Browse the corpus

udy, we sought to investigate whether there is an association between cough frequency, and its duration, with radiologic characteristics, such as cavitary volume and cavitary proximity to the airway. We also evaluated whether bacillary burden and culture conversion were associated with these radiologic characteristics. Materials and Methods Study Design This was a prospective cohort study conducted in two tertiary hospitals in Lima, Peru. The detailed study protocol has been published previously.20 Study participants were at least 18 years old, and their pulmonary TB diagnosis and drug sensitivity were assessed by means of microscopic observation drug susceptibility (MODS) broth culture assay.20, 21, 22, 23, 24, 25, 26 In this report, we restricted analyses to participants who had a strain that was susceptible to isoniazid and rifampicin and who did not have HIV (Fig 1) because immune status and drug-resistant strains affect radiologic manifestation.27, 28Figure 1 Flowchart for the Cayetano Cough Monitor CT scanning study. Radiologic features are based on readings from a US-board-certified radiologist. Cavity volume and distance to the airway are based on results from a computer-automated algorithm. TTP = time to positivity.

Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) has transformed diagnostic evaluation of mediastinal lymphadenopathy in respiratory medicine. This has revolutionized lung cancer staging, providing a median sensitivity of 89% for detection of malignant cells, and also affords high sensitivity for detection of granulomatous lymphadenitis in sarcoidosis.1, 2, 3 Two specific diagnostic limitations have emerged as research priorities to maximize the benefits from EBUS-TBNA: (1) to increase sensitivity for detection of malignancy and hence the negative predictive value, to reduce the need for more invasive surgical sampling; and (2) to increase the sensitivity for detection of active TB and improve discrimination between TB and sarcoidosis in granulomatous lymphadenitis. This distinction currently relies on assessment of demographic risk of TB or microbiological confirmation, which is unavailable in > 50% of cases.4, 5

e surgical sampling; and (2) to increase the sensitivity for detection of active TB and improve discrimination between TB and sarcoidosis in granulomatous lymphadenitis. This distinction currently relies on assessment of demographic risk of TB or microbiological confirmation, which is unavailable in > 50% of cases.4, 5 Genomewide transcriptional profiling can identify molecular signatures in peripheral blood or tumor specimens that could be used to improve diagnosis and risk stratification of patients with infectious and inflammatory diseases or to guide targeted therapeutic strategies for individuals with cancer.6, 7, 8, 9, 10 Feasibility of transcriptional profiling to detect gene expression associated with molecular pathogenesis of lung cancer in a small number of tumor-infiltrated nodes has been reported.11, 12 Granulomatous lymphadenitis has never been assessed by this method, but three previous studies reported extensive overlap of peripheral blood transcriptional signatures associated with TB and sarcoidosis.13, 14, 15 Although computational machine learning techniques were successfully applied to peripheral blood gene signatures to discriminate between TB and sarcoidosis cases, confidence in this approach has been partly undermined because of minimal overlap between the discriminating genes identified in each study.13, 14, 15 Since the inflammatory processes in sarcoidosis are reputed to be compartmentalized,16 we speculated that transcriptional profiles at the site of disease may offer better discrimination. In the present study we tested the hypothesis that genomewide transcriptional profiling of EBUS-TBNA samples will improve diagnostic differentiation of TB and sarcoidosis and detection of cancer in patients undergoing staging investigations.

anscriptional profiles at the site of disease may offer better discrimination. In the present study we tested the hypothesis that genomewide transcriptional profiling of EBUS-TBNA samples will improve diagnostic differentiation of TB and sarcoidosis and detection of cancer in patients undergoing staging investigations. Materials and Methods Ethics Statement The study was approved by the North London Research Ethics Committee (10/H0724/72). Written informed consent was obtained from all participants. Study Design Lymph node samples were obtained from adult patients undergoing EBUS-TBNA for investigation of mediastinal lymphadenopathy (see e-Appendix 1 for a description of conventional assessments). Cases were classified according to predefined diagnostic criteria given in e-Table 1.17

Materials and Methods Ethics Statement The study was approved by the North London Research Ethics Committee (10/H0724/72). Written informed consent was obtained from all participants. Study Design Lymph node samples were obtained from adult patients undergoing EBUS-TBNA for investigation of mediastinal lymphadenopathy (see e-Appendix 1 for a description of conventional assessments). Cases were classified according to predefined diagnostic criteria given in e-Table 1.17 Transcriptional Profiling by cDNA Microarray Lymph node cores harvested into RNALater (Qiagen) and homogenized in Qiazol (Qiagen) were used to obtain total RNA using the RNEasy Micro kit (Qiagen). Samples were processed for Agilent microarrays, and data were normalized as previously described (e-Appendix 1).18 Principal component analysis (PCA) was performed using the prcomp function in R to compare global gene expression profiles. Significant gene expression differences between samples were identified by t tests (P < .05) using MultiExperiment Viewer (version 4.6.0) and restricted to those with greater than twofold differences. Pathway overrepresentation analysis of differentially expressed genes was conducted using InnateDB, and transcriptional regulation of specific gene expression profiles was assessed by oPossum single transcription factor binding site enrichment analysis as previously described.19, 20 Additional details for the analysis of microarray data are provided in e-Appendix 1 and e-Figs 1, 2, 6. Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-2547.

single transcription factor binding site enrichment analysis as previously described.19, 20 Additional details for the analysis of microarray data are provided in e-Appendix 1 and e-Figs 1, 2, 6. Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-2547. Case Classification Using Machine Learning Support vector machines (SVMs) for binary computational classification of high-dimensional data were trained to classify samples using selected gene signatures.21 Given a set of training data, an SVM establishes a model that optimizes separation of data points from two groups. The SVM algorithms were implemented using the kernlab package in R3.0.2 with a linear kernel, which allows the influence of each gene in the model to be weighted (see e-Appendix 1 for a detailed description of SVM classification). Bootstrap sampling with replacement was used to select training cases to optimize the SVM model. This model was then used to classify remaining data that were not included in the training cohort, and the process was repeated 100 times to evaluate its performance. Sensitivity and specificity values for multiple SVM models were then presented in receiver operating characteristic (ROC) curve analysis using the “pROC” package in R3.0.2. In additional assessments of the SVM models, we used leave-one-out cross-validation, in which the training set comprises all but one of the samples, which is then tested, and the process is repeated to test each individual sample. The minimum number of genes needed for accurate SVM classification was assessed by training the SVM on 10 randomly chosen cases to determine weight values for each gene, iteratively refined in 100 training sets. A cumulative sequence of genes ranked by weight was then used to determine SVM classification accuracy in distinct test cases using ROC curves.

for accurate SVM classification was assessed by training the SVM on 10 randomly chosen cases to determine weight values for each gene, iteratively refined in 100 training sets. A cumulative sequence of genes ranked by weight was then used to determine SVM classification accuracy in distinct test cases using ROC curves. Results Transcriptomes Reflect Associated Molecular Pathologies We performed transcriptional profiling of EBUS-TBNA samples from 88 patients (Table 1). Detailed information about study subjects is given in e-Table 2. First, we focused our analysis on patients with “definite” diagnoses of sarcoidosis, TB, cancer, or reactive lymphadenopathy, ascertained by routine clinical and laboratory assessments. Comparison of EBUS-TBNA genomewide data from these patients by PCA to visualize the greatest co-correlated differences between individual samples shows that sarcoidosis, TB, and reactive lymph node samples clustered together, and most but not all cancer samples clustered separately (Fig 1).

and laboratory assessments. Comparison of EBUS-TBNA genomewide data from these patients by PCA to visualize the greatest co-correlated differences between individual samples shows that sarcoidosis, TB, and reactive lymph node samples clustered together, and most but not all cancer samples clustered separately (Fig 1). Despite the overlap in the clustering analysis described above, direct comparison of gene expression data identified differentially expressed genes between granulomatous and nongranulomatous lymph node samples (488 genes), between TB and sarcoidosis samples (58 genes), and between malignant and reactive samples (1,223 genes) (see e-Tables 3-5 for differentially expressed gene lists). Granulomatous lymph nodes were significantly enriched for genes associated with immunologic processes integral to cell-mediated immunity and granuloma formation and regulated by canonical transcription factors involved in proinflammatory and cytokine responses (e-Fig 3, e-Table 6). In keeping with previous peripheral blood transcriptional profiling studies,13, 14, 15 genomewide transcriptional profiles of TB and sarcoidosis lymph node samples were very similar. We identified only 16 genes with significantly higher expression in sarcoidosis and 42 genes with higher expression levels in TB lymph node samples (e-Fig 4).

ous peripheral blood transcriptional profiling studies,13, 14, 15 genomewide transcriptional profiles of TB and sarcoidosis lymph node samples were very similar. We identified only 16 genes with significantly higher expression in sarcoidosis and 42 genes with higher expression levels in TB lymph node samples (e-Fig 4). In the nongranulomatous samples, malignant lymph node samples were significantly enriched for genes involved in cell cycle control and extracellular matrix interactions, consistent with processes related to cancer development and metastasis (e-Fig 5), and under the transcriptional regulation of Kruppel-like factors and other zinc finger protein family members (e-Table 7). These control cell proliferation, differentiation, migration, and pluripotency in normal tissues and regulate cancer cell proliferation, apoptosis, and metastasis in many human tumors.22

(e-Fig 5), and under the transcriptional regulation of Kruppel-like factors and other zinc finger protein family members (e-Table 7). These control cell proliferation, differentiation, migration, and pluripotency in normal tissues and regulate cancer cell proliferation, apoptosis, and metastasis in many human tumors.22 Machine Learning Discriminates Diagnostic Categories SVM are data-driven computational algorithms that can “learn” to discriminate between high-dimensional data such as genomewide transcriptomes. Therefore, we assessed the performance of SVM discrimination between diagnostic categories using the differentially expressed transcriptional signatures described above. In this analysis, we used repeated bootstrap subsampling cross-validation23, 24 with five, 15, or 25 training cases to evaluate the classification performance of the SVM, ensuring the test set was always independent of the training set. ROC curves showed that the SVM performance improved as the cohort sample size increased, achieving area under the curve (AUC) values > 0.9 in each case (Fig 2). A recognized limitation of SVM is risk of overfitting of training data, but the high levels of accuracy consistently achieved in multiple iterations using different combinations of training and test data strongly suggest overfitting does not confound the present analysis.

curve (AUC) values > 0.9 in each case (Fig 2). A recognized limitation of SVM is risk of overfitting of training data, but the high levels of accuracy consistently achieved in multiple iterations using different combinations of training and test data strongly suggest overfitting does not confound the present analysis. Next we assessed the performance of SVM classification in a two-step decision tree sequence, using leave-one-out cross-validation as an alternative cross-validation strategy that optimizes the size of the training cohort (Fig 3). In step one, we classified each sample as granulomatous or nongranulomatous. In step two, samples classified as granulomatous were then subclassified as TB or sarcoidosis, and those classified as nongranulomatous were subclassified as cancer or reactive lymphadenopathy. We achieved excellent specificity across all four diagnostic groups and high sensitivity for detection of malignancy (93%) and sarcoidosis (85%) (Table 2, e-Tables 8, 9). This analysis was less sensitive for identification of reactive lymphadenopathy (80%) and TB (67%), but we noted a positive relationship between test sensitivity and sample size for each diagnosis (Table 2).

ic groups and high sensitivity for detection of malignancy (93%) and sarcoidosis (85%) (Table 2, e-Tables 8, 9). This analysis was less sensitive for identification of reactive lymphadenopathy (80%) and TB (67%), but we noted a positive relationship between test sensitivity and sample size for each diagnosis (Table 2). We also sought to identify the minimum number of genes needed to accurately classify cases using each SVM. The most discriminating genes were identified by their weighting in the training data sets using repeated bootstrap subsampling cross-validation (see e-Table 10 for lists of the most discriminating genes). Cumulative inclusion of these genes in order of their weighting in the SVM training data improved the ROC curve AUCs (Fig 4A). To achieve ROC curve AUC > 0.9, expression data from at least five genes were required to discriminate granulomatous and nongranulomatous cases, 19 genes to discriminate TB and sarcoidosis, and 150 genes to discriminate cancer and reactive cases. Peripheral blood gene signatures that distinguish TB and sarcoidosis have been reported previously.13, 14, 15 These show only modest overlap with our lymph node-derived differential gene expression signature (Fig 4B) and clearly perform less well in discriminating TB from sarcoidosis lymph node samples by SVM classification (Figs 4C, 4D). This suggests that the gene signatures exhibit context specificity and that peripheral blood data may not faithfully reflect the site of disease.

ed differential gene expression signature (Fig 4B) and clearly perform less well in discriminating TB from sarcoidosis lymph node samples by SVM classification (Figs 4C, 4D). This suggests that the gene signatures exhibit context specificity and that peripheral blood data may not faithfully reflect the site of disease. SVM Classification of Undiagnosed Lymphadenopathy Finally, we tested the two-step SVM decision-tree sequence described above on cases in which a “definite” diagnosis could not be made at the time of EBUS-TBNA. We compared SVM classification of these samples with the final diagnosis based on follow-up data (Table 3). SVM analysis identified “granulomatous disease” in almost all specimens with histologic evidence of granulomas and two samples without granulomas on histology (Possible S2 and S3). A definite diagnosis of sarcoidosis was later confirmed in one case (Possible S2), after assessment of further lymph node samples obtained by mediastinoscopy revealed noncaseating granulomas. Another case (Probable TB2) was classified by SVM as “sarcoidosis” but subsequently confirmed as “TB.” The majority of “possible cancer” samples were classified as “reactive,” consistent with histologic assessments. However, SVM classified two cases as “cancer,” in the absence of histologic evidence. Subsequent examination of surgically resected specimens confirmed metastatic carcinoma in the lymph node, which had shown no evidence of malignancy on EBUS-TBNA 6 weeks previously (Possible C10). Strikingly, SVM also predicted the presence of “cancer” 4 months before tumor involvement was discernible on histology in another case (Possible C11). SVM predicted “cancer” in one further individual (U2) with a presumptive diagnosis of sarcoidosis and no evidence of malignancy during long-term follow up.

ossible C10). Strikingly, SVM also predicted the presence of “cancer” 4 months before tumor involvement was discernible on histology in another case (Possible C11). SVM predicted “cancer” in one further individual (U2) with a presumptive diagnosis of sarcoidosis and no evidence of malignancy during long-term follow up. Discussion The development of EBUS has revolutionized investigation of mediastinal lymphadenopathy by facilitating minimally invasive sampling of lymph nodes at the site of disease. Transcriptional profiling of EBUS-TBNA in granulomatous and malignant lymphadenopathy revealed clear evidence of the key biologic processes relevant to each disease. Samples from granulomatous diseases were enriched for immune cell recruitment and activation as well as antigen presentation and interferon-γ signaling, which characterize granulomatous inflammation.25 Malignant lymph node samples were enriched for molecules involved in control of cell division and interactions with the extracellular matrix or adjacent cells, the molecular mechanisms that underpin cancer development and metastasis. Bioinformatic analysis of the transcriptional control of genes involved in these processes confirmed the importance of nuclear factor-κB as a principal regulator of inflammatory responses in granulomatous disease and Kruppel-like factors in transcriptional regulation of key events in tumor generation in humans in vivo.22 Macroscopic blood contamination of EBUS-TBNA was variable, but globin transcript levels were comparable between the diagnostic groups and therefore did not produce a systematic bias. Although globin depletion is advocated in blood samples to improve sensitivity,26 biologically plausible gene expression differences between groups in our study were still evident. Nonetheless, assessment of globin depletion or RNA sequencing to increase sensitivity should be considered in future studies.

tematic bias. Although globin depletion is advocated in blood samples to improve sensitivity,26 biologically plausible gene expression differences between groups in our study were still evident. Nonetheless, assessment of globin depletion or RNA sequencing to increase sensitivity should be considered in future studies. The similarity between TB and sarcoidosis profiles suggests that by the time of clinical presentation, the molecular pathology in these diseases is comparable. Sarcoidosis samples showed higher expression of genes with putative roles in granuloma formation, including cathepsin K, transmembrane 7 superfamily member 4, and chemokine (C-C motif) ligand 21.27, 28, 29 We hypothesize that this may reflect the more organized, noncaseating granuloma phenotype typical of sarcoidosis.30 The most highly expressed gene in sarcoidosis compared with TB lymph nodes was chitotriosidase (CHIT)1, which encodes a lysosomal hydrolase that degrades fungal cell wall chitin.31 Elevated levels of chitotriosidase have also been reported in BAL fluid and serum of individuals with sarcoidosis.32, 33 A putative role for inhaled antigens is recognized in sarcoidosis.34 Therefore, aberrant immune responses to fungi in pathogenesis of sarcoidosis merit further investigation.

ngal cell wall chitin.31 Elevated levels of chitotriosidase have also been reported in BAL fluid and serum of individuals with sarcoidosis.32, 33 A putative role for inhaled antigens is recognized in sarcoidosis.34 Therefore, aberrant immune responses to fungi in pathogenesis of sarcoidosis merit further investigation. The performance of SVM using differential gene expression signatures to distinguish between granulomatous and nongranulomatous disease, cancer and reactive lymphadenopathy, and TB and sarcoidosis showed excellent promise to improve diagnostic classification, providing AUC statistics of > 0.9 in each case. Importantly, the SVM models identified the most influential genes and opportunity to test the feasibility of using targeted transcriptional analysis rather than genomewide technologies in future studies. Our analysis suggested that this might be possible for discriminating granulomatous from nongranulomatous disease and TB from sarcoidosis using multiplex quantitative polymerase chain reaction analysis of 20 to 30 genes. However, expression data from more than 150 genes were required to discriminate cancer and reactive lymphadenopathy.

ested that this might be possible for discriminating granulomatous from nongranulomatous disease and TB from sarcoidosis using multiplex quantitative polymerase chain reaction analysis of 20 to 30 genes. However, expression data from more than 150 genes were required to discriminate cancer and reactive lymphadenopathy. The application of SVM in a two-step decision tree model was significantly more sensitive than mycobacterial culture for identification of TB and equivalent to histologic detection of noncaseating granulomas for sarcoidosis.4, 5, 35, 36 Notably, SVM predicted the presence of granulomatous disease in one EBUS-guided biopsy with no granulomas on histology, from an individual subsequently diagnosed with sarcoidosis after lymph node samples obtained by mediastinoscopy confirmed granulomatous inflammation. Importantly, differentially expressed genes that distinguished sarcoidosis from TB lymph node transcriptional profiles performed better than previously published gene signatures derived from peripheral blood,13, 14, 15 suggesting that peripheral blood does not wholly reflect the profiles at the site of disease and strengthening the case for lymph node sampling for maximum diagnostic accuracy.

from TB lymph node transcriptional profiles performed better than previously published gene signatures derived from peripheral blood,13, 14, 15 suggesting that peripheral blood does not wholly reflect the profiles at the site of disease and strengthening the case for lymph node sampling for maximum diagnostic accuracy. SVM also classified specimens from two individuals undergoing EBUS-guided lymph node sampling for lung cancer staging as “cancer,” despite no histologic evidence of malignancy. In keeping with this, SVM classification was slightly more sensitive than the reported median sensitivity of histologic identification of malignancy.2 In both cases, further biopsies taken from these nodes 6 weeks to 4 months after the initial specimens, demonstrated tumor infiltration that may reflect progression of micrometastatic disease. These examples highlight the potential power of combining transcriptional profiling with a computational classification algorithm to provide a more sensitive approach for identification of subtle molecular indicators of granulomatous disease or malignancy that are discernible before histologic or cytologic abnormalities become evident. Importantly, this strategy has the potential to identify individuals who might benefit from more frequent surveillance and to improve the current low negative predictive value of 40% for EBUS-guided lymph node sampling in isolated mediastinal lymphadenopathy,37 which will avoid unnecessary invasive mediastinoscopy procedures and minimize delays in definitive treatments. We recognize that SVM analysis of transcriptome data did not correctly classify every case in our study; therefore, this approach does not currently supersede clinical assessment and laboratory investigations. However, both accuracy and sensitivity of SVM classification improved as the training dataset sample size increased, suggesting the potential for more precise discrimination and increased sensitivity with expansion of the current cohort. Our data pave the way for larger-scale observational cohorts to validate the findings presented here and controlled trials to investigate the impact of this approach on clinical outcomes.

increased, suggesting the potential for more precise discrimination and increased sensitivity with expansion of the current cohort. Our data pave the way for larger-scale observational cohorts to validate the findings presented here and controlled trials to investigate the impact of this approach on clinical outcomes. Conclusions We propose that transcriptional profiling of lymph node samples from the site of disease combined with machine learning data analysis offers a novel strategy with molecular-level resolution that could be applied to augment conventional investigation of clinically ambiguous cases of mediastinal lymphadenopathy. This merits further evaluation in future large-scale clinical trials. Supplementary Data e-Online Data Acknowledgments Author contributions: G. S. T. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis, including and especially any adverse effects. G. S. T., M. N., and S. M. J. contributed to study design; G. S. T., G. H., J. B., N. N., S. M. J., and A. B. contributed to sample collection; G. S. T., R. F. M., A. B., and M. N. contributed to data collection; G. S. T. and N. S. performed the experiments; G. S. T., M. N., N. T., K. B., and B. M. C. analyzed the data; G. S. T. and M. N. contributed to writing the paper; G. S. T., N. T., B. M. C., K. B., N. S., G. H., J. B., A. B., N. N., S. M. J., R. F. M., and M. N. contributed to critical revision of the manuscript for important intellectual content and final approval of the manuscript.

. T., K. B., and B. M. C. analyzed the data; G. S. T. and M. N. contributed to writing the paper; G. S. T., N. T., B. M. C., K. B., N. S., G. H., J. B., A. B., N. N., S. M. J., R. F. M., and M. N. contributed to critical revision of the manuscript for important intellectual content and final approval of the manuscript. Financial/nonfinancial disclosures: None declared. Role of sponsors: The funding bodies had no involvement in the development of this research or manuscript. Additional information: The e-Appendix, e-Figures, and e-Tables can be found in the Supplemental Materials section of the online article. FUNDING/SUPPORT: This work was supported by a Rosetrees Trust grant to G. S. T., a Wellcome Trust Senior Clinical Fellowship [Grant WT091730AIA] to S. M. J., and UK National Institute for Health Research Biomedical Research Centre funding to University College London Hospital and University College London. Figure 1 Clustering analysis of genomewide lymph node profiles does not distinguish disease groups. Comparison of genomewide transcriptional profiles of lymph node samples by principal component analysis (PCA) shows that the majority of cancer samples cluster away from all other disease groups in PC1, responsible for the greatest differences within the data. However, sarcoidosis, TB, and reactive lymph nodes as well as some cancer samples clustered together in both PC1 and PC2 in this analysis, which was unable to segregate individual disease groups. Each symbol represents a sample.

all other disease groups in PC1, responsible for the greatest differences within the data. However, sarcoidosis, TB, and reactive lymph nodes as well as some cancer samples clustered together in both PC1 and PC2 in this analysis, which was unable to segregate individual disease groups. Each symbol represents a sample. Figure 2 Support vector machines (SVMs) classification performance improves as cohort sample size increases. The performance of SVM to classify cases by training on signatures comprising differentially expressed genes (greater than twofold difference and P < .05, t test) between comparator groups using bootstrap sampling of five, 15, or 25 cases with replacement (100 iterations), is represented by receiver operating characteristic (ROC) curves. Increasing the training dataset sample size progressively improves the ability of SVMs to correctly distinguish granulomatous (n = 28) from nongranulomatous (n = 37) disease (A), sarcoidosis (n = 19) from TB (n = 9) (B), and malignant (n = 27) from reactive (n = 10) lymph nodes (C), with AUC values of > 0.9 once the cohort comprises 25 samples. AUC = area under the curve.

ely improves the ability of SVMs to correctly distinguish granulomatous (n = 28) from nongranulomatous (n = 37) disease (A), sarcoidosis (n = 19) from TB (n = 9) (B), and malignant (n = 27) from reactive (n = 10) lymph nodes (C), with AUC values of > 0.9 once the cohort comprises 25 samples. AUC = area under the curve. Figure 3 Support vector machines classification sequence. In this analysis, SVMs are trained using leave-one-out cross-validation. Step 1: Cases are subjected to initial SVM analysis using the 488-gene signature, which distinguishes granulomatous from nongranulomatous disease. Step 2: Samples classified as granulomatous disease subsequently undergo SVM testing using the 58-gene signature, which discriminates sarcoidosis from TB, and those classified as nongranulomatous disease undergo further SVM evaluation using the 1,223-gene signature, which distinguishes cancer from reactive lymph node. See Figure 2 legend for expansion of abbreviation.

e subsequently undergo SVM testing using the 58-gene signature, which discriminates sarcoidosis from TB, and those classified as nongranulomatous disease undergo further SVM evaluation using the 1,223-gene signature, which distinguishes cancer from reactive lymph node. See Figure 2 legend for expansion of abbreviation. Figure 4 Lymph node transcriptional signatures show greater power to discriminate TB from sarcoidosis than peripheral blood signatures. A, Performance of cumulative genes in rank order of their weighting in each SVM is represented by the mean (± 95% CI) of ROC curve AUCs in 100 bootstrap subsampling cross-validation iterations. B, The published peripheral blood signatures (Maertzdorf, Bloom, Koth) and the LN signature identified in this study that differentiate TB from sarcoidosis exhibit minimal overlap. C, ROC curves represent the ability of SVM algorithms to classify sarcoidosis and TB lymph node samples when trained by bootstrap sampling with replacement on transcriptional signatures that distinguish sarcoidosis from TB lymph nodes (Tomlinson) or peripheral blood profiles (Bloom, Maertzdorf, Koth). D, The LN-derived gene signature performs significantly better (AUC = 0.92) than peripheral blood-derived signatures (AUC all < 0.7) in classifying TB and sarcoidosis samples. LN = lymph node. See Figure 2 legend for expansion of other abbreviations. Table 1 Summary of Study Subjects’ Demographic Data

Figure 4 Lymph node transcriptional signatures show greater power to discriminate TB from sarcoidosis than peripheral blood signatures. A, Performance of cumulative genes in rank order of their weighting in each SVM is represented by the mean (± 95% CI) of ROC curve AUCs in 100 bootstrap subsampling cross-validation iterations. B, The published peripheral blood signatures (Maertzdorf, Bloom, Koth) and the LN signature identified in this study that differentiate TB from sarcoidosis exhibit minimal overlap. C, ROC curves represent the ability of SVM algorithms to classify sarcoidosis and TB lymph node samples when trained by bootstrap sampling with replacement on transcriptional signatures that distinguish sarcoidosis from TB lymph nodes (Tomlinson) or peripheral blood profiles (Bloom, Maertzdorf, Koth). D, The LN-derived gene signature performs significantly better (AUC = 0.92) than peripheral blood-derived signatures (AUC all < 0.7) in classifying TB and sarcoidosis samples. LN = lymph node. See Figure 2 legend for expansion of other abbreviations. Table 1 Summary of Study Subjects’ Demographic Data Diagnosis No. Age Range, y Sex M (F) Ethnicity Definite sarcoidosis 19 24-80 12 (7) 16 Eurasian, 3 African Definite TB 9 21-71 8 (1) 7 Eurasian, 2 African Reactive 10 33-78 9 (1) 10 Eurasian Definite cancer 27 49-86 16 (11) 25 Eurasian, 1 East Asian, 1 African Possible sarcoidosis 3 38-50 3 (0) 2 Eurasian, 1 Latin American Probable TB 2 30-35 0 (2) 2 Eurasian Possible cancer 12 56-82 5 (7) 12 Eurasian Undetermined 6 44-58 2 (4) 5 Eurasian, 1 African Table 2 Sensitivity and Specificity of SVM Classification

Definite cancer 27 49-86 16 (11) 25 Eurasian, 1 East Asian, 1 African Possible sarcoidosis 3 38-50 3 (0) 2 Eurasian, 1 Latin American Probable TB 2 30-35 0 (2) 2 Eurasian Possible cancer 12 56-82 5 (7) 12 Eurasian Undetermined 6 44-58 2 (4) 5 Eurasian, 1 African Table 2 Sensitivity and Specificity of SVM Classification Diagnosis No. Sensitivity, % Specificity, % Sarcoidosis 19 85 96 TB 9 67 98 Reactive 10 80 93 Cancer 27 93 92 SVM = support vector machine. Table 3 SVM Classification of Cases Without a Definite Clinical Diagnosis

Definite cancer 27 49-86 16 (11) 25 Eurasian, 1 East Asian, 1 African Possible sarcoidosis 3 38-50 3 (0) 2 Eurasian, 1 Latin American Probable TB 2 30-35 0 (2) 2 Eurasian Possible cancer 12 56-82 5 (7) 12 Eurasian Undetermined 6 44-58 2 (4) 5 Eurasian, 1 African Table 2 Sensitivity and Specificity of SVM Classification Diagnosis No. Sensitivity, % Specificity, % Sarcoidosis 19 85 96 TB 9 67 98 Reactive 10 80 93 Cancer 27 93 92 SVM = support vector machine. Table 3 SVM Classification of Cases Without a Definite Clinical Diagnosis Diagnosis Clinical Question Histology SVM 1 SVM 2 SVM 3 Clinical Outcome Possible S1 Sarcoidosis? G NG … R Clinical diagnosis of sarcoidosis. Spontaneous resolution of symptoms, under observation only. Possible S2 Sarcoidosis? NG G S … Definite sarcoidosis confirmed by noncaseating granulomas on lymph node sample from mediastinoscopy. Clinical improvement with steroid treatment. Possible S3 Sarcoidosis? NG G S … Spontaneous improvement in clinical symptoms and chest radiograph lymphadenopathy. Probable TB1 TB? G G TB … Good clinical and radiologic response to empirical TB treatment. Probable TB2 TB? G G S … Pleural fluid acid- alcohol-fast bacilli positive, Mycobacterium tuberculosis culture negative. Good clinical response to empirical TB treatment. Possible C1 Metastatic endometrial cancer? G G … R Palliative chemotherapy for presumed lung metastases. Possible C2 Sarcoidosis/lymphoma? G G … R Died of non-Hodgkin’s lymphoma diagnosed on bone marrow biopsy. Possible C3 Metastatic bowel cancer? G G … R Patient declined empirical TB treatment and did not attend further respiratory follow-up appointment. Possible C4 Metastatic bowel cancer? G G … R Died. Possible C5 Metastatic lung cancer? NG NG … R Developed further lung lesion and fluorodeoxyglucose-avid lymph node following wedge resection of right lower lobe tumor. Possible C6 Cancer? NG NG … R Remains in remission from lung cancer—no further treatment. Possible C7 Metastatic lung cancer? NG NG … R Died. Possible C8 Metastatic bladder/renal cancer? NG NG … R Clinically well after neoadjuvant chemotherapy, right nephroureterectomy, cystoprostatectomy, and ileal conduit formation. Possible C9 Metastatic lung cancer? NG NG … R Neoadjuvant chemotherapy, left upper lobe resection, consolidation chemotherapy. Clinically stable but mediastinal lymphadenopathy still evident on CT scan. Possible C10 Metastatic lung cancer? NG NG … C Left upper lobe resection specimen showed metastatic carcinoma in the lymph node, which had shown no histologic evidence of malignancy on samples obtained via EBUS. Possible C11 Metastatic lung cancer?

ically stable but mediastinal lymphadenopathy still evident on CT scan. Possible C10 Metastatic lung cancer? NG NG … C Left upper lobe resection specimen showed metastatic carcinoma in the lymph node, which had shown no histologic evidence of malignancy on samples obtained via EBUS. Possible C11 Metastatic lung cancer? NG NG … C Previous right middle lobe resection for lung adenocarcinoma and had confirmed metastatic adenocarcinoma on a second EBUS procedure performed 4 mo later due to progressive lymph node enlargement. Possible C12 Metastatic bladder cancer? NG NG … R Clinically stable, static appearance of lymphadenopathy on interval CT scan; discharged from respiratory follow-up. U1 Sarcoidosis/TB? G G S … Empirical TB treatment stopped after 2 mo as no response. Now under observation—remains clinically stable. U2 Sarcoidosis? NG NG … C Presumptive diagnosis of sarcoidosis-induced peripheral neuropathy. Minor clinical improvement with steroid and cyclophosphamide treatment. U3 Sarcoidosis/lymphoma? NG G S … Unknown—patient did not attend follow-up clinical appointments. U4 Sarcoidosis? G G S … Remains clinically stable—under observation only. U5 Sarcoidosis/TB? G G S … Spontaneous improvement in clinical symptoms and chest radiograph lymphadenopathy. Subsequently completed 6 mo empirical TB treatment (Mantoux 45 mm, interferon-γ release assay negative). Now well—discharged. U6 Sarcoidosis/TB/lymphoma? NG NG … R Clinically well after 6 mo empirical TB treatment—discharged. C = cancer; EBUS = endobronchial ultrasound; G = granulomatous; NG = nongranulomatous; R = reactive; S = sarcoidosis; U = undetermined, See Table 2 legend for expansion of other abbreviation.

Normal mucociliary clearance is essential in pulmonary defense.1 Abnormal mucociliary clearance is a feature of asthma, particularly in severe disease, as a consequence of ciliary dysfunction.2 This ciliary dysfunction might contribute to the persistent inflammation and susceptibility to infection in the asthmatic airway, as evidenced by higher bacterial DNA level3, 4 and fungal colonization, in particular Aspergillus fumigatus.5

sthma, particularly in severe disease, as a consequence of ciliary dysfunction.2 This ciliary dysfunction might contribute to the persistent inflammation and susceptibility to infection in the asthmatic airway, as evidenced by higher bacterial DNA level3, 4 and fungal colonization, in particular Aspergillus fumigatus.5 Importantly, asthma is a heterogeneous disease and in addition to the Th2-mediated eosinophilic paradigm, neutrophilic-predominant inflammation is a feature of one third of patients with asthma.6, 7, 8, 9 Although the cause of neutrophilic asthma is unclear, it is associated with increased presence of proinflammatory and Th1 cytokines in sputum and bacterial colonization.10 Pathogens can exert direct toxic effects on ciliary function or indirectly via oxidative stress,11 and stimuli such as infection,12 pollutants13 and proinflammatory mediators14 can induce production of reactive oxygen species (ROS). The nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX)/dual oxidase (DUOX) family plays an important role in the generation of ROS and contains seven members: NOX1-5 and DUOX1/2.15 DUOX expression in the bronchial epithelium has been reported,16, 17, 18 and DUOXs have been shown to be important for neutrophil recruitment to the airways.19 In addition, we have previously reported that NOX4 expression was increased in airway smooth muscle in asthma, leading to increased ROS production and intrinsic airway smooth muscle hypercontractility.20 Whether NOX/DUOX expression is altered in the bronchial epithelium and possibly contributes to an increased susceptibility to ciliary dysfunction in asthma is unknown.

expression was increased in airway smooth muscle in asthma, leading to increased ROS production and intrinsic airway smooth muscle hypercontractility.20 Whether NOX/DUOX expression is altered in the bronchial epithelium and possibly contributes to an increased susceptibility to ciliary dysfunction in asthma is unknown. We hypothesized that ciliary dysfunction in asthma is due to a combination of an intrinsic abnormality in ciliary function and airway inflammation. To test our hypothesis, we assessed: (1) the ciliary beat frequency (CBF) in fresh ex vivo epithelial cells from patients with asthma with and without sputum neutrophilia, (2) the role of NADPH oxidases in ciliary function and their specificity to the neutrophilic asthma phenotype, and (3) the effects of NADPH oxidase inhibition on ciliary function in a murine in vivo ovalbumin (OVA) sensitization and challenge model. Methods A more detailed description of the methods is available in e-Appendix 1. Subjects Patients with asthma and healthy control subjects were recruited from a single center (Glenfield Hospital). Asthma severity was defined by using the Global Initiative for Asthma (GINA) treatment steps (mild to moderate asthma, GINA steps 1-3; severe asthma, GINA steps 4-5).21 The study was approved by the Leicestershire Ethics Committee (UHL 10613), and patients gave their written informed consent.

nter (Glenfield Hospital). Asthma severity was defined by using the Global Initiative for Asthma (GINA) treatment steps (mild to moderate asthma, GINA steps 1-3; severe asthma, GINA steps 4-5).21 The study was approved by the Leicestershire Ethics Committee (UHL 10613), and patients gave their written informed consent. Epithelial Cells Primary epithelial cells were isolated from bronchial brushes during bronchoscopy. Experiments were undertaken by using epithelial human bronchial epithelial cells, characterized by cytokeratin 5 and 14 (Abcam) expression.22 Ciliary function was assessed in ciliated air-liquid interface (ALI) cultures or fresh epithelial strips by using video microscopy as previously described.23, 24

ng bronchoscopy. Experiments were undertaken by using epithelial human bronchial epithelial cells, characterized by cytokeratin 5 and 14 (Abcam) expression.22 Ciliary function was assessed in ciliated air-liquid interface (ALI) cultures or fresh epithelial strips by using video microscopy as previously described.23, 24 Immunohistochemistry and Immunofluorescence Human bronchial biopsy specimens were embedded in glycomethacrylate.20 Sections were stained by using an 8-oxo-dG monoclonal antibody, anti-NADPH oxidase 4 antibody (Abcam), anti-DUOX1 (Abcam), and anti-DUOX2 (Millipore) or corresponding isotype control (DAKO and ImmunoStep). Staining intensity above isotype control for 8-oxo-dG expression was assessed by using a semiquantitative scoring ranging from none to low, moderate, or high staining (0-3). For NOX4 and DUOX1/2, staining intensity was measured in all areas of epithelium by thresholding using CellˆF software (Olympus). All assessments were made by an observer blinded to the subjects' clinical characteristics. Cytospins of human bronchial epithelial cells were labeled with polyclonal rabbit antibodies to NOX1 and NOX4 (4 μg/mL, Insight Biotechnology; 4 μg/mL, Abcam, respectively) or the corresponding isotype control (BD Bioscience). They were indirectly labeled with an R-Phycoerythrin-conjugated secondary antibody (AbD Serotec). Cells were counterstained with 4′,6′-diamidino-2 phenylindole (1 μg/mL; Sigma-Aldrich).

to NOX1 and NOX4 (4 μg/mL, Insight Biotechnology; 4 μg/mL, Abcam, respectively) or the corresponding isotype control (BD Bioscience). They were indirectly labeled with an R-Phycoerythrin-conjugated secondary antibody (AbD Serotec). Cells were counterstained with 4′,6′-diamidino-2 phenylindole (1 μg/mL; Sigma-Aldrich). RNA Extraction, Real-Time Quantitative Polymerase Chain Reaction, and Gene Array Total RNA was extracted from epithelial ALI cultures by using an RNeasy Mini Kit with DNase I treatment (Qiagen). Complementary DNA synthesis and quantitative polymerase chain reaction were performed with a two-step real-time quantitative polymerase chain reaction kit (Invitrogen) by using the Chromo4 Real-Time Detector and Opticon Monitor 3 (Bio-Rad Laboratories). The internal normalizer gene was 18S RNA amplified with 18S primer forward (h18SRNA.891F:GTTGGTTTTCGGAACTGAGG) and 18S reverse primer (h18SRNA.1090R:GCATCGTTTATGGTCGGAAC); amplification of NOX4 was with primers forward (hNox4.598F:TGGCTGCCCATCTGGTGAATG) and reverse (hNox4.878R:CAGCAGCCCTCCTGAAACATGC). Primers (Operon MWG Biotech) and reaction mix (Invitrogen) were used as described previously.20 Relative quantification of NOX4 messenger RNA (mRNA) was performed by using the comparative 2−ΔΔCt method and expressed as fold-change. The internal normalizer gene was 18S RNA.

d reverse (hNox4.878R:CAGCAGCCCTCCTGAAACATGC). Primers (Operon MWG Biotech) and reaction mix (Invitrogen) were used as described previously.20 Relative quantification of NOX4 messenger RNA (mRNA) was performed by using the comparative 2−ΔΔCt method and expressed as fold-change. The internal normalizer gene was 18S RNA. RNA expression levels from basal epithelial cells were examined by using the Human Genome U133A probe array (GeneChip; Affymetrix). Hybridized biotinylated complementary RNA was stained with streptavidin phycoerythrin (Molecular Probes), scanned with a HP GeneArray Scanner (Hewlett-Packard). Image data from each microarray were individually scaled to an intensity of 200 by using GeneChip Operating Software (Affymetrix). Scaled average differences and absolute call data were exported to text files for further analysis. Intracellular ROS Assay Intracellular ROS generation in human bronchial epithelial cells was quantified by using 5-(and-6)-carboxy-2′, 7′-dichlorofluorescein diacetate oxidation.20 In Vivo Model of OVA Sensitization and Challenge In a standard murine model of OVA sensitization and challenge protocol that was adapted from Caceres et al,25 airway inflammation, remodeling, and ciliary function were assessed. The University of Leicester Ethics Committee and the UK Home Office approved the experimental protocols.

l of OVA Sensitization and Challenge In a standard murine model of OVA sensitization and challenge protocol that was adapted from Caceres et al,25 airway inflammation, remodeling, and ciliary function were assessed. The University of Leicester Ethics Committee and the UK Home Office approved the experimental protocols. NOX1/4 Inhibitor GKT137831 To investigate the effects of NADPH oxidase inhibition in asthmatic epithelial cells and a murine model, we used GKT137831 (Genkyotex),26 a NADPH oxidase inhibitor that is selective for NOX1 and NOX4 vs NOX2, NOX3, and NOX5. GKT137831 has greater selectivity and potency for NOX4 than for NOX1. Its potential inhibitory effects on DUOX1/2 have not been fully explored.

bition in asthmatic epithelial cells and a murine model, we used GKT137831 (Genkyotex),26 a NADPH oxidase inhibitor that is selective for NOX1 and NOX4 vs NOX2, NOX3, and NOX5. GKT137831 has greater selectivity and potency for NOX4 than for NOX1. Its potential inhibitory effects on DUOX1/2 have not been fully explored. Statistical Analysis Statistical analyses were performed by using PRISM version 6 (GraphPad Software). Parametric data were described as mean ± SEM; geometric mean ± 95% confidence intervals (CIs) were used for log normally distributed data, and nonparametric data are presented as median ± interquartile range. Comparisons between groups were assessed by using t tests (Mann-Whitney U test for nonparametric data). Comparisons across ≥ 3 groups were analyzed by using an analysis of variance (ANOVA) model (Kruskal-Wallis test for nonparametric data) with post hoc pairwise comparisons with the Tukey test (the Dunn test for nonparametric analyses). Sputum eosinophil (> 3%) and neutrophil (> 61%) counts were used as cutoffs to define inflammatory asthma phenotypes.27 Statistical analyses were performed as indicated in the figure legends, and P < .05 was taken as the level of statistical significance. Results Clinical characteristics of the subjects who provided bronchial biopsy specimens for immunohistochemistry or epithelial brushings for ex vivo bronchial epithelial strips are shown Table 1.

Statistical Analysis Statistical analyses were performed by using PRISM version 6 (GraphPad Software). Parametric data were described as mean ± SEM; geometric mean ± 95% confidence intervals (CIs) were used for log normally distributed data, and nonparametric data are presented as median ± interquartile range. Comparisons between groups were assessed by using t tests (Mann-Whitney U test for nonparametric data). Comparisons across ≥ 3 groups were analyzed by using an analysis of variance (ANOVA) model (Kruskal-Wallis test for nonparametric data) with post hoc pairwise comparisons with the Tukey test (the Dunn test for nonparametric analyses). Sputum eosinophil (> 3%) and neutrophil (> 61%) counts were used as cutoffs to define inflammatory asthma phenotypes.27 Statistical analyses were performed as indicated in the figure legends, and P < .05 was taken as the level of statistical significance. Results Clinical characteristics of the subjects who provided bronchial biopsy specimens for immunohistochemistry or epithelial brushings for ex vivo bronchial epithelial strips are shown Table 1. CBF was reduced in patients with neutrophilic vs nonneutrophilic asthma (mean difference ± 95% CIs, –2.97 ± 0.87; P = .003) (Fig 1A) and was correlated with the percent sputum neutrophil count (r = –0.70; P < .001) (Fig 1B). Compared with our previously published CBF in healthy control subjects (10.7 ± 0.4 Hz),2 CBF was reduced in those patients with asthma with sputum neutrophilia (5.8 ± 0.6 Hz; P < .0001) but not in those without sputum neutrophilia (8.8 ± 0.5 Hz; P = .089). There was no relationship between sputum eosinophil count and CBF (r = 0.19; P = .40) (Fig 1C).

lished CBF in healthy control subjects (10.7 ± 0.4 Hz),2 CBF was reduced in those patients with asthma with sputum neutrophilia (5.8 ± 0.6 Hz; P < .0001) but not in those without sputum neutrophilia (8.8 ± 0.5 Hz; P = .089). There was no relationship between sputum eosinophil count and CBF (r = 0.19; P = .40) (Fig 1C). Oxidative stress-induced DNA damage, represented by semiquantitative assessment of 8-oxo-dG expression, was increased in the bronchial epithelium in patients with neutrophilic asthma (Kruskal-Wallis test, P = .005) (Figs 2A, 2B) and in those with severe disease vs mild to moderate disease and healthy control subjects (Kruskal-Wallis test, P = .036; data not shown). The intensity of bronchial epithelial 8-oxo-dG expression was also correlated to airflow obstruction (rs = –0.68; P < .001) (Fig 2C) and the percentage of sputum neutrophils (rs = 0.48; P = .02) (Fig 2D).

e disease vs mild to moderate disease and healthy control subjects (Kruskal-Wallis test, P = .036; data not shown). The intensity of bronchial epithelial 8-oxo-dG expression was also correlated to airflow obstruction (rs = –0.68; P < .001) (Fig 2C) and the percentage of sputum neutrophils (rs = 0.48; P = .02) (Fig 2D). Gene array analysis revealed the presence of NOX4 and DUOX1 and DUOX2 expression, but not the other NOX family members, in primary basal epithelial cells from ≥ 3 of the six donors with asthma. There was no significant difference in the proportion of healthy vs asthmatic donors who expressed NOX4, DUOX1, or DUOX 2, and their relative expression was not significantly different between cells from patients with asthma vs healthy control subjects (NOX4, 1.6-fold [0.79-3.1]; DUOX1, 0.87-fold [0.46-1.63]; and DUOX2, 1.12-fold [0.38-3.3]). The geometric mean (95% CI) percent β-actin expression was 0.07% (0.05-0.12) for NOX4, 0.69% (0.56-0.95) for DUOX1, and 0.45% (0.06-1.3) for DUOX2. We therefore focused our immunohistochemical analysis of bronchial biopsy specimens on NOX4 and DUOX1/2.

1]; DUOX1, 0.87-fold [0.46-1.63]; and DUOX2, 1.12-fold [0.38-3.3]). The geometric mean (95% CI) percent β-actin expression was 0.07% (0.05-0.12) for NOX4, 0.69% (0.56-0.95) for DUOX1, and 0.45% (0.06-1.3) for DUOX2. We therefore focused our immunohistochemical analysis of bronchial biopsy specimens on NOX4 and DUOX1/2. NOX4 protein expression in epithelial cells of tissue bronchial biopsy specimens was significantly increased in patients with neutrophilic asthma (29.5% ± 3.7%) compared with patients with nonneutrophilic asthma (18.5% ± 3.2%; P = .041) and with healthy control subjects (18.6% ± 2.9%; P = .032) (ANOVA, P = .040) (Figs 3A, 3C), and it was significantly correlated with sputum neutrophil count (r = 0.52; P =.042) but not with lung function. DUOX1 protein expression was increased in the epithelium in patients with nonneutrophilic (29.0% ± 7.4%) and neutrophilic (22.3% ± 4.7%) asthma vs healthy control subjects (9.8% ± 3.2%; P = .026 and P = .041, respectively); it was not differentially expressed in neutrophilic vs nonneutrophilic phenotypes (P = .465) (ANOVA, P = .048) (Figs 3B, 3D) and did not correlate with sputum neutrophil count (r = –0.06; P = .832). Although DUOX2 was expressed in a much lower percentage of epithelium, a significant increase was seen in patients with nonneutrophilic asthma (0.97% ± 0.29%; P = .026) but not neutrophilic asthma (0.58% ± 0.15%; P = .072) compared with healthy control subjects (0.27% ± 0.07%), and expression was not differentially expressed in neutrophilic vs nonneutrophilic phenotypes (P = .258) (ANOVA, P = .045) (e-Fig 1).

e was seen in patients with nonneutrophilic asthma (0.97% ± 0.29%; P = .026) but not neutrophilic asthma (0.58% ± 0.15%; P = .072) compared with healthy control subjects (0.27% ± 0.07%), and expression was not differentially expressed in neutrophilic vs nonneutrophilic phenotypes (P = .258) (ANOVA, P = .045) (e-Fig 1). NOX4 mRNA in epithelial ALI cultures was significantly increased in neutrophilic asthma samples compared with healthy control samples (P = .009) and nonneutrophilic asthma samples (P = .0061) (ANOVA, P = .004) (Fig 3E). NOX4 mRNA expression, as with DNA damage, was correlated to airflow obstruction (r = –0.54; P = .039) (Fig 3F). In basal epithelial cells, NOX4 mRNA expression was similarly elevated, although not significantly, in the neutrophilic subtype compared with the health and the nonneutrophilic subtype (data not shown). At the protein level, basal epithelial cells expressed NOX4 by immunofluorescence (Fig 3G) but not NOX1 (data not shown).

asal epithelial cells, NOX4 mRNA expression was similarly elevated, although not significantly, in the neutrophilic subtype compared with the health and the nonneutrophilic subtype (data not shown). At the protein level, basal epithelial cells expressed NOX4 by immunofluorescence (Fig 3G) but not NOX1 (data not shown). Baseline ROS production did not differ between health and nonneutrophilic or neutrophilic asthma. However, ROS-induced ROS production in response to H202 stimulation increased in neutrophilic asthma compared with health and nonneutrophilic asthma as the concentration of H202 increased. Indeed, the maximal H2O2-induced intracellular ROS generation in basal epithelial cells was significantly increased in patients with neutrophilic asthma (mean ± SEM optical density, OD arbitrary units) (3,380 ± 270) compared with health (2,371 ± 252; P = .02) but not significantly increased compared with patients with nonneutrophilic asthma (2,542 ± 398; P = .11) (ANOVA, P = .037) (Figs 4A, 4B). The response to this maximum dose demonstrated a trend to a correlation in the percentage of sputum neutrophils (r = 0.4; P = .052). GKT137831 at 20 μM (Genkyotex) significantly reduced H2O2-induced (10 mM) intracellular ROS generation, in health from 1,849 ± 281 to 826 ± 194 (mean difference [95% CI], 1022 [–1407 to –638]; P = .001), and in asthma from 2,356 ± 246 to 1,247 ± 134 (–1,130 [–1,429 to –830]; P = .0002) (Fig 4C). GKT137831 similarly attenuated 1 mM H2O2-induced ROS generation. In patients with asthma, this reduction by GKT137831 (–1,362 ± 105) was greater than with N-acetylcysteine (5 mM) (–1,044 ± 90) (mean difference [95% CI], –318 [–626 to –10]; P = .04) (Fig 4D). The reduction in H2O2-induced (10 mM or 1 mM) ROS generation in response to GKT137831 or N-acetylcysteine was not significantly different between health and asthma.

n by GKT137831 (–1,362 ± 105) was greater than with N-acetylcysteine (5 mM) (–1,044 ± 90) (mean difference [95% CI], –318 [–626 to –10]; P = .04) (Fig 4D). The reduction in H2O2-induced (10 mM or 1 mM) ROS generation in response to GKT137831 or N-acetylcysteine was not significantly different between health and asthma. CBF and pattern were assessed in primary ALI cultured epithelial cells from subjects with and without asthma by using high-speed video microscopy (Videos 1A-1D). In contrast to our ex vivo observations, there was no significant difference in the mean ± SEM CBF between health and disease according to the scraping method (difference between means, 0.12 ± 1.4 Hz; P = .93) (Fig 5A) or the overhead viewing method (difference between means, 0.80 ± 1.40; P = .57 [five healthy control subjects and 20 patients with asthma]; data not shown). There were no relationships with asthma severity (data not shown) or sputum neutrophil count (rs = –0.32; P = .34). Beat patterns were similar in the healthy control and asthmatic groups, including percentage of normal cilia (50 ± 7% vs 40% ± 5%; P = .28) (Fig 5B), dyskinetic cilia (41% ± 7% vs 53% ± 4%; P = .15) (Fig 5C), and static cilia (9% ± 3% vs 7% ± 2%; P = .58) (Fig 5D). These findings, together with similar levels of ciliogenesis in culture between asthma and health (60% ± 6% vs 63% ± 4%; P = .72) and surface morphology index (1.6 ± 0.14 vs 1.7 ± 0.12; P = .54), suggest the bronchial epithelial cells from patients with asthma do not have a primary deficiency in differentiation.

g 5D). These findings, together with similar levels of ciliogenesis in culture between asthma and health (60% ± 6% vs 63% ± 4%; P = .72) and surface morphology index (1.6 ± 0.14 vs 1.7 ± 0.12; P = .54), suggest the bronchial epithelial cells from patients with asthma do not have a primary deficiency in differentiation. To further establish the role of NADPH oxidase in ciliary dysfunction in asthma, we investigated the effect of GKT137831 on ciliary function of fresh bronchial epithelial strips studied within 1.5 h of bronchoscopy. These experiments were performed without any additional stimuli in contrast to the in vitro experiments described earlier. Using these fresh samples from patients with asthma, GKT137831 (5 μM and 20 μM) significantly increased CBF from baseline over the 1 h of incubation. GKT137831 20 μM at 1 h resulted in elevated CBF (10.9 ± 0.6 Hz), significantly different from baseline (6.4 ± 0.5 Hz) (mean difference [95% CI], 4.5 [3.5-5.5]; P < .0001) and from the diluent controls (8.9 ± 0.9 Hz, 2.0 [0.6-3.4]; P = .01) (Fig 6A). With GKT137831 20 μM, the percentage of normal cilia increased from 32% ± 5% to 48% ± 6% (mean difference [95% CI], 16% [10-22]; P < .0001) at 0.5 h and from 28% ± 5% to 45% ± 5% (mean difference [95% CI], 17% [6-27]; P = .01) at 1 h vs diluent (Fig 6B). The percentage of dyskinetic cilia was reduced from 57% ± 4% to 50 ± 5% (mean difference [95% CI], –7% [–14 to 0.3]; P = .04) at 0.5 h (Fig 6C) vs diluent. This improvement in CBF in response to NADPH oxidase inhibition was in fact due to cilia motility. NADPH oxidase inhibition lowered the proportion of static cilia compared with diluent, from 11% ± 4% to 2% ± 1% at 0.5 h (mean difference [95% CI], –9% [–16 to –2]; P = .01) and from 17% ± 7% to 2% ± 1% at 1 h (mean difference [95% CI], –15% [–29 to –0.6]; P = .04) (Fig 6D). Sputum neutrophil count was correlated to the GKT137831-induced reduction in the percentage of static cilia (r = 0.61; P = .03) (Fig 6E). It was also correlated to the percentage of normal cilia (r = 0.55; P = 0.05 [data not shown]). These improvements were significantly different in patients with neutrophilic vs nonneutrophilic asthma (Fig 6F).

correlated to the GKT137831-induced reduction in the percentage of static cilia (r = 0.61; P = .03) (Fig 6E). It was also correlated to the percentage of normal cilia (r = 0.55; P = 0.05 [data not shown]). These improvements were significantly different in patients with neutrophilic vs nonneutrophilic asthma (Fig 6F). In a murine model of OVA sensitization and challenge with the use of a standard protocol (Fig 7A), there was evidence of inflammation in the bronchoalveolar lavage, which was predominantly neutrophilic (Fig 7B) as well as ciliary dysfunction (Figs 7C, 7D). Administration of GKT137831 following OVA sensitization resulted in no significant change in bronchoalveolar lavage total inflammatory cell count (OVA sensitization alone vs GKT137831 alone, 303 [243-660] cells vs 240 [14-413] cells; P = .33) or cell differential. Following sensitization and challenge with OVA, the CBF was markedly impaired (P < .001 compared with baseline), and the addition of GKT137831 improved this finding almost back to baseline (P = .003). Similarly, the percentage of ciliated cells was reduced following sensitization and challenge with OVA compared with baseline (P = .007) and was restored following the addition of GKT137831 (P = .024).

.001 compared with baseline), and the addition of GKT137831 improved this finding almost back to baseline (P = .003). Similarly, the percentage of ciliated cells was reduced following sensitization and challenge with OVA compared with baseline (P = .007) and was restored following the addition of GKT137831 (P = .024). Discussion To our knowledge, we report here for the first time that bronchial epithelial ciliary dysfunction in asthma is related to neutrophilic inflammation. This ciliary dysfunction, evident in vivo, does not persist in vitro, suggesting that the asthmatic airway environment is essential in the development and maintenance of ciliary dysfunction. In contrast to DUOX1, which was increased in asthma independent of inflammatory phenotype, and DUOX2, which was only increased in nonneutrophilic asthma, NOX4 expression was upregulated only in the bronchial epithelium from patients with asthma with neutrophilic inflammation in vivo. This increased NOX4 expression was also present in vitro with NADPH oxidase-dependent increased ROS generation in primary epithelial cells from patients with asthma with neutrophilic inflammation. Critically, NADPH oxidase inhibition completely abrogated ciliary dysfunction in ex vivo epithelial strips, as evidenced by restoration of CBF to the frequency we previously described in epithelial strips from healthy subjects2 and in an in vivo murine model of asthma with evidence of neutrophilic lung inflammation, but it did not significantly reduce this inflammation. Taken together, these findings support the view that ciliary dysfunction in asthma is both NADPH oxidase-dependent (in particular NOX4) and inflammation-dependent, with both necessary but neither sufficient, as persistent ciliary dysfunction was not observed in vitro despite increased NOX4 expression, and NADPH oxidase inhibition abolished ciliary dysfunction without affecting airway inflammation.

ma is both NADPH oxidase-dependent (in particular NOX4) and inflammation-dependent, with both necessary but neither sufficient, as persistent ciliary dysfunction was not observed in vitro despite increased NOX4 expression, and NADPH oxidase inhibition abolished ciliary dysfunction without affecting airway inflammation. Noneosinophilic, and specifically neutrophilic, asthma is common9 but responds poorly to corticosteroid therapy.7 Neither the mechanisms nor the clinical management of neutrophilic asthma are fully understood. Oxidative stress and neutrophilia are often associated,14 and, consistent with this view, we found that neutrophilic asthma and airflow obstruction were associated with oxidative DNA damage, as evidenced by increased bronchial epithelial 8-oxo-dG expression assessed semiquantitatively. Critically, we extended our earlier observation that ciliary dysfunction is impaired in asthma and associated with increasing severity2 to demonstrate that this ciliary dysfunction is a particular feature of neutrophilic asthma. Normal ciliary function is a key component of innate immunity, and its impairment is likely to promote persistence of inflammation and increase risk of airway infection and colonization by pathogens.3, 4, 10 Indeed, macrolide antibiotics have demonstrated efficacy in noneosinophilic asthma,28 but concerns related to antibiotic resistance as a consequence of long-term widespread antibiotic use highlight the need for new therapies to treat neutrophilic asthma. The biologic relevance of the magnitude of CBF changes we observed in fresh bronchial strips following GKT137831 treatment must be addressed in future studies. However, following exposure to GKT137831, the CBF in the patients with asthma improved to a level we have reported previously in health2 and the improvement above diluent (2 Hz) was similar to the differences observed between health and COPD.29

l strips following GKT137831 treatment must be addressed in future studies. However, following exposure to GKT137831, the CBF in the patients with asthma improved to a level we have reported previously in health2 and the improvement above diluent (2 Hz) was similar to the differences observed between health and COPD.29 NADPH oxidases play a critical role in ROS generation. NOX4 and DUOX1/2 expression in the epithelium was evident in asthma. DUOX1 was increased in asthma independent of inflammatory phenotype, and DUOX2 was only increased in nonneutrophilic asthma. In contrast, NOX4 was only increased in neutrophilic asthma. NOX4 plays a crucial role in regulating altered oxidative pathways in idiopathic pulmonary fibrosis30, 31, 32 and several nonrespiratory diseases,26, 33, 34 and its expression is upregulated by oxidative stress itself and inflammation. We found that increased ROS in the bronchial epithelium was a feature of neutrophilic asthma that persisted in vitro in association with increased NOX4 expression. Although NOX4 might be upregulated by neutrophilic inflammation, its overexpression can occur independently of the asthmatic environment. Importantly, ciliary dysfunction did not persist in vitro, suggesting this NOX4 upregulation was not itself sufficient to drive ciliary dysfunction but also required the presence of airway inflammation.

be upregulated by neutrophilic inflammation, its overexpression can occur independently of the asthmatic environment. Importantly, ciliary dysfunction did not persist in vitro, suggesting this NOX4 upregulation was not itself sufficient to drive ciliary dysfunction but also required the presence of airway inflammation. Beyond asthma, bronchial epithelial ciliary dysfunction is a fundamental abnormality in bronchiectasis,35 as well as in COPD.36 Antioxidant therapy, such as N-acetylcysteine, has been considered in these conditions; however, studies are inconsistent with benefits demonstrated in some studies37 but not in others.38, 39 The magnitude of benefit is also relatively small, possibly due to poor tolerance of high concentrations of the therapeutic agent, questionable bioavailability, and use of nonspecific antioxidant therapies that enhance the clearance of ROS rather than targeting ROS generation. Our findings suggest that NAPDH oxidase inhibition, which reduces intracellular ROS generation, improves ciliary dysfunction and thus extends the potential role of NADPH oxidases (and particularly NOX4) in asthma beyond the described impact of upregulated NOX4 on airway smooth muscle hypercontractility.20

S generation. Our findings suggest that NAPDH oxidase inhibition, which reduces intracellular ROS generation, improves ciliary dysfunction and thus extends the potential role of NADPH oxidases (and particularly NOX4) in asthma beyond the described impact of upregulated NOX4 on airway smooth muscle hypercontractility.20 A major limitation of our study is that although GKT137831 has good specificity and potency for NOX4, it also has effects on NOX1 and is likely to exert effects on DUOX, as has been observed with similar inhibitors in this class (eg, GKT136901).40 Our findings support a key role for NOX4; the expression of this NOX family member was increased in neutrophilic asthma and associated with ciliary dysfunction. However, we cannot fully exclude possible contributions from other mechanisms, including other members of the NOX family (particularly DUOX1/2). In addition, due to the considerable homology in the amino acid sequence of DUOX1 and DUOX2, the DUOX2 antibody may cross-react with DUOX1. However the amount of staining seen with the DUOX2 antibody is considerably less than that with the DUOX1 antibody. Thus, we are confident that DUOX2 is not highly expressed in the bronchial epithelium. One further potential limitation of our study is the lack of direct in vivo human data demonstrating that NADPH oxidase inhibition modifies ciliary function and inflammation. However, we have confirmed the importance of NADPH oxidase in ciliary dysfunction in fresh ex vivo primary cells studied immediately after obtaining the cells at bronchoscopy, and we supported the observations by using an in vivo murine model of asthma with neutrophilic lung inflammation. This asthma model clearly showed that NADPH oxidase inhibition markedly improved ciliary function but did not significantly attenuate lung inflammation. These findings suggest that although the presence of inflammation might be important for NOX4 activation, it is the consequent ROS generation rather than the inflammation per se that is critical in ciliary dysfunction.

inhibition markedly improved ciliary function but did not significantly attenuate lung inflammation. These findings suggest that although the presence of inflammation might be important for NOX4 activation, it is the consequent ROS generation rather than the inflammation per se that is critical in ciliary dysfunction. Conclusions Bronchial epithelial cells from patients with asthma with neutrophilic airway inflammation exhibit heightened NOX4 expression and ROS generation. NADPH oxidase inhibition in an in vivo murine model of asthma improved ciliary function, which was also confirmed in ex vivo bronchial epithelial cells from patients with neutrophilic asthma. These findings support the view that NADPH oxidase inhibition (and in particular NOX4) might present a new stratified approach for the treatment of neutrophilic asthma. Supplementary Data e-Online Data Video 1A Video 1B Video 1C Video 1D Acknowledgments Author contributions: C. E. B. is the guarantor for the manuscript. C. S., A. W., C. O., P. W. A., and C. E. B. conceived the study. W-Y. H. W., F. H., L. H., L. W., and R. A. H. collected the data. W-Y. H. W., F. H., L. H., L. W., R. A. H., E. G., V. M., and A. S. performed the assays. W-Y. H. W., F. H., L. H., L. W., R. A. H., S. B., R. S., D. D., L. C., and C. E. B. analyzed the data. W-Y. H. W., F. H., R. S., and C. E. B. wrote the manuscript. All authors provided critical review of the manuscript and approved of its contents and submission.

R. A. H., E. G., V. M., and A. S. performed the assays. W-Y. H. W., F. H., L. H., L. W., R. A. H., S. B., R. S., D. D., L. C., and C. E. B. analyzed the data. W-Y. H. W., F. H., R. S., and C. E. B. wrote the manuscript. All authors provided critical review of the manuscript and approved of its contents and submission. Financial/nonfinancial disclosures: The authors have reported to CHEST the following: D. D. has received speakers fees from AstraZeneca, Boehringer Ingelheim, and Chiesi Pharmaceuticals. C. S. is a paid employee and owns shares of Genkyotex SA. A. W. serves on advisory boards for GlaxoSmithKline and receives research support from GlaxoSmithKline, Pfizer, and AstraZeneca. C. E. B. serves on advisory boards for GlaxoSmithKline, AstraZeneca, MedImmune, Roche, and Aerovance; receives honoraria from Novartis; and receives research support from GlaxoSmithKline, AstraZeneca, and MedImmune. None declared (W.-Y.H.W., F.H., L.H., L.W., R.A.H., S.B., E.G., A.S., L.C., V.M., A.W., R.S., C.O., P.W.A.). Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Other contributions: The authors thank Genkyotex for donating compound GKT137831. The authors also thank all the clinical staff for helping with collecting samples and patient details, Myriam Bahri, BSc, for her help with immunohistochemistry, and the patients for their participation in this study.

Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Other contributions: The authors thank Genkyotex for donating compound GKT137831. The authors also thank all the clinical staff for helping with collecting samples and patient details, Myriam Bahri, BSc, for her help with immunohistochemistry, and the patients for their participation in this study. Additional information: The e-Appendix, e-Figure, and Videos can be found in the Supplemental Materials and Multimedia sections of the online article. Drs Wan and Hollins were joint first authors on this manuscript. FUNDING/SUPPORT: The following funding bodies were involved in this study: Wellcome Trust Senior Clinical Fellowship (C. E. B.), Airway Disease Predicting Outcomes through Patient Specific Computational Modelling (AirPROM) project (funded through FP7 EU grant), MAARA, NC3R, and the European Regional Development Fund (ERDF 05567) part funded the research laboratories. This article presents independent research funded by the Leicester Respiratory National Institute for Health Research Biomedical Research Unit. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.

und (ERDF 05567) part funded the research laboratories. This article presents independent research funded by the Leicester Respiratory National Institute for Health Research Biomedical Research Unit. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health. Figure 1 A-C, Baseline ciliary function of ciliated fresh ex vivo bronchial epithelial strips. (A) CBF was measured in fresh ex vivo epithelial strips from 10 patients with nonneutrophilic asthma and 11 patients with neutrophilic asthma. Comparisons were made by using unpaired t tests. (B) Correlation between CBF of the fresh ex vivo epithelial strips and sputum neutrophil count (n = 21). Patients with asthma with or without sputum neutrophilia (> 61%) are represented as triangles or squares, respectively. (C) The CBF of the fresh ex vivo epithelial strips and sputum eosinophil count is not correlated (n = 21). CBF = ciliary beat frequency.

ex vivo epithelial strips and sputum neutrophil count (n = 21). Patients with asthma with or without sputum neutrophilia (> 61%) are represented as triangles or squares, respectively. (C) The CBF of the fresh ex vivo epithelial strips and sputum eosinophil count is not correlated (n = 21). CBF = ciliary beat frequency. Figure 2 A-D, Oxidative damage (8-oxo-dG levels) in human bronchial epithelium in vivo. (A) Representative photomicrographs showing 8-oxo-dG staining in the epithelium in bronchial biopsy specimens from isotype control subjects (i); from a patient with asthma with positive staining of medium intensity (ii) and high intensity (iii) (×400). (B) Expression of 8-oxo-dG using a semiquantitative score (0 = none, 1 = low, 2 = moderate, 3 = high expression) in healthy control subjects (circles) and in patients with asthma with or without sputum neutrophilia (> 61%) (triangles or squares, respectively). Horizontal bars represent medians. Kruskal-Wallis test, P = .005. (C) and (D) Correlation of the expression of 8-oxo-dG with airflow obstruction and sputum neutrophil count, respectively (Spearman correlations are as shown).

s with asthma with or without sputum neutrophilia (> 61%) (triangles or squares, respectively). Horizontal bars represent medians. Kruskal-Wallis test, P = .005. (C) and (D) Correlation of the expression of 8-oxo-dG with airflow obstruction and sputum neutrophil count, respectively (Spearman correlations are as shown). Figure 3 A-G, Bronchial epithelial nicotinamide adenine dinucleotide phosphate (NADPH) oxidase expression in asthma in vivo and in vitro. (A) Representative immunohistochemistry photomicrographs of (i) bronchial biopsy specimen stained with an isotype control and epithelial NOX4 protein expression in bronchial biopsy tissue from patients with (ii) nonneutrophilic and (iii) neutrophilic asthma. (B) Representative immunohistochemistry photomicrographs of (i) bronchial biopsy specimens stained with an isotype control and epithelial DUOX1 protein expression in bronchial biopsy tissue from patients with (ii) nonneutrophilic and (iii) neutrophilic asthma. Intensity of staining of (C) NOX4 and (D) DUOX1 was measured by using thresholding and expressed as percent area of positive staining. Bar represents mean (SEM). (E) NOX4 gene expression in air-liquid interface cultures, quantified by real-time quantitative polymerase chain reaction, in healthy control subjects (n = 7; circles), patients with nonneutrophilic asthma (n = 4; squares), and patients with neutrophilic asthma (n = 4; triangles). Analysis of variance and post hoc Tukey tests as shown. (F) Correlation between NOX4 gene expression in air-liquid interface cultures and airflow obstruction (Spearman correlation as shown). (G) Immunofluorescence staining of cytospun basal epithelial cells for NOX4. NOX4 is stained red, nuclei is stained blue, and the isotype control is shown as inset; ×20 magnification, scale bar represents 50 μm. mRNA = messenger RNA.

air-liquid interface cultures and airflow obstruction (Spearman correlation as shown). (G) Immunofluorescence staining of cytospun basal epithelial cells for NOX4. NOX4 is stained red, nuclei is stained blue, and the isotype control is shown as inset; ×20 magnification, scale bar represents 50 μm. mRNA = messenger RNA. Figure 4 A-D, The role of NADPH oxidase in oxidant-induced reactive oxygen species (ROS) generation. (A) Generation of dose-dependent intracellular ROS, induced by H2O2, in human bronchial epithelial cells from healthy control subjects (n = 8; circles), patients with nonneutrophilic asthma (n = 4; squares), and patients with neutrophilic asthma (n = 9; triangles) quantified by using the 5-(and-6)-carboxy-2′, 7′-dichlorofluorescein diacetate assay. (B) Corresponding maximum dose response for individual subjects. (C) The effect of GKT137831 on H2O2-induced intracellular ROS generation in human bronchial epithelial cells from healthy control subjects (n = 3; open bars) and patients with asthma (n = 6; closed bars). *Comparison with corresponding GKT137831-only groups; # comparison with corresponding H2O2-only groups, P < .05; ## P < .01; ∗∗∗, ### P < .001. (D) Reduction of H2O2 induced-intracellular ROS generation in response to GKT137831 (20 μM) and NAC (5 mM). Unpaired t test, P < .05. EC50 = half maximal effective concentration; NAC = N-acetylcysteine; OD = optical density; ROS = reactive oxygen species.

onding H2O2-only groups, P < .05; ## P < .01; ∗∗∗, ### P < .001. (D) Reduction of H2O2 induced-intracellular ROS generation in response to GKT137831 (20 μM) and NAC (5 mM). Unpaired t test, P < .05. EC50 = half maximal effective concentration; NAC = N-acetylcysteine; OD = optical density; ROS = reactive oxygen species. Figure 5 A-D, Ciliary function of human primary bronchial epithelial cell cultures. (A) CBF of ciliated epithelial cells from healthy control subjects (circles), patients with nonneutrophilic asthma (squares), and patients with neutrophilic asthma (triangles). Ciliary beat patterns were expressed in the percentage of (B) normally beating cilia, (C) dyskinetic cilia, and (D) static cilia. Unpaired t test, P values as shown. See Figure 1 legend for expansion of abbreviation.

es), patients with nonneutrophilic asthma (squares), and patients with neutrophilic asthma (triangles). Ciliary beat patterns were expressed in the percentage of (B) normally beating cilia, (C) dyskinetic cilia, and (D) static cilia. Unpaired t test, P values as shown. See Figure 1 legend for expansion of abbreviation. Figure 6 A-F, Effect of GKT137831 on ciliary function of fresh ex vivo bronchial epithelial strips. The effect of GKT137831 (20 μM) on the ciliary function over 1 h was assessed by using fresh asthmatic bronchial epithelial strips (n = 13) from bronchoscopy. Six to 10 side profiles were recorded per sample for analysis. Ciliary function was represented by CBF (A), and beat patterns are presented as percentage of (B) normal cilia, (C) dyskinetic cilia, and (D) static cilia at baseline (open bars), 0.5 h (light grey bars), and 1 h (dark grey bars). *Comparison with baseline; # comparison with corresponding diluent controls: paired t test, P < .05. (E) Correlation between percentage of sputum neutrophils and GKT137831 20 μM-induced absolute change as percentage of static cilia in 1 h. Pearson’s correlation, P < .05. (F) Improvement in GKT137831-induced beat patterns in nonneutrophilic and neutrophilic asthma subtypes. Unpaired t test, P values as shown. ## P < .01; ∗∗∗, ### P < .001. See Figure 1 legend for expansion of abbreviation.

0 μM-induced absolute change as percentage of static cilia in 1 h. Pearson’s correlation, P < .05. (F) Improvement in GKT137831-induced beat patterns in nonneutrophilic and neutrophilic asthma subtypes. Unpaired t test, P values as shown. ## P < .01; ∗∗∗, ### P < .001. See Figure 1 legend for expansion of abbreviation. Figure 7 A-D, Effect of GKT137831 on ciliary function in an in vivo murine model of OVA sensitization and challenge. (A) Dosing regimen: up to 14 animals were used for each of the 4 conditions: (i) saline, (ii) GKT137831 (40 mg/kg) alone, (iii) model with saline control, or (iv) model in the presence of GKT137831 (40 mg/kg). (B) Differential cell counts of immune cells in the BALF of CBA/Ca mice were assessed at the time of culling. Counts are based on the number of immune cells counted in 10 fields of view at ×400 magnification. Lymphocytes were included in the counts but were too infrequent or absent to be included in color in the figure bars. Ciliary function is represented by (C) CBF and (D) percent ciliated cells. Analysis of variance and Kruskal-Wallis test for comparisons across groups with Tukey and Dunn tests for post hoc pairwise comparisons, respectively. BALF = bronchoalveolar lavage fluid; OVA = ovalbumin. See Figure 1 legend for expansion of other abbreviation. Table 1 Clinical Characteristics of Subjects Who Provided Bronchial Epithelial Cells for Ex Vivo Studies or Bronchial Biopsy Sampling

Figure 7 A-D, Effect of GKT137831 on ciliary function in an in vivo murine model of OVA sensitization and challenge. (A) Dosing regimen: up to 14 animals were used for each of the 4 conditions: (i) saline, (ii) GKT137831 (40 mg/kg) alone, (iii) model with saline control, or (iv) model in the presence of GKT137831 (40 mg/kg). (B) Differential cell counts of immune cells in the BALF of CBA/Ca mice were assessed at the time of culling. Counts are based on the number of immune cells counted in 10 fields of view at ×400 magnification. Lymphocytes were included in the counts but were too infrequent or absent to be included in color in the figure bars. Ciliary function is represented by (C) CBF and (D) percent ciliated cells. Analysis of variance and Kruskal-Wallis test for comparisons across groups with Tukey and Dunn tests for post hoc pairwise comparisons, respectively. BALF = bronchoalveolar lavage fluid; OVA = ovalbumin. See Figure 1 legend for expansion of other abbreviation. Table 1 Clinical Characteristics of Subjects Who Provided Bronchial Epithelial Cells for Ex Vivo Studies or Bronchial Biopsy Sampling Characteristic Ex Vivo Epithelial Cell Donors Bronchial Biopsy Sample Donors Healthy Control (n = 8) Neutrophilic (n = 11) Nonneutrophilic (n = 10) Healthy Control (n = 17) Neutrophilic Asthma (n = 20) Nonneutrophilic Asthma (n = 20) Sex (male/female) 5/3 7/4 7/3 12/5 12/8 5/15 Age, y 52 ± 6 58 ± 4 57 ± 3 46 ± 4 55 ± 2 49 ± 3 Current/ex-/never smoker 0/4/4 0/4/7 0/2/10 0/1/16 0/2/18 0/0/20 Pack-yearsa 8 ± 5 3 ± 2 1 ± 1 0 (0-1) 0 (0-4) 0 (0-0) Predicted FEV1, % 96 ± 4 80 ± 7 82 ± 6 96 ± 4 82 ± 5 79 ± 6 FEV1/FVC, % 78 ± 2 59 ± 5 62 ± 5 78 ± 2 67 ± 3 69 ± 3 Blood total IgE, kU/mLa NA 139 (68-438) 172 (122-665) NA 204 (36-1318) 243 (150-653) Blood eosinophils, kU/mLa NA 0.22 (0.14-0.56) 0.28 (0.19-0.40) NA 0.47 (0.17-0.55) 0.23 (0.17-0.58) Treatment (μg/24 BDP)a,b NA 1,000 (800-2,000) 1,600 (750-2,000) NA 1,600 (1,200-2,000) 1,600 (1,600-2,000) GINA treatment step (1-3 or 4/5) NA 4, 7 1, 9 NA 5, 15 8, 12 No. with atopy NA 5 7 NA 15 16 Age of disease onset, y NA 38 ± 7 26 ± 5 NA 22 ± 8 25 ± 6 Sputum characterization Eosinophils, %a NA 3.5 (1.3-22.5) 15.2 (3.8-36.5) NA 2.0 (0.3-5.2) 9.0 (0.4-23.7) Neutrophils, %a NA 80.0 (65.0-87.2) 37.2 (3.8-36.5) NA 76.7 (67.9-85.8) 35.8 (20.0-46.5) Data are presented as mean ± SD unless otherwise indicated. GINA = Global Initiative for Asthma; IgE = immunoglobulin E; NA = not applicable.

ization Eosinophils, %a NA 3.5 (1.3-22.5) 15.2 (3.8-36.5) NA 2.0 (0.3-5.2) 9.0 (0.4-23.7) Neutrophils, %a NA 80.0 (65.0-87.2) 37.2 (3.8-36.5) NA 76.7 (67.9-85.8) 35.8 (20.0-46.5) Data are presented as mean ± SD unless otherwise indicated. GINA = Global Initiative for Asthma; IgE = immunoglobulin E; NA = not applicable. a Median (interquartile range). b Beclomethasone dipropionate (BDP) equivalence.

