two X chromosomes is inactivated in female cells. Thus, a loss-of-function mutation in one of these X-linked immune response genes would be expressed in one-half of the cells in women, but all the cells in men, resulting in increased X-linked immunodeficiency in men. We know that inflammation can be both good and bad. Good is controlling infection; bad is contributing to the symptoms and tissue damage in influenza. Although men are less able than women to mount an immune defense against the influenza virus, the findings of Rebuli and colleagues suggest that men have increased expression of inflammation-related genes in the nasal mucosa relative to women at baseline, and further increase inflammatory gene responses to LAIV infection with prior exposure to wood smoke. Thus, men have more difficulty than women in fighting off influenza, and prior wood smoke exposure may enhance the inflammatory response, and hence the severity, of influenza. There are important limitations to this study. We do not know the degree to which the sex differences in gene expression observed by Rebuli and colleagues in response to wood smoke and LAIV are the result of shifts in the type of cells recovered from the nose. We are not provided with a differential cell count for the nasal lavage, but it is likely that the observed changes in gene expression reflect in part an influx of inflammatory cells into the nose in response to these combined challenges, rather than just changes in gene expression of resident nasal epithelial cells.