COPD is the third leading cause of death worldwide and a risk predictor for atherosclerosis.1, 2, 3 Several pathophysiological processes may contribute to disease progression and increased cardiovascular risk in COPD, including systemic effects of smoking, chronic inflammation,4 and endothelial dysfunction.5 Patients with COPD are also more likely to have other cardiovascular comorbidities, including central abdominal obesity and metabolic syndrome, particularly in earlier stages of COPD.6, 7, 8 Endothelium-derived hyperpolarizing factor (EDHF) and, particularly, epoxyeicosatrienoic acid (EET) are involved in the modulation of vascular tone,9 attenuation of inflammation,10 and activation of fibrinolysis by augmenting tissue plasminogen activator (tPA) expression.11 EETs are synthesized by cytochrome P450 (CYP) enzymes, and metabolized to their less biologically active diols by soluble epoxide hydrolase (sEH) enzymes.12 Smoking has a synergistic effect with CYP450 and sEH polymorphisms,13 resulting in enhanced sEH activity, reduced plasma EETs, and increasing overall risk of myocardial infarction.14 Plasma EET levels are reduced in patients with coronary artery disease who are obese or who smoke.15 EETs are also produced in lung epithelial cells, and they may become dysfunctional in COPD.12 In vivo, smokers exhibit reduced endothelial responses to bradykinin,5 and this may be associated with impaired EDHF-mediated vasodilation.16, 17 However, the functional role of EETs has not yet been characterized in humans.

5 EETs are also produced in lung epithelial cells, and they may become dysfunctional in COPD.12 In vivo, smokers exhibit reduced endothelial responses to bradykinin,5 and this may be associated with impaired EDHF-mediated vasodilation.16, 17 However, the functional role of EETs has not yet been characterized in humans. Upregulation of EETs by sEH inhibition in animals improves metabolic syndrome18 and lung function and attenuates smoking-related inflammation and emphysema.19 GSK2256294 is a novel potent sEH inhibitor in phase I clinical development and may have the potential to impact systemic and pulmonary endothelial function. As this was a phase I clinical trial mainly focused on safety and tolerability in healthy people, we used a cohort of overweight smokers as representative of patients with early-stage COPD.

nt sEH inhibitor in phase I clinical development and may have the potential to impact systemic and pulmonary endothelial function. As this was a phase I clinical trial mainly focused on safety and tolerability in healthy people, we used a cohort of overweight smokers as representative of patients with early-stage COPD. We hypothesized that EET synthesis is reduced in patients with COPD and otherwise healthy overweight smokers and that sEH inhibition would upregulate EETs and endothelial dysfunction. We completed a physiological study in which we assessed EET-mediated basal tone, and the EET component of bradykinin stimulated vasodilation in patients with COPD and in overweight smokers to maximize the impact of cardiovascular risk factors in otherwise healthy subjects. Subsequently, we examined the effects of a novel sEH inhibitor, GSK2256294, in human resistance arteries in vitro and in vivo in a phase I clinical trial with an experimental medicine arm to provide early proof of mechanism for target engagement in overweight smokers. The study design, safety, and pharmacokinetic data from the phase I trial were reported separately,20 and we only report the effects of sEH inhibition on endothelial function in this manuscript.

ical trial with an experimental medicine arm to provide early proof of mechanism for target engagement in overweight smokers. The study design, safety, and pharmacokinetic data from the phase I trial were reported separately,20 and we only report the effects of sEH inhibition on endothelial function in this manuscript. Methods All study procedures were conducted in accordance with the Declaration of Helsinki, were approved by appropriate institutional review boards, and received favorable opinions from local ethics committees (13/EE/0032, 12/LO/1832), and the Medicines and Healthcare products Regulatory Agency. Analysis and statistical methods are described in e-Appendix 1. All subjects were recruited following written consent. We used forearm venous occlusion plethysmography21 to assess vascular function in vivo with intraarterial infusion of challenge agents through a 27-gauge needle (Coopers Needleworks) inserted into the brachial artery. Venous plasma concentrations of EET/DHET were assessed as representative of sEH activity at baseline and during the forearm blood flow studies. Oscillometric BP was monitored in the noninfused arm. Detailed methods and statistical analyses can be seen in e-Appendix 1.

ers Needleworks) inserted into the brachial artery. Venous plasma concentrations of EET/DHET were assessed as representative of sEH activity at baseline and during the forearm blood flow studies. Oscillometric BP was monitored in the noninfused arm. Detailed methods and statistical analyses can be seen in e-Appendix 1. Study 1 Twelve male patients with COPD (FEV1/FVC < 0.7 and FEV1 < 80% postbronchodilator use), and 12 healthy sex-matched control groups (matched control group 1) underwent a single forearm blood flow study to assess EET-mediated vasodilation (UK Clinical Research Network Portfolio ID: 14339). Patients taking concomitant medications that interfere with CYP450 or cyclooxygenase enzymes were asked to stop for at least 4 days prior to the forearm blood flow procedure. Overall endothelium-dependent function was assessed by infusing bradykinin (100, 300, and 1,000 pmol/min; Bachem Distribution Services GmbH), and stimulated EET release was assessed by coinfusing bradykinin with 0.4 μmol/min fluconazole, a CYP inhibitor that inhibits EET synthesis (Pfizer Ltd.) (e-Fig 1).9 Endothelium-independent responses were assessed using 12 and 38 nmol/min (3 and 10 μg/min) sodium nitroprusside (SNP) (Nitroprussiat FIDES). Study 2 Twelve overweight smokers (≥ 10 cigarettes/d and > 5 pack-year history, weight > 60 kg, and BMI 28-35 kg/m2) and equal numbers of healthy sex- and age-matched nonsmoker control groups (matched control group 2) underwent the same forearm blood flow protocol as did subjects in study 1.

Study 1 Twelve male patients with COPD (FEV1/FVC < 0.7 and FEV1 < 80% postbronchodilator use), and 12 healthy sex-matched control groups (matched control group 1) underwent a single forearm blood flow study to assess EET-mediated vasodilation (UK Clinical Research Network Portfolio ID: 14339). Patients taking concomitant medications that interfere with CYP450 or cyclooxygenase enzymes were asked to stop for at least 4 days prior to the forearm blood flow procedure. Overall endothelium-dependent function was assessed by infusing bradykinin (100, 300, and 1,000 pmol/min; Bachem Distribution Services GmbH), and stimulated EET release was assessed by coinfusing bradykinin with 0.4 μmol/min fluconazole, a CYP inhibitor that inhibits EET synthesis (Pfizer Ltd.) (e-Fig 1).9 Endothelium-independent responses were assessed using 12 and 38 nmol/min (3 and 10 μg/min) sodium nitroprusside (SNP) (Nitroprussiat FIDES). Study 2 Twelve overweight smokers (≥ 10 cigarettes/d and > 5 pack-year history, weight > 60 kg, and BMI 28-35 kg/m2) and equal numbers of healthy sex- and age-matched nonsmoker control groups (matched control group 2) underwent the same forearm blood flow protocol as did subjects in study 1. Study 3 We first assessed the effects of sEH inhibition in vitro by application of GSK2256294 to human resistance arteries treated with L-nitroarginine methyl ester (LNAME) and indomethacin (detailed description of methods in e-Appendix 1) and in vivo using forearm blood flow before a dose, after a single dose (acute effects), and after 14 days (chronic effects) of oral GSK2256294. Responses to bradykinin (300, 600, and 1,000 pmol/min) were assessed in the presence of 8 μmol/min NG-monomethyl-L-arginine (LNMMA; Bachem) and 6 mmol (1 g) IV aspirin (Aspergic Sanofi-Aventis) to inhibit nitric oxide (NO) and prostaglandin I2 synthesis to maximize EDHF and EET (e-Fig 2). Venous concentrations of tPA and plasminogen activator inhibitor type 1 (PAI-1) were measured before and after each dose of bradykinin.22 Challenge agent doses were chosen based on previous studies.5

Aspergic Sanofi-Aventis) to inhibit nitric oxide (NO) and prostaglandin I2 synthesis to maximize EDHF and EET (e-Fig 2). Venous concentrations of tPA and plasminogen activator inhibitor type 1 (PAI-1) were measured before and after each dose of bradykinin.22 Challenge agent doses were chosen based on previous studies.5 To assess the effects of GSK2256294 in vivo, we studied healthy overweight smokers (no concomitant medications) as a paradigm for a COPD population in a phase I clinical trial to provide early proof of mechanism (ClinicalTrials.gov NCT01762774). Thirty male overweight smokers, were allocated in a 2:1 ratio between GSK2256294 (6 mg or 18 mg) and placebo for 14 days of repeated doses. GSK2256294 doses were chosen based on enzyme inhibition and pharmacokinetic data from the single-dosing cohorts.20 Results Study 1 Subject demographics are presented in Table 1. The average FEV1 was 53% ± 13% predicted and the FEV1/FVC ratio was 0.5 ± 0.1 in the subjects with COPD. There was a trend toward a higher plasma concentration of the basal EET/DHET ratio in the matched control group 1 compared with patients with COPD (0.54 ± 0.12 vs 0.45 ± 0.14; P = .08) (Fig 1).

presented in Table 1. The average FEV1 was 53% ± 13% predicted and the FEV1/FVC ratio was 0.5 ± 0.1 in the subjects with COPD. There was a trend toward a higher plasma concentration of the basal EET/DHET ratio in the matched control group 1 compared with patients with COPD (0.54 ± 0.12 vs 0.45 ± 0.14; P = .08) (Fig 1). There was a dose-dependent increase in the forearm blood flow ratio following bradykinin in both groups (P < .0001). Bradykinin response was significantly higher in the matched control group 1 than in patients with COPD (maximal dilatation 1,314% ± 191% vs 552% ± 103%; P = .005) (Fig 2A). In the presence of fluconazole, maximum dilatation to bradykinin was reduced in matched control group 1 (406% ± 64%; P < .0001) but not in patients with COPD (447% ± 124%; P = .32), showing a significant between-group difference in inhibition (P = .03). There was no difference in SNP response between groups (data not shown). BP values remained constant throughout the studies. Although not significant, plasma concentrations of the EET/DHET ratio in response to bradykinin was higher in the matched control group 1 compared with patients with COPD (maximum 8.6 ± 3.4 vs 6.8 ± 1.1; P = .83). Although it was not significant, in the presence of fluconazole, total EET/DHET levels were slightly lower in matched control group 1 (maximum 4.7 ± 0.4; P = .27) but not in patients with COPD (5.2 ± 0.9; P = .70) (Fig 3A).

ched control group 1 compared with patients with COPD (maximum 8.6 ± 3.4 vs 6.8 ± 1.1; P = .83). Although it was not significant, in the presence of fluconazole, total EET/DHET levels were slightly lower in matched control group 1 (maximum 4.7 ± 0.4; P = .27) but not in patients with COPD (5.2 ± 0.9; P = .70) (Fig 3A). Study 2 Although not significant, the basal EET/DHET ratio was higher in the matched control group 2 compared with overweight smokers (0.46 ± 0.06 vs 0.39 ± 0.04; P = .33) (Fig 1). Bradykinin response was higher in the matched control group 2 than in overweight smokers (maximal dilatation: 930% ± 81% vs 575% ± 112%; P = .02) (Fig 2B). In the presence of fluconazole, maximum dilatation to bradykinin was reduced in the matched control group 2 (400% ± 49%; P < .0001) but not in overweight smokers (437% ± 57%; P = .16), resulting in a significant between-group difference (P = .002). There was no difference in SNP response between groups (data not shown). BP values remained constant throughout the studies. There was no difference in the bradykinin response between subjects with COPD and overweight smokers (P = .72). Although not significant, the increase in the EET/DHET ratio in response to bradykinin was higher in the healthy matched control group 2 compared with overweight smokers (maximum 10.31 ± 4.43 vs 5.66 ± 0.46; P = .80). In the presence of fluconazole, EET/DHET was reduced in the matched control group 2 but were slightly increased in overweight smokers (maximum, 5.02 ± 0.38 vs 8.19 ± 2.18; P = .003) (Fig 3B).

s higher in the healthy matched control group 2 compared with overweight smokers (maximum 10.31 ± 4.43 vs 5.66 ± 0.46; P = .80). In the presence of fluconazole, EET/DHET was reduced in the matched control group 2 but were slightly increased in overweight smokers (maximum, 5.02 ± 0.38 vs 8.19 ± 2.18; P = .003) (Fig 3B). Study 3 In LNAME- and indomethacin-treated resistance vessels, GSK2256294 10 μM increased 11,12-EET-mediated vasodilation compared with vehicle (n = 6 in each group; 90% ± 4% vs 73% ± 6% maximal dilatation) (Fig 4A) and shifted the bradykinin EC50 (n = 6; –8.33 ± 0.17 logM vs –8.10 ± 0.12 logM; P = .001) (Fig 4B). However, vasodilation from 8,9-EET was unaltered (maximal dilatation, 82% ± 16% vs 72% ± 19%), suggesting that the effects were regioisomer specific. The vasodilation from papaverine (100 μM), a test of direct smooth muscle vasodilation, was unchanged with GSK2256294 administration. In vivo, 28 subjects, including the 11 who took part in the physiological study, completed forearm blood flow studies before dosing, after a single dose, and after 14 days of repeated dosing with placebo (n = 6) or GSK2256294, 6 mg or 18 mg (n = 11 in each group) (Table 1). There was a trend toward increased bradykinin response after single and repeated dosing in the active treatment groups. In subjects who received 6 mg, response to bradykinin increased by 23% ± 17% on day 1 and by 22% ± 22% on day 14. In those who received 18 mg, bradykinin response increased by 14% ± 17% on day 1 and 12% ± 14% on day 14. Responses to SNP did not change.

se after single and repeated dosing in the active treatment groups. In subjects who received 6 mg, response to bradykinin increased by 23% ± 17% on day 1 and by 22% ± 22% on day 14. In those who received 18 mg, bradykinin response increased by 14% ± 17% on day 1 and 12% ± 14% on day 14. Responses to SNP did not change. In a post hoc analysis of the forearm blood flow ratio, there was an improvement in bradykinin-induced responses following dosing with the active drug compared with placebo (P = .007), with the greatest effect in the active-drug 18-mg group. In this group, the maximum bradykinin response improved from 338% ± 46% before dosing to 566% ± 110% after a single dose (P = .02) and to 503% ± 123% after chronic dosing (P = .003) (Fig 5). LNMMA and aspirin inhibited basal flow equally on all 3 days in the three treatment arms (e-Table 1). BP remained stable, and there were no changes to tPA in response to BK or in PAI-1 release (data not shown). Discussion The findings from these studies suggest that COPD and smoking are associated with impaired overall endothelial function and reduced stimulated vascular EET production. Proof-of-mechanism data demonstrate that sEH inhibition with GSK2256294 results in improvements in vascular function both in vitro and in vivo.

Discussion The findings from these studies suggest that COPD and smoking are associated with impaired overall endothelial function and reduced stimulated vascular EET production. Proof-of-mechanism data demonstrate that sEH inhibition with GSK2256294 results in improvements in vascular function both in vitro and in vivo. We elected to study patients with COPD and overweight smokers, as the mechanisms behind COPD, smoking, and cardiovascular disease remain poorly understood. Both smokers and patients with COPD exhibit low-grade systemic inflammation,1 which plays a key role in endothelial activation, resulting in endothelial dysfunction and the initiation of atherosclerosis.23 It has been demonstrated that patients with COPD,5 smokers,24 and ex-smokers25 exhibit a similar degree of endothelial dysfunction, suggesting that smoking may be the key contributing factor. Cardiovascular risk factors are more likely to cluster in obesity, manifesting as a syndrome of increased adipocytes, hyperglycemia, and dyslipidemia, with underlying low-grade inflammation. In normotensive overweight subjects with metabolic syndrome, acetylcholine-induced rather than bradykinin-induced vasodilation is reduced, possibly suggesting a lesser degree of endothelial dysfunction.9 However, the extent to which EETs contributed to this endothelial dysfunction remained unclear. Our study was the first to interrogate this further, and forearm blood flow data suggest that EET production is impaired similarly in patients with COPD and overweight smokers, supported by plasma quantification of EET/DHET as a representative of sEH activity.

ntributed to this endothelial dysfunction remained unclear. Our study was the first to interrogate this further, and forearm blood flow data suggest that EET production is impaired similarly in patients with COPD and overweight smokers, supported by plasma quantification of EET/DHET as a representative of sEH activity. We observed a trend toward reduced baseline EET/DHET in patients with COPD and overweight smokers, and when comparing the two matched control groups, the baseline EET/DHET ratio was slightly less in the younger matched control subjects for overweight smokers (matched control group 2) than those for COPD (matched control group 1). However, human plasma EET and DHET levels are notoriously difficult to quantify due to their instability; thus, definitive conclusions cannot be drawn from these insignificant results but can only be taken in context of our forearm blood flow data and previous published data. In animals, obesity is associated with reduced hepatic expression of EET-producing CYP2C enzymes.26 In mesenteric arteries of obese Zucker rats, there are reduced CYP2C and CYP2J enzymes, with enhanced activity of sEH enzymes.27 Increased sEH activity may represent more advanced inflammation, as in coronary artery disease; those who are obese or who smoke exhibit the lowest EET/DHET ratio.15 sEH activity is associated with forearm blood flow, as subjects with the Lys55Arg polymorphism in the sEH encoding gene (EPHX2) exhibit higher sEH activity and reduced vasodilator responses to bradykinin.28 Smoking can also significantly upregulate EPHX2,29 and this is associated with increased coronary artery calcification in humans.13

sociated with forearm blood flow, as subjects with the Lys55Arg polymorphism in the sEH encoding gene (EPHX2) exhibit higher sEH activity and reduced vasodilator responses to bradykinin.28 Smoking can also significantly upregulate EPHX2,29 and this is associated with increased coronary artery calcification in humans.13 The reduced EET synthesis and endothelial dysfunction observed in patients with COPD and overweight smokers may be a result of chronic low-grade inflammation secondary to smoking.30 In animals, dimethyl sulfoxide-soluble smoke particles can upregulate endothelium-derived vasoconstrictors through the nuclear factor kappa light-chain enhancer of activated B cells (NF-κB),31 a pivotal protein controlling the transcription of genes relevant to the pathophysiology of the blood vessel wall, including adhesion molecules and cytokines.10 EETs exert their antiinflammatory effects by inhibiting the activation of NF-κB.10 Inflammatory states are associated with downregulation of hepatic and extrahepatic CYP450 enzymes, resulting in a vicious cycle of reduced EET production and an ineffective EET-mediated antiinflammatory effect both locally and systemically.32

exert their antiinflammatory effects by inhibiting the activation of NF-κB.10 Inflammatory states are associated with downregulation of hepatic and extrahepatic CYP450 enzymes, resulting in a vicious cycle of reduced EET production and an ineffective EET-mediated antiinflammatory effect both locally and systemically.32 GSK2256294 is a potent sEH inhibitor that exerts high levels of sEH enzyme inhibition both in vitro19 and in vivo.20 In human left internal mammary arteries, 11,12-EETs are the most potent regioisomer,33 and we confirmed that both 11,12-EET- mediated and bradykinin-mediated vasodilation were enhanced in the presence of GSK2256294 in human resistance arteries. In animal models of cigarette smoking and obesity, sEH inhibition improves lung34 and endothelial function35 and attenuates pulmonary inflammation, as reflected by reduced inflammatory cells, including neutrophils and macrophages.19 In human bronchial cells, treatment with exogenous EETs protects against cigarette smoke extract-induced injury.36 Consistent with in vitro results, both acute and chronic sEH inhibition for up to 2 weeks improves responses to bradykinin.

reflected by reduced inflammatory cells, including neutrophils and macrophages.19 In human bronchial cells, treatment with exogenous EETs protects against cigarette smoke extract-induced injury.36 Consistent with in vitro results, both acute and chronic sEH inhibition for up to 2 weeks improves responses to bradykinin. No changes were observed in tPA release following sEH inhibition. tPA is a fibrinolytic serine protease that is released from the endothelium and regulates degradation of intravascular fibrin. Impaired tPA release can be associated with coronary atherosclerosis and cigarette smoking.25 Treatment of human endothelial cells with exogenous EETs, particularly 11,12-EETs, can increase tPA protein expression in a dose- and time-dependent manner, possibly due to activation of a G-protein, while not affecting PAI-1, the endogenous inhibitor of tPA.11 tPA release may also be dependent on the agonist, and in this group of overweight smokers, substance P may elicit a greater response.24

EETs, can increase tPA protein expression in a dose- and time-dependent manner, possibly due to activation of a G-protein, while not affecting PAI-1, the endogenous inhibitor of tPA.11 tPA release may also be dependent on the agonist, and in this group of overweight smokers, substance P may elicit a greater response.24 Some limitations of this study warrant consideration. Since the main focus of the phase I clinical trial was on safety, tolerability, and pharmacokinetics of GSK2256294 in healthy volunteers, we were not able to test this novel drug in patients with COPD. In addition, the lack of a nonsmoking control group in the phase I clinical trial means that the magnitudes of the effects of both doses of GSK2256294 were relatively small and similar to the variance in bradykinin responses in the placebo group. Therefore, phase II studies in larger patient groups are required to draw definitive conclusions.

a nonsmoking control group in the phase I clinical trial means that the magnitudes of the effects of both doses of GSK2256294 were relatively small and similar to the variance in bradykinin responses in the placebo group. Therefore, phase II studies in larger patient groups are required to draw definitive conclusions. Some evidence also suggests that in NO-deficient conditions, EETs may be upregulated.9 Thus, by creating an NO-deficient milieu during the forearm blood flow study with LNMMA, we may have masked any further upregulation of EETs by sEH inhibition. Larger clinical trials in patients with COPD, without concomitant inhibition of NO synthase, would be required to further understand the clinical impact of sEH inhibition. This must also be approached with caution because of the potential of EETs to stimulate angiogenesis, and possibly modulate cancer genesis and metastasis,37 although, interestingly, dual-action cyclooxygenase and sEH inhibition may in fact suppress cancer.38 We found no changes in serum vascular endothelial growth factor, the active drug group with this dosing regimen, after 14 days.20 Conclusions Patients with COPD and overweight smokers have impaired endothelial function and dysregulated EETs signaling. sEH inhibition can augment bradykinin-induced vasodilation in human resistance vessels both in vitro and in vivo, suggesting that sEH inhibition may be a novel therapeutic target to ameliorate cardiovascular risk in patients with smoking-related endothelial dysfunction. Supplementary Data e-Online Data

Conclusions Patients with COPD and overweight smokers have impaired endothelial function and dysregulated EETs signaling. sEH inhibition can augment bradykinin-induced vasodilation in human resistance vessels both in vitro and in vivo, suggesting that sEH inhibition may be a novel therapeutic target to ameliorate cardiovascular risk in patients with smoking-related endothelial dysfunction. Supplementary Data e-Online Data Acknowledgments Author contributions: I. B. W. is the guarantor of the manuscript. I. B. W., D. E. N., R. T-S., R. J. M., and J. C. contributed to the study design. L. Y., J. C., R. J. M., A. L. L.. L., D. G., Z. A., J. L. G., and I. B. W. contributed to data collection and interpretation. L. Y., J. C., D. G., Z. A., J. L. G., A. L. L.. L., D. E. N., R. T-S., and I. B. W. contributed to drafting the manuscript. Financial/nonfinancial disclosures: The authors have reported to CHEST the following: J. C. is employed by Cambridge University Hospitals NHS Foundation Trust and spends 50% of his time on GSK clinical trial research but receives no GSK benefits. D. E. N. has received consultancy fees from GSK. I. B. W. has received academic grants from GSK. R. J. M., A. L. L.., and R. T-S. are GSK employees and shareholders. None declared (L. Y., D. D. G., Z. A., J. L. G.). Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Other contributions: We thank participating subjects and acknowledge all clinical study site personnel who contributed to the clinical trial.

Financial/nonfinancial disclosures: The authors have reported to CHEST the following: J. C. is employed by Cambridge University Hospitals NHS Foundation Trust and spends 50% of his time on GSK clinical trial research but receives no GSK benefits. D. E. N. has received consultancy fees from GSK. I. B. W. has received academic grants from GSK. R. J. M., A. L. L.., and R. T-S. are GSK employees and shareholders. None declared (L. Y., D. D. G., Z. A., J. L. G.). Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Other contributions: We thank participating subjects and acknowledge all clinical study site personnel who contributed to the clinical trial. Additional information: The e-Appendix, e-Figures, and e-Table can be found in the Supplemental Materials section of the online article. Drs Yang and Cheriyan contributed equally to this manuscript.

Other contributions: We thank participating subjects and acknowledge all clinical study site personnel who contributed to the clinical trial. Additional information: The e-Appendix, e-Figures, and e-Table can be found in the Supplemental Materials section of the online article. Drs Yang and Cheriyan contributed equally to this manuscript. FUNDING/SUPPORT: This work was supported by GSK [SEH114068] and Innovate UK [ERICA Consortium 10037625]; the Wellcome Trust Translational Medicine and Therapeutics fellowship programme awarded to L. Y. [fellowship ref no: 100780/Z/12/Z]. The Wellcome Trust grant WT103782AIA was awarded to D. E. N.; the Raymond and Beverley Sackler fellowship awarded to L. Y.; National Institute for Health Research funding awarded to I. B. W., and J. C. in the Cambridge Comprehensive Biomedical Research; and the British Heart Foundation [Grant Nos. CH/09/002, and RG66885 RCZA/008 awarded to D. E. N. and I. B. W., respectively]. J. L. G. and Z. A. are funded by the Medical Research Council (Medical Research Council Lipid Profiling and Signalling, MC UP A90 1006 & Lipid Dynamics and Regulation, MC PC 13030). Figure 1 Plasma concentration of basal EET/DHET in patients with COPD and overweight smokers. There was a trend toward a higher EET/DHET ratio in matched control group 1 (blue) than in patients with COPD (red; P = .08) and a higher EET/DHET ratio in matched control group 2 (blue) compared with overweight smokers (gray; not significant). DHET = dihydroxyepoxyeicosatrienoic acid; EET = epoxyeicosatrienoic acid.

Financial/nonfinancial disclosures: The authors have reported to CHEST the following: F. L. M. R. has served on the advisory boards of AstraZeneca, Novartis, and Mundipharma. He has served as a consultant for AstraZeneca, Boehringer Ingelheim, Chiesi, Mundipharma, Malesci-Guidotti, Novartis, and Teva and has received grants from AstraZeneca, Boehringer Ingelheim, Chiesi, and GlaxoSmithKline. A. P. reports grants, personal fees, or reimbursement for travel expenses from AstraZeneca, Chiesi, Boehringer Ingelheim, GlaxoSmithKline, Menarini, Merck Sharp & Dohme, Mundipharma, Novartis, Teva, Sanofi, and Zambon. None declared (A. D. S., C. S., I. G., P. C., P. B., M. C., P. M., P. R., G. G., F. C., S. P., Y. G., K. F. C., B. J. B., I. M. A, B. B., G. C.). Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Additional information: The e-Appendix, e-Figures, and e-Tables can be found in the Supplemental Materials section of the online article. FUNDING/SUPPORT: This work was supported by Istituti Clinici Scientifici Maugeri, SpA, SB, IRCCS, Ricerca Corrente, FARs of the University of Ferrara (to G. C.). I. M. A. is supported by the BBSRC, the Wellcome Trust [Grant 093080/Z/10/Z], the Dunhill Medical Trust [Grant R368/0714], and the MRC-ABPI COPD-MAP consortium [Grant G1001367/1].

. There was a trend toward a higher EET/DHET ratio in matched control group 1 (blue) than in patients with COPD (red; P = .08) and a higher EET/DHET ratio in matched control group 2 (blue) compared with overweight smokers (gray; not significant). DHET = dihydroxyepoxyeicosatrienoic acid; EET = epoxyeicosatrienoic acid. Figure 2 Forearm blood flow responses in (A) patients with COPD and (B) overweight smokers. Bradykinin-induced vasodilation (solid lines) was greater in healthy matched control groups (blue) than in patients with COPD (red; *P = .005) and overweight smokers (gray; §P = .02). In the presence of fluconazole (dotted lines), bradykinin-induced vasodilation was reduced in healthy matched control subjects (**P < .0001 and §§P < .0001) but not in patients with COPD or overweight smokers. BK = bradykinin; FBF = forearm blood flow. Figure 3 Sum of baseline corrected EET/DHET ratio in response to bradykinin in patients with (A) COPD and (B) overweight smokers. Although not significant, there was a trend toward a greater increase in total EET/DHET ratio in response to bradykinin (solid lines) in healthy subjects (blue) compared with patients with COPD (red) and overweight smokers (gray). In the presence of fluconazole (dotted lines), there was a trend toward a reduced total EET/DHET ratio in the healthy group but not in patients with COPD or overweight smokers. See Figure 1 and 2 legends for expansion of abbreviations.

ubjects (blue) compared with patients with COPD (red) and overweight smokers (gray). In the presence of fluconazole (dotted lines), there was a trend toward a reduced total EET/DHET ratio in the healthy group but not in patients with COPD or overweight smokers. See Figure 1 and 2 legends for expansion of abbreviations. Figure 4 In vitro study. Effect of GSK2256294 on (A) 11,12-EET-induced vasodilation and (B) BK-induced vasodilation in LNAME- and indomethacin-treated human resistance arteries. (A) Isolated human arterioles (n = 6) were preconstricted with endothelin-1, and 11,12-EET-induced dilatation was examined in the absence and presence of 10 μM GSK2256294 (blue). (B) Bradykinin-induced dilatation was examined in the absence (gray) and presence of 1 μM (red) and 10 μM (blue) GSK2256294. *P < .05 compared with control group. LNAME = L-nitroarginine methyl ester. See Figure 1 and 2 legends for expansion of other abbreviations. Figure 5 Phase I clinical trial. Responses to bradykinin in overweight smokers who received (A) placebo, (B) 6 mg, and (C) 18 mg of active drug. Bradykinin induced significant vasodilation on all 3 days in all three treatment groups (P < .0001). Forearm blood flow improved overall in the active drug group (P = .007), with the greatest effect in the 18-mg active drug group, after acute dosing (*P = .02 in C) and after 14 days chronic dosing (**P = .003 in C). Solid lines represent predose; small dotted lines represent acute dose, and long dotted lines represent chronic dose. See Figure 2 legend for expansion of abbreviations.

.007), with the greatest effect in the 18-mg active drug group, after acute dosing (*P = .02 in C) and after 14 days chronic dosing (**P = .003 in C). Solid lines represent predose; small dotted lines represent acute dose, and long dotted lines represent chronic dose. See Figure 2 legend for expansion of abbreviations. Table 1 Demographics for All Study Subjects

.007), with the greatest effect in the 18-mg active drug group, after acute dosing (*P = .02 in C) and after 14 days chronic dosing (**P = .003 in C). Solid lines represent predose; small dotted lines represent acute dose, and long dotted lines represent chronic dose. See Figure 2 legend for expansion of abbreviations. Table 1 Demographics for All Study Subjects Subject Demographics (mean ± SD) COPD Group Overweight Smoker Group Phase I Clinical Group (Overweight Smokers) COPD (n = 12) Control (n = 12) P Value Overweight Smokers (n = 11) Control Group (n = 12) P Value 6-mg Dose (n = 11) 18-mg Dose (n = 11) Placebo (n = 6) P Value Age, y 63 ± 6 64 ± 7 .70 48 ± 8 49 ± 10 .47 43 ± 10 42 ± 9 41 ± 8 .92 BMI, kg/m2 27 ± 3 26 ± 3 .33 30 ± 3 25 ± 2 .0001 31 ± 2 31 ± 2 31 ± 3 .72 Height, m 1.75 ± 0.3 1.77 ± 0.1 .33 1.82 ± 0.1 1.80 ± 0.1 .73 1.83 ± 0.06 1.78 ± 0.04 1.76 ± 0.09 .11 Weight, kg 84 ± 12 82 ± 11 .77 103 ± 13 80 ± 7 .0001 103 ± 10 98 ± 10 95 ± 9 .25 Supine SBP, mm Hg 130 ± 17 133 ± 7 .52 130 ± 14 124 ± 14 .26 128 ± 16 141 ± 13 130 ± 13 .09 Supine DBP, mm Hg 80 ± 3 82 ± 5 .77 82 ± 5 78 ± 8 .34 79 ± 7 82 ± 8 76 ± 5 .26 Pack-years 39 ± 17 0 NA 21 ± 11 0 NA 20 ± 10 17 ± 9 16 ± 6 .58 Subject demographics for the physiological study and phase I clinical trial. There were no significant differences in the demographics between overweight smokers and healthy matched control group in the physiological study and no differences between placebo, 6 mg, and 18 mg active drug. For the physiological study, the P value was calculated using a Student t test for COPD vs the matched control groups, and for the phase I clinical trial, the P value was calculated using one-way analysis of variance comparison between three treatment groups.

no differences between placebo, 6 mg, and 18 mg active drug. For the physiological study, the P value was calculated using a Student t test for COPD vs the matched control groups, and for the phase I clinical trial, the P value was calculated using one-way analysis of variance comparison between three treatment groups. DBP = diastolic BP; NA = not available; SBP = systolic BP.

During 2016, there were estimated to be 10.4 million new TB cases worldwide, causing 1.7 million deaths.1 TB is transmitted mostly through coughing,2, 3, 4 which has been associated with increased bacillary burden.5 Cough can be assessed easily throughout treatment, but its relationship with cavitary lung disease, to our knowledge, has not been studied.6 Identifying radiologic characteristics associated with increased cough frequency is important in understanding transmission and evaluating treatment response.2, 5, 7, 8 Infectivity of TB is different for each individual, with some infecting more than others, so transmission in TB is considered heterogeneous.9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 In this longitudinal study, we sought to investigate whether there is an association between cough frequency, and its duration, with radiologic characteristics, such as cavitary volume and cavitary proximity to the airway. We also evaluated whether bacillary burden and culture conversion were associated with these radiologic characteristics.

Sirtuin-1 (SIRT1) is the mammalian homolog of silent information regulator (Sir2) family, initially described in yeast,1 and this highly preserved gene encodes nicotinamide adenine dinucleotide (NAD)-dependent protein deacetylases.2 Through modulating acetylating/deacetylating balances of multiple substrate proteins, SIRT1 regulates various cellular responses such as apoptosis, cellular senescence, endocrine metabolism, glucose homeostasis, and aging.2, 3, 4, 5, 6, 7 Although SIRT1 was originally described as a nuclear protein,8, 9 it has recently been shown that SIRT1 shuttles between the nucleus and cytoplasm,10, 11, 12 where it may associate with different target proteins in responding to divergent extracellular stimuli.13, 14, 15 Interestingly, SIRT1 has recently been measured in the serum,16 although its precise origin is unknown. In previous reports, serum SIRT1 (s120S) was consistently decreased with aging,17 and there was an accelerated reduction of serum SIRT1 in neurologic disorders such as Alzheimer’s disease,16 as well as in frailty18 and obesity,19, 20 all of which suggest that serum SIRT1 may be a potential biomarker for various aging-associated diseases. By contrast, an increase in serum SIRT1 has been reported in patients with asthma.21 However, the measurement of serum SIRT1 in other pulmonary diseases has not yet been elucidated.

in frailty18 and obesity,19, 20 all of which suggest that serum SIRT1 may be a potential biomarker for various aging-associated diseases. By contrast, an increase in serum SIRT1 has been reported in patients with asthma.21 However, the measurement of serum SIRT1 in other pulmonary diseases has not yet been elucidated. COPD is a major global health problem.22, 23 In contrast to asthma, COPD is mainly caused by noxious gases such as cigarette smoke24, 25 and is characterized by poorly reversible small airway obstruction, emphysema, and corticosteroid-insensitive inflammation.26 COPD progresses slowly; therefore, most patients are elderly, and there is increasing evidence that it reflects accelerated aging of the lungs.27, 28, 29 SIRT1 is decreased in the peripheral lung and peripheral blood mononuclear cells in patients with COPD.30 In this study, we measured the serum levels of SIRT1 by Western blot in patients with COPD and age-matched control subjects and examined how it relates to characteristics of the disease.

ing of the lungs.27, 28, 29 SIRT1 is decreased in the peripheral lung and peripheral blood mononuclear cells in patients with COPD.30 In this study, we measured the serum levels of SIRT1 by Western blot in patients with COPD and age-matched control subjects and examined how it relates to characteristics of the disease. Methods Reagents Commercially available reagents were obtained as follows: Roswell Park Memorial Institute medium (RPMI) medium 1640 (RPMI 1640) (No. 32404-014) and Dulbecco's Modified Eagle Medium (DMEM) (31053-028) were obtained from Life Technologies; fetal bovine serum (FBS), complete protease inhibitor cocktail (11836153001), and rabbit-derived anti-SIRT1 antibody (No. 5322) were obtained from Sigma-Aldrich; anti-β-actin antibody (ab6276) was obtained from Abcam; and goat-derived, peroxidase-conjugated antimouse (P0447) or antirabbit (P0448) secondary antibodies were obtained from DAKO.

), complete protease inhibitor cocktail (11836153001), and rabbit-derived anti-SIRT1 antibody (No. 5322) were obtained from Sigma-Aldrich; anti-β-actin antibody (ab6276) was obtained from Abcam; and goat-derived, peroxidase-conjugated antimouse (P0447) or antirabbit (P0448) secondary antibodies were obtained from DAKO. Patients and Healthy Volunteers for Serum This project was approved by the Ethics Committee of Sismanogleio General Hospital (approval No. 5210-07/03/2012), and written informed consent was obtained from patients and healthy volunteers. COPD was defined and categorized according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines.22 Blood samples were taken from never smoker healthy subjects with normal lung function (NS, 12 subjects), smokers without COPD (SM, 19 subjects), and 26 patients with mild to very severe COPD (stages 1-2, 13 subjects; stages 3-4, 13 subjects) (Table 1). All patients with COPD were considered to be clinically stable, because none had required a change in their regular therapy during the 8 weeks preceding the sampling nor had they been treated with systemic corticosteroids or antibiotics. Patients with asthma, pneumonia, or lung cancer were excluded from the study. The smoking history of each subject was represented by the mean number of pack-years of cigarette consumption by ex-smokers and current smokers. All patients with COPD had a history of smoking, but all patients were asked to refrain from smoking for 3 hours before the serum sampling. Emphysema was characterized by high-resolution CT (HRCT).31 The degree of emphysema was determined using a visual emphysema score, as previously described.32 Briefly, emphysema was identified as areas of hypovascular low attenuation and was graded with a five-point scale based on the percentage of lung involved: 0 = no emphysema, 1 = up to 25% of the lung parenchyma involved, 2 = between 26% and 50% of lung parenchyma involved, 3 = between 51% and 75% of the lung parenchyma involved, and 4 = between 76% and 100% of lung parenchyma involved. Grades of the axial images of each lung were added and divided by the number of images evaluated to yield emphysema scores that ranged from 0 to 4.

ved, 2 = between 26% and 50% of lung parenchyma involved, 3 = between 51% and 75% of the lung parenchyma involved, and 4 = between 76% and 100% of lung parenchyma involved. Grades of the axial images of each lung were added and divided by the number of images evaluated to yield emphysema scores that ranged from 0 to 4. Patients with COPD were characterized as having frequent exacerbations if there were two or more severe exacerbations in 1 year.33 The Medical Research Council (MRC) dyspnea scale,34 Borg scale (dyspnea and fatigue),35 6-min minute walking distance (6MWD),36 BMI, airflow obstruction, dyspnea, exercise capacity (BODE) index,37 and the Charlson index38 were examined according to the original reports. We also examined the air trapping by residual volume (RV)/total lung capacity (TLC) and the oxygenation capacity of lung by Pao2/Fio2 or by the alveolar-arterial oxygen difference (Aado2).Table 1 Characteristics of Study Subjects

(BODE) index,37 and the Charlson index38 were examined according to the original reports. We also examined the air trapping by residual volume (RV)/total lung capacity (TLC) and the oxygenation capacity of lung by Pao2/Fio2 or by the alveolar-arterial oxygen difference (Aado2).Table 1 Characteristics of Study Subjects Variable Nonsmokers Smokers Without COPD COPD Stages 1-2 COPD Stages 3-4 No. (M/F) 12 (3/9) 19 (11/8) 13 (10/3) 13 (11/2) Age, y 65.3 ± 11.1 58.9 ± 12.3 64.6 ± 10.1 64.2 ± 11.1 Pack-years 0 38.9 ± 25.3 70.9 ± 27.2a 68.1 ± 25.8a Emphysema 1 of 12 6 of 19 12 of 13 13 of 13 Emphysema score 0.29 ± 1.01 0.39 ± 0.74 1.33 ± 1.17 2.25 ± 1.20a,b MRC dyspnea score 0.33 ± 0.89 0.68 ± 0.95 1.46 ± 0.78c 2.38 ± 1.12a,b Charlson index 0.58 ± 1.00 0.95 ± 1.08 1.54 ± 1.27 1.92 ± 1.12c Pao2 (mm Hg) 81.2 ± 7.2 77.7 ± 5.8 73.6 ± 6.9 66.7 ± 9.0a,b Paco2 (mm Hg) 38.3 ± 2.4 39.5 ± 1.7 39.6 ± 2.7 39.8 ± 7.4 Pao2/Fio2 (mm Hg/%) 386.7 ± 34.4 370.2 ± 27.8 350.5 ± 33.0 307.7 ± 55.5a,b Aado2 20.6 ± 9.4 22.6 ± 6.3 25.4 ± 7.2 41.0 ± 23.1a,b Pao2/Paco2 2.12 ± 0.15 1.97 ± 0.17 1.82 ± 0.19b 1.72 ± 0.33b,d BMI (kg/m2) 25.8 ± 3.2 29.0 ± 7.5 24.8 ± 3.6 23.0 ± 3.5d FEV1 % predicted (%) 89.4 ± 12.6 89.7 ± 14.4 73.1 ± 14.6a,c 32.2 ± 7.8a,b FVC % predicted (%) 84.2 ± 11.3 87.2 ± 14.9 90.3 ± 18.9 57.2 ± 11.7a,b FEV1/FVC 84.4 ± 5.6 81.8 ± 9.2 61.4 ± 7.2a,b 45.9 ± 7.8a,b Dlco % predicted (%) 78.9 ± 19.7 77.3 ± 17.3 63.5 ± 20.8 43.0 ± 13.7a,b KCO % predicted (%) 85.1 ± 16.9 89.2 ± 10.9 67.8 ± 19.5a 59.9 ± 14.0a,b TLC % predicted (%) 83.7 ± 9.4 86.0 ± 25.6 94.6 ± 13.4 102.7 ± 36.2 FRC % predicted (%) 81.7 ± 8.3 82.2 ± 24.6 106.0 ± 27.1 110.3 ± 25.6c,d RV % predicted (%) 82.8 ± 8.6 80.2 ± 20.7 107.9 ± 23.9a,c 120.6 ± 18.5a,b RV/TLC 0.365 ± 0.062 0.324 ± 0.075 0.401 ± 0.085 0.473 ± 0.085a,b IC % predicted (%) 84.8 ± 10.1 86.6 ± 23.8 85.5 ± 16.3 56.7 ± 16.5a,b 6MWD (m) 480.2 ± 110.2 508.6 ± 108.5 439.2 ± 122.4 359.6 ± 147.4a DBOrgDyspnea 0.92 ± 2.27 0.68 ± 1.70 2.00 ± 1.73 3.15 ± 1.91a,c DBorgFatigue 0.92 ± 1.68 0.68 ± 1.11 1.23 ± 1.17 2.38 ± 1.66a Dsat (%) –1.2 ± 4.4 –0.7 ± 3.2 –2.5 ± 4.6 –7.2 ± 4.7a,b DHR 30.6 ± 8.3 21.3 ± 9.5 32.5 ± 23.1 22.3 ± 15.4 LABA/ LAMA/ ICS 1/ 1/ 0 2/ 2/ 3 8/ 9/ 3 10/ 10/ 10 CVD/ DM 2/ 2 7/ 2 6/ 1 5/ 3 Data are expressed as mean values ± SD. Pack-year represents the number of cigarettes smoked per day/20 × duration of smoking in years.

Dsat (%) –1.2 ± 4.4 –0.7 ± 3.2 –2.5 ± 4.6 –7.2 ± 4.7a,b DHR 30.6 ± 8.3 21.3 ± 9.5 32.5 ± 23.1 22.3 ± 15.4 LABA/ LAMA/ ICS 1/ 1/ 0 2/ 2/ 3 8/ 9/ 3 10/ 10/ 10 CVD/ DM 2/ 2 7/ 2 6/ 1 5/ 3 Data are expressed as mean values ± SD. Pack-year represents the number of cigarettes smoked per day/20 × duration of smoking in years. Patients with COPD were categorized by the Global Initiative for Chronic Obstructive Lung Disease definition guidelines.22 Emphysematous phenotype was characterized according to the presence of significant emphysematous lesions (> 15% of lung parenchyma) by high-resolution CT.31 Patients with COPD were characterized as having frequent exacerbations if there were two or more exacerbations in 1 year.33 6MWD = six-min walking distance; Aado2 = alveolar-arterial oxygen difference; CVD = cardiovascular disease; DHR = difference in heart rate during 6MWD; DBorgDyspnea = the difference in Dyspnea level according to Borg scale between the end and the beginning of the 6-min walk test; DBorgFatigue = the difference in Fatigue level according to Borg scale between the end and the beginning of the 6-min walk test; Dlco = diffusing capacity of lung for carbon monoxide; Dsat = Desaturation on movement; DM = diabetes mellitus; FRC = functional residual capacity; IC = inspiratory capacity; ICS = inhaled corticosteroids; KCO = the carbon monoxide transfer coefficient; LABA = long acting beta-agonists; LAMA = long-acting muscarinic antagonists; MRC = Medical Research Council; RV = residual volume; TLC = total lung capacity. a P < .01 vs smokers without COPD. b P < .01 vs nonsmokers.

6MWD = six-min walking distance; Aado2 = alveolar-arterial oxygen difference; CVD = cardiovascular disease; DHR = difference in heart rate during 6MWD; DBorgDyspnea = the difference in Dyspnea level according to Borg scale between the end and the beginning of the 6-min walk test; DBorgFatigue = the difference in Fatigue level according to Borg scale between the end and the beginning of the 6-min walk test; Dlco = diffusing capacity of lung for carbon monoxide; Dsat = Desaturation on movement; DM = diabetes mellitus; FRC = functional residual capacity; IC = inspiratory capacity; ICS = inhaled corticosteroids; KCO = the carbon monoxide transfer coefficient; LABA = long acting beta-agonists; LAMA = long-acting muscarinic antagonists; MRC = Medical Research Council; RV = residual volume; TLC = total lung capacity. a P < .01 vs smokers without COPD. b P < .01 vs nonsmokers. c P < .05 vs nonsmokers. d P < .05 vs smokers without COPD. Blood Sampling Blood samples were collected in BD Vacutainer Plus Plastic Serum and SST Tubes, which are coated with silicone and micronized silica particles to accelerate clotting. Samples were then centrifuged at 1500g for 15 min at room temperature, and supernatants were aliquoted as serum samples and immediately stored at –70 oC until measurement.

collected in BD Vacutainer Plus Plastic Serum and SST Tubes, which are coated with silicone and micronized silica particles to accelerate clotting. Samples were then centrifuged at 1500g for 15 min at room temperature, and supernatants were aliquoted as serum samples and immediately stored at –70 oC until measurement. Pulmonary Function Tests Pulmonary function tests were performed using MasterScreen (Erich Jaeger GmbH) and included postbronchodilator FEV1, FVC, FEV1/FVC ratio, TLC, RV, inspiratory capacity (IC), and diffusing capacity for carbon monoxide (Dlco). Dlco and diffusing capacity for carbon monoxide adjusted for alveolar volume (DLCO/VA or KCO) were assessed by the single-breath method with the patient in the sitting position. Lung function measurements were expressed as percentage of predicted values. Tests were performed according to the American Thoracic Society/European Respiratory Society guidelines by the same technician to ensure consistency of results. All lung function data are shown in Table 1.

the patient in the sitting position. Lung function measurements were expressed as percentage of predicted values. Tests were performed according to the American Thoracic Society/European Respiratory Society guidelines by the same technician to ensure consistency of results. All lung function data are shown in Table 1. Serum SIRT1 Serum samples were diluted in radioimmunoprecipitation assay buffer (Sigma; 150 mM NaCl, 1.0% IGEPAL CA-630, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, and 50 mM tris(hydroxymethyl)aminomethane, pH 8.0) completed with protease inhibitor, as previously published,39 separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to nitrocellulose membrane, and incubated with anti-SIRT1 antibody or with anti-β-actin antibody overnight. The membranes were then incubated with the appropriate peroxidase-conjugated secondary antibodies. The bound antibodies were visualized by chemiluminescence (ECL Plus; GE Healthcare). The band density in each sample was normalized to the level of standard sample (from one healthy donor), and each SIRT1 protein level was shown as the relative ratio against that in standard sample.

ase-conjugated secondary antibodies. The bound antibodies were visualized by chemiluminescence (ECL Plus; GE Healthcare). The band density in each sample was normalized to the level of standard sample (from one healthy donor), and each SIRT1 protein level was shown as the relative ratio against that in standard sample. Cell Culture BEAS-2B cells (SV40-immortalized human airway bronchial epithelial cell line) and A549 cells (human lung adenocarcinoma epithelial cell line) were purchased from the American Culture of Tissue Collection and grown in complete growth medium (RPMI 1640 and DMEM supplemented with heat-inactivated 10% FBS and 1% L-glutamine, respectively) at 37oC/5% CO2. Before use, cells were starved in minimum medium (RPMI 1640 or DMEM supplemented with 1% FBS and 1% L-glutamine), and cell culture supernatants were harvested at different time points. To eliminate the contamination of supernatant by free-floating cells, the supernatant was centrifuged and the upper half of the medium was taken as the sample.

tarved in minimum medium (RPMI 1640 or DMEM supplemented with 1% FBS and 1% L-glutamine), and cell culture supernatants were harvested at different time points. To eliminate the contamination of supernatant by free-floating cells, the supernatant was centrifuged and the upper half of the medium was taken as the sample. Human primary bronchial epithelial cells obtained from three subjects without COPD and three subjects with COPD were cultured as monolayers in LHC-9 media (Invitrogen) on collagen (1% w/v)-coated plates as previously reported.40 Cells were extracted from lung tissue of patients undergoing lung resection at the Royal Brompton Hospital. All subjects gave informed written consent and the study was approved by the National Research Ethics Service London-Chelsea Research Ethics committee, study No. 09/H0801/85. All cells were serum starved 16 hours before stimulation. Cells were stimulated with 3% cigarette-smoke-conditioned media prepared as previously reported.41 Statistical Analysis Data from clinical samples were expressed as mean values ± SD. For the analysis of SIRT1, statistical significance was assessed using a nonparametric Kruskal-Wallis test with a Bonferroni multiple comparison procedure to exclude possible interaction between various variables within subgroups (Statcel 2, OMS Publishing Inc.). The analysis of correlation between each factor was performed by Spearman correlation coefficient rank sum test. All reported P values were two-sided, and P values < .05 were considered to be statistically significant.

ude possible interaction between various variables within subgroups (Statcel 2, OMS Publishing Inc.). The analysis of correlation between each factor was performed by Spearman correlation coefficient rank sum test. All reported P values were two-sided, and P values < .05 were considered to be statistically significant. Results In a previous report, s120S was found to be detectable by Western blot,18 which showed an excellent correlation with the enzyme-linked immunosorbent assay (ELISA). As shown in Figure 1, anti-SIRT1 antibody used in this study detected different sizes of SIRT1, including 75, 102, and 120 kDa (the size originally reported) in BEAS-2B cells or A549 cells; therefore, we determined these SIRT1 fractions in serum samples separately. Compared with healthy subjects, the patients with COPD showed decreased levels of 120-kDa s120S (SIRT1 ratio in healthy subjects [NS + SM], 0.90 ± 0.34 vs subjects with COPD, 0.68 ± 0.24; P = .014) (Fig 2A), whereas SIRT1 with lower molecular weights (102 kDa and 75 kDa) did not (Figs 2B, 2C, e-Figs 1A, 1B, 1C). s120S showed a significant positive correlation with airway obstruction (FEV1/ FVC ratio; r, 0.31; P = .020) (Fig 2D, Table 2) and also with the severity of airway obstruction, measured by FEV1 % predicted (r = 0.29; P = .029) (Fig 2E), suggesting that s120S protein levels decrease with COPD progression (Fig 2F).Figure 1 SIRT1 protein in serum. Western blot analysis of serum sample (S) was compared with whole-cell extracts of BEAS-2B cells (B) and A549 cells (A). SIRT1 = silent information regulator 2 homolog 1; WCE = whole cell extract.

29) (Fig 2E), suggesting that s120S protein levels decrease with COPD progression (Fig 2F).Figure 1 SIRT1 protein in serum. Western blot analysis of serum sample (S) was compared with whole-cell extracts of BEAS-2B cells (B) and A549 cells (A). SIRT1 = silent information regulator 2 homolog 1; WCE = whole cell extract. Figure 2 Reduced levels of 120-kDa serum SIRT1 (s120S) protein in COPD. A, s120S protein level in serum from healthy subjects and subjects with COPD (NS + SM) (C1-4 disease stage). Protein levels of (B) 102-kDa and (C) 75-kDa SIRT1 with or without COPD. D, Correlation between the s120S protein level and FEV1/FVC ratio. E, Correlation between the s120S protein level and FEV1 % predicted. F, s120S protein levels of healthy subjects and patients with COPD (NS + SM) of different stages (C1-2, C3-4). NS = nonsmoking subjects; PC = positive control from healthy subject; SM = smokers without COPD. Table 2 Spearman Correlation Coefficient Rank Sum Test Between the s120S (120-kDa) SIRT1 and Patient Characteristics

Figure 2 Reduced levels of 120-kDa serum SIRT1 (s120S) protein in COPD. A, s120S protein level in serum from healthy subjects and subjects with COPD (NS + SM) (C1-4 disease stage). Protein levels of (B) 102-kDa and (C) 75-kDa SIRT1 with or without COPD. D, Correlation between the s120S protein level and FEV1/FVC ratio. E, Correlation between the s120S protein level and FEV1 % predicted. F, s120S protein levels of healthy subjects and patients with COPD (NS + SM) of different stages (C1-2, C3-4). NS = nonsmoking subjects; PC = positive control from healthy subject; SM = smokers without COPD. Table 2 Spearman Correlation Coefficient Rank Sum Test Between the s120S (120-kDa) SIRT1 and Patient Characteristics Variable All Subjects Normal Subjects COPD Only r P Value r P Value r P Value BMI 0.36 .0077 0.28 .13 0.34 .092 Pack-year –0.33 .014 –0.16 .38 –0.13 .51 FEV1/FVC 0.31 .021 –0.072 .69 0.32 .11 Emphysema score –0.38 .0048 –0.082 .33 –0.34 .091 Kco % predicted 0.32 .025 0.059 .77 0.32 .13 FEV1 % predicted 0.29 .032 –0.088 .63 0.40 .046 Pao2/Paco2 0.28 .034 0.22 .22 0.058 .77 RV % predicted –0.27 .054 0.11 .57 –0.23 .26 IC % predicted 0.28 .064 0.26 .20 0.25 .28 Dlco % predicted 0.26 .069 –0.043 .83 0.33 .10 Pao2/Fio2 0.23 .079 0.11 .55 0.036 .86 Pao2 0.22 .098 0.11 .55 –0.026 .90 6MWD 0.22 .11 –0.097 .60 0.45 .023 Dsat 0.20 .14 0.14 .45 0.12 .56 FVC % predicted 0.20 .14 –0.059 .75 0.27 .18 Aado2 –0.18 .17 –0.032 .86 –0.052 .80 FRC % predicted –0.19 .18 0.10 .60 –0.083 .69 RV/TLC –0.16 .24 –0.013 .95 –0.025 .90 Paco2 –0.06 .65 –0.097 .60 0.039 .85 TLC % predicted –0.06 .66 0.21 .30 –0.10 .61 See Table 1 legend for expansion of abbreviations.

4 0.14 .45 0.12 .56 FVC % predicted 0.20 .14 –0.059 .75 0.27 .18 Aado2 –0.18 .17 –0.032 .86 –0.052 .80 FRC % predicted –0.19 .18 0.10 .60 –0.083 .69 RV/TLC –0.16 .24 –0.013 .95 –0.025 .90 Paco2 –0.06 .65 –0.097 .60 0.039 .85 TLC % predicted –0.06 .66 0.21 .30 –0.10 .61 See Table 1 legend for expansion of abbreviations. In addition, s120S showed a negative correlation with the amount of cigarette consumption (pack-year; r = –0.33; P = .014) (Fig 3A). Patients with a higher degree of emphysema on HRCT had lower levels of s120S (r = –0.40; P = .027) (Fig 3B) when analyzed in all subjects showing some degree of emphysema. A good correlation was also observed in all subjects (P = .0048) (Table 2) and COPD subjects only (P = .091) (Table 2). In addition, patients with emphysema showed decreased levels of s120S when compared with the patients with normal lungs (SIRT1 ratio in control population, 0.92 ± 0.37 vs 0.71 ± 0.24 in those with emphysema; P = .026) (e-Fig 1D). This was confirmed by the significant positive correlation between the s120S SIRT1 and KCO % predicted (r = 0.32; P = .025) (Fig 3C, Table 2). The s120S was not correlated with age, probably because of the elderly bias of samples included. In contrast, the BMI showed a significant positive correlation with s120S (r = 0.36; P = .0077) (Fig 3D, Table 2). In addition, s120S decreased significantly as symptoms (MRC dyspnea score) increased (Fig 3E). The severity of hypoxia (Pao2 or desaturation on movement) and oxygenation capacity of lung (Pao2/Fio2 or Aado2) did not show any correlation with s120S (Table 2); however, s120S showed positive correlation with Pao2/Paco2 ratio representing the combined effect on gas exchange42 (r = 0.28; P = .034) (Fig 3F), which suggested that the impairment of aerobic metabolism might contribute to the s120S protein level. Other patient background characteristics (Table 2) or subject comorbidities (such as cardiovascular disease or diabetes mellitus) and Charlson index did not show any association with serum levels of SIRT1.Figure 3 Serum SIRT1 (s120S) 120-kDa protein and patient characteristics. A, Relationship between the s120S protein level and cigarette smoke exposure in pack-years, (B) with emphysema score in all subjects demonstrating emphysema, (C) KCO % predicted, (D) BMI, (E) MRC dyspnea score, and (F) Pao2/Paco2 ratio in all subjects. Pack-year represents the number of cigarettes smoked per day/20 × duration of smoking in years.

s120S protein level and cigarette smoke exposure in pack-years, (B) with emphysema score in all subjects demonstrating emphysema, (C) KCO % predicted, (D) BMI, (E) MRC dyspnea score, and (F) Pao2/Paco2 ratio in all subjects. Pack-year represents the number of cigarettes smoked per day/20 × duration of smoking in years. KCO = the carbon monoxide transfer coefficient; MRC = Medical Research Council.

s120S protein level and cigarette smoke exposure in pack-years, (B) with emphysema score in all subjects demonstrating emphysema, (C) KCO % predicted, (D) BMI, (E) MRC dyspnea score, and (F) Pao2/Paco2 ratio in all subjects. Pack-year represents the number of cigarettes smoked per day/20 × duration of smoking in years. KCO = the carbon monoxide transfer coefficient; MRC = Medical Research Council. When we limited analysis to the patients with COPD only, we identified two additional findings. First, patients with COPD with frequent exacerbations tend to have lower s120S levels compared with those with stable disease (Fig 4A). Second, s120S had a positive correlation not only with the FEV1 % predicted (r = 0.40; P = .046) (Fig 4B) but also with 6MWD (r = 0.45; P = 0.023) (Fig 4C). This was also confirmed by the fact that s120S was negatively associated with the MRC dyspnea score (Fig 4D) and with the BODE index (Fig 4E), which is known to be a strong predictor of long-term prognosis in COPD.37 We further analyzed the correlation with all parameters in the GOLD stage 1-2 population only and the 3-4 population only. We did not see any significant correlation except for IC % in stages 1-2 (e-Table 1).Figure 4 The s120S protein levels of patients with COPD. A, The s120S protein level in relation to frequent exacerbations of COPD. Relationship between the s120S protein level and (B) FEV1 % predicted, (C) 6MWD, (D) MRC dyspnea score, and (E) s120S protein levels in patients with COPD with different BODE index. F, Time dependency of SIRT1 in cell-culture supernatant from BEAS-2B cells (B2B) or (G) A549 cells. 6MWD = 6-min walking distance; BODE = BMI, airflow obstruction, dyspnea, and exercise capacity. See Figure 1 legend for expansion of other abbreviations.

0S protein levels in patients with COPD with different BODE index. F, Time dependency of SIRT1 in cell-culture supernatant from BEAS-2B cells (B2B) or (G) A549 cells. 6MWD = 6-min walking distance; BODE = BMI, airflow obstruction, dyspnea, and exercise capacity. See Figure 1 legend for expansion of other abbreviations. Regarding the origin of s120S, we were unable to detect any active secretion from lung epithelial cell lines such as BEAS-2B (airway) cells or A549 (parenchymal) cells (Figs 4F, 4G). In A549 cells, SIRT1 was detected in supernatant after 72 hours of culturing, but as the housekeeping protein β-actin was also detected, it was likely that this may have been due to increased cell permeability related to loss of function. Furthermore, we also investigated SIRT1 protein release in supernatant from non-COPD (n = 3) and COPD (n = 3) primary bronchial epithelial cells. We did not find original or degraded SIRT1 protein or β-actin in supernatant in the absence or presence of cigarette-smoke-conditioned media (data not shown). Thus, it was unlikely that SIRT1 was excreted from bronchial epithelial cells.

rnatant from non-COPD (n = 3) and COPD (n = 3) primary bronchial epithelial cells. We did not find original or degraded SIRT1 protein or β-actin in supernatant in the absence or presence of cigarette-smoke-conditioned media (data not shown). Thus, it was unlikely that SIRT1 was excreted from bronchial epithelial cells. Discussion In the current study, we showed for the first time, to our knowledge, that the protein levels of s120S in its 120-kDa form were significantly decreased in patients with COPD. The s120S protein levels were positively correlated with the severity of airway obstruction and showed a strong negative correlation with the amount of cigarette consumption, suggesting that oxidative stress may lead to the reduction of s120S. This is contrast to bronchial asthma,21 in which s120S detected by ELISA was increased. In addition, we found that s120S was significantly correlated with the severity of emphysema (HRCT reading and KCO % predicted) and functional disability represented by an MRC dyspnea score, 6MWD, and BODE score. These results suggest that s120S may be a useful marker for assessing certain disease characteristics in patients with COPD.

ound that s120S was significantly correlated with the severity of emphysema (HRCT reading and KCO % predicted) and functional disability represented by an MRC dyspnea score, 6MWD, and BODE score. These results suggest that s120S may be a useful marker for assessing certain disease characteristics in patients with COPD. Among the seven sirtuin isozymes,43 SIRT1 is the most widely studied in mammals from the viewpoint of regulation by oxidative stress, which is relevant to cellular senescence and chronic inflammation.2, 3, 4 In fact, dysregulation of SIRT1 has been described not only in aging-associated diseases but also in those associated with long-term cigarette smoking,44, 45 and all are characterized by oxidant/antioxidant imbalance.46 We previously reported a reduction in SIRT1 in the peripheral lungs of patients with COPD.30 Although reports of reduced SIRT1 relate to intracellular SIRT1 (mRNA or protein), Kumar et al16 first reported that SIRT1 was detectable in the serum. In this report, s120S was measured by various methods, including Western blot, ELISA, and surface plasmon resonance, with good correlation with each method, confirming that SIRT1 is a serum protein. Interestingly, they also showed significant reduction of s120S protein levels as dementia progressed, suggesting that s120S may be a useful biomarker for assessing cognitive disease. This report was surprising, as SIRT1 had originally been described only as a nuclear protein.8, 9 However, recent reports have demonstrated that SIRT1 can shuttle between the nucleus and cytoplasm10, 11, 12; therefore, SIRT1 is potentially present in the extracellular component.

omarker for assessing cognitive disease. This report was surprising, as SIRT1 had originally been described only as a nuclear protein.8, 9 However, recent reports have demonstrated that SIRT1 can shuttle between the nucleus and cytoplasm10, 11, 12; therefore, SIRT1 is potentially present in the extracellular component. The strength of our study is the selective determination of the fraction of full-length SIRT1 (120 kDa) separate from other truncated SIRT1 proteins by Western blot. This is in contrast to previous reports that used ELISA for s120S detection.18, 19, 20, 21 Despite its good quantitative capability, ELISA does not appear to be specific for the full-length functional fraction of SIRT1, because antibodies recognize several fractions with the target motif, irrespective of their function (Fig 1A). In previous reports, several bands of SIRT1 protein (original and truncated proteins) were described, indicating different molecular weights by Western blot,47, 48 each of which may function differently, although this has not yet been elucidated. Therefore, our results are unique, as we were able to analyze only the fully functional fraction of full-length SIRT1 (120-kDa SIRT1) that was separated by Western blot. Thus, Western blot should be used for s120S detection.

blot,47, 48 each of which may function differently, although this has not yet been elucidated. Therefore, our results are unique, as we were able to analyze only the fully functional fraction of full-length SIRT1 (120-kDa SIRT1) that was separated by Western blot. Thus, Western blot should be used for s120S detection. In addition, this is the first report to show that s120S is reduced in patients with COPD. Compared with healthy subjects, patients with COPD showed decreased levels of s120S, which correlated not only with the airway obstruction but also with its severity, resultant lung emphysema, and decreased diffusion capacity. These results were compatible with the previous reports that found that the SIRT1 protein level was decreased in peripheral lung or peripheral mononuclear cells in patients with COPD.30 Furthermore, we could also detect the association of s120S protein levels with BMI, MRC dyspnea score, and Pao2/Paco2 imbalance, all of which suggested that s120S is not just an indicator of lung damage but is a surrogate marker for oxygen metabolism and systemic metabolic status. Interestingly, our result appears to be opposite to that reported in patients with asthma21; therefore, s120S might be a potential biomarker to help to differentiate these two diseases. Future studies might be necessary for comparing the s120S levels directly between patients with asthma and COPD populations. Since reduced s120S is also reported in association with frailty in elderly people, it may also be useful in understanding the multimorbidity associated with COPD. We further analyzed the correlation with all parameters in a GOLD stage 1-2 population only and in a GOLD stage 3-4 population only. We did not see any significant correlation in all parameters except for IC % predicted in the stage 1-2 population. However, we observed a nearly significant correlation in FEV1/FVC (P = .19), emphysema score (P = .080), KCO % (P = .17), RV % (P = .090), and 6MWD (P = .11) in the stage 1-2 population. We did not have enough power to demonstrate the association in the current study, but a future large study will reveal the usefulness of s120S as a potential biomarker to determine the early stage of COPD.

19), emphysema score (P = .080), KCO % (P = .17), RV % (P = .090), and 6MWD (P = .11) in the stage 1-2 population. We did not have enough power to demonstrate the association in the current study, but a future large study will reveal the usefulness of s120S as a potential biomarker to determine the early stage of COPD. A limitation of the present study is that we have not identified the precise source of SIRT1 in serum. We could not detect any fractions of SIRT1 in the cell culture supernatant in A549 and BEAS-2B epithelial cells, indicating that cellular leakage or active secretion is unlikely. In primary bronchial epithelial cells, we could not find full or degraded SIRT1 proteins or β-actin in supernatant in the presence or absence of cigarette-smoke-conditioned media, suggesting that SIRT1 is unlikely secreted by bronchial epithelial cells. However, SIRT1 protein was detected in supernatant from A549 cells at a later stage, which was associated with an increase in β-actin expression. This suggests that epithelial cells are still a possible source of SIRT1 when cells are damaged. Considering that SIRT1 in patients with COPD has been reported to be decreased not only in the lung30 but also in endothelial progenitor cells49 and circulating leukocytes,30, 50 decreased s120S might reflect the reduction of SIRT1 in cells as a result of oxidative stress. Peripheral blood mononuclear cells or alveolar macrophages might be potential sources of SIRT1. It might be necessary in a future study to identify the precise origin of s120S and the factors that modulate s120S levels in COPD and other chronic aging diseases. Secondarily, even though there were statistical differences in SIRT1 levels between the comparison groups, there was significant overlap in the values of all groups. GOLD stage is defined by lung function (mainly by FEV1 % predicted), but, as discussed earlier, there was good correlation between SIRT1 and certain disease characteristics such as emphysema, MRC dyspnea score, 6MWD, and BODE score. Therefore, SIRT1 level is influenced by several factors rather than lung function alone. The current study is too small to evaluate further, but we believe that a future large study with more patients will provide a novel approach to classify disease stage based on SIRT1.

ysema, MRC dyspnea score, 6MWD, and BODE score. Therefore, SIRT1 level is influenced by several factors rather than lung function alone. The current study is too small to evaluate further, but we believe that a future large study with more patients will provide a novel approach to classify disease stage based on SIRT1. Conclusions In summary, we report for the first time that s120S was reduced in patients with COPD and that this reduction was correlated with the extent of emphysema and reduced functional measurements that correlate with disease progression. Serum SIRT1 might therefore serve as a potential biomarker for COPD. Supplementary Data e-Figure 1 and e-Table 1 Acknowledgments Author contributions: S. Y. conducted the assays, carried out the data analysis, and drafted the manuscript. A. I. P., A. P., and C. V. were involved in sample preparation and participated in the design of the original study. K. I. contributed to the data analysis, design of the study, and the manuscript preparation, and is the guarantor of the manuscript. P. B. and S. L. participated in the design of the study and contributed substantially to preparation of manuscript.

ample preparation and participated in the design of the original study. K. I. contributed to the data analysis, design of the study, and the manuscript preparation, and is the guarantor of the manuscript. P. B. and S. L. participated in the design of the study and contributed substantially to preparation of manuscript. Financial/nonfinancial disclosures: The authors have reported to CHEST the following: S. Y. is a recipient of a Banyu Life Science Foundation International fellowship. C. V. and J. B. are recipients of the Wellcome Trust grant. K. I. is currently an employee of Pulmocide Ltd and has honorary contract with Imperial College. P. J. B. has served on scientific advisory boards of AstraZeneca plc, Boehringer Ingelheim GmbH, Bespak (Consort Medical plc), Chiesi Pharmaceuticals Inc, Daiichi-Sankyo Co Ltd, Deep Breeze Ltd, GlaxoSmithKline plc, Glenmark Pharmaceuticals, Johnson & Johnson, Merck & Co Inc, Novartis AG, Nycomed International Management GmbH (Takeda Pharmaceutical Co Ltd), Pfizer Inc, Prosonix Ltd, Sun Pharmaceutical Industries Ltd, Teva Pharmaceutical Industries Ltd, and UCB Inc, and has received research funding from Aquinox Pharmaceuticals Inc, AstraZeneca plc, Boehringer Ingelheim GmbH, Chiesi Pharmaceuticals Inc, Daiichi-Sankyo Co Ltd, GlaxoSmithKline plc, Novartis AG, Nycomed International Management GmbH (Takeda Pharmaceutical Co Ltd), Pfizer Inc, Prosonix Ltd, and Sun Pharmaceuticals Industries Ltd. None declared (A. I. P., A. P., S. L.).

aceuticals Inc, AstraZeneca plc, Boehringer Ingelheim GmbH, Chiesi Pharmaceuticals Inc, Daiichi-Sankyo Co Ltd, GlaxoSmithKline plc, Novartis AG, Nycomed International Management GmbH (Takeda Pharmaceutical Co Ltd), Pfizer Inc, Prosonix Ltd, and Sun Pharmaceuticals Industries Ltd. None declared (A. I. P., A. P., S. L.). Other contributions: A special thanks to Peter Fenwick, MSc, and Prof Louise Donnelly, PhD, for kindly providing us with the primary epithelial cells. Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Additional information: The e-Figure and e-Table can be found in the Supplemental Materials section of the online article. FUNDING/SUPPORT: This project was supported by the Wellcome Trust Programme. [Grant 093080/Z/10/Z].

The transforming growth factor-β (TGF-β) family regulates cell proliferation, differentiation, extracellular matrix synthesis, and apoptosis, which are all important processes in COPD pathogenesis. Attenuation of TGF-β signaling leads to pulmonary emphysema in animal models,1, 2 which may reflect TGF-β1 effects on vascular endothelial growth factor and angiogenesis.3, 4 TGF-β1 also has a pivotal role in maintaining peripheral tolerance against self-antigens5 and controlling autoimmune responses.6, 7 The TGF-β superfamily has several members: TGF-β exists as three isoforms—TGF-β1, TGF-β2, and TGF-β3—which exhibit similar functions in vitro but may have distinct activities in vivo.8 TGF-βs act through specific receptors (TGF-β receptors [TGF-βRs]) I, II, and III. All TGF-βs bind a heteromeric type I/II receptor complex, which activates both “canonical” signals involving SMADs (small mother against decapentaplegic) and “noncanonical” pathways involving mitogen-activated protein kinases (MAPKs) and phosphoinositide 3-kinase (PI3K).9

ors (TGF-β receptors [TGF-βRs]) I, II, and III. All TGF-βs bind a heteromeric type I/II receptor complex, which activates both “canonical” signals involving SMADs (small mother against decapentaplegic) and “noncanonical” pathways involving mitogen-activated protein kinases (MAPKs) and phosphoinositide 3-kinase (PI3K).9 The SMAD family of transcription factors consists of the receptor-regulated SMADs, a common pathway SMAD, and inhibitory SMADs. Receptor-regulated SMADs include SMAD2 and SMAD3, which are recognized by TGF-βRs and activin receptors, and SMADs 1, 5, 8, and 9, which are activated by bone morphogenetic protein (BMP) receptors. SMAD4 or cooperating SMAD is not phosphorylated by the TGF-βRs, whereas inhibitory SMADs (anti-SMADs), including SMAD6 and SMAD7, downregulate TGF-β signaling. SMADs also act as signal integrators and interact with MAPK, nuclear factor-κB, PI3K, and hypoxia-inducible factor 1 signaling pathways.8, 10 TGF-βRIII (also known as β-glycan) acts as a coreceptor and directly binds TGF-β1, 2, and 3 to enhance their binding to the TGF-βRI/II complex and increase SMAD-dependent signaling.11 TGF-β signaling is modulated by connective tissue growth factor (CTGF or CCN2). CNN directly binds TGF-β1 and facilitates binding to the TGF-βRI/II complex enhancing downstream signaling.12

or and directly binds TGF-β1, 2, and 3 to enhance their binding to the TGF-βRI/II complex and increase SMAD-dependent signaling.11 TGF-β signaling is modulated by connective tissue growth factor (CTGF or CCN2). CNN directly binds TGF-β1 and facilitates binding to the TGF-βRI/II complex enhancing downstream signaling.12 The latency-associated peptide (LAP) is associated with latent TGF-β binding proteins (LTBPs) forming a complex known as the large latent complex (LLC).13, 14 Most TGF-β is secreted as part of the LLC. LTBPs belong to the fibrillin-LTBP family of extracellular matrix (ECM) proteins. LTBP-1 anchors to the ECM and creates traction when LAP binds cell surface integrins; this traction deforms LAP, which releases active TGF-β.13, 15 Conversely, TGF-βR-associated binding protein (TRAP) 1 inhibits TGF-β1 function by interfering with SMAD3 signaling.16 The 5′-TG-3′-interacting factors (TGIFs), TGIF1 and TGIF2, cause repression of TGF-β1-activated genes by direct competition with Smad2.17 In addition, BMP and activin membrane-bound inhibitor (BAMBI) acts as a competitive receptor antagonist for TGF-βRI.18 BAMBI expression is upregulated by TGF-β1 in a feedback loop.19 TGF-β-induced protein (TGFBI, also known as BIG-H3 and keratoepithelin) is an extracellular matrix protein used as a TGF-β1 bioactivity marker.20 The aim of this study is to investigate the expression of TGF-β signaling pathways in the lower airways (bronchial mucosa and peripheral lung) of patients with stable COPD and control subjects.

Conversely, TGF-βR-associated binding protein (TRAP) 1 inhibits TGF-β1 function by interfering with SMAD3 signaling.16 The 5′-TG-3′-interacting factors (TGIFs), TGIF1 and TGIF2, cause repression of TGF-β1-activated genes by direct competition with Smad2.17 In addition, BMP and activin membrane-bound inhibitor (BAMBI) acts as a competitive receptor antagonist for TGF-βRI.18 BAMBI expression is upregulated by TGF-β1 in a feedback loop.19 TGF-β-induced protein (TGFBI, also known as BIG-H3 and keratoepithelin) is an extracellular matrix protein used as a TGF-β1 bioactivity marker.20 The aim of this study is to investigate the expression of TGF-β signaling pathways in the lower airways (bronchial mucosa and peripheral lung) of patients with stable COPD and control subjects. Methods Subjects All patients with COPD and healthy control subjects were recruited from the Respiratory Medicine Unit of the Istituti Clinici Scientifici Maugeri, Veruno, Italy and the Section of Respiratory Diseases of the University Hospital of Ferrara, Italy. Archival material was used in the present study.21 We obtained bronchial biopsy samples from 55 subjects to study immunohistochemically. The characteristics of these subjects are reported in Table 1. Twenty-four subjects undergoing lung resection for a solitary peripheral neoplasm were recruited for the immunohistochemical study of peripheral lung tissue. The characteristics of these subjects are reported in Table 2. COPD and chronic bronchitis were defined according to international guidelines, that is, COPD is the presence of a postbronchodilator FEV1/FVC ratio < 70% and chronic bronchitis is the presence of cough and sputum production for at least 3 months in each of two consecutive years according to Global Initiative for Chronic Obstructive Lung Disease criteria (http://www.goldcopd.org). In patients with COPD, the severity of the airflow obstruction was graded using the 2011 GOLD criteria. The GOLD criteria used here to stratify the severity of stable COPD was based on the degree of airflow obstruction, as symptoms and exacerbation rate data were not collected routinely before 2012. Furthermore, the addition of the symptoms and the number of exacerbations are of unproved value in designing studies on the pathogenesis of COPD.Table 1 Clinical Characteristics of Subjects for Immunohistochemical Studies on Bronchial Biopsy Samples

s and exacerbation rate data were not collected routinely before 2012. Furthermore, the addition of the symptoms and the number of exacerbations are of unproved value in designing studies on the pathogenesis of COPD.Table 1 Clinical Characteristics of Subjects for Immunohistochemical Studies on Bronchial Biopsy Samples Group No. Age (y) Male/Female Pack-years of Smoking Ex/Current Smokers FEV1 (% Predicted) Pre-β2 FEV1 (% Predicted) Post-β2 FEV1/FVC (%) Control nonsmokers 11 67 ± 10 10/1 0 0 116 ± 14 ND 85 ± 10 Control smokers with normal lung function 12 61 ± 7 9/3 43 ± 26 2/10 104 ± 13 ND 81 ± 6 COPD grades I and II (mild/moderate) 14 67 ± 8 12/2 40 ± 19 5/9 66 ± 14a 72 ± 12 60 ± 8a COPD grades III and IV (severe/very severe) 18 66 ± 9 11/7 54 ± 36 13/5 35 ± 8a,b 38 ± 9 44 ± 10a,b Patients with COPD were classified according to Global Initiative for Chronic Obstructive Lung Disease 2011 (http://www.goldcopd.org) grades of severity using only the severity of airflow obstruction. For patients with COPD, FEV1/FVC (%) are postbronchodilator values. ND = not determined; pre-β2 = values obtained before bronchodilator use; post-β2 = values obtained after bronchodilator use. Statistical analysis with analysis of variance test: a P < .0001, which was significantly different from control smokers with normal lung function and control never-smokers. b P < .0001, which was significantly different from mild/moderate COPD. Table 2 Characteristics of Subjects for Immunohistochemical Studies on the Peripheral Lung Tissue

Statistical analysis with analysis of variance test: a P < .0001, which was significantly different from control smokers with normal lung function and control never-smokers. b P < .0001, which was significantly different from mild/moderate COPD. Table 2 Characteristics of Subjects for Immunohistochemical Studies on the Peripheral Lung Tissue Groups No. Age (y) Male/Female Ex/Current Smokers Pack-Years of Smoking Chronic Bronchitis FEV1 (% Predicted) FEV1/FVC (%) Control smokers 12 63.6 ± 3 10/2 6/6 51.3 ± 11.6 No 87.9 ± 4.5 77.4 ± 1.7 Patients with COPD 12 69.9 ± 1.3 12/0 6/6 45.8 ± 6.1 No 68.6 ± 4.2a 58.7 ± 2.5a For COPD and control smoker subjects, FEV1 % predicted and FEV1/FVC % are postbronchodilator values. Data expressed as mean ± SEM. a Analysis of variance = P < .01. All patients with COPD were stable, and none of the subjects with COPD were treated with theophylline, antibiotics, antioxidants, mucolytic agents, or glucocorticoids, or any combination thereof, in the month prior to bronchoscopy or lung resection. The study conformed to the Declaration of Helsinki and was approved by the institutional review boards of Istituti Clinici Scientifici Maugeri (protocol p81) and the University Hospital of Ferrara. Other Methods A detailed description of the lung function and fiberoptic bronchoscopy results; the collection, processing, and immunohistochemical analysis of the bronchial biopsy samples and the peripheral lung (e-Tables 1, 2); and the statistical analysis are provided in e-Appendix 1.

All patients with COPD were stable, and none of the subjects with COPD were treated with theophylline, antibiotics, antioxidants, mucolytic agents, or glucocorticoids, or any combination thereof, in the month prior to bronchoscopy or lung resection. The study conformed to the Declaration of Helsinki and was approved by the institutional review boards of Istituti Clinici Scientifici Maugeri (protocol p81) and the University Hospital of Ferrara. Other Methods A detailed description of the lung function and fiberoptic bronchoscopy results; the collection, processing, and immunohistochemical analysis of the bronchial biopsy samples and the peripheral lung (e-Tables 1, 2); and the statistical analysis are provided in e-Appendix 1. Results Clinical Characteristics of Subjects Providing Bronchial Biopsy and Peripheral Lung Samples We obtained and studied bronchial biopsy samples from 55 subjects. Thirty-two subjects had mild/moderate or severe stable COPD, 12 were current or ex-smokers with normal lung function, and 11 were lifelong nonsmokers with normal lung function (Table 1). In addition, we studied peripheral lung tissue from 24 subjects: 12 smokers with mild/moderate stable COPD and 12 smokers with normal lung function (Table 2). Measurement of the Inflammatory Cells in the Bronchial Lamina Propria of COPD and Control Subjects The data obtained from patients with stable COPD by immunohistochemical analysis confirmed previous results21 showing elevated numbers of CD8+ T cells, CD68+ macrophages, and neutrophils in COPD.

Results Clinical Characteristics of Subjects Providing Bronchial Biopsy and Peripheral Lung Samples We obtained and studied bronchial biopsy samples from 55 subjects. Thirty-two subjects had mild/moderate or severe stable COPD, 12 were current or ex-smokers with normal lung function, and 11 were lifelong nonsmokers with normal lung function (Table 1). In addition, we studied peripheral lung tissue from 24 subjects: 12 smokers with mild/moderate stable COPD and 12 smokers with normal lung function (Table 2). Measurement of the Inflammatory Cells in the Bronchial Lamina Propria of COPD and Control Subjects The data obtained from patients with stable COPD by immunohistochemical analysis confirmed previous results21 showing elevated numbers of CD8+ T cells, CD68+ macrophages, and neutrophils in COPD. Immunohistochemical Results for TGF-β Signaling Pathway Members in the Bronchial Epithelium No differences in the expression of TGF-β1, TGF-β2, TGF-β3, TGF-βRI, TGF-βRII, TGF-βRIII, TGFβ-I/BIGH3, TGIF2, SMAD2, SMAD3, SMAD6, SMAD7, CCN2, LTBP-1, or TRAP-1 were seen in the bronchial epithelia of subjects with COPD compared with nonsmoking control subjects. The number of BAMBI+ immunostained cells was significantly increased in the bronchial epithelia of subjects with COPD compared with control subjects (Fig 1A-E, Table 3).Figure 1 Photomicrographs showing the bronchial mucosa from (A) control nonsmoker, (B) control healthy smoker with normal lung function, (C) mild/moderate stable COPD, (D) severe/very severe stable COPD immunostained for identification of BAMBI+ cells (arrows) in the epithelium and bronchial lamina propria. Results are representative of those from 11 nonsmokers, 12 healthy smokers, 14 subjects with mild/moderate COPD, and 18 subjects with severe/very severe COPD. Graphs indicate median (interquartile range) values of (E) BAMBI scored in the epithelium, BAMBI (score), and (F) BAMBI+ cells quantified in the lamina propria (BAMBI/mm2) of the groups of subjects studied. P values were obtained using the Mann-Whitney test for comparison between groups. BAMBI = bone morphogenetic proteins and activin membrane-bound inhibitor.

e) values of (E) BAMBI scored in the epithelium, BAMBI (score), and (F) BAMBI+ cells quantified in the lamina propria (BAMBI/mm2) of the groups of subjects studied. P values were obtained using the Mann-Whitney test for comparison between groups. BAMBI = bone morphogenetic proteins and activin membrane-bound inhibitor. Table 3 Immunohistochemical Quantification of TGF-β Signaling Pathways in Bronchial Biopsy Samples

e) values of (E) BAMBI scored in the epithelium, BAMBI (score), and (F) BAMBI+ cells quantified in the lamina propria (BAMBI/mm2) of the groups of subjects studied. P values were obtained using the Mann-Whitney test for comparison between groups. BAMBI = bone morphogenetic proteins and activin membrane-bound inhibitor. Table 3 Immunohistochemical Quantification of TGF-β Signaling Pathways in Bronchial Biopsy Samples Target Nonsmokers With Normal Lung Function Smokers With Normal Lung Function Mild/Moderate COPD Severe/Very Severe COPD Kruskal-Wallis P Value Bronchial epithelium score (0-3) TGF-β1 0.25 (0.0-0.75) 0.25 (0.0-0.75) 0.25 (0.0-1.0) 0.37 (0.0-0.75) .945 TGF-β2 0.12 (0.0-1.5) 0.0 (0.0-1.5) 0.0 (0.0-1.5) 0.25 (0.0-2.0) .846 TGF-β3 0.0 (0.0-0.0) 0.0 (0.0-0.0) 0.0 (0.0-0.0) 0.0 (0.0-0.0) n.v. TGF-βRI 0.0 (0.0-0.25) 0.0 (0.0-0.25) 0.0 (0.0-1.0) 0.0 (0.0-0.25) .661 TGF-βRII 0.5 (0.2-2.0) 0.5 (0.0-3.0) 0.5 (0.0-2.0) 1.0 (0.0-2.0) .405 TGFβ-RIII 0.25 (0.0-0.5) 0.25 (0.0-0.75) 0.25 (0.0-1.0) 0.25 (0.0-0.75) .334 TGFBI/BIGH3 0.0 (0.0-0.0) 0.0 (0.0-0.25) 0.0 (0.0-0.0) 0.0 (0.0-0.25) .971 TGIF2 1.5 (1.25-2.0) 1.5 (0.75-1.75) 1.75 (0.75-2.5) 1.5 (1.0-2.0) .185 SMAD2 1.0 (0.0-2.5) 0.5 (0.0-3.0) 0.0 (0.0-2.0) 1.0 (0.0-2.5) .249 SMAD3 1.0 (0.0-2.0) 0.0 (0.0-3.0) 0.5 (0.0-1.5) 1.0 (0.0-2.5) .468 SMAD6 0.50 (0.25-1.0) 0.50 (0.0-1.5) 0.75 (0.0-1.5) 0.25 (0.25-1.25) .296 SMAD7 0.75 (0.0-2.0) 0.5 (0.0-2.5) 0.35 (0.0-2.5) 0.5 (0.0-2.0) .797 CCN2 1.5 (1.0-1.5) 1.5 (1.0-2.5) 1.75 (1.0-3.0) 1.5 (1.0-2.5) .866 LTBP-1 0.25 (0.0-0.25) 0.0 (0.0-0.25) 0.25 (0.0-0.75) 0.25 (0.0-0.5) .263 TRAP-1 0.0 (0.0-0.0) 0.0 (0.0-0.0) 0.0 (0.0-0.0) 0.0 (0.0-0.0) n.v.

-1.5) 0.75 (0.0-1.5) 0.25 (0.25-1.25) .296 SMAD7 0.75 (0.0-2.0) 0.5 (0.0-2.5) 0.35 (0.0-2.5) 0.5 (0.0-2.0) .797 CCN2 1.5 (1.0-1.5) 1.5 (1.0-2.5) 1.75 (1.0-3.0) 1.5 (1.0-2.5) .866 LTBP-1 0.25 (0.0-0.25) 0.0 (0.0-0.25) 0.25 (0.0-0.75) 0.25 (0.0-0.5) .263 TRAP-1 0.0 (0.0-0.0) 0.0 (0.0-0.0) 0.0 (0.0-0.0) 0.0 (0.0-0.0) n.v. BAMBI 0.5 (0.25-1.25) 0.5 (0.25-0.75) 1.0 (0.25-1.75)a 0.75 (0.25-1.5)b .033 Bronchial lamina propria score (cells/mm2) TGF-β1 27.0 (5.0-60.0) 12.5 (0.0-143.0) 16.0 (0.0-64.0) 28.0 (5.0-141.0) .752 TGF-β2 28.0 (14.0-69.0) 13.0 (0.0-37.0) 19.5 (0.0-56.0) 23.0 (0.0-68.0) .211 TGF-β3 0.0 (0.0-13.0) 16.0 (6.0-58.0)a 10.0 (0.0-39.0)a 4.5 (0.0-65.0)b .007 TGF-βRI 5.0 (0.0-52.0) 6.0 (0.0-75.0) 13.0 (0.0-97.0) 5.0 (0.0-73.0) .186 TGF-βRII 74.0 (0.0-225.0) 48.0 (0.0-216.0) 72.5 (8.0-505.0) 45.0 (0.0-376.0) .528 TGF-βRIII 6.0 (4.0-23.0) 6.0 (0.0-32.0) 11.0 (0.0-97.0) 8.0 (5.0-121.0) .612 TGFBI/BIGH3 251.0 (138-484) 304.0 (174-548) 338.0 (218-408) 361.0 (244-468) .142 TGIF2 204 (64-352) 169 (77-277) 177 (77-322) 157 (122-235) .712 SMAD2 181.5 (0.0-750.0) 166.5 (50.0-514.0) 51.0 (0.0-960.0) 91.5 (4.0-627.0) .530 SMAD3 77.0 (0.0-690.0) 87.0 (6.0-754.0) 111.0 (0.0-909.0) 120.0 (0.0-353,0) .822 SMAD6 45.5 (12.0-145.0) 60.5 (12.0-148.0) 83.0 (21.0-134.0) 32.0 (9.0-148.0) .143 SMAD7 67.0 (11.0-300.0) 48.5 (0.0-584.0) 96.0 (0.0-520.0) 56.5 (0.0-620.0) .949 CCN2 147.0 (84.0-210.0) 86 (52.0-234.0) 111.0 (55.0-312.0) 102.0 (48.0-168.0) .768 LTBP-1 3.0 (0.0-13.0) 1.5 (0.0-8.0) 5.0 (0.0-65.0)b 9.5 (0.0-109.0)b .017 TRAP-1 0.0 (0.0-6.0) 0.0 (0.0-5.0) 0.0 (0.0-12.0) 0.0 (0.0-5.0) .243 BAMBI 16.0 (7.0-86.0) 26.0 (8.0-35.0) 53.0 (24.0-364.0)a,b 48.5 (16.0-258.0)a,b .0001 Data are expressed as median (range).

2.0-234.0) 111.0 (55.0-312.0) 102.0 (48.0-168.0) .768 LTBP-1 3.0 (0.0-13.0) 1.5 (0.0-8.0) 5.0 (0.0-65.0)b 9.5 (0.0-109.0)b .017 TRAP-1 0.0 (0.0-6.0) 0.0 (0.0-5.0) 0.0 (0.0-12.0) 0.0 (0.0-5.0) .243 BAMBI 16.0 (7.0-86.0) 26.0 (8.0-35.0) 53.0 (24.0-364.0)a,b 48.5 (16.0-258.0)a,b .0001 Data are expressed as median (range). BAMBI = bone morphogenetic protein and activin membrane-bound inhibitor. CTGF = connective tissue growth factor; LTBP-1 = latent transforming growth factor-β1 binding protein 1; n.v. = no value. SMAD = small mother against decapentaplegic TGFBI = transforming growth factor-β-induced protein; TGIF2 = TGF-β-induced factor 2; TRAP-1 = transforming growth factor-β receptor-associated binding protein. Statistics: The Kruskal-Wallis test was used for multiple comparisons followed by the Mann-Whitney U test for comparison between groups. a P < .05, which was significantly different from control nonsmokers. b P < .05, which was significantly different from control smokers. Immunohistochemical Results for TGF-β Signaling Pathway Members in the Lamina Propria No differences in TGF-β1, TGF-β2, TGF-βRI, TGF-βRII, TGF-βRIII, TGFBI/BIGH3, TGIF2, SMAD2, SMAD3, SMAD6, SMAD7, CCN2, or TRAP-1 expression were observed between groups in the lamina propria (Table 3).

b P < .05, which was significantly different from control smokers. Immunohistochemical Results for TGF-β Signaling Pathway Members in the Lamina Propria No differences in TGF-β1, TGF-β2, TGF-βRI, TGF-βRII, TGF-βRIII, TGFBI/BIGH3, TGIF2, SMAD2, SMAD3, SMAD6, SMAD7, CCN2, or TRAP-1 expression were observed between groups in the lamina propria (Table 3). The number of TGF-β3+ immunostained cells was significantly increased in the bronchial lamina propria of patients with mild/moderate COPD and control smokers with normal lung function compared with nonsmoking control subjects (Table 3). However, subjects with severe/very severe COPD had significantly decreased TGF-β3+ cells compared with control smokers with normal lung function. LTBP-1+ immunostaining (Table 3) and the number of BAMBI+ cells were significantly increased in all severities of COPD compared with control subjects (Fig 1A-F, Table 3). No differences in TGF-β1, TGF-β2, TGF-βRI, TGF-βRII, TGF-βRIII, TGFBI/BIGH3, TGIF2, SMAD2, SMAD3, SMAD6, SMAD7, CCN2, or TRAP-1 expression were observed between groups in the lamina propria.

number of BAMBI+ cells were significantly increased in all severities of COPD compared with control subjects (Fig 1A-F, Table 3). No differences in TGF-β1, TGF-β2, TGF-βRI, TGF-βRII, TGF-βRIII, TGFBI/BIGH3, TGIF2, SMAD2, SMAD3, SMAD6, SMAD7, CCN2, or TRAP-1 expression were observed between groups in the lamina propria. Immunohistochemical Results of the TGF-β Signaling Pathway in the Peripheral Lung At variance with the bronchial epithelium, the percentage of TGF-β1+ (Fig 2A, e-Fig 1), TGF-β3+ (Fig 2E, e-Fig 2), and CCN2+ (Fig 2G, e-Fig 3) bronchiolar epithelial cells in the peripheral airways was significantly decreased in patients with COPD compared with control smokers with normal lung function (Table 4). The percentage of TGF-β1+ alveolar macrophages was decreased in patients with COPD compared with control smokers with normal lung function (Fig 2B, Table 4). There was also a trend for decreased TGF-β3+ and CCN2+ immunostained alveolar macrophages in patients with COPD (Figs 2F, 2H, Table 4). No other significant differences were observed between groups for any other molecules studied (Fig 2, Table 4, e-Fig 4).Figure 2 The percentage of the (A) bronchiolar epithelial and (B) alveolar macrophage cells immunostained for TGF-β1 and (G and H) CCN2 (CTGF). Values for (C and D) TGF-β2 and (E and F) TGF-β3 are shown. Results are representative of those from 12 subjects with stable COPD and 12 control smokers with normal lung function. The Mann-Whitney U test was used for statistical analysis. Exact P values are shown above each graph.

for TGF-β1 and (G and H) CCN2 (CTGF). Values for (C and D) TGF-β2 and (E and F) TGF-β3 are shown. Results are representative of those from 12 subjects with stable COPD and 12 control smokers with normal lung function. The Mann-Whitney U test was used for statistical analysis. Exact P values are shown above each graph. Table 4 Immunohistochemical Quantification of TGF-β Signaling Pathways in the Peripheral Lung

for TGF-β1 and (G and H) CCN2 (CTGF). Values for (C and D) TGF-β2 and (E and F) TGF-β3 are shown. Results are representative of those from 12 subjects with stable COPD and 12 control smokers with normal lung function. The Mann-Whitney U test was used for statistical analysis. Exact P values are shown above each graph. Table 4 Immunohistochemical Quantification of TGF-β Signaling Pathways in the Peripheral Lung Localization and Antigen Control Smokers COPD Mann-Whitney U Test P Value Bronchiolar epithelium (cells, percentage) TGF-β1 95.5 (69.8-98.8) 68.0 (37.0-88.0) .0266 TGF-β2 73.0 (40.8-94.0) 12.5 (2.3-90.3) .1651 TGF-β3 7.0 (2.3-81.8) 2.5 (0.4-5.0) .0321 TGF-β-RI 85.0 (80.0-92.3) 91.5 (84.0-96.5) .1081 TGF-β-RII (bronchiolar smooth muscle) 15.5 (8.3-23.5) 11.0 (0.5-20.3) .4003 TGF-βRIII 0.0 (0.0-0.0) 0.0 (0.0-0.0) NA TGIF2 (nuclear) 3.0 (0.0-21.0) 12.5 (7.5-19.8) .1623 TGIF2 (apical) 60.0 (34.5-85.3) 52.5 (45.5-73.0) .9539 SMAD2 (nuclear) 9.0 (4.0-21.5) 6.0 (1.3-34.8) .7947 SMAD2 (cytosolic) 10.5 (1.5-22.5) 1.0 (0.0-7.8) .0610 SMAD3 5.0 (2.3-30.8) 15.5 (1.0-45.5) .7066 SMAD6 100.0 (99.8-100.0) 100.0 (99.2-100.0) .6148 SMAD7 99.0 (97.3-99.7) 99.0 (93.0-99.0) .5516 TRAP-1 (nuclear) 5.0 (0.3-12.8) 6.0 (0.3-15.5) .8612 TRAP-1 (cytosolic) 0.0 (0.0-0.0) 0.0 (0.0-7.5) .4255 CCN2 100.0 (98.8-100.0) 73.5 (43.8-92.8) .0017 LTBP-1 0.0 (0.0-0.0) 0.0 (0.0-0.0) .8939 BAMBI (nuclear) 55.5 (38.2-75.1) 51.3 (40.0-60.3) .5833 BAMBI (cytosolic) 4.0 (2.1-6.6) 3.5 (1.2-5.8) .5031 Alveolar macrophages (cells, percentage) TGF-β1 98.5 (90.5-99.8) 63.0 (50.0-66.0) .0145 TGF-β2 93.0 (84.0-98.0) 88.5 (55.5-95.0) .2843 TGF-β3 90.5 (61.5-98.0) 61.0 (17.8-79.8) .0528 TGF-βRI 87.5 (34.0-93.5) 85.0 (70.8-97.8) .5435 TGF-βRII 11.0 (0.0-40.5) 29.5 (12.0-47.8) .1737 TGF-βRIII 18.0 (11.0-63.8) 42.5 (20.8-76.0) .2850 TGIF2 (nuclear) 7.5 (2.3-23.0) 7.5 (1.3-15.8) .7501 TGIF2 (cytosolic) 73.5 (45.3-93.3) 73.5 (72.3-81.0) .9769 SMAD2 (nuclear) 0.0 (0.0-0.0) 0.0 (0.0-1.8) .3496 SMAD2 (cytosolic) 63.0 (41.0-82.0) 60.5 (39.3-85.8) .9770 SMAD3 (nuclear) 0.0 (0.0-1.0) 0.0 (0.0-1.0) .9175 SMAD3 (cytosolic) 39.5 (32.0-68.3) 27.0 (10.0-79.8) .4024 SMAD6 100.0 (100.0-100.0) 100.0 (100.0-100.0) NA SMAD7 100.0 (100.0-100.0) 100.0 (100.0-100.0) .3384 TRAP-1 (nuclear) 14.0 (11.0-19.0) 22.5 (1.8-24.0) .1558 TRAP-1 (cytosolic) 79.0 (65.0-82.8) 76.5 (63.0-79.5) .3545 CCN2 98.0 (94.5-100.0) 90.5 (87.8-98.9) .0506 LTBP-1 29.5 (15.8-46.5) 45.0 (29.5-77.3) .1120 BAMBI (nuclear) 53.8 (7.0-66.8) 18.5 (10.8-31.1) .0832 BAMBI (cytosolic) 32.5 (24.2-40.0) 42.5 (26.0-59.5) .2038 Lung vessels (score) TGFBI/BIGH3 1.0 (1.0-1.8) 1.0

558 TRAP-1 (cytosolic) 79.0 (65.0-82.8) 76.5 (63.0-79.5) .3545 CCN2 98.0 (94.5-100.0) 90.5 (87.8-98.9) .0506 LTBP-1 29.5 (15.8-46.5) 45.0 (29.5-77.3) .1120 BAMBI (nuclear) 53.8 (7.0-66.8) 18.5 (10.8-31.1) .0832 BAMBI (cytosolic) 32.5 (24.2-40.0) 42.5 (26.0-59.5) .2038 Lung vessels (score) TGFBI/BIGH3 1.0 (1.0-1.8) 1.0 (1.0-2.0) .3132 SMAD2 1.0 (1.0-2.0) 1.5 (1.0-2.8) .3440 Data expressed as median (range). Statistics: The Mann-Whitney U test was applied for comparison between groups. Note: when not specified, the positive staining is intended to be nuclear, apart from TGFβ-RI, in which it is apical in the bronchiolar epithelium and cytosolic in the alveolar macrophages, and TGF-βRIII and LTBP-1, for which it is cytosolic. NA = not applicable. See Table 3 legend for expansion of other abbreviations.

Statistics: The Mann-Whitney U test was applied for comparison between groups. Note: when not specified, the positive staining is intended to be nuclear, apart from TGFβ-RI, in which it is apical in the bronchiolar epithelium and cytosolic in the alveolar macrophages, and TGF-βRIII and LTBP-1, for which it is cytosolic. NA = not applicable. See Table 3 legend for expansion of other abbreviations. Analysis of Gene Expression Data in Large and Small Airway Epithelial Cells We examined the relative expression of TGF-β1, TGF-β2, TGF-β3, CCN2, LTBP1, and BAMBI mRNA in epithelial cells from the small (GSE11784) and large airways (GSE37147) of patients with COPD compared with control subjects (e-Table 3) using previously published data sets. These data demonstrated a lack of concordance between protein and mRNA for TGF-β pathway members (compare results in e-Table 4 with those in Tables 3 and 4). BAMBI mRNA expression was not different between bronchial epithelial cells from subjects with COPD (n = 30) and healthy smokers (n = 69), although active smoking had a small significant (adjusted P = .03) 1.14-fold increase in BAMBI mRNA (e-Table 3). In contrast, LTBP-1 mRNA expression in subjects with COPD was significantly less than that in nonsmokers (adjusted P = 8.74 × 10–14).

lial cells from subjects with COPD (n = 30) and healthy smokers (n = 69), although active smoking had a small significant (adjusted P = .03) 1.14-fold increase in BAMBI mRNA (e-Table 3). In contrast, LTBP-1 mRNA expression in subjects with COPD was significantly less than that in nonsmokers (adjusted P = 8.74 × 10–14). In contrast to the reduction in TGF-β1, TGF-β3, and CCN2 protein expression in COPD compared with cells from healthy smokers, there was no difference in TGF-β3 mRNA in small airway epithelial cells, but there was an increase in TGF-β1 and CCN2 mRNA expression (e-Table 3). TGF-β1 mRNA expression was significantly increased at 1.22-fold (adjusted P = .00194) and CCN2 was increased by 1.3-fold (adjusted P = .0024) in cells from subjects with COPD (n = 36) vs those from healthy nonsmokers (n = 69).

y epithelial cells, but there was an increase in TGF-β1 and CCN2 mRNA expression (e-Table 3). TGF-β1 mRNA expression was significantly increased at 1.22-fold (adjusted P = .00194) and CCN2 was increased by 1.3-fold (adjusted P = .0024) in cells from subjects with COPD (n = 36) vs those from healthy nonsmokers (n = 69). Correlations Between Clinical Parameters, Inflammatory Cell Counts, and TGF-β Signaling Pathway in Bronchial Biopsy Samples There was a significant correlation between the number of cigarettes smoked (pack-years) and the number of TGF-β3+ immunostained cells/mm2 in the bronchial lamina propria when subjects with or without COPD were grouped together (Fig 3A). This correlation was maintained within the COPD group alone (Fig 3B). The numbers of BAMBI+ immunostained cells/mm2 in the bronchial lamina propria were significantly correlated with numbers of CD8+ cells/mm2 (Fig 3C) and CD68+ cells/mm2 (Figure 3D). No other significant correlations were observed between groups for all the other molecules studied.Figure 3 Regression analysis between pack-years and number of TGF-β3+ cells infiltrating the bronchial lamina propria in (A) all smokers (with and without COPD) and (B) patients with COPD alone. In the latter patients, there is a significant positive correlation between the number of BAMBI+ cells in the bronchial lamina propria and those of (C) CD8+ and (D) CD68+ cells. Correlation coefficients were calculated by using the Spearman rank method. See Figure 1 legend for expansion of abbreviations.

th COPD alone. In the latter patients, there is a significant positive correlation between the number of BAMBI+ cells in the bronchial lamina propria and those of (C) CD8+ and (D) CD68+ cells. Correlation coefficients were calculated by using the Spearman rank method. See Figure 1 legend for expansion of abbreviations. Discussion We report here for the first time, to our knowledge, the comprehensive expression and localization of TGF-β regulatory proteins in the lower airways of patients with stable COPD compared with control subjects. We observed decreased expression of TGF-β1 and TGF-β3 in the bronchiolar but not bronchial epithelium and of TGF-β1 in alveolar macrophages of patients with stable COPD compared with control smokers with normal lung function. Furthermore, TGF-β3 expression was increased in the bronchial lamina propria of control smokers with normal lung function and mild/moderate stable COPD compared with control nonsmokers and correlated significantly with pack-years. These data suggest that the TGF-β signaling is selectively impaired in the small airway epithelia of patients with stable COPD. The expression of the TGF-β pseudoreceptor BAMBI was also elevated in the bronchial mucosa of patients with COPD.

compared with control nonsmokers and correlated significantly with pack-years. These data suggest that the TGF-β signaling is selectively impaired in the small airway epithelia of patients with stable COPD. The expression of the TGF-β pseudoreceptor BAMBI was also elevated in the bronchial mucosa of patients with COPD. Conflicting results have previously been reported regarding the expression of the TGF-β signaling pathway in patients with stable COPD in bronchial biopsy samples and peripheral lung (e-Table 4). Our data for TGF-β1 agree with some previous studies22, 23 but not others.24, 25 The latter studies found increased TGF-β1 expression in the small airway and alveolar epithelial cells, but when the control subjects used for the study are classified according to current GOLD criteria using FEV1/FVC ratios, many would be reclassified as having COPD.24 Another study reported increased TGF-β1 protein for both COPD and control smokers compared with nonsmoking subjects.26 A more recent study found increased release of total TGF-β1 from bronchial epithelium in vitro obtained from patients with COPD compared with control subjects without any significant differences in active TGF-β1 release.27 Finally, another study found no significant differences in BAL TGF-β1 levels between patients with COPD and control subjects.28 These apparent discrepancies may be explained by incomplete clinical features or the selectivity of the primary anti-TGF-β1 antibodies used, or both.29, 30

ferences in active TGF-β1 release.27 Finally, another study found no significant differences in BAL TGF-β1 levels between patients with COPD and control subjects.28 These apparent discrepancies may be explained by incomplete clinical features or the selectivity of the primary anti-TGF-β1 antibodies used, or both.29, 30 Llinàs et al31 observed decreased TGF-β1 mRNA expression in the peripheral lungs of patients with severe stable COPD compared with control nonsmoking subjects, and in agreement with our present data, Kokturk et al32 found no difference in TGF-β1 immunohistochemical expression in the bronchial biopsy samples from patients with stable COPD and control nonsmoking subjects. Vignola et al33 demonstrated increased TGF-β1 immunostaining in the bronchial biopsy samples of patients with chronic bronchitis compared with control young nonsmoking subjects (mean age, 46 years). However, these patients had a mean FEV1 of 56% to 90% predicted, but only nine subjects were considered to have COPD using unspecified criteria.33 Hence, these differences render the comparison of our results with the previous studies difficult.33, 34 Decreased expression of TGF-β1 in the small airways in COPD shown here and its critical role in regulating self-tolerance5 may explain the autoimmunity seen in some patients with COPD.6, 7 The complex signaling control points within the TGF-β activation pathway may enable targeted treatment of this complication.

33, 34 Decreased expression of TGF-β1 in the small airways in COPD shown here and its critical role in regulating self-tolerance5 may explain the autoimmunity seen in some patients with COPD.6, 7 The complex signaling control points within the TGF-β activation pathway may enable targeted treatment of this complication. TGF-β3 expression was increased in the bronchial lamina propria of patients with COPD and control smokers compared with control nonsmoking subjects, although TGF-β3+ cells were decreased in patients with severe/very severe COPD compared with control smokers. There was a significant correlation between pack-years and the number of TGF-β3+ immunostained cells/mm2 in the bronchial lamina propria. To our knowledge, this is the first report of TGF-β3 protein expression and localization in the lower airways of patients with stable COPD and is in keeping with the results of decreased TGF-β3 mRNA expression in the peripheral lung of patients with more severe COPD31, 35 and of increased TGF-β3 mRNA expression in the peripheral lung of smokers compared with nonsmokers.31 This concordance was not observed with our present mRNA results.

ts with stable COPD and is in keeping with the results of decreased TGF-β3 mRNA expression in the peripheral lung of patients with more severe COPD31, 35 and of increased TGF-β3 mRNA expression in the peripheral lung of smokers compared with nonsmokers.31 This concordance was not observed with our present mRNA results. TGF-β3 protein release from bronchial epithelium in vitro and BAL TGF-β3 levels were similar between patients with COPD and control subjects.27, 28 We were unable to find any significant differences between patients with stable COPD and control subjects in TGF-β2 and TGF-βR expression and localization in the lower airways. In addition, we could not confirm the decreased TGF-βRI protein expression observed in the peripheral lungs of patients with moderate stable COPD compared with control subjects in previous studies.23, 31 This discrepancy may be explained by differences in anti-TGF-βRI antibody used. In contrast, our results on TGF-βRII protein expression are in agreement with the data from the same study23and with the mRNA data from Llinàs et al.31 To our knowledge, we have provided the first data on TGF-βRIII protein expression, showing its low immune expression in the lower airways of patients with stable COPD and control subjects. Single nucleotide polymorphisms are likely to be associated with decreased TGF-βRIII function and are linked to an increased risk of pulmonary emphysema developing.36 This area requires further research.

xpression, showing its low immune expression in the lower airways of patients with stable COPD and control subjects. Single nucleotide polymorphisms are likely to be associated with decreased TGF-βRIII function and are linked to an increased risk of pulmonary emphysema developing.36 This area requires further research. We found no significant differences between patients with stable mild/moderate COPD and control subjects (including smokers with normal lung function) in SMAD2, SMAD3, SMAD6, and SMAD7 expression and localization in the lower airways, confirming previous studies.23, 37, 38 In contrast, reduced SMAD6 and SMAD7 mRNA expression was reported in bronchial biopsy samples from patients with stable COPD,39 as well as decreased SMAD7 mRNA in the bronchial epithelium40 and decreased SMAD mRNA expression41 and SMAD protein expression in the peripheral lung.42 These discrepancies may be explained by the lack of concordance between mRNA and protein, as we described in our study comparing gene expression data, or by the potential different pathogenic pathways behind the onset of COPD vs pulmonary emphysema.43, 44, 45 Overall, these data suggest limited involvement of SMAD signaling in the pathogenesis of lower airway inflammation and damage in patients with stable mild/moderate COPD.

our study comparing gene expression data, or by the potential different pathogenic pathways behind the onset of COPD vs pulmonary emphysema.43, 44, 45 Overall, these data suggest limited involvement of SMAD signaling in the pathogenesis of lower airway inflammation and damage in patients with stable mild/moderate COPD. We also demonstrated decreased CCN2 (CTGF) expression in the bronchiolar epithelium but not in the bronchial mucosa of patients with stable COPD compared with control smokers with normal lung function. Conflicting results have previously been reported for CCN2 mRNA expression in patients with COPD.31, 46 TGF-β1 induces CCN2 expression in fibroblasts in vitro,47 suggesting that the decreased expression of both TGF-β1 and CCN2 observed in the bronchiolar epithelium of the patients in our study with stable COPD may be related. Lung specimens from subjects with a solitary peripheral neoplasm were used in the present study, raising the question as to whether lung tumors may influence the results. The presence of similar pathologic conditions in smokers and patients with COPD and the large number of studies already published examining inflammatory markers and cytokine pathways in similar sets of subjects support the use of these groups of patients for peripheral lung studies.

ng tumors may influence the results. The presence of similar pathologic conditions in smokers and patients with COPD and the large number of studies already published examining inflammatory markers and cytokine pathways in similar sets of subjects support the use of these groups of patients for peripheral lung studies. LTBP-1 is the only LTBP that both interacts with latent TGF-βs48 and is predominantly expressed in the lungs.49, 50, 51, 52, 53 LTBP-1 immunostaining was increased in the bronchial lamina propria of patients with COPD compared with control smokers with normal lung function. LTBP-1 may activate TGF-β3, but the concomitant upregulation of TGF-β3 observed in the bronchial lamina propria in our study is smoking related and not disease related, suggesting the presence of alternative mechanisms regulating the expression of these two molecules. In fact, a substantial amount of LTBP-1 can be secreted by cells without being bound to latent TGF-βs, and TGF-β-independent functions for LTBP-1 need to be determined.54

study is smoking related and not disease related, suggesting the presence of alternative mechanisms regulating the expression of these two molecules. In fact, a substantial amount of LTBP-1 can be secreted by cells without being bound to latent TGF-βs, and TGF-β-independent functions for LTBP-1 need to be determined.54 We observed a marked increase in the expression of the TGF-β pseudoreceptor BAMBI in the bronchial mucosa but not in the peripheral airways of patients with stable COPD compared with control subjects in keeping with previous studies in COPD peripheral lung.55 There was a positive correlation between the number of BAMBI+ and CD8+ and CD68+ cells. These data are in keeping with the enhanced plasma BAMBI levels recently described in stable COPD that positively correlated with the blood Th17/regulatory T cells (Treg) ratio.56 In a mouse model of autoimmune arthritis, BAMBI deficiency protected mice against the development of disease by modulating Th17/Treg differentiation.57 This may account for the autoimmune and the Th17/Treg imbalance that we and others have previously described in the bronchial mucosa of patients with stable COPD.6, 7, 58

a mouse model of autoimmune arthritis, BAMBI deficiency protected mice against the development of disease by modulating Th17/Treg differentiation.57 This may account for the autoimmune and the Th17/Treg imbalance that we and others have previously described in the bronchial mucosa of patients with stable COPD.6, 7, 58 The mechanisms regulating BAMBI expression are poorly understood. In vitro, BAMBI expression can be upregulated by TGF-β1.19 However, the discrepancy observed between TGF-β1 and BAMBI expression in the different compartments of the lower airways suggest that non-TGF-β1-dependent pathways could be involved in BAMBI upregulation in stable COPD. As a potential limitation of this study, we did not apply multiple corrections in the statistical analysis of differences between groups for our “ex vivo” data of the lower airways. In fact, applying multiple corrections can lead to false-negative findings, but by not applying them, findings might be false positive. We were confident that applying a multiple test (analysis of variance or Kruskal-Wallis) followed by a “restricted” test analyzing differences between groups could be sufficiently stringent for identification of true differences. In conclusion, the reported differences in TGF-β1 and BAMBI expression may contribute to the pathogenesis of stable COPD creating a microenvironment facilitating local autoimmune responses associated with COPD. Supplementary Data e-Online Data

The mechanisms regulating BAMBI expression are poorly understood. In vitro, BAMBI expression can be upregulated by TGF-β1.19 However, the discrepancy observed between TGF-β1 and BAMBI expression in the different compartments of the lower airways suggest that non-TGF-β1-dependent pathways could be involved in BAMBI upregulation in stable COPD. As a potential limitation of this study, we did not apply multiple corrections in the statistical analysis of differences between groups for our “ex vivo” data of the lower airways. In fact, applying multiple corrections can lead to false-negative findings, but by not applying them, findings might be false positive. We were confident that applying a multiple test (analysis of variance or Kruskal-Wallis) followed by a “restricted” test analyzing differences between groups could be sufficiently stringent for identification of true differences. In conclusion, the reported differences in TGF-β1 and BAMBI expression may contribute to the pathogenesis of stable COPD creating a microenvironment facilitating local autoimmune responses associated with COPD. Supplementary Data e-Online Data Acknowledgments Author contributions: A. D. S., G. C., I. M..A., F. L. M. R., and F. C., contributed to study design, interpretation of data, and writing the manuscript. P. J. B., K. F. C., P. B., P. M., P. R., S. P., Y. G., G. G., and B. B. contributed to critical revision of the manuscript. C. S., I. G., P. C., P. B., M. C., A. D. S., G. C., and A. P. contributed to the production of the data and accuracy of the data analysis.

sign, interpretation of data, and writing the manuscript. P. J. B., K. F. C., P. B., P. M., P. R., S. P., Y. G., G. G., and B. B. contributed to critical revision of the manuscript. C. S., I. G., P. C., P. B., M. C., A. D. S., G. C., and A. P. contributed to the production of the data and accuracy of the data analysis. Financial/nonfinancial disclosures: The authors have reported to CHEST the following: F. L. M. R. has served on the advisory boards of AstraZeneca, Novartis, and Mundipharma. He has served as a consultant for AstraZeneca, Boehringer Ingelheim, Chiesi, Mundipharma, Malesci-Guidotti, Novartis, and Teva and has received grants from AstraZeneca, Boehringer Ingelheim, Chiesi, and GlaxoSmithKline. A. P. reports grants, personal fees, or reimbursement for travel expenses from AstraZeneca, Chiesi, Boehringer Ingelheim, GlaxoSmithKline, Menarini, Merck Sharp & Dohme, Mundipharma, Novartis, Teva, Sanofi, and Zambon. None declared (A. D. S., C. S., I. G., P. C., P. B., M. C., P. M., P. R., G. G., F. C., S. P., Y. G., K. F. C., B. J. B., I. M. A, B. B., G. C.). Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Additional information: The e-Appendix, e-Figures, and e-Tables can be found in the Supplemental Materials section of the online article.

strains affect radiologic manifestation.27, 28Figure 1 Flowchart for the Cayetano Cough Monitor CT scanning study. Radiologic features are based on readings from a US-board-certified radiologist. Cavity volume and distance to the airway are based on results from a computer-automated algorithm. TTP = time to positivity. The Cayetano Cough Monitor (CayeCoM) was used to record participants’ data daily during the first 14 days of treatment and at days 21, 30, and 60. Recordings started at 9:00 am.20 A cough episode included all independent cough events that occurred without a 2-second pause, no cough was a cough frequency ≤ 0.7 cough events per hour, and cough cessation was two consecutive recordings with no cough.5 Sputum was obtained on days 0, 3, 7, 14, 21, and 60 of treatment. Bacillary burden was assessed through time to positivity (TTP) of cultures5, 29, 30 in all MODS culture-positive sputum samples, and culture conversion was defined as the first negative culture with no subsequent positive cultures.5 The study data for cough frequency and bacillary burden has been published.5, 31 A baseline chest CT scan was obtained within 31 days of treatment initiation in all participants enrolled in the study who consented, similar to methods used in a previous TB study in participants who were drug susceptible and HIV negative.28

study data for cough frequency and bacillary burden has been published.5, 31 A baseline chest CT scan was obtained within 31 days of treatment initiation in all participants enrolled in the study who consented, similar to methods used in a previous TB study in participants who were drug susceptible and HIV negative.28 Radiologic Imaging Scans were obtained (Aquilion 64, Toshiba) and analyzed by using a free Digital Imaging and Communications in Medicine viewer. Our computer-automated algorithm detected and measured the volume of the cavitary lesion and its proximity to the airway. A previous algorithm used in small animals32 has been improved in performance for human CT scans by using a more accurate lung segmentation algorithm.33, 34 The validation methods of this higher-resolution algorithm are described in the supplementary methods section of e-Appendix 1. In the case of multiple cavities, we used the cumulative volume of all cavities for analyses.

roved in performance for human CT scans by using a more accurate lung segmentation algorithm.33, 34 The validation methods of this higher-resolution algorithm are described in the supplementary methods section of e-Appendix 1. In the case of multiple cavities, we used the cumulative volume of all cavities for analyses. Fuzzy connectedness methods35 were used to segment the airway in high-resolution CT scans (< 4-mm section thickness). The proximity of the cavitary lesion to the airway was determined using Euclidean distance transform.36, 37, 38 If multiple cavities were present, then the cavity closest to the bronchi was used to determine proximity to the airway. To evaluate other radiologic features, a US board-certified radiologist (P. C.) evaluated each scan to indicate presence or absence of consolidation, cavitation, pneumatocele, atelectasis, fibrosis, bronchiectasis, pericardial effusion, pleural effusion, lymphadenopathy, miliary spread, and pneumothorax. Statistical Analysis Data analysis was performed using software (Stata/SE 14.0, Stata Corp). P values ≤ .05 were considered statistically significant, and data are shown following recommended numeric presentation. Percentages presented as integers, mean difference (MD) is shown to one decimal place, rate ratio (RR) and hazard ratio (HR) are shown based on the rule of four.39, 40

re (Stata/SE 14.0, Stata Corp). P values ≤ .05 were considered statistically significant, and data are shown following recommended numeric presentation. Percentages presented as integers, mean difference (MD) is shown to one decimal place, rate ratio (RR) and hazard ratio (HR) are shown based on the rule of four.39, 40 Cavitary disease was evaluated based on its volume and proximity to the airway according to data from the computer-automated algorithm. We chose 7 mL as the cutoff between a small and a large cavity and 10 mm as the cutoff between a cavity positioned closer to and farther from the airway to the closest edge of the cavity (inner wall). Cutoff analyses showed significance at these values (P < .001 for both) (e-Fig 1). In addition, the presence of bronchiectasis, atelectasis, pleural effusion, and lymphadenopathy were assessed by the radiologist. Other features were too skewed to be compared.

m the airway to the closest edge of the cavity (inner wall). Cutoff analyses showed significance at these values (P < .001 for both) (e-Fig 1). In addition, the presence of bronchiectasis, atelectasis, pleural effusion, and lymphadenopathy were assessed by the radiologist. Other features were too skewed to be compared. We evaluated baseline cavitary lung disease (cavitary volume and proximity from the cavitary lesion to the airway) with pretreatment cough frequency (negative binomial model) and pretreatment TTP (linear regression), adjusting for age and sex, respectively. We also assessed the association between baseline cavitary lung disease and longitudinal cough frequency results during treatment by using a negative binomial model adjusting for age, culture positivity or negativity, sex, treatment day, and treatment day squared, with a random intercept for study participant; covariates were chosen based on previous analyses.5 Baseline cavitary lung disease and longitudinal TTP during treatment were assessed using a linear regression model adjusting for age, cough rate, sex, treatment day, and treatment day squared. A Cox proportional hazards model, unadjusted and adjusted to age and sex, was used to evaluate baseline cavitary lung disease and its effect on cough cessation and culture conversion.

ng treatment were assessed using a linear regression model adjusting for age, cough rate, sex, treatment day, and treatment day squared. A Cox proportional hazards model, unadjusted and adjusted to age and sex, was used to evaluate baseline cavitary lung disease and its effect on cough cessation and culture conversion. In addition, we used the same analyses to evaluate the presence of baseline atelectasis, bronchiectasis, pleural effusion, and lymphadenopathy on cough frequency before and during treatment, TTP, cough cessation, and culture conversion. For all analyses described, given the small sample size and the exploratory nature of these analyses, no correction for multiple comparisons was made. Ethics This study was conducted in accordance with the Declaration of Helsinki.41 This study also was conducted with institutional review board approval by each participating hospital; Universidad Peruana Cayetano Heredia (SIDISI:57183); Asociación Benefica PRISMA in Lima, Peru; and Johns Hopkins University in Baltimore, Maryland (IRB00001676).

ucted in accordance with the Declaration of Helsinki.41 This study also was conducted with institutional review board approval by each participating hospital; Universidad Peruana Cayetano Heredia (SIDISI:57183); Asociación Benefica PRISMA in Lima, Peru; and Johns Hopkins University in Baltimore, Maryland (IRB00001676). Results There were 64 participants with available CT scans, but three scans were of poor image quality (recorded as JPEG format instead of Digital Imaging and Communications in Medicine), and two were incomplete (not enough cross-sectional images) and therefore could not be read. After excluding participants with drug-resistant strains, HIV-positive status, or CT scanning performed after 1 month of treatment, 41 participants were available for analysis. The 41 participants in the study group had a total of 695 recordings, but 37% had to be excluded for technical reasons (e-Table 1). After exclusion, there were 18 participants with pretreatment cough recordings. The median length of recordings was 21 hours. Sixty-eight percent of participants were male, with a median age at enrollment of 30 years (interquartile range, 23-50 years). CT scans were obtained a median of 13 days after treatment initiation (interquartile range, 7-21 days). Demographic and radiologic characteristics of participants are shown in Table 1.Table 1 Baseline Demographic Characteristics of the Study Group

median age at enrollment of 30 years (interquartile range, 23-50 years). CT scans were obtained a median of 13 days after treatment initiation (interquartile range, 7-21 days). Demographic and radiologic characteristics of participants are shown in Table 1.Table 1 Baseline Demographic Characteristics of the Study Group Characteristic Data No. of participants 41 Male participants, % (95% CI) 28 (68%, 53%-83%) Age at study enrollment, median (IQR), y 30 (23-50) Pretreatment culture positive, No. (%, 95% CI) 38 (93%, 84%-100%) Pretreatment TTP, median (IQR), d 6 (6-8) Pretreatment negative auramine smear, No. (%, 95% CI) 13 (32%, 17%-47%) Pretreatment paucibacillary auramine smear,a No. (%, 95% CI) 2 (5%, 0%-12%) Pretreatment auramine smear,b No. (%, 95% CI) 9 (22%, 9%-35%) Pretreatment auramine smear,c No. (%, 95% CI) 6 (15%, 3%-26%) Pretreatment auramine smear,d No. (%, 95% CI) 11 (27%, 13%-41%) Participants who were drug susceptible No. (%, 95% CI) 41 (100%, 100%-100%) Lung volume,e median (IQR), mL 4,700 (4,000-6,000) No cavity,e No. (%, 95% CI) 3 (7%, 0%-16%) Cavity in right lung only,e No. (%, 95% CI) 16 (39%, 23%-55%) Cavity in left lung only,e No. (%, 95% CI) 16 (39%, 23%-55%) Cavity in both lungs,e No. (%, 95% CI) 6 (15%, 3%-26%) Cavity volume,e median (IQR), mL 4 (1-13) Distance to airway,e median (IQR), mm 7 (2-16) IQR = interquartile range; TTP = time to positivity of microscopic observation drug susceptibility culture. a 1 to 19 acid-fast bacilli per 40 fields at ×400 magnification. b 20 to 199 acid-fast bacilli per 40 fields at ×400 magnification.

Characteristic Data No. of participants 41 Male participants, % (95% CI) 28 (68%, 53%-83%) Age at study enrollment, median (IQR), y 30 (23-50) Pretreatment culture positive, No. (%, 95% CI) 38 (93%, 84%-100%) Pretreatment TTP, median (IQR), d 6 (6-8) Pretreatment negative auramine smear, No. (%, 95% CI) 13 (32%, 17%-47%) Pretreatment paucibacillary auramine smear,a No. (%, 95% CI) 2 (5%, 0%-12%) Pretreatment auramine smear,b No. (%, 95% CI) 9 (22%, 9%-35%) Pretreatment auramine smear,c No. (%, 95% CI) 6 (15%, 3%-26%) Pretreatment auramine smear,d No. (%, 95% CI) 11 (27%, 13%-41%) Participants who were drug susceptible No. (%, 95% CI) 41 (100%, 100%-100%) Lung volume,e median (IQR), mL 4,700 (4,000-6,000) No cavity,e No. (%, 95% CI) 3 (7%, 0%-16%) Cavity in right lung only,e No. (%, 95% CI) 16 (39%, 23%-55%) Cavity in left lung only,e No. (%, 95% CI) 16 (39%, 23%-55%) Cavity in both lungs,e No. (%, 95% CI) 6 (15%, 3%-26%) Cavity volume,e median (IQR), mL 4 (1-13) Distance to airway,e median (IQR), mm 7 (2-16) IQR = interquartile range; TTP = time to positivity of microscopic observation drug susceptibility culture. a 1 to 19 acid-fast bacilli per 40 fields at ×400 magnification. b 20 to 199 acid-fast bacilli per 40 fields at ×400 magnification. c 5 to 50 acid-fast bacilli per field at ×400 magnification. d > 50 acid-fast bacilli per field at ×400 magnification.

Characteristic Data No. of participants 41 Male participants, % (95% CI) 28 (68%, 53%-83%) Age at study enrollment, median (IQR), y 30 (23-50) Pretreatment culture positive, No. (%, 95% CI) 38 (93%, 84%-100%) Pretreatment TTP, median (IQR), d 6 (6-8) Pretreatment negative auramine smear, No. (%, 95% CI) 13 (32%, 17%-47%) Pretreatment paucibacillary auramine smear,a No. (%, 95% CI) 2 (5%, 0%-12%) Pretreatment auramine smear,b No. (%, 95% CI) 9 (22%, 9%-35%) Pretreatment auramine smear,c No. (%, 95% CI) 6 (15%, 3%-26%) Pretreatment auramine smear,d No. (%, 95% CI) 11 (27%, 13%-41%) Participants who were drug susceptible No. (%, 95% CI) 41 (100%, 100%-100%) Lung volume,e median (IQR), mL 4,700 (4,000-6,000) No cavity,e No. (%, 95% CI) 3 (7%, 0%-16%) Cavity in right lung only,e No. (%, 95% CI) 16 (39%, 23%-55%) Cavity in left lung only,e No. (%, 95% CI) 16 (39%, 23%-55%) Cavity in both lungs,e No. (%, 95% CI) 6 (15%, 3%-26%) Cavity volume,e median (IQR), mL 4 (1-13) Distance to airway,e median (IQR), mm 7 (2-16) IQR = interquartile range; TTP = time to positivity of microscopic observation drug susceptibility culture. a 1 to 19 acid-fast bacilli per 40 fields at ×400 magnification. b 20 to 199 acid-fast bacilli per 40 fields at ×400 magnification. c 5 to 50 acid-fast bacilli per field at ×400 magnification. d > 50 acid-fast bacilli per field at ×400 magnification. e Presence, location, and volume of a cavitary lesion are based on computer-automated algorithm results of CT scans, which estimated volumes by using the voxel size. Distance from airway to cavitary lesion was calculated only for participants with a CT scan obtained with a section thickness of 4 mm at most on the basis of the computer-automated algorithm results of CT scans, which estimated distances on the basis of Euclidean distance transform.

ted volumes by using the voxel size. Distance from airway to cavitary lesion was calculated only for participants with a CT scan obtained with a section thickness of 4 mm at most on the basis of the computer-automated algorithm results of CT scans, which estimated distances on the basis of Euclidean distance transform. According to the radiologist, the most common findings were cavitary lesions (98%), consolidations (93%), bronchiectasis (68%), atelectasis (29%), lymphadenopathy (20%), and pleural effusion (17%). Only three participants had pneumatocele or pericardial effusion reported, only one participant had fibrosis reported, and miliary spread and pneumothorax were not reported. CT scans with adequate quality were used (n = 41) (Fig 1). In a sensitivity-specificity analysis of cavitation detection, the computer-automated algorithm had a sensitivity of 95% and a specificity of 100% (e-Table 2). The validation of the higher-resolution computer-automated algorithm is shown in the supplementary results (e-Appendix 1, e-Fig 2).

were used (n = 41) (Fig 1). In a sensitivity-specificity analysis of cavitation detection, the computer-automated algorithm had a sensitivity of 95% and a specificity of 100% (e-Table 2). The validation of the higher-resolution computer-automated algorithm is shown in the supplementary results (e-Appendix 1, e-Fig 2). Cough Frequency Associations Baseline cavitary volume and proximity to the airway were not associated with pretreatment cough frequency (e-Tables 3-5). However, results of our multivariable analyses showed that cough frequency during treatment in participants with larger cavities was nearly double that of participants with smaller cavities RR, 1.98; 95% CI, 1.17-3.35; P = .01) (Table 2). Similarly, participants with cavity lesions located farther from the airway had significantly less cough frequency during treatment than did patients with closer proximities (RR, 0.41; 95% CI, 0.248-0.68; P = .001) (Table 3). When we analyzed both cavity volume and distance to the airway, combined, we found that only distance to the airway was significant during treatment (RR, 0.376; 95% CI, 0.196-0.72; P = .003) (Table 4). Older age had a strong trend for more cough frequency during treatment in our models (Table 2, Table 3, Table 4).Table 2 Cavity Volume as Risk Factor for Cough Frequency During Treatment

ined, we found that only distance to the airway was significant during treatment (RR, 0.376; 95% CI, 0.196-0.72; P = .003) (Table 4). Older age had a strong trend for more cough frequency during treatment in our models (Table 2, Table 3, Table 4).Table 2 Cavity Volume as Risk Factor for Cough Frequency During Treatment Risk Factor for Cough Frequency Partially Adjusted Model (N = 41, Obs = 428) Fully Adjusted Model (N = 41, Obs = 188) RR P Value 95% CI RR P Value 95% CI Treatment day 0.90 < .001 0.88-0.93 0.95 < .001 0.92-0.98 Treatment day squared 1.00 < .001 1.00-1.00 1.00 < .001 1.00-1.00 MODS culture positive … … … 1.55 .08 0.94-2.54 Small vs large cavity (categorical) Small cavity (≤ 7 mL) … … … … … … Large cavity (> 7 mL) 1.90 < .001 1.35-2.69 1.98 .01 1.17-3.35 Sex, female … … … 1.31 .3 0.78-2.19 Age per 10 years, y … … … 1.23 .008 1.05-1.42 Cough frequency was used as an outcome in a negative binomial regression to test for risk factors that would increase cough frequency during treatment. In the partially adjusted model, we adjusted for treatment day and treatment day squared. In the fully adjusted model, we adjusted for treatment day, treatment day squared, MODS culture positivity, age, and sex. The volume of the cavity in milliliters was calculated through a computer-automated algorithm that analyzed CT scans on the basis of the voxel size of the cavitary lesion. Participants with no cavities were included in this analysis as having 0-mL volume. MODS = microscopic observation drug susceptibility; Obs = observations; RR = rate ratio.

in milliliters was calculated through a computer-automated algorithm that analyzed CT scans on the basis of the voxel size of the cavitary lesion. Participants with no cavities were included in this analysis as having 0-mL volume. MODS = microscopic observation drug susceptibility; Obs = observations; RR = rate ratio. Table 3 Distance to Airway as a Risk Factor for Cough Frequency During Treatment Risk Factor for Cough Frequency Partially Adjusted Model (N = 33, Obs = 353) Fully Adjusted Model (N = 33, Obs = 154) RR P Value 95% CI RR P Value 95% CI Treatment day 0.91 < .001 0.89-0.93 0.95 .001 0.92-0.98 Treatment day squared 1.00 < .001 1.00-1.00 1.00 .003 1.00-1.00 MODS culture positive … … … 1.47 .1 0.88-2.48 Distance to airway (categorical) Closer distance (≤ 10 mm) … … … … … … Farther distance (> 10 mm) 0.331 < .001 0.236-0.47 0.41 .001 0.248-0.68 Sex, female … … … 0.92 .8 0.55-1.57 Age per 10 years, y … … … 1.20 .06 1.00-1.46 Cough frequency was used as an outcome in a negative binomial regression to test for risk factors that would increase cough frequency during treatment. In the partially adjusted model, we adjusted for treatment day and treatment day squared. In the fully adjusted model, we adjusted for treatment day, treatment day squared, MODS culture positivity, age, and sex. Distance to the airway from the cavitary lesion was calculated through a computer-automated algorithm that analyzed CT scans with high resolution (< 4-mm section thickness) on the basis of Euclidean distance transform. See Table 2 legend for expansion of abbreviations.

ay squared, MODS culture positivity, age, and sex. Distance to the airway from the cavitary lesion was calculated through a computer-automated algorithm that analyzed CT scans with high resolution (< 4-mm section thickness) on the basis of Euclidean distance transform. See Table 2 legend for expansion of abbreviations. Table 4 Combined Risk Factors for Cough Frequency During Treatment

ay squared, MODS culture positivity, age, and sex. Distance to the airway from the cavitary lesion was calculated through a computer-automated algorithm that analyzed CT scans with high resolution (< 4-mm section thickness) on the basis of Euclidean distance transform. See Table 2 legend for expansion of abbreviations. Table 4 Combined Risk Factors for Cough Frequency During Treatment Risk Factor for Cough Frequency Partially Adjusted Model (N = 33, Obs = 353) Fully Adjusted Model (N = 33, Obs = 154) RR P Value 95% CI RR P Value 95% CI Treatment day 0.91 < .001 0.89-0.93 0.95 < .001 0.91-0.98 Treatment day squared 1.00 < .001 1.00-1.00 1.00 .003 1.00-1.00 MODS culture positive … … … 1.47 .1 0.88-2.46 Small vs large cavity (categorical) Small cavity (≤ 7 mL) Ref … … Ref … … Large cavity (> 7 mL) 1.03 .9 0.68-1.55 0.86 .7 0.42-1.76 Distance to airway (categorical) Closer distance (≤ 10 mm) Ref … … Ref … … Farther distance (> 10 mm) 0.336 < .001 0.227-0.50 0.376 .003 0.196-0.72 Sex, female … … … 0.87 .6 0.49-1.56 Age per 10 years, y … … … 1.20 .06 0.99-1.45 Cough frequency was used as an outcome in a negative binomial regression to test for risk factors that would increase cough frequency during treatment. In the partially adjusted model, we adjusted for treatment day and treatment day squared. In the fully adjusted model, we adjusted for treatment day, treatment day squared, MODS culture positivity, age, and sex. The volume of the cavity in milliliters was calculated through a computer-automated algorithm that analyzed CT scans on the basis of the voxel size of the cavitary lesion. Participants with no cavities were included in this analysis as having 0-mL volume. Distance to the airway from the cavitary lesion was calculated through a computer-automated algorithm that analyzed CT scans with high resolution (< 4-mm section thickness) on the basis of Euclidean distance transform. Ref = reference. See Table 2 legend for expansion of other abbreviations.

ysis as having 0-mL volume. Distance to the airway from the cavitary lesion was calculated through a computer-automated algorithm that analyzed CT scans with high resolution (< 4-mm section thickness) on the basis of Euclidean distance transform. Ref = reference. See Table 2 legend for expansion of other abbreviations. There was a nonsignificant trend for association between atelectasis and higher pretreatment cough frequency (RR, 2.71; 95% CI, 0.91-8.1; P = .07). Atelectasis (RR, 1.89; 95% CI, 1.17-3.08; P = .01) and pleural effusion (RR, 1.99; 95% CI, 1.06-3.73; P = .03) were associated with higher cough frequency during treatment (e-Table 6).

nonsignificant trend for association between atelectasis and higher pretreatment cough frequency (RR, 2.71; 95% CI, 0.91-8.1; P = .07). Atelectasis (RR, 1.89; 95% CI, 1.17-3.08; P = .01) and pleural effusion (RR, 1.99; 95% CI, 1.06-3.73; P = .03) were associated with higher cough frequency during treatment (e-Table 6). Bacillary Burden Associations Pretreatment TTP with faster-growing cultures, denoting higher bacillary burden, was associated with larger cavity volumes, in a nonsignificant trend (MD, −1.3; 95% CI, −3.0 to 0.4; P = .1) (e-Table 7), but there was no clear trend with proximity to the airway (e-Table 8). However, when analyzing both, combined, farther distance to the airway showed a nonsignificant trend with slower growing cultures, denoting lower bacillary burden (MD, 1.6; 95% CI, −0.6 to 3.9; P = .1) (e-Table 9). During treatment, we noted an association between larger cavity volumes and faster culture growth, higher bacillary burden, in sputum (MD, −2.4; 95% CI, −4.6 to −0.3; P = .03). Farther distance also was associated with longer time for culture growth, lower bacillary burden, during treatment (MD, 3.3; 95% CI, 1.4-5.2; P = .001). When analyzing both volume and distance, combined, only distance to the airway remained significant during treatment (MD, 2.8; 95% CI, 1.0-4.5; P = .002). Other radiologic features were not associated with bacillary burden before or during treatment.

illary burden, during treatment (MD, 3.3; 95% CI, 1.4-5.2; P = .001). When analyzing both volume and distance, combined, only distance to the airway remained significant during treatment (MD, 2.8; 95% CI, 1.0-4.5; P = .002). Other radiologic features were not associated with bacillary burden before or during treatment. Cough Cessation Assessment Cough cessation tended to be three times faster among participants with smaller cavities than among those with larger cavities, but this finding was not statistically significant (adjusted HR, 2.89; 95% CI, 0.95-8.8; P = .06). The probabilities of cough cessation by day 60, were 69% for small cavities and 31% for large cavities (Fig 2). Furthermore, the hazard for cough cessation was significantly three times higher among participants with cavities located > 10 mm from the airway than among those with cavities located ≤ 10 mm from the airway (adjusted HR, 3.61; 95% CI, 1.26-10.4; P = .02). By day 60, the probabilities for cough cessation were 37% for closer distances and 75% for farther distances (Fig 3). The presence of other radiologic features was not associated with cough cessation.Figure 2 Kaplan-Meier curves for cough cessation and culture conversion by cavity volume size in the study group. Survival curves for cough cessation and microscopic observation drug susceptibility (MODS) culture conversion. Cough cessation represents the time to a cough frequency of ≤ 0.7 cough per hour (considered no cough) for two consecutive recordings. A small cavity is ≤ 7 mL, and a large cavity is > 7 mL on the basis of the results from a computer-automated algorithm. By day 14, the unadjusted probability of cough cessation for small cavities was 58% (95% CI, 40%-77%; adjusted, 97%), whereas for larger cavities this probability was 31% (95% CI, 13%-63%; adjusted, 4%); by day 60, these probabilities were 69% (95% CI, 52%-85%; adjusted, 99%) and 31% (95% CI, 13%-63%; adjusted, 4%), respectively. MODS culture conversion represents time to the first negative culture with no subsequent positive culture. By day 14, the unadjusted probability of culture conversion for small cavities was 37% (95% CI, 22%-58%; adjusted, 32%), whereas for larger cavities this probability was 14% (95% CI, 4%-46%; adjusted, 6%); by day 60, these probabilities were 100% (95% CI, 100%-100%; adjusted, 100%) and 73% (95% CI, 47%-93%; adjusted, 82%), respectively.

djusted probability of culture conversion for small cavities was 37% (95% CI, 22%-58%; adjusted, 32%), whereas for larger cavities this probability was 14% (95% CI, 4%-46%; adjusted, 6%); by day 60, these probabilities were 100% (95% CI, 100%-100%; adjusted, 100%) and 73% (95% CI, 47%-93%; adjusted, 82%), respectively. Figure 3 Kaplan-Meier curves for cough cessation and culture conversion by distance from cavity to airway in the study group. Cough cessation represents the time to a cough frequency of ≤ 0.7 cough per hour (considered no cough) for two consecutive recordings. A closer distance is ≤ 10 mm, and a farther distance is > 10 mm on the basis of the results from a computer-automated algorithm. By day 14, the probability of cough cessation for closer distances was 32% (95% CI, 16%-57%; adjusted, 11%), whereas for farther distances this probability was 65% (95% CI, 45%-84%; adjusted, 94%); by day 60, these probabilities were 37% (95% CI, 20%-63%; adjusted, 13.1%) and 75% (95% CI, 55%-91%; adjusted, 98%), respectively. MODS culture conversion represents time to the first negative culture with no subsequent positive culture. By day 14, the probability of culture conversion for closer distances was 15% (95% CI, 5%-40%; adjusted, 2%), whereas for farther distances this probability was 43% (95% CI, 25%-66%; adjusted, 42%); by day 60, these probabilities were 83% (95% CI, 62%-96%; adjusted, 47%) and 100% (95% CI, 100%-100%; adjusted, 100%), respectively. See Figure 2 legend for expansion of abbreviation.

stances was 15% (95% CI, 5%-40%; adjusted, 2%), whereas for farther distances this probability was 43% (95% CI, 25%-66%; adjusted, 42%); by day 60, these probabilities were 83% (95% CI, 62%-96%; adjusted, 47%) and 100% (95% CI, 100%-100%; adjusted, 100%), respectively. See Figure 2 legend for expansion of abbreviation. Culture Conversion Assessment Culture conversion hazard tended to be two times higher among patients with smaller cavities than among those with larger cavities, but this finding was not statistically significant (adjusted HR, 2.07; 95% CI, 0.90-4.7; P = .09). By day 60, the probabilities of culture conversion are 100% for small cavities and 73% for large cavities (Fig 2). Similarly, those with lesions located farther from the airway tended to have a higher culture conversion hazard but this was not statistically significant (adjusted HR, 2.00; 95% CI, 0.95-4.2; P = .07). Culture conversion probabilities, by day 60, were 83% for closer distances and 100% for farther distances (Fig 3). The presence of other radiologic features was not associated with culture conversion.

er culture conversion hazard but this was not statistically significant (adjusted HR, 2.00; 95% CI, 0.95-4.2; P = .07). Culture conversion probabilities, by day 60, were 83% for closer distances and 100% for farther distances (Fig 3). The presence of other radiologic features was not associated with culture conversion. Discussion Despite the importance of cough in TB transmission, there is a lack of research on this topic,2, 7 and a recent clinical guideline demonstrated that cough duration and cavitary lung disease have not been studied.6 An increase in cough frequency, as well as delayed cough cessation, heightens the theoretical chances for that patient to expel TB aerosols into the air,42, 43 increasing the risk of transmission.44, 45 Our study demonstrated that higher cough frequency during treatment, as well as delayed time to cough cessation, are associated with larger cavitary volume, especially cavities closer to the airway.

theoretical chances for that patient to expel TB aerosols into the air,42, 43 increasing the risk of transmission.44, 45 Our study demonstrated that higher cough frequency during treatment, as well as delayed time to cough cessation, are associated with larger cavitary volume, especially cavities closer to the airway. Patients suspected of having pulmonary TB possibly could be risk stratified for transmission and prognosis within 24 hours through use of the CayeCoM and chest CT scan by using a diagnostic algorithm, based on an underlying mathematical framework,46 in a much shorter time frame compared with that for culture (median culture of MODS is 1 week).47 This risk stratification is particularly important in TB, for which transmission is heterogeneous,9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 especially in certain environments.48 A diagnostic algorithm could determine quickly the most likely contagious patients, as well as identify potential patients who might not respond well to treatment given their increased disease burden.46 However, the most important factor to diminish transmission is effective treatment,5, 49, 50 and other factors (cough strength, sputum viscosity, cough hygiene, social interaction) also would need to be evaluated for this algorithm to build on current scores.51

well to treatment given their increased disease burden.46 However, the most important factor to diminish transmission is effective treatment,5, 49, 50 and other factors (cough strength, sputum viscosity, cough hygiene, social interaction) also would need to be evaluated for this algorithm to build on current scores.51 We observed that a larger cavity volume and a closer proximity to the airway was associated with higher cough frequency during treatment, higher bacillary burden before and during treatment, delayed cough cessation, and culture conversion. Previous studies support the association between larger cavitary volume and higher bacillary burden before treatment,52 as well as a relationship between closer proximity to the airway and higher bacillary burden before treatment.53 When evaluating both volume and proximity, combined, we found that of these two, proximity to the airway seems to play a larger role for both cough frequency during treatment and bacillary burden before and during treatment. The closer the cavity is to the airway, the more inflammation causes increased cough frequency during treatment, and the better oxygen access is provide an optimum microenvironment for Mycobacterium tuberculosis growth.54, 55 Previous studies show that M tuberculosis grows better within the macrophages of the luminal surface of the cavitary lesion because of better oxygen access, coupled with a lack of T lymphocytes, which diminishes the interactions between T cells and macrophages that clear mycobacteria.54, 55

m tuberculosis growth.54, 55 Previous studies show that M tuberculosis grows better within the macrophages of the luminal surface of the cavitary lesion because of better oxygen access, coupled with a lack of T lymphocytes, which diminishes the interactions between T cells and macrophages that clear mycobacteria.54, 55 Patients with more severe infection might have bronchial obstruction that can lead to a lung collapse (atelectasis), which in turn can act as a one-way valve that ultimately increases cough frequency.56, 57 However, pleural effusion is a hypersensitivity reaction that could cause a systemic response resulting in cough, independent of bacillary burden.58, 59, 60 Our study also supports the suggested relationship between radiologic extent of the disease, based on CXR,61 and cough frequency.62

ncreases cough frequency.56, 57 However, pleural effusion is a hypersensitivity reaction that could cause a systemic response resulting in cough, independent of bacillary burden.58, 59, 60 Our study also supports the suggested relationship between radiologic extent of the disease, based on CXR,61 and cough frequency.62 A limitation is that 18 participants had pretreatment recordings, and nearly one-third of recordings had to be excluded due to technical limitations. We did not identify bias when comparing participants with at least 10 excluded recordings with those with fewer than 10 excluded recordings. Chest radiography (CXR) is usually the imaging modality of choice in TB control programs, but CT scans are more sensitive for detecting pleural and parenchymal lesions.63, 64, 65, 66, 67, 68 Nearly all participants had cavities, so we could not evaluate or compare the cough frequency between patients with a cavity and patients without a cavity. The presence of a cavity has not been associated with cough-generated aerosols.49 Given that CXR are obtained in a two-dimensional fashion, it would not have enabled us to evaluate three-dimensional volume and proximity to the airway. A strength of our investigation is that our cough measurements with CayeCoM were validated previously,69, 70, 71 as was the algorithm used to evaluate cavity volume and proximity to the airway.32, 38 Our sample size was similar to those in other CT scanning studies in TB, and the small delay in CT scanning after starting treatment is unlikely to affect results because major changes in cavity structure are uncommon in the first month of treatment.28, 52, 53, 64

aluate cavity volume and proximity to the airway.32, 38 Our sample size was similar to those in other CT scanning studies in TB, and the small delay in CT scanning after starting treatment is unlikely to affect results because major changes in cavity structure are uncommon in the first month of treatment.28, 52, 53, 64 Conclusions To our knowledge, this is the first report regarding an association between cough frequency during treatment and cavitary lung disease. Our study demonstrates an association between cough frequency during treatment, and its duration, with cavitary volume and cavitary proximity to the airway. Younger patients, with small cavitary lesions, especially lesions farther from the airway, may present with minimal cough and sputum samples with low bacillary burden (ie, be smear negative). These patients likely would have cough symptoms later than those with cavities close to the bronchi and, if they are not cultured, may be missed by smear alone. Similarly, if a patient is has a large cavity diagnosed, especially close to the airway, this patient has an increased risk for coughing more during treatment and should be monitored closely for the possibility of expelling more M tuberculosis to the environment. Supplementary Data e-Online Data

Conclusions To our knowledge, this is the first report regarding an association between cough frequency during treatment and cavitary lung disease. Our study demonstrates an association between cough frequency during treatment, and its duration, with cavitary volume and cavitary proximity to the airway. Younger patients, with small cavitary lesions, especially lesions farther from the airway, may present with minimal cough and sputum samples with low bacillary burden (ie, be smear negative). These patients likely would have cough symptoms later than those with cavities close to the bronchi and, if they are not cultured, may be missed by smear alone. Similarly, if a patient is has a large cavity diagnosed, especially close to the airway, this patient has an increased risk for coughing more during treatment and should be monitored closely for the possibility of expelling more M tuberculosis to the environment. Supplementary Data e-Online Data Acknowledgments Author contributions: A. P. and R. H. G. are the guarantors and take responsibility for the content of this manuscript, including the data and analysis. All authors were substantially involved in the study design and drafting the manuscript for intellectual content, and all reviewed the final manuscript before submission. J. W. L. and M. A. B. directly contributed to the study design and were responsible for supervision of data gathering. A. P., D. P. B., G. O. L., and R. H. G. directly contributed to data management and statistical analysis. Z. X. and D. J. M. directly contributed to image analysis.

final manuscript before submission. J. W. L. and M. A. B. directly contributed to the study design and were responsible for supervision of data gathering. A. P., D. P. B., G. O. L., and R. H. G. directly contributed to data management and statistical analysis. Z. X. and D. J. M. directly contributed to image analysis. Financial/nonfinancial disclosures: The authors have reported to CHEST the following: B. H. T. reports consulting on cough as a biomarker for respiratory diseases (not TB) for Pfizer Research. None declared (A. P., D. P. B., J. W. L., N. M. V., M. A. B., G. O. L., Z. X., G. C., E. T., D. J. M., J. S. F., D. A. J. M., C. A. E., P. C., R. H. G.). Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript.

Financial/nonfinancial disclosures: The authors have reported to CHEST the following: B. H. T. reports consulting on cough as a biomarker for respiratory diseases (not TB) for Pfizer Research. None declared (A. P., D. P. B., J. W. L., N. M. V., M. A. B., G. O. L., Z. X., G. C., E. T., D. J. M., J. S. F., D. A. J. M., C. A. E., P. C., R. H. G.). Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. *Collaborators from the Tuberculosis Working Group in Peru: Lilia Cabrera, RN and Marco Varela, BSc (Asociación Benefica PRISMA, Lima, Peru); Francisco Vigil-Romani, MD (Hospital Nacional Cayetano Heredia); Jesus Chacaltana, MD and José L. Cabrera, MD (Hospital Nacional Daniel Alcides Carrión, Lima, Peru); Antonio Salas, MD, Felix Llanos, MD, and Marcos Ñavincopa, MD (Hospital Nacional Dos De Mayo, Lima, Peru); Daniela E. Kirwan, MD and Sumona Datta, MD (Imperial College London, London, England); Jessica D. Rothstein, MSPH (Johns Hopkins University, Baltimore, MD); Nicole A. Doria, MD (George Washington University, Washington, DC); Gustavo Hérnandez-Córdova, MD and Richard Oberhelman, MD (Tulane University, New Orleans, LA); Jorge Coronel, BSc, Luz Caviedes, BSc, and Mirko Zimic, PhD (Universidad Peruana Cayetano Heredia, Lima, Peru); Eyal Oren, PhD (University of Arizona, Tucson, AZ); and nurses from the Peruvian National Tuberculosis Control Program. Data sharing statement: Data from this study are publicly available through the Dryad Digital Repository at https://doi.org/10.5061/dryad.8pt77k0.

*Collaborators from the Tuberculosis Working Group in Peru: Lilia Cabrera, RN and Marco Varela, BSc (Asociación Benefica PRISMA, Lima, Peru); Francisco Vigil-Romani, MD (Hospital Nacional Cayetano Heredia); Jesus Chacaltana, MD and José L. Cabrera, MD (Hospital Nacional Daniel Alcides Carrión, Lima, Peru); Antonio Salas, MD, Felix Llanos, MD, and Marcos Ñavincopa, MD (Hospital Nacional Dos De Mayo, Lima, Peru); Daniela E. Kirwan, MD and Sumona Datta, MD (Imperial College London, London, England); Jessica D. Rothstein, MSPH (Johns Hopkins University, Baltimore, MD); Nicole A. Doria, MD (George Washington University, Washington, DC); Gustavo Hérnandez-Córdova, MD and Richard Oberhelman, MD (Tulane University, New Orleans, LA); Jorge Coronel, BSc, Luz Caviedes, BSc, and Mirko Zimic, PhD (Universidad Peruana Cayetano Heredia, Lima, Peru); Eyal Oren, PhD (University of Arizona, Tucson, AZ); and nurses from the Peruvian National Tuberculosis Control Program. Data sharing statement: Data from this study are publicly available through the Dryad Digital Repository at https://doi.org/10.5061/dryad.8pt77k0. Additional information: The e-Appendix, e-Figures, and e-Tables can be found in the Supplemental Materials section of the online article.

*Collaborators from the Tuberculosis Working Group in Peru: Lilia Cabrera, RN and Marco Varela, BSc (Asociación Benefica PRISMA, Lima, Peru); Francisco Vigil-Romani, MD (Hospital Nacional Cayetano Heredia); Jesus Chacaltana, MD and José L. Cabrera, MD (Hospital Nacional Daniel Alcides Carrión, Lima, Peru); Antonio Salas, MD, Felix Llanos, MD, and Marcos Ñavincopa, MD (Hospital Nacional Dos De Mayo, Lima, Peru); Daniela E. Kirwan, MD and Sumona Datta, MD (Imperial College London, London, England); Jessica D. Rothstein, MSPH (Johns Hopkins University, Baltimore, MD); Nicole A. Doria, MD (George Washington University, Washington, DC); Gustavo Hérnandez-Córdova, MD and Richard Oberhelman, MD (Tulane University, New Orleans, LA); Jorge Coronel, BSc, Luz Caviedes, BSc, and Mirko Zimic, PhD (Universidad Peruana Cayetano Heredia, Lima, Peru); Eyal Oren, PhD (University of Arizona, Tucson, AZ); and nurses from the Peruvian National Tuberculosis Control Program. Data sharing statement: Data from this study are publicly available through the Dryad Digital Repository at https://doi.org/10.5061/dryad.8pt77k0. Additional information: The e-Appendix, e-Figures, and e-Tables can be found in the Supplemental Materials section of the online article. FUNDING/SUPPORT: This study was funded by the National Institute of Allergy and Infectious Diseases [Grant R21AI094143 to R. H. G.] and Fogarty International Center [Grant D43TW006581 to R. H. G.] at the National Institutes of Health and Grand Challenges Canada [Grant 0539-01-10 to G. C.]. This study also was funded by the Center for Infectious Disease Imaging, the intramural research program of the National Institute of Allergy and Infectious Diseases and the National Institute of Biomedical Imaging and Bioengineering from the National Institutes of Health. Contributions by coauthors were funded as follows: Fogarty International Center at the National Institutes of Health [Grants D43TW001140, D43TW010074 to A. P.; Grant R25TW009340 to D. P. B.; Grant R24TW007988 to J. W. L. and M. A. B.; and Grant D43TW009349 to G. O. L. and G. C.]; Grand Challenges Canada [Grant 0537-01-10 to J. W. L.]; Wellcome Trust [Grant 078067/Z/05/Z to D. A. J. M.; and Grants 105788/Z/14/Z and 201251/Z/16/Z to C. A. E.]; Imperial Biomedical Research Centre to J. S. F., and C. A. E.; Joint Global Health Trials [Grant MR/K007467/1 to C. A. E. and R. H. G.]; Stop TB Partnership’s TB REACH initiative funded by the government of Canada [Grant W5_PER_CDT1_PRISMA to C. A. E.] and Bill & Melinda Gates Foundation [Grant OPP1118545 to C. A. E.]; and Innovation for Health and Development funding C. A. E.. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies.

FOR EDITORIAL COMMENT, SEE PAGE 6 ARDS is a type of acute diffuse alveolar damage with an onset within 7 days of known clinical risk factors or new/worsening respiratory symptoms. The hallmarks for ARDS are hypoxemia and bilateral opacities, using either chest radiography or CT scan.1 Globally, ARDS is responsible for 10.4% of all ICU admissions, and approximately 23% of patients with ARDS need mechanical ventilation.2 ARDS is associated with high morbidity and mortality.3, 4 A 2009 systematic review assessing the mortality of ARDS over time demonstrated an overall mortality rate of 44% and 36.2% for observational studies and random controlled trials, respectively, and found that these rates were unchanged since 1994.5 Risk factors for the development of ARDS and for the closely related diagnosis of acute lung injury (ALI), a term also used before definitions of ARDS were standardized in 2012,6 include increased age and clinical factors such as sepsis, pneumonia, aspiration, trauma, pancreatitis, shock, blood transfusions, and smoke or toxic gas inhalation.4, 7, 8, 9 Alcohol abuse has also been reported to increase the risk of ARDS,10, 11 perhaps because acute alcohol intoxication increases the risk of aspiration and pulmonary infection, while chronic alcohol ingestion disturbs both immunologic and nonimmunologic host defense mechanisms within the airway, resulting in alveolar macrophage immune dysregulation and alveolar epithelial barrier dysfunction.12

erhaps because acute alcohol intoxication increases the risk of aspiration and pulmonary infection, while chronic alcohol ingestion disturbs both immunologic and nonimmunologic host defense mechanisms within the airway, resulting in alveolar macrophage immune dysregulation and alveolar epithelial barrier dysfunction.12 To date, however, there remains limited and inconsistent evidence on the relation between alcohol consumption and the risk of ARDS. To synthesize this mixed evidence to estimate an overall magnitude of risk, and to explore whether this varies by predisposing condition for ARDS, we therefore now report a systematic review and meta-analysis of observational studies of the association between alcohol consumption and ARDS. Methods The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)13 and MOOSE (Meta-analysis of Observational Studies in Epidemiology)14 guidelines were used for the conduction of this systematic review and meta-analysis (e-Table 1). The protocol was published in the PROSPERO (International Prospective Register of Systematic Reviews database; registration number CRD42015029910). Study Selection We used the Population-Exposure-Outcome-Study Design criteria throughout the review process, based on type of participants, type of exposure, type of outcome, and study design. Type of Participants All studies of adults aged 18 years and over were eligible for inclusion in this review.

Methods The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)13 and MOOSE (Meta-analysis of Observational Studies in Epidemiology)14 guidelines were used for the conduction of this systematic review and meta-analysis (e-Table 1). The protocol was published in the PROSPERO (International Prospective Register of Systematic Reviews database; registration number CRD42015029910). Study Selection We used the Population-Exposure-Outcome-Study Design criteria throughout the review process, based on type of participants, type of exposure, type of outcome, and study design. Type of Participants All studies of adults aged 18 years and over were eligible for inclusion in this review. Type of Exposure We included all studies that had assessed alcohol consumption, either by self-report or a proxy such as clinical records, defined either as drinking level (low, moderate, heavy, alcohol abuse, alcoholism) or as frequency (grams per day). Type of Outcome The outcome of interest was ARDS. We excluded studies limited to specific clinical diagnoses (HIV, hepatitis B and C viruses). Study Design All the primary comparative observational studies were included (longitudinal/cohort, case control, cross sectional).

Type of Exposure We included all studies that had assessed alcohol consumption, either by self-report or a proxy such as clinical records, defined either as drinking level (low, moderate, heavy, alcohol abuse, alcoholism) or as frequency (grams per day). Type of Outcome The outcome of interest was ARDS. We excluded studies limited to specific clinical diagnoses (HIV, hepatitis B and C viruses). Study Design All the primary comparative observational studies were included (longitudinal/cohort, case control, cross sectional). Search Strategy Medline (via Ovid), EMBASE (via Ovid), and Web of Science were searched independently by two authors from December 1985 to December 2015. Search filters for observational study designs were used,15 and search terms for both outcome and exposure were developed from relevant Cochrane Reviews groups16 (e-Table 2). The search terms using every possible combination were the following: Respiratory Distress Syndrome, Adult/or Adult Respiratory Distress Syndrome/or Acute Lung Injury/or Acute Respiratory Distress Syndrome/or ARDS or ALI. The reference lists were also screened in order to identify additionally eligible studies. There was no language limitation, and where necessary translations of foreign language articles were conducted. In case of duplication the most informative study was used. Two reviewers (E. S., J. L.-B.) independently screened the titles and abstracts. All relevant studies were obtained and the full text was screened independently by two reviewers (E. S., J. L.-B.). Any disagreements were resolved through discussion or with the help of the third reviewer (J. B.).

ormative study was used. Two reviewers (E. S., J. L.-B.) independently screened the titles and abstracts. All relevant studies were obtained and the full text was screened independently by two reviewers (E. S., J. L.-B.). Any disagreements were resolved through discussion or with the help of the third reviewer (J. B.). Data Extraction The data extraction was performed independently by two reviewers, using a previous pilot data extraction form. Variables of interest included author, year of study, study design, definitions of exposure (alcohol) and outcome (ARDS), geographic location, reference population, demographic of study population setting, number of people recruited, and adjustment for confounders.

eviewers, using a previous pilot data extraction form. Variables of interest included author, year of study, study design, definitions of exposure (alcohol) and outcome (ARDS), geographic location, reference population, demographic of study population setting, number of people recruited, and adjustment for confounders. For categorical measures of alcohol drinking, where possible we compared any alcohol vs no alcohol consumption (reference group). When the nonalcohol category was not reported in the studies, the lowest exposed category was used as the reference group. Where exposure to alcohol was reported as quantiles or as categories, we compared the highest exposure groups with lowest exposed group. Also, in the analysis, categorical measures of alcohol consumption were further defined as levels of consumption: light/moderate/heavy drinking; alcohol abuse (including alcoholism). Grams of daily alcohol consumption were used as a standard measure, defining one drink as 0.6 ounce, 14.0 g, or 1.2 tablespoons of pure alcohol.17 According to the Centers for Disease Control and Prevention guidelines, we defined heavy drinking as a weekly consumption of 15 or more drinks for men, and eight or more drinks for women, whereas binge drinking was defined either as five or more drinks during a single occasion for men, and four or more for women. Excessive drinking was defined as the presence of either binge or heavy drinking.17 Moderate alcohol drinking was defined as the daily consumption of up to one drink for women and two drinks for men.18

binge drinking was defined either as five or more drinks during a single occasion for men, and four or more for women. Excessive drinking was defined as the presence of either binge or heavy drinking.17 Moderate alcohol drinking was defined as the daily consumption of up to one drink for women and two drinks for men.18 Assessment of Study Quality The quality of the studies was assessed by the Newcastle-Ottawa Scale.19 High quality was defined as a grade of ≥ 6. Both case-control and cohort studies had a maximum score of 9; whereas cross-sectional studies had a score of 7. The quality assessment was not conducted for articles published as abstracts, due to the lack of information. Two reviewers (E. S., J. L.-B.) independently assessed the quality of the included studies. Discrepancies were resolved through discussion and consensus.

of 9; whereas cross-sectional studies had a score of 7. The quality assessment was not conducted for articles published as abstracts, due to the lack of information. Two reviewers (E. S., J. L.-B.) independently assessed the quality of the included studies. Discrepancies were resolved through discussion and consensus. Statistical Analysis Relative measures of effect were estimated as odds ratios (ORs), relative risks (RRs), or hazard ratios (HRs) with 95% confidence intervals. Results were extracted as either adjusted effect measures, crude measures of effect, or using raw data. We used adjusted estimates in preference. Where more than one adjusted estimate was presented in the paper, we used the estimate that was adjusted for smoking and other socioeconomic factors, where available. For case-control studies we estimated the OR whereas for cohort and cross-sectional studies we estimated the RR. When alcohol exposure was reported either as quantiles or categories, we extracted the effect estimates, taking the highest vs the lowest exposure group. We pooled odds ratios and relative risks together in cases of a rare outcome. Also, studies assessing the effect of definite transfusion-related ALI were analyzed separately and thus not combined in the meta-analysis with other predisposing condition resulting in ALI.

s, taking the highest vs the lowest exposure group. We pooled odds ratios and relative risks together in cases of a rare outcome. Also, studies assessing the effect of definite transfusion-related ALI were analyzed separately and thus not combined in the meta-analysis with other predisposing condition resulting in ALI. Because of the anticipated heterogeneity between the studies, DerSimonian and Laird random-effects models were used to weight each study. The I2 statistic was used to indicate between the studies the percentage of variation due to heterogeneity.20 Subgroup analyses were carried out to explain the identified heterogeneity, based on predisposing condition for ARDS, study design, study quality, year of publication, geographic location, and adjustment for confounders. We used Egger’s statistical test for assessment of publication bias, and a funnel plot for visual assessment. Stata software version 14 (StataCorp) and Review manager software version 5.3 (Cochrane Collaboration) were both used for the statistical analysis. A P value < .05 was thought to represent a statistically significant level.

’s statistical test for assessment of publication bias, and a funnel plot for visual assessment. Stata software version 14 (StataCorp) and Review manager software version 5.3 (Cochrane Collaboration) were both used for the statistical analysis. A P value < .05 was thought to represent a statistically significant level. Results Database searches and reference lists yielded a total of 4,392 articles (Fig 1). After the removal of 739 duplicates we identified 3,653 articles for titles/abstracts screening, from which we identified 200 articles for full text review. Of these, 183 were excluded because the study design was a review or a letter (eight studies); or because there was no comparison group (37 studies); insufficient information on exposure and outcome (13 studies); ineligible outcomes such as sleep apnea, pneumonia, asthma, COPD, airway obstruction, oxygen desaturation index (68 studies); irrelevant exposure (55 studies); or duplicate data (two studies). Thus 17 studies met our criteria for inclusion in the review.Figure 1 Flow chart of studies.

exposure and outcome (13 studies); ineligible outcomes such as sleep apnea, pneumonia, asthma, COPD, airway obstruction, oxygen desaturation index (68 studies); irrelevant exposure (55 studies); or duplicate data (two studies). Thus 17 studies met our criteria for inclusion in the review.Figure 1 Flow chart of studies. Study Characteristics The characteristics of the 17 included studies in the review are shown in Table 1. Twelve studies used a cohort design21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32; four were case-control studies33, 34, 35, 36, and one was a cross-sectional study using survey data.37 A total population of 177,674 people was included. Patients with ARDS had a mean age ranging from 33 to 72.7 years, were more likely to be male (range, 50% to 85%; 13 studies), and the majority were white (range, 50% to 88%; eight studies).Table 1 Characteristics of the Included Studies Study/Year Study Design Country Population/Main Predisposing Condition Characteristics of Patients With ARDS No.

Study Characteristics The characteristics of the 17 included studies in the review are shown in Table 1. Twelve studies used a cohort design21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32; four were case-control studies33, 34, 35, 36, and one was a cross-sectional study using survey data.37 A total population of 177,674 people was included. Patients with ARDS had a mean age ranging from 33 to 72.7 years, were more likely to be male (range, 50% to 85%; 13 studies), and the majority were white (range, 50% to 88%; eight studies).Table 1 Characteristics of the Included Studies Study/Year Study Design Country Population/Main Predisposing Condition Characteristics of Patients With ARDS No. of People Included Alcohol Ascertainment Definition of Exposure to Alcohol Definition Used to Ascertain ARDS Adjustment Afshar et al21/2014 Cohort USA Hospital/Trauma Age: 33 ya Male: 80.6% White: 57.7% 26,305 Blood alcohol content > 0 mg/dL Berlin Adjusted for: age, sex, race, tobacco, diabetes mellitus, immunosuppression medication Ahmed et al33/2014 Nested case control USA Hospital Age: — Male: — White: — 828 … Any use … Matched for: age, sepsis, sex, surgery, ratio of oxygen saturation to fraction of inspired oxygen, and lung injury prediction score Calfee et al22/2011b Cohort USA Hospital/Trauma Age: 44 y Male: 81% White: 66% 144 AUDIT Questionnaire Alcohol abuse AECC No adjustment/matching performed Calfee et al23/2015 Cohort USA Hospital Age: 56 y Male: 53% White: 88% 426 AUDIT Questionnaire Alcohol abuse AECC Adjusted for: log-NNAL, APACHE II scores, race, diabetes, time elapsed between admission and enrollment Cardinal-Fernandez et al24/2013 Cohort Europe Hospital/Sepsis Age: 57 y Male: 71.4% White: — 149 Questionnaire Alcoholism AECC No adjustment/matching performed Gajic et al34/2007b Nested case control USA Hospital/ICU Age: 61 ya Male: 50% White: — 74 Medical records Alcohol abuse AECC Matched for: age, sex, and admission diagnosis Gajic et al25/2011b Cohort USA Hospital Age: 57 ya Male: 65% White: 60% 5,584 Questionnaire Alcohol abuse AECC Adjusted for predisposing conditions, high-risk surgery, high-risk trauma, male sex, body mass index, chemotherapy, diabetes, smoking, emergency surgery, tachypnea, hypoalbuminemia, acidosis, Spo2, Fio2 Ge et al26/2014 Cohort China Hospital/ICU Age: — Male: — White: — 343 Questionnaire Alcohol abuse AECC Adjusted for: age, sex, smoking, use of alcohol, history of diabetes, sepsis, septic shock, trauma, pneumonia, aspiration, massive blood transfusion, bacteremia, pulmonary contusion Iribarren et al27/2000 Cohort USA Hospital Age: 52.8 y Male: 59% White: 73% 121,012 Questionnaire ≥ 3 drinks/d in previous year AECC Adjusted for: age, sex, race, smoking, body mass index, education Iscimen et al28/2008b Cohort Europe Hospital/Septic shock Age: — Male: — White: — 160 Medical records Alcohol abuse … Adjusted for: delay

7/2000 Cohort USA Hospital Age: 52.8 y Male: 59% White: 73% 121,012 Questionnaire ≥ 3 drinks/d in previous year AECC Adjusted for: age, sex, race, smoking, body mass index, education Iscimen et al28/2008b Cohort Europe Hospital/Septic shock Age: — Male: — White: — 160 Medical records Alcohol abuse … Adjusted for: delay ed goal-directed resuscitation, delayed antibiotics, chemotherapy, transfusion, diabetes mellitus Kojicic et al35/2012b Case control USA Hospital/Pneumonia Age: 64.5 ya Male: 50% White: — 596 Medical records Alcohol abuse AECC Matched for: specific pathogen, isolation site, sex, and age Licker et al29/2003b Cohort USA Hospital Age: 67 y Male: — White: — 869 Medical records Alcohol abuse > 60 g/d AECC Adjusted for: pneumonectomy, ventilator hyperpressure index, fluid infused Moss et al30/1996 Cohort USA Hospital/Sepsis, trauma Age: 45.2 y Male: 63% White: 50% 351 Medical records Alcohol abuse AECC Adjusted for: sex, at- risk diagnosis, APACHE II score Moss et al31/2003 Cohort USA Hospital/Septic shock Age: 50.1 y Male: 68% White: — 220 SMAST Questionnaire Alcohol abuse AECC Adjusted for: source of infection, sex, age, chronic hepatic dysfunction, diabetes, severity of illness, nutritional status, and smoking status TenHoor et al37/2001 Cross sectional USA Hospital/Decedents Age: 72.7 y Male: 51% White: 86% 19,003 Interview ≥ 3 drinks/wk Death certificate Adjusted for: sepsis, cirrhosis, medical or surgical misadventure, injury, nonwhite, male, age > 64 y, current smoking/former smoking Thakur et al32/2009 Cohort USA Hospital/ICU Age: 55 y Male: 85% White: — 1,357 Interview > 14 drinks/wk AECC Adjusted for: aspiration, chemotherapy, high-risk surgery, pancreatitis, sepsis, shock, smoking, cirrhosis, and sex Toy et al36/2012b Case control USA Hospital Age: 54 y Male: 49% White: 71% 253 Medical records Alcohol abuse AECC No adjustment/matching performed AECC = American-European Consensus Conference definition; APACHE II = Acute Physiology and Chronic Health Evaluation II; AUDIT = Alcohol Use Disorders Identification Test; Fio2 = fraction of inspired oxygen; log-NNAL = log-transformed NNAL [4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol] level; SMAST = Short Michigan Alcohol Screening Test; Spo2 = oxygen saturation as measured by pulse oximetry.

hysiology and Chronic Health Evaluation II; AUDIT = Alcohol Use Disorders Identification Test; Fio2 = fraction of inspired oxygen; log-NNAL = log-transformed NNAL [4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol] level; SMAST = Short Michigan Alcohol Screening Test; Spo2 = oxygen saturation as measured by pulse oximetry. a Median presented. b Outcome definition used within the study is acute lung injury. All studies were conducted in a hospital setting, with 14 being conducted in the United States, two in Europe,24, 28 and one in China.26 Fourteen studies adjusted for confounders21, 23, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 37 and seven of these had reported results adjusted for smoking. Study quality was assessed using the Newcastle-Ottawa Scale for 15 studies (two studies were published as an abstract only) and of these, eight (53.3%) were found to be of high quality. The median risk of bias score was 6, indicating a medium risk of bias (Table 2). The main reasons for lower scores in risk of bias were as follows: flawed study design (lack of objective/validated methods for exposure definition), selection bias (representativeness of sample population) and information bias (lack of provided information description in outcome assessment), or nonadequacy of follow-up.Table 2 Critical Appraisal of the Included Studies, Using Newcastle-Ottawa Scale

sign (lack of objective/validated methods for exposure definition), selection bias (representativeness of sample population) and information bias (lack of provided information description in outcome assessment), or nonadequacy of follow-up.Table 2 Critical Appraisal of the Included Studies, Using Newcastle-Ottawa Scale Study/Year No. of Stars Selectiona Comparabilityb Exposurec Overall Score Afshar et al21/2014 3 2 3 8 Ahmed et al33/2014d … … … … Calfee et al22/2011 3 0 2 5 Calfee et al23/2015 3 1 2 6 Cardinal-Fernandez et al24/2013 1 0 3 4 Gajic et al34/2007 2 1 1 4 Gajic et al25/2011 2 0 2 4 Ge et al26/2014 2 2 3 7 Iribarren et al27/2000 2 2 2 6 Iscimen et al28/2008d … … … … Kojicic et al35/2012 2 1 1 4 Licker et al29/2003 2 1 3 6 Moss et al31/2003 2 2 3 7 Moss et al30/1996 1 1 2 4 TenHoor et al37/2001 2 2 2 6 Thakur et al32/2009 2 2 2 6 Toy et al36/2012 2 0 1 3 a Maximum, four stars. b Maximum, two stars. c Maximum, three stars. d Only abstract available—not quality assessment.

Study/Year No. of Stars Selectiona Comparabilityb Exposurec Overall Score Afshar et al21/2014 3 2 3 8 Ahmed et al33/2014d … … … … Calfee et al22/2011 3 0 2 5 Calfee et al23/2015 3 1 2 6 Cardinal-Fernandez et al24/2013 1 0 3 4 Gajic et al34/2007 2 1 1 4 Gajic et al25/2011 2 0 2 4 Ge et al26/2014 2 2 3 7 Iribarren et al27/2000 2 2 2 6 Iscimen et al28/2008d … … … … Kojicic et al35/2012 2 1 1 4 Licker et al29/2003 2 1 3 6 Moss et al31/2003 2 2 3 7 Moss et al30/1996 1 1 2 4 TenHoor et al37/2001 2 2 2 6 Thakur et al32/2009 2 2 2 6 Toy et al36/2012 2 0 1 3 a Maximum, four stars. b Maximum, two stars. c Maximum, three stars. d Only abstract available—not quality assessment. Exposure Reporting Sixteen studies investigated the effects of chronic alcohol exposure, and one the effect of acute exposure assessed by blood alcohol levels.21 Most of the studies reported chronic alcohol exposure assessed alcohol by self-report from a questionnaire22, 23, 24, 25, 26, 27, 31 or interview32, 37; six used alcohol consumption documented in medical records28, 29, 30, 34, 35, 36 and in one study the method of assessment and the definition of alcohol consumption were not defined.33 Measures of alcohol consumption included drinks per day,27 drinks per week,32, 37 milligrams of alcohol per deciliter of blood,21 alcoholism,24 and alcohol abuse ascertained either from medical records or questionnaire.22, 23, 25, 26, 28, 29, 30, 31, 34, 35, 36 Specifically, alcohol abuse was defined in three of the 11 studies using a validated questionnaire, two defined alcohol abuse using the AUDIT (Alcohol Use Disorders Identification Test),22, 23 and one using the SMAST (Short Michigan Alcohol Screening Test).31 All studies analyzed the effects of alcohol exposure as a binary measure, contrasting high with low intake, or a history of abuse with no history of abuse, or any alcohol intake with none.

ing the AUDIT (Alcohol Use Disorders Identification Test),22, 23 and one using the SMAST (Short Michigan Alcohol Screening Test).31 All studies analyzed the effects of alcohol exposure as a binary measure, contrasting high with low intake, or a history of abuse with no history of abuse, or any alcohol intake with none. Outcome Reporting Outcome definitions for ARDS included the American-European Consensus Conference definition,22, 23, 24, 25, 26, 27, 29, 30, 31, 32, 34, 35, 36 death certificates,37 and the Berlin definition.21Two studies did not provide clear information on outcome definition.28, 33 Meta-Analysis Thirteen of the studies provided data that could be included in a pooled analysis, which demonstrated that any measure of high exposure to alcohol significantly increased the risk of ARDS by a ratio of 1.89 (95% CI, 1.45-2.48; I2 = 48%) (Fig 2). No evidence of publication bias was found (funnel plot [Fig 3 and Egger’s asymmetry test], P = .150).Figure 2 Forest plot of alcohol consumption and the risk of ARDS; subgroup analysis based on alcohol abuse vs high alcohol consumption. Figure 3 Funnel plot of any high alcohol consumption and the risk of ARDS.

Meta-Analysis Thirteen of the studies provided data that could be included in a pooled analysis, which demonstrated that any measure of high exposure to alcohol significantly increased the risk of ARDS by a ratio of 1.89 (95% CI, 1.45-2.48; I2 = 48%) (Fig 2). No evidence of publication bias was found (funnel plot [Fig 3 and Egger’s asymmetry test], P = .150).Figure 2 Forest plot of alcohol consumption and the risk of ARDS; subgroup analysis based on alcohol abuse vs high alcohol consumption. Figure 3 Funnel plot of any high alcohol consumption and the risk of ARDS. Similar magnitudes of increased risk were seen in sensitivity analyses limited to studies categorizing alcohol intake as alcohol abuse (OR, 1.90; 95% CI, 1.40-2.60; I2 = 56%) (Fig 2), and limited to studies comparing only high alcohol with low or no alcohol consumption (OR, 1.96; 95% CI, 1.07-3.57; I2 = 17%) (Fig 2). However, the only study to use a zero intake as the reference group27 found no significant effect of consuming of ≥ 3 drinks per day during the last year (OR, 0.97; 95% CI, 0.30-3.16). A further sensitivity analysis excluding one study, which compared decedents with a diagnosis of ARDS compared with decedents with other diagnoses,37 had a marginal effect on the magnitude of the association (OR, 1.91; 95% CI, 1.43-2.54; 12 studies) compared with the unrestricted analysis.

OR, 0.97; 95% CI, 0.30-3.16). A further sensitivity analysis excluding one study, which compared decedents with a diagnosis of ARDS compared with decedents with other diagnoses,37 had a marginal effect on the magnitude of the association (OR, 1.91; 95% CI, 1.43-2.54; 12 studies) compared with the unrestricted analysis. Subgroup analysis found that the predisposing condition (trauma, sepsis/septic shock, pneumonia) for ARDS explained heterogeneity between the studies (P value for subgroup differences, .003); where an increased risk of ARDS associated with alcohol consumption was apparent only in patients with sepsis/septic shock (OR, 2.76; 95% CI, 1.80-4.24; four studies) (Fig 4). Further analyses to explore reasons for heterogeneity in the meta-analysis (e-Table 3) showed no statistically significant interaction by study design (case control, longitudinal/cohort, cross sectional; P = .22), study quality (high vs low; P = .09), country of study (United States, Europe, China; P = .19), effect estimate (adjusted vs unadjusted analysis; P = .21), and year of publication (1995-2005 vs 2006-2015; P = .20).Figure 4 Forest plot of alcohol consumption and the risk of ARDS; subgroup analysis in patients with trauma, sepsis, and pneumonia. aData presented for the subset of trauma patients; bData presented for the subset of sepsis patients.

adjusted analysis; P = .21), and year of publication (1995-2005 vs 2006-2015; P = .20).Figure 4 Forest plot of alcohol consumption and the risk of ARDS; subgroup analysis in patients with trauma, sepsis, and pneumonia. aData presented for the subset of trauma patients; bData presented for the subset of sepsis patients. Two studies were identified that assessed the effects of alcohol on the risk of transfusion-related ALI.34, 36 Both studies found that alcohol increased the risk of transfusion-related ALI (results: P = .006 [37% vs 18%]; OR, 3.0; 95% CI, 1.07-8.7). A meta-analysis of these two studies could not be performed as the first study34 did not provide sufficient information to allow ORs to be estimated, due to the study using individual matching to identify the control subjects. Two further studies could not be included in the meta-analysis. The first of these compared risks of ARDS in those with alcohol detected in blood compared with those with no detectable alcohol21; as the effects of acute alcohol intoxication are very different from those of chronic alcohol exposure, this study was not included in the meta-analysis. This study found that the presence of alcohol in blood was associated with an increased risk of ARDS (OR, 1.50). The second study was published only in abstract form,33 which did not provide sufficient information to allow ORs to be estimated, due to the study using individual matching. Briefly, this study showed that patients with ARDS were more likely to consume alcohol (17% vs 10%) compared with control subjects.

S (OR, 1.50). The second study was published only in abstract form,33 which did not provide sufficient information to allow ORs to be estimated, due to the study using individual matching. Briefly, this study showed that patients with ARDS were more likely to consume alcohol (17% vs 10%) compared with control subjects. Discussion This article reports the first meta-analysis of observational studies of the association between alcohol consumption and the risk of ARDS among adults. We found evidence of a 1.89-fold increase in the odds of ARDS in persons with high alcohol consumption, which in subgroup analyses appeared to be attributable to the effect of exposure defined as alcohol abuse and also in those with sepsis or septic shock as the predisposing condition for ARDS. Our review is based on a comprehensive search of the worldwide literature held in key medical databases and using search terms from recognized sources, complemented by searches of reference lists from identified publications. We imposed no language restriction in our searches. It is therefore likely that our results are representative and generalizable. The absence of publication bias further validates our findings.

ases and using search terms from recognized sources, complemented by searches of reference lists from identified publications. We imposed no language restriction in our searches. It is therefore likely that our results are representative and generalizable. The absence of publication bias further validates our findings. Being based largely on observational studies raises the possibility of bias, which may be introduced in our analysis. However, misclassification bias due to the inclusion of former/lower drinkers in the reference group is likely, if anything, to have reduced the magnitudes of estimated effects. However, the subgroup analyses were conducted in an attempt to explore reasons for heterogeneity, and we found that there were no significant differences according to study quality, study design, effect estimate, continent, or year of publication. A previous narrative review has drawn attention to the potential importance of chronic alcohol abuse in the etiology of ARDS,38 finding an increased incidence of ARDS in alcohol abusers. Also, a narrative review published in 2009, which included only four studies on alcohol and ARDS, concluded that alcohol abuse is a risk factor for the development of ARDS.7 Our findings extend the conclusions of this work, identifying a summary effect estimate and that the increased risk applies predominantly to ARDS arising from sepsis.

view published in 2009, which included only four studies on alcohol and ARDS, concluded that alcohol abuse is a risk factor for the development of ARDS.7 Our findings extend the conclusions of this work, identifying a summary effect estimate and that the increased risk applies predominantly to ARDS arising from sepsis. The mechanism or mechanisms by which alcohol consumption might increase the risk of ARDS, particularly among patients with sepsis, are not fully understood. However, effects on membrane permeability,39, 40 glutathione depletion,41, 42, 43 Toll-like receptor up-regulation,44 expression of transforming growth factor-β1,45, 46 and impairment of macrophage function are all potential explanations.47 Our study thus provides comprehensive evidence that high alcohol consumption increases the risk of ARDS. Supplementary Data e-Online Data Acknowledgments Author contributions: E. S. acts as guarantor of the manuscript, and all authors approved the final version of the article to be published. E. S., J. B., and J. L.-B. designed the study and wrote the protocol. E. S. wrote the search strategy and undertook the literature searches, and wrote the draft of the manuscript. E. S. and J. L.-B. undertook study screening, data extraction, and quality assessment. E. S. undertook all data analysis, supervised by J. L.-B. All authors contributed to the interpretation of the findings. J. B. and J. L.-B. provided critical revisions to the article. Financial/nonfinancial disclosure: None declared.

Acknowledgments Author contributions: E. S. acts as guarantor of the manuscript, and all authors approved the final version of the article to be published. E. S., J. B., and J. L.-B. designed the study and wrote the protocol. E. S. wrote the search strategy and undertook the literature searches, and wrote the draft of the manuscript. E. S. and J. L.-B. undertook study screening, data extraction, and quality assessment. E. S. undertook all data analysis, supervised by J. L.-B. All authors contributed to the interpretation of the findings. J. B. and J. L.-B. provided critical revisions to the article. Financial/nonfinancial disclosure: None declared. Other contributions: The author thanks Erica Brasil, Magdalena Opazo-Breton, PhD, and Yue Huang, PhD, from the University of Nottingham for help in translations. Additional information: The e-Tables can be found in the Supplemental Materials section of the online article. FUNDING/SUPPORT: This work was supported by the Medical Research Council [Grant No. MR/K023195/1]; the UK Centre for Tobacco and Alcohol Studies (http://www.ukctas.net); and the British Heart Foundation, Cancer Research UK, the Economic and Social Research Council, and the National Institute of Health Research, under the auspices of the UK Clinical Research Collaboration, is gratefully acknowledged.

Pneumonia remains the most common infectious cause of morbidity and mortality in young children worldwide.1 Most of the 1.1 million deaths in children < 2 years of age occur in resource-limited settings.2 Early and accurate diagnosis of bacterial pneumonia presents a major challenge toward successful treatment. Current international guidelines rely on clinical presentation and physical examination, with imaging used in ambiguous or severe cases.3, 4 A lack of trained physicians and access to diagnostic tools, such as laboratory tests and imaging, make it difficult to follow international guidelines in resource-limited settings.

rrent international guidelines rely on clinical presentation and physical examination, with imaging used in ambiguous or severe cases.3, 4 A lack of trained physicians and access to diagnostic tools, such as laboratory tests and imaging, make it difficult to follow international guidelines in resource-limited settings. The World Health Organization (WHO) developed a pneumonia case management algorithm for resource-limited settings, allowing diagnosis based on symptoms and clinical signs.5 This algorithm has been shown to have low diagnostic specificity.6, 7, 8 Furthermore, no individual clinical features, including those in the WHO case management algorithm, were sufficient to reliably predict radiographically confirmed pneumonia.9 Although chest radiography (CXR) is a standard diagnostic tool for the identification of pneumonia, it has poor validity.10 Without a standardized training approach, such as the WHO CXR methodology, CXR also has high interobserver variability, and current clinical guidelines do not require a CXR for the diagnosis of pneumonia.4 There is evidence that lung auscultation and pulse oximetry improve the ability to correctly identify pneumonia11, 12; however, a recent prospective study had mixed results with pulse oximetry improving diagnosis but not auscultation.13 Lung ultrasound (LUS) has been shown to have good sensitivity and specificity compared with CXR.14, 15, 16, 17

auscultation and pulse oximetry improve the ability to correctly identify pneumonia11, 12; however, a recent prospective study had mixed results with pulse oximetry improving diagnosis but not auscultation.13 Lung ultrasound (LUS) has been shown to have good sensitivity and specificity compared with CXR.14, 15, 16, 17 Previous studies have attempted to develop predictive models in children with suspicion of pneumonia.18, 19, 20, 21, 22 Current algorithms have not been reliable and have been limited by small samples or the exclusion of common pediatric respiratory diseases such asthma or bronchiolitis.21, 22 We sought to assess the diagnostic value of clinical prediction models based on lung auscultation, pulse oximetry, and LUS to identify radiographically confirmed clinical pneumonia in Peruvian children < 5 years of age. This assessment may elucidate the value of implementing these clinical tools where CXR may not be available or appropriate in the diagnosis of pneumonia.

prediction models based on lung auscultation, pulse oximetry, and LUS to identify radiographically confirmed clinical pneumonia in Peruvian children < 5 years of age. This assessment may elucidate the value of implementing these clinical tools where CXR may not be available or appropriate in the diagnosis of pneumonia. Methods Study Design We consecutively enrolled children aged 2 to 59 months presenting to the ED, inpatient wards, and outpatient clinics with an acute respiratory illness at the Instituto Nacional de Salud del Niño in Lima, Peru, between January 2012 and September 2013.23 The Instituto Nacional de Salud del Niño is the largest freestanding pediatric hospital in Lima. It is a public government-run hospital (www.insn.gob.pe) serving predominantly low-income populations, and it is also a national referral center. We excluded children with a history of significant heart disease or chronic respiratory disease other than asthma and children who required invasive airway management. We also recruited 230 children without any acute illness,23 but we limited the use of their data to oxyhemoglobin saturation (Spo2) values in this analysis. This cohort of children was used in a previous LUS validation study,16 and study protocol was published elsewhere.23 The study was approved by the institutional review board committees of the Instituto Nacional de Salud del Niño (Lima, Peru) (No. CL-4311), A.B. PRISMA (Lima, Peru) (No. CE1457.11), and the Johns Hopkins School of Medicine (Baltimore, MD) (No. 64148). Written informed consent was obtained from a parent or guardian prior to enrollment into the study.

nstitutional review board committees of the Instituto Nacional de Salud del Niño (Lima, Peru) (No. CL-4311), A.B. PRISMA (Lima, Peru) (No. CE1457.11), and the Johns Hopkins School of Medicine (Baltimore, MD) (No. 64148). Written informed consent was obtained from a parent or guardian prior to enrollment into the study. Data Collection Child participants who met inclusion criteria underwent a standard clinical assessment, after consent was obtained from their parents, for signs and symptoms, including lung auscultation, pulse oximetry, and imaging. Clinical assessment, including auscultation, was conducted by the treating pediatrician. A study team member recorded the clinical findings, including auscultation findings, vital signs, presenting history, and Spo2.23 All children underwent LUS and had an anterior-posterior CXR taken. Spo2 was assessed using pediatric probes on either Rad 5v pulse oximeters (Masimo Corp) or, in few instances, the peripheral pulse oximeters available at Instituto Nacional de Salud del Niño. Lung auscultation was performed on the anterior and posterior zones of the thorax, with the patient supine or upright, as previously described.23 Pediatricians conducting auscultation were asked to report the presence of the following findings: crackles, wheeze, decreased breath sounds, or bronchial breath sounds. Ausculatory findings were obtained with acoustic stethoscopes.

ior zones of the thorax, with the patient supine or upright, as previously described.23 Pediatricians conducting auscultation were asked to report the presence of the following findings: crackles, wheeze, decreased breath sounds, or bronchial breath sounds. Ausculatory findings were obtained with acoustic stethoscopes. Definitions We used WHO growth standards to define wasting (weight-for-height z score < −2 SD), stunting (height-for-age z score < −2 SD), and severe malnutrition (weight-for-height z score of ≤ −3 SD).24 We used age-specific respiratory rate cutoffs to define tachypnea: ≥ 50 breaths/min for children 2 to 11 months of age and ≥ 40 breaths/min for children 12 to 59 months of age.4 Tachycardia was defined as ≥ 190 beats/min for children 2 to 11 months of age and ≥ 140 beats/min for children 12 to 59 months of age.4 Pulse oximetry was included as a continuous variable, but we conducted sensitivity analyses with Spo2 cutoffs of ≤ 92% and ≤ 95%.

≥ 40 breaths/min for children 12 to 59 months of age.4 Tachycardia was defined as ≥ 190 beats/min for children 2 to 11 months of age and ≥ 140 beats/min for children 12 to 59 months of age.4 Pulse oximetry was included as a continuous variable, but we conducted sensitivity analyses with Spo2 cutoffs of ≤ 92% and ≤ 95%. Pneumonia The definitions of clinical pneumonia, asthma, bronchiolitis, or an upper respiratory tract infection were based on standard of care, using patient history and physical examination, including Spo2 and CXR results. A clinical diagnosis was made by the treating pediatrician. Children had WHO-defined pneumonia if they had an acute presentation of either cough or difficulty breathing and also had either lower chest wall indrawing or age-specific tachypnea.4 Severe pneumonia or very severe disease was defined as WHO pneumonia with at least one of the following danger signs: persistent vomiting, convulsions, lethargy, no oral intake, stridor, or severe malnutrition.25 Severe clinical pneumonia was defined as a clinical diagnosis of pneumonia, by the treating pediatrician, and the presence of at least one of the danger signs previously listed.25

h at least one of the following danger signs: persistent vomiting, convulsions, lethargy, no oral intake, stridor, or severe malnutrition.25 Severe clinical pneumonia was defined as a clinical diagnosis of pneumonia, by the treating pediatrician, and the presence of at least one of the danger signs previously listed.25 LUS Study children received a complete LUS using a MicroMaxx portable ultrasound machine (Sonosite/FujiFilm) with an HFL38/13-6 MHz linear transducer. LUS assessment was conducted by one of three trained general practitioners following a standardized protocol developed using international recommendations.16, 26 Interpretation and conduct of LUS were performed independent of clinical evaluation and CXR findings.16 We defined pneumonia on LUS as the presence of a hypoechoic area consistent with a consolidation and occupying of more than one intercostal space in longitudinal view, or a smaller consolidation with a pleural effusion, and interstitial abnormalities was defined as three or more B lines within a single acoustic window. We required agreement by two of three ultrasound readers for a final LUS diagnosis.16

solidation and occupying of more than one intercostal space in longitudinal view, or a smaller consolidation with a pleural effusion, and interstitial abnormalities was defined as three or more B lines within a single acoustic window. We required agreement by two of three ultrasound readers for a final LUS diagnosis.16 Radiographic Pneumonia We obtained anteroposterior CXR on all children with an acute respiratory illness. Radiographic pneumonia was defined as the presence of a lobar consolidation with or without pleural effusion.27 All chest radiographs were reviewed by two members of a team of three expert pediatric radiologists blinded to clinical information and results from LUS.16 Radiographic diagnosis was made as a consensus of the team using a standardized protocol, as previously described.16

nsolidation with or without pleural effusion.27 All chest radiographs were reviewed by two members of a team of three expert pediatric radiologists blinded to clinical information and results from LUS.16 Radiographic diagnosis was made as a consensus of the team using a standardized protocol, as previously described.16 Biostatistical Methods Our primary objective was to assess the ability of different diagnostic algorithms to correctly classify children diagnosed with clinical pneumonia that is corroborated by the finding of a lobar consolidation on CXR. As such, we compared each additive clinical scenario against radiographically confirmed clinical pneumonia. We evaluated the following four additive scenarios: WHO-defined pneumonia,5 addition of lung auscultation findings, addition of Spo2 by pulse oximetry, followed by the addition of LUS findings. We used multivariable logistic regression to model the presence of radiographically confirmed clinical pneumonia as a function of the four additive scenarios, adjusted for malnutrition (both wasting and stunting), having tachycardia, and a having previous history of pneumonia. We used logistic regression to calculate a concordance statistic (C statistic), which is statistically equivalent to the area under the curve (AUC).28 Models with a higher AUC did better at identifying radiographically confirmed clinical pneumonia. Analyses were performed using STATA version 13 (Stata Corp) and R (The R Foundation).

ic regression to calculate a concordance statistic (C statistic), which is statistically equivalent to the area under the curve (AUC).28 Models with a higher AUC did better at identifying radiographically confirmed clinical pneumonia. Analyses were performed using STATA version 13 (Stata Corp) and R (The R Foundation). Results Participant Characteristics There were 832 children recruited to the study and who underwent diagnostic imaging for pneumonia. Two children were missing clinical data (< 1%) and were excluded from the analysis. We summarized participant characteristics in Table 1. Mean participant age was 21.3 months, 59% of which were boys, 8% were wasted, and 17% were stunted. The study primarily included an inner city population that is low to middle income. We summarized socioeconomic status in Table 1. Overall, final clinical diagnoses, as reported by the treating pediatrician, were as follows: 453 (55%) had clinical pneumonia, 133 (16%) had asthma, 103 (12%) had bronchiolitis, and 143 (17%) had an upper respiratory infection. Radiologists identified 221 consolidations (27%) and 264 interstitial opacities (32%) on chest radiographs in children. A total of 191 children (23%) met criteria for radiographically confirmed clinical pneumonia and 429 children (51.6%) met criteria for WHO-defined pneumonia.Table 1 Demographic Information and Clinical Characteristics According to Study Group

(27%) and 264 interstitial opacities (32%) on chest radiographs in children. A total of 191 children (23%) met criteria for radiographically confirmed clinical pneumonia and 429 children (51.6%) met criteria for WHO-defined pneumonia.Table 1 Demographic Information and Clinical Characteristics According to Study Group Characteristics Full Sample 2-11 mo of Age 12-59 mo of Age Demographic characteristics Sample size 832 (100) 39 (322) 61 (510) Age, mo 21.3 ± 16.2 6.5 ± 2.8 30.6 ± 14.1 No. of boys 59 (488) 60 (193) 58 (295) Social demographics No. of people in household 5.0 ± 0.07 5.3 ± 0.11 4.9 ± 0.09 No. of people in household < 3 3 (22) 2 (5) 3 (17) 3-6 78 (648) 76 (243) 80 (405) 7-10 17 (138) 20 (64) 15 (74) > 10 3 (22) 3 (9) 3 (13) Employment status of parents Both parents employed 23 (193) 13 (43) 30 (150) Only father employed 72 (596) 82 (260) 66 (336) Only mother employed 2 (17) 2 (7) 2 (10) Neither parent employed 2 (20) 3 (9) 2 (11) Father education level, y < 6 2 (17) 2 (7) 2 (10) 6-10 15 (120) 18 (56) 13 (64) 11-12 59 (483) 62 (195) 57 (288) > 12 22 (182) 16 (51) 26 (131) Mother education level, y < 6 3 (25) 4 (14) 2 (11) 6-10 25 (211) 28 (91) 24 (120) 11-12 50 (413) 53 (171) 48 (242) > 12 22 (179) 14 (45) 26 (134) Water supply Home water supply 91 (755) 88 (283) 93 (472) External water supply 9 (77) 12 (39) 7 (38) Toilet waste elimination Connection to city drainage 90 (746) 86 (277) 92 (469) Home septic tank < 1 (1) 0 (0) < 1 (1) Latrine 10 (85) 14 (45) 8 (40) Clinical characteristics Weight-for-height z score 0.26 ± 1.70 0.48 ± 1.68 0.12 ± 1.69 % < −2 SD 8 (69) 7 (21) 9 (48) Height-for-age z score −0.42 ± 1.95 −0.48 ± 1.96 −0.39 ± 1.95 % < −2 SD 17 (140) 19 (61) 15 (79) Symptoms Cough 99 (825) 99 (321) 99 (504) Difficulty breathing 84 (696) 88 (282) 81 (414) Fever 64 (529) 64 (205) 64 (324) Chest indrawing 36 (300) 47 (150) 29 (150) Temperature (°C) 36.8 ± 0.65 36.8 ± 0.60 36.8 ± 0.68 No. with ≥ 38.0°C 9 (74) 8 (26) 9 (48) Heart rate 130 ± 18 135 ± 17 126 ± 18 Tachycardia 15 (126) 0 (0) 25 (126) Respiratory rate 39 ± 12 44 ± 12 36 ± 11 Tachypnea 36 (301) 31 (99) 40 (202) Oxygen saturation 96 ± 3 97 ± 3 96 ± 3 No. ≤ 95% 31 (259) 31 (99) 31 (160) No. ≤ 92% 9 (73) 8 (25) 9 (48) Auscultation findings Wheeze 45 (373) 46 (148) 44 (225) Crackles 53 (445) 57 (183) 51 (262) Decreased breath sounds 12 (98) 8 (27) 14 (71) Values are mean ± SD or %. (No).

6 ± 11 Tachypnea 36 (301) 31 (99) 40 (202) Oxygen saturation 96 ± 3 97 ± 3 96 ± 3 No. ≤ 95% 31 (259) 31 (99) 31 (160) No. ≤ 92% 9 (73) 8 (25) 9 (48) Auscultation findings Wheeze 45 (373) 46 (148) 44 (225) Crackles 53 (445) 57 (183) 51 (262) Decreased breath sounds 12 (98) 8 (27) 14 (71) Values are mean ± SD or %. (No). Distribution of Spo2 by Pneumonia Status We plotted the distribution of Spo2 values by categories of acute respiratory illness, ranging from none to having severe pneumonia, and stratified by CXR findings (Fig 1). Mean Spo2 was lowest in children with a clinical diagnosis of pneumonia, followed by children with either asthma, bronchiolitis, or an upper respiratory tract infection. It was highest among children without an acute illness. Overall, mean Spo2 was lower in children with clinical pneumonia than in those who did not have clinical pneumonia (95.9% vs 97.1%, respectively; P < .001). There was no difference in Spo2 values between children with nonsevere clinical pneumonia and those with severe clinical pneumonia (95.9% vs 95.5%, respectively; P = .78). However, a difference was seen between children with WHO-defined pneumonia and those with WHO-defined severe pneumonia (96.1% vs 95.3%, respectively; P = .02). Additionally, no difference was found between children with radiographically confirmed clinical pneumonia and those with clinical pneumonia without a consolidation (95.9% vs 95.9%, respectively; P = .96).Figure 1 Oxyhemoglobin saturation (Spo2) among children by acute respiratory illness status. Boxplots, from left to right, represent Spo2 values for children without an acute respiratory illness, with an acute respiratory illness that was not pneumonia (asthma, bronchiolitis, or upper respiratory infections), with clinical pneumonia, and with severe clinical pneumonia. The gray dots represent outliers (ie, values that lie > 1½ times the interquartile range). Diamonds/circles and vertical bars to the left of each boxplot represent the mean Spo2 and corresponding 95% CI, respectively, stratified by chest radiography (CXR) findings (diamonds show CXR without a consolidation, circles show CXR with a consolidation).

iers (ie, values that lie > 1½ times the interquartile range). Diamonds/circles and vertical bars to the left of each boxplot represent the mean Spo2 and corresponding 95% CI, respectively, stratified by chest radiography (CXR) findings (diamonds show CXR without a consolidation, circles show CXR with a consolidation). Classification of Radiographically Confirmed Pneumonia We summarized the diagnostic validity for each individual clinical tool (Table 2) and plotted AUCs (Fig 2) and corresponding receiver operating characteristic curves (Fig 3) for the different diagnostic tools based on the described four additive scenarios to classify pneumonia when using radiographically confirmed clinical pneumonia. WHO-defined pneumonia had a 66% sensitivity and 53% specificity for correctly identifying radiographically confirmed clinical pneumonia. The presence of cough or shortness of breath with intercostal indrawing or age-specific tachypnea, crackles on auscultation, and decreased breath sounds on auscultation had sensitivities > 60%, whereas decreased breath sounds on auscultation, Spo2 ≤ 92%, and consolidation on LUS had specificities > 90% (Table 2). Consolidation on LUS had the highest positive likelihood ratio for radiographically confirmed clinical pneumonia, whereas consolidation on LUS and presence of crackles had the lowest negative likelihood ratio. As noted in Figure 2, the use of WHO-defined pneumonia was limited in its ability to classify radiographically confirmed clinical pneumonia. The addition of lung auscultation improved the classification of radiographically confirmed clinical pneumonia, with decreased breath sounds, presence of crackles, and absence of wheezes independently associated (Fig 4). The addition of pulse oximetry to identify hypoxemia as a continuous variable (Fig 2) did not improve the classification of radiographically confirmed clinical pneumonia beyond WHO-defined pneumonia and lung auscultation. Additionally, no improvement in classification is seen using hypoxemia cutoffs of Spo2 ≤ 92% (AUC = 0.73; 95% CI, 0.69-0.77) or ≤ 95% (AUC = 0.73; 95% CI, 0.69-0.77). Both hypoxemia cutoffs, ≤ 92% and ≤ 95%, were not associated with radiographically confirmed clinical pneumonia (Fig 4). Finally, the addition of Spo2 alone to WHO-defined pneumonia did not improve classification (AUC = 0.63; 95% CI, 0.59-0.67).Table 2 Assessment of Diagnostic Validity of Each Clinical Tool

0.69-0.77). Both hypoxemia cutoffs, ≤ 92% and ≤ 95%, were not associated with radiographically confirmed clinical pneumonia (Fig 4). Finally, the addition of Spo2 alone to WHO-defined pneumonia did not improve classification (AUC = 0.63; 95% CI, 0.59-0.67).Table 2 Assessment of Diagnostic Validity of Each Clinical Tool Diagnostic Tools Se % (95% CI) Sp % (95% CI) LR+ (95% CI) LR– (95% CI) PPV % (95% CI) NPV % (95% CI) WHO pneumonia 66.3 (59.1-73.0) 52.7 (48.7-56.6) 1.40 (1.23-1.60) 0.64 (0.52-0.79) 29.4 (25.1-33.9) 84.0 (80.1-87.5) Auscultation findings Presence of crackles 75.3 (68.5-81.2) 53.0 (49.0-56.9) 1.60 (1.43-1.80) 0.47 (0.36-0.60) 32.2 (27.9-36.8) 87.8 (84.1-90.9) Absence of wheezes 64.2 (57.0-71.0) 47.3 (43.4-51.3) 1.22 (1.07-1.39) 0.76 (0.61-0.93) 26.6 (22.6-30.9) 81.7 (77.4-85.5) Decreased breath sounds 23.2 (17.4-29.8) 91.6 (89.1-93.6) 2.74 (1.91-3.95) 0.84 (0.77-0.91) 44.9 (34.8-55.3) 80.1 (77.0-82.9) Oxyhemoglobin saturation ≤ 95% 9.5 (5.7-14.6) 91.4 (89.0-93.5) 1.10 (0.66-1.83) 0.99 (0.94-1.04) 24.7 (15.3-36.1) 77.3 (74.1-80.2) Oxyhemoglobin saturation ≤ 92% 40.0 (33.0-47.3) 71.6 (67.9-75.0) 1.41 (1.14-1.74) 0.84 (0.74-0.95) 29.5 (24.0-35.4) 80.1 (76.6-83.3) Consolidation on lung ultrasound 55.3 (47.9-62.5) 95.0 (93.0-96.6) 11.05 (7.70-15.9) 0.47 (0.40-0.55) 76.6 (68.7-83.4) 87.7 (85.1-90.1) LR+ = positive likelihood ratio, LR– = negative likelihood ratio; NPV = negative predictive value; PPV = positive predictive value; Se = sensitivity; Sp = specificity; WHO = World Health Organization.

lung ultrasound 55.3 (47.9-62.5) 95.0 (93.0-96.6) 11.05 (7.70-15.9) 0.47 (0.40-0.55) 76.6 (68.7-83.4) 87.7 (85.1-90.1) LR+ = positive likelihood ratio, LR– = negative likelihood ratio; NPV = negative predictive value; PPV = positive predictive value; Se = sensitivity; Sp = specificity; WHO = World Health Organization. Figure 2 Area under the curve (AUC) (C statistic) for each of the four additive clinical scenarios used to classify radiographically confirmed clinical pneumonia, stratified by age group. We plotted AUCs and corresponding 95% CIs derived from multivariable logistic regression models for each additive clinical scenario (described in y axis), for all children in the sample (first panel) and stratified by age group: 2 to 11 mo of age (middle panel) and 12 to 59 mo of age (bottom panel). We also present numerical AUCs and 95% CIs for each row. LUS = lung ultrasound; Spo2 = oxyhemoglobin saturation; WHO = World Health Organization. Figure 3 Receiver operating characteristic (ROC) curves for each of the four additive clinical scenarios used to classify radiographically confirmed clinical pneumonia, stratified by age group. We plotted ROC curves derived from multivariable logistic regression models for each additive clinical scenario (described in y axis), for all children in the sample (left panel) and stratified by age group: 2 to 11 mo of age (middle panel) and 12 to 59 mo of age (right panel). See Figure 2 legend for expansion of other abbreviations.

urves derived from multivariable logistic regression models for each additive clinical scenario (described in y axis), for all children in the sample (left panel) and stratified by age group: 2 to 11 mo of age (middle panel) and 12 to 59 mo of age (right panel). See Figure 2 legend for expansion of other abbreviations. Figure 4 Forest plot of the odds of having radiographically confirmed clinical pneumonia using four types of clinical tools. We plotted the adjusted OR of having radiographically confirmed clinical pneumonia for each four clinical tools in an additive scenario using overall study sample, and then stratified by age groups (2-11 and 12-59 mo of age). Adjusted ORs are represented with diamonds and 95% CIs are represented by horizontal lines. We show the additive scenarios on the y axis. Four sets of logistic regression models were built. The first model included a composite for variables to WHO pneumonia, and adjusted for confounders (medical history of pneumonia, age-specific tachycardia, and weight-for-height and height-for-age z scores). The second model included WHO pneumonia, three auscultatory variables (absence of wheezes, presence of crackles, and decreased breath sounds), and vide supra confounders. The third model was a set of models that included WHO pneumonia, the three auscultatory variables, Spo2 expressed three different ways, and vide supra confounders. Specifically, we ran three independent models with Spo2 as a continuous variable, and Spo2 with the thresholds of ≤ 95% and ≤ 92%. The fourth model included WHO pneumonia, the three auscultatory variables, continuous Spo2, two lung ultrasound variables (interstitial abnormalities or consolidation), and vide supra confounders. Adjusted ORs and 95% CIs are also presented numerically for each row. See Figure 2 legend for expansion of abbreviations.

. The fourth model included WHO pneumonia, the three auscultatory variables, continuous Spo2, two lung ultrasound variables (interstitial abnormalities or consolidation), and vide supra confounders. Adjusted ORs and 95% CIs are also presented numerically for each row. See Figure 2 legend for expansion of abbreviations. LUS contributed to the largest improvement in the classification of radiographically confirmed clinical pneumonia. When LUS alone, without auscultation and Spo2, was added to WHO-defined pneumonia, it improved the classification of radiographically confirmed clinical pneumonia (AUC = 0.82; 95% CI, 0.78-0.85). Consolidation on LUS was associated with radiographically confirmed clinical pneumonia (Fig 4). Finding interstitial abnormalities on LUS was indicative of not having radiographically confirmed clinical pneumonia. The addition of Spo2 and LUS, without lung auscultation, improved classification of radiographically confirmed pneumonia beyond clinical signs and symptoms (AUC = 0.82; 95% CI, 0.79-0.86). However, the model that included lung auscultation, with Spo2 and LUS, had better discrimination (AUC = 0.85 vs AUC = 0.82).

monia. The addition of Spo2 and LUS, without lung auscultation, improved classification of radiographically confirmed pneumonia beyond clinical signs and symptoms (AUC = 0.82; 95% CI, 0.79-0.86). However, the model that included lung auscultation, with Spo2 and LUS, had better discrimination (AUC = 0.85 vs AUC = 0.82). Subgroup Analyses In children 12 to 59 months of age, the use of WHO-defined pneumonia resulted in poor classification of radiographically confirmed clinical pneumonia (Fig 2). The addition of lung auscultation improved classification (Fig 2), with crackles, decreased breath sounds, and absence of wheezes independently associated with radiographically confirmed clinical pneumonia (Fig 4). The addition of pulse oximetry to identify hypoxemia did not improve classification (Fig 2) and was not independently associated with radiographically confirmed clinical pneumonia (Fig 4). LUS contributed the largest improvement in the classification of radiographically confirmed clinical pneumonia.

pneumonia (Fig 4). The addition of pulse oximetry to identify hypoxemia did not improve classification (Fig 2) and was not independently associated with radiographically confirmed clinical pneumonia (Fig 4). LUS contributed the largest improvement in the classification of radiographically confirmed clinical pneumonia. In infants 2 to 11 months of age, the use of WHO-defined pneumonia also resulted in poor classification (Fig 2). The addition of lung auscultation improved classification, albeit less so than in children 12 to 59 months of age (Fig 2). When the contributions of lung auscultation were also assessed independently, we found that the presence of crackles and absence of wheezes were not associated with radiographically confirmed clinical pneumonia, whereas decreased breath sounds were associated with radiographically confirmed clinical pneumonia (Fig 4). The addition of pulse oximetry resulted in an improvement in the classification of radiographically confirmed clinical pneumonia (Fig 2). A lower Spo2 was associated with a higher odds of having radiographically confirmed clinical pneumonia (Fig 4). Finally, LUS again was associated with the largest improvement in the classification of radiographically confirmed clinical pneumonia (Fig 2).

ication of radiographically confirmed clinical pneumonia (Fig 2). A lower Spo2 was associated with a higher odds of having radiographically confirmed clinical pneumonia (Fig 4). Finally, LUS again was associated with the largest improvement in the classification of radiographically confirmed clinical pneumonia (Fig 2). Discussion We found that the WHO definition of pneumonia based on clinical symptoms and signs alone had poor discrimination for radiographically confirmed clinical pneumonia among Peruvian children who presented with an acute respiratory illness. Although children with pneumonia had lower Spo2, the use of pulse oximetry to identify hypoxemia did not add value, above WHO-defined pneumonia, to the classification of radiographically confirmed clinical pneumonia in the overall population. In contrast, both the use of auscultation or LUS improved the classification of children with radiographically confirmed clinical pneumonia.

se oximetry to identify hypoxemia did not add value, above WHO-defined pneumonia, to the classification of radiographically confirmed clinical pneumonia in the overall population. In contrast, both the use of auscultation or LUS improved the classification of children with radiographically confirmed clinical pneumonia. Lung auscultation remains an important component of pneumonia diagnosis with more predictive accuracy than an initial clinical assessment alone. In the study population, the presence of crackles, decreased breath sounds, and absence of wheezes were all important predictors for radiographically confirmed clinical pneumonia. This is consistent with other studies where auscultation is a useful predictor of radiographically confirmed pneumonia.29 Presence of physicians, trained personnel, or even a device30 that identifies lung sounds may be critical for increasing the accuracy of a clinical algorithm for the diagnosis of pneumonia in resource-limited settings.

ith other studies where auscultation is a useful predictor of radiographically confirmed pneumonia.29 Presence of physicians, trained personnel, or even a device30 that identifies lung sounds may be critical for increasing the accuracy of a clinical algorithm for the diagnosis of pneumonia in resource-limited settings. The use of pulse oximetry to identify pneumonia is supported by studies showing that a low Spo2 may help to identify more cases of pneumonia than a clinical approach alone.12, 31 Our findings in Peru demonstrated that although Spo2 was significantly lower in children with pneumonia when compared with those who did not have pneumonia, it did not add value to other diagnostic tools to identify radiographically confirmed clinical pneumonia. One possibility that may explain our findings is that our study was conducted in a tertiary referral hospital, where physicians were able to more easily recognize hypoxemia without having to use a pulse oximeter. The sample size may also limit our ability to provide adequate inferences because only 73 children (9%) had an Spo2 ≤ 92%. Our analysis suggests that Spo2 may aid in the diagnosis of pneumonia when assessing infants 2 to 11 months of age, possibly because clinical signs of hypoxemia are more difficult to ascertain in this age group.

e may also limit our ability to provide adequate inferences because only 73 children (9%) had an Spo2 ≤ 92%. Our analysis suggests that Spo2 may aid in the diagnosis of pneumonia when assessing infants 2 to 11 months of age, possibly because clinical signs of hypoxemia are more difficult to ascertain in this age group. Finally, consolidation on LUS, when added to a clinical model, pulse oximetry, and lung auscultation, had the strongest prediction values for radiographically confirmed clinical pneumonia. Children with interstitial opacities alone on LUS were less likely to have radiographically confirmed clinical pneumonia. LUS has recently been proposed as an alternative to CXR because of its high accuracy to diagnose pneumonia in both adults14 and children,15 with pediatric studies confirming these findings in low-resource communities.16 In addition, studies have shown that LUS is a safe alternative to CXR in children with suspected pneumonia.32, 33 Our data support these safety and efficacy trials in the use of LUS as a good predictor of radiographically confirmed clinical pneumonia. Moreover, LUS could be a substitute for CXR in settings that do not have the capability or resources to manage a CXR system.

native to CXR in children with suspected pneumonia.32, 33 Our data support these safety and efficacy trials in the use of LUS as a good predictor of radiographically confirmed clinical pneumonia. Moreover, LUS could be a substitute for CXR in settings that do not have the capability or resources to manage a CXR system. Our study has several strengths. First, we obtained data on clinical signs and symptoms, auscultation, Spo2, and imaging in a large number of children with acute respiratory symptoms. Second, LUS was interpreted by practitioners blinded to clinical or CXR information to avoid potential biases in the interpretation of LUS images. Finally, this study included a variety of acute lower respiratory conditions that could be confused with pneumonia.

large number of children with acute respiratory symptoms. Second, LUS was interpreted by practitioners blinded to clinical or CXR information to avoid potential biases in the interpretation of LUS images. Finally, this study included a variety of acute lower respiratory conditions that could be confused with pneumonia. Our study also has some potential shortcomings. First, the study population was mostly derived from a tertiary referral center. Although children were recruited from outpatient clinics and not all children were referred for respiratory illness, the generalizability of our findings may be limited to children seeking care at tertiary medical centers. Second, we excluded children with chronic lung disease other than asthma and congenital cardiac diseases from the study, further limiting generalizability. Third, we only conducted longitudinal scans when performing LUS. It is possible that the addition of transverse scanning would have resulted in higher diagnostic performance for pneumonia.34 Finally, a gold standard for the diagnosis of pediatric pneumonia is not well defined, and we acknowledge that the interobserver variability in the interpretation of CXR, especially on absence of clinical findings, is high.

ion of transverse scanning would have resulted in higher diagnostic performance for pneumonia.34 Finally, a gold standard for the diagnosis of pediatric pneumonia is not well defined, and we acknowledge that the interobserver variability in the interpretation of CXR, especially on absence of clinical findings, is high. Conclusions Different algorithms, including use of signs and symptoms, laboratory data, and imaging, have been proposed to better diagnose pneumonia in children. Still, there is no consensus of which predictors have the highest yield for discrimination of radiographically confirmed clinical pneumonia, and results may vary depending on population, age, and setting. Our analysis found that lung auscultation and LUS may improve diagnosis of pediatric pneumonia, beyond clinical signs and symptoms. The next steps should be validation studies to assessing utility, ease of use, and feasibility of auscultation and LUS tools in resource-limited settings, and their impact on clinical outcomes.

nalysis found that lung auscultation and LUS may improve diagnosis of pediatric pneumonia, beyond clinical signs and symptoms. The next steps should be validation studies to assessing utility, ease of use, and feasibility of auscultation and LUS tools in resource-limited settings, and their impact on clinical outcomes. Acknowledgments Author contributions: W. C. had ultimate oversight over study conduct, analysis, and interpretation of results. F. P. contributed to design of manuscript, conducted the data analysis and interpretation, and was responsible for writing the manuscript. M. A. C. was responsible for supervision of data gathering and ultrasound, participated in sonographic grading, contributed equally to interpretation of findings, and contributed to writing of the manuscript. L. E. E. contributed to the study design, was responsible for the conduct of the study, was responsible for supervision of data gathering and ultrasound, participated in sonographic grading, contributed to the analysis and interpretation, and contributed to writing of the manuscript. M. G. contributed to data analysis and interpretation and contributed to writing of the manuscript. R. H. G. contributed to study design, contributed to analysis and interpretation, and contributed to writing of the manuscript. C. H. M. contributed to data interpretation and contributed to writing of the manuscript. D. F.-Q. was responsible for the conduct of the study and contributed to writing of the manuscript. P. C.-C. was responsible for radiographic analysis for chest radiographs and contributed to writing of the manuscript. J. M.-C. was responsible for supervision of data gathering and ultrasound, participated in sonographic grading, and contributed to writing of the manuscript. E. D. M. contributed equally to data interpretation and contributed to writing of the manuscript. W. C. conceived the study design, contributed to analysis and interpretation, and was responsible for writing of the manuscript.

ltrasound, participated in sonographic grading, and contributed to writing of the manuscript. E. D. M. contributed equally to data interpretation and contributed to writing of the manuscript. W. C. conceived the study design, contributed to analysis and interpretation, and was responsible for writing of the manuscript. Financial/nonfinancial disclosures: None declared. Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Other contributions: We thank the support provided by Asociacion Benefica PRISMA, Instituto Nacional de Salud del Niño, and collaborators at Johns Hopkins University, Cincinnati Children’s Hospital, and Hospital Nacional Eduardo Rebagliati Martins. FUNDING/SUPPORT: W. C. is supported by the Bill & Melinda Gates Foundation [Grant OPP1117483] and by the National Institutes of Health [Grant 1UM1HL134590]. L. E. E. was supported by the Doris Duke Charitable Foundation Clinical Research Fellowship.

Asthma affects 5% of the adult population and is characterized by symptoms of variable airflow limitation, airway inflammation, and airway hyperresponsiveness. Polarization of the inflammatory response toward increased expression of T helper 2 (Th2) cytokines is considered an important component of the asthma paradigm.1 A major effector axis resulting in induction of Th2 polarization is the recognition of allergen presented by dendritic cells in local lymph nodes to CD4+ T cell. This axis is an optimal target for drug development because it orchestrates inflammation and the development of an allergen-specific humoral response and the development of T- and B-cell memory. The differentiation of naive T cells or reactivation of memory T cells depends on various costimulatory molecules expressed on the T-cell surface, and their cognate ligands.2 One of the most promising costimulatory targets is OX40 and its ligand, OX40L, which has relative T-cell specificity and is not expressed on naive T cells.3

ntiation of naive T cells or reactivation of memory T cells depends on various costimulatory molecules expressed on the T-cell surface, and their cognate ligands.2 One of the most promising costimulatory targets is OX40 and its ligand, OX40L, which has relative T-cell specificity and is not expressed on naive T cells.3 OX40 is primarily induced on T cells in the effector phase and is predominantly expressed by Th2 cells.4 In mouse models of allergic inflammation, OX40 knockout mice have significantly reduced Th2 responses to allergen, accompanied by a reduction in airway eosinophilia and airway hyperreactivity.5 OX40L is expressed mainly by antigen-presenting cells, particularly dendritic cells,6 but also B cells, macrophages, and Langerhans cells.6 In the airway, dendritic cells express OX40L in response to stimulation by epithelial cell-derived thymic stromal lymphopoietin (TSLP).7 In murine and nonhuman primate models of asthma in vivo, blockade of OX40L inhibits TSLP-mediated Th2 inflammation and attenuates the number of OX40L+ dendritic cells in the lung.8 Whether OX40 and OX40L expression are increased in human airway tissue from subjects with asthma is unknown. We hypothesized that OX40 and OX40L expression is increased in the bronchial lamina propria and that this expression is related to disease severity and Th2 cytokine expression. To test our hypothesis, we enumerated OX40, OX40L, interleukin (IL)-4, and IL-4 receptor α (IL4-Rα) expression in bronchial biopsies from subjects with mild, moderate, and severe asthma, compared with healthy controls.

l lamina propria and that this expression is related to disease severity and Th2 cytokine expression. To test our hypothesis, we enumerated OX40, OX40L, interleukin (IL)-4, and IL-4 receptor α (IL4-Rα) expression in bronchial biopsies from subjects with mild, moderate, and severe asthma, compared with healthy controls. Materials and Methods Subjects Twenty-seven patients with asthma and 13 healthy controls were recruited from Glenfield Hospital, Leicester, England. All subjects were nonsmokers with a smoking history of < 10 pack-years, had been free of exacerbations, and were on stable treatment of 8 weeks prior to entry into the study. Asthma was defined by one or more of the following objective criteria: significant bronchodilator reversibility of > 200 mL, a provocation concentration of methacholine causing a 20% fall in FEV1 < 8 mg/mL, and/or a peak flow amplitude % mean over 2 weeks of > 20%. Asthma severity was defined according to the Global Initiative for Asthma (GINA) treatment steps.9 Normal subjects had no history of respiratory disease and normal spirometry. The study was approved by the Leicestershire Research Ethics Committee and informed consent was obtained from all subjects.

er 2 weeks of > 20%. Asthma severity was defined according to the Global Initiative for Asthma (GINA) treatment steps.9 Normal subjects had no history of respiratory disease and normal spirometry. The study was approved by the Leicestershire Research Ethics Committee and informed consent was obtained from all subjects. Protocol and Clinical Characterization Subjects attended on two occasions. At the first visit, they underwent spirometry; allergen skin prick tests for Dermatophagoides pteronyssinus, dog, cat and grass pollen; a methacholine inhalation test;10 and sputum induction.11 At the second visit 1 week later, subjects underwent bronchoscopy.12 Mucosal biopsy specimens were processed into the water-soluble resin glycol methacrylate (Polysciences; Northampton, England) for embedding.13

hagoides pteronyssinus, dog, cat and grass pollen; a methacholine inhalation test;10 and sputum induction.11 At the second visit 1 week later, subjects underwent bronchoscopy.12 Mucosal biopsy specimens were processed into the water-soluble resin glycol methacrylate (Polysciences; Northampton, England) for embedding.13 Immunohistochemistry Two-micrometer sections were cut and immunostained with the following mouse monoclonal antibodies: OX40 and OX40L (BD Biosciences; Oxford, England) and IL-4 (clones 4D9 and 3H4; AMS Biotechnology; Oxford, England) and IL-4Rα (R&D Systems; Abingdon, Oxfordshire, England) in quadruplicate; α-smooth muscle actin (Dako; Cambridge, England); and major basic protein (Caltag; Paiseley, England) with appropriate isotype controls (Dako). The number of positively stained nucleated cells was enumerated per square millimeter of the lamina propria or airway smooth muscle (ASM) by a blinded observer. A minimum ASM area of 0.1 mm2 was considered assessable, as described previously.14 The number of subjects with adequate smooth muscle was 10/13 controls, 10/10 GINA 1 asthma, 6/7 GINA 2-3 asthma, and 8/10 GINA 4-5 asthma. IL-4Rα expression by the ASM was also assessed using a semiquantitative intensity score of no staining = 0, low = 1, moderate = 2, and high = 3. Post hoc in three subjects with asthma, sequential sections were stained for OX40 and CD3 (T cells), and OX40L and CD1a (dendritic cells). The proportion of the OX40 or OX40L+ cells that were colocalized to T cells or dendritic cells, respectively, was determined as previously described.15

, moderate = 2, and high = 3. Post hoc in three subjects with asthma, sequential sections were stained for OX40 and CD3 (T cells), and OX40L and CD1a (dendritic cells). The proportion of the OX40 or OX40L+ cells that were colocalized to T cells or dendritic cells, respectively, was determined as previously described.15 Analysis Statistical analysis was performed using PRISM software, version 4 (GraphPad Software; La Jolla, CA). Parametric data were presented as mean (SEM) and nonparametric data as median (interquartile range [IQR]). Parametric data were analyzed with one-way analysis of variance (ANOVA) and Bonferroni’s posttest correction for intergroup comparison. Nonparametric data were analyzed using the Kruskal-Wallis tests and Dunn’s test for post hoc comparison. The Fisher exact test was used to assess the categorical differences in expression in ASM. Correlations between parametric data were assessed by Pearson’s correlation and nonparametric data by Spearman’s rank correlation. P < .05 was considered significant. Results The baseline clinical characteristics are as shown in Table 1. The subjects with asthma and the controls were matched for age and gender. Subjects with asthma had a higher BMI compared with controls, and evidence of airway hyperresponsiveness. Table 1 —Baseline Clinical Characteristics

Analysis Statistical analysis was performed using PRISM software, version 4 (GraphPad Software; La Jolla, CA). Parametric data were presented as mean (SEM) and nonparametric data as median (interquartile range [IQR]). Parametric data were analyzed with one-way analysis of variance (ANOVA) and Bonferroni’s posttest correction for intergroup comparison. Nonparametric data were analyzed using the Kruskal-Wallis tests and Dunn’s test for post hoc comparison. The Fisher exact test was used to assess the categorical differences in expression in ASM. Correlations between parametric data were assessed by Pearson’s correlation and nonparametric data by Spearman’s rank correlation. P < .05 was considered significant. Results The baseline clinical characteristics are as shown in Table 1. The subjects with asthma and the controls were matched for age and gender. Subjects with asthma had a higher BMI compared with controls, and evidence of airway hyperresponsiveness. Table 1 —Baseline Clinical Characteristics Control (n = 13) Asthma GINA 1 (n = 10) GINA 2-3 (n = 7) GINA 4-5 (n = 10) Age (y) 40 (3) 45 (5) 55 (7) 49 (5) Sex, male (female) 6 (7) 5 (5) 4 (3) 6 (4) Atopy, % 33 50 57 67 Disease duration, y NA 12 (4) 12 (5) 17 (5) BDP equivalent/24 h, mcg NA 0 960 (194)a 1,092 (136)a OCS (n = 2) BMI, kg/m2 24 (1) 32 (2) 31 (2) 30 (2) Postbronchodilator FEV1% 103 (4) 95 (6) 99 (4) 83 (8) Postbronchodilator FEV1/FVC, % 81 (3) 77 (4) 78 (2) 74 (3) PC20,b mg/mL > 16 0.5 (0.1-2.2) 0.5 (0.2-1.9) 0.4 (0.1-2.1) Bronchodilator response FEV1% 0.6 (1) 12 (6) 4 (3) 16 (7) Sputum eosinophils,b % 0.3 (0.2-0.5) 0.7 (0.2-2.8) 1.2 (0.4-3.8) 2.6 (0.4-17)a Sputum neutrophils, % 64 (20) 51 (10) 58 (9) 52 (14) Eosinophils/mm2 lamina propriac 1.9 (3.7) 21.3 (34.5)a 12.1 (12.7) 6.1 (38.4) Data expressed as mean (SEM) unless otherwise indicated. BDP = beclamethasone diproprionate equivalent dose; GINA = Global Initiative for Asthma; NA = not applicable; OCS = oral corticosteroid; PC20 = provocation concentration causing a 20% fall in FEV1.

mm2 lamina propriac 1.9 (3.7) 21.3 (34.5)a 12.1 (12.7) 6.1 (38.4) Data expressed as mean (SEM) unless otherwise indicated. BDP = beclamethasone diproprionate equivalent dose; GINA = Global Initiative for Asthma; NA = not applicable; OCS = oral corticosteroid; PC20 = provocation concentration causing a 20% fall in FEV1. a P < .05. b Geometric mean (95% CI). c Median (interquartile range). Photomicrographs illustrating OX40, OX40L, IL-4, IL-4Rα, and the colocalization of OX40, OX40 with T cells, and dendritic cells, respectively, in the lamina propria and their appropriate isotype controls are as shown (Fig 1). In a subset of asthmatics (n = 3), we determined that mean (SEM) 56 (6)% of the OX40L+ cells in the lamina propria were dendritic cells and 94 (6)% of the OX40+ cells were T cells. Figure 1. Examples of photomicrographs of immunohistochemical staining for isotype control (A), OX40 (B), OX40L (C), IL-4(3H4) (D), IL-4(4D9) (E), IL-4Rα (F), and sequential sections staining for OX40 (G), CD3 (H), CD1a (I) and OX40L (J) (original magnification × 400). Positive cells are highlighted by arrows in B-F. In G and I, positive cells are highlighted by arrows and the corresponding cell if positive in H and J. IL = interleukin; IL-4Rα = IL-4 receptor α; OX40L = OX40 ligand.

ential sections staining for OX40 (G), CD3 (H), CD1a (I) and OX40L (J) (original magnification × 400). Positive cells are highlighted by arrows in B-F. In G and I, positive cells are highlighted by arrows and the corresponding cell if positive in H and J. IL = interleukin; IL-4Rα = IL-4 receptor α; OX40L = OX40 ligand. The mean (SEM) number of OX40+ cells in the lamina propria was significantly increased in subjects with mild corticosteroid-naive (GINA 1) asthma (1.2 [0.5]/mm2), but not in those with moderate asthma (GINA 2-3) (0.31 [0.16]/mm2) or severe asthma (GINA 4-5) (0.3 [0.16]/mm2), compared with healthy controls (0.2 [0.1]/mm2) (P = .048, ANOVA; P < .05, control vs GINA 1) (Fig 2A). Similarly, the number of OX40L+ cells in the lamina propria was significantly increased in subjects with mild asthma (1.9 [0.7]/mm2), but not in those with moderate asthma (0.17 [0.1]/mm2) or severe asthma (0.51 [0.21]/mm2), compared with healthy controls (0.55 [0.3]/mm2) (P = .04, ANOVA; P < .05, control vs GINA 1) (Fig 2B). The median (IQR) number of OX40+ cells in the ASM bundle was not significantly different across the groups (P = .08, Kruskal-Wallis), but the proportion of patients with OX40+ cells in the ASM was significantly increased in mild asthma (5/10), but not in moderate (1/6) or severe asthma (3/8), compared with the controls (0/10) (P = .03, Fisher exact test) (Fig 2 C). OX40L+ cells were not identified within the ASM bundle in health or asthma. There was a strong correlation between the number of OX40 and OX40L+ cells in the lamina propria (r = 0.83; P < .0001) (Fig 2D).

in moderate (1/6) or severe asthma (3/8), compared with the controls (0/10) (P = .03, Fisher exact test) (Fig 2 C). OX40L+ cells were not identified within the ASM bundle in health or asthma. There was a strong correlation between the number of OX40 and OX40L+ cells in the lamina propria (r = 0.83; P < .0001) (Fig 2D). Figure 2. The number of OX40+ (A) and OX40L+ cells (B) in the lamina propria and OX40+ cells (C) in the ASM bundle in subjects with asthma and healthy controls. The correlation between OX40 and OX40L expression in the lamina propria (D). □ = controls; ◆ = GINA 1; ● = GINA 2; ▲ = GINA 3; ▼ = GINA 4; X = GINA 5. *P < .05. ASM = airway smooth muscle; GINA = Global Initiative for Asthma. See Figure 1 for expansion of other abbreviations.

subjects with asthma and healthy controls. The correlation between OX40 and OX40L expression in the lamina propria (D). □ = controls; ◆ = GINA 1; ● = GINA 2; ▲ = GINA 3; ▼ = GINA 4; X = GINA 5. *P < .05. ASM = airway smooth muscle; GINA = Global Initiative for Asthma. See Figure 1 for expansion of other abbreviations. The mean (SEM) number of IL-4 (3H4)+ cells in the lamina propria was significantly increased in subjects with mild asthma (2.7 [1.0]/mm2), but not in those with moderate asthma (0.64 [0.4]/mm2) or severe asthma (0.84 [0.25]/mm2), compared with healthy controls (0.26 [0.15]/mm2) (P = .01, ANOVA; P < .01, control vs GINA 1) (Fig 3A). Similarly, the number of IL-4 (4D9)+ cells in the lamina propria was significantly increased in subjects with mild asthma (3.4 [1.4]/mm2), but not in those with moderate asthma (0.61 [0.23]/mm2) or severe asthma (0.45 [0.2]/mm2), compared with healthy controls (0.17 [0.09]/mm2) (P = .007, ANOVA; P < .01, control vs GINA 1) (Fig 3B). The median (IQR) number of IL-4+ (3H4 and 4D9) cells in the ASM bundle was significantly increased in subjects with mild asthma (0.5 [10.8] and 1.2 [6.8]/mm2), but not in those with moderate asthma (0 [0.5] and 0 [0.8]/mm2) or severe asthma (0 [0] and 0 [0]/mm2), compared with healthy controls (0 [0] and 0 [0]/mm2) (P = .04 and P = .027, Kruskal-Wallis) (Figs 3C, D).

lls in the ASM bundle was significantly increased in subjects with mild asthma (0.5 [10.8] and 1.2 [6.8]/mm2), but not in those with moderate asthma (0 [0.5] and 0 [0.8]/mm2) or severe asthma (0 [0] and 0 [0]/mm2), compared with healthy controls (0 [0] and 0 [0]/mm2) (P = .04 and P = .027, Kruskal-Wallis) (Figs 3C, D). Figure 3. The number of IL-4 (3H4)+ cells (A) and IL-4 (4D9)+ cells (B) in the lamina propria and IL-4 (3H4)+ cells (C) and IL-4 (4D9)+ cells (D) in the ASM bundle in subjects with asthma and healthy controls. □ = controls; ◆ = GINA 1; ● = GINA 2; ▲ = GINA 3; ▼ = GINA 4; X = GINA 5. *P < .05. See Figure 1 and 2 legends for expansion of abbreviations. In contrast, the mean (SEM) number of IL-4Rα+ cells in the lamina propria was increased in moderate asthma (3.2 [0.9] cells/mm2) and severe asthma (3.1 [0.5] cells/mm2), but not in mild disease (1.9 [0.5] cells/mm2), compared with healthy controls (P = .002, ANOVA) (Fig 4A). The intensity of IL-4Rα expression by ASM was not significantly different between asthma and healthy controls (P = .3, ANOVA) (Fig 4B). Figure 4. The number of IL-4Rα+ cells in the lamina propria (A) and IL-4Rα expression by the ASM bundle (B) in subjects with asthma and healthy controls. See Figures 1 and 2 legends for expansion of abbreviations. □ = controls; ◆ = GINA 1; ● = GINA 2; ▲ = GINA 3; ▼ = GINA 4; X = GINA 5. *P < .05.

In contrast, the mean (SEM) number of IL-4Rα+ cells in the lamina propria was increased in moderate asthma (3.2 [0.9] cells/mm2) and severe asthma (3.1 [0.5] cells/mm2), but not in mild disease (1.9 [0.5] cells/mm2), compared with healthy controls (P = .002, ANOVA) (Fig 4A). The intensity of IL-4Rα expression by ASM was not significantly different between asthma and healthy controls (P = .3, ANOVA) (Fig 4B). Figure 4. The number of IL-4Rα+ cells in the lamina propria (A) and IL-4Rα expression by the ASM bundle (B) in subjects with asthma and healthy controls. See Figures 1 and 2 legends for expansion of abbreviations. □ = controls; ◆ = GINA 1; ● = GINA 2; ▲ = GINA 3; ▼ = GINA 4; X = GINA 5. *P < .05. In the subjects with asthma, OX40 and OX40L expression in the lamina propria was strongly correlated with the IL-4 (3H4) (r = 0.67, P < .0001 and r = 0.59, P = .001) and (4D9) (r = 0.48, P = .01 and r = 0.4, P = .038) expression, respectively (Fig 5). The number of OX40+ cells in the ASM bundle was not significantly correlated with the number of IL-4+ (3H4 or 4D9) cells (data not shown). We also observed a strong correlation between the number of OX40L+ cells and eosinophils in the lamina propria (rs = 0.64, P < .0001), but not with OX40/IL-4 expression (data not shown). There were no significant correlations among OX40, OX40L, or IL-4 expression in the lamina propria or ASM bundle and FEV1% predicted, bronchodilator response, or airway hyperresponsiveness (data not shown).

and eosinophils in the lamina propria (rs = 0.64, P < .0001), but not with OX40/IL-4 expression (data not shown). There were no significant correlations among OX40, OX40L, or IL-4 expression in the lamina propria or ASM bundle and FEV1% predicted, bronchodilator response, or airway hyperresponsiveness (data not shown). Figure 5. The correlations between OX40/OX40L and IL-4 expression in the lamina propria. See Figure 1 legend for expansion of abbreviations. Discussion We report here, to our knowledge for the first time, that the number of OX40+ and OX40L+ cells are increased in the lamina propria and OX40+ cells in the ASM bundle in mild asthma, but not in moderate-to-severe disease. OX40L+ cells were not observed in the ASM bundle in health or disease. The number of OX40/OX40L cells in the lamina propria was associated with IL-4 expression and eosinophilic inflammation.

ells are increased in the lamina propria and OX40+ cells in the ASM bundle in mild asthma, but not in moderate-to-severe disease. OX40L+ cells were not observed in the ASM bundle in health or disease. The number of OX40/OX40L cells in the lamina propria was associated with IL-4 expression and eosinophilic inflammation. To date, there have been no reports of OX40 or OX40L expression in bronchial submucosal tissue in asthma. Therefore, our observation that OX40 and OX40L expression is increased in mild asthma provides further evidence of a potential role of these molecules in asthma. We identified that the majority of OX40+ cells were T cells and OX40L+ cells were dendritic cells. Previously, animal studies have implicated these costimulatory molecules in the immunopathogenesis of asthma. In mouse models of allergic inflammation, OX40 knockout mice have significantly reduced Th2 responses to allergen with a reduction in airway eosinophilia and airway hyperreactivity.5,16,17 Dendritic cell-derived OX40L is induced by TSLP, a IL-7-like cytokine produced by epithelial cells, which provides an important interaction between the epithelial-dendritic cell.7 Critically, OX40L neutralization in mouse and nonhuman primate models of asthma inhibited TSLP-induced immune responses, including Th2 inflammatory cell infiltration, cytokine secretion, and IgE production8. Consistent with these observations, our findings showed a close association between the number of eosinophils and IL-4 expression in the lamina propria and the OX40/OX40L axis. However, in contrast to the animal models, a relationship was not indicated between OX40/OX40L expression and disordered airway physiology, including airway hyperresponsiveness. This dissociation between airway function and Th2-mediated eosinophilic inflammation is consistent with observations made in noneosinophilic bronchitis18,19 and the response to highly specific anti-IL-5 therapy.20,21 These observations in human disease may challenge a potential role for the OX40/OX40L axis in airway dysfunction in asthma and need to be examined further.

diated eosinophilic inflammation is consistent with observations made in noneosinophilic bronchitis18,19 and the response to highly specific anti-IL-5 therapy.20,21 These observations in human disease may challenge a potential role for the OX40/OX40L axis in airway dysfunction in asthma and need to be examined further. Interestingly, OX40L expression by dendritic cells is also increased by environmental fungal22 and endotoxin23 exposure, which have both been implicated in asthma and asthma exacerbations.24,25 OX40L production is also not restricted to dendritic cells, because other cells such as ASM may be important potential sources.26,27 Ex vivo data in human ASM cells have shown that OX40/OX40L expression was upregulated by tumor necrosis factor-α, and, interestingly, was significantly greater in asthmatic vs nonasthmatic ASM cells. We did not observe OX40/OX40L expression by ASM, but mast cells located within the ASM bundle14,28 are an important source of tumor necrosis factor-α29 and following activation may induce ASM-derived OX40L. ASM cells have established contractile and synthetic function and to this may be added a possible immunomodulatory role. Therefore, the role of mesenchymal cells and their interactions with inflammatory cells in the OX40/OX40L axis need to be investigated further.

and following activation may induce ASM-derived OX40L. ASM cells have established contractile and synthetic function and to this may be added a possible immunomodulatory role. Therefore, the role of mesenchymal cells and their interactions with inflammatory cells in the OX40/OX40L axis need to be investigated further. Importantly, we were unable to demonstrate increased OX40/OX40L expression in moderate-to-severe disease, in contrast to mild asthma. This may reflect real differences among disease severities, perhaps as a consequence of treatment. Importantly, the number of subjects included in this study was relatively small and therefore our inability to demonstrate an increase in moderate-to-severe disease may be because of inadequate powering of the study. Furthermore, we may have underestimated OX40 and OX40L expression in asthma, because it may be transient and may increase in response to activation such as allergen challenge. Similarly, the potential benefits of therapeutic blockade of OX40/OX40L may depend on sensitization and the duration of post-allergen challenge. In mouse models of asthma, conflicting data exist regarding the most efficacious time point to neutralize OX40L, with one report suggesting that presensitization alone, and not subsequent challenge, led to a reduction in the asthmatic response,17 and another finding that airway inflammation was attenuated in postsensitized animals.30 Therefore, although our inability to demonstrate increased expression in severe asthma does question its potential role as a therapeutic target for this group of patients, our findings do not exclude the possibility that the OX40/OX40L axis may be important across all severities of asthma.

in postsensitized animals.30 Therefore, although our inability to demonstrate increased expression in severe asthma does question its potential role as a therapeutic target for this group of patients, our findings do not exclude the possibility that the OX40/OX40L axis may be important across all severities of asthma. In keeping with the OX40/OX40L expression, IL-4 upregulation was restricted to those subjects with mild disease. This is consistent with an earlier report,31 but is in stark contrast to IL-13 expression, which resulted in an increase in the sputum and bronchial mucosa of severe asthmatics.32,33 Interestingly, the number of IL-4Rα+ cells was increased in the lamina propria in moderate-to-severe disease, with a small, nonsignificant increase in mild asthmatics. This differential expression of IL-4 and IL-13 in severe disease has important implications for potential therapeutic strategies in severe asthma directed toward either cytokine neutralization34,35 or receptor antagonism.36 In addition, the balance between the expression of cytokines and their receptors in severe disease is likely to be clinically important and requires further consideration.

ant implications for potential therapeutic strategies in severe asthma directed toward either cytokine neutralization34,35 or receptor antagonism.36 In addition, the balance between the expression of cytokines and their receptors in severe disease is likely to be clinically important and requires further consideration. Conclusions In conclusion, we have shown that the OX40/OX40L axis is upregulated in mild asthma. but not in moderate-tosevere disease, and is related to the intensity of IL-4 expression and eosinophilic inflammation. Whether these costimulatory molecules present new therapeutic targets for asthma, and how we can determine which severity of asthma and what stage of the natural history of disease may benefit, require further investigation. Author contributions: Dr Siddiqui: contributed to patient recruitment, bronchoscopies, manuscript preparation, and data analysis. Mr Mistry: contributed to tissue processing and immunohistochemistry. Ms Doe: contributed to tissue processing and immunohistochemistry. Ms Stinson: contributed to technical advice provision and coauthorship of the manuscript. Dr Foster: contributed to technical advice provision and coauthorship of the manuscript. Dr Brightling: contributed to patient recruitment, bronchoscopies, manuscript preparation, and data analysis.

Ms Doe: contributed to tissue processing and immunohistochemistry. Ms Stinson: contributed to technical advice provision and coauthorship of the manuscript. Dr Foster: contributed to technical advice provision and coauthorship of the manuscript. Dr Brightling: contributed to patient recruitment, bronchoscopies, manuscript preparation, and data analysis. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Brightling has received consultancy fees from MedImmune, AstraZeneca, GlaxoSmithKline, Roche, and Genentech and research grants from AZ, MedImmune, and GSK. Dr Foster and Ms Stinson are employees of AstraZeneca. Dr Siddiqui, Mr Mistry, and Ms Doe have reported that no potential conflicts exist with any companies/organizations whose products or services may be discussed in this article. Funding/Support: This study was supported by AstraZeneca, a Department of Health Clinician Scientist award, and a Wellcome Senior Clinical Fellowship. Abbreviations ASMairway smooth muscle GINAGlobal Initiative for Asthma ILinterleukin IL-4RαIL-4 receptor α IQRinterquartile range OX40LOX40 ligand Th2T helper 2 TSLPthymic stromal lymphopoietin

Asthma affects over 300 million people worldwide, and its prevalence is increasing. Asthma is a complex disease characterized by airway hyperresponsiveness, variable airflow obstruction, airway inflammation, varying degrees of subepithelial fibrosis, mucus hyperproduction, and remodeling.1 Atopic asthma has classically been associated with increased expression of T helper (Th) 2 cytokines, which are increased in sputum,2 bronchoalveolar T cells,3 and bronchial biopsies.4 A major effector axis resulting in induction of Th2 polarization is the recognition of allergen presented by dendritic cells in local lymph nodes to CD4+ T cells. This axis is an optimal target for drug development because it orchestrates the inflammation and development of an allergen-specific humoral response and the development of T- and B-cell memory. The differentiation of naive T cells or reactivation of memory T cells depends on various costimulatory molecules primarily expressed on the surface of T cells and their cognate ligands. One of the most promising costimulatory targets is OX40 and its ligand, OX40L. OX40L is directly mediated by thymic stromal lymphopoietin (TSLP), which is produced by epithelial cells,5 mast cells,6 airway smooth muscle,7 and dendritic cells,8 which are all involved in Th2 responses. TSLP was originally identified as a growth-promoting factor found in cultured supernatants of a thymic stromal cell line in 1994 to support the development of murine B cells.9 TSLP plays an important role in many allergic diseases, such as atopic dermatitis and asthma. TSLP is also up-regulated in COPD,10 but its role and relationship to OX40/OX40L signaling in this disease is unclear. TSLP binds to its TSLP receptor and the IL-7 receptor α chain. Dendritic cells play a crucial role in the pathogenesis of allergic disease. TSLP activates immature CD11c+ dendritic cells to express OX40L, and these cells then become mature dendritic cells, which migrate to the draining lymph nodes.

disease is unclear. TSLP binds to its TSLP receptor and the IL-7 receptor α chain. Dendritic cells play a crucial role in the pathogenesis of allergic disease. TSLP activates immature CD11c+ dendritic cells to express OX40L, and these cells then become mature dendritic cells, which migrate to the draining lymph nodes. There they activate the differentiation of naive CD4+ T cells by binding to the OX40 receptor, where they become inflammatory cells producing IL-4, IL-5, IL-13, tumor necrosis factor-α (TNF-α), and little or no IL-10 (Fig 1).11

disease is unclear. TSLP binds to its TSLP receptor and the IL-7 receptor α chain. Dendritic cells play a crucial role in the pathogenesis of allergic disease. TSLP activates immature CD11c+ dendritic cells to express OX40L, and these cells then become mature dendritic cells, which migrate to the draining lymph nodes. There they activate the differentiation of naive CD4+ T cells by binding to the OX40 receptor, where they become inflammatory cells producing IL-4, IL-5, IL-13, tumor necrosis factor-α (TNF-α), and little or no IL-10 (Fig 1).11 Figure 1. Drawing shows the pathophysiologic characteristics of OX40/OX40L and TSLP in allergic inflammation. Cellular damage caused by allergens or viruses triggers mucosal epithelial cells or skin cells (keratinocytes, fibroblasts, and mast cells) to produce TSLP. TSLP initiates the innate phase of allergic immune responses by activating immature DCs by binding to their TSLPR. TSLP/TSLPR-activated DCs produce the chemokines IL-8 and eotaxin-2, TARC, and MDC, and by costimulating mast cells, produce IL-5, IL-13, GM-CSF, and IL-6. The activated immature DCs then mature and produce the OX40L and migrate into the draining lymph nodes, where they trigger the differentiation of allergen-specific naive CD4+ T cells by binding to the receptor OX40 and differentiating the CD4+ T cells into inflammatory Th2 cells, producing IL-4, IL-5, IL-13, and TNF-α, but no IL-10. These Th2 cytokines then initiate inflammation by triggering IgE production, eosinophilia, mucus production, and fibroblast proliferation. DC =dendritic cell; GM-CSF =granulocyte-macrophage colony-stimulated factor; MDC = macrophase-derived chemokine; MHC = major histocompatibility complex; TARC =T-helper 2 attracting chemokines activation-regulated chemokine; Th = T-helper; TNF = tumor necrosis factor; TSLP =thymic stromal lymphopoietin; TSLPR =thymic stromal lymphopoietin receptor.

ranulocyte-macrophage colony-stimulated factor; MDC = macrophase-derived chemokine; MHC = major histocompatibility complex; TARC =T-helper 2 attracting chemokines activation-regulated chemokine; Th = T-helper; TNF = tumor necrosis factor; TSLP =thymic stromal lymphopoietin; TSLPR =thymic stromal lymphopoietin receptor. The sentinel roles of the OX40/OX40L axis in the adaptive immune response and TSLP in both the innate and adaptive responses suggest these molecular targets may present attractive novel therapeutic targets. In this article, we consider the evidence that the OX40/OX40L axis plays a role in asthma, its potential importance as a therapeutic target, and the likely target population. OX40 and OX40L OX40 (ACT35, CD134, TNFRSF4) was identified in 1987 and found to be bound to activated T cells.12 Since then, it has been cloned in rat, mouse, and human cells. The OX40 receptor is preferentially expressed on the surface of activated regulatory CD4+ T cells,13 natural killer T cells, natural killer cells, and neutrophils, and more recently, we have found it to be expressed in human airway smooth muscle cells.14 OX40 signaling strongly regulates T-cell division, survival, and cytokine release.15 The OX40 ligand (OX40L, CD252, TNFSF4) was originally identified in 1985 as gp34 (GP34) protein on human T cell leukemic virus-transformed cells16 and is expressed on antigen-expressing cells, for instance, B cells,17 dendritic cells,18 and macrophages19 as well as airway smooth muscle cells.20

and cytokine release.15 The OX40 ligand (OX40L, CD252, TNFSF4) was originally identified in 1985 as gp34 (GP34) protein on human T cell leukemic virus-transformed cells16 and is expressed on antigen-expressing cells, for instance, B cells,17 dendritic cells,18 and macrophages19 as well as airway smooth muscle cells.20 Evidence of a Critical Role for OX40/OX40L in the Pathogenesis of Asthma Animal Models In murine asthma models, OX40−/− mice challenged with ovalbumin showed significantly reduced Th2 response, lung inflammation, mucus secretion, 80% to 90% reduction in eosinophilia, decreased goblet cell hyperplasia, and significantly attenuated airway hyperreactivity compared with wild-type mice.21 Studies have also demonstrated that OX40L−/− mice sensitized with ovalbumin have significantly reduced total serum IgE, pulmonary eosinophils, cytokines, and pulmonary inflammation compared with wild-type control mice.22,23 Inhibition of OX40-OX40L binding via the administration of anti-OX40L mAb in wild-type mice dramatically reduced airway hyperresponsiveness and associated asthma symptoms, compared with mice challenged with isotype control.23,24 Mouse splenic CD11c+ dendritic cells stimulated for 48 h with TSLP upregulated OX40L expression compared with CD40L or TNF-α and unstimulated cells. A blockade of OX40/OX40L interactions using a specific α-mouse OX40L 4F5 monoclonal antibody significantly inhibited Th2 cytokine production compared with a control antibody. This confirms that OX40L activity on dendritic cells was important for the effects of TSLP in driving Th2 polarization.25 Both in murine and nonhuman primate models of asthma in vivo, a blockade of OX40L inhibited TSLP-mediated Th2 inflammation.25 Studies have demonstrated that OX40 can inhibit the development of adaptive Foxp3+ T regulatory cells that differentiate from naive CD4 T-cell populations in response to TGF-β.26,27 Foxp3, an X chromosome-encoded fork-head transcription factor family member, is critical for the differentiation of regulatory T cells. These cells have an important role in preventing autoimmunity and pathologic changes inflicted by uncontrolled immune responses to infections. Deficiency or mutation in Foxp3 in humans and mice leads to early onset and susceptibility to diseases such as asthma.28

critical for the differentiation of regulatory T cells. These cells have an important role in preventing autoimmunity and pathologic changes inflicted by uncontrolled immune responses to infections. Deficiency or mutation in Foxp3 in humans and mice leads to early onset and susceptibility to diseases such as asthma.28 Targeting OX40L may have the potential to improve the efficacy of immunotherapy to promote tolerance. In wild-type mice exposed to intranasal antigen and specific CD4+Foxp3+, regulatory T cells were generated, which outnumbered IL-4 and interferon γ-producing CD4 T cells. Inhaled lipopolysaccharide downregulated the regulatory T cells, but up-regulated IL-4+ and interferon-γ T cells, and it also increased OX40L expression on dendritic cells and B cells. Inhibiting OX40/OX40L interactions with an anti-OX40L antibody upregulated regulatory T cells suppressing lipopolysaccharide stimulation.29 Sensitization to fungi such as Aspergillus fumigatus and Alternaria is associated with poor lung function30 and exacerbations. When bone marrow-derived mouse dendritic cells were stimulated with Alternaria for 48 h, upregulation in OX40L expression was detected using flow cytometry,31 suggesting that Alternaria may play a role in Th2 cytokine release.

pergillus fumigatus and Alternaria is associated with poor lung function30 and exacerbations. When bone marrow-derived mouse dendritic cells were stimulated with Alternaria for 48 h, upregulation in OX40L expression was detected using flow cytometry,31 suggesting that Alternaria may play a role in Th2 cytokine release. The activation of pattern-recognition receptors such as toll-like receptors plays a critical role in Th1 cell differentiation, yet their contribution to the generation of Th2 responses is poorly understood. Interestingly, when mice deficient in either MyD88−/− or TLR4−/− were sensitized intranasally to the common allergen house dust mites and challenged 2 weeks later, they showed diminished Th2 responses as well as fewer OX40Ls presenting dendritic cells in the draining lymph node compared with wild-type mice.32 The activation marker CD30, a member of the TNF receptor family, is expressed on activated T cells. In an acute asthma model, CD30−/− mice developed reduced expression of OX40,33 whereas in contrast, OX40 expression was not downregulated in a chronic murine asthma model.34 These differences in expression may be the result of the fact that in a chronic asthma model, T cells are able to proliferate, leading to chronic airway inflammation. Airway tolerance is vital for protecting the lung from inflammatory disease-driven allergens, but factors that lead to this susceptibility remain elusive. The pattern recognition receptors nucleotide-binding oligomerization domain (Nod)-like receptors Nod1 and Nod2 are both highly expressed by epithelial cells. Intranasal exposure of Nod2, but not Nod1, induced TSLP-promoting OX40L expression. The generation of these ligands also blocked CD4+ fork-head box protein 3+ adaptive regulatory T cells and concomitantly drove IL-4-producing CD4 T cells, leading to allergic disease and asthmatic lung inflammation.35

epithelial cells. Intranasal exposure of Nod2, but not Nod1, induced TSLP-promoting OX40L expression. The generation of these ligands also blocked CD4+ fork-head box protein 3+ adaptive regulatory T cells and concomitantly drove IL-4-producing CD4 T cells, leading to allergic disease and asthmatic lung inflammation.35 The animal-model data, therefore, present compelling evidence of a central role for OX40/OX40L in the development of Th2 polarization in response to a number of insults that are considered important in asthma. These data support a role for this axis in the immunopathogenesis of asthma.

epithelial cells. Intranasal exposure of Nod2, but not Nod1, induced TSLP-promoting OX40L expression. The generation of these ligands also blocked CD4+ fork-head box protein 3+ adaptive regulatory T cells and concomitantly drove IL-4-producing CD4 T cells, leading to allergic disease and asthmatic lung inflammation.35 The animal-model data, therefore, present compelling evidence of a central role for OX40/OX40L in the development of Th2 polarization in response to a number of insults that are considered important in asthma. These data support a role for this axis in the immunopathogenesis of asthma. Human Models The role of OX40/OX40L in humans is limited compared with murine-model systems. TSLP was preferentially induced in peripheral blood-isolated myeloid dendritic cells from healthy volunteers to express mRNA for OX40L using microarray analysis.11 Blocking OX40/OX40L interactions using a specific OX40L-neutralizing antibody inhibited the production of Th2 cytokines and TNF-α, but increased the production of IL-10 in CD4+ cells cocultured with TSLP-primed dendritic cells.11 Airway smooth muscle cells are critical in the development of bronchoconstriction in asthma, are the major contributors to airway remodeling and persistent airflow obstruction, and release a number of chemokines/cytokines that bind specifically to activated T cells, resulting in increased cellular proliferation. Studies have reported OX40L to be expressed and released by airway smooth muscle cells of people with and without asthma, but with no significant difference.20 However, following TNF-α stimulation, there was an increase in OX40L, and a decrease after stimulation of TNF-α and interferon-γ combined.36 Cells activated with rOX40:Fc over 24 h released IL-6, which was significantly higher in the patients with asthma compared with people without asthma.36 More recently, our group has reported OX40/OX40L expression to be increased in the bronchial submucosa of patients with mild asthma, but not in those with moderate to severe disease, and this was related to the degree of tissue eosinophilia and IL-4 expression.14

ients with asthma compared with people without asthma.36 More recently, our group has reported OX40/OX40L expression to be increased in the bronchial submucosa of patients with mild asthma, but not in those with moderate to severe disease, and this was related to the degree of tissue eosinophilia and IL-4 expression.14 Air pollution, particularly from diesel-exhaust particulates, is associated with worsening in asthma symptoms and control. Bronchial epithelial cells are the first major targets for inhaled pollutants. Bronchial epithelial cells treated with diesel-exhaust particulates express an increase in OX40L expression.37 A recent study also identified TSLP to be highly expressed in isolated nasal epithelial cells from patients with nasal polyposis compared with those without. The TSLP receptor and OX40L receptor were also increased in dendritic cells from the nasal mucosa of patients with nasal polyposis.38 The evidence in humans of a role of OX40/OX40L in asthma is, therefore, circumstantial and the expression data are weak. However, the interaction between OX40/OX40L in Th2 polarization may occur early in the disease pathogenesis and, more importantly, is primarily located in the lymph node rather than the bronchial submucosa. Current strategies to explore this axis have not addressed this compartment, and, therefore, its role has not been fully studied.

eraction between OX40/OX40L in Th2 polarization may occur early in the disease pathogenesis and, more importantly, is primarily located in the lymph node rather than the bronchial submucosa. Current strategies to explore this axis have not addressed this compartment, and, therefore, its role has not been fully studied. TSLP Asthma TSLP is both necessary and sufficient for the development of Th2 cytokine-associated inflammation of the airways in rodents. Mice expressing a TSLP transgene in the airway epithelium develop a spontaneous, progressive inflammatory disease with all the characteristics of human asthma,39 whereas direct intranasal delivery of TSLP (in the presence of antigen) leads to rapid onset of features similar to severe disease.40 Studies have also reported increased TSLP mRNA in bronchial epithelium in asthma in response to allergen, viruses, and other environmental stimuli.41 In human disease, genetic analysis has shown an association of polymorphisms in TSLP with asthma and airway hyperresponsiveness, IgE concentrations, and eosinophilia.42-44 In addition, patients with asthma have higher concentrations of TSLP in their lungs.45,46 The role of TSLP in both the innate and adaptive immune response may suggest that its potential efficacy is broader than targeting the OX40/OX40L axis alone.

d airway hyperresponsiveness, IgE concentrations, and eosinophilia.42-44 In addition, patients with asthma have higher concentrations of TSLP in their lungs.45,46 The role of TSLP in both the innate and adaptive immune response may suggest that its potential efficacy is broader than targeting the OX40/OX40L axis alone. Targeting the TSLP and OX40/OX40L Axis in Asthma There is an increasing recognition that asthma is a heterogeneous condition.47 Complex gene-environment interactions activate several biologic pathways that consequently result in the disordered airway physiologic aspects and symptoms that characterize asthma. The view that asthma is primarily an allergic disease mediated by Th2 cytokines has been challenged because asthma can develop in the absence of atopy. Indeed, allergic sensitization is likely to be more important in early-onset disease and particularly in children, whereas this feature of disease is less prominent in late-onset asthma.48 The application of noninvasive measures of airway inflammation has also led us to observe different inflammatory phenotypes, including eosinophilic- and neutrophilic-predominant asthma.49 The OX40/OX40L axis is particularly important in allergic sensitization and Th2 polarization. Therefore, predictably, patients with Th2-mediated eosinophilic inflammation are likely to be the most appropriate target population. However, the timing of the intervention is important and may be most effective prior to the onset of allergic sensitization and disease presentation, which obviously is unpredictable, and hence this is not a practical strategy. Ongoing activation of the OX40/OX40L axis may be important is some patients with asthma, but to date, this is uncertain and biomarkers to identify this group are unknown. At present, one clinical trial has been completed using a huMAb OX40L in the prevention of allergen-induced airway in adults with mild asthma. This study was funded by Genentech and completed in January 2011. However, the report from this study is still awaited.50

tain and biomarkers to identify this group are unknown. At present, one clinical trial has been completed using a huMAb OX40L in the prevention of allergen-induced airway in adults with mild asthma. This study was funded by Genentech and completed in January 2011. However, the report from this study is still awaited.50 TSLP is likely to be effective in the same population as for OX40/OX40L, but also is critical in the innate immune response, suggesting its potential efficacy may target a broader group of patients with asthma. Critically, the potential efficacy of either approach will need to outweigh the potential side effects. Importantly, to date, early safety studies suggest that the safety profile of anti-OX40 therapy is good. Conclusion Current asthma therapies improve symptoms, improve disease control, and reduce exacerbations.1 None are disease modifying whereby they alter the natural history of the underlying disease. Data from animal models present a compelling argument that targeting the TSLP or OX40/OX40L axis will alter allergic sensitization and T-cell polarization. This presents a tantalizing possibility that this therapeutic approach in asthma may be disease modifying. Patients with asthma and researchers share the ambition to achieve a cure for asthma that can only be achieved by disease modification. Therefore, targeting TSLP or the OX40/OX40L axis presents an exciting opportunity that may provide a step-change in the treatment of asthma. Its fate will become apparent over the forthcoming couple of years as the eagerly awaited clinical trials are reported.

r asthma that can only be achieved by disease modification. Therefore, targeting TSLP or the OX40/OX40L axis presents an exciting opportunity that may provide a step-change in the treatment of asthma. Its fate will become apparent over the forthcoming couple of years as the eagerly awaited clinical trials are reported. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Brightling has received consultancy fees and research funding from AstraZeneca, MedImmune, GlaxoSmithKline, Chiesi, Roche, and Novartis. Dr Kaur has reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or in the preparation of the manuscript. Funding/Support: Dr Brightling is supported by a Wellcome Senior Clinical Fellowship. This is an Open Access article distributed under the terms of the Creative Commons Attribution-Noncommercial License (http://creativecommons.org/licenses/by-nc/3.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Information for commercial entities is available online (http://www.chestpubs.org/site/misc/reprints.xhtml). Abbreviations Nodnucleotide-binding oligomerization domain ThT helper TNFtumor necrosis factor TSLPthymic stromal lymphopoietin

COPD is a serious public health problem in the United States. In 2008, chronic lower respiratory diseases, of which COPD represents the principal component, became the third leading cause of mortality.1 Because smoking is the dominant risk factor for COPD and contributed to about 80% of COPD deaths in 2000 to 2004,2 much of this disease is potentially preventable. People with COPD experience worse health-related quality of life, more disabilities, and higher rates of comorbidities than people without COPD.3‐5 The direct economic cost attributable to COPD and asthma in 2008 has been estimated at $53.7 billion in the United States.6 These costs include those for prescription medicines ($20.4 billion), outpatient or office-based providers ($13.2 billion), hospital inpatient stays ($13.1 billion), home health care ($4.0 billion), and ED visits ($3.1 billion).

table to COPD and asthma in 2008 has been estimated at $53.7 billion in the United States.6 These costs include those for prescription medicines ($20.4 billion), outpatient or office-based providers ($13.2 billion), hospital inpatient stays ($13.1 billion), home health care ($4.0 billion), and ED visits ($3.1 billion). COPD consists of chronic bronchitis, emphysema, and small airways disease. This common lung disease is characterized by inflammation and thickening of the mucosae of the airways, weakening or destruction of alveolar walls, and excess mucus production. These mechanical and physiologic changes lead to airflow limitation with limited reversibility. Patients affected by this disorder may be asymptomatic or experience cough, dyspnea, wheezing, and chest tightness. With progression of the disease, dyspnea worsens and oxygenation impairment develops. As the capacity of the lung continues to decline, patients may have increasing difficulty in performing activities of daily living. Although the clinical course of COPD is variable, it is progressive in many patients. Increasingly, research is examining the relationships between COPD and comorbid disease.7,8

develops. As the capacity of the lung continues to decline, patients may have increasing difficulty in performing activities of daily living. Although the clinical course of COPD is variable, it is progressive in many patients. Increasingly, research is examining the relationships between COPD and comorbid disease.7,8 The condition has a diverse etiology.4,9 Although smoking is the chief cause of COPD in most populations, substantial proportions of COPD occur among nonsmokers.10‐12 Other important causes include indoor air pollution from burning of biomass, occupational exposures to a variety of dusts and smoke, asthma, and repeated respiratory infections. In addition, genetic causes, such as α1-antitrypsin deficiency, can result in emphysema. In 2002, the Centers for Disease Control and Prevention (CDC) released the initial surveillance report about COPD that contained surveillance data through the year 2000.13 This report summarized data from national data systems regarding prevalence, physician outpatient visits, ED visits, hospitalizations, and mortality. Of note was that the age-adjusted mortality rate had increased from 1980 to 2000, especially in women. The current surveillance report seeks to characterize recent aspects of the burden of COPD by providing additional information from national datasets through 2011.

t visits, ED visits, hospitalizations, and mortality. Of note was that the age-adjusted mortality rate had increased from 1980 to 2000, especially in women. The current surveillance report seeks to characterize recent aspects of the burden of COPD by providing additional information from national datasets through 2011. Materials and Methods The following data sources were used to produce the estimates in this report: Behavioral Risk Factor Surveillance System (BRFSS) (2011), National Health Interview Survey (NHIS) (1999-2011), National Ambulatory Medical Care Survey (NAMCS) (1999-2010), National Hospital Ambulatory Medical Care Survey (NHAMCS) (1999-2010), National Hospital Discharge Survey (NHDS) (1999-2010), death certificate data from the National Vital Statistics System (NVSS) (1999-2010), and Medicare Part A hospital claims administrative data (1999-2010). We did not include data from the National Health and Nutrition Examination Survey in this report because data from NHIS has commonly been used to provide national estimates of the prevalence of COPD. Furthermore, prevalence estimates of obstructive impairment using recent National Health and Nutrition Examination Survey data have been published.14 Except for Medicare hospital claims, the data presented in this report are limited to adults aged ≥ 25 years, to remain consistent with the prior surveillance report. Because all the data that were used in the analyses are freely available in the public domain, our study was exempt from human subject review.

blished.14 Except for Medicare hospital claims, the data presented in this report are limited to adults aged ≥ 25 years, to remain consistent with the prior surveillance report. Because all the data that were used in the analyses are freely available in the public domain, our study was exempt from human subject review. Behavioral Risk Factor Surveillance System BRFSS data from 2011 were used to estimate the state specific and US prevalence of COPD. An annual sample representing the noninstitutionalized US adult population aged ≥ 18 years in each state was selected by state health departments in collaboration with the CDC using a complex multistage sampling design.15 Data from 475,616 respondents aged ≥ 25 years were analyzed for this report. The BRFSS is a random-digit-dialed telephone survey of landline and cellphone households, and one adult is selected for the telephone interview. The median survey response rate in 2011 for all states and the District of Columbia was 49.7% and ranged from 33.8% to 64.1%. The median cooperation rate (percentage of people who completed interviews among all eligible contacted people) was 74.2% and ranged from 52.7% to 84.3%. The following question was used to define COPD: “Have you ever been told by a doctor or other health professional that you have chronic obstructive pulmonary disease (COPD), emphysema, or bronchitis?” An affirmative response was defined as physician-diagnosed COPD. Demographic information was self-reported.

% to 84.3%. The following question was used to define COPD: “Have you ever been told by a doctor or other health professional that you have chronic obstructive pulmonary disease (COPD), emphysema, or bronchitis?” An affirmative response was defined as physician-diagnosed COPD. Demographic information was self-reported. National Health Interview Survey NHIS data from 1999 to 2011 were used to estimate the prevalence of COPD.16 The NHIS is implemented annually by the National Center for Health Statistics, CDC. During each year, a sample representing the civilian, noninstitutionalized US population aged ≥ 18 years was selected by using a complex multistage sampling design that involves stratification, clustering, and oversampling. The universe of primary sampling units (PSUs) (single counties or groups of adjacent counties—or equivalent jurisdictions—or metropolitan area) is organized into strata from which a sample of PSUs representing areas is drawn. From substrata (census blocks or combined blocks) created in these selected PSUs, secondary sampling units are systematically selected. From each substratum, households with African American, Hispanic, and Asian (since 2006) were oversampled, and a sample of all other households was selected. Only one randomly selected adult per family was asked to participate in the Sample Adult questionnaire. Participants were visited in their homes, where US Census Bureau interviewers conducted a computer-assisted personal interview with the participants. The number of adult participants and the response rates of the surveys are summarized in e-Table 1. Data from adult respondents aged ≥ 25 years were analyzed for this report. The following two questions were used to define COPD: “Have you ever been told by a doctor or other health professional that you had emphysema?” and “During the past 12 months, have you been told by a doctor or other health professional that you had chronic bronchitis?” An affirmative response to one or both of these questions was defined as physician-diagnosed COPD for this report. Demographic information was self-reported.

rofessional that you had emphysema?” and “During the past 12 months, have you been told by a doctor or other health professional that you had chronic bronchitis?” An affirmative response to one or both of these questions was defined as physician-diagnosed COPD for this report. Demographic information was self-reported. National Ambulatory Medical Care Survey NAMCS data from 1999 to 2010 were used to estimate the annual number of physician office visits with the first-listed diagnosis of COPD.17 The NAMCS is an annual, national probability sample survey of ambulatory visits to nonfederally employed office-based physicians conducted by the National Center for Health Statistics, CDC. Beginning in 2006, visits to Community Health Centers (CHCs) were also included. NAMCS used a multistage design that involved probability samples of PSUs, physicians within PSUs, and patient visits within practices. The first-stage sample included 112 PSUs. In each sample PSU, a probability sample of practicing nonfederal office-based physicians was selected from master files maintained by the American Medical Association and American Osteopathic Association. The final stage involved systematic random samples of office visits during randomly assigned 7-day reporting periods. Starting in 2006, a dual-sampling procedure was used to select CHC physicians and other providers. First, the traditional NAMCS sample was selected using the methods described previously. Second, information from the Health Resources and Services Administration and the Indian Health Service was used to select a sample of CHCs. Within CHCs, a maximum of three health-care providers were selected, including physicians, physician assistants, nurse practitioners, or nurse midwives. After selection, CHC providers followed traditional NAMCS methods for selecting patient visits. The physician-patient encounter or visit represents the basic sampling unit in NAMCS.

maximum of three health-care providers were selected, including physicians, physician assistants, nurse practitioners, or nurse midwives. After selection, CHC providers followed traditional NAMCS methods for selecting patient visits. The physician-patient encounter or visit represents the basic sampling unit in NAMCS. Data are collected by the physician or the physician’s staff or by US Census Bureau field representatives. Information concerning race and ethnicity was based on the physician’s knowledge of the patient or on the physician’s or assistant’s judgment rather than the patient self-report. The number of physician office visits and the physicians’ response rates are shown in e-Table 2. Because the percent of office visit medical records that were missing race information ranged from 16.9% to 32.8% (e-Table 2), we used information for race (whites and blacks only) that was imputed by the National Center for Health Statistics. Three visit diagnosis fields were available to participating physicians. A diagnosis of COPD was established from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes 490 (bronchitis not specified as acute or chronic), 491 (chronic bronchitis), 492 (emphysema), or 496 (chronic airway obstruction, not elsewhere classified, which includes COPD) for the first-listed diagnosis. Rates for office visits were calculated using US civilian population estimates provided in the data file documentation for each year (e-Table 3). SEs were produced with statistical software.

2 (emphysema), or 496 (chronic airway obstruction, not elsewhere classified, which includes COPD) for the first-listed diagnosis. Rates for office visits were calculated using US civilian population estimates provided in the data file documentation for each year (e-Table 3). SEs were produced with statistical software. National Hospital Ambulatory Medical Care Survey NHAMCS data for the years 1999 to 2010 were used to estimate the number of ED visits for COPD.17 The NHAMCS is an annual, national probability sample survey of ambulatory visits made to nonfederal, general, short-stay hospitals in the US conducted by the National Center for Health Statistics, CDC. NHAMCS uses a multistage probability design with samples of PSUs, hospitals within PSUs, EDs plus clinics within outpatient departments, and patient visits within EDs and outpatient clinics. Sample hospitals are randomly assigned to 16 panels that rotate across 13 4-week reporting periods throughout the year. The initial sample frame of hospitals was based on the 1991 SMG hospital database now maintained by IMS Health Incorporated. Hospital staff or US Census Bureau field representatives performed data collection for NHAMCS. The annual number of patient record forms submitted by EDs is shown in e-Table 4.

roughout the year. The initial sample frame of hospitals was based on the 1991 SMG hospital database now maintained by IMS Health Incorporated. Hospital staff or US Census Bureau field representatives performed data collection for NHAMCS. The annual number of patient record forms submitted by EDs is shown in e-Table 4. The NHAMCS files contained three visit diagnosis fields. An ICD-9-CM code of 490-492 or 496 for the first-listed diagnosis was defined as an ED visit for COPD. Because the percentage of ED records that were missing race information ranged from 10.4% to 15.3% (e-Table 4), we used information for race (whites and blacks only) that was imputed by the National Center for Health Statistics. The US civilian population estimates that we used to calculate rates of ED visits were obtained from the data file documentation for each year (e-Table 3). SEs were produced with statistical software.

), we used information for race (whites and blacks only) that was imputed by the National Center for Health Statistics. The US civilian population estimates that we used to calculate rates of ED visits were obtained from the data file documentation for each year (e-Table 3). SEs were produced with statistical software. National Hospital Discharge Survey NHDS data from 1999 to 2010 were used to estimate the annual number of hospital discharges for COPD.18 NHDS is an annual survey of inpatient discharges from nonfederal, short-stay hospitals in the US conducted from 1965 to 2010 by the National Center for Health Statistics, CDC. Using the SMG Hospital Market Data File or its successors as the sampling frame, the NHDS samples inpatient discharges from nonfederal, general, short-stay hospitals located in the 50 states and the District of Columbia. A three-stage design has been used since 1988. Units selected at the first stage of sampling consisted of either hospitals or geographic areas, such as counties, groups of counties, or metropolitan statistical areas in the 50 states and the District of Columbia. Within sampled geographic areas, additional hospitals were selected. Finally, at the last stage, discharges were selected within the sampled hospitals using systematic random sampling. Data collection was performed with manual and automated systems. The annual number of sampled records and hospital response rates are provided in e-Table 5.

phic areas, additional hospitals were selected. Finally, at the last stage, discharges were selected within the sampled hospitals using systematic random sampling. Data collection was performed with manual and automated systems. The annual number of sampled records and hospital response rates are provided in e-Table 5. Using the first-listed diagnosis, hospital discharges for COPD were identified by using the ICD-9-CM codes 490-492 or 496 as the first-listed diagnosis or ICD-9-CM code 466-466.1 (acute bronchitis) if the first-listed diagnosis of acute bronchitis was accompanied by another listed diagnosis of COPD (490-492 or 496). The percent of hospital records missing race information ranged from 16.0% to 31.0% (e-Table 5). US civilian population estimates used to calculate hospital discharge rates were obtained from the NHDS data documentation (e-Table 6). Relative SEs were calculated from the following formula: RSE(X) = (a + b/X)1/2, where a and b represent coefficients provided in the data documentation, and X represents the number of discharges.

population estimates used to calculate hospital discharge rates were obtained from the NHDS data documentation (e-Table 6). Relative SEs were calculated from the following formula: RSE(X) = (a + b/X)1/2, where a and b represent coefficients provided in the data documentation, and X represents the number of discharges. Medicare Part A Hospital Claims Medicare data from 1999 to 2010 were used to estimate the annual number of hospital discharges for COPD among Medicare enrollees aged ≥ 65 years. Hospitalization information from 100% of Medicare Part A hospital claims data were obtained from an administrative claims dataset maintained by the Centers for Medicare and Medicaid Services. Information was limited to approximately 10 million annual claims submitted for short-term fee-for-service hospital stays among Medicare enrollees aged ≥ 65 years residing in one of the 50 states or the District of Columbia in a given year. A hospital discharge for COPD was defined for a first-listed discharge diagnosis with ICD-9-CM codes 490-492 or 496—about 3% of annual Medicare claims. Few Medicare claims (< 0.05%) were submitted for acute bronchitis (ICD-9-CM code 466-466.1) with concomitant COPD; therefore, we did not include these discharges in our analyses. Race/ethnicity information on the claims data for Medicare enrollees represents information provided by most Medicare enrollees at the time of enrollment into the Medicare system or is information updated for older enrollees. Less than 0.5% of COPD claims were missing race information. State of residence was also obtained from the claims data. Medicare enrollment records were obtained from the Centers for Medicare and Medicaid Services and were used as the denominator file to calculate hospital rates after restricting the denominator to Medicare enrollees who met all the following criteria on July 1 of any given year (alive, aged ≥ 65 years, entitled to Part A benefits, residing in one of the 50 states or the District of Columbia, and not enrolled in a managed care plan).

enominator file to calculate hospital rates after restricting the denominator to Medicare enrollees who met all the following criteria on July 1 of any given year (alive, aged ≥ 65 years, entitled to Part A benefits, residing in one of the 50 states or the District of Columbia, and not enrolled in a managed care plan). National Vital Statistics System The number of deaths with COPD as the underlying cause for the years 1999 to 2010 come from the NVSS and are made available from CDC’s WONDER system (Compressed Mortality File).19 This interactive Web-based tool allows queries to obtain numbers of death for underlying causes, crude death rates, age-adjusted death rates, 95% CIs, and SEs for groups defined by various characteristics including year, place of residence (state, county, region, or division), sex, age group, race, and Hispanic origin.20 Data from the NVSS are based on information from all resident death certificates filed in the 50 States and the District of Columbia. Cause-of-death statistics presented in this report are classified in accordance with the International Classification of Diseases, Tenth Revision (ICD-10). ICD-10 codes J40-J44 were used to identify deaths from COPD as the underlying cause of death. These causes include chronic bronchitis (J40-J42), emphysema (J43), and other COPD (J44).

istics presented in this report are classified in accordance with the International Classification of Diseases, Tenth Revision (ICD-10). ICD-10 codes J40-J44 were used to identify deaths from COPD as the underlying cause of death. These causes include chronic bronchitis (J40-J42), emphysema (J43), and other COPD (J44). Mortality rates were calculated by using population estimates produced by the Bureau of the Census in collaboration with the National Center for Health Statistics.20 The 1999 population estimates are US Census Bureau bridged-race intercensal estimates of the July 1 resident population, based on the 1990 census and the bridged-race 2000 census. The 2000 and 2010 population estimates are April 1 modified 2000 and 2010 census counts with bridged-race categories, whereas the 2001 to 2009 population estimates are bridged-race intercensal estimates of July 1 resident populations, based on the year 2000 and the year 2010 census counts (released by CDC on October 26, 2012). Age-adjusted death rates for 2001 to 2009 may vary from previous reports because of the 2012 revision of the 2001 to 2009 population denominator estimates.

are bridged-race intercensal estimates of July 1 resident populations, based on the year 2000 and the year 2010 census counts (released by CDC on October 26, 2012). Age-adjusted death rates for 2001 to 2009 may vary from previous reports because of the 2012 revision of the 2001 to 2009 population denominator estimates. Data Analysis SAS-callable SUDAAN (Research Triangle Institute) was used to obtain weighted US estimates and prevalence from NHIS and state-specific and US estimates and prevalence from BRFSS. SAS or SAS-callable SUDAAN analyses for data from NAMCS, NHAMCS, and NHDS were weighted to obtain national US estimates. SAS was also used to obtain the number of COPD hospital discharges from Medicare hospital claims. The reported numbers of deaths, age-specific death rates, and age-adjusted death rates from COPD were obtained from CDC WONDER.19 Estimates were produced for all adults aged ≥ 25 years as well as for groups defined by age (25-44, 45-54, 55-64, 65-74, and ≥ 75 years), sex, and race/ethnicity. Racial/ethnic categories varied between surveillance systems because of differences in Medicare definitions of race/ethnicity categories; absence of racial/ethnic information on many medical records abstracted for NAMCS, NHAMCS, and NHDS; or small numbers of NHIS respondents in some racial/ethnic categories in the population samples selected. Except for Medicare estimates, age-adjusted estimates were standardized to the 2000 standard US population aged ≥ 25 years using the direct method.21 Medicare estimates were age-standardized to the 2000 standard US population aged ≥ 65 years. Because of the well-known relationship between age and COPD and because of the aging of the US population, we calculated age-adjusted estimates of prevalence and rates. State-specific age-adjusted estimates for BRFSS prevalence, Medicare hospitalizations, and mortality for COPD were also obtained to examine geographic clustering of COPD burden.

relationship between age and COPD and because of the aging of the US population, we calculated age-adjusted estimates of prevalence and rates. State-specific age-adjusted estimates for BRFSS prevalence, Medicare hospitalizations, and mortality for COPD were also obtained to examine geographic clustering of COPD burden. The statistical significance of temporal trends for age-specific prevalence of COPD in NHIS was examined by using log-linear regression analysis with time as the independent variable; analyses for trends in the age-adjusted prevalence included age as a continuous variable. The statistical significance for linear trends in age-specific and age-adjusted rates of physician-office visits, ED visits, NHDS and Medicare hospitalizations, and mortality was examined using weighted least-squares regression, where the weights were the inverse of the squared SE.

ncluded age as a continuous variable. The statistical significance for linear trends in age-specific and age-adjusted rates of physician-office visits, ED visits, NHDS and Medicare hospitalizations, and mortality was examined using weighted least-squares regression, where the weights were the inverse of the squared SE. Results Prevalence (BRFSS Telephone Survey) After age adjustment, 6.5% of US adults (unadjusted prevalence, 6.8%) representing 13.7 million noninstitutionalized adults aged ≥ 25 years in 2011 were estimated to have a self-reported physician diagnosis of COPD based on a telephone survey (Table 1). The age-adjusted prevalence displayed a strong age gradient, and the age-adjusted prevalence was higher in women (7.3%) than in men (5.7%) and higher in American Indian/Alaska Natives (11.0%) than in non-Hispanic whites (6.9%), non-Hispanic blacks (6.5%), Hispanics (4.1%), and Asian/Pacific Islanders (2.5%). The age-adjusted prevalence varied between states (Table 2). The highest age-adjusted prevalence of COPD in 2011 was clustered in the southern states and along the Ohio River Valley (Fig 1). Table 1 —Estimated Number and Prevalence of Self-Reported, Physician-Diagnosed COPD (Ever COPD, Chronic Bronchitis, or Emphysema) Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, Behavioral Risk Factor Surveillance System, 2011

Results Prevalence (BRFSS Telephone Survey) After age adjustment, 6.5% of US adults (unadjusted prevalence, 6.8%) representing 13.7 million noninstitutionalized adults aged ≥ 25 years in 2011 were estimated to have a self-reported physician diagnosis of COPD based on a telephone survey (Table 1). The age-adjusted prevalence displayed a strong age gradient, and the age-adjusted prevalence was higher in women (7.3%) than in men (5.7%) and higher in American Indian/Alaska Natives (11.0%) than in non-Hispanic whites (6.9%), non-Hispanic blacks (6.5%), Hispanics (4.1%), and Asian/Pacific Islanders (2.5%). The age-adjusted prevalence varied between states (Table 2). The highest age-adjusted prevalence of COPD in 2011 was clustered in the southern states and along the Ohio River Valley (Fig 1). Table 1 —Estimated Number and Prevalence of Self-Reported, Physician-Diagnosed COPD (Ever COPD, Chronic Bronchitis, or Emphysema) Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, Behavioral Risk Factor Surveillance System, 2011 Characteristics Estimated No.a Age-Adjusted,b,c % Unadjusted, %c Race/ethnicity White, non-Hispanic 10,460,000 6.9 7.6 Black, non-Hispanic 1,418,000 6.5 6.4 Hispanic 1,030,000 4.1 3.6 Asian/Pacific Islander 173,000 2.5 2.2 American Indian/Alaska Native 247,000 11.0 11.5 Other, non-Hispanic 397,000 11.2 11.3 Sex Women 8,197,000 7.3 7.8 Men 5,681,000 5.7 5.8 Age group, y 25-44 2,755,000 … 3.4 45-54 2,913,000 … 6.6 55-64 3,263,000 … 9.2 65-74 2,719,000 … 12.1 ≥ 75 2,227,000 … 11.6 Total 13,724,000 6.5 6.8 a Numbers for each variable may not add to total because of rounding.

000 11.0 11.5 Other, non-Hispanic 397,000 11.2 11.3 Sex Women 8,197,000 7.3 7.8 Men 5,681,000 5.7 5.8 Age group, y 25-44 2,755,000 … 3.4 45-54 2,913,000 … 6.6 55-64 3,263,000 … 9.2 65-74 2,719,000 … 12.1 ≥ 75 2,227,000 … 11.6 Total 13,724,000 6.5 6.8 a Numbers for each variable may not add to total because of rounding. b Age-adjusted to the 2000 US standard population aged ≥ 25 y. c All relative SEs are ≤ 30%. Table 2 —Estimated Number and Prevalence of Self-Reported, Physician-Diagnosed COPD (Ever COPD, Chronic Bronchitis, or Emphysema) Among Adults Aged ≥ 25 Years, By State—United States, Behavioral Risk Factor Surveillance System, 2011

b Age-adjusted to the 2000 US standard population aged ≥ 25 y. c All relative SEs are ≤ 30%. Table 2 —Estimated Number and Prevalence of Self-Reported, Physician-Diagnosed COPD (Ever COPD, Chronic Bronchitis, or Emphysema) Among Adults Aged ≥ 25 Years, By State—United States, Behavioral Risk Factor Surveillance System, 2011 State Estimated No.a Age-Adjusted,b,c % Unadjusted, %c Alabama 330,000 9.9 10.4 Alaska 24,000 6.1 5.5 Arizona 253,000 5.8 6.1 Arkansas 171,000 8.1 8.9 California 1,073,000 4.7 4.9 Colorado 167,000 5.1 5.0 Connecticut 155,000 6.1 6.5 Delaware 35,000 5.4 5.8 District of Columbia 20,000 5.0 4.9 Florida 1,086,000 7.5 8.4 Georgia 462,000 7.4 7.4 Hawaii 43,000 4.5 4.7 Idaho 58,000 5.7 5.9 Illinois 549,000 6.4 6.6 Indiana 390,000 8.9 9.3 Iowa 109,000 5.0 5.5 Kansas 134,000 6.9 7.3 Kentucky 306,000 10.1 10.6 Louisiana 213,000 7.0 7.3 Maine 79,000 7.5 8.5 Maryland 239,000 6.1 6.2 Massachusetts 283,000 6.0 6.4 Michigan 574,000 8.2 8.8 Minnesota 148,000 4.1 4.2 Mississippi 170,000 8.6 9.0 Missouri 353,000 8.3 8.9 Montana 44,000 6.0 6.6 Nebraska 65,000 5.2 5.5 Nevada 143,000 7.9 8.1 New Hampshire 61,000 6.4 6.9 New Jersey 329,000 5.3 5.6 New Mexico 92,000 6.5 6.9 New York 822,000 6.0 6.3 North Carolina 458,000 6.9 7.3 North Dakota 21,000 4.5 5.0 Ohio 646,000 7.9 8.4 Oklahoma 225,000 8.6 9.3 Oregon 168,000 5.9 6.5 Pennsylvania 626,000 6.7 7.3 Rhode Island 49,000 6.5 7.0 South Carolina 252,000 7.7 8.2 South Dakota 31,000 5.2 5.9 Tennessee 391,000 8.6 9.2 Texas 928,000 6.0 5.9 Utah 68,000 4.4 4.3 Vermont 24,000 4.9 5.6 Virginia 363,000 6.6 6.8 Washington 205,000 4.4 4.5 West Virginia 124,000 8.8 9.7 Wisconsin 219,000 5.4 5.8 Wyoming 25,000 6.2 6.7 Total 13,724,000 6.5 6.8 a Numbers may not add to total because of rounding.

Dakota 31,000 5.2 5.9 Tennessee 391,000 8.6 9.2 Texas 928,000 6.0 5.9 Utah 68,000 4.4 4.3 Vermont 24,000 4.9 5.6 Virginia 363,000 6.6 6.8 Washington 205,000 4.4 4.5 West Virginia 124,000 8.8 9.7 Wisconsin 219,000 5.4 5.8 Wyoming 25,000 6.2 6.7 Total 13,724,000 6.5 6.8 a Numbers may not add to total because of rounding. b Age-adjusted to the 2000 US standard population aged ≥ 25 y. c All relative SEs are ≤ 30%. Figure 1. Age-adjusted prevalence (%) of self-reported physician-diagnosed COPD among adults aged ≥ 25 years, by state—United States, Behavioral Risk Factor Surveillance System, 2011. Prevalence (NHIS Interview Survey) During the period from 1999 to 2011, the estimated numbers (Table 3) and age-adjusted prevalence of COPD (Table 4) fluctuated. Prevalence increased among successive age groups up to age 65 years and older, and the age-adjusted prevalence was usually higher among non-Hispanic whites compared with non-Hispanic blacks or Hispanics. The annual age-adjusted prevalence was higher in women than in men (Fig 2). The highest age-adjusted prevalence for both men and women was observed in 2001. Despite substantial interyear variation in age-adjusted prevalence estimates, significant tests for linear trend suggested declines during 1999 to 2011 in the age-adjusted prevalence among all adults (P = .019) and adults aged 25 to 44 years (P < .001).

e-adjusted prevalence for both men and women was observed in 2001. Despite substantial interyear variation in age-adjusted prevalence estimates, significant tests for linear trend suggested declines during 1999 to 2011 in the age-adjusted prevalence among all adults (P = .019) and adults aged 25 to 44 years (P < .001). Table 3 —Estimated Annual Number of Adults Aged ≥ 25 Years With Self-Reported Physician-Diagnosed COPD (Lifetime Emphysema or Chronic Bronchitis During the Preceding 12 Months), by Race/Ethnicity, Sex, and Age Group—United States, National Health Interview Survey, 1999-2011

e-adjusted prevalence for both men and women was observed in 2001. Despite substantial interyear variation in age-adjusted prevalence estimates, significant tests for linear trend suggested declines during 1999 to 2011 in the age-adjusted prevalence among all adults (P = .019) and adults aged 25 to 44 years (P < .001). Table 3 —Estimated Annual Number of Adults Aged ≥ 25 Years With Self-Reported Physician-Diagnosed COPD (Lifetime Emphysema or Chronic Bronchitis During the Preceding 12 Months), by Race/Ethnicity, Sex, and Age Group—United States, National Health Interview Survey, 1999-2011 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Race White, non-Hispanic 8,193,000 8,792,000 10,034,000 8,449,000 8,050,000 8,792,000 8,751,000 9,105,000 7,789,000 9,275,000 9,902,000 9,153,000 9,038,000 Black, non-Hispanic 773,000 969,000 1,177,000 1,127,000 958,000 937,000 1,049,000 1,149,000 889,000 1,036,000 1,178,000 1,227,000 1,433,000 Hispanic 512,000 573,000 655,000 621,000 601,000 693,000 655,000 679,000 683,000 655,000 910,000 927,000 987,000 Other, non-Hispanic 224,000 182,000 273,000 253,000 214,000 261,000 236,000 376,000 285,000 324,000 347,000 323,000 441,000 Sex Women 6,126,000 6,717,000 7,550,000 6,514,000 6,168,000 6,750,000 6,677,000 6,891,000 5,849,000 7,266,000 7,682,000 7,066,000 7,658,000 Men 3,576,000 3,798,000 4,588,000 3,936,000 3,655,000 3,934,000 4,013,000 4,419,000 3,796,000 4,024,000 4,655,000 4,564,000 4,241,000 Age group, y 25-44 3,087,000 3,157,000 3,899,000 3,129,000 2,526,000 2,987,000 2,868,000 2,552,000 2,159,000 2,795,000 2,597,000 2,699,000 2,560,000 45-54 1,811,000 2,184,000 2,671,000 2,311,000 1,964,000 2,294,000 2,274,000 2,461,000 2,039,000 2,703,000 2,773,000 2,383,000 2,430,000 55-64 1,725,000 1,879,000 2,135,000 2,014,000 2,126,000 2,043,000 2,199,000 2,747,000 2,351,000 2,330,000 2,937,000 2,740,000 3,053,000 65-74 1,639,000 1,721,000 1,773,000 1,678,000 1,791,000 1,702,000 1,845,000 1,703,000 1,624,000 1,902,000 2,120,000 2,018,000 2,253,000 ≥ 75 1,439,000 1,573,000 1,661,000 1,318,000 1,414,000 1,658,000 1,504,000 1,847,000 1,473,000 1,560,000 1,910,000 1,790,000 1,604,000 Total 9,702,000 10,515,000 12,138,000 10,450,000 9,822,000 10,683,000 10,690,000 11,310,000 9,646,000 11,290,000 12,337,000 11,630,000 11,899,000 Numbers for each variable may not add to total because of rounding.

61,000 1,318,000 1,414,000 1,658,000 1,504,000 1,847,000 1,473,000 1,560,000 1,910,000 1,790,000 1,604,000 Total 9,702,000 10,515,000 12,138,000 10,450,000 9,822,000 10,683,000 10,690,000 11,310,000 9,646,000 11,290,000 12,337,000 11,630,000 11,899,000 Numbers for each variable may not add to total because of rounding. Table 4 —Estimated Annual Prevalence of Self-Reported Physician-Diagnosed COPD (Lifetime Emphysema or Chronic Bronchitis During the Preceding 12 Months) Among Adults Aged ≥ 25 Years, by Race/Ethnicity, Sex, and Age Group—United States, National Health Interview Survey, 1999-2011

61,000 1,318,000 1,414,000 1,658,000 1,504,000 1,847,000 1,473,000 1,560,000 1,910,000 1,790,000 1,604,000 Total 9,702,000 10,515,000 12,138,000 10,450,000 9,822,000 10,683,000 10,690,000 11,310,000 9,646,000 11,290,000 12,337,000 11,630,000 11,899,000 Numbers for each variable may not add to total because of rounding. Table 4 —Estimated Annual Prevalence of Self-Reported Physician-Diagnosed COPD (Lifetime Emphysema or Chronic Bronchitis During the Preceding 12 Months) Among Adults Aged ≥ 25 Years, by Race/Ethnicity, Sex, and Age Group—United States, National Health Interview Survey, 1999-2011 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 P for Linear Trend Racea White, non-Hispanic 6.1 6.6 7.5 6.2 5.7 6.3 6.1 6.3 5.4 6.4 6.7 6.1 6.0 .130 Black, non-Hispanic 4.3 5.4 6.3 6.0 5.0 4.8 5.2 5.4 4.2 4.6 5.3 5.5 6.2 .443 Hispanic 3.6 3.9 4.3 3.7 3.6 3.9 3.3 3.5 3.3 3.1 4.1 3.9 4.3 .805 Other, non-Hispanic 4.0 2.8 3.9 3.2 3.3 3.6 2.9 4.1 2.9 3.1 3.4 3.1 3.9 .626 Sexa Women 6.7 7.3 8.1 6.9 6.3 6.8 6.6 6.7 5.6 6.9 7.1 6.5 7.0 .136 Men 4.6 4.8 5.6 4.8 4.3 4.5 4.6 4.9 4.1 4.3 4.9 4.7 4.3 .063 Age group, y 25-44 3.7 3.9 4.8 3.9 3.1 3.6 3.5 3.1 2.6 3.4 3.2 3.3 3.2 < .001 45-54 5.1 5.9 7.0 5.9 4.9 5.6 5.4 5.7 4.7 6.2 6.3 5.4 5.6 .655 55-64 7.5 8.0 8.8 7.9 7.7 7.1 7.3 8.8 7.2 7.0 8.4 7.7 8.2 .929 65-74 9.2 9.6 10.0 9.5 9.9 9.3 10.0 8.9 8.4 9.6 10.3 9.5 10.3 .566 ≥ 75 9.8 10.6 11.0 8.6 8.8 10.2 9.1 11.1 8.7 9.0 11.1 10.3 9.0 .679 Totala 5.7 6.1 6.9 5.9 5.3 5.7 5.6 5.8 4.9 5.6 6.0 5.7 5.7 .019 Totalb 5.6 6.0 6.9 5.9 5.3 5.7 5.6 5.9 5.0 5.8 6.2 5.8 5.9 .372 Annual prevalence per 100 population. All relative SEs are ≤ 30%.

2 9.6 10.0 9.5 9.9 9.3 10.0 8.9 8.4 9.6 10.3 9.5 10.3 .566 ≥ 75 9.8 10.6 11.0 8.6 8.8 10.2 9.1 11.1 8.7 9.0 11.1 10.3 9.0 .679 Totala 5.7 6.1 6.9 5.9 5.3 5.7 5.6 5.8 4.9 5.6 6.0 5.7 5.7 .019 Totalb 5.6 6.0 6.9 5.9 5.3 5.7 5.6 5.9 5.0 5.8 6.2 5.8 5.9 .372 Annual prevalence per 100 population. All relative SEs are ≤ 30%. a Age-adjusted to the 2000 US standard population aged ≥ 25 y. b Unadjusted prevalence. Figure 2. Age-adjusted prevalence (%) of self-reported physician-diagnosed COPD among adults aged ≥ 25 years, by sex and year—United States, National Health Interview Survey, 1999-2011. Physician Office Visits (NAMCS) In 2010, there were an estimated 10.3 million (unadjusted, 516.1 per 10,000 US civilian population; age-adjusted, 494.8 per 10,000 US civilian population) physician office visits with a first-listed diagnosis of COPD among adults aged ≥ 25 years. The age-adjusted rate of office visits for COPD was higher among men than women in 2010 (Fig 3) and higher among whites than blacks during 2009 to 2010 (Fig 4). There was considerable temporal variability in the estimated number of physician-based office visits (Table 5). As expected for a chronic disease, age-specific rates for office visits for COPD increased substantially within each given year (Table 6), and age-specific rates declined during 1999 to 2010 among those aged 45 to 54 years (P = .033). No clear time trend was evident for age-adjusted rates among any group defined by sex or race (Table 6).

hronic disease, age-specific rates for office visits for COPD increased substantially within each given year (Table 6), and age-specific rates declined during 1999 to 2010 among those aged 45 to 54 years (P = .033). No clear time trend was evident for age-adjusted rates among any group defined by sex or race (Table 6). Figure 3. Sex-specific age-adjusted rates (per 10,000 US civilian population) of physician office visits, ED visits, and hospital visits for COPD as the first-listed diagnosis among adults aged ≥ 25 years—United States, National Ambulatory Medical Care Survey, National Hospital Ambulatory Medical Care Survey, National Hospital Discharge Survey, 2010. Figure 4. Race-specific age-adjusted rates (per 10,000 US civilian population) of physician office visits, ED visits, and hospital visits for COPD as the first-listed diagnosis among adults aged ≥ 25 years—United States, National Ambulatory Medical Care Survey, National Hospital Ambulatory Medical Care Survey, National Hospital Discharge Survey, 2009-2010. Table 5 —Estimated Annual Number of Physician Office Visits for COPD as the First-Listed Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Ambulatory Medical Care Survey, 1999-2010

Figure 4. Race-specific age-adjusted rates (per 10,000 US civilian population) of physician office visits, ED visits, and hospital visits for COPD as the first-listed diagnosis among adults aged ≥ 25 years—United States, National Ambulatory Medical Care Survey, National Hospital Ambulatory Medical Care Survey, National Hospital Discharge Survey, 2009-2010. Table 5 —Estimated Annual Number of Physician Office Visits for COPD as the First-Listed Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Ambulatory Medical Care Survey, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Racea White 9,138,000 6,996,000 9,907,000 13,800,000 10,485,000 10,034,000 12,684,000 11,236,000 8,853,000 9,009,000 11,434,000 8,527,000 Black … … … … … … … … … … … … Sex Women 6,080,000 4,041,000 6,260,000 9,391,000 6,667,000 6,878,000 6,405,000 6,929,000 6,099,000 5,947,000 8,001,000 5,210,000 Men 4,275,000 3,956,000 4,483,000 5,697,000 5,672,000 4,606,000 6,667,000 6,016,000 4,488,000 3,663,000 4,940,000 5,081,000 Age group, y 25-44 1,784,000 1,446,000 1,850,000 3,022,000 2,709,000 2,126,000 2,106,000 1,301,000 1,913,000 1,902,000 1,649,000 … 45-54 1,295,000 … … 1,970,000 2,405,000 1,599,000 … 1,758,000 1,409,000 1,005,000 2,158,000 … 55-64 2,276,000 … … 2,538,000 1,704,000 2,440,000 2,415,000 3,171,000 1,919,000 2,077,000 3,024,000 2,153,000 65-74 2,854,000 2,175,000 2,563,000 3,878,000 2,871,000 2,499,000 3,349,000 3,418,000 3,191,000 1,895,000 2,303,000 3,191,000 ≥ 75 2,147,000 2,084,000 2,920,000 3,680,000 2,649,000 2,820,000 3,522,000 3,298,000 2,154,000 2,730,000 3,808,000 3,431,000 Total 10,355,000 7,997,000 10,743,000 15,087,000 12,339,000 11,484,000 13,072,000 12,945,000 10,586,000 9,609,000 12,941,000 10,291,000 Numbers for each variable may not add to total because of rounding. COPD includes ICD-9-CM codes 490-492 or 496. Ellipses indicate unreliable estimate (relative SE > 30% and/or number of records < 30). ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification.

,000 12,941,000 10,291,000 Numbers for each variable may not add to total because of rounding. COPD includes ICD-9-CM codes 490-492 or 496. Ellipses indicate unreliable estimate (relative SE > 30% and/or number of records < 30). ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification. a Data not available for other specific race groups. Table 6 —Estimated Annual Rate of Physician Office Visits for COPD as the First-Listed Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Ambulatory Medical Care Survey, 1999-2010

,000 12,941,000 10,291,000 Numbers for each variable may not add to total because of rounding. COPD includes ICD-9-CM codes 490-492 or 496. Ellipses indicate unreliable estimate (relative SE > 30% and/or number of records < 30). ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification. a Data not available for other specific race groups. Table 6 —Estimated Annual Rate of Physician Office Visits for COPD as the First-Listed Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Ambulatory Medical Care Survey, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Racea,b White 623.4 473.1 651.0 897.2 676.1 633.4 791.7 679.1 541.6 537.4 661.1 481.0 .585 Black … … … … … … … … … … … … … Sexa Women 660.2 432.4 647.4 955.5 678.8 679.0 622.1 656.2 587.7 557.3 726.3 458.3 .725 Men 572.0 519.5 583.3 714.6 701.4 547.5 789.3 688.6 515.6 410.5 525.4 554.4 .233 Age group, y 25-44 216.2 176.3 223.0 364.8 329.5 258.9 256.8 158.5 234.2 234.0 203.8 … .082 45-54 363.7 … … 495.5 594.3 387.3 … 409.5 323.7 228.4 488.0 … .033 55-64 987.5 … … 960.0 614.4 843.8 799.8 1,009.1 589.7 619.7 873.5 600.7 .826 65-74 1,603.7 1,224.9 1,417.7 2,150.8 1,586.8 1,371.6 1,820.3 1,830.4 1,670.3 953.4 1,120.6 1,505.7 .264 ≥ 75 1,464.0 1,393.7 1,867.6 2,309.7 1,636.7 1,716.1 2,102.2 1,946.0 1,257.2 1,573.2 2,189.8 1,936.1 .380 Totala 609.2 466.6 604.8 836.2 673.5 614.2 691.6 663.1 543.0 483.3 632.3 494.8 .541 Totalc 596.4 456.2 594.5 824.5 668.0 614.2 689.9 674.1 545.8 490.1 654.6 516.1 .848 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. Ellipses indicate unreliable estimate (relative SE > 30% and/or number of records < 30). See Table 5 legend for expansion of abbreviation.

.541 Totalc 596.4 456.2 594.5 824.5 668.0 614.2 689.9 674.1 545.8 490.1 654.6 516.1 .848 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. Ellipses indicate unreliable estimate (relative SE > 30% and/or number of records < 30). See Table 5 legend for expansion of abbreviation. a Age-adjusted to the 2000 US standard population aged ≥ 25 y. b Data not available for other specific race groups. c Unadjusted rate. ED Visits (NHAMCS) In 2010, there were an estimated 1.5 million (unadjusted rate, 73.6 per 10,000 US civilian population; age-adjusted rate, 72.0 per 10,000 US civilian population) ED visits with a first-listed diagnosis of COPD among adults aged ≥ 25 years. The age-adjusted rate of ED visits for COPD was higher among women than men in 2010 (Fig 3) and among blacks than whites during 2009 to 2010 (Fig 4). The estimated annual number of ED visits for COPD fluctuated (Table 7). There was a considerable increase each year in age-specific rates for ED visits with advancing age up to ages 65 years and older (Table 8), but there were no significant temporal trends during 1999 to 2010 in age-specific and age-adjusted rates for any group defined by age, race, or sex. Table 7 —Estimated Annual Numbers of ED Visits for COPD as the First-Listed Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Hospital Ambulatory Medical Care Survey, 1999-2010

ED Visits (NHAMCS) In 2010, there were an estimated 1.5 million (unadjusted rate, 73.6 per 10,000 US civilian population; age-adjusted rate, 72.0 per 10,000 US civilian population) ED visits with a first-listed diagnosis of COPD among adults aged ≥ 25 years. The age-adjusted rate of ED visits for COPD was higher among women than men in 2010 (Fig 3) and among blacks than whites during 2009 to 2010 (Fig 4). The estimated annual number of ED visits for COPD fluctuated (Table 7). There was a considerable increase each year in age-specific rates for ED visits with advancing age up to ages 65 years and older (Table 8), but there were no significant temporal trends during 1999 to 2010 in age-specific and age-adjusted rates for any group defined by age, race, or sex. Table 7 —Estimated Annual Numbers of ED Visits for COPD as the First-Listed Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Hospital Ambulatory Medical Care Survey, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Racea White 1,205,000 1,278,000 1,039,000 1,046,000 1,230,000 957,000 1,112,000 973,000 1,067,000 1,295,000 1,452,000 1,179,000 Black 285,000 243,000 226,000 229,000 273,000 172,000 342,000 327,000 207,000 265,000 272,000 253,000 Sex Women 802,000 898,000 717,000 769,000 907,000 619,000 841,000 729,000 842,000 861,000 1,029,000 945,000 Men 730,000 651,000 582,000 523,000 648,000 528,000 647,000 587,000 455,000 726,000 734,000 523,000 Age group, y 25-44 448,000 481,000 488,000 418,000 372,000 356,000 492,000 314,000 397,000 358,000 446,000 388,000 45-54 270,000 194,000 193,000 183,000 294,000 151,000 215,000 293,000 255,000 277,000 297,000 284,000 55-64 269,000 315,000 197,000 226,000 256,000 184,000 268,000 254,000 212,000 321,000 293,000 290,000 65-74 233,000 267,000 207,000 219,000 317,000 253,000 201,000 251,000 234,000 321,000 388,000 286,000 ≥ 75 312,000 292,000 212,000 246,000 315,000 202,000 311,000 204,000 198,000 311,000 340,000 221,000 Total 1,532,000 1,549,000 1,299,000 1,292,000 1,555,000 1,147,000 1,488,000 1,316,000 1,297,000 1,588,000 1,763,000 1,468,000 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation.

,000 311,000 204,000 198,000 311,000 340,000 221,000 Total 1,532,000 1,549,000 1,299,000 1,292,000 1,555,000 1,147,000 1,488,000 1,316,000 1,297,000 1,588,000 1,763,000 1,468,000 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. a Data not available for other specific race groups. Table 8 —Estimated Annual Rate of ED Visits for COPD as the First-Listed Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Hospital Ambulatory Medical Care Survey, 1999-2010

,000 311,000 204,000 198,000 311,000 340,000 221,000 Total 1,532,000 1,549,000 1,299,000 1,292,000 1,555,000 1,147,000 1,488,000 1,316,000 1,297,000 1,588,000 1,763,000 1,468,000 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. a Data not available for other specific race groups. Table 8 —Estimated Annual Rate of ED Visits for COPD as the First-Listed Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Hospital Ambulatory Medical Care Survey, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Racea,b White 82.7 86.7 69.0 68.4 78.9 61.4 70.6 60.4 66.0 77.8 86.9 70.6 .541 Black 160.8 133.0 120.8 120.0 143.4 87.7 166.9 153.2 94.4 121.1 123.4 112.6 .411 Sexa Women 87.7 96.9 75.0 79.0 91.5 62.5 83.5 71.3 82.1 82.4 97.9 88.0 .769 Men 95.2 82.4 70.4 62.4 77.0 62.5 74.3 64.9 49.1 78.4 79.3 54.5 .072 Age group, y 25-44 54.3 58.6 58.9 50.4 45.2 43.4 60.0 38.2 48.7 44.1 55.0 48.2 .166 45-54 75.8 52.5 49.7 46.0 72.5 36.5 51.1 68.2 58.7 62.8 67.1 64.4 .281 55-64 116.8 133.5 78.4 85.4 92.4 63.8 88.8 80.9 65.1 95.8 84.7 80.8 .478 65-74 130.9 150.3 114.7 121.6 175.4 139.1 109.2 134.7 122.2 161.3 188.6 135.0 .505 ≥ 75 212.9 195.2 135.8 154.6 194.9 123.0 185.8 120.4 115.8 179.2 195.5 124.5 .142 Totala 89.8 89.6 72.5 71.2 84.6 61.9 78.8 67.8 66.8 79.5 88.3 72.0 .432 Totalc 88.3 88.4 71.9 70.6 84.2 61.3 78.5 68.5 66.9 81.0 89.2 73.6 .428 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. All relative SEs are ≤ 30%. See Table 5 legend for expansion of abbreviation.

5 .142 Totala 89.8 89.6 72.5 71.2 84.6 61.9 78.8 67.8 66.8 79.5 88.3 72.0 .432 Totalc 88.3 88.4 71.9 70.6 84.2 61.3 78.5 68.5 66.9 81.0 89.2 73.6 .428 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. All relative SEs are ≤ 30%. See Table 5 legend for expansion of abbreviation. a Age-adjusted to the 2000 US standard population aged ≥ 25 y. b Data not available for other specific race groups. c Unadjusted rate. Hospitalizations (NHDS) In 2010, there were an estimated 699,000 hospitalizations (unadjusted rate, 34.4 per 10,000 US civilian population; age-adjusted rate, 32.2 per 10,000 US civilian population) for COPD as the first-listed diagnosis among adults aged ≥ 25 years. Age-adjusted rates of hospitalizations for COPD varied little between men and women in 2010 (Fig 3) or between blacks and whites during 2009 to 2010 (Fig 4). The annual number of hospitalizations for COPD fluctuated between 1999 and 2010 (Table 9). The age-specific hospital rates for COPD increased with advancing age each year (Fig 5), and there was a decline in age-specific rates during 1999 to 2010 among adults aged 25 to 44 years (P = .039), adults aged 55 to 64 years (P = .001), adults aged 65 to 74 years (P = .005), and adults aged ≥ 75 years (P = .018) (Table 10). Declining trends for age-adjusted rates for COPD hospitalization during 1999 to 2010 were observed among all adults (P = .001), men (P < .001), and women (P = .022) (Table 10).

ars (P = .039), adults aged 55 to 64 years (P = .001), adults aged 65 to 74 years (P = .005), and adults aged ≥ 75 years (P = .018) (Table 10). Declining trends for age-adjusted rates for COPD hospitalization during 1999 to 2010 were observed among all adults (P = .001), men (P < .001), and women (P = .022) (Table 10). Table 9 —Estimated Annual Number of Hospitalizations for COPD as the First-Listed Discharge Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Hospital Discharge Survey, 1999-2010

ars (P = .039), adults aged 55 to 64 years (P = .001), adults aged 65 to 74 years (P = .005), and adults aged ≥ 75 years (P = .018) (Table 10). Declining trends for age-adjusted rates for COPD hospitalization during 1999 to 2010 were observed among all adults (P = .001), men (P < .001), and women (P = .022) (Table 10). Table 9 —Estimated Annual Number of Hospitalizations for COPD as the First-Listed Discharge Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Hospital Discharge Survey, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Racea White 500,000 441,000 457,000 462,000 460,000 429,000 505,000 452,000 440,000 493,000 573,000 543,000 Black 59,000 47,000 54,000 53,000 49,000 45,000 50,000 55,000 52,000 59,000 62,000 80,000 Sex Women 402,000 350,000 362,000 368,000 369,000 336,000 387,000 344,000 345,000 414,000 416,000 398,000 Men 300,000 297,000 288,000 293,000 304,000 291,000 324,000 312,000 294,000 296,000 312,000 301,000 Age group, y 25-44 28,000 28,000 21,000 27,000 24,000 19,000 21,000 23,000 25,000 20,000 18,000 17,000 45-54 59,000 64,000 62,000 66,000 70,000 72,000 81,000 72,000 75,000 75,000 86,000 81,000 55-64 134,000 129,000 115,000 132,000 128,000 123,000 145,000 133,000 126,000 138,000 150,000 145,000 65-74 219,000 188,000 190,000 200,000 189,000 173,000 193,000 183,000 179,000 193,000 211,000 205,000 ≥ 75 261,000 239,000 263,000 236,000 263,000 241,000 271,000 245,000 234,000 284,000 262,000 251,000 Total 702,000 647,000 650,000 662,000 673,000 628,000 711,000 657,000 639,000 710,000 728,000 699,000 Numbers for each variable may not add to total because of rounding. COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation.

241,000 271,000 245,000 234,000 284,000 262,000 251,000 Total 702,000 647,000 650,000 662,000 673,000 628,000 711,000 657,000 639,000 710,000 728,000 699,000 Numbers for each variable may not add to total because of rounding. COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. a Data not available for other specific race groups. Race was not imputed. Percent missing data for race are shown in e-Table 5. Figure 5. Age-specific rates (per 10,000 US civilian population) of hospitalizations for COPD as the first-listed discharge diagnosis among adults aged ≥ 25 years, by year—United States, National Hospital Discharge Survey, 1999-2010. Table 10 —Estimated Annual Rates of Hospitalizations for COPD as the First-Listed Discharge Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Hospital Discharge Survey, 1999-2010

Figure 5. Age-specific rates (per 10,000 US civilian population) of hospitalizations for COPD as the first-listed discharge diagnosis among adults aged ≥ 25 years, by year—United States, National Hospital Discharge Survey, 1999-2010. Table 10 —Estimated Annual Rates of Hospitalizations for COPD as the First-Listed Discharge Diagnosis Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, National Hospital Discharge Survey, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Racea,b White 32.7 28.5 28.5 28.6 28.0 25.8 29.8 26.3 25.3 27.6 31.6 29.5 .104 Black 37.6 29.2 29.2 30.4 28.5 24.9 26.9 29.9 26.2 28.8 30.2 39.5 .563 Sexa Women 40.8 35.3 35.3 35.3 34.5 31.1 35.2 30.9 30.4 35.6 35.3 33.4 .022 Men 39.9 39.0 39.0 36.7 37.4 35.3 38.4 36.4 33.1 32.9 33.9 31.6 < .001 Age group, y 25-44 3.3 3.3 2.4 3.2 2.9 2.3 2.5 2.8 3.0 2.4 2.2 2.1 .039 45-54 16.5 17.3 15.9 16.5 17.1 17.2 19.0 16.7 17.1 16.9 19.4 18.4 .102 55-64 57.9 54.0 45.5 49.7 45.8 42.3 47.8 42.1 38.6 41.0 43.1 40.2 .001 65-74 121.6 104.7 103.6 109.7 103.0 93.9 103.7 96.8 92.7 96.0 101.5 95.5 .005 ≥ 75 161.3 144.8 154.7 136.1 149.7 135.0 149.1 133.7 126.2 151.4 139.8 131.4 .018 Totala 40.2 36.6 36.6 35.5 35.5 32.5 36.1 32.8 31.4 34.1 34.3 32.2 .001 Totalc 39.7 36.2 35.4 35.5 35.8 33.0 36.8 33.6 32.4 35.6 36.2 34.4 .018 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. All relative SEs are ≤ 30%. See Table 5 legend for expansion of abbreviation. a Age-adjusted to the 2000 US standard population aged ≥ 25 y.

Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Racea,b White 32.7 28.5 28.5 28.6 28.0 25.8 29.8 26.3 25.3 27.6 31.6 29.5 .104 Black 37.6 29.2 29.2 30.4 28.5 24.9 26.9 29.9 26.2 28.8 30.2 39.5 .563 Sexa Women 40.8 35.3 35.3 35.3 34.5 31.1 35.2 30.9 30.4 35.6 35.3 33.4 .022 Men 39.9 39.0 39.0 36.7 37.4 35.3 38.4 36.4 33.1 32.9 33.9 31.6 < .001 Age group, y 25-44 3.3 3.3 2.4 3.2 2.9 2.3 2.5 2.8 3.0 2.4 2.2 2.1 .039 45-54 16.5 17.3 15.9 16.5 17.1 17.2 19.0 16.7 17.1 16.9 19.4 18.4 .102 55-64 57.9 54.0 45.5 49.7 45.8 42.3 47.8 42.1 38.6 41.0 43.1 40.2 .001 65-74 121.6 104.7 103.6 109.7 103.0 93.9 103.7 96.8 92.7 96.0 101.5 95.5 .005 ≥ 75 161.3 144.8 154.7 136.1 149.7 135.0 149.1 133.7 126.2 151.4 139.8 131.4 .018 Totala 40.2 36.6 36.6 35.5 35.5 32.5 36.1 32.8 31.4 34.1 34.3 32.2 .001 Totalc 39.7 36.2 35.4 35.5 35.8 33.0 36.8 33.6 32.4 35.6 36.2 34.4 .018 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. All relative SEs are ≤ 30%. See Table 5 legend for expansion of abbreviation. a Age-adjusted to the 2000 US standard population aged ≥ 25 y. b Data not available for other specific race groups. Race was not imputed. Percent missing data for race are shown in e-Table 5. c Unadjusted rate.

Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Racea,b White 32.7 28.5 28.5 28.6 28.0 25.8 29.8 26.3 25.3 27.6 31.6 29.5 .104 Black 37.6 29.2 29.2 30.4 28.5 24.9 26.9 29.9 26.2 28.8 30.2 39.5 .563 Sexa Women 40.8 35.3 35.3 35.3 34.5 31.1 35.2 30.9 30.4 35.6 35.3 33.4 .022 Men 39.9 39.0 39.0 36.7 37.4 35.3 38.4 36.4 33.1 32.9 33.9 31.6 < .001 Age group, y 25-44 3.3 3.3 2.4 3.2 2.9 2.3 2.5 2.8 3.0 2.4 2.2 2.1 .039 45-54 16.5 17.3 15.9 16.5 17.1 17.2 19.0 16.7 17.1 16.9 19.4 18.4 .102 55-64 57.9 54.0 45.5 49.7 45.8 42.3 47.8 42.1 38.6 41.0 43.1 40.2 .001 65-74 121.6 104.7 103.6 109.7 103.0 93.9 103.7 96.8 92.7 96.0 101.5 95.5 .005 ≥ 75 161.3 144.8 154.7 136.1 149.7 135.0 149.1 133.7 126.2 151.4 139.8 131.4 .018 Totala 40.2 36.6 36.6 35.5 35.5 32.5 36.1 32.8 31.4 34.1 34.3 32.2 .001 Totalc 39.7 36.2 35.4 35.5 35.8 33.0 36.8 33.6 32.4 35.6 36.2 34.4 .018 Annual rate per 10,000 US civilian population. COPD includes ICD-9-CM codes 490-492 or 496. All relative SEs are ≤ 30%. See Table 5 legend for expansion of abbreviation. a Age-adjusted to the 2000 US standard population aged ≥ 25 y. b Data not available for other specific race groups. Race was not imputed. Percent missing data for race are shown in e-Table 5. c Unadjusted rate. Medicare Hospitalizations (Medicare Part A Hospital Claims) In 2010, there were 312,654 (unadjusted rate, 11.11 per 1,000 Medicare enrollees aged ≥ 65 years; age-adjusted rate, 11.18 per 1,000 Medicare enrollees aged ≥ 65 years) hospital discharge claims for COPD as the first-listed diagnosis. The annual number of Medicare hospitalizations for COPD fluctuated during 1999 to 2010 (Table 11). Age-specific rates for those aged 65 to 74 years declined significantly (P = .033) (Table 12). Age-adjusted rates were highest among Native American enrollees and lowest among Asian enrollees in most years (Fig 6). Age-adjusted rates for Medicare hospitalizations for COPD declined during 1999 to 2010 for all enrollees overall (P = .045) and men (P = .022), but the decline was not significant for women (P = .138) or for specific race groups (Table 12).

tive American enrollees and lowest among Asian enrollees in most years (Fig 6). Age-adjusted rates for Medicare hospitalizations for COPD declined during 1999 to 2010 for all enrollees overall (P = .045) and men (P = .022), but the decline was not significant for women (P = .138) or for specific race groups (Table 12). Table 11 —Annual Number of Medicare Hospitalizations for COPD as the First-Listed Discharge Diagnosis Among Medicare Beneficiaries Aged ≥ 65 Years, by Race/Ethnicity, Sex, and Age Group—United States, Medicare Part A Hospital Claims, 1999-2010

tive American enrollees and lowest among Asian enrollees in most years (Fig 6). Age-adjusted rates for Medicare hospitalizations for COPD declined during 1999 to 2010 for all enrollees overall (P = .045) and men (P = .022), but the decline was not significant for women (P = .138) or for specific race groups (Table 12). Table 11 —Annual Number of Medicare Hospitalizations for COPD as the First-Listed Discharge Diagnosis Among Medicare Beneficiaries Aged ≥ 65 Years, by Race/Ethnicity, Sex, and Age Group—United States, Medicare Part A Hospital Claims, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Race/ethnicity White, non-Hispanic 311,551 282,944 286,225 288,338 280,631 255,896 277,529 266,810 234,796 277,693 265,149 273,918 Black, non-Hispanic 25,468 23,545 24,280 25,530 24,313 22,003 24,312 22,740 21,344 23,893 24,611 27,106 Hispanic 4,422 4,030 4,256 4,347 4,183 3,945 4,505 4,196 3,892 4,531 4,517 4,770 Native American 581 519 577 1,018 1,210 1,229 1,228 1,553 1,176 1,468 1,432 1,547 Asian 1,818 1,553 1,683 1,772 1,750 2,865 1,901 1,719 1,879 2,327 2,369 2,442 Sex Women 194,756 177,658 179,941 181,588 176,902 162,180 174,986 167,743 149,181 174,940 168,625 175,597 Men 154,141 139,843 141,875 143,795 139,164 126,169 138,283 132,445 116,711 138,212 132,266 137,057 Age group, y 65-74 152,179 136,721 138,118 137,777 136,354 122,701 131,321 125,471 111,455 130,057 128,891 134,072 ≥ 75 196,718 180,780 183,698 187,606 179,802 165,648 181,948 174,717 154,437 183,095 172,000 178,582 Total 348,897 317,501 321,816 325,383 316,156 288,349 313,269 300,188 265,892 313,152 300891 312,654 COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation.

130,057 128,891 134,072 ≥ 75 196,718 180,780 183,698 187,606 179,802 165,648 181,948 174,717 154,437 183,095 172,000 178,582 Total 348,897 317,501 321,816 325,383 316,156 288,349 313,269 300,188 265,892 313,152 300891 312,654 COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. Table 12 —Annual Rates of Medicare Hospitalizations for COPD Among Medicare Beneficiaries Aged ≥ 65 Years, by Race/Ethnicity, Sex, and Age Group—United States, Medicare Part A Hospital Claims, 1999-2010

130,057 128,891 134,072 ≥ 75 196,718 180,780 183,698 187,606 179,802 165,648 181,948 174,717 154,437 183,095 172,000 178,582 Total 348,897 317,501 321,816 325,383 316,156 288,349 313,269 300,188 265,892 313,152 300891 312,654 COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. Table 12 —Annual Rates of Medicare Hospitalizations for COPD Among Medicare Beneficiaries Aged ≥ 65 Years, by Race/Ethnicity, Sex, and Age Group—United States, Medicare Part A Hospital Claims, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Racea White, non-Hispanic 13.47 12.00 11.80 11.58 11.09 10.02 10.86 10.73 9.61 11.50 11.03 11.31 .052 Black, non-Hispanic 12.59 11.29 11.19 11.42 10.63 9.48 10.49 10.29 9.98 11.37 11.55 12.39 .831 Hispanic 13.39 10.89 10.68 10.24 9.30 8.12 9.09 8.77 8.14 9.66 9.44 9.73 .081 Native American 19.21 14.65 15.62 11.80 13.14 12.73 11.61 14.84 10.88 13.32 12.62 13.23 .394 Asian 8.42 5.50 5.42 5.20 4.75 6.96 4.36 3.92 4.11 4.91 4.81 4.77 .108 Sexa Women 12.41 11.22 11.08 10.90 10.50 9.52 10.28 10.17 9.22 10.97 10.64 10.99 .138 Men 14.81 13.14 12.83 12.53 11.81 10.54 11.47 11.23 9.99 11.91 11.34 11.56 .022 Age group, y 65-74 11.26 10.01 9.81 9.51 9.24 8.19 8.75 8.58 7.70 8.90 8.72 8.88 .033 ≥ 75 15.49 13.98 13.79 13.67 12.87 11.72 12.86 12.69 11.47 13.94 13.24 13.69 .175 Totala 13.28 11.91 11.71 11.49 10.97 9.88 10.71 10.55 9.50 11.31 10.87 11.18 .045 Totalb 13.31 11.94 11.74 11.53 11.00 9.91 10.74 10.58 9.51 11.29 10.83 11.11 .034 Annual rate per 1,000 Medicare beneficiaries, aged ≥ 65 y, alive, entitled to Medicare Part A, and not in a managed care plan on July 1 of the given year. COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation.

1 11.94 11.74 11.53 11.00 9.91 10.74 10.58 9.51 11.29 10.83 11.11 .034 Annual rate per 1,000 Medicare beneficiaries, aged ≥ 65 y, alive, entitled to Medicare Part A, and not in a managed care plan on July 1 of the given year. COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. a Age-adjusted to the 2000 US standard population aged ≥ 65 y. b Unadjusted rate. Figure 6. Race-specific age-adjusted rates (per 1,000 Medicare enrollees) of Medicare hospitalizations for COPD as the first-listed discharge diagnosis among Medicare enrollees aged ≥ 65 years, by year—United States, Medicare Part A hospital claims, 1999-2010.

a Age-adjusted to the 2000 US standard population aged ≥ 65 y. b Unadjusted rate. Figure 6. Race-specific age-adjusted rates (per 1,000 Medicare enrollees) of Medicare hospitalizations for COPD as the first-listed discharge diagnosis among Medicare enrollees aged ≥ 65 years, by year—United States, Medicare Part A hospital claims, 1999-2010. Medicare hospital claims data provide an opportunity to obtain state-specific estimates (Table 13). Changes in age-adjusted rates during 1999 to 2010 varied between states (Table 14). A comparison of state-specific Medicare hospital rates in 1999 to 2000 to those in 2009 to 2010 (Fig 7) demonstrates geographic clustering of the 10 states in 1999 to 2000, with the highest hospitalization rates (14.0-26.6 per 1,000 Medicare enrollees) along the Mississippi River and Ohio River valleys. By 2009 to 2010, there was a marked improvement in rates in many of those states. States with the highest age-adjusted Medicare hospitalization rates in 2009 to 2010 in Figure 7 are similar to those states in Figure 1, with the highest age-adjusted prevalence of COPD in 2011. Figure 8 shows that there were no significant increases in age-adjusted Medicare hospitalization rates in any state during 1999 to 2010 and identifies those states which have experienced no significant change or a significant decline (P < .05) during the past decade. Table 13 —Annual Number of Medicare Hospitalizations for COPD as the First-Listed Discharge Diagnosis Among Medicare Beneficiaries Aged ≥ 65 Years, by State—United States, Medicare Part A Hospital Claims, 1999-2010

Medicare hospital claims data provide an opportunity to obtain state-specific estimates (Table 13). Changes in age-adjusted rates during 1999 to 2010 varied between states (Table 14). A comparison of state-specific Medicare hospital rates in 1999 to 2000 to those in 2009 to 2010 (Fig 7) demonstrates geographic clustering of the 10 states in 1999 to 2000, with the highest hospitalization rates (14.0-26.6 per 1,000 Medicare enrollees) along the Mississippi River and Ohio River valleys. By 2009 to 2010, there was a marked improvement in rates in many of those states. States with the highest age-adjusted Medicare hospitalization rates in 2009 to 2010 in Figure 7 are similar to those states in Figure 1, with the highest age-adjusted prevalence of COPD in 2011. Figure 8 shows that there were no significant increases in age-adjusted Medicare hospitalization rates in any state during 1999 to 2010 and identifies those states which have experienced no significant change or a significant decline (P < .05) during the past decade. Table 13 —Annual Number of Medicare Hospitalizations for COPD as the First-Listed Discharge Diagnosis Among Medicare Beneficiaries Aged ≥ 65 Years, by State—United States, Medicare Part A Hospital Claims, 1999-2010 State 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Alabama 8,996 8,172 8,751 8,779 8,442 7,216 7,842 7,098 6,805 8,132 7,686 7,704 Alaska 391 404 414 378 387 341 320 408 290 355 335 373 Arizona 3,317 2,890 2,914 3,152 3,101 3,006 3,638 2,967 2,652 3,543 3,482 4,082 Arkansas 5,444 5,033 4,991 5,202 4,798 4,148 4,563 4,598 3,651 3,957 3,858 3,956 California 20,181 17,174 17,142 17,382 17,045 14,601 15,331 14,392 14,367 17,093 17,302 18,743 Colorado 2,830 2,264 2,332 2,317 2,350 2,013 2,337 2,406 1,954 2,535 2,117 2,397 Connecticut 3,032 2,964 3,125 3,315 3,276 3,227 3,487 3,359 3,300 3,886 3,698 3,768 Delaware 983 1,038 992 1,028 1,047 1,088 1,188 1,052 1,069 1,105 1,066 1,257 District of Columbia 463 458 443 418 391 343 438 363 336 445 404 451 Florida 23,513 21,200 22,448 23,194 22,096 22,210 22,732 21,121 19,590 23,592 23,926 24,952 Georgia 9,983 9,269 9,434 9,591 9,089 8,215 9,146 9,092 7,491 8,617 8,660 8,517 Hawaii 470 412 421 406 379 377 386 369 373 424 410 424 Idaho 1,249 1,001 994 930 852 659 729 884 567 609 616 603 Illinois 16,794 14,861 15,180 15,556 14,669 13,354 14,618 14,566 13,146 16,660 16,089 16,021 Indiana 10,357 9,831 9,774 9,551 9,109 8,178 9,334 9,334 7,543 9,624 8,835 8,682 Iowa 5,008 4,308 4,234 3,734 3,369 2,977 2,940 3,736 2,424 2,774 2,546 2,645 Kansas 3,822 3,321 3,247 3,205 3,282 2,759 3,066 3,542 2,440 2,749 2,450 2,441 Kentucky 10,655 9,496 9,959 9,839 9,730 8,546 9,544 9,752 7,731 9,423 9,188 8,986 Louisiana 6,520 6,031 6,373 6,418 6,228 5,258 5,887 5,155 4,490 5,389 5,150 5,287 Maine 2,380 2,477 2,056 1,863 1,837 1,635 1,761 1,938 1,438 1,625 1,712 1,588 Maryland 5,437 5,283 5,631 5,746 5,695 5,211 5,898 6,221 6,223 7,086 6,470 6,522 Massachusetts 7,890 7,447 7,117 7,053 7,170 6,754 7,277 7,108 6,828 8,437 8,185 8,689 Michigan 14,207 12,606 12,377 13,285 12,984 12,892 13,915 13,152 11,491 13,053 12,627 14,282 Minnesota 4,764 4,130 4,291 4,061 3,745 3,560 3,667 3,354 2,513 3,097 2,862 2,930 Mississippi 6,361 6,017 6,191 6,117 5,851 5,128 5,666 5,090 4,164 4,858 5,005 4,983 Missouri 8,527 7,535 8,037 7,709 7,455 7,023 7,909 7,254 6,082 7,451 6,884 6,932 Montana 1,306 1,142 1,203 1,072 1,015 874 867 1,100 704 759 718 717

,282 Minnesota 4,764 4,130 4,291 4,061 3,745 3,560 3,667 3,354 2,513 3,097 2,862 2,930 Mississippi 6,361 6,017 6,191 6,117 5,851 5,128 5,666 5,090 4,164 4,858 5,005 4,983 Missouri 8,527 7,535 8,037 7,709 7,455 7,023 7,909 7,254 6,082 7,451 6,884 6,932 Montana 1,306 1,142 1,203 1,072 1,015 874 867 1,100 704 759 718 717 Nebraska 2,108 1,713 1,542 1,452 1,443 1,208 1,590 1,981 1,337 1,693 1,516 1,705 Nevada 1,969 1,551 1,488 1,576 1,529 1,332 1,544 1,547 1,346 1,754 1,890 2,062 New Hampshire 1,329 1,450 1,361 1,305 1,335 1,452 1,318 1,640 1,358 1,626 1,541 1,499 New Jersey 10,193 9,420 9,584 10,325 10,742 10,172 11,049 10,265 9,882 11,328 10,836 11,166 New Mexico 1,569 1,309 1,342 1,400 1,356 1,138 1,430 1,214 1,116 1,348 1,316 1,418 New York 20,526 18,680 17,835 17,878 17,360 16,983 18,046 17,177 15,787 17,987 17,643 19,369 North Carolina 12,146 11,326 11,079 10,940 11,079 9,511 10,610 9,911 8,819 10,097 9,655 9,530 North Dakota 970 770 742 633 576 548 631 838 553 495 393 410 Ohio 18,466 16,823 17,282 17,601 17,255 15,239 16,863 16,229 14,217 16,242 14,712 15,278 Oklahoma 5,558 5,057 5,456 5,392 5,256 4,779 5,186 5,055 4,973 5,587 5,227 5,574 Oregon 2,329 1,925 2,073 2,201 2,029 1,672 1,646 1,768 1,354 1,469 1,544 1,551 Pennsylvania 19,100 17,406 16,841 17,294 16,401 15,426 16,677 14,169 12,866 14,900 13,724 15,694 Rhode Island 1,178 1,167 1,063 1,014 1,054 954 1,021 972 969 1,140 1,115 1,288 South Carolina 5,427 5,104 5,054 5,327 5,277 4,815 5,333 4,689 4,444 5,153 4,910 5,002 South Dakota 1,219 1,052 938 890 860 649 751 1,104 534 653 668 605 Tennessee 10,251 9,771 9,989 10,025 10,101 9,285 9,799 9,102 8,215 10,032 9,587 9,188 Texas 21,794 20,395 22,665 23,830 23,043 20,590 22,387 20,379 19,180 22,262 21,264 22,557 Utah 750 623 718 692 645 521 628 547 425 528 498 551 Vermont 719 774 632 629 601 618 568 615 458 499 549 548 Virginia 9,254 8,672 8,793 8,712 8,561 7,331 8,008 7,356 6,865 8,030 7,579 7,306 Washington 3,660 3,303 3,459 3,403 3,180 2,852 3,516 3,562 2,925 3,516 3,478 3,540 West Virginia 6,678 6,295 5,889 5,880 5,830 4,979 5,249 5,541 4,474 4,859 4,611 4,706 Wisconsin 6,064 5,573 5,353 5,143 4,747 4,290 4,441 4,548 3,639 4,088 3,828 3,735 Wyoming 760 578 567 540 504 412 492 568 494 588 526 440 Total 348,897 317,501 321,816 288,338 316,156 288,349 313,269 300,188 265,892 313,152 300,891 312,654 COPD includes ICD-9-CM codes 490-492 or 496.

5,249 5,541 4,474 4,859 4,611 4,706 Wisconsin 6,064 5,573 5,353 5,143 4,747 4,290 4,441 4,548 3,639 4,088 3,828 3,735 Wyoming 760 578 567 540 504 412 492 568 494 588 526 440 Total 348,897 317,501 321,816 288,338 316,156 288,349 313,269 300,188 265,892 313,152 300,891 312,654 COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. Table 14 —Age-Adjusted Annual Rates for Medicare Hospitalizations With COPD as the First-Listed Discharge Diagnosis Among Medicare Beneficiaries Aged ≥ 65 Years, by State—United States, Medicare Part A Hospital Claims, 1999-2010

5,249 5,541 4,474 4,859 4,611 4,706 Wisconsin 6,064 5,573 5,353 5,143 4,747 4,290 4,441 4,548 3,639 4,088 3,828 3,735 Wyoming 760 578 567 540 504 412 492 568 494 588 526 440 Total 348,897 317,501 321,816 288,338 316,156 288,349 313,269 300,188 265,892 313,152 300,891 312,654 COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. Table 14 —Age-Adjusted Annual Rates for Medicare Hospitalizations With COPD as the First-Listed Discharge Diagnosis Among Medicare Beneficiaries Aged ≥ 65 Years, by State—United States, Medicare Part A Hospital Claims, 1999-2010 Statea 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Alabama 18.10 16.42 17.33 17.04 16.17 13.84 15.11 13.99 13.56 16.73 16.15 15.68 .121 Alaska 12.62 12.31 12.26 10.89 10.60 9.05 8.01 9.85 6.59 7.76 6.88 7.49 < .001 Arizona 9.88 7.97 7.61 7.63 7.14 6.64 7.83 6.89 6.05 8.01 7.66 8.62 .541 Arkansas 16.14 14.84 14.70 14.67 13.44 11.53 12.54 13.02 10.70 11.67 11.36 11.53 < .001 California 11.33 9.44 9.06 8.51 7.98 6.71 6.95 6.56 6.49 7.57 7.54 8.04 .027 Colorado 11.93 9.22 8.96 8.35 8.17 6.79 7.71 8.01 6.44 8.27 6.80 7.49 .015 Connecticut 8.52 8.21 8.02 7.70 7.56 7.41 8.01 7.81 7.90 9.55 9.30 9.48 .057 Delaware 10.97 11.34 10.48 10.52 10.44 10.61 11.33 9.87 9.89 10.06 9.53 10.79 .060 District of Columbia 8.17 8.00 7.60 7.21 6.87 6.11 7.89 6.66 6.22 8.33 7.41 8.08 .736 Florida 13.32 11.69 11.77 11.52 10.65 10.57 10.88 10.58 9.88 12.02 12.15 12.55 .747 Georgia 14.76 13.50 13.29 13.28 12.35 10.70 11.67 11.98 10.00 11.25 11.25 11.63 .004 Hawaii 5.18 4.34 4.50 4.29 3.92 3.80 3.81 3.72 3.70 4.17 4.05 4.22 .078 Idaho 9.87 7.81 7.57 6.93 6.23 4.72 5.13 6.50 4.28 4.65 4.78 4.62 < .001 Illinois 13.49 11.91 11.91 11.72 10.89 9.88 10.79 10.94 9.95 12.53 11.94 11.71 .439 Indiana 14.92 14.09 13.85 13.38 12.57 11.21 12.73 13.09 10.88 14.05 13.11 12.83 .202 Iowa 12.18 10.46 10.28 9.04 8.16 7.26 7.24 9.54 6.31 7.18 6.65 6.91 < .001 Kansas 11.95 10.41 10.21 9.91 9.93 8.29 9.21 10.81 7.53 8.54 7.67 7.57 .001 Kentucky 23.83 21.10 21.76 21.07 20.58 17.87 19.67 20.45 16.93 20.80 20.34 19.77 .080 Louisiana 16.95 15.23 15.38 15.02 14.35 12.03 13.93 12.43 11.12 13.62 13.23 13.52 .015 Maine 13.60 14.05 11.48 10.28 10.05 8.83 9.44 10.31 7.65 8.75 9.67 9.03 .005 Maryland 11.52 10.63 10.26 10.35 10.17 9.23 10.36 11.10 11.06 12.40 10.91 10.80 .338 Massachusetts 13.15 12.28 11.53 11.28 10.85 10.14 10.95 10.66 10.31 12.72 12.30 12.93 .908 Michigan 12.71 11.24 11.00 11.32 10.93 10.74 11.53 11.24 10.99 13.38 13.43 13.05 .139 Minnesota 9.83 8.41 8.60 8.08 7.42 7.08 7.46 7.62 5.97 7.69 7.37 8.10 .036 Mississippi 19.86 18.74 19.27 18.73 17.72 15.34 16.89 16.06 13.05 15.16 15.51 15.19 < .001 Missouri 13.92 12.41 13.12 12.53 11.82 11.05 12.40 11.52 9.76 12.06 11.24 11.31

.71 11.24 11.00 11.32 10.93 10.74 11.53 11.24 10.99 13.38 13.43 13.05 .139 Minnesota 9.83 8.41 8.60 8.08 7.42 7.08 7.46 7.62 5.97 7.69 7.37 8.10 .036 Mississippi 19.86 18.74 19.27 18.73 17.72 15.34 16.89 16.06 13.05 15.16 15.51 15.19 < .001 Missouri 13.92 12.41 13.12 12.53 11.82 11.05 12.40 11.52 9.76 12.06 11.24 11.31 .014 Montana 11.44 9.73 10.13 8.95 8.36 7.10 6.98 9.27 6.10 6.63 6.34 6.26 < .001 Nebraska 9.92 7.90 7.11 6.66 6.64 5.54 7.32 9.32 6.38 8.14 7.26 8.09 .882 Nevada 15.60 11.66 10.85 10.78 9.79 8.31 9.31 9.20 7.83 9.83 10.37 10.85 .177 New Hampshire 10.37 10.12 9.35 8.84 8.83 9.41 8.40 10.25 8.42 10.01 9.48 9.08 .504 New Jersey 11.69 10.45 10.42 10.73 11.02 10.40 11.34 10.56 10.11 11.55 11.13 11.36 .576 New Mexico 10.62 8.60 8.26 8.52 8.10 6.73 8.29 7.24 6.59 7.89 7.68 8.09 .071 New York 11.64 10.52 9.94 9.92 9.64 9.09 9.79 9.62 9.05 10.52 10.46 11.56 .933 North Carolina 13.92 12.78 12.33 12.02 11.91 10.11 11.22 10.76 9.75 11.07 10.49 10.08 < .001 North Dakota 10.51 8.34 8.04 6.91 6.27 5.98 6.88 9.44 6.31 5.72 4.53 4.71 .003 Ohio 15.65 14.13 14.11 14.24 13.77 12.06 13.36 13.13 11.71 14.78 13.59 15.60 .494 Oklahoma 14.40 13.13 14.07 13.60 13.09 11.83 12.79 12.95 12.46 14.01 13.03 13.77 .524 Oregon 9.23 7.52 7.83 8.11 7.33 5.89 5.72 6.38 4.99 5.34 5.67 5.52 < .001 Pennsylvania 14.25 13.18 12.16 12.41 11.86 11.21 12.28 11.15 10.46 12.82 12.00 13.65 .289 Rhode Island 12.72 12.46 11.56 10.93 11.66 10.60 11.63 11.13 11.26 13.29 13.03 14.62 .223 South Carolina 12.28 11.30 10.98 11.38 11.06 9.88 10.73 9.62 9.24 10.73 10.11 10.04 .008 South Dakota 11.49 9.86 8.81 8.39 7.92 5.94 6.84 10.06 5.04 6.28 6.19 5.55 .004 Tennessee 16.34 15.49 15.54 15.68 15.75 14.43 15.28 14.82 13.66 16.73 15.82 15.12 .442 Texas 13.94 12.81 13.11 13.11 12.29 10.82 11.68 10.79 10.23 11.87 11.23 11.69 .009 Utah 4.44 3.60 4.07 3.85 3.50 2.76 3.31 3.22 2.61 3.37 3.29 3.62 .073 Vermont 9.92 10.33 8.37 8.24 7.79 7.89 7.18 7.62 5.65 6.08 6.61 6.48 < .001 Virginia 13.44 12.15 11.90 11.62 11.20 9.44 10.13 9.54 8.98 10.47 9.82 9.27 < .001 Washington 8.19 7.13 6.96 6.51 5.93 5.14 6.19 6.26 5.16 6.20 6.13 6.09 .065 West Virginia 27.38 25.81 24.10 23.95 23.62 20.11 21.06 22.62 21.20 23.50 22.52 22.50 .033 Wisconsin 9.32 8.56 8.16 7.60 7.04 6.43 6.84 7.54 6.35 7.41 7.20 7.12 .022 Wyoming 14.19 10.65 10.22 9.67 8.83 7.12 8.31 9.66 8.28 9.72 8.58 7.02 .023 Totala 13.28 11.91 11.71 11.49 10.97 9.88 10.71 10.55 9.50 11.31 10.87 11.18 .045 Annual rate per 1,000 Medicare enrollees aged ≥ 65 y, alive, entitled to Med

22.50 .033 Wisconsin 9.32 8.56 8.16 7.60 7.04 6.43 6.84 7.54 6.35 7.41 7.20 7.12 .022 Wyoming 14.19 10.65 10.22 9.67 8.83 7.12 8.31 9.66 8.28 9.72 8.58 7.02 .023 Totala 13.28 11.91 11.71 11.49 10.97 9.88 10.71 10.55 9.50 11.31 10.87 11.18 .045 Annual rate per 1,000 Medicare enrollees aged ≥ 65 y, alive, entitled to Med icare Part A, and not in a managed care plan on July 1 of the given year. COPD includes ICD-9-CM codes 490-492 or 496. See Table 5 legend for expansion of abbreviation. a Age-adjusted to the 2000 US standard population aged ≥ 65 y. Figure 7. Age-adjusted rates (per 1,000 Medicare enrollees) of Medicare hospitalizations for COPD as the first-listed discharge diagnosis among Medicare enrollees aged ≥ 65 years—United States, Medicare Part A hospital claims, 1999-2000 and 2009-2010. Figure 8. Significant linear change (P < .05) in state-specific age-adjusted rates (per 1,000 Medicare enrollees) of Medicare hospitalizations for COPD as the first-listed discharge diagnosis among Medicare enrollees aged ≥ 65 years—United States, Medicare Part A hospital claims, 1999-2010.

Figure 7. Age-adjusted rates (per 1,000 Medicare enrollees) of Medicare hospitalizations for COPD as the first-listed discharge diagnosis among Medicare enrollees aged ≥ 65 years—United States, Medicare Part A hospital claims, 1999-2000 and 2009-2010. Figure 8. Significant linear change (P < .05) in state-specific age-adjusted rates (per 1,000 Medicare enrollees) of Medicare hospitalizations for COPD as the first-listed discharge diagnosis among Medicare enrollees aged ≥ 65 years—United States, Medicare Part A hospital claims, 1999-2010. Deaths (Death Certificates) In 2010, there were 133,575 deaths (crude rate, 65.5 per 100,000 US population; age-adjusted rate, 63.1 per 100,000 population) among adults aged ≥ 25 years. Although the annual number of deaths increased somewhat during 1999 to 2010 (Table 15), the age-adjusted death rate for COPD declined during 1999 to 2010 among men (P = .001) but did not change significantly in women (P = .127) or overall (P = .163) (Table 16). Age-specific rates increased among adults aged 45 to 54 years (P < .001) but declined among those aged 55 to 64 years (P = .002) and 65 to 74 years (P < .001). The age-specific rates each year were highest among those aged ≥ 75 years and 65 to 74 years (Fig 9). Age-adjusted rates were highest among non-Hispanic whites followed by American Indian/Alaska Natives, non-Hispanic blacks, Hispanics, and Asian/Pacific Islanders (Fig 10). During 1999 to 2010, age-adjusted rates increased among American Indian/Alaska Natives (P = .008) and declined among Hispanics (P = .038) and Asian/Pacific Islanders (P < .001) but did not change significantly among non-Hispanic whites or non-Hispanic blacks.

nic blacks, Hispanics, and Asian/Pacific Islanders (Fig 10). During 1999 to 2010, age-adjusted rates increased among American Indian/Alaska Natives (P = .008) and declined among Hispanics (P = .038) and Asian/Pacific Islanders (P < .001) but did not change significantly among non-Hispanic whites or non-Hispanic blacks. Table 15 —Annual Number of Adults Aged ≥ 25 Years With COPD as the Underlying Cause of Death, by Race, Sex, and Age Group—United States, Mortality Component of the National Vital Statistics System, 1999-2010

nic blacks, Hispanics, and Asian/Pacific Islanders (Fig 10). During 1999 to 2010, age-adjusted rates increased among American Indian/Alaska Natives (P = .008) and declined among Hispanics (P = .038) and Asian/Pacific Islanders (P < .001) but did not change significantly among non-Hispanic whites or non-Hispanic blacks. Table 15 —Annual Number of Adults Aged ≥ 25 Years With COPD as the Underlying Cause of Death, by Race, Sex, and Age Group—United States, Mortality Component of the National Vital Statistics System, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Race/ethnicitya White, non-Hispanic 107,706 106,198 107,145 108,481 110,088 106,393 113,939 108,435 111,477 123,171 119,715 119,894 Black, non-Hispanic 6,640 6,327 6,355 6,585 6,565 6,274 7,086 6,660 6,896 7,743 7,489 7,700 Hispanic 2,488 2,341 2,512 2,724 2,827 2,779 3,166 2,994 3,238 3,623 3,672 3,817 AIAN 360 381 378 413 455 457 467 468 559 579 555 643 API 923 925 985 969 1,037 1,013 1,136 1,144 1,146 1,288 1,265 1,265 Sex Women 58,040 58,436 59,789 60,673 62,363 60,194 65,193 62,290 63,813 71,031 69,334 69,797 Men 60,416 58,058 57,908 58,807 58,904 56,940 60,812 57,633 59,678 65,606 63,583 63,778 Age group, y 25-44 494 500 554 586 573 554 558 519 521 566 506 453 45-54 2,472 2,618 2,695 2,842 2,883 2,920 3,356 3,326 3,596 3869 4,083 3,861 55-64 10,643 10,130 10,545 10,670 11,451 11,183 12,173 11,823 12,273 13518 13,636 13,674 65-74 31,699 30,249 29,942 29,040 29,241 27,740 29,296 27,640 28,100 31390 30,762 31,254 ≥ 75 73,148 72,997 73,961 76,342 77,119 74,737 80,622 76,615 79,001 87,294 83,930 84,333 Total 118,456 116,494 117,697 119,480 121,267 117,134 126,005 119,923 123,491 136,637 132,917 133,575 COPD includes ICD-10 codes J40–J44 from the WHO. AIAN = non-Hispanic American Indian/Alaska Natives; API = non-Hispanic Asian/Pacific Islanders; ICD-10 = International Classification of Diseases, tenth revision; WHO = World Health Organization.

697 119,480 121,267 117,134 126,005 119,923 123,491 136,637 132,917 133,575 COPD includes ICD-10 codes J40–J44 from the WHO. AIAN = non-Hispanic American Indian/Alaska Natives; API = non-Hispanic Asian/Pacific Islanders; ICD-10 = International Classification of Diseases, tenth revision; WHO = World Health Organization. a A summation of the annual numbers will not equal the total annual number because of small numbers of death in other race/ethnicity or unknown categories. Table 16 —Annual Rates for Deaths With COPD as the Underlying Cause Of Death Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, Mortality Component of the National Vital Statistics System, 1999-2010

a A summation of the annual numbers will not equal the total annual number because of small numbers of death in other race/ethnicity or unknown categories. Table 16 —Annual Rates for Deaths With COPD as the Underlying Cause Of Death Among Adults Aged ≥ 25 Years, by Race, Sex, and Age Group—United States, Mortality Component of the National Vital Statistics System, 1999-2010 Variable 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Race/ethnicitya White, non-Hispanic 72.0 70.5 70.4 70.5 70.7 67.7 71.5 67.2 68.1 74.1 71.0 70.2 .870 Black, non-Hispanic 45.9 43.0 42.6 43.6 42.5 39.8 43.8 40.3 40.8 44.6 41.7 41.8 .240 Hispanic 32.9 29.7 29.7 30.8 30.5 28.2 30.4 27.4 28.1 29.8 28.8 28.5 .038 AIAN 54.2 53.0 51.6 54.1 57.3 55.0 54.7 53.2 60.8 59.3 56.1 62.9 .008 API 25.7 24.8 24.2 22.2 22.7 21.2 22.1 21.0 19.8 21.0 19.7 19.0 < .001 Sexd Women 54.6 54.4 54.9 55.0 55.9 53.3 56.8 53.6 54.0 59.1 56.8 56.3 .127 Men 88.2 83.8 81.8 81.7 79.9 75.7 78.8 73.0 73.7 79.2 74.8 73.6 .001 Age group, y 25-44 0.6 0.6 0.7 0.7 0.7 0.7 0.7 0.6 0.6 0.7 0.6 0.6 … 45-54 6.8 6.9 6.8 7.1 7.1 7.0 7.9 7.7 8.2 8.7 9.1 8.6 < .001 55-64 44.8 41.7 42.0 40.0 40.9 38.2 39.7 37.0 37.0 39.6 38.5 37.5 .002 65-74 172.1 164.5 162.9 157.9 158.1 148.6 155.2 143.9 142.6 153.1 144.9 143.9 < .001 ≥ 75 446.6 439.7 437.5 445.6 444.2 426.2 453.7 426.6 435.8 477.7 456.4 454.5 .212 Totala 67.0 65.2 64.9 65.0 64.9 61.8 65.3 61.1 61.7 66.9 63.8 63.1 .163 Totalb 65.7 64.0 63.9 64.2 64.5 61.6 65.4 61.5 62.6 68.4 65.8 65.5 .515 Annual rate per 100,000 US population. COPD includes ICD-10 codes J40–J44 from the WHO. Death rates for 2001-2009 will differ from previous reports because 2001-2009 population denominators have been revised In CDC Wonder (Oct 2012). See Table 15 legend for expansion of abbreviations.

65.4 61.5 62.6 68.4 65.8 65.5 .515 Annual rate per 100,000 US population. COPD includes ICD-10 codes J40–J44 from the WHO. Death rates for 2001-2009 will differ from previous reports because 2001-2009 population denominators have been revised In CDC Wonder (Oct 2012). See Table 15 legend for expansion of abbreviations. a Age-adjusted to the 2000 US standard population aged ≥ 25 y. b Unadjusted rate. Figure 9. Age-specific death rates (per 100,000) for COPD as the underlying cause of death among adults aged ≥ 25 years, by year—United States, Mortality Component of the National Vital Statistics System, 1999-2010. Figure 10. Race-specific age-adjusted death rates (per 100,000) for COPD as the underlying cause of death among adults aged ≥ 25 years, by year—United States, Mortality Component of the National Vital Statistics System, 1999-2010.

Figure 9. Age-specific death rates (per 100,000) for COPD as the underlying cause of death among adults aged ≥ 25 years, by year—United States, Mortality Component of the National Vital Statistics System, 1999-2010. Figure 10. Race-specific age-adjusted death rates (per 100,000) for COPD as the underlying cause of death among adults aged ≥ 25 years, by year—United States, Mortality Component of the National Vital Statistics System, 1999-2010. Numbers of deaths (Table 17) and age-adjusted death rates varied during 1999 to 2010 in most states (Table 18). Figure 11 compares aggregated age-adjusted state-specific death rates for COPD in 1999 to 2000 to those for 2009 to 2010. In 1999 to 2000, states with the highest death rates (75.0-103.9 per 100,000) were along the Ohio River valley and in multiple western states. Geographic clustering of COPD death rates aggregated for 2009 to 2010 was observed in states along the Ohio River Valley and in several western states and also in several southern states (Fig 11). Although death rates for COPD declined in many states during 1999 to 2010, five states (Alabama, Mississippi, Arkansas, Oklahoma, and South Dakota) experienced significant increases in deaths from COPD (Fig 12). Table 17 —Annual Number of Adults Aged ≥ 25 Years With COPD as the Underlying Cause of Death, by State—United States, Mortality Component of the National Vital Statistics System, 1999-2010

Numbers of deaths (Table 17) and age-adjusted death rates varied during 1999 to 2010 in most states (Table 18). Figure 11 compares aggregated age-adjusted state-specific death rates for COPD in 1999 to 2000 to those for 2009 to 2010. In 1999 to 2000, states with the highest death rates (75.0-103.9 per 100,000) were along the Ohio River valley and in multiple western states. Geographic clustering of COPD death rates aggregated for 2009 to 2010 was observed in states along the Ohio River Valley and in several western states and also in several southern states (Fig 11). Although death rates for COPD declined in many states during 1999 to 2010, five states (Alabama, Mississippi, Arkansas, Oklahoma, and South Dakota) experienced significant increases in deaths from COPD (Fig 12). Table 17 —Annual Number of Adults Aged ≥ 25 Years With COPD as the Underlying Cause of Death, by State—United States, Mortality Component of the National Vital Statistics System, 1999-2010 State 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Alabama 2,095 1,964 2,125 2,245 2,347 2,256 2,318 2,239 2,463 2,655 2,693 2,784 Alaska 133 125 134 132 132 128 145 129 164 176 187 165 Arizona 2,414 2,410 2,382 2,462 2,448 2,328 2,723 2,695 2,605 2,860 2,773 2,836 Arkansas 1,296 1,328 1,297 1,383 1,439 1,375 1,508 1437 1,587 1,833 1,772 1,732 California 12,488 12,092 12,356 12,088 12,833 11,971 12,608 12,223 11,995 12,870 12,393 12,455 Colorado 1,791 1,715 1,762 1,773 1,864 1,823 1,864 1,873 1,948 2,128 2,012 2,143 Connecticut 1,366 1,469 1,429 1,385 1,395 1,375 1,421 1,400 1,310 1,461 1,389 1,237 Delaware 316 322 286 335 333 334 398 338 370 460 421 427 District of Columbia 150 160 138 125 126 144 123 114 122 133 127 139 Florida 8,815 8,345 8,621 8,738 8,778 8,703 9,173 8,668 9,092 9,957 9,891 10,076 Georgia 2,903 2,914 2,950 3,032 3,105 2,980 3,262 3,241 3,269 3,426 3,615 3,694 Hawaii 250 233 243 234 249 270 260 258 260 266 269 265 Idaho 542 549 562 574 575 544 691 624 639 682 699 701 Illinois 4,851 4,486 4,499 4,539 4,601 4,493 4,817 4,521 4,552 5,384 5,093 4,998 Indiana 2,915 2,948 3,053 3,032 3,167 3,030 3,365 3,193 3,130 3,768 3,649 3,697 Iowa 1,574 1,454 1,482 1,521 1,620 1,492 1,650 1,603 1,605 1,803 1,777 1,633 Kansas 1,323 1,351 1,393 1,328 1,407 1,272 1,529 1,447 1,437 1,581 1,537 1,532 Kentucky 2,260 2,090 2,204 2,339 2,327 2,202 2,507 2,331 2,577 2,874 2,791 2,721 Louisiana 1,525 1,591 1,684 1,598 1,660 1,547 1,830 1,630 1,633 1,831 1,826 1,895 Maine 730 755 773 769 762 744 813 763 717 774 801 788 Maryland 1,836 1,828 1,813 1,838 1,887 1,808 1,823 1,761 1,813 1,916 1,980 1,955 Massachusetts 2,729 2,799 2,699 2,630 2,668 2,466 2,529 2,457 2,260 2,510 2,481 2,306 Michigan 4,130 4,150 3,974 4,251 4,316 4,099 4,304 4,334 4,466 5,050 4,814 4,943 Minnesota 1,880 1,794 1,816 1,864 1,742 1,762 1,883 1,706 1,686 2,023 1,879 1,923 Mississippi 1,217 1,189 1,275 1,320 1,352 1,295 1,416 1,324 1,363 1,464 1,505 1,602 Missouri 2,949 2,692 2,781 2,765 2,836 2,628 3,002 2,922 2,990 3,663 3,354 3,453 Montana 538 499 559 554 568 563 569 561 595 681 588 586 Nebraska 893 794 831 885 851 774 897 845 882 1,009 945 976 Nevada 987 959 1,108 1,149 1,155 1,097

Mississippi 1,217 1,189 1,275 1,320 1,352 1,295 1,416 1,324 1,363 1,464 1,505 1,602 Missouri 2,949 2,692 2,781 2,765 2,836 2,628 3,002 2,922 2,990 3,663 3,354 3,453 Montana 538 499 559 554 568 563 569 561 595 681 588 586 Nebraska 893 794 831 885 851 774 897 845 882 1,009 945 976 Nevada 987 959 1,108 1,149 1,155 1,097 1,209 1,046 1,027 1,233 1,215 1,155 New Hampshire 578 560 596 553 509 581 611 589 593 676 640 594 New Jersey 2,993 2,874 2,761 2,737 2,775 2,895 3,009 2,732 2,881 3,159 3,010 2,998 New Mexico 810 725 734 815 893 722 824 857 852 973 955 996 New York 6,653 6,419 6,514 6,581 6,336 6,430 6,472 6,047 6,281 6,619 6,440 6,509 North Carolina 3,412 3,533 3,343 3,531 3,725 3,474 4,005 3,858 4,071 4,413 4,196 4,357 North Dakota 256 280 298 306 284 260 260 274 255 341 330 341 Ohio 5,656 5,773 5,686 5,840 5,739 5,727 6,406 5,871 6,263 6,771 6,479 6,520 Oklahoma 1,683 1,906 1,853 1,920 2,093 1,923 2,296 2,133 2,333 2,645 2,539 2,679 Oregon 1,664 1,599 1,646 1,754 1,738 1,711 1,767 1,730 1,816 1,868 1,847 1,888 Pennsylvania 5,922 5,837 5,646 5,797 5,816 5,774 5,935 5,420 5,871 6,531 6,254 6,025 Rhode Island 478 493 494 509 474 447 502 467 406 462 500 498 South Carolina 1,675 1,645 1,640 1,805 1,819 1,699 1,879 1,854 1,949 2,176 2,245 2,175 South Dakota 315 364 335 364 365 375 424 363 437 474 431 430 Tennessee 2,655 2,765 2,826 2,874 2,939 2,885 3,076 2,875 3,064 3,462 3,408 3,460 Texas 7,139 6,960 7,404 7,400 7,264 7,110 7,666 7,334 7,814 8,605 8,365 8,667 Utah 523 482 483 555 525 549 559 548 589 604 547 637 Vermont 288 295 298 267 295 286 370 308 307 330 353 322 Virginia 2,549 2,667 2,607 2,620 2,840 2,607 2,770 2,592 2,656 2,899 2,901 2,865 Washington 2,604 2,533 2,520 2,604 2,545 2,448 2,591 2,553 2,597 2,832 2,835 2,634 West Virginia 1,208 1,300 1,242 1,195 1,257 1,207 1,315 1,233 1,289 1,567 1,462 1,455 Wisconsin 2,172 2,202 2,286 2,246 2,223 2,220 2,352 2,273 2,325 2,455 2,397 2,384 Wyoming 327 277 256 319 270 301 279 329 285 304 307 324 Total 118,456 116,494 117,697 119,480 121,267 117,134 126,005 119,923 123,491 136,637 132,917 133,575 COPD includes ICD-10 Codes J40–J44 from the WHO.

5 1,233 1,289 1,567 1,462 1,455 Wisconsin 2,172 2,202 2,286 2,246 2,223 2,220 2,352 2,273 2,325 2,455 2,397 2,384 Wyoming 327 277 256 319 270 301 279 329 285 304 307 324 Total 118,456 116,494 117,697 119,480 121,267 117,134 126,005 119,923 123,491 136,637 132,917 133,575 COPD includes ICD-10 Codes J40–J44 from the WHO. See Table 15 legend for expansion of abbreviations. Table 18 —Age-Adjusted Annual Rates for Deaths With COPD as the Underlying Cause of Death Among Adults Aged ≥ 25 Years, by State—United States, Mortality Component of the National Vital Statistics System, 1999-2010

5 1,233 1,289 1,567 1,462 1,455 Wisconsin 2,172 2,202 2,286 2,246 2,223 2,220 2,352 2,273 2,325 2,455 2,397 2,384 Wyoming 327 277 256 319 270 301 279 329 285 304 307 324 Total 118,456 116,494 117,697 119,480 121,267 117,134 126,005 119,923 123,491 136,637 132,917 133,575 COPD includes ICD-10 Codes J40–J44 from the WHO. See Table 15 legend for expansion of abbreviations. Table 18 —Age-Adjusted Annual Rates for Deaths With COPD as the Underlying Cause of Death Among Adults Aged ≥ 25 Years, by State—United States, Mortality Component of the National Vital Statistics System, 1999-2010 Statea 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P for Linear Trend Alabama 72.2 67.1 72.0 75.5 78.1 74.2 75.4 71.3 77.2 81.4 81.5 83.2 .002 Alaska 81.1 71.0 74.3 69.2 67.0 57.7 62.3 55.2 67.6 67.2 73.6 60.6 .202 Arizona 74.9 73.2 70.3 71.4 68.9 63.8 71.6 68.5 64.4 68.5 64.5 64.5 .003 Arkansas 68.9 70.0 68.0 72.0 74.3 70.3 76.3 71.2 77.8 88.1 83.9 80.8 .001 California 68.9 65.6 65.5 63.0 65.4 60.2 61.9 59.2 56.8 59.3 55.7 54.9 < .001 Colorado 85.4 80.4 80.3 78.9 80.8 77.0 76.8 74.1 74.4 79.1 71.8 75.0 .002 Connecticut 55.5 59.3 56.7 54.3 53.8 52.5 53.7 51.9 48.3 53.0 49.6 43.6 < .001 Delaware 63.8 64.3 55.4 64.0 61.6 60.4 70.0 57.5 61.0 73.5 65.9 65.8 .236 District of Columbia 41.7 44.2 38.3 34.5 35.1 40.3 34.7 32.4 34.2 37.4 35.2 37.7 .099 Florida 64.0 60.0 61.1 60.7 60.1 58.4 60.3 56.3 57.8 61.9 60.3 60.3 .338 Georgia 72.5 72.0 71.4 71.6 72.3 67.7 71.9 68.8 67.7 68.1 70.1 70.0 .030 Hawaii 31.9 29.0 29.4 27.4 28.4 29.9 27.7 27.1 26.7 27.0 26.1 24.8 < .001 Idaho 73.1 72.9 72.7 72.7 70.9 65.6 80.6 70.4 69.9 72.4 71.8 70.1 .597 Illinois 63.0 57.9 57.6 57.6 57.9 56.1 59.4 55.3 55.0 64.2 59.9 58.2 .908 Indiana 76.2 76.6 78.4 77.0 79.2 75.1 82.1 76.6 73.9 87.2 83.6 83.3 .060 Iowa 69.3 63.6 64.2 66.0 69.6 63.6 70.3 67.6 66.4 74.3 73.0 66.2 .157 Kansas 71.8 72.6 74.7 70.8 74.6 67.1 79.4 74.4 73.1 79.4 76.0 75.0 .155 Kentucky 88.6 81.5 85.2 89.4 87.9 82.3 91.6 83.3 90.7 99.0 94.1 90.7 .050 Louisiana 59.0 61.0 63.9 60.1 61.4 56.9 66.3 61.0 60.1 65.8 64.1 65.9 .072 Maine 79.0 80.5 81.1 79.5 77.2 74.5 80.5 73.9 67.7 72.0 73.3 71.3 .002 Maryland 61.0 60.1 58.2 57.9 58.2 54.8 54.0 51.3 52.0 53.5 53.8 52.0 < .001 Massachusetts 61.3 62.3 59.6 57.6 58.0 53.2 54.0 52.1 47.4 51.8 50.6 46.5 < .001 Michigan 67.0 66.8 63.1 66.5 66.5 62.3 64.6 64.1 64.8 72.0 67.4 68.6 .271 Minnesota 60.2 56.9 57.1 57.4 52.8 52.9 55.6 49.6 48.3 56.5 51.6 51.9 .015 Mississippi 70.1 68.1 72.6 74.7 76.1 71.8 77.4 71.9 73.0 76.7 78.7 82.3 .004 Missouri 76.7 69.7 71.2 70.2 71.4 65.5 74.0 70.9 71.3 86.3 77.8 78.8 .097 Montana 87.3 80.0 87.9 85.9 86.9 84.1 83.1 79.9 83.2 92.9 78.5 77.2 .218 Nebraska 73.8 66.1 67.8 71.8 69.2 62.0 71.3 66.0 68.2 76.9 71.5 73.1 .389 Nevada 96.5 92.1 99.2 100.0

ssippi 70.1 68.1 72.6 74.7 76.1 71.8 77.4 71.9 73.0 76.7 78.7 82.3 .004 Missouri 76.7 69.7 71.2 70.2 71.4 65.5 74.0 70.9 71.3 86.3 77.8 78.8 .097 Montana 87.3 80.0 87.9 85.9 86.9 84.1 83.1 79.9 83.2 92.9 78.5 77.2 .218 Nebraska 73.8 66.1 67.8 71.8 69.2 62.0 71.3 66.0 68.2 76.9 71.5 73.1 .389 Nevada 96.5 92.1 99.2 100.0 95.9 86.7 92.2 76.5 74.0 85.7 80.6 74.9 .001 New Hampshire 77.0 73.6 77.0 70.3 63.4 69.9 72.3 67.7 66.4 74.2 69.2 62.9 .053 New Jersey 53.0 50.4 47.9 46.9 47.1 48.7 50.1 44.9 46.7 50.4 47.3 46.8 .186 New Mexico 78.2 68.6 67.3 73.1 78.8 62.2 68.8 69.4 67.3 74.7 70.7 71.8 .816 New York 53.6 51.0 51.1 50.9 48.5 48.8 48.6 45.1 46.3 48.1 46.4 46.2 < .001 North Carolina 70.6 72.3 67.1 69.3 71.5 65.1 73.4 68.0 69.8 73.5 68.0 69.1 .897 North Dakota 51.9 56.6 59.1 61.4 56.5 51.1 50.8 51.7 48.7 64.3 62.2 63.8 .383 Ohio 74.4 75.5 73.7 74.8 72.6 71.8 79.4 71.8 75.5 80.5 75.9 75.7 .291 Oklahoma 73.2 82.2 79.5 81.7 88.4 80.7 95.3 86.6 93.8 104.4 98.3 102.6 < .001 Oregon 73.5 70.0 70.7 73.8 71.7 69.5 70.0 66.8 68.4 68.9 66.6 67.1 .001 Pennsylvania 61.1 59.7 57.4 58.5 58.1 57.6 58.5 52.9 56.7 62.4 59.4 56.8 .463 Rhode Island 60.5 61.5 61.7 62.6 57.5 54.2 60.2 56.8 48.4 55.8 60.2 58.4 .147 South Carolina 69.5 67.4 65.7 71.2 69.7 63.7 68.6 65.2 66.5 72.0 72.2 68.6 .478 South Dakota 55.9 64.0 59.5 63.6 63.7 63.5 71.5 59.8 70.9 75.5 68.0 67.8 .012 Tennessee 74.6 77.1 77.6 77.8 78.3 75.7 79.0 71.7 74.8 82.6 79.5 79.3 .298 Texas 69.0 66.3 69.2 68.0 65.2 62.5 65.6 60.8 63.1 67.8 63.8 64.8 .105 Utah 54.4 49.4 48.2 54.3 50.1 50.8 50.3 47.3 49.0 48.9 43.1 48.8 .031 Vermont 73.2 73.9 73.3 64.4 70.0 67.1 85.7 69.2 67.5 71.2 74.0 67.3 .729 Virginia 64.2 66.2 63.5 62.3 66.5 59.8 61.9 56.5 56.8 60.3 59.0 56.9 .002 Washington 76.2 73.1 71.2 72.1 69.2 65.1 67.4 64.6 64.1 68.1 66.4 60.3 < .001 West Virginia 87.4 93.8 89.3 85.1 88.5 84.6 91.1 84.6 87.0 104.9 96.8 95.1 .138 Wisconsin 59.8 60.3 61.8 59.8 58.5 57.8 60.4 57.5 57.8 59.7 58.0 56.5 .019 Wyoming 113.4 94.6 85.1 104.6 86.7 94.8 85.4 99.4 84.6 88.4 87.8 89.6 .119 Totala 67.0 65.2 64.9 65.0 64.9 61.8 65.3 61.1 61.7 66.9 63.8 63.1 .163 Annual rate per 100,000 US population.

3.8 89.3 85.1 88.5 84.6 91.1 84.6 87.0 104.9 96.8 95.1 .138 Wisconsin 59.8 60.3 61.8 59.8 58.5 57.8 60.4 57.5 57.8 59.7 58.0 56.5 .019 Wyoming 113.4 94.6 85.1 104.6 86.7 94.8 85.4 99.4 84.6 88.4 87.8 89.6 .119 Totala 67.0 65.2 64.9 65.0 64.9 61.8 65.3 61.1 61.7 66.9 63.8 63.1 .163 Annual rate per 100,000 US population. COPD includes ICD-10 codes J40–J44 from the WHO. Death rates for 2001-2009 will differ from previous reports because 2001-2009 population denominators have been revised in CDC Wonder (Oct 2012). See Table 15 legend for expansion of abbreviations. a Age-adjusted to the 2000 US population aged ≥ 25 y. Figure 11. Age-adjusted state-specific death rates (per 100,000) for COPD as the underlying cause of death among adults aged ≥ 25 years, by state—United States, Mortality Component of the National Vital Statistics System, 1999-2000 and 2009-2010. Figure 12. Significant linear change (P < .05) in state-specific age-adjusted death rates for COPD as the underlying cause of death among adults aged ≥ 25 years, by state—United States, Mortality Component of the National Vital Statistics System, 1999-2010.

Figure 11. Age-adjusted state-specific death rates (per 100,000) for COPD as the underlying cause of death among adults aged ≥ 25 years, by state—United States, Mortality Component of the National Vital Statistics System, 1999-2000 and 2009-2010. Figure 12. Significant linear change (P < .05) in state-specific age-adjusted death rates for COPD as the underlying cause of death among adults aged ≥ 25 years, by state—United States, Mortality Component of the National Vital Statistics System, 1999-2010. Discussion The previous COPD surveillance report noted that rates of hospitalizations and mortality for COPD had increased from 1980 to 2000.13 However, the mortality rate in men and some age groups and hospitalization rates in both men and women have declined since 1999. Rates of physician-based office visits and ED visits for COPD from 1999 to 2010 demonstrated substantial interyear variability and showed no particular trend; however, it is encouraging that there were no increases in office visit rates or ED rates for COPD.

ization rates in both men and women have declined since 1999. Rates of physician-based office visits and ED visits for COPD from 1999 to 2010 demonstrated substantial interyear variability and showed no particular trend; however, it is encouraging that there were no increases in office visit rates or ED rates for COPD. Because smoking is the most important etiologic driver of COPD,22 trends in the prevalence of smoking impacted many of the metrics examined in this surveillance report, although the exact temporal relationship between changes in the smoking prevalence and changes in health-care use and mortality for COPD are not well defined. Since 1965, the prevalence of smoking has decreased considerably. In 1965, 42.4% (unadjusted percentage) of adults aged ≥ 18 years were current smokers compared with 19.3% in 2010.23 The crude prevalence of smoking in 2010 was one-half that in 1965 for both men (21.5% vs 51.9%, respectively) and women (17.3% vs 33.9%, respectively). In 1999 to 2001, American Indian/Alaska Native adults had a higher age-adjusted prevalence of current smoking (30.3% in men and 34.7% in women) compared with white adults (25.1% in men and 22.2% in women),23 which may explain the increase in COPD mortality during 1999 to 2010 in that population. The prevalence of current smoking among American Indian/Alaska Native adults has since declined to 25.1% in men and 21.0% in women for 2008 to 201023; therefore, a decline in mortality from COPD may be expected for that population in the future. However, a recent report observed that almost 39% of 15 million adults with self-reported COPD in 2011 in the United States continued to smoke.12 This large population represents an important opportunity for physician counseling and referral to smoking cessation interventions such as 1-800-QUIT-NOW.

tion in the future. However, a recent report observed that almost 39% of 15 million adults with self-reported COPD in 2011 in the United States continued to smoke.12 This large population represents an important opportunity for physician counseling and referral to smoking cessation interventions such as 1-800-QUIT-NOW. Two broad currents influence mortality rates estimated from death certificate data: changes in the prevalence of COPD and changes in the case-fatality rate among people with COPD. Although the estimates of the prevalence of self-reported COPD from the NHIS suggest that the prevalence may have declined since 1999, the rates since 2002 have remained fairly stable. A number of treatment strategies have been shown to have the potential to reduce mortality in patients with COPD and include newer medications and evolving guidelines to treat COPD, oxygen therapy, respiratory management, pulmonary rehabilitation, and influenza vaccinations.24,25 The lag times between changes in the prevalence of COPD and the uptake of treatments and COPD mortality rates may differ. The balance of these temporal changes is likely to have a substantial impact on the trajectory of the mortality rate. With continued declines in the smoking prevalence and improved management of patients with COPD, mortality rates can be expected to decline in future years.

eatments and COPD mortality rates may differ. The balance of these temporal changes is likely to have a substantial impact on the trajectory of the mortality rate. With continued declines in the smoking prevalence and improved management of patients with COPD, mortality rates can be expected to decline in future years. The generally small reduction in the age-adjusted mortality rate was limited to men. It is unclear why the mortality rate in women did not fall as well, given the decline in smoking prevalence in women since 1965. If the estimates are valid, these results suggest that research will be needed to address possible explanations for the poor progress among women. These data are consistent with the results of a study showing that the mortality rate among women with an obstructive impairment changed little in contrast to the mortality rate among men with an obstructive impairment.26

t that research will be needed to address possible explanations for the poor progress among women. These data are consistent with the results of a study showing that the mortality rate among women with an obstructive impairment changed little in contrast to the mortality rate among men with an obstructive impairment.26 The use of spirometry is critical to establishing the diagnosis and severity of COPD. Additional tests that can help in the diagnosis include lung diffusion capacity test, chest radiograph, and arterial blood gas test. GOLD (Global Initiative for Chronic Obstructive Lung Disease) established four levels of COPD on the basis of spirometric measurements: mild, moderate, severe, and very severe.27 The results reported here should be considered in the context of several limitations. Depending on the spirometric criteria used, estimates of prevalence of COPD based on spirometry tests may be as much as double the estimates derived from self-reported information.13,28,29 Consequently, the estimates of self-reported prevalence of COPD in the current surveillance report almost certainly underestimate the true prevalence of this condition. Furthermore, not accounting for the undiagnosed percentage of adults with COPD can also potentially distort demographic comparisons. As shown in the previous surveillance report, men had a higher prevalence than women when the presence of COPD was based on spirometric criteria.13 When self-reported data were used to estimate the prevalence of COPD, however, women had a higher prevalence than men, as was also observed in the present report.

parisons. As shown in the previous surveillance report, men had a higher prevalence than women when the presence of COPD was based on spirometric criteria.13 When self-reported data were used to estimate the prevalence of COPD, however, women had a higher prevalence than men, as was also observed in the present report. If COPD is underdiagnosed, then the mortality rates presented in the present report likely underestimate the true mortality rates from COPD.30‐32 Another factor that may contribute to underestimating COPD mortality rates is the possibility that comorbidities may displace COPD as the underlying cause of death that is reported on the death certificate.33 Assuming that underestimates of the COPD mortality rates were approximately constant during the study period, the interpretation of the direction of the trends is valid.

mortality rates is the possibility that comorbidities may displace COPD as the underlying cause of death that is reported on the death certificate.33 Assuming that underestimates of the COPD mortality rates were approximately constant during the study period, the interpretation of the direction of the trends is valid. Race was self-reported by participants of the BRFSS and NHIS but was recorded by medical or other personnel in the other data systems. The comparability of race designations among surveys is unknown. For some data systems, such as the NAMCS, NHAMCS, and NHDS, race was missing for a large proportion of records. For example, 16% of the NHDS discharges for 2010 and 23% of NAMCS records in 2010 lacked information about the racial status of the patient. Medicare and death certificate data represented the only data that allowed trend analyses for American Indian/Alaska Natives, Hispanics, and Asian populations. Because race and ethnicity designations are subject to misclassification,34 caution is urged in interpreting racial- and ethnic-specific disparities. In the future, the BRFSS, with its large annual sample size of almost one-half million respondents, will allow trend analyses of prevalence of self-reported COPD among those racial/ethnic groups.

y designations are subject to misclassification,34 caution is urged in interpreting racial- and ethnic-specific disparities. In the future, the BRFSS, with its large annual sample size of almost one-half million respondents, will allow trend analyses of prevalence of self-reported COPD among those racial/ethnic groups. Since 1997, GOLD has striven to increase awareness of COPD as a major public health problem across the globe, to spur efforts to prevent this disease, and to develop guidelines to improve the diagnosis and treatment of COPD. In 2013, it released updated versions of Global Strategy for Diagnosis, Management, and Prevention of COPD.24 Several studies have reported imperfect implementation of the GOLD guidelines in clinical practice.35 Additional efforts may be needed to educate physicians about the management of this condition.

ment of COPD. In 2013, it released updated versions of Global Strategy for Diagnosis, Management, and Prevention of COPD.24 Several studies have reported imperfect implementation of the GOLD guidelines in clinical practice.35 Additional efforts may be needed to educate physicians about the management of this condition. Healthy People objectives provide science-based, 10-year national objectives for improving the health of all Americans; identify nationwide health improvement priorities; and strive to engage multiple sectors (public health agencies, communities, organizations, academia, and medicine) to take actions to strengthen policies and improve practices that are driven by the best available evidence and knowledge. The Healthy People 2010 objective for COPD called for a 50% reduction in the mortality rate from COPD among adults aged ≥ 45 years at baseline in 1999 (123.9 per 100,000)36; however, that objective was not met by 2010 (116.6 per 100,000)—possibly for many reasons described above. The new Healthy People 2020 effort37 has been expanded to include the following objectives that pertain to the evaluation and management of COPD among adults aged ≥ 45 years:• Reduce activity limitations among adults with COPD. • Reduce deaths from COPD. • Reduce hospitalizations for COPD. • Reduce hospital ED visits for COPD • Increase the proportion of adults with abnormal lung function whose underlying obstructive disease has been diagnosed.

Healthy People objectives provide science-based, 10-year national objectives for improving the health of all Americans; identify nationwide health improvement priorities; and strive to engage multiple sectors (public health agencies, communities, organizations, academia, and medicine) to take actions to strengthen policies and improve practices that are driven by the best available evidence and knowledge. The Healthy People 2010 objective for COPD called for a 50% reduction in the mortality rate from COPD among adults aged ≥ 45 years at baseline in 1999 (123.9 per 100,000)36; however, that objective was not met by 2010 (116.6 per 100,000)—possibly for many reasons described above. The new Healthy People 2020 effort37 has been expanded to include the following objectives that pertain to the evaluation and management of COPD among adults aged ≥ 45 years:• Reduce activity limitations among adults with COPD. • Reduce deaths from COPD. • Reduce hospitalizations for COPD. • Reduce hospital ED visits for COPD • Increase the proportion of adults with abnormal lung function whose underlying obstructive disease has been diagnosed. The CDC and the National Heart Lung Blood Institute (NHLBI) have a formal collaboration to increase public awareness and identify critical communication, research, evaluation, and data collection needs to prevent and manage COPD. This collaboration has resulted in the annual BRFSS collection since 2011 of COPD prevalence data at state and local levels, which will enhance the COPD Learn More Breathe Better Campaign supported by the NHLBI. Such state-level and county-level data as the BRFSS, Medicare, and vital statistics can identify geographic clustering of, as well as racial/ethnic disparities in, COPD indicators to provide guidance to public health agencies in leveraging and targeting resources to those geographic areas and local populations with the greatest burden of COPD. These data will also be critical in identifying communities that will likely benefit best from awareness and outreach campaigns and in evaluating the effectiveness of public health efforts to prevent, treat, and control COPD.

ng resources to those geographic areas and local populations with the greatest burden of COPD. These data will also be critical in identifying communities that will likely benefit best from awareness and outreach campaigns and in evaluating the effectiveness of public health efforts to prevent, treat, and control COPD. COPD remains a significant source of morbidity and mortality in the United States. In 2007, chronic lower respiratory diseases constituted the fourth leading cause of death and rose to the third leading cause of death in 2008 primarily because cerebrovascular disease deaths continued a consistent decline and to a lesser extent as a result of adjustments to coding and classification.1 The data examined in this surveillance report testify to the heavy public health burden that COPD continues to levy in the United States. Prior to 1999, rates of mortality and hospitalizations had shown worrisome increases. Thus, the apparent leveling of the mortality rate and a decrease in the rate of hospitalization represent cause for cautious optimism. Future surveillance efforts will be critical to tracking the course of COPD in the United States. Supplementary Material Online Supplement Click here for additional data file.

COPD remains a significant source of morbidity and mortality in the United States. In 2007, chronic lower respiratory diseases constituted the fourth leading cause of death and rose to the third leading cause of death in 2008 primarily because cerebrovascular disease deaths continued a consistent decline and to a lesser extent as a result of adjustments to coding and classification.1 The data examined in this surveillance report testify to the heavy public health burden that COPD continues to levy in the United States. Prior to 1999, rates of mortality and hospitalizations had shown worrisome increases. Thus, the apparent leveling of the mortality rate and a decrease in the rate of hospitalization represent cause for cautious optimism. Future surveillance efforts will be critical to tracking the course of COPD in the United States. Supplementary Material Online Supplement Click here for additional data file. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Mannino has received honoraria/consulting fees and served on speaker bureaus for GlaxoSmithKline; Novartis AG; Pfizer, Inc; AstraZeneca; Forest Laboratories, Inc; and Creative Educational Concepts. Furthermore, he has received royalties from UptoDate, Inc. Drs Ford, Croft, Wheaton, Zhang, and Giles have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

ca; Forest Laboratories, Inc; and Creative Educational Concepts. Furthermore, he has received royalties from UptoDate, Inc. Drs Ford, Croft, Wheaton, Zhang, and Giles have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Other contributions: Work was performed at the Centers for Disease Control and Prevention, Atlanta, Georgia. The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention. Additional information: The e-Tables can be found in the “Supplemental Materials” area of the online article. Abbreviations BRFSSBehavioral Risk Factor Surveillance System CDCCenters for Disease Control and Prevention CHCCommunity Health Center GOLDGlobal Initiative for Chronic Obstructive Lung Disease ICD-9-CMInternational Classification of Diseases, Ninth Revision, Clinical Modification ICD-10International Classification of Diseases, 10th Revision NAMCSNational Ambulatory Medical Care Survey NHAMCSNational Hospital Ambulatory Medical Care Survey NHDSNational Hospital Discharge Survey NHISNational Health Interview Survey NHLBINational Heart, Lung, and Blood Institute NVSSNational Vital Statistics System PSUprimary sampling unit

The airway epithelium is at the interface between the host and the environment, plays a critical role in normal wound repair, and is implicated as key in the immunopathogenesis of asthma.1 Epithelial cells in vivo are in an activated state, with increased expression of chemokines such as CXCL82 and CCL11.3 Structural changes observed consistently in the asthmatic epithelium include increased permeability,4 reduced ciliary beat frequency and coordinated cilia movement, increased cell protrusion and cytoplasmic blebbing,5 goblet cell hyperplasia,6 increased mucin production,7 and increased levels of epithelial proliferation and apoptosis8 when compared with nonasthmatic epithelium.

ithelium include increased permeability,4 reduced ciliary beat frequency and coordinated cilia movement, increased cell protrusion and cytoplasmic blebbing,5 goblet cell hyperplasia,6 increased mucin production,7 and increased levels of epithelial proliferation and apoptosis8 when compared with nonasthmatic epithelium. Whether these abnormalities persist in vitro is important to determine the relative contribution of the asthmatic environment and intrinsic changes in cellular behavior in defining disease expression. Following wounding, asthmatic epithelium demonstrates aberrant repair, dyssynchronous mitosis,9 and defective epithelial tight junctions, suggesting persistence of abnormalities in wound repair.10 In contrast, whether asthmatic epithelial cells have an enhanced synthetic response is contentious. Some reports have found constitutive chemokine and cytokine release by epithelial cells from subjects with asthma compared with healthy control subjects is upregulated,11 downregulated,12 or unchanged.9 Similarly, both an increased13 and deficient14 interferon (IFN)-β response following exposure of epithelial cells from subjects with asthma to virus is reported. The phenotype of epithelial cells may also vary throughout the airway tree as demonstrated by differences in transepithelial resistance between epithelial cells from conducting airways and the nose.13 Therefore, the synthetic function of asthmatic epithelium from different sites within the airway and their response to antiinflammatory therapy need to be further defined.

ughout the airway tree as demonstrated by differences in transepithelial resistance between epithelial cells from conducting airways and the nose.13 Therefore, the synthetic function of asthmatic epithelium from different sites within the airway and their response to antiinflammatory therapy need to be further defined. Inflammatory gene expression often involves the transcription factor nuclear factor κB (NF-κB), and this signaling pathway represents a site for antiinflammatory intervention. Phosphorylation of the inhibitory κB (IκB) proteins by the IκB kinase (IKK) 2-containing IKK complex and subsequent degradation of the IκB proteins are prerequisites for NF-κB activation. Therefore, inhibition of IKK2 would specifically prevent NF-κB transcription and signaling. One of the mechanisms of action of glucocorticosteroids also involves targeting the NF-κB pathway, and glucocorticoids are the most effective antiinflammatory treatments for asthma.15 Although the response to these compounds has been well characterized in inflammatory cells within the airway, there is a lack of data examining the response to glucocorticosteroids in primary epithelial cells.16

ting the NF-κB pathway, and glucocorticoids are the most effective antiinflammatory treatments for asthma.15 Although the response to these compounds has been well characterized in inflammatory cells within the airway, there is a lack of data examining the response to glucocorticosteroids in primary epithelial cells.16 We, therefore, hypothesized that synthetic capacity would be altered in primary airway epithelial cells from subjects with asthma vs healthy subjects and that there would be differential effects of antiinflammatory therapy. To test our hypothesis, we aimed to examine: (1) the synthetic function of airway epithelial cells from different locations in health vs disease with and without stimulation by measuring a panel of epithelial-derived chemokines and cytokines,14,17,18 and (2) to determine the effects of corticosteroids and novel antiinflammatory therapies upon the synthetic capacity of these epithelial cells. Materials and Methods Subjects Subjects were recruited from Glenfield Hospital, Leicester, England and by local advertising. Asthma was defined according to GINA (Global Initiative for Asthma) guidelines.19 Subject characterization included demographics, spirometry, allergen skin prick tests, sputum induction, methacholine bronchial challenge, nasal brushings, and bronchoscopy. The study was approved by the Leicestershire ethics committees, and all patients gave their written informed consent.

ve for Asthma) guidelines.19 Subject characterization included demographics, spirometry, allergen skin prick tests, sputum induction, methacholine bronchial challenge, nasal brushings, and bronchoscopy. The study was approved by the Leicestershire ethics committees, and all patients gave their written informed consent. This study was conducted in accordance with the amended Declaration of Helsinki. The Leicestershire, Rutland, and Northamptonshire ethics committee (ethics reference 4977/project approval number 6347) approved the protocol, and written informed consent was obtained from all patients.

ve for Asthma) guidelines.19 Subject characterization included demographics, spirometry, allergen skin prick tests, sputum induction, methacholine bronchial challenge, nasal brushings, and bronchoscopy. The study was approved by the Leicestershire ethics committees, and all patients gave their written informed consent. This study was conducted in accordance with the amended Declaration of Helsinki. The Leicestershire, Rutland, and Northamptonshire ethics committee (ethics reference 4977/project approval number 6347) approved the protocol, and written informed consent was obtained from all patients. Epithelial Cell Culture Epithelial cells were obtained from nasal and bronchial brushings from the second- or third-generation bronchi and were grown on 12-well tissue culture plates in bronchial epithelial growth medium (BEGM; Lonza Group Ltd), including supplement SingleQuot BulletKit (Lonza Group Ltd), 0.3% Fungizone antimycotic (Life Technologies Corporation), and 1% antibiotic-antimycotic (Life Technologies Corporation) for 2 to 7 days. Basal cells were then expanded into 75-cm2 flasks and upon confluence seeded at 105 cells/cm2 on 1.2-cm2-diameter transwell clear inserts (Corning Incorporated) under BEGM for 2 days. All culture surfaces were collagen coated (Nutacon B.V.). After reaching confluence, the basal cell monolayer was fed on the basolateral side only with air-liquid interface media (ALI) media. This ALI media consisted of 50% BEGM and 50% hi-glucose minimal essential medium (Life Technologies Corporation) containing 100 nM retinoic acid (Sigma-Aldrich Co LLC), including supplements as previously detailed. Supplements were removed 24 h prior to experiments. Nasal and bronchial basal epithelial cells were characterized using immunofluorescence for cytokeratin 5 and 14 expression (Abcam plc). Spontaneous mucus production occurred from 14 to 21 days, suggesting the presence of goblet cells, and ciliation occurred between 25 and 35 days. There was a 100% success rate of nonasthmatic and asthmatic cultures from the initial brushing to the passage into ALI and a 58% success rate of these ALI cultures becoming ciliated. Differentiated bronchial epithelial cells were characterized using high-speed video microscopy for the presence of active cilia

35 days. There was a 100% success rate of nonasthmatic and asthmatic cultures from the initial brushing to the passage into ALI and a 58% success rate of these ALI cultures becoming ciliated. Differentiated bronchial epithelial cells were characterized using high-speed video microscopy for the presence of active cilia Mediator Analysis Mediators from basal and differentiated epithelial cells were analyzed using both multiplex and single enzyme-linked immunosorbent assay (ELISA) kits. We measured IL1β, tumor necrosis factor (TNF)-α, CCL2, CXCL8, CXCL10, CCL11, CCL13, CCL17, CCL22, and CCL26 (MesoScale Discovery [MSD]; Meso Scale Diagnostics, LLC). Limits of detections were 2.4 to 10,000 pg/mL. ELISA kits analyzing single mediators were used to examine CCL5 (R&D Systems, Inc) and IFN-β (Pestka Biomedical Laboratories, Inc) production from epithelial cells. Limits of detection for the CCL5 and IFN-β ELISAs were 31.2 to 1,000 pg/mL and 25 to 2,000 pg/mL.

ics, LLC). Limits of detections were 2.4 to 10,000 pg/mL. ELISA kits analyzing single mediators were used to examine CCL5 (R&D Systems, Inc) and IFN-β (Pestka Biomedical Laboratories, Inc) production from epithelial cells. Limits of detection for the CCL5 and IFN-β ELISAs were 31.2 to 1,000 pg/mL and 25 to 2,000 pg/mL. Study Protocol The compounds IKK2i (100 nM, 300 nM, 1 μM, or 3 μM), (gifts from GlaxoSmithKline, Stevenage, England), prednisolone (10 nM, 100 nM, 1 μM, 10 μM) (GlaxoSmithKline) were added where appropriate to cell cultures for 1 h before the addition of any stimulus. Stimuli used included Poly-IC (0.5, 2.5, 12.5, 25 μg/mL) (InvivoGen) with Poly-dIdC (0.5, 2.5, 12.5, 25 μg/mL) (Sigma-Aldrich Co LLC) as a negative control, IL-1β (10 ng/mL) or IL-1β (10 ng/mL) and IFN-γ (10 ng/mL) (R&D Systems). Nasal epithelial basal cells were used only to obtain a dose response to prednisolone and poly IC. Bronchial epithelial basal and differentiated cells were then used to examine the response to the full panel of compounds/stimuli. Epithelial basal cells were cultured on 24-well culture plates, and compounds/stimuli were added to the media. Differentiated epithelial cultures were cultured on transwells with compounds/stimuli added basolaterally. Cell culture supernatants were collected from all experiments after 24 h. Mediator concentrations were corrected for the volume of growth medium and cell count for each condition. Cell viability was assessed using trypan blue.

d epithelial cultures were cultured on transwells with compounds/stimuli added basolaterally. Cell culture supernatants were collected from all experiments after 24 h. Mediator concentrations were corrected for the volume of growth medium and cell count for each condition. Cell viability was assessed using trypan blue. Statistical Analysis Statistical analysis was performed using GraphPad Prism 4 (GraphPad Software, Inc). Constitutive levels of mediators were log normally distributed and, thus, expressed as geometric mean (95% CI). Therefore, stimulated levels of mediators were expressed as fold change over constitutive levels. Paired and unpaired data were analyzed using paired and unpaired t tests, respectively. Comparison across groups was assessed using analysis of variance (ANOVA). Differences were considered significant when P ≤ .05. Correction for multiple comparisons was not conducted.20

ressed as fold change over constitutive levels. Paired and unpaired data were analyzed using paired and unpaired t tests, respectively. Comparison across groups was assessed using analysis of variance (ANOVA). Differences were considered significant when P ≤ .05. Correction for multiple comparisons was not conducted.20 Results The clinical characteristics of the subjects with asthma and healthy control subjects are as shown in e-Table 1. The patterns of cytokine and chemokine expression constitutively released over 24 h by nasal, basal bronchial, and differentiated ALI culture bronchial epithelial cells were significantly different for IFN-β, IL-1β, CCL2, CCL5, and CXCL8 (e-Table 2, Fig 1). These differences between cell types were similar for cells derived from subjects with and without disease, but no differences were observed between subjects with asthma and control subjects (e-Figs 1a-1c, e-Table 3). No differences were found between subjects who were treated with corticosteroids and those who were corticosteroid naive (data not shown). Figure 1. Constitutive mediator concentration between primary epithelial cell types. Constitutive expression (pg/mL/106 cells) after 24 h of mediators from nonasthmatic and asthmatic nasal basal (white), bronchial basal (gray), and bronchial differentiated (black) epithelial cells. Cytokines are ordered according to the statistical difference of mediators between cell types (*P ≤ .05) (n = 6 nasal basal, n = 17 bronchial basal, and n = 21 bronchial differentiated epithelial cells). IFN = interferon; TNF = tumor necrosis factor.

basal (gray), and bronchial differentiated (black) epithelial cells. Cytokines are ordered according to the statistical difference of mediators between cell types (*P ≤ .05) (n = 6 nasal basal, n = 17 bronchial basal, and n = 21 bronchial differentiated epithelial cells). IFN = interferon; TNF = tumor necrosis factor. There was a concentration-dependent release of mediators by nasal basal epithelial cells in response to poly-IC activation compared with the Poly-dIdC control as illustrated in e-Fig 2. The concentration of Poly IC that induced maximal mediator release above constitutive levels in nasal basal cells was then used to stimulate bronchial basal and differentiated cells. Stimulation of the epithelial cells by poly IC, IL-1β or IL-1β, and IFN-γ significantly increased mediator levels above constitutive levels for the majority of mediators, and the pattern of mediator release was distinctive for the stimulus and epithelial cell type (Tables 1-3). Differences were only observed in four of the 111 parameters measured between nonasthmatic and asthmatic groups in any of the cell types examined (e-Tables 4-6). Table 1 —Mediator Concentration Between Primary Epithelial Cell Types Following Poly IC Stimulation

There was a concentration-dependent release of mediators by nasal basal epithelial cells in response to poly-IC activation compared with the Poly-dIdC control as illustrated in e-Fig 2. The concentration of Poly IC that induced maximal mediator release above constitutive levels in nasal basal cells was then used to stimulate bronchial basal and differentiated cells. Stimulation of the epithelial cells by poly IC, IL-1β or IL-1β, and IFN-γ significantly increased mediator levels above constitutive levels for the majority of mediators, and the pattern of mediator release was distinctive for the stimulus and epithelial cell type (Tables 1-3). Differences were only observed in four of the 111 parameters measured between nonasthmatic and asthmatic groups in any of the cell types examined (e-Tables 4-6). Table 1 —Mediator Concentration Between Primary Epithelial Cell Types Following Poly IC Stimulation Mediators Poly IC (12.5 μg/mL) Stimulated Epithelial Cells, Fold Change Over Constitutive Levels P Value Nasal Basal (n = 6) Bronchial Basal (n = 17) Bronchial Differentiated (n = 17) IFN-β 3.1 (1.0-9.6)a 4.8 (2.2-10.6)b 1.1 (0.9-1.5) < .01c CCL5 534.6 (274.0-1043.0)b 994.6 (549.3-1801.0) 42.6 (14.5-125.3)b < .01b,c,d IL-1β 35.5 (11.1-111.3)b 2.5 (1.4-4.4)e 2.8 (1.1-7.1)a < .01a,d,f CXCL8 25.9 (13.1-51.2)b 4.7 (2.51-8.9) 2.1 (1.5-2.7)b < .01b,c,d,f CCL2 52.5 (23.7-116.2)b 8.0 (3.0-21.2) 6.4 (2.7-15.2)b .02b,d,f CCL13 18.3 (10.8-30.8)b 5.7 (1.7-19.5)e 1.7 (1.3-2.2)b < .01b,c,d CCL4 485 0.0 (219.6-1071.0)b 135.1 (50.1-364.8) 18.1 (6.8-48.3)b < .01b,c,d TNF-α 145.3 (41.1-513.1)b 11.1 (6.3-19.5) 5.0 (2.4-10.3)b < .01b,d,f CCL22 25.0 (14.4-43.5)b 5.2 (2.3-11.6) 2.2 (1.7-2.9)b < .01b,c,d,f CCL17 65.2 (28.0-151.7)b 10.7 (2.7-41.9)e 2.9 (2.1-3.9)b < .01b,d,f CCL26 131.5 (29.4-588.6)b 4.7 (1.0-21.7) 2.5 (1.3-4.9)e < .01d,e,f CXCL10 376.2 (134.8-1050.0)b 168.9 (35.7-799.5)b 48.9 (21.2-112.7)b .07b CCL11 29.5 (15.6-55.7)b 10.7 (5.2-22.1)b 5.9 (3.1-11.5)b .02b,d Fold change (geometric mean [95% CI]) over constitutive levels after stimulation for 24 h. ANOVA = analysis of variance; IFN = interferon; TNF = tumor necrosis factor.

.9)e < .01d,e,f CXCL10 376.2 (134.8-1050.0)b 168.9 (35.7-799.5)b 48.9 (21.2-112.7)b .07b CCL11 29.5 (15.6-55.7)b 10.7 (5.2-22.1)b 5.9 (3.1-11.5)b .02b,d Fold change (geometric mean [95% CI]) over constitutive levels after stimulation for 24 h. ANOVA = analysis of variance; IFN = interferon; TNF = tumor necrosis factor. a P ≤ .05 compared with constitutive levels. b P < .001 compared with constitutive levels. c Bronchial basal vs bronchial differentiated, one-way ANOVA with Tukey post hoc multiple comparison. d Nasal vs bronchial differentiated, one-way ANOVA with Tukey post hoc multiple comparison. e P ≤ .01 compared with constitutive levels. f Nasal vs bronchial basal, one-way ANOVA with Tukey post hoc multiple comparison. Table 2 —Mediator Concentration Between Primary Epithelial Cell Types Following IL-1β Stimulation

c Bronchial basal vs bronchial differentiated, one-way ANOVA with Tukey post hoc multiple comparison. d Nasal vs bronchial differentiated, one-way ANOVA with Tukey post hoc multiple comparison. e P ≤ .01 compared with constitutive levels. f Nasal vs bronchial basal, one-way ANOVA with Tukey post hoc multiple comparison. Table 2 —Mediator Concentration Between Primary Epithelial Cell Types Following IL-1β Stimulation Mediators IL-1β (10 ng/mL) Stimulated Epithelial Cells, Fold Change Over Constitutive Levels P Value Nasal Basal (n = 6) Bronchial Basal (n = 11) Bronchial Differentiated (n = 10) IFN-β 1.2 (0.6-2.6) 0.9 (0.7-1.2) 1.0 (0.8-1.1) .39 CCL5 1.9 (0.9-3.8) 9.2 (3.4-25.1)a 11.7 (3.3-41.0)b .06b IL-1β N/A N/A N/A N/A CXCL8 10.0 (2.0-51.6)c 3.8 (1.8-7.8)b 1.9 (1.2-3.0)c .02c,d CCL2 2.0 (1.6-2.4)a 1.4 (0.6-3.3) 23.2 (5.4-99.5)a < .01a,d,e CCL13 1.8 (1.4-2.4)b 3.2 (0.5-21.5) 1.7 (1.1-2.5)c .57c CCL4 9.6 (3.6-25.5)b 4.2 (1.7-10.7)b 19.1 (4.6-79.8)b .14b TNF-α 4.3 (1.9-9.6)b 2.7 (1.9-3.8)a 6.4 (2.1-18.8)b .19b CCL22 2.0 (1.5-2.8)b 1.5 (1.0-2.3) 2.4 (1.8-3.2)a .11a CCL17 1.6 (1.3-2.0)b 2.1 (0.5-9.2) 2.6 (2.0-3.5)a .67a CCL26 2.2 (1.7-2.7)a 1.5 (0.1-16.0) 2.0 (0.7-5.4) .90 CXCL10 2.1 (1.6-2.7)a 3.3 (0.6-18.1) 6.0 (3.6-10.2)a .28a CCL11 2.3 (1.9-2.8)a 1.4 (1.0-2.0) 5.3 (1.6-17.7)c .07c Fold change (geometric mean [95% CI]) over constitutive levels after stimulation for 24 h. N/A = not applicable. See Table 1 legend for expansion of abbreviations. a P < .001 compared with constitutive levels. b P ≤ .01 compared with constitutive levels. c P ≤ .05 compared with constitutive levels.

Mediators IL-1β (10 ng/mL) Stimulated Epithelial Cells, Fold Change Over Constitutive Levels P Value Nasal Basal (n = 6) Bronchial Basal (n = 11) Bronchial Differentiated (n = 10) IFN-β 1.2 (0.6-2.6) 0.9 (0.7-1.2) 1.0 (0.8-1.1) .39 CCL5 1.9 (0.9-3.8) 9.2 (3.4-25.1)a 11.7 (3.3-41.0)b .06b IL-1β N/A N/A N/A N/A CXCL8 10.0 (2.0-51.6)c 3.8 (1.8-7.8)b 1.9 (1.2-3.0)c .02c,d CCL2 2.0 (1.6-2.4)a 1.4 (0.6-3.3) 23.2 (5.4-99.5)a < .01a,d,e CCL13 1.8 (1.4-2.4)b 3.2 (0.5-21.5) 1.7 (1.1-2.5)c .57c CCL4 9.6 (3.6-25.5)b 4.2 (1.7-10.7)b 19.1 (4.6-79.8)b .14b TNF-α 4.3 (1.9-9.6)b 2.7 (1.9-3.8)a 6.4 (2.1-18.8)b .19b CCL22 2.0 (1.5-2.8)b 1.5 (1.0-2.3) 2.4 (1.8-3.2)a .11a CCL17 1.6 (1.3-2.0)b 2.1 (0.5-9.2) 2.6 (2.0-3.5)a .67a CCL26 2.2 (1.7-2.7)a 1.5 (0.1-16.0) 2.0 (0.7-5.4) .90 CXCL10 2.1 (1.6-2.7)a 3.3 (0.6-18.1) 6.0 (3.6-10.2)a .28a CCL11 2.3 (1.9-2.8)a 1.4 (1.0-2.0) 5.3 (1.6-17.7)c .07c Fold change (geometric mean [95% CI]) over constitutive levels after stimulation for 24 h. N/A = not applicable. See Table 1 legend for expansion of abbreviations. a P < .001 compared with constitutive levels. b P ≤ .01 compared with constitutive levels. c P ≤ .05 compared with constitutive levels. d Nasal vs bronchial differentiated, one-way ANOVA with Tukey post hoc multiple comparison. e Bronchial basal vs bronchial differentiated, one-way ANOVA with Tukey post hoc multiple comparison. Table 3 —Mediator Concentration Between Primary Epithelial Cell Types Following IL-1β and IFN-γ

c P ≤ .05 compared with constitutive levels. d Nasal vs bronchial differentiated, one-way ANOVA with Tukey post hoc multiple comparison. e Bronchial basal vs bronchial differentiated, one-way ANOVA with Tukey post hoc multiple comparison. Table 3 —Mediator Concentration Between Primary Epithelial Cell Types Following IL-1β and IFN-γ Mediators IL-1β (10 ng/mL) + IFN-γ (10 ng/mL) Stimulated Epithelial Cells, Fold Change Over Constitutive Levels P Value Nasal Basal (n = 6) Bronchial Basal (n = 17) Bronchial Differentiated (n = 11) IFN-β 0.7 (0.4-1.2) 0.9 (0.7-1.3) 0.9 (0.7-1.1) .60 CCL5 8.1 (2.4-27.6)a 46.6 (19.0-114.2)b 50.2 (18.5-136.1)b .07b IL-1β N/A N/A N/A N/A CXCL8 30.5 (10.3-89.7)b 4.4 (2.4-7.8)b 1.9 (1.1-3.1)c < .01c,d,e CCL2 39.1 (15.5-98.4)b 63.5 (25.6-157.3)b 73.8 (15.9-342.3)b .78b CCL13 19.4 (7.8-48.0)b 6.4 (2.2-18.4)a 4.3 (2.9-6.5)b .11b CCL4 87.8 (38.5-68.0)b 42.7 (15.0-121.4)b 65.1 (19.7-214.5)b .64b TNF-α 5.3 (3.6-7.9)b 4.4 (3.1-6.2)b 5.6 (1.6-19.2)c .85c CCL22 14.3 (6.3-32.5)b 6.4 (3.3-12.4)b 6.9 (5.0-9.7)b .18b CCL17 29.7 (9.9-88.5)b 16.1 (5.4-48.4)b 13.8 (9.3-20.6)b .56b CCL26 28.2 (11.7-68.0)b 6.8 (1.8-25.7)a 6.9 (3.0-16.0)b .23b CXCL10 390.8 (122.1-1251.0)b 323.1 (115.5-903.3)b 645.8 (253.0-1652.0)b .54b CCL11 30.6 (11.6-80.6)b 13.7 (8.0-23.5)b 30.3 (9.9-93.0)b .22b Fold change (geometric mean [95% CI]) over constitutive levels after stimulation for 24 h. See Table 1 and 2 legends for expansion of abbreviations. a P ≤ .01 compared with constitutive levels. b P < .001 compared with constitutive levels. c P ≤ .05 compared with constitutive levels.

Mediators IL-1β (10 ng/mL) + IFN-γ (10 ng/mL) Stimulated Epithelial Cells, Fold Change Over Constitutive Levels P Value Nasal Basal (n = 6) Bronchial Basal (n = 17) Bronchial Differentiated (n = 11) IFN-β 0.7 (0.4-1.2) 0.9 (0.7-1.3) 0.9 (0.7-1.1) .60 CCL5 8.1 (2.4-27.6)a 46.6 (19.0-114.2)b 50.2 (18.5-136.1)b .07b IL-1β N/A N/A N/A N/A CXCL8 30.5 (10.3-89.7)b 4.4 (2.4-7.8)b 1.9 (1.1-3.1)c < .01c,d,e CCL2 39.1 (15.5-98.4)b 63.5 (25.6-157.3)b 73.8 (15.9-342.3)b .78b CCL13 19.4 (7.8-48.0)b 6.4 (2.2-18.4)a 4.3 (2.9-6.5)b .11b CCL4 87.8 (38.5-68.0)b 42.7 (15.0-121.4)b 65.1 (19.7-214.5)b .64b TNF-α 5.3 (3.6-7.9)b 4.4 (3.1-6.2)b 5.6 (1.6-19.2)c .85c CCL22 14.3 (6.3-32.5)b 6.4 (3.3-12.4)b 6.9 (5.0-9.7)b .18b CCL17 29.7 (9.9-88.5)b 16.1 (5.4-48.4)b 13.8 (9.3-20.6)b .56b CCL26 28.2 (11.7-68.0)b 6.8 (1.8-25.7)a 6.9 (3.0-16.0)b .23b CXCL10 390.8 (122.1-1251.0)b 323.1 (115.5-903.3)b 645.8 (253.0-1652.0)b .54b CCL11 30.6 (11.6-80.6)b 13.7 (8.0-23.5)b 30.3 (9.9-93.0)b .22b Fold change (geometric mean [95% CI]) over constitutive levels after stimulation for 24 h. See Table 1 and 2 legends for expansion of abbreviations. a P ≤ .01 compared with constitutive levels. b P < .001 compared with constitutive levels. c P ≤ .05 compared with constitutive levels. d Nasal vs bronchial basal, ANOVA with Tukey post hoc multiple comparison. e Nasal vs bronchial differentiated, ANOVA with Tukey post hoc multiple comparison.

Mediators IL-1β (10 ng/mL) + IFN-γ (10 ng/mL) Stimulated Epithelial Cells, Fold Change Over Constitutive Levels P Value Nasal Basal (n = 6) Bronchial Basal (n = 17) Bronchial Differentiated (n = 11) IFN-β 0.7 (0.4-1.2) 0.9 (0.7-1.3) 0.9 (0.7-1.1) .60 CCL5 8.1 (2.4-27.6)a 46.6 (19.0-114.2)b 50.2 (18.5-136.1)b .07b IL-1β N/A N/A N/A N/A CXCL8 30.5 (10.3-89.7)b 4.4 (2.4-7.8)b 1.9 (1.1-3.1)c < .01c,d,e CCL2 39.1 (15.5-98.4)b 63.5 (25.6-157.3)b 73.8 (15.9-342.3)b .78b CCL13 19.4 (7.8-48.0)b 6.4 (2.2-18.4)a 4.3 (2.9-6.5)b .11b CCL4 87.8 (38.5-68.0)b 42.7 (15.0-121.4)b 65.1 (19.7-214.5)b .64b TNF-α 5.3 (3.6-7.9)b 4.4 (3.1-6.2)b 5.6 (1.6-19.2)c .85c CCL22 14.3 (6.3-32.5)b 6.4 (3.3-12.4)b 6.9 (5.0-9.7)b .18b CCL17 29.7 (9.9-88.5)b 16.1 (5.4-48.4)b 13.8 (9.3-20.6)b .56b CCL26 28.2 (11.7-68.0)b 6.8 (1.8-25.7)a 6.9 (3.0-16.0)b .23b CXCL10 390.8 (122.1-1251.0)b 323.1 (115.5-903.3)b 645.8 (253.0-1652.0)b .54b CCL11 30.6 (11.6-80.6)b 13.7 (8.0-23.5)b 30.3 (9.9-93.0)b .22b Fold change (geometric mean [95% CI]) over constitutive levels after stimulation for 24 h. See Table 1 and 2 legends for expansion of abbreviations. a P ≤ .01 compared with constitutive levels. b P < .001 compared with constitutive levels. c P ≤ .05 compared with constitutive levels. d Nasal vs bronchial basal, ANOVA with Tukey post hoc multiple comparison. e Nasal vs bronchial differentiated, ANOVA with Tukey post hoc multiple comparison. Bronchial basal epithelial cell mediator release in response to Poly IC or IL-1β and IFN-γ activation were inhibited by IKK2i in a concentration-dependent manner but not by prednisolone (e-Fig 3). The addition of inflammatory stimuli and compounds did not affect cell viability. Prednisolone (10 μM) and IKK2i (1 μm) were used to assess the effect upon basal cell and ALI epithelial synthetic capacity. IKK2i reduced the release of nearly all mediators by basal and ALI epithelial cells in response to poly-IC, IL-1β or IL-1β, and IFN-γ, and the effect was greater in basal epithelial cells. Prednisolone had minimal effects upon the mediator release (e-Fig 4, Tables 4-6). Differences between efficacies of the IKK2 inhibitor and prednisolone were significant for the majority of mediators and stimuli for basal cells (e-Table 7). For poly IC-stimulated ALI epithelial cells, differences between the efficacy of the IKK2 inhibitor and prednisolone were significant for CCL11, CXCL10, CCL2, CCL4, CCL5, (P < .001), CCL17 (P < .01), and CCL26 and CCL22, (P < .05) (e-Table 8).

icant for the majority of mediators and stimuli for basal cells (e-Table 7). For poly IC-stimulated ALI epithelial cells, differences between the efficacy of the IKK2 inhibitor and prednisolone were significant for CCL11, CXCL10, CCL2, CCL4, CCL5, (P < .001), CCL17 (P < .01), and CCL26 and CCL22, (P < .05) (e-Table 8). Table 4 —Prednisolone and IKK2i Modulation of Poly IC-Mediated Bronchial Epithelial Cell Mediator Release

icant for the majority of mediators and stimuli for basal cells (e-Table 7). For poly IC-stimulated ALI epithelial cells, differences between the efficacy of the IKK2 inhibitor and prednisolone were significant for CCL11, CXCL10, CCL2, CCL4, CCL5, (P < .001), CCL17 (P < .01), and CCL26 and CCL22, (P < .05) (e-Table 8). Table 4 —Prednisolone and IKK2i Modulation of Poly IC-Mediated Bronchial Epithelial Cell Mediator Release Mediators Bronchial Basal (n = 17) Bronchial Differentiated (n = 11) Poly IC (12.5 μg/mL) Poly IC (12.5 μg/mL) Poly IC (12.5 μg/mL) Poly IC (12.5 μg/mL) pg/mL/106 cells + Pred + Ikk2i pg/mL/106 cells + Pred + Ikk2i IFN-β 72.4 (22.5-233.1) 1.3 (1.0-1.3) 0.7 (0.4-1.0)a 15.8 (11.6-21.4) 0.9 (0.9-1.0) 0.8 (0.5-1.3) CCL5 11,352 (5,614-22,956) 1.1 (0.9-1.3) 0.1 (0.1-0.3)b 626.0 (194.6-2,014.0) 1.0 (0.5-1.9) 0.2 (0.1-0.4)b IL-1β 541.2 (313.2-935.1) 0.9 (0.7-1.1) 0.4 (0.2-0.7)c 60.3 (20.3-178.9) 0.9 (0.7-1.1) 1.0 (0.5-2.0) CXCL8 53,004 (34,652-81,076) 1.0 (0.9-1.1) 0.1 (0.1-0.3)b 22,319 (16,126.0-30,891.0) 0.9 (0.8-1.0)a 0.8 (0.7-1.1) CCL2 412.8 (180.9-941.9) 1.2 (0.9-1.1) 0.1 (0.0-0.2)b 1,915.0 (582.3-6,297.0) 0.9 (0.8-1.1) 0.2 (0.1-0.4)b CCL13 2,287 (1,323-3,955) 1.1 (0.9-1.4) 0.2 (0.1-0.3)b 1,326.0 (717.2-2,450.0) 1.0 (0.8-1.3) 0.7 (0.6-1.0) CCL4 2,912 (1,756-4,829) 1.0 (0.8-1.2) 0.0 (0.0-0.0)b 865.3 (304.4-2,460.0) 0.8 (0.6-1.2) 0.2 (0.1-0.6)c TNF-α 1,630 (1,015-2,618) 0.7 (0.6-0.9)c 0.1 (0.0-0.1)b 540.4 (234.8-1,244.0) 0.7 (0.4-1.1) 0.4 (0.1-1.1) CCL22 4,981 (2,502-9,916) 1.1 (0.9-1.4) 0.1 (0.1-0.3)b 2,621.0 (1,428.0-4,811.0) 0.9 (0.8-1.1) 0.6 (0.5-0.9)c CCL17 3,143 (1,468-6,730) 1.2 (0.9-1.6) 0.1 (0.0-0.3)b 1,434.0 (707.0-2,910.0) 0.9 (0.6-1.5) 0.5 (0.3-0.8)a CCL26 14,855 (8,918-2,4745) 1.0 (0.9-1.4) 0.2 (0.0-0.3)b 10,671.0 (5,806-19,614.0) 0.9 (0.7-1.3) 0.5 (0.3-0.9)a CXCL10 70,950 (28,845-174,515) 1.0 (0.9-1.2) 0.0 (0.0-0.1)b 27,465 (10,525-71,667) 0.9 (0.6-1.5) 0.1 (0.0-0.3)b CCL11 2,433 (1,376-4,302) 1.1 (0.9-1.4) 0.1 (0.0-0.2)b 1,209.0 (539.4-2,711.0) 1.0 (0.8-1.1) 0.4 (0.2-0.7) Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 legend for expansion of abbreviations.

0.2)b 1,209.0 (539.4-2,711.0) 1.0 (0.8-1.1) 0.4 (0.2-0.7) Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 legend for expansion of abbreviations. a P ≤ .05 compared with constitutive levels. b P < .001 compared with constitutive levels. c P ≤ .01 compared with constitutive levels. Table 5 —Prednisolone and IKK2i Modulation of IL-1β-Mediated Bronchial Epithelial Cell Mediator Release

0.2)b 1,209.0 (539.4-2,711.0) 1.0 (0.8-1.1) 0.4 (0.2-0.7) Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 legend for expansion of abbreviations. a P ≤ .05 compared with constitutive levels. b P < .001 compared with constitutive levels. c P ≤ .01 compared with constitutive levels. Table 5 —Prednisolone and IKK2i Modulation of IL-1β-Mediated Bronchial Epithelial Cell Mediator Release Mediators Bronchial Basal (n = 11) Bronchial Differentiated (n = 11) IL-1β (10 ng/mL) IL-1β (10 ng/mL) IL-1β (10 ng/mL) IL-1β (10 ng/mL) pg/mL/106 Cells + Pred + Ikk2i pg/mL/106 Cells + Pred + Ikk2i IFN-β 34.8 (13.1-92.8) 1.1 (0.8-1.4) 1.1 (0.8-1.3) 13.5 (11.3-16.2) 1.0 (0.8-1.4) 1.0 (0.8-1.2) CCL5 82.6 (29.1-234.7) 1.2 (0.7-2.2) 0.2 (0.1-0.7)a 92.24 (25.9-328.3) 1.1 (0.9-1.3) 0.1 (0.0-0.4)b IL-1β N/A N/A N/A N/A N/A N/A CXCL8 51,154 (25,440-102,859) 0.9 (0.8-1.0)a 0.2 (0.1-0.4)c 17,828 (9,334-34,052) 1.0 (0.7-1.5) 1.2 (0.7-2.2) CCL2 101.3 (31.0-330.3) 1.0 (0.8-1.2) 0.4 (0.2-0.8)a 1,908.0 (488.9-7,446.0) 1.1 (0.8-1.5) 0.2 (0.1-0.4)c CCL13 1,140 (431.4-3,010.0) 1.0 (0.9-1.1) 0.3 (0.1-1.5) 664.4 (359.7-1,227.0) 1.1 (0.9-1.2) 0.9 (0.8-1.1) CCL4 171.6 (41.3-713.5) 1.0 (0.6-1.5) 0.2 (0.1-0.5)b 351.0 (136.0-905.9) 1.1d (1.0-1.3)a 0.4 (0.2-0.8)a TNF-α 919.8 (436.1-1,940.0) 1.2 (0.8-1.6) 0.3 (0.2-0.5)c 277.3 (100.4-765.9) 0.9 (0.6-1.2) 0.3 (0.1-0.9)a CCL22 1,498 (563-3,986) 0.9 (0.7-1.1) 0.4 (0.2-1.1) 1,657.0 (875.4-3,135.0) 0.9 (0.8-1.0) 0.7 (0.6-0.9)a CCL17 554.7 (203.0-1,515.0) 1.2 (0.9-1.6) 0.9 (0.6-1.3) 715.5 (383.2-1,336.0) 1.1 (0.9-1.2) 0.8 (0.6-1.0) CCL26 1,581.0 (397.5-6,286.0) 3.1d (1.4-6.4)b 0.6 (0.1-4.0) 12,606.0 (4,420.0-35,952.0) 1.3 (0.7-2.1) 0.8 (0.4-1.6) CXCL10 1,189.0 (367.9-3,843.0) 0.8 (0.5-1.3) 0.3 (0.1-1.2) 1,988.0 (754.4-5,237.0) 0.9 (0.5-1.5) 0.4 (0.2-0.8)a CCL11 419.7 (143.3-1,229.0) 1.0 (0.7-1.4) 0.7 (0.4-1.1) 433.5 (231.6-811.5) 1.0 (0.9-1.2) 0.8 (0.6-1.0)a Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 and 2 legends for expansion of abbreviations.

) 433.5 (231.6-811.5) 1.0 (0.9-1.2) 0.8 (0.6-1.0)a Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 and 2 legends for expansion of abbreviations. a P ≤ .05 compared with constitutive levels. b P ≤ .01 compared with constitutive levels. c P < .001 compared with constitutive levels. d Significant upregulation. Table 6 —Prednisolone and IKK2i Modulation of IL-1β and IFN-γ-Mediated Bronchial Epithelial Cell Mediator Release

) 433.5 (231.6-811.5) 1.0 (0.9-1.2) 0.8 (0.6-1.0)a Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 and 2 legends for expansion of abbreviations. a P ≤ .05 compared with constitutive levels. b P ≤ .01 compared with constitutive levels. c P < .001 compared with constitutive levels. d Significant upregulation. Table 6 —Prednisolone and IKK2i Modulation of IL-1β and IFN-γ-Mediated Bronchial Epithelial Cell Mediator Release Mediators Bronchial Basal (n = 17) Bronchial Differentiated (n = 11) IL-1β/IFN-γ (10 ng/mL) IL-1β/IFN-γ (10 ng/mL) IL-1β/ IFN-γ (10 ng/mL) IL-1β/IFN-γ (10 ng/mL) pg/mL/106 Cells + Pred + Ikk2i pg/mL/106 Cells + Pred + Ikk2i IFN-β 35.6 (14.4-87.7) 0.9 (0.6-1.5) 1.4 (0.7-2.5) 13.2 (12.2-14.2) 1.3 (0.9-1.8) 1.0 (0.9-1.1) CCL5 1,076 (509.6-2,271.0) 1.2 (0.9-1.7) 0.0 (0.0-0.0)a 397.2 (144.9-1,089.0) 1.0 (0.7-1.3) 0.1 (0.1-0.2)a IL-1β N/A N/A N/A N/A N/A N/A CXCL8 49,032 (30,759-78,160) 1.0 (0.8-1.2) 0.1 (0.1-0.2)a 17,947 (11,053-29,142) 1.0 (0.9-1.1) 1.0 (0.9-1.2) CCL2 3,841 (1,588-9,287) 1.3 (0.7-2.5) 0.2 (0.1-0.5)b 6,365.0 (2,222.0-18,230.0) 0.8 (0.6-1.1) 0.4 (0.2-0.6)a CCL13 2,649 (1,447-4,849) 1.2 (0.9-1.7) 0.6 (0.3-0.9)c 1,744.0 (1,059.0-2,873.0) 1.1 (0.9-1.4) 0.8 (0.6-1.1) CCL4 996.4 (469.8-2,113.0) 1.4 (0.8-2.4) 0.4 (0.2-0.9)c 1,326.0 (720.9-2,440.0) 1.2 (0.9-1.7) 0.7 (0.4-1.1) TNF-α 648.6 (327.0-1,286.0) 1.0 (0.9-1.2) 0.3 (0.2-0.5)a 262.4 (110.7-621.9) 1.0 (0.6-1.4) 0.5 (0.2-1.0) CCL22 6,206 (3,124-12,328) 1.3 (0.8-1.9) 0.6 (0.4-1.1) 4,879.0 (2,951.0-8,068.0) 1.1 (0.9-1.4) 0.8 (0.6-1.1) CCL17 4,871 (2,392-9,917) 1.4 (0.8-2.2) 0.6 (0.3-1.1) 3,688.0 (2,239.0-6,074.0) 1.1 (0.9-1.5) 0.7 (0.5-1.1) CCL26 21,244 (13,104-344,442) 1.1 (0.9-1.4) 0.5 (0.3-0.8)b 47,002.0 (23,802.0-92,816.0) 1.3 (0.7-2.1) 0.6 (0.3-1.3) CXCL10 141,993 (84,327-239,091) 1.1 (1.0-1.2) 0.8 (0.7-1.0)c 222,457 (10,0350-493,146) 0.8 (0.7-0.9)c 0.7 (0.5-0.9)c CCL11 3,145 (1,675-5,907) 1.4 (0.9-2.2) 0.7 (0.4-1.3) 2,409.0 (1,431.0-4,055.0) 1.1 (0.9-1.3) 0.7 (0.5-1.0) Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 and 2 legends for expansion of abbreviations.

09.0 (1,431.0-4,055.0) 1.1 (0.9-1.3) 0.7 (0.5-1.0) Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 and 2 legends for expansion of abbreviations. a P < .001 compared with constitutive levels. b P ≤ .01 compared with constitutive levels. c P ≤ .05 compared with constitutive levels.

09.0 (1,431.0-4,055.0) 1.1 (0.9-1.3) 0.7 (0.5-1.0) Stimulated levels (pg/mL/10 cells) in conjunction with modulation by prednisolone (Pred) (10 μM) or iKK2 (1 μm) (fold change over stimulated levels) after 24 h. All data are expressed as geometric mean (95% CI). See Table 1 and 2 legends for expansion of abbreviations. a P < .001 compared with constitutive levels. b P ≤ .01 compared with constitutive levels. c P ≤ .05 compared with constitutive levels. Discussion We report here for the first time, to our knowledge, differences in the constitutive production of chemokines and cytokines in vitro between nasal basal, bronchial basal, and bronchial differentiated epithelial cells, highlighting the importance of the location of the airway epithelium in relation to its mediator production. We were unable to demonstrate differences in the constitutive release of mediators from epithelial cells derived from subjects with asthma vs healthy control subjects from any of the epithelial cell types. There was marked up-regulation in mediators in response to proinflammatory cytokines and toll-like receptor (TLR)-3 stimulation, which was greater in nasal than bronchial epithelial cells. However, we did not identify differential response to these stimuli between health and disease. This increased expression was broadly corticosteroid unresponsive in epithelial cells from both subjects with asthma and healthy control subjects, suggesting this activation of synthetic capacity is via corticosteroid-insensitive pathways that are not disease specific. In contrast, the release of these mediators was markedly attenuated by IKK2 inhibition, suggesting that this novel antiinflammatory therapy may have potential to modulate corticosteroid-unresponsive pathways in asthma.

ation of synthetic capacity is via corticosteroid-insensitive pathways that are not disease specific. In contrast, the release of these mediators was markedly attenuated by IKK2 inhibition, suggesting that this novel antiinflammatory therapy may have potential to modulate corticosteroid-unresponsive pathways in asthma. The view that intrinsic differences persist in vitro between epithelial cells from subjects with asthma vs healthy control subjects is controversial. Here we add to this debate, as our data were unable to identify differential expression of a broad range of chemokines or cytokines from primary epithelial cells in health compared with disease either constitutively or following stimulation. Deficiency in type 1 and 3 IFNs following viral infection in asthma is proposed to be a fundamental impairment in the innate immune response promoting persistence of a viral infection and the development of an exacerbation of asthma.14,21 This deficient response may be in part due to chronic airway inflammation in asthma or due to intrinsic differences in asthmatic epithelium. However, we and others13 were unable to confirm this finding in vitro. Taken together, these data suggest that the interferon release by epithelial cells is heterogeneous, and although this abnormality in innate immunity may be important in some individuals, it is unlikely to represent a defining characteristic of asthma.

ever, we and others13 were unable to confirm this finding in vitro. Taken together, these data suggest that the interferon release by epithelial cells is heterogeneous, and although this abnormality in innate immunity may be important in some individuals, it is unlikely to represent a defining characteristic of asthma. Interestingly there was a differential synthetic capacity of epithelial cells from different locations and states of differentiation. This was particularly apparent for IFN-β, CCL5, IL-1β, CXCL8, and CCL2. These may represent important functional differences in wound repair and host defense. In addition to these differences in constitutive mediator release, there were key differences in release following stimulation. For example, CCL2 has been implicated in wound repair and recruitment of monocytes22,23 and fibrocytes24; its release was increased constitutively, and following stimulation by IL-1β there was marked up-regulation of release by the differentiated epithelial cells compared with increased release by nasal-derived basal cells following stimulation with poly-IC. IFN-β was increased in nasal and bronchial basal cells constitutively compared with differentiated cells. Similarly, there was a generalized increase in most chemokines and cytokines in the nasal epithelial basal cells following Poly IC stimulation compared with the bronchial epithelial cells. This perhaps suggests a heightened sensitivity of these cells to TLR3 activation, which may be a consequence of epigenetic changes promoted by previous exposure to viruses. These differences have hitherto been unexplored and highlight the potential importance of epithelial cell type in the study of epithelial function in asthma.

ps suggests a heightened sensitivity of these cells to TLR3 activation, which may be a consequence of epigenetic changes promoted by previous exposure to viruses. These differences have hitherto been unexplored and highlight the potential importance of epithelial cell type in the study of epithelial function in asthma. The concentration of prednisolone used in this investigation was representative of the plasma concentration present in subjects with severe asthma taking such medication.25 It is very interesting and an important finding that the corticosteroid used had little effect on cytokine production, in contrast to IKK2 inhibition. The observed corticosteroid epithelial relative resistance, but responsiveness toward inhibition of IKK2 in donors with and without asthma, would suggest that corticosteroid epithelial unresponsiveness is related to features of the canonical NF-κB signaling pathway. These data, therefore, do not support the view that there is an intrinsic corticosteroid unresponsiveness in asthmatic epithelium as a consequence of differential corticosteroid receptor expression, altered ligand affinity, or decreased corticosteroid binding to DNA.26 Indeed, although IL-8 expression has been shown to be significantly repressed by dexamethasone, the induction of NF-κB-dependent transcription in primary human bronchial epithelial cells was unaffected by the application of the corticosteroid, suggesting that the targeting of NF-κB transcription by corticosteroids may represent a more minor mechanism of action than has been previously believed.27

y dexamethasone, the induction of NF-κB-dependent transcription in primary human bronchial epithelial cells was unaffected by the application of the corticosteroid, suggesting that the targeting of NF-κB transcription by corticosteroids may represent a more minor mechanism of action than has been previously believed.27 There is a potential for using small molecule inhibitors of IKK2 to treat asthma. These compounds bypass problems associated with corticosteroid-resistant therapy, such as reduced corticosteroid receptor expression and translocation.28 To date, such small molecule inhibitors have been proven to significantly inhibit NF-κB transcription and inflammatory gene expression in primary human airway epithelial cells,27 comprehensively reduce chronic pulmonary inflammation in murine models,29 and inhibit pathologic features of airway remodelling and ameliorate airway responsiveness in a chronic allergen exposure model of bronchial asthma in mice.30

nscription and inflammatory gene expression in primary human airway epithelial cells,27 comprehensively reduce chronic pulmonary inflammation in murine models,29 and inhibit pathologic features of airway remodelling and ameliorate airway responsiveness in a chronic allergen exposure model of bronchial asthma in mice.30 The effects of the site of sampling, selection of subjects, activators, and inhibitors upon mediator release are important to consider. Our study focused upon synthetic capacity of epithelial cells in vitro to determine the persistence of abnormalities in asthmatic cells independent of the in vivo environment. However, one potential criticism is that we may have failed to observe other important functional differences. Indeed, our findings do not exclude the possibility that there remain fundamental differences in epithelial cell function in asthma. Critical differences between epithelial cells from asthma and health for other functional responses are reported for proliferation9,31 and wound repair.9 Asthma severity may also be an important factor, although we recruited subjects across the spectrum of asthma severity and were unable to demonstrate differences in synthetic capacity and disease severity. However, our study was underpowered to fully explore this question, and, thus, it remains a possibility that disease severity may exert an important influence upon epithelial cell function. Although the cells were stimulated basolaterally, it is acknowledged that mucus production on the apical surface may have still influenced the function of the cell culture as a whole. To reduce the interference of mucus production with the production of cytokines, mucus was removed from ALI cultures prior to simulation by gentle washing with PBS. There is also the possibility that an alternative duration of stimulation and even apical stimulation of the differentiated epithelial cultures may have produced differences between the donors with and without asthma. Indeed, there is evidence to suggest variation in apical and basolateral secretion of mediators, but this may be due to the location of receptors for certain stimuli.32 The IFN-γ receptor is expressed basolaterally in both cultured airway epithelia and normal human airway tissue. Therefore, stimulating the ALI cultures from the basolateral surface allowed for the maximal response to be examined from the differentiated epithelium.

due to the location of receptors for certain stimuli.32 The IFN-γ receptor is expressed basolaterally in both cultured airway epithelia and normal human airway tissue. Therefore, stimulating the ALI cultures from the basolateral surface allowed for the maximal response to be examined from the differentiated epithelium. Furthermore, in vitro conditions may also be important determinants of functional responses, such as the matrix environment may influence some of the differences observed between reports9,11,12; indeed, the effects of culture conditions upon the differentiation process and its dynamic during culture could have resulted in problems regarding the interpretation of data. This is unlikely to have affected our study, as we controlled carefully the culture conditions for the epithelial cells. Samples were obtained from the second- or third-generation bronchi and cultured on collagen-coated surfaces. Spontaneous mucus production occurred from 14 to 21 days and ciliation between 25 to 35 days. Those cultures that did not reach the final stages of differentiation were not used for this investigation. One further potential criticism is that we did not undertake corrections for multiple comparisons. We chose not to undertake these corrections, as there is a debate among statisticians about the validity of various methods. If a Bonferroni correction was applied, the revised P value for significance would have been adjusted from P = .05 to P = .0039.

is that we did not undertake corrections for multiple comparisons. We chose not to undertake these corrections, as there is a debate among statisticians about the validity of various methods. If a Bonferroni correction was applied, the revised P value for significance would have been adjusted from P = .05 to P = .0039. In conclusion, the synthetic capacity of epithelial cells differs between location and the degree of differentiation, but in vitro we were unable to identify differential expression between health and disease, suggesting that activation of epithelial cells in vivo maybe largely a consequence of the asthmatic environment. Activation of epithelial cells by proinflammatory cytokines and TLR3 agonists is corticosteroid unresponsiveness independent of disease, but is sensitive to IKK2 inhibition, suggesting that IKK2 inhibitors may be important novel therapies for asthma. Supplementary Material Online Supplement Click here for additional data file. Author contributions: Dr Brightling is guarantor for the manuscript. Dr Woodman: contributed to performing the multiplex assays, analyzing the data, writing the manuscript, and approving the final manuscript. Ms Wan: contributed to performing the multiplex assays and approving the final manuscript. Ms Milone: contributed to performing the multiplex assays and approving the final manuscript. Dr Grace: contributed to performing the multiplex assays and approving the final manuscript. Dr Sousa: contributed to writing the manuscript and approving the final manuscript.

Ms Wan: contributed to performing the multiplex assays and approving the final manuscript. Ms Milone: contributed to performing the multiplex assays and approving the final manuscript. Dr Grace: contributed to performing the multiplex assays and approving the final manuscript. Dr Sousa: contributed to writing the manuscript and approving the final manuscript. Dr Williamson: contributed to writing the manuscript and approving the final manuscript. Dr Brightling: contributed to writing the manuscript and approving the final manuscript. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Ms Milone was an employee of GlaxoSmithKline (GSK) at the time of manuscript preparation. Dr Grace was an employee and shareholder of GSK at the time of manuscript preparation. Dr Sousa was an employee of GSK at the time of manuscript preparation. Dr Williamson was an employee of GSK at the time of manuscript preparation. Dr Brightling has received grant monies and consultancy fees from pharmaceutical companies. Dr Woodman and Ms Wan have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: This study was sponsored by the University Hospitals of Leicester NHS Trust. The sponsor and funders approve this study; all data were generated, analyzed, and interpreted by the authors. Additional information: The e-Figures and e-Tables can be found in the “Supplemental Materials” area of the online article.

Role of sponsors: This study was sponsored by the University Hospitals of Leicester NHS Trust. The sponsor and funders approve this study; all data were generated, analyzed, and interpreted by the authors. Additional information: The e-Figures and e-Tables can be found in the “Supplemental Materials” area of the online article. Funding/Support: This study was funded by GlaxoSmithKline and a Wellcome Senior Clinical Fellowship (Dr Brightling). Novel compounds were supplied by GlaxoSmithKline. Abbreviations ALIair-liquid interface ANOVAanalysis of variance BEGMbronchial epithelial growth medium ELISAenzyme-linked immunosorbent assay IFNinterferon IκBinhibitory κB IKKIκB kinase NF-κBnuclear factor κB TLRToll-like receptor TNFtumor necrosis factor

Pulmonary hypertension (PH) is an uncommon but progressive condition. In the past, it has been called an orphan disease because it affects small numbers of individuals, is associated with many diseases, and is often overlooked by doctors.1 Previous surveillance from the Centers for Disease Control and Prevention2 from 1980 to 2002 identified decreasing mortality rates associated with PH among men, but increasing mortality rates among women, along with stable rates among whites but increasing rates among blacks. Increasing rates of hospitalization associated with PH were identified as well. The symptoms of PH during the initial stage of the disease are common to many other medical conditions (eg, difficulty breathing, fatigue), often resulting in a delayed diagnosis until more severe symptoms arise (eg, dizziness, chest pain, ankle swelling, palpitations).3,4

on associated with PH were identified as well. The symptoms of PH during the initial stage of the disease are common to many other medical conditions (eg, difficulty breathing, fatigue), often resulting in a delayed diagnosis until more severe symptoms arise (eg, dizziness, chest pain, ankle swelling, palpitations).3,4 PH is characterized by increased pressure in the pulmonary arteries (resting mean pulmonary artery pressure ≥ 25 mm Hg) and increased pulmonary arterial resistance but it is associated with many underlying conditions.5 The World Health Organization classification of PH, known as the Dana Point classification, was last updated in 20086 (Table 1), after the most recent PH surveillance summary from the Centers for Disease Control and Prevention.2 Some common underlying causes include pulmonary arterial hypertension (PAH) from congenital heart disease, connective tissue disease, or persistent PH of the newborn; PH due to left-sided heart disease; chronic lung diseases and hypoxemia; and chronic thromboembolic pulmonary disease. Genetics also plays a role in PH, and although PH occurs at all ages, the incidence increases with age. Registries from France and the United Kingdom report an incidence rate of 1.1 to 2.4 cases per million population per year, a prevalence of 6.6 to 15.0 cases per million population per year for PAH, and a 5-year mortality of approximately 40%.7 TABLE 1 ] World Health Organization Dana Point Classification of Pulmonary Hypertension (2008)

PH is characterized by increased pressure in the pulmonary arteries (resting mean pulmonary artery pressure ≥ 25 mm Hg) and increased pulmonary arterial resistance but it is associated with many underlying conditions.5 The World Health Organization classification of PH, known as the Dana Point classification, was last updated in 20086 (Table 1), after the most recent PH surveillance summary from the Centers for Disease Control and Prevention.2 Some common underlying causes include pulmonary arterial hypertension (PAH) from congenital heart disease, connective tissue disease, or persistent PH of the newborn; PH due to left-sided heart disease; chronic lung diseases and hypoxemia; and chronic thromboembolic pulmonary disease. Genetics also plays a role in PH, and although PH occurs at all ages, the incidence increases with age. Registries from France and the United Kingdom report an incidence rate of 1.1 to 2.4 cases per million population per year, a prevalence of 6.6 to 15.0 cases per million population per year for PAH, and a 5-year mortality of approximately 40%.7 TABLE 1 ] World Health Organization Dana Point Classification of Pulmonary Hypertension (2008) Classification 1. Pulmonary arterial hypertension 1.1. Idiopathic 1.2. Heritable 1.2.1. BMPR2 1.2.2. ALK1 endoglin (with or without hereditary hemorrhagic telangiectasia 1.2.3. Unknown 1.3. Drug and toxin induced 1.4. Associated with 1.4.1. Connective tissue diseases 1.4.2. HIV infection 1.4.3. Portal hypertension 1.4.4. Congenital heart diseases 1.4.5. Shistosomiasis 1.4.6. Chronic hemolytic anemia 1.5. Persistent pulmonary hypertension of the newborn 1.6. Pulmonary venoocclusive disease 2. Pulmonary hypertension owing to left-sided heart disease 2.2. Systolic dysfunction 2.2. Diastolic dysfunction 2.3. Valvular disease 3. Pulmonary hypertension owing to lung diseases and/or hypoxia 3.1. COPD 3.2. Interstitial lung disease 3.3. Other pulmonary diseases with mixed restrictive and obstructive pattern 3.4. Sleep-disordered breathing 3.5. Alveolar hypoventilation disorders 3.6. Chronic exposure to high altitude 3.7. Development abnormalities 4. Chronic thromboembolic pulmonary hypertension 5. Pulmonary hypertension with unclear multifactorial mechanisms 5.1. Hematologic disorders: myeloproliferative disorders, splenectomy 5.2. Systemic disorders: sarcoidosis, pulmonary Langerhans’ cell histiocytosis 5.3. Metabolic disorders: glycogen storage disease, Gaucher disease, thyroid disorders 5.4. Others: tumoral obstruction fibrosing mediastinitis, chronic renal failure or dialysis ALK1 = activin receptor-like kinase type 1; BMPR2 = bone morphogenetic protein receptor type 2. (Adapted with permission from Simonneau et al.6)

. Metabolic disorders: glycogen storage disease, Gaucher disease, thyroid disorders 5.4. Others: tumoral obstruction fibrosing mediastinitis, chronic renal failure or dialysis ALK1 = activin receptor-like kinase type 1; BMPR2 = bone morphogenetic protein receptor type 2. (Adapted with permission from Simonneau et al.6) Much of what we know about PH comes from specialized disease registries.8‐10 With expanding research into the diagnosis and treatment of PH, it is important to provide updated surveillance on the impact of this disease on hospitalizations and mortality. The surveillance report by Hyduk et al2 described trends in mortality and hospitalization rates associated with PH among adults aged 45 years and older from 1980 through 2002 by demographic characteristics. This study builds on previous PH surveillance of mortality and hospitalization using data from the National Vital Statistics System and the National Hospital Discharge Survey (NHDS).11 The purpose of our report is to describe trends in diagnosed PH-related mortality and hospitalizations during the period 2001 to 2010. Because PH is frequently reported as a secondary diagnosis, our report presents data for PH as any contributing cause of death or as any listed hospital diagnosis.

arge Survey (NHDS).11 The purpose of our report is to describe trends in diagnosed PH-related mortality and hospitalizations during the period 2001 to 2010. Because PH is frequently reported as a secondary diagnosis, our report presents data for PH as any contributing cause of death or as any listed hospital diagnosis. Materials and Methods Mortality Mortality data from the National Vital Statistics System for the period 2001 to 2010 were analyzed. Bridged-race July 1 population estimates produced by the US Census Bureau in collaboration with the National Center for Health Statistics were compiled using intercensal estimates for the period 2001 to 2009 and postcensal estimates for 2010. For this report, all diseases and conditions reported on death certificates were classified according to codes from the International Classification of Diseases, 10th Revision (ICD-10). For this analysis, PH deaths are defined as those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including underlying cause) on the death certificate. An ICD coding change for mortality occurred in 2003 with the addition of ICD-10 code I27.2 for secondary PH. This resulted in a shift from coding most cases of death related to PH from primary PH (ICD-10 I27.0) to other secondary PH (Fig 1,12 Table 2). These changes require careful interpretation of PH surveillance data over time when relying on individual ICD-10 codes.13

th the addition of ICD-10 code I27.2 for secondary PH. This resulted in a shift from coding most cases of death related to PH from primary PH (ICD-10 I27.0) to other secondary PH (Fig 1,12 Table 2). These changes require careful interpretation of PH surveillance data over time when relying on individual ICD-10 codes.13 Figure 1 – International Classification of Diseases coding for pulmonary hypertension mortality: United States, 2001-2010. I27.0 = primary pulmonary hypertension; I27.2 = other secondary pulmonary hypertension; I27.8 = other specified pulmonary heart diseases; and I27.9 = pulmonary heart disease, unspecified. Data are from the Multiple Cause of Death Files, 1999-2010, as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program. (Reprinted from the Centers for Disease Control and Prevention, National Center for Health Statistics.12) TABLE 2 ] ICD-CM Coding for Diagnosis During Hospitalizations and ICD for Mortality Used in the United States Since 2001

Figure 1 – International Classification of Diseases coding for pulmonary hypertension mortality: United States, 2001-2010. I27.0 = primary pulmonary hypertension; I27.2 = other secondary pulmonary hypertension; I27.8 = other specified pulmonary heart diseases; and I27.9 = pulmonary heart disease, unspecified. Data are from the Multiple Cause of Death Files, 1999-2010, as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program. (Reprinted from the Centers for Disease Control and Prevention, National Center for Health Statistics.12) TABLE 2 ] ICD-CM Coding for Diagnosis During Hospitalizations and ICD for Mortality Used in the United States Since 2001 ICD-CM Diagnoses During Hospitalizations ICD for Mortality ICD-9-CM 2001 ICD-9-CM 2010 ICD-10-CM 2013 ICD-10 2001 ICD-10 2003 416. Pulmonary heart disease 416. Pulmonary heart disease I27. Other pulmonary heart diseases I27. Other pulmonary heart disease I27. Other pulmonary heart disease 416.0. Primary pulmonary hypertension; idiopathic pulmonary arteriosclerosis; pulmonary hypertension (essential) (idiopathic) (primary) 416.0. Primary pulmonary hypertension; idiopathic pulmonary arteriosclerosis; pulmonary hypertension (essential) (idiopathic) (primary) I27.0. Primary pulmonary hypertension I27.0. Primary pulmonary hypertension I27.0. Primary pulmonary hypertension 416.1 Kyphoscoliotic heart disease 416.1. Kyphoscoliotic heart disease I27.1. Kyphoscoliotic heart disease I27.1. Kyphoscoliotic heart disease I27.1. Kyphoscoliotic heart disease 416.2. Chronic pulmonary embolism I27.2. Other secondary pulmonary hypertension (pulmonary hypertension NOS) I27.2. Other secondary pulmonary hypertension 416.8. Other specified diseases of pulmonary circulation (pulmonary arteritis, endarteritis, rupture or stricture of pulmonary vessel 416.8. Other specified diseases of pulmonary circulation (pulmonary arteritis, endarteritis, rupture or stricture of pulmonary vessel I27.8. Other specified pulmonary heart diseases I27.8. Other specified pulmonary heart diseases I27.8. Other specified pulmonary heart diseases I27.81. Cor pulmonale (chronic) I27.82. Chronic pulmonary embolism I27.89. Other specified pulmonary heart diseases (Eisenmenger’s complex and syndrome) 417.9. Unspecified disease of pulmonary circulation 417.9. Unspecified disease of pulmonary circulation I27.9. Pulmonary heart disease, unspecified (chronic cardiopulmonary disease) I27.9. Pulmonary heart disease, unspecified I27.9. Pulmonary heart disease, unspecified ICD = International Classification of Diseases; ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification; ICD-10 = International Classification of Diseases, Tenth Revision; ICD-10-CM = International Classification of Diseases, Tenth Revision, Clinical Modification; NOS = not otherwise specified.

rnational Classification of Diseases; ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification; ICD-10 = International Classification of Diseases, Tenth Revision; ICD-10-CM = International Classification of Diseases, Tenth Revision, Clinical Modification; NOS = not otherwise specified. Rates are expressed per 100,000 population and were directly age standardized to the 2000 US standard population and eight age groups (0-12 months followed by those aged 1 to 34 years, 35 to 44 years, 45 to 54 years, 55 to 64 years, 65 to 74 years, 75 to 84 years, and ≥ 85 years). Age-standardized death rates and 95% CIs were calculated by sex, race/ethnicity (ie, non-Hispanic [NH] white, NH black, NH American Indian or Alaska native (AI/AN), NH Asian/Pacific Islander, Hispanic), and the decedent’s state of residence at time of death. Age-specific death rates were calculated. It should be noted that race and ethnicity may not be captured accurately on the death certificate, especially for Hispanics and races other than black or white.14 This may affect the observed death rates for Hispanics and other nonwhite/nonblack individuals, especially AI/AN decedents.

fic death rates were calculated. It should be noted that race and ethnicity may not be captured accurately on the death certificate, especially for Hispanics and races other than black or white.14 This may affect the observed death rates for Hispanics and other nonwhite/nonblack individuals, especially AI/AN decedents. Trend analyses and comparability tests for age-standardized or age-specific death rates plotted over time by year were conducted using Joinpoint software, developed by the National Cancer Institute. (Joinpoint version 4.0.3; National Cancer Institute). The number of trend segments is based on a segmented line regression analysis of best fit with the smallest number of “joinpoints” or points (0 or 1) at which the direction of the trend line changes. Annual percent change (APC) was calculated for each of the trend segments. Average annual percent change (AAPC) was calculated for the time period 2001 to 2010 to quantify the average trend over this time period. To determine whether the trend lines were parallel or coincident, tests were conducted to assess pairwise differences between race/ethnicity, with NH white as the referent; age-specific differences, with 0 to 12 months as the referent; and sex differences, with men as the referent. Tests determined whether (1) two joinpoint regression functions were identical (test of coincidence) or (2) two regression mean functions were parallel (test of parallelism) at P < .05.

H white as the referent; age-specific differences, with 0 to 12 months as the referent; and sex differences, with men as the referent. Tests determined whether (1) two joinpoint regression functions were identical (test of coincidence) or (2) two regression mean functions were parallel (test of parallelism) at P < .05. The distribution of selected disease categories reported as the underlying cause of death (UCOD) among decedents with reported PH was examined for 2001 through 2010 by race/ethnicity and sex. Rates and trends for these distributions were calculated over that time period.

H white as the referent; age-specific differences, with 0 to 12 months as the referent; and sex differences, with men as the referent. Tests determined whether (1) two joinpoint regression functions were identical (test of coincidence) or (2) two regression mean functions were parallel (test of parallelism) at P < .05. The distribution of selected disease categories reported as the underlying cause of death (UCOD) among decedents with reported PH was examined for 2001 through 2010 by race/ethnicity and sex. Rates and trends for these distributions were calculated over that time period. Hospitalizations The NHDS, conducted annually from 1965 to 2010 by the Centers for Disease Control and Prevention’s National Center for Health Statistics, is a national probability sample survey of hospital discharge information abstracted from the medical records of inpatients from nonfederal, short-stay general hospitals in the United States.15 Only hospitals with an average length of stay of fewer than 30 days for all patients, and general and children’s general hospitals, were included in the survey. Excluded were federal, military, and Veterans Affairs hospitals, as well as hospital units of institutions, such as prison hospitals, and hospitals with fewer than six beds staffed for patient use. NHDS data from 2001 to 2010 were analyzed in five 2-year increments (2001-2002, 2003-2004, 2005-2006, 2007-2008, and 2009-2010). Annual discharge and population estimates were combined in 2-year intervals to increase the reliability of rate estimates. This report provides estimates of the number and rates of hospitalization discharges attributed to PH, defined as having any one of up to seven listed medical diagnoses with an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code of 416.0, 416.8, or 416.9. In 2010, a change occurred in the ICD-9-CM codes for PH, with the addition of a code for chronic pulmonary embolism (416.2). Estimates of the number and rate of hospitalizations associated with PH were calculated by using weights (ie, inflation factors) that allowed national estimation from the sample. More details about the design of NHDS have been published elsewhere.15 Because people with multiple discharges during the year may be sampled more than once, estimates are for hospital discharges, not persons. Data for newborn infants, defined as patients admitted to a hospital by birth, were excluded from this report.

More details about the design of NHDS have been published elsewhere.15 Because people with multiple discharges during the year may be sampled more than once, estimates are for hospital discharges, not persons. Data for newborn infants, defined as patients admitted to a hospital by birth, were excluded from this report. Intercensal estimates of the US noninstitutionalized civilian population from 2001 to 2002 to 2009 to 2010 were provided by the US Bureau of the Census and were used to calculate age- and sex-specific hospitalization rates per 100,000 people. Age-specific hospitalization rates were calculated for those aged < 35 years, 35 to 44 years, 45 to 54 years, 55 to 64 years, 65 to 74 years, 75 to 84 years, and ≥ 85 years. Hospitalization rates were also directly age standardized to the 2000 US standard population using seven age groups.16 Race-specific estimates are not provided because of incomplete reporting of race on hospital records. In addition, the distribution of selected disease categories as the principal (ie, first listed) diagnosis was examined for each period among hospitalizations with any listed PH, and the average lengths of stay for PH-related hospitalizations were compared by age group. Average length of stay was estimated as the ratio of total days of care per discharge to the total number of discharges.

ncipal (ie, first listed) diagnosis was examined for each period among hospitalizations with any listed PH, and the average lengths of stay for PH-related hospitalizations were compared by age group. Average length of stay was estimated as the ratio of total days of care per discharge to the total number of discharges. A weighted least-squares regression method17 was used to test for linear trends in hospitalization rates. A P value of < .05 indicated statistical significance. Differences among subgroups were evaluated using unrounded numbers with two-tailed t tests. Data analyses were performed using the statistical packages SAS, versions 9.2 and 9.3 (SAS Institute Inc), and SUDAAN, versions 10.0 and 11.0 (Research Triangle Institute).

A P value of < .05 indicated statistical significance. Differences among subgroups were evaluated using unrounded numbers with two-tailed t tests. Data analyses were performed using the statistical packages SAS, versions 9.2 and 9.3 (SAS Institute Inc), and SUDAAN, versions 10.0 and 11.0 (Research Triangle Institute). Results Mortality The death rate for PH as any contributing cause of death was 5.5 per 100,000 in 2001 and 6.5 per 100,000 in 2010 (Tables 3, 4). PH death rates for men and women were the same in 2001 (5.6 per 100,000). Although rates have increased for both men and women, the AAPC in death rates from 2001 to 2010 shows that rates rose more sharply for women (AAPC, 2.5; P < .05) than for men (AAPC, 0.9; P < .05) (Fig 2, Table 5). Although there was a 1.3% annual decrease in PH death rates among men from 2001 to 2006, this was followed by a significant increase (APC, 3.7; P < .05) from 2006 to 2010. PH death rates among women show significant increases during both time segments (APC, 1.7 in 2001-2006 and 3.5 in 2006-2010). Overall, trends in PH deaths increased significantly only from 2006 to 2010, with an APC of 3.6 (P < .05) and a total AAPC from 2001 to 2010 of 1.9 (95% CI, 1.5-2.4) (Fig 2, Table 5). TABLE 3 ] ASDRs (per 100,000) for Pulmonary Hypertension as Any Cause of Death Among Americans of All Ages by Sex and Race/Ethnicity: United States, 2001-2005

Results Mortality The death rate for PH as any contributing cause of death was 5.5 per 100,000 in 2001 and 6.5 per 100,000 in 2010 (Tables 3, 4). PH death rates for men and women were the same in 2001 (5.6 per 100,000). Although rates have increased for both men and women, the AAPC in death rates from 2001 to 2010 shows that rates rose more sharply for women (AAPC, 2.5; P < .05) than for men (AAPC, 0.9; P < .05) (Fig 2, Table 5). Although there was a 1.3% annual decrease in PH death rates among men from 2001 to 2006, this was followed by a significant increase (APC, 3.7; P < .05) from 2006 to 2010. PH death rates among women show significant increases during both time segments (APC, 1.7 in 2001-2006 and 3.5 in 2006-2010). Overall, trends in PH deaths increased significantly only from 2006 to 2010, with an APC of 3.6 (P < .05) and a total AAPC from 2001 to 2010 of 1.9 (95% CI, 1.5-2.4) (Fig 2, Table 5). TABLE 3 ] ASDRs (per 100,000) for Pulmonary Hypertension as Any Cause of Death Among Americans of All Ages by Sex and Race/Ethnicity: United States, 2001-2005 Sex and Race/Ethnicity 2001 2002 2003 2004 2005 Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Total 15,596 5.5 (5.4-5.6) 15,667 5.5 (5.4-5.6) 15,910 5.5 (5.4-5.6) 16,385 5.6 (5.5-5.6) 16,880 5.6 (5.5-5.7) Sex Male 6,513 5.6 (5.4-5.7) 6,483 5.5 (5.3-5.6) 6,357 5.3 (5.2-5.4) 6,524 5.3 (5.2-5.4) 6,649 5.3 (5.2-5.5) Female 9,083 5.6 (5.5-5.7) 9,184 5.6 (5.4-5.7) 9,553 5.7 (5.6-5.8) 9,861 5.8 (5.7-5.9) 10,231 5.9 (5.8-6.0) Race NH white 12,648 5.5 (5.4-5.6) 12,730 5.5 (5.4-5.6) 12,876 5.5 (5.4-5.5) 13,118 5.5 (5.4-5.6) 13,503 5.6 (5.5-5.7) NH black 2,022 7.6 (7.2-7.9) 2,055 7.5 (7.2-7.9) 2,064 7.6 (7.2-7.9) 2,205 7.8 (7.5-8.2) 2,304 8.2 (7.8-8.5) NH AI/AN 66 4.2 (3.1-5.2) 69 4.2 (3.1-5.3) 74 4.9 (3.7-6.1) 83 4.8 (3.7-6.0) 88 5.2 (4.0-6.3) NH API 215 2.6 (2.3-3.0) 196 2.2 (1.9-2.5) 245 2.7 (2.3-3.0) 244 2.7 (2.3-3.0) 246 2.5 (2.2-2.8) Hispanic 645 3.2 (2.9-3.5) 617 3.1 (2.8-3.4) 651 3.2 (2.9-3.5) 735 3.2 (3.0-3.5) 739 3.2 (2.9-3.4) AI/AN = American Indian/Alaska native; API = Asian/Pacific Islander; ASDR = age-standardized death rate; NH = non-Hispanic.

NH API 215 2.6 (2.3-3.0) 196 2.2 (1.9-2.5) 245 2.7 (2.3-3.0) 244 2.7 (2.3-3.0) 246 2.5 (2.2-2.8) Hispanic 645 3.2 (2.9-3.5) 617 3.1 (2.8-3.4) 651 3.2 (2.9-3.5) 735 3.2 (3.0-3.5) 739 3.2 (2.9-3.4) AI/AN = American Indian/Alaska native; API = Asian/Pacific Islander; ASDR = age-standardized death rate; NH = non-Hispanic. TABLE 4 ] ASDRs (per 100,000) for Pulmonary Hypertension as Any Cause of Death Among Americans of All Ages by Sex and Race/Ethnicity: United States, 2006-2010 Sex and Race/Ethnicity 2006 2007 2008 2009 2010 Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Total 17,214 5.6 (5.5-5.7) 18,164 5.8 (5.7-5.9) 19,374 6.1 (6.0-6.2) 20,330 6.3 (6.2-6.4) 21,292 6.5 (6.4-6.5) Sex Male 6,570 5.2 (5.0-5.3) 7,047 5.4 (5.3-5.5) 7,400 5.6 (5.5-5.7) 7,792 5.8 (5.6-5.9) 8,261 6.0 (5.9-6.1) Female 10,644 6.0 (5.9-6.1) 11,117 6.2 (6.0-6.3) 11,974 6.5 (6.4-6.6) 12,538 6.7 (6.6-6.8) 13,031 6.8 (6.7-6.9) Race NH white 13,771 5.6 (5.5-5.7) 14,516 5.8 (5.7-5.9) 15,485 6.1 (6.0-6.2) 16,116 6.2 (6.2-6.3) 16,938 6.5 (6.4-6.6) NH black 2,345 8.0 (7.7-8.4) 2,488 8.4 (8.1-8.8) 2,595 8.6 (8.3-9.0) 2,812 9.0 (8.7-9.4) 2,862 9.1 (8.7-9.4) NH AI/AN 88 5.3 (4.1-6.5) 82 4.7 (3.6-5.8) 87 4.8 (3.7-5.9) 102 5.6 (4.5-6.8) 121 6.7 (5.4-8.0) NH API 256 2.6 (2.2-2.9) 301 2.8 (2.5-3.2) 342 3.1 (2.7-3.4) 374 3.2 (2.9-3.6) 399 3.3 (3.0-3.7) Hispanic 754 3.3 (3.0-3.5) 777 3.1 (2.9-3.4) 865 3.4 (3.1-3.6) 926 3.5 (3.2-3.7) 972 3.6 (3.4-3.8) See Table 3 legend for expansion of abbreviations.

.5) 82 4.7 (3.6-5.8) 87 4.8 (3.7-5.9) 102 5.6 (4.5-6.8) 121 6.7 (5.4-8.0) NH API 256 2.6 (2.2-2.9) 301 2.8 (2.5-3.2) 342 3.1 (2.7-3.4) 374 3.2 (2.9-3.6) 399 3.3 (3.0-3.7) Hispanic 754 3.3 (3.0-3.5) 777 3.1 (2.9-3.4) 865 3.4 (3.1-3.6) 926 3.5 (3.2-3.7) 972 3.6 (3.4-3.8) See Table 3 legend for expansion of abbreviations. Figure 2 – Age-standardized death rates for pulmonary hypertension as a contributing cause of death and trend lines among individuals of all ages, by sex: United States, 2001-2010. The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having International Classification of Diseases, 10th Revision (ICD-10) codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). All trend lines are compared with the referent group, men; P < .05. Both parallelism and coincident comparison of trend lines were rejected. Rates are per 100,000 population and are age standardized to the 2000 US standard population (eight age groups). See Table 5 for additional data. TABLE 5 ] Change in ASDRs (per 100,000) for Pulmonary Hypertension as Contributing Cause of Death Among Individuals of All Ages, by Sex: United States, 2001-2010

Figure 2 – Age-standardized death rates for pulmonary hypertension as a contributing cause of death and trend lines among individuals of all ages, by sex: United States, 2001-2010. The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having International Classification of Diseases, 10th Revision (ICD-10) codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). All trend lines are compared with the referent group, men; P < .05. Both parallelism and coincident comparison of trend lines were rejected. Rates are per 100,000 population and are age standardized to the 2000 US standard population (eight age groups). See Table 5 for additional data. TABLE 5 ] Change in ASDRs (per 100,000) for Pulmonary Hypertension as Contributing Cause of Death Among Individuals of All Ages, by Sex: United States, 2001-2010 Joinpoint Segment Sex APC AAPC (95% CI) (2001-2010) 2001-2006 All 0.6 1.9b (1.5-2.4) 2006-2010 All 3.6a … 2001-2006 Men −1.3a 0.9b (0.3-1.5) 2006-2010 Men 3.7a … 2001-2006 Women 1.7a 2.5b (1.9-3.1) 2006-2010 Women 3.5a … The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). AAPCs are provided for the time period 2001 to 2010; the joinpoint varies and is based on the joinpoint regression analysis of best fit with the smallest number of deaths. Rates are per 100,000 population and are age standardized to the 2000 US standard population (eight age groups). AAPC = average annual percent change; APC = annual percent change. See Table 2 and 3 legends for expansion of other abbreviations.

oinpoint regression analysis of best fit with the smallest number of deaths. Rates are per 100,000 population and are age standardized to the 2000 US standard population (eight age groups). AAPC = average annual percent change; APC = annual percent change. See Table 2 and 3 legends for expansion of other abbreviations. a APC is significantly different from zero at α = 0.05. b AAPC is significantly different from zero at α = 0.05.

oinpoint regression analysis of best fit with the smallest number of deaths. Rates are per 100,000 population and are age standardized to the 2000 US standard population (eight age groups). AAPC = average annual percent change; APC = annual percent change. See Table 2 and 3 legends for expansion of other abbreviations. a APC is significantly different from zero at α = 0.05. b AAPC is significantly different from zero at α = 0.05. From 2001 to 2010, PH death rates were consistently higher for NH blacks than for NH whites (2010 rate, 9.1 per 100,000 for NH blacks vs 6.5 per 100,000 for NH whites) (Tables 3, 4). When compared with those for NH whites, PH death rates for NH AI/AN, NH Asian/Pacific Islanders, and Hispanics were consistently either lower or not statistically significantly different. However, it should be noted that because of concerns regarding the accuracy of coding race/ethnicity on death certificates, the comparisons of NH whites with race/ethnic groups other than NH blacks are provided with caution on interpretation.18 PH death rates increased over time for all races/ethnicities (Fig 3, Table 6). For NH whites, PH deaths did not increase during the first time segments from 2001 to 2006 (APC, 0.5) but increased significantly from 2006 to 2010 (APC, 3.8; P < .05). NH blacks experienced a significant increase in death rates during the entire time period from 2001 to 2010 (APC, 2.3; P < .05), as did NH AI/AN and NH Asian/Pacific Islanders (APC, 4.0 and 3.7, respectively; P < .05). Although there was no significant change in PH death rates among Hispanics from 2001 to 2007 (APC, 0.2), there was a significant increase from 2007 to 2010 (APC, 4.0; P < .05). The AAPC in the PH death rate for 2001 to 2010 shows significant increases for all races/ethnicities. Compared with NH whites, all race/ethnicity trend lines were parallel but not coincident (P < .05).

among Hispanics from 2001 to 2007 (APC, 0.2), there was a significant increase from 2007 to 2010 (APC, 4.0; P < .05). The AAPC in the PH death rate for 2001 to 2010 shows significant increases for all races/ethnicities. Compared with NH whites, all race/ethnicity trend lines were parallel but not coincident (P < .05). Figure 3 – Age-standardized death rates for pulmonary hypertension as a contributing cause of death and trend lines among individuals of all ages, by race/ethnicity: United States, 2001-2010. The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). aRates are per 100,000 population and are age standardized to the 2000 US standard population (eight age groups); ball race/ethnicity trend lines are parallel to the referent group of NHW, P < .05. There are no statistically significant coincident trend lines to the referent group of NHW. See Table 6 for additional data. NHAIAN = non-Hispanic American Indian/Alaska native; NHAPI = non-Hispanic Asian/Pacific Islander; NHB = non-Hispanic black; NHW = non-Hispanic white. See Figure 2 legend for expansion of other abbreviation. TABLE 6 ] Change in ASDRs for Pulmonary Hypertension as Contributing Cause of Death Among Individuals of All Ages, by Race/Ethnicity: United States, 2001-2010

Figure 3 – Age-standardized death rates for pulmonary hypertension as a contributing cause of death and trend lines among individuals of all ages, by race/ethnicity: United States, 2001-2010. The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). aRates are per 100,000 population and are age standardized to the 2000 US standard population (eight age groups); ball race/ethnicity trend lines are parallel to the referent group of NHW, P < .05. There are no statistically significant coincident trend lines to the referent group of NHW. See Table 6 for additional data. NHAIAN = non-Hispanic American Indian/Alaska native; NHAPI = non-Hispanic Asian/Pacific Islander; NHB = non-Hispanic black; NHW = non-Hispanic white. See Figure 2 legend for expansion of other abbreviation. TABLE 6 ] Change in ASDRs for Pulmonary Hypertension as Contributing Cause of Death Among Individuals of All Ages, by Race/Ethnicity: United States, 2001-2010 Joinpoint Segment Race/Ethnicity APC AAPC (95% CI) (2001-2010) 2001-2006 NH white 0.5 1.9a (1.5-2.4) 2006-2010 NH white 3.8b 2001-2010 NH black 2.3b 2.3a (1.9-2.8) 2001-2010 NH AI or AN 4.0b 4.0a (1.6-6.4) 2001-2010 NH API 3.7b 3.7a (1.9-5.5) 2001-2007 Hispanic 0.2 1.5a (0.8-2.2) 2007-2010 Hispanic 4.0b The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). See Table 2, 3, and 5 legends for expansion of abbreviations.

n deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). See Table 2, 3, and 5 legends for expansion of abbreviations. a AAPC is significantly different from zero at α = 0.05. b APC is significantly different from zero at α = 0.05. Death rates for PH were highest in the oldest age group (85 years and older) for every year (Tables 7, 8) and increased by > 65% between 2001 and 2010 (65.6 per 100,000 to 108.7 per 100,000). PH death rates in the 0- to 12-month age group were higher in 2001 (14.9 per 100,000) than in 2010 (8.2 per 100,000). The PH death rate trend analysis by age group for 2001 to 2010 shows differences in the direction of the slope segments (Fig 4, Table 9). Neonates through 12 months had the only statistically significant decrease in rates (APC, −6.9; P < .05). The greatest increases in PH death rates were among the oldest age group (85 years and older) from 2001 to 2005 (APC, 3.9; P < .05) and from 2005 to 2010 (APC, 7.2; P < .05). The AAPC for 2001 to 2010 shows significant increases in PH death rates for those aged 75 to 84 years and for those aged 85 years and older. Both parallelism and coincident comparisons of trend lines were rejected. TABLE 7 ] Deaths and Age-Specific Death Rates (per 100,000) of Pulmonary Hypertension as Any Cause of Death Among Americans by Age: United States, 2001-2005

Death rates for PH were highest in the oldest age group (85 years and older) for every year (Tables 7, 8) and increased by > 65% between 2001 and 2010 (65.6 per 100,000 to 108.7 per 100,000). PH death rates in the 0- to 12-month age group were higher in 2001 (14.9 per 100,000) than in 2010 (8.2 per 100,000). The PH death rate trend analysis by age group for 2001 to 2010 shows differences in the direction of the slope segments (Fig 4, Table 9). Neonates through 12 months had the only statistically significant decrease in rates (APC, −6.9; P < .05). The greatest increases in PH death rates were among the oldest age group (85 years and older) from 2001 to 2005 (APC, 3.9; P < .05) and from 2005 to 2010 (APC, 7.2; P < .05). The AAPC for 2001 to 2010 shows significant increases in PH death rates for those aged 75 to 84 years and for those aged 85 years and older. Both parallelism and coincident comparisons of trend lines were rejected. TABLE 7 ] Deaths and Age-Specific Death Rates (per 100,000) of Pulmonary Hypertension as Any Cause of Death Among Americans by Age: United States, 2001-2005 Age Groups 2001 2002 2003 2004 2005 Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Total 15,596 5.5 (5.4-5.6) 15,667 5.5 (5.4-5.6) 15,910 5.5 (5.4-5.6) 16,385 5.6 (5.5-5.6) 16,880 5.6 (5.5-5.7) 0-12 mo 598 14.9 (13.7-16.1) 596 15.1 (13.9-16.3) 349 8.8 (7.9-9.7) 566 14.1 (12.9-15.3) 385 9.6 (8.7-10.6) 1-34 y 457 0.3 (0.3-0.4) 430 0.3 (0.3-0.3) 414 0.3 (0.3-0.3) 412 0.3 (0.3-0.3) 408 0.3 (0.3-0.3) 35-44 y 580 1.3 (1.2-1.4) 529 1.2 (1.1-1.3) 561 1.3 (1.2-1.4) 528 1.2 (1.1-1.3) 514 1.2 (1.1-1.3) 45-54 y 1,081 2.7 (2.6-2.9) 1,122 2.8 (2.6-3.0) 1,055 2.6 (2.4-2.7) 1,184 2.8 (2.7-3.0) 1,166 2.7 (2.6-2.9) 55-64 y 1,793 7.1 (6.8-7.5) 1,848 6.9 (6.6-7.2) 1,950 7.0 (6.7-7.3) 2,039 7.0 (6.7-7.3) 2,120 6.9 (6.6-7.2) 65-74 y 3,503 19.1 (18.4-19.7) 3,343 18.2 (17.6-18.8) 3,426 18.5 (17.5-19.1) 3,430 18.4 (17.8-19) 3,466 18.4 (17.7-19.0) 75-84 y 4,755 37.8 (36.7-38.8) 4,830 37.8 (36.8-38.9) 5,044 39.1 (38.0-40.2) 4,921 37.9 (36.8-38.9) 5,207 39.8 (38.7-40.9) ≥ 85 y 2,829 65.6 (63.2-68.0) 2,969 68.0 (65.5-70.4) 3,111 69.7 (67.2-72.1) 3,305 72.7 (70.2-75.2) 3,614 77.0 (74.5-79.5) See Table 3 legend for expansion of abbreviation.

9) 3,466 18.4 (17.7-19.0) 75-84 y 4,755 37.8 (36.7-38.8) 4,830 37.8 (36.8-38.9) 5,044 39.1 (38.0-40.2) 4,921 37.9 (36.8-38.9) 5,207 39.8 (38.7-40.9) ≥ 85 y 2,829 65.6 (63.2-68.0) 2,969 68.0 (65.5-70.4) 3,111 69.7 (67.2-72.1) 3,305 72.7 (70.2-75.2) 3,614 77.0 (74.5-79.5) See Table 3 legend for expansion of abbreviation. TABLE 8 ] Deaths and Age-Specific Death Rates (per 100,000) of Pulmonary Hypertension as Any Cause of Death Among Americans by Age: United States, 2006-2010

9) 3,466 18.4 (17.7-19.0) 75-84 y 4,755 37.8 (36.7-38.8) 4,830 37.8 (36.8-38.9) 5,044 39.1 (38.0-40.2) 4,921 37.9 (36.8-38.9) 5,207 39.8 (38.7-40.9) ≥ 85 y 2,829 65.6 (63.2-68.0) 2,969 68.0 (65.5-70.4) 3,111 69.7 (67.2-72.1) 3,305 72.7 (70.2-75.2) 3,614 77.0 (74.5-79.5) See Table 3 legend for expansion of abbreviation. TABLE 8 ] Deaths and Age-Specific Death Rates (per 100,000) of Pulmonary Hypertension as Any Cause of Death Among Americans by Age: United States, 2006-2010 Age Groups 2006 2007 2008 2009 2010 Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Deaths ASDR (95% CI) Total 17,214 5.6 (5.5-5.7) 18,164 5.8 (5.7-5.9) 19,373 6.1 (6.0-6.2) 20,330 6.3 (6.2-6.4) 21,292 6.5 (6.4-6.5) 0-12 mo 316 7.8 (7.0-8.7) 332 8.0 (7.1-8.9) 376 9.1 (8.2-10.0) 367 9.2 (8.2-10.1) 325 8.2 (7.3-9.1) 1-34 y 415 0.3 (0.3-0.3) 429 0.3 (0.3-0.3) 422 0.3 (0.3-0.3) 434 0.3 (0.3-0.3) 386 0.3 (0.2-0.3) 35-44 y 488 1.1 (1.0-1.2) 447 1.0 (0.9-1.1) 509 1.2 (1.1-1.3) 491 1.2 (1.1-1.3) 465 1.1 (1.0-1.2) 45-54 y 1,210 2.8 (2.6-3.0) 1,212 2.8 (2.6-2.9) 1,233 2.8 (2.6-2.9) 1,334 3.0 (2.8-3.1) 1,238 2.8 (2.6-2.9) 55-64 y 2,179 6.8 (6.5-7.1) 2,331 7.0 (6.8-7.3) 2,325 6.8 (6.5-7.1) 2,504 7.1 (6.8-7.3) 2,516 6.9 (6.6-7.2) 65-74 y 3,401 17.7 (17.1-18.3) 3,538 18.0 (17.4-18.6) 3,717 18.1 (17.5-18.7) 3,938 18.5 (18.0-19.1) 4,133 19 (18.5-19.6) 75-84 y 5,279 40.3 (39.2-41.4) 5,482 41.9 (40.8-43.0) 5,873 44.9 (43.8-46.1) 5,911 45.4 (44.2-46.5) 6,256 47.9 (46.7-49.1) ≥ 85 y 3,926 80.7 (78.2-83.2) 4,393 87.2 (84.6-89.7) 4,918 94.7 (92.0-97.3) 5,351 99.7 (97.0-102.4) 5,973 108.7 (106.0-111.5) See Table 3 legend for expansion of abbreviation.

4,133 19 (18.5-19.6) 75-84 y 5,279 40.3 (39.2-41.4) 5,482 41.9 (40.8-43.0) 5,873 44.9 (43.8-46.1) 5,911 45.4 (44.2-46.5) 6,256 47.9 (46.7-49.1) ≥ 85 y 3,926 80.7 (78.2-83.2) 4,393 87.2 (84.6-89.7) 4,918 94.7 (92.0-97.3) 5,351 99.7 (97.0-102.4) 5,973 108.7 (106.0-111.5) See Table 3 legend for expansion of abbreviation. Figure 4 – Age-specific death rates for pulmonary hypertension as a contributing cause of death and trend lines among individuals of all ages, by six age groups: United States, 2001-2010. The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). aRates are per 100,000 population and age-specific death rates were calculated for each of six age groups; ball age-specific trend lines are compared with the referent age group of 0 to 12 mo, P < .05. Both parallelism and coincident comparison of trend lines were rejected. See Table 9 for additional data. See Figure 2 legend for expansion of abbreviation. TABLE 9 ] Death and Age-Specific Death Rates for Pulmonary Hypertension as Contributing Cause of Death Among Individuals of All Ages, by Six Age Groups: United States, 2001-2010

Figure 4 – Age-specific death rates for pulmonary hypertension as a contributing cause of death and trend lines among individuals of all ages, by six age groups: United States, 2001-2010. The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). aRates are per 100,000 population and age-specific death rates were calculated for each of six age groups; ball age-specific trend lines are compared with the referent age group of 0 to 12 mo, P < .05. Both parallelism and coincident comparison of trend lines were rejected. See Table 9 for additional data. See Figure 2 legend for expansion of abbreviation. TABLE 9 ] Death and Age-Specific Death Rates for Pulmonary Hypertension as Contributing Cause of Death Among Individuals of All Ages, by Six Age Groups: United States, 2001-2010 Joinpoint Segment Age Group, y APC 2001-2010 AAPC (95% CI) 2001-2010 0-12a −6.9b −6.9c (−11.1 to −2.5) 2001-2010 1-54 0.1 0.1 (−0.6 to 0.7) 2001-2010 55-64 −0.2 −0.2 (−0.6 to 0.2) 2001-2007 65-74 −0.9 0.1 (−0.7 to 0.9) 2007-2010 65-74 2.0 … 2001-2004 75-84 0.1 2.5c (1.6 to 3.4) 2004-2010 75-84 3.7b … 2001-2005 85+ 3.9b 5.7c (5.3 to 6.2) 2005-2010 85+ 7.2b … The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). Rates are per 100,000 population and age-specific death rates were calculated for each of six age groups. See Table 2 and 5 legends for expansion of abbreviations.

I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). Rates are per 100,000 population and age-specific death rates were calculated for each of six age groups. See Table 2 and 5 legends for expansion of abbreviations. a Mo. b APC is significantly different from zero at α = 0.05. c AAPC is significantly different from zero at α = 0.05. In 2001, the northern mountain states plus the District of Columbia, Vermont, New Hampshire, Delaware, North Carolina, and Ohio experienced the highest PH age-standardized death rates (Fig 5). By 2010, several other western and midwestern states, including Iowa, Minnesota, Nebraska, Oregon, and Washington, had experienced significant increases in death rates (P < .05) and had moved into the highest rate category. South Carolina and Tennessee also experienced significant increases in death rates between 2001 and 2010 (P < .05).

her western and midwestern states, including Iowa, Minnesota, Nebraska, Oregon, and Washington, had experienced significant increases in death rates (P < .05) and had moved into the highest rate category. South Carolina and Tennessee also experienced significant increases in death rates between 2001 and 2010 (P < .05). Figure 5 – Age-standardized death rates for pulmonary hypertension as any cause of death among all ages by state. A, Death rates in 2001. B, Death rates in 2010. The National Vital Statistics System was used to ascertain deaths due to pulmonary hypertension, which were considered those with decedents having ICD-10 codes I27.0, I27.2, I27.8, or I27.9 reported as any contributing cause of death (ie, any of the possible 20 conditions, including the underlying cause of death). Rates are per 100,000 population and are age standardized to the 2000 US standard population (eight age groups). See Figure 2 legend for expansion of abbreviation. In 2001, the most common UCOD among decedents with PH listed as any contributing cause of death was PH (29.3%), followed by chronic lower respiratory disease (26.2%), coronary heart disease (8.8%), and interstitial lung disease (4.5%) (Table 10). Within the category of chronic lower respiratory disease, emphysema was indicated in 4.2% of deaths (Table 10). The most commonly reported underlying causes of death for all race/sex groups were PH and chronic lower respiratory disease. TABLE 10 ] Distribution of Underlying Cause of Death Among Decedents With Reported Pulmonary Hypertension by Race/Sex Groups, 2001

In 2001, the most common UCOD among decedents with PH listed as any contributing cause of death was PH (29.3%), followed by chronic lower respiratory disease (26.2%), coronary heart disease (8.8%), and interstitial lung disease (4.5%) (Table 10). Within the category of chronic lower respiratory disease, emphysema was indicated in 4.2% of deaths (Table 10). The most commonly reported underlying causes of death for all race/sex groups were PH and chronic lower respiratory disease. TABLE 10 ] Distribution of Underlying Cause of Death Among Decedents With Reported Pulmonary Hypertension by Race/Sex Groups, 2001 Disease Category: ICD-10 Codes Total NH White Men NH Black Men Hispanic Men NH White Women NH Black Women Hispanic Women Diseases of the circulatory system (49.6%) Pulmonary hypertension: I27.0, I27.2, I27.8, I27.9 29.3 (4,573) 24.6 (1,318) 33.5 (254) 31.4 (86) 30.4 (2,219) 37.5 (474) 35.0 (130) Coronary heart disease: I20-I25 8.8 (1,380) 10.6 (570) 6.7 (51) 8.4 (23) 8.4 (616) 5.6 (71) 6.7 (25) Valvular heart disease, nonrheumatic: I34-I38 2.6 (405) 1.7 (92) 2.1 (16) 2.2 (6) 3.4 (247) 1.9 (24) 3.5 (13) Aortic stenosis: I06.0, I35.0, I35.2 0.9 (147) 0.8 (43) a a 1.1 (80) 0.6 (8) 2.4 (9) Hypertension: I10-I15 1.0 (156) 0.8 (41) 1.7 (13) a 0.9 (68) 2.1 (27) a Rheumatic heart disease: I00-I09 1.4 (219) 0.5 (28) a a 2.1 (155) 1.5 (19) 1.3 (5) Pulmonary embolism: I26 1.3 (200) 0.9 (49) 2.0 (15) a 1.5 (112) 1.1 (14) 1.3 (5) Heart failure: I50 0.6 (99) 0.8 (41) 0.7 (5) a 0.6 (44) 0.6 (7) a Other cardiovascular/cerebrovascular disease: I27.1, I28-I33, I40-I49, I51-I78 3.6 (560) 3.4 (181) 5.3 (40) 3.3 (9) 3.5 (256) 3.5 (44) 5.7 (21) Respiratory disorder or infection (37.5%) Chronic lower respiratory disease: J40-J47 26.2 (4,080) 33.3 (1,782) 17.1 (130) 14.6 (40) 26.0 (1,896) 12.9 (163) 8.6 (32) Emphysema only: J43 4.2 (654) 6.0 (322) 2.4 (18) 1.8 (5) 3.8 (276) 1.7 (21) a Interstitial lung disease: J84 4.5 (701) 4.6 (248) 3.7 (28) 5.5 (15) 4.3 (311) 5.6 (71) 4.9 (18) Influenza and pneumonia: J09-J18 0.9 (143) 0.9 (49) a a 0.9 (69) 0.8 (10) a All other respiratory diseases: J00-06, J20-J39, J60-J70, J85-J98 1.7 (266) 1.9 (101) 2.9 (22) 3.3 (9) 1.4 (101) 1.7 (21) 1.3 (5) Autoimmune diseases: M05-M06, M08, M30-35 1.3 (209) 0.3 (16) 0.8 (6) a 1.6 (119) 3.2 (40) 5.1 (19) Congenital malformations, deformation, and chromosomal abnormalities: Q00-Q99 3.0 (473) 2.5 (133) 4.7 (36) 10.6 (29) 2.6 (190) 3.3 (42) 5.4 (20) Malignant neoplasms of the trachea, bronchus, and lung: C34 1.8 (278) 2.5 (132) 1.8 (14) a 1.6 (116) 0.6 (7) a Diabetes mellitus: E10-E14 1.0 (160) 0.9 (47) 1.1 (8) a 1.0 (74) 1.1 (14) 2.7 (10) Nephritis, nephrotic syndrome, and nephrosis: N00-N07, N17-N19, N25-N27 0.7 (112) 0.6 (33) 1.3 (10) a 0.7 (52) 1.0 (12) a Chronic liver disease and cirrhosis: K70, K73-74 0.3 (5

ung: C34 1.8 (278) 2.5 (132) 1.8 (14) a 1.6 (116) 0.6 (7) a Diabetes mellitus: E10-E14 1.0 (160) 0.9 (47) 1.1 (8) a 1.0 (74) 1.1 (14) 2.7 (10) Nephritis, nephrotic syndrome, and nephrosis: N00-N07, N17-N19, N25-N27 0.7 (112) 0.6 (33) 1.3 (10) a 0.7 (52) 1.0 (12) a Chronic liver disease and cirrhosis: K70, K73-74 0.3 (5 0) 0.4 (20) a a 0.3 (19) a 1.6 (6) Malignant neoplasms of lymphoid, hematopoietic, and related tissue: C81-C96 0.4 (70) 0.4 (23) a a 0.4 (30) 0.6 (7) a Sleep apnea: G47.3 0.1 (17) 0.2 (10) a a a a a Certain conditions originating in the perinatal period: P00-P96 1.0 (160) 0.9 (47) 3.4 (26) 6.9 (19) 0.5 (37) 1.6 (20) 1.9 (7) All other causes 7.3 (1,139) 6.6 (353) 8.8 (67) 5.5 (15) 6.6 (479) 12.9 (163) 8.6 (32) Total 15,597 5,357 759 274 7,292 1,263 371 Data are presented as % (No.). See Table 2 and 3 legends for expansion of abbreviations. a Death counts and percentages are suppressed if the number of deaths is fewer than five. In 2010, PH (at 30.2%) was the most common UCOD among decedents with PH listed as any cause of death, followed by chronic lower respiratory disease (19.0%), coronary heart disease (10.3%), and interstitial lung disease (4.9%) (Table 11). Interestingly, from 2001 to 2010, the percentage of PH deaths caused by emphysema dropped from 4.2% to 1.4% of all PH deaths, whereas PH deaths caused by nonrheumatic valvular heart disease rose from 2.6% in 2001 to 4.1% in in 2010. Autoimmune diseases listed as the UCOD were higher for women than for men regardless of race or Hispanic origin.

to 2010, the percentage of PH deaths caused by emphysema dropped from 4.2% to 1.4% of all PH deaths, whereas PH deaths caused by nonrheumatic valvular heart disease rose from 2.6% in 2001 to 4.1% in in 2010. Autoimmune diseases listed as the UCOD were higher for women than for men regardless of race or Hispanic origin. TABLE 11 ] Distribution of Underlying Cause of Death Among Decedents With Reported Pulmonary Hypertension by Race/Sex Groups, 2010

to 2010, the percentage of PH deaths caused by emphysema dropped from 4.2% to 1.4% of all PH deaths, whereas PH deaths caused by nonrheumatic valvular heart disease rose from 2.6% in 2001 to 4.1% in in 2010. Autoimmune diseases listed as the UCOD were higher for women than for men regardless of race or Hispanic origin. TABLE 11 ] Distribution of Underlying Cause of Death Among Decedents With Reported Pulmonary Hypertension by Race/Sex Groups, 2010 Disease Category: ICD-10 codes Total NH White Men NH Black Men Hispanic Men NH White Women NH Black Women Hispanic Women Diseases of the circulatory system (54.2%) Pulmonary hypertension: I27.0, I27.2, I27.8, I27.9 30.2 (6,436) 26.1 (1,741) 31.4 (322) 23.2 (82) 31.8 (3,273) 35.9 (659) 32.5 (201) Coronary heart disease: I20-I25 10.3 (2,184) 13.8 (921) 8.5 (87) 9.3 (33) 9.2 (945) 6.6 (121) 5.2 (32) Valvular heart disease, nonrheumatic: I34-I38 4.1 (879) 3.5 (235) 1.7 (17) 2.8 (10) 5.4 (551) 1.6 (29) 3.1 (19) Aortic stenosis: I06.0, I35.0, I35.2 1.6 (350) 1.4 (92) a a 2.2 (230) 0.5 (10) 0.8 (5) Hypertension: I10-I15 1.6 (336) 1.1 (70) 2.8 (29) 2.0 (7) 1.6 (168) 2.7 (50) 1.3 (8) Rheumatic heart disease: I00-I09 1.5 (332) 0.8 (54) a a 2.2 (225) 0.9 (16) 1.6 (10) Pulmonary embolism: I26 0.7 (145) 0.7 (45) 0.8 (8) a 0.7 (68) 1.0 (18) a Heart failure: I50 0.4 (95) 0.5 (31) a a 0.5 (48) 0.5 (9) a Other cardiovascular/cerebrovascular disease: I27.1, I28-I33, I40-I49, I51-I78 3.7 (784) 3.4 (226) 5.8 (60) 3.7 (13) 3.7 (381) 4.0 (74) 2.7 (17) Respiratory disorder or infection (28.4%) Chronic lower respiratory disease: J40-J47 19.0 (4,041) 23.2 (1,547) 15.4 (158) 12.2 (43) 18.9 (1939) 11.8 (216) 11.1 (69) Emphysema only: J43 1.4 (295) 1.8 (123) 1.1 (11) a 1.3 (138) 0.8 (15) a Interstitial lung disease: J84 4.9 (1,043) 6.2 (413) 4.6 (47) 11.0 (39) 3.9 (401) 4.0 (73) 6.1 (38) Influenza and pneumonia: J09-J18 1.2 (254) 0.9 (63) 1.7 (17) 1.7 (6) 1.2 (127) 0.9 (17) 2.1 (13) All other respiratory diseases: J00-06, J20-J39, J60-J70, J85-J98 1.9 (408) 2.4 (159) 3.6 (37) 3.4 (12) 1.3 (135) 2.5 (45) 1.8 (11) Autoimmune diseases: M05-M06, M08, M30-35 2.4 (514) 0.9 (61) 1.5 (15) a 2.8 (283) 4.9 (90) 6.9 (43) Congenital malformations, deformation, and chromosomal abnormalities: Q00-Q99 1.6 (334) 1.3 (89) 2.3 (24) 5.4 (19) 1.1 (114) 2.2 (41) 4.0 (25) Malignant neoplasms of the trachea, bronchus, and lung: C34 1.5 (320) 1.8 (121) 1.7 (17) a 1.4 (141) 0.8 (15) 1.6 (10) Diabetes mellitus: E10-E14 1.5 (312) 1.5 (97) 1.8 (18) 1.4 (5) 1.3 (132) 1.7 (32) 2.6 (16) Nephritis, nephrotic syndrome, and nephrosis: N00-N07, N17-N19, N25-N27 1.1 (238) 0.9 (63) 1.2 (12) 1.4 (5) 1.1 (117) 1.3 (23) 1.3 (8)

e trachea, bronchus, and lung: C34 1.5 (320) 1.8 (121) 1.7 (17) a 1.4 (141) 0.8 (15) 1.6 (10) Diabetes mellitus: E10-E14 1.5 (312) 1.5 (97) 1.8 (18) 1.4 (5) 1.3 (132) 1.7 (32) 2.6 (16) Nephritis, nephrotic syndrome, and nephrosis: N00-N07, N17-N19, N25-N27 1.1 (238) 0.9 (63) 1.2 (12) 1.4 (5) 1.1 (117) 1.3 (23) 1.3 (8) Chronic liver disease and cirrhosis: K70, K73-74 0.8 (175) 0.8 (56) 1.0 (10) 2.8 (10) 0.7 (69) 0.5 (9) 2.4 (15) Malignant neoplasms of lymphoid, hematopoietic, and related tissue: C81-C96 0.7 (150) 0.8 (50) 0.9 (9) a 0.7 (69) 0.7 (13) a Sleep apnea: G47.3 0.6 (134) 0.7 (48) 1.0 (10) a 0.5 (48) 1.0 (18) a Certain conditions originating in the perinatal period: P00-P96 0.1 (27) a 0.5 a a a a All other causes 8.5 (1,811) 7.1 (474) 11.4 (117) 13.6 (48) 7.9 (808) 13.8 (254) 10.0 (62) Total (21,292) (6,660) (1,026) (353) (1,027) (1,836) (619) Data are presented as % (No.). See Table 2 and 3 legends for expansion of abbreviations. a Death counts and percentages are suppressed if the number of deaths is fewer than five. The UCOD associated with PH showed several interesting trends (Table 12). The AAPC increased significantly for several UCOD conditions, including coronary heart disease, aortic stenosis, hypertension, autoimmune disease, influenza, and pneumonia; it showed no significant change for malignancies of the trachea and bronchus; and it decreased for heart failure, chronic lower respiratory disease (emphysema in particular), and congenital malformations. TABLE 12 ] Trends Over Time by Underlying Cause of Death Among Decedents With Reported PH

The UCOD associated with PH showed several interesting trends (Table 12). The AAPC increased significantly for several UCOD conditions, including coronary heart disease, aortic stenosis, hypertension, autoimmune disease, influenza, and pneumonia; it showed no significant change for malignancies of the trachea and bronchus; and it decreased for heart failure, chronic lower respiratory disease (emphysema in particular), and congenital malformations. TABLE 12 ] Trends Over Time by Underlying Cause of Death Among Decedents With Reported PH Disease Category: ICD-10 Codes 2001 2010 AAPC Rate/100,000 % of Total PH Deaths Rate/100,000 % of Total PH Deaths Diseases of the circulatory system Pulmonary hypertension: I27.0, I27.2, I27.8, I27.9 1.63 29.3 1.95 30.2 1.5 Coronary heart disease: I20-I25 0.48 8.8 0.66 10.3 3.8a Valvular heart disease, nonrheumatic: I34-I38 0.14 2.6 0.25 4.1 6.7a Aortic stenosis: I06.0, I35.0, I35.2 0.04 0.9 0.10 1.6 10.8a Hypertension: I10-I15 0.04 1.0 0.10 1.6 9.2a Rheumatic heart disease: I00-I09 0.07 1.4 0.10 1.5 3.2a Pulmonary embolism: I26 0.07 1.3 0.04 0.7 −7.2a Heart failure: I50 0.03 0.6 0.02 0.4 −3.6a Other cardiovascular/cerebrovascular disease: I27.1, I28-I33, I40-I49, I51-I78 0.21 3.6 0.23 3.7 2.0 Respiratory disorder or infection Chronic lower respiratory disease: J40-J47 1.46 26.2 1.24 19.0 −1.3a Emphysema only: J43 0.24 4.2 0.08 1.4 −8.7a Interstitial lung disease: J84 0.24 4.5 0.31 4.9 4.5a Influenza and pneumonia: J09-J18 0.04 0.9 0.07 1.2 5.5a All other respiratory diseases: J00-06, J20-J39, J60-J70, J85-J98 0.08 1.7 0.12 1.9 3.3a Autoimmune diseases: M5-M6, M8, M30-35 0.07 1.3 0.16 2.4 10.9a Congenital malformations, deformation, and chromosomal abnormalities: Q00-Q99 0.17 3.0 0.11 1.6 −5.2a Malignant neoplasms of the trachea, bronchus, and lung: C34 0.09 1.8 0.09 1.5 0.1 Diabetes mellitus: E10-E14 0.05 1.0 0.09 1.5 7.6a Nephritis, nephrotic syndrome, and nephrosis: N00-N07, N17-N19, N25-N27 0.03 0.7 0.07 1.1 9.2a Chronic liver disease and cirrhosis: K70, K73-74 0.02 0.3 0.05 0.8 17.0a Malignant neoplasms of lymphoid, hematopoietic, and related tissue: C81-C96 0.02 0.4 0.04 0.7 6.4a Sleep apnea: G47.3 … 0.1 0.04 0.6 NA Certain conditions originating in the perinatal period: P00-P96 0.05 1.0 0.01 0.1 −12.5a Total 5.5 … 6.5 … … NA = not applicable; PH = pulmonary hypertension. See Table 2 and 5 legends for expansion of other abbreviations.

opoietic, and related tissue: C81-C96 0.02 0.4 0.04 0.7 6.4a Sleep apnea: G47.3 … 0.1 0.04 0.6 NA Certain conditions originating in the perinatal period: P00-P96 0.05 1.0 0.01 0.1 −12.5a Total 5.5 … 6.5 … … NA = not applicable; PH = pulmonary hypertension. See Table 2 and 5 legends for expansion of other abbreviations. a P < .05. Hospitalizations From 2001/2002 to 2009/2010, the age-adjusted rate of hospitalizations associated with PH increased by 44%, from 91 per 100,000 discharges to 131 per 100,000 discharges (P < .001). The rate of hospitalization for women increased by 52%, from 97 per 100,000 to 147 per 100,000 (P < .001). The rate of hospitalization for men increased by 33%, from 83 per 100,000 to 110 per 100,000 (P < .01). Crude and age-adjusted hospitalization rates were both higher for women than for men at each time interval, except for 2007 to 2008 (P < .05) (Table 13). The proportion of total PH hospitalizations among women remained consistently higher than among men: Women accounted for 61% of all PH hospitalizations in 2001/2002 and 63% in 2009/2010 (Table 14). TABLE 13 ] Hospitalization Rates (per 100,000) With Any Listed Pulmonary Hypertension Diagnosis by Age and Sex: United States, 2001/2002 Through 2009/2010

Hospitalizations From 2001/2002 to 2009/2010, the age-adjusted rate of hospitalizations associated with PH increased by 44%, from 91 per 100,000 discharges to 131 per 100,000 discharges (P < .001). The rate of hospitalization for women increased by 52%, from 97 per 100,000 to 147 per 100,000 (P < .001). The rate of hospitalization for men increased by 33%, from 83 per 100,000 to 110 per 100,000 (P < .01). Crude and age-adjusted hospitalization rates were both higher for women than for men at each time interval, except for 2007 to 2008 (P < .05) (Table 13). The proportion of total PH hospitalizations among women remained consistently higher than among men: Women accounted for 61% of all PH hospitalizations in 2001/2002 and 63% in 2009/2010 (Table 14). TABLE 13 ] Hospitalization Rates (per 100,000) With Any Listed Pulmonary Hypertension Diagnosis by Age and Sex: United States, 2001/2002 Through 2009/2010 Sex and Age Groups 2001/2002 2003/2004 2005/2006 2007/2008 2009/2010 Age-standardizeda rate Total 91 (85-97) 87 (81-93) 94 (88-100) 112 (100-124) 131 (115-147) Female 97 (89-105) 93 (85-101) 100 (92-108) 115 (101-129) 147 (129-165) Male 83 (75-91) 80 (70-90) 85 (77-93) 108 (94-122) 110 (92-128) Age-specific rate Total 91 (82-101) 89 (78-99) 96 (85-108) 117 (95-140) 140 (106-173) < 35 y 7 (6-9) 10 (7-12) 11 (8-14) 10 (7-13) 17b 35-44 y 35 (25-46) 27 (21-33) 40 (31-49) 40 (25-54) 38 (24-52) 45-54 y 58 (48-69) 69 (52-87) 68 (55-82) 85 (63-107) 85 (61-110) 55-64 y 178 (151-206) 158 (130-187) 137 (111-163) 154 (114-194) 166 (122-210) 65-74 y 314 (271-357) 303 (259-348) 297 (246-348) 383 (301-465) 418 (308-529) 75-84 y 553 (478-629) 529 (444-614) 606 (521-691) 711 (552-869) 872 (658-1,086) ≥ 85 y 834 (647-1,020) 728 (559-897) 890 (746-1,035) 1,182 (896-1,468) 1,527 (1,158-1,896) Female 109 (96-122) 104 (92-117) 115 (100-129) 134 (108-161) 175 (133-216) < 35 y 8 (6-11) 9 (6-11) 14 (9-19) 11 (7-15) 19 (9-28) 35-44 y 38 (19-57) 34 (25-43) 41 (29-54) 42 (24-59) 42 (25-60) 45-54 y 62 (47-78) 73 (55-91) 72 (55-89) 92 (64-120) 108 (71-146) 55-64 y 203 (160-246) 160 (120-199) 141 (113-169) 142 (110-175) 173 (118-228) 65-74 y 334 (275-394) 309 (259-360) 329 (259-400) 405 (302-508) 458 (329-587) 75-84 y 545 (464-626) 589 (491-686) 620 (526-713) 710 (530-891) 986 (726-1,247) ≥ 85 y 931 (704-1,159) 787 (581-992) 967 (789-1,144) 1,223 (915-1,531) 1,697 (1,290-2,104) Male 73 (65-82) 72 (61-83) 77 (67-88) 100 (79-121) 104 (77-131) < 35 y 7 (4-9) 11 (8-14) 8 (5-12) 9 (5-12) 16b 35-44 y 33 (22-43) 20 (13-27) 38 (23-53) 38 (16-60) 33 (16-50) 45-54 y 54 (41-67) 65 (43-88) 64 (45-82) 77 (51-104) 62 (41-82) 55-64 y 151 (116-187) 157 (117-196) 134 (102-165) 167 (104-229) 159 (110-208) 65-74 y 290 (234-346) 296 (233-358) 259 (210-308) 357 (277-438) 373 (260-486) 75-84 y 566 (460-672) 440 (312-569) 586 (467-705) 711 (539-884) 713 (505-921) ≥ 85 y 606 (425-787) 595 (406-785) 725 (531-920) 1,095 (707-1,483) 1,173 (794-1,551) Data are presented as rate (95% CI). (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11)

486) 75-84 y 566 (460-672) 440 (312-569) 586 (467-705) 711 (539-884) 713 (505-921) ≥ 85 y 606 (425-787) 595 (406-785) 725 (531-920) 1,095 (707-1,483) 1,173 (794-1,551) Data are presented as rate (95% CI). (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11) a To the 2000 US standard population. b Estimate does not meet standards of reliability or precision. The National Center for Health Statistics considers an estimate to be reliable if it is based on at least 30 discharge records and it has a relative SE of ≤ 30% (ie, the SE is ≤ 30% of the estimate). TABLE 14 ] Average Annual Estimates of Hospitalizations per 2 Years for Pulmonary Hypertension by Age Group and Sex: United States, 2001/2002 Through 2009/2010

b Estimate does not meet standards of reliability or precision. The National Center for Health Statistics considers an estimate to be reliable if it is based on at least 30 discharge records and it has a relative SE of ≤ 30% (ie, the SE is ≤ 30% of the estimate). TABLE 14 ] Average Annual Estimates of Hospitalizations per 2 Years for Pulmonary Hypertension by Age Group and Sex: United States, 2001/2002 Through 2009/2010 Sex and Age Groups 2001/2002 2003/2004 2005/2006 2007/2008 2009/2010 Estimate (95% CI)a % Estimate (95% CI)a % Estimate (95% CI)a % Estimate (95% CI)a % Estimate (95% CI)a % Total 261 (234-288) … 258 (227-289) … 286 (252-320) … 354 (286-422) … 429 (326-531) … < 35 y 10 (8-13) 4 14 (11-17) 5 16 (11-20) 6 14 (10-18) 4 25b 6 35-44 y 16 (11-20) 6 12 (9-15) 5 17 (13-21) 6 17 (11-23) 5 15 (10-21) 4 45-54 y 23 (19-27) 9 29 (21-36) 11 29 (23-35) 10 37 (28-47) 11 38 (27-49) 9 55-64 y 46 (39-53) 18 45 (37-53) 17 43 (34-51) 15 51 (38-64) 14 59 (43-74) 14 65-74 y 58 (50-65) 22 56 (48-64) 22 56 (46-65) 20 76 (59-92) 21 88 (65-112) 21 75-84 y 70 (61-80) 27 68 (57-79) 26 79 (68-90) 28 93 (72-113) 26 115 87-143) 27 ≥ 85 y 38 (29-46) 14 35 (27-43) 13 46 (39-54) 16 66 (50-82) 19 88 (67-109) 20 Female 159 (140-177) 61 155 (137-173) 60 173 (152-194) 61 206 (165-246) 58 272 (207-337) 63 < 35 y 6 (4-7) 54 6 (4-8) 44 10 (6-13) 62 8 (5-10) 54 13 (7-20) 53 35-44 y 9 (4-13) 54 8 (6-9) 63 9 (6-12) 52 9 (5-13) 53 9 (5-12) 57 45-54 y 13 (9-16) 55 15 (11-19) 54 16 (12-19) 54 21 (14-27) 55 24 (16-33) 64 55-64 y 27 (22-33) 59 24 (18-29) 52 23 (18-27) 53 24 (19-30) 48 32 (22-42) 54 65-74 y 33 (27-39) 58 31 (26-36) 56 34 (26-41) 60 43 (32-54) 57 52 (37-67) 59 75-84 y 42 (35-48) 59 46 (38-53) 67 48 (41-55) 61 55 (41-69) 59 76 (56-96) 66 ≥ 85 y 29 (22-37) 78 26 (19-33) 75 34 (28-41) 74 46 (35-58) 70 66 (50-82) 75 Male 102 (90-114) 39 103 (88-119) 40 113 (97-128) 39 148 (117-179) 42 157 (116-198) 37 < 35 y 5 (3-6) 46 8 (5-10) 56 6 (4-9) 38 6 (4-9) 46 12b 47 35-44 y 7 (5-10) 46 4 (3-6) 37 8 (5-12) 48 8 (3-13) 47 7 (3-10) 43 45-54 y 10 (8-13) 45 13 (9-18) 46 13 (10-17) 46 17 (11-23) 45 13 (9-18) 36 55-64 y 19 (14-23) 41 21 (16-27) 48 20 (15-25) 47 27 (17-37) 52 27 (19-36) 46 65-74 y 24 (19-29) 42 25 (20-30) 44 22 (18-27) 40 32 (25-40) 43 36 (25-47) 41 75-84 y 29 (23-34) 41 23 (16-30) 33 31 (25-37) 39 38 (29-47) 41 39 (28-51) 34 ≥ 85 y 8 (6-11) 22 9 (6-12) 25 12 (9-15) 26 20 (13-27) 30 22 (15-29) 25 a Estimate and 95% CI data are presented in thousands. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11)

75-84 y 29 (23-34) 41 23 (16-30) 33 31 (25-37) 39 38 (29-47) 41 39 (28-51) 34 ≥ 85 y 8 (6-11) 22 9 (6-12) 25 12 (9-15) 26 20 (13-27) 30 22 (15-29) 25 a Estimate and 95% CI data are presented in thousands. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11) b Estimate does not meet standards of reliability or precision. The National Center for Health Statistics considers an estimate to be reliable if it is based on at least 30 discharge records and it has a relative SE of ≤ 30% (ie, the SE is ≤ 30% of the estimate).

75-84 y 29 (23-34) 41 23 (16-30) 33 31 (25-37) 39 38 (29-47) 41 39 (28-51) 34 ≥ 85 y 8 (6-11) 22 9 (6-12) 25 12 (9-15) 26 20 (13-27) 30 22 (15-29) 25 a Estimate and 95% CI data are presented in thousands. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11) b Estimate does not meet standards of reliability or precision. The National Center for Health Statistics considers an estimate to be reliable if it is based on at least 30 discharge records and it has a relative SE of ≤ 30% (ie, the SE is ≤ 30% of the estimate). Rates of PH hospitalization increased over the time period for those aged < 35 years, 45 to 54 years, 75 to 84 years, and ≥ 85 years (Table 13). There was no significant change in hospitalization rates for those aged 35 to 44 years, 55 to 64 years, or 75 to 84 years. The rate of PH hospitalization for those aged ≥ 85 years had the greatest increase, rising by 83%, from 834 per 100,000 to 1,527 per 100,000 (P < .001). The 75- to 84-year-old age group had the highest proportion of hospitalizations of all age groups throughout the period (27% in 2001-2002 and in 2009-2010) (Table 14). Among men, the only significant increase in the PH hospitalization rate from 2001 to 2002 to 2009 to 2010 was for those aged ≥ 85 years (606 in 2001 to 1,173 in 2010; P < .01) (Table 13). Among women, the rate of hospitalization increased for almost all age groups (< 35 years, 45-54 years, 75-84 years, and ≥ 85 years) (Table 13). The hospitalization rate for women < 35 years old was the lowest among all women and had the greatest percentage increase over time. For both men and women, rates of hospitalization increased as age increased. Hospitalizations of those ≥ 85 years old were predominately among women from 2001 to 2002 through 2009 to 2010; in fact, in 2009 to 2010, 75% of hospitalizations in this age group were of women (Table 14).

eatest percentage increase over time. For both men and women, rates of hospitalization increased as age increased. Hospitalizations of those ≥ 85 years old were predominately among women from 2001 to 2002 through 2009 to 2010; in fact, in 2009 to 2010, 75% of hospitalizations in this age group were of women (Table 14). The three most commonly reported principal diagnoses at inpatient discharges with any listed PH remained the same throughout the 10-year period (Table 15). The most common principal diagnosis was congestive heart failure, followed by other heart diseases (which included PH) and chronic and unspecified bronchitis. Congestive heart failure was the primary diagnosis for 19% of the PH hospitalizations in 2001 to 2002 and for 18% in 2009 to 2010. The top three diagnoses accounted for 43% of the hospitalizations in 2001 to 2002 and for 37% in 2009 to 2010. Pneumonia and cardiac dysrhythmias rounded out the five leading principal diagnoses, except in 2007 to 2008, when respiratory failure was the fifth most common principal diagnosis. Acute myocardial infarction moved from the sixth most frequent principal diagnosis in 2001 to 2002 to the 10th most frequent in 2009 to 2010. Nephritis, nephrotic syndrome, and nephrosis first appeared as the 10th most common primary diagnosis in 2007 to 2008 and moved up to the seventh most common primary diagnosis in 2009 to 2010. TABLE 15 ] Primary Diagnoses for Hospitalizations with Any Listed Pulmonary Hypertension Diagnosis: United States, 2001/2002 Through 2009/2010

The three most commonly reported principal diagnoses at inpatient discharges with any listed PH remained the same throughout the 10-year period (Table 15). The most common principal diagnosis was congestive heart failure, followed by other heart diseases (which included PH) and chronic and unspecified bronchitis. Congestive heart failure was the primary diagnosis for 19% of the PH hospitalizations in 2001 to 2002 and for 18% in 2009 to 2010. The top three diagnoses accounted for 43% of the hospitalizations in 2001 to 2002 and for 37% in 2009 to 2010. Pneumonia and cardiac dysrhythmias rounded out the five leading principal diagnoses, except in 2007 to 2008, when respiratory failure was the fifth most common principal diagnosis. Acute myocardial infarction moved from the sixth most frequent principal diagnosis in 2001 to 2002 to the 10th most frequent in 2009 to 2010. Nephritis, nephrotic syndrome, and nephrosis first appeared as the 10th most common primary diagnosis in 2007 to 2008 and moved up to the seventh most common primary diagnosis in 2009 to 2010. TABLE 15 ] Primary Diagnoses for Hospitalizations with Any Listed Pulmonary Hypertension Diagnosis: United States, 2001/2002 Through 2009/2010 Primary Diagnosis (ICD-9-CM Diagnosis Codes) 2001/2002 2003/2004 2005/2006 2007/2008 2009/2010 Congestive heart failure: 428.0,428.2-428.4 19.2 19.9 19.5 17.0 17.6 Other heart disease (includes pulmonary hypertension): 391.0-392.0, 393-398, 415-416, 420-426, 428.1, 428.9-429.9 12.7 16.0 13.1 12.9 12.0 Chronic and unspecified bronchitis: 490-491 10.6 7.8 9.2 6.9 7.3 Pneumonia: 480-486 7.3 6.3 5.9 4.4 6.2 Cardiac dysrhythmias: 427 4.3 4.8 5.9 5.8 5.1 Acute myocardial infarction: 410 3.9 3.0 2.4 3.4 2.3 Respiratory failure: 518.81, 518.83, 518.84 3.5 3.8 4.2 4.7 4.3 Coronary atherosclerosis: 414.0, 414.2, 414.3 3.4 4.1 3.4 3.3 2.5 Other diseases of the respiratory system: 416.2, 471-472, 475-478, 487, 488.0, 488.1, 500-511.1, 511.8-518.6, 518.82, 518.89-519 2.8 2.9 2.1 2.5 2.1 Hypertensive heart disease: 402, 404 2.3 1.6 1.2 0.7 1.5 Other COPD and allied conditions: 492, 494-496 1.8 2.1 0.9 0.3 0.5 Cerebrovascular disease: 430-438 1.3 1.3 2.0 1.8 2.1 Nephritis, nephrotic syndrome, and nephrosis: 580-589 0.6 0.8 0.8 2.3 2.7 Asthma: 493 1.4 1.6 1.9 1.1 2.4 Anemias: 280-285 1.0 1.5 1.8 0.9 1.6 Complications of medical and surgical care, not elsewhere classified: 996-999 1.2 1.5 1.5 1.4 1.6 Other diseases of the circulatory system: 390, 392.9, 403, 405, 417, 451-454, 456-459 0.7 1.4 1.5 1.5 1.9 All other diagnoses 21.8 19.8 22.6 29.2 26.6 Data are presented as %. See Table 2 legend for expansion of abbreviation. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11)

circulatory system: 390, 392.9, 403, 405, 417, 451-454, 456-459 0.7 1.4 1.5 1.5 1.9 All other diagnoses 21.8 19.8 22.6 29.2 26.6 Data are presented as %. See Table 2 legend for expansion of abbreviation. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11) There were no significant trends in the average length of stay or the discharge status for any age group or for men or women from 2001 to 2002 to 2009 to 2010 (Fig 6). However, in 2001 to 2002, and again in 2009 to 2010, the average length of stay for hospitalizations for those aged < 35 years was significantly longer than for any other age group. In 2009 to 2010, significantly more men than women were discharged home (P < .02), and significantly more women than men were discharged to long-term care (P < .001) (Fig 7). As age increased, the percentage discharged to home decreased. Figure 6 – Comparison of average length of hospital stay by age groups for discharges for pulmonary hypertension (PH) as any listed International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis: United States, 2001/2002 and 2009/2010. In both 2001/2002 and 2009/2010, the average length of stay for PH hospitalizations for those aged ≤ 35 y was longer than for all other age groups (P < .05). (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11)

nosis: United States, 2001/2002 and 2009/2010. In both 2001/2002 and 2009/2010, the average length of stay for PH hospitalizations for those aged ≤ 35 y was longer than for all other age groups (P < .05). (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11) Figure 7 – Hospital discharge status by sex for pulmonary hypertension as any listed ICD-9-CM diagnosis: United States, 2009/2010. In 2009/2010, more men than women were discharged home (P < .03), and more women than men were discharged to LTC (P < .01). LTC = long-term care; STC = short-term care. See Figure 6 legend for expansion of other abbreviation. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11)

n than women were discharged home (P < .03), and more women than men were discharged to LTC (P < .01). LTC = long-term care; STC = short-term care. See Figure 6 legend for expansion of other abbreviation. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11) Discussion The findings in our report indicate significant increases in death rates associated with PH for women, men, all racial/ethnic groups, and especially among those aged 75 years and older. Considering all PH deaths from 2001 to 2010, we identified significant decreases from PH associated with chronic lower respiratory disease (including emphysema specifically), conditions arising in the perinatal period, congenital malformations, and pulmonary embolism. At the same time, we identified increases from 2001 to 2010 in death rates for PH associated with aortic stenosis, hypertension, coronary heart disease, autoimmune diseases, diabetes, renal disease, and chronic liver disease. Given that PH mortality due to chronic lower respiratory disease represents a significant proportion of total PH mortality, the decline from 2001 to 2010 in these death rates may be affecting the increasing rates identified with other PH-associated conditions; therefore, the significant rates associated with other conditions such as aortic stenosis, hypertension, diabetes, renal disease, and chronic liver disease should be interpreted with caution, given their individually smaller contributions to total PH mortality.

sing rates identified with other PH-associated conditions; therefore, the significant rates associated with other conditions such as aortic stenosis, hypertension, diabetes, renal disease, and chronic liver disease should be interpreted with caution, given their individually smaller contributions to total PH mortality. Death rates among those ≥ 85 years old have increased rapidly since 2001. The Registry to Evaluate Early and Long-term PAH Disease Management (REVEAL Registry) compared demographics from their data on patients with group 1 PAH with those from a National Institutes of Health (NIH) registry from 1982. They found that their current registry patients had a mean age of 45 years compared with 36 years in the NIH registry in 1982, and a higher prevalence of women (78% vs 63%).18

ry) compared demographics from their data on patients with group 1 PAH with those from a National Institutes of Health (NIH) registry from 1982. They found that their current registry patients had a mean age of 45 years compared with 36 years in the NIH registry in 1982, and a higher prevalence of women (78% vs 63%).18 Differences in death rates by sex and race/ethnicity were reported previously,2,19,20 and this report adds important new information about those trends. A previous report2 showed that the trend in the PH mortality rate for men actually declined from 1980 through 2002; information in this report shows that the rates for men continued to decline until 2006 and then began increasing. In general, between 2001 and 2010, mortality rates from COPD declined among men12 and this may explain, in part, the long-term decline in mortality among men with PH that was trending from 1980 to 2006. Reasons for the reversal in trend for men are unclear, but the previous decline may have been associated with reductions in COPD and emphysema, which the current analysis identifies as well. On the other hand, we identified increases in the rates of PH mortality in association with coronary heart disease and hypertension as well as autoimmune disease in men. Few studies have found male sex to be a risk factor for death among patients with PH associated with autoimmune disease.21

nalysis identifies as well. On the other hand, we identified increases in the rates of PH mortality in association with coronary heart disease and hypertension as well as autoimmune disease in men. Few studies have found male sex to be a risk factor for death among patients with PH associated with autoimmune disease.21 Death rates associated with PH for women from 2001 to 2010 were consistently higher than those for men. These sex differences are consistent with data from the REVEAL Registry, in which the patients were more likely to be women.18 Women have higher rates of connective tissue disease than do men.22 The French National Pulmonary Hypertension registry found that women have higher rates of idiopathic and inherited PH compared with men.23 In the REVEAL Registry, which includes only patients with PAH, it was found that roughly one-half of patients with PAH had idiopathic PAH and one-half had associated PAH, with connective tissue disease accounting for roughly 25% of the patients in the registry.24 With current ICD-10 mortality codes, there are challenges in separating PAH as idiopathic vs associated (Fig 1, Table 2). Indeed, most causes of death from PH in the multiple cause of death files are listed as “other secondary PH” or “PH, not otherwise specified.” Few cases are listed as “primary PH.” In a retrospective cohort study of 55 patients with PAH, Kawut et al25 found that patients with systemic sclerosis associated with PH had a higher risk of death despite having similar demographic and hemodynamic characteristics. Research is also ongoing regarding a hormonal mechanism that may explain the sex differences.9,26 Differential sex exposure to anorexigens may also help explain some sex differences,18 although it is unclear to what extent this applies to the US population.

te having similar demographic and hemodynamic characteristics. Research is also ongoing regarding a hormonal mechanism that may explain the sex differences.9,26 Differential sex exposure to anorexigens may also help explain some sex differences,18 although it is unclear to what extent this applies to the US population. Mortality rates for blacks are increasing at a greater rate than they are for whites. The race/sex differences found in these data are consistent with data from the REVEAL Registry, which noted that the greatest ratios of female to male patients in the registry by race/ethnicity were among blacks and the lowest were among whites.18 In addition to increased connective tissue disease among black women, blacks in general have higher rates of connective tissue disease, hypertension, and hemoglobinopathies in association with PH than do whites.22

e to male patients in the registry by race/ethnicity were among blacks and the lowest were among whites.18 In addition to increased connective tissue disease among black women, blacks in general have higher rates of connective tissue disease, hypertension, and hemoglobinopathies in association with PH than do whites.22 Mortality data for all decedents with PH listed on the death certificate are consistent with a widening race/sex gap in recent years, although age-adjusted race/sex mortality gaps are less prominent than are patient ratios in the REVEAL Registry, which is not age adjusted and only includes PAH. Differential race/ethnicity changes over time may also be related to changing population demographics for PH mortality over the past decades. Previous studies have reported that blacks have a shorter survival from the time of diagnosis of PH,27 and have higher rates of PH associated with connective tissue disease. Our findings of increasing rates of PH mortality associated with hypertension, diabetes, and renal failure may also contribute to the racial gap in PH mortality. Poms et al28 found that those with PAH and diabetes have decreased survival. Although they did not report on race/ethnicity associations of PH with chronic kidney disease, Bolignano et al29 suggested that renal failure itself may be a trigger for PH. Overall mortality from diabetes (listed as the UCOD) declined from 2001 to 2010,30 whereas deaths from diabetes in patients with PH have increased.

Although they did not report on race/ethnicity associations of PH with chronic kidney disease, Bolignano et al29 suggested that renal failure itself may be a trigger for PH. Overall mortality from diabetes (listed as the UCOD) declined from 2001 to 2010,30 whereas deaths from diabetes in patients with PH have increased. Interestingly, the geographic pattern of PH death rates by state does not mirror the patterns seen in maps showing all heart disease deaths, where the highest rates are typically located in the deep south and the lowest rates in the northwest.31 This highlights the differences in risk factors and populations at risk of PH. We also found increasing hospitalization rates for both men and women, and there has been some shift in the most common primary reason for admission. Although heart failure is the most common underlying reason for admission associated with PH, the proportion of hospitalizations for heart failure associated with PH has declined, as have most major reasons for hospitalization associated with PH. Yet hospitalizations for asthma, renal conditions, and the category of less common conditions associated with PH have increased. This may reflect better documentation, but it may also suggest important changes in the underlying conditions associated with PH.

ve most major reasons for hospitalization associated with PH. Yet hospitalizations for asthma, renal conditions, and the category of less common conditions associated with PH have increased. This may reflect better documentation, but it may also suggest important changes in the underlying conditions associated with PH. The fact that we found that women have twice as many hospitalizations associated with PH than do men likely reflects the difference in life expectancy but also the disproportionate burden of PH disease experienced by women. Compared with the previous report covering 1980 to 2002, women continue to have higher rates of hospitalizations than do men, and the greatest increase in rates over time is in the oldest age group. For women, increases in the PH hospitalization rate were noted across most age groups, whereas for men, the rate increased only among those ≥ 85 years old. Most hospitalizations for those ≥ 85 years old were among women. Discharge status is likely influenced by these demographic differences, with more women discharged to long-term care and more men discharged to their home.

noted across most age groups, whereas for men, the rate increased only among those ≥ 85 years old. Most hospitalizations for those ≥ 85 years old were among women. Discharge status is likely influenced by these demographic differences, with more women discharged to long-term care and more men discharged to their home. Although some of the increase in mortality from PH-associated conditions may be explained by a heightened awareness of PH and referral to specialists able to make the diagnosis, the significant declines in mortality from congenital malformations, and conditions originating in the perinatal period, may be explained in part by both increased awareness and improved treatment. Although there has been a significant increase in the overall mortality rate for PH since 1980, it is interesting to note that the REVEAL Registry found similar severity at diagnosis (as evidenced by New York Heart Association and World Health Organization functional classes) between the NIH registry and the REVEAL Registry for PAH, and yet one in five patients diagnosed with PAH presented with a history of having symptoms for > 2 years prior to diagnosis.3 In the United Kingdom and Ireland cohort study from 2001 to 2009, patients were still presenting with severe symptoms late in the course of disease.32 Given the accelerating rate of mortality among those aged ≥ 85 years, the overall increase in mortality from PH may reflect different characteristics of disease manifestation among the older patients with additional comorbid conditions. Canadian researchers found that patients ≥ 65 years old were more likely to have PH associated with scleroderma than were younger patients, although they also noted that younger patients with PH associated with scleroderma were more likely to die during a median follow-up of 3 years.33

h additional comorbid conditions. Canadian researchers found that patients ≥ 65 years old were more likely to have PH associated with scleroderma than were younger patients, although they also noted that younger patients with PH associated with scleroderma were more likely to die during a median follow-up of 3 years.33 This report is subject to several limitations related to both mortality and hospitalization data. First, death certificates may not accurately capture the UCOD or contributing causes of death because of inaccuracy of diagnosis or limited knowledge of the deceased’s medical history by the person completing the death certificate. This limitation may result in underreporting of PH on death certificates. Second, race and ethnicity may not be accurately captured on death certificates or hospitalization data, especially for Hispanics and races other than black or white. This may affect the observed mortality and hospitalization rates for Hispanics and other nonwhite/nonblack individuals, especially AI/AN decedents.14 Third, comparisons of mortality data with epidemiologic data from disease registries are limited by the fact that some disease registries are prevalence registries, whereas others are incident registries, or they separate incident and prevalent cases, and by the fact that disease registries are predominately for PAH rather than for all PH causes. Fourth, although trends in UCOD over time can be identified, secular changes in the incidence or prevalence of the diseases reported as the UCOD over time are not reflected in the mortality trends shown here. Therefore, these trends must be interpreted with caution. Fifth, the NHDS does not include patients admitted to federal, military or Veterans Affairs hospitals; therefore, results underestimate the total number of hospitalizations for PH among adults. Sixth, unfortunately, the terminology used in the ICD and ICD-CM does not crosswalk easily with the Dana Point classification of PH, and although using the identical fourth digits in both the ICD-9-CM and the ICD-10, the descriptions do not crosswalk well with each other, nor do they crosswalk well with the Dana Point PH classification of PH. This contributes to challenges in correct classification of causes of death and hospital diagnoses. Lastly, the NHDS increased the number of diagnoses collected in 2010 from seven to 15.

D-10, the descriptions do not crosswalk well with each other, nor do they crosswalk well with the Dana Point PH classification of PH. This contributes to challenges in correct classification of causes of death and hospital diagnoses. Lastly, the NHDS increased the number of diagnoses collected in 2010 from seven to 15. The number of diagnoses collected has a significant impact on estimates, particularly for conditions that are typically secondary diagnoses such as PH. Larger percentage differences can be seen in hospitalizations for older people, who are more likely to carry multiple diagnoses. The estimated number of PH hospitalizations in 2010 using seven diagnoses was 471,864, but the estimate increased to 728, 983 when using 15 diagnoses (a 54% increase in the estimated number of hospitalizations [Table 16]). For those aged 85 years and older, there is a 66% increase in the estimated number of hospitalizations if 15 diagnoses are used, compared with seven diagnoses. This analysis helps explain differences in previously reported hospitalizations associated with PH from the Healthcare Cost and Utilization Project on hospital stays related to PH.34 TABLE 16 ] Difference in Hospital Discharges for Pulmonary Hypertension as Any Listed ICD-9-CM Diagnosis Using Seven vs 15 Diagnoses by Age Group: United States, 2010

The number of diagnoses collected has a significant impact on estimates, particularly for conditions that are typically secondary diagnoses such as PH. Larger percentage differences can be seen in hospitalizations for older people, who are more likely to carry multiple diagnoses. The estimated number of PH hospitalizations in 2010 using seven diagnoses was 471,864, but the estimate increased to 728, 983 when using 15 diagnoses (a 54% increase in the estimated number of hospitalizations [Table 16]). For those aged 85 years and older, there is a 66% increase in the estimated number of hospitalizations if 15 diagnoses are used, compared with seven diagnoses. This analysis helps explain differences in previously reported hospitalizations associated with PH from the Healthcare Cost and Utilization Project on hospital stays related to PH.34 TABLE 16 ] Difference in Hospital Discharges for Pulmonary Hypertension as Any Listed ICD-9-CM Diagnosis Using Seven vs 15 Diagnoses by Age Group: United States, 2010 Age Groups Pulmonary Hypertension Discharges 2010 (Seven Diagnoses) Pulmonary Hypertension Discharges 2010 (15 Diagnoses) Percent Difference Total 472 729 54 < 35 y 27a 30a 10 35-44 y 19 25 33 45-54 y 41 58 43 55-64 y 61 93 52 65-74 y 99 155 57 75-84 y 130 210 62 ≥ 85 y 96 158 66 Numbers are presented in thousands. See Table 2 legend for expansion of abbreviations. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11)

y 27a 30a 10 35-44 y 19 25 33 45-54 y 41 58 43 55-64 y 61 93 52 65-74 y 99 155 57 75-84 y 130 210 62 ≥ 85 y 96 158 66 Numbers are presented in thousands. See Table 2 legend for expansion of abbreviations. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11) a Estimate does not meet standards of reliability or precision. The National Center for Health Statistics considers an estimate to be reliable if it is based on at least 30 discharge records and it has a relative SE of ≤ 30% (ie, the SE is ≤ 30% of the estimate).

y 27a 30a 10 35-44 y 19 25 33 45-54 y 41 58 43 55-64 y 61 93 52 65-74 y 99 155 57 75-84 y 130 210 62 ≥ 85 y 96 158 66 Numbers are presented in thousands. See Table 2 legend for expansion of abbreviations. (Adapted with permission from Centers for Disease Control and Prevention, National Center for Health Statistics.11) a Estimate does not meet standards of reliability or precision. The National Center for Health Statistics considers an estimate to be reliable if it is based on at least 30 discharge records and it has a relative SE of ≤ 30% (ie, the SE is ≤ 30% of the estimate). Conclusions Our report documents increases in mortality associated with PH among men, women, and all race and ethnic groups, and from several conditions commonly associated with PH (hypertension and coronary heart disease, aortic stenosis, liver disease and cirrhosis, and autoimmune disease) but also from renal disease and diabetes, while finding that PH-associated mortality decreased over time from deaths due to congenital malformations among newborns, and for emphysema, chronic lower respiratory disease, and pulmonary embolism. Approximately one-half of deaths associated with PH occur among those under 75 years of age. Although PH death rates have been stable for those aged 1 to 74 years over the past decade, the identified increases in PH death among those with a UCOD from hypertension, coronary heart disease, and valvular disease may be avoidable with improved public health efforts in the primary prevention of heart disease. The increased mortality from autoimmune conditions in association with PH requires further study. Increases in hospitalizations may reflect both improved recognition of PH as well as an increase in treatment options. The decline in PH mortality due to congenital malformations, chronic lower respiratory disease, and emphysema over time is encouraging. The ability to code mortality and hospitalizations using a modified diagnostic classification based on the five major classes in the Dana Point classification could improve the surveillance of PH, which would also aid in understanding age, race, and sex differences in PH mortality.

d emphysema over time is encouraging. The ability to code mortality and hospitalizations using a modified diagnostic classification based on the five major classes in the Dana Point classification could improve the surveillance of PH, which would also aid in understanding age, race, and sex differences in PH mortality. Financial/nonfinancial disclosures: The authors have reported to CHEST that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the US Centers for Disease Control and Prevention. FOR EDITORIAL COMMENT SEE PAGE 239 SEE RELATED ARTICLE PAGE 449 ABBREVIATIONS AAPCaverage annual percent change AI/ANAmerican Indian/Alaska native APCannual percent change ICD-9-CMInternational Classification of Diseases, Ninth Revision, Clinical Modification ICD-10International Classification of Diseases, 10th Revision NHnon-Hispanic NHDSNational Hospital Discharge Survey NIHNational Institutes of Health PAHpulmonary arterial hypertension PHpulmonary hypertension REVEAL RegistryRegistry to Evaluate Early and Long-term PAH Disease Management UCODunderlying cause of death

Pulmonary arterial hypertension (PAH) is a rare complication in patients with connective tissue diseases (CTDs), and it is associated with high mortality rates, particularly in patients with systemic sclerosis (SSc).1 Studies have shown that patients with CTD-associated PAH (CTD-APAH) experience poorer survival compared with patients with idiopathic PAH (IPAH).2‐4 In addition, despite similar baseline hemodynamics, patients with PAH associated with SSc (SSc-APAH) have the poorest survival rates when compared with other CTD-APAH subgroups, including patients with systemic lupus erythematosus, mixed CTD, and rheumatoid arthritis, in both incident and prevalent populations.3,5

.2‐4 In addition, despite similar baseline hemodynamics, patients with PAH associated with SSc (SSc-APAH) have the poorest survival rates when compared with other CTD-APAH subgroups, including patients with systemic lupus erythematosus, mixed CTD, and rheumatoid arthritis, in both incident and prevalent populations.3,5 Risk score calculators have been developed for patients with PAH as a whole, incorporating variables predictive of high mortality, including World Health Organization (WHO) group 1 subgroup, age, sex, New York Heart Association (NYHA) functional class (FC), vital signs, 6-min walk distance (6MWD), brain natriuretic peptide (BNP) level, presence of pericardial effusion, diffusion capacity of the lung for carbon monoxide (Dlco), and baseline hemodynamic variables such as mean right atrial pressure (mRAP), pulmonary vascular resistance (PVR), and cardiac output.6,7 A study focusing on the CTD-APAH population found that higher mRAP, lower 6MWD, higher FC, and the presence of a pericardial effusion were predictive of death.8 In contrast, studies including patients with SSc-APAH alone have identified male sex, lower Dlco, older age, and FC IV status as independent predictors of death.9,10 No studies have evaluated a large cohort of patients with CTD-APAH to identify unique predictors of mortality in patients with SSc-APAH. We sought to use the large Registry to Evaluate Early and Long-Term PAH Management (REVEAL Registry) cohort of patients with CTD-APAH to identify unique predictors of mortality in the patients with SSc-APAH compared with patients with CTD other than SSc (non-SSc-CTD)-APAH that may account for the mortality differences between these groups.

large Registry to Evaluate Early and Long-Term PAH Management (REVEAL Registry) cohort of patients with CTD-APAH to identify unique predictors of mortality in the patients with SSc-APAH compared with patients with CTD other than SSc (non-SSc-CTD)-APAH that may account for the mortality differences between these groups. Materials and Methods REVEAL Registry The REVEAL Registry is a longitudinal registry involving 54 pulmonary hypertension centers in the United States (e-Appendix 1). Each participating center obtained institutional review board approval prior to patient enrollment. The design and objectives of the REVEAL Registry are described elsewhere.11 All patients provided informed consent prior to enrollment, and “enrollment” was defined as the date consent was given. “Diagnosis” was defined as the date of diagnostic right-sided heart catheterization (RHC) occurring at or before the date of enrollment. Patients with new diagnoses were defined as those whose diagnostic RHC occurred within 90 days of enrollment. All consecutive patients who, in the opinion of the enrolling investigator, had a clinical diagnosis of PAH WHO group 112 and met the following inclusion criteria were eligible for enrollment: (1) mean pulmonary artery pressure of > 25 mm Hg at rest or 30 mm Hg with exercise, (2) mean pulmonary capillary wedge pressure or left ventricular end diastolic pressure of ≤ 18 mm Hg, (3) PVR of ≥ 240 dynes/s/cm5 (divide by 80 for Wood units [WU]), and (4) ≥ 3 months of age.

ion criteria were eligible for enrollment: (1) mean pulmonary artery pressure of > 25 mm Hg at rest or 30 mm Hg with exercise, (2) mean pulmonary capillary wedge pressure or left ventricular end diastolic pressure of ≤ 18 mm Hg, (3) PVR of ≥ 240 dynes/s/cm5 (divide by 80 for Wood units [WU]), and (4) ≥ 3 months of age. Data Collection The data in the REVEAL Registry was collected prospectively, but the analyses for this study were performed retrospectively. Data collection methods have been described previously.3 Patients were enrolled from March 2006 through December 2009. Demographics, clinical characteristics, and outcomes were assessed at enrollment and quarterly thereafter. The database of 3,515 patients was locked on February 4, 2013, for the current analyses. We developed an algorithm (Fig 1) to exclude patients with exercise-induced PAH, in accordance with the Dana Point Classification Criteria,12 and those with pulmonary capillary wedge pressure > 15 mm Hg, who have been shown to differ in many respects from those meeting the traditional hemodynamic definition of PAH,13 and included only patients with CTD-APAH. We also excluded those with evidence of significant interstitial lung disease (ILD), defined as those with evidence of “severe” fibrosis on high-resolution CT scan of the chest or “moderate” fibrosis if pulmonary function testing revealed a total lung capacity of < 60% predicted.14 We divided the patients with CTD-APAH into those with SSc-APAH (SSc group) and those with non-SSc-CTD-APAH (non-SSc group).

efined as those with evidence of “severe” fibrosis on high-resolution CT scan of the chest or “moderate” fibrosis if pulmonary function testing revealed a total lung capacity of < 60% predicted.14 We divided the patients with CTD-APAH into those with SSc-APAH (SSc group) and those with non-SSc-CTD-APAH (non-SSc group). Figure 1 – STROBE diagram of the Registry to Evaluate Early and Long-Term PAH Management (REVEAL) Registry patients used in this analysis. We included only patients with CTD-APAH who met the strict criteria of World Health Organization group 1 pulmonary arterial hypertension. CTD-APAH = pulmonary arterial hypertension associated with connective tissue disease; HRCT = high-resolution CT scan of the chest; ILD = interstitial lung disease; non-SSc-CTD = connective tissue disease other than systemic sclerosis; PCWP = pulmonary capillary wedge pressure; SSc = systemic sclerosis; TLC = total lung capacity.

pulmonary arterial hypertension associated with connective tissue disease; HRCT = high-resolution CT scan of the chest; ILD = interstitial lung disease; non-SSc-CTD = connective tissue disease other than systemic sclerosis; PCWP = pulmonary capillary wedge pressure; SSc = systemic sclerosis; TLC = total lung capacity. Statistical Analysis Baseline characteristics at the time of enrollment were compared between the SSc and non-SSc groups, using the Student t or Wilcoxon test to compare continuous variables and the χ2 or Fisher exact test to compare categorical variables. Because BNP levels were highly skewed, the variables were log transformed for comparison as continuous variables. Cumulative probabilities of survival at 3 years were calculated using the Kaplan-Meier estimator for both the previously and newly diagnosed populations, and differences between the SSc and non-SSc groups were compared using the log-rank test. Follow-up time was calculated from the date of enrollment. Cox regression models identified significant predictors of mortality in the SSc and non-SSc populations. All variables identified previously as candidate predictors of mortality in the overall REVEAL Registry population were evaluated in univariate and multivariate models. Stepwise selection was used to determine the final model, retaining only variables with P < .05. SAS, version 9.1 (SAS Institute Inc) statistical software was used for all analyses.

ously as candidate predictors of mortality in the overall REVEAL Registry population were evaluated in univariate and multivariate models. Stepwise selection was used to determine the final model, retaining only variables with P < .05. SAS, version 9.1 (SAS Institute Inc) statistical software was used for all analyses. Results Baseline Characteristics in Patients With CTD-APAH Of 3,515 patients enrolled in the REVEAL Registry, 815 were identified as having CTD-APAH (Fig 1). Of these, 804 (500 SSc and 304 non-SSc) who did not have significant ILD were selected for these analyses. The majority of patients in the non-SSc group had systemic lupus erythematosus-APAH or mixed CTD-APAH (Table 1). Patients with SSc were older and had a shorter time between diagnostic RHC and enrollment into the database than did the patients with non-SSc-CTD-APAH (Table 2). Patients with SSc-APAH had more severe disease overall, with a higher NYHA FC, shorter 6MWD, higher Borg dyspnea index, lower Dlco, and higher BNP level. Patients with SSc-APAH were also more likely to have renal insufficiency and pericardial effusions than patients with non-SSc-CTD-APAH. Although there was a strong trend toward higher mRAP in the SSc group, there were no significant differences in hemodynamics or PAH-specific therapies at the time of enrollment in the SSc vs non-SSc groups. TABLE 1 ]  Types of CTD-APAH

Results Baseline Characteristics in Patients With CTD-APAH Of 3,515 patients enrolled in the REVEAL Registry, 815 were identified as having CTD-APAH (Fig 1). Of these, 804 (500 SSc and 304 non-SSc) who did not have significant ILD were selected for these analyses. The majority of patients in the non-SSc group had systemic lupus erythematosus-APAH or mixed CTD-APAH (Table 1). Patients with SSc were older and had a shorter time between diagnostic RHC and enrollment into the database than did the patients with non-SSc-CTD-APAH (Table 2). Patients with SSc-APAH had more severe disease overall, with a higher NYHA FC, shorter 6MWD, higher Borg dyspnea index, lower Dlco, and higher BNP level. Patients with SSc-APAH were also more likely to have renal insufficiency and pericardial effusions than patients with non-SSc-CTD-APAH. Although there was a strong trend toward higher mRAP in the SSc group, there were no significant differences in hemodynamics or PAH-specific therapies at the time of enrollment in the SSc vs non-SSc groups. TABLE 1 ]  Types of CTD-APAH Type of CTD No. (%) All SSc-APAH 500 (62.2) SSc, limited 299 (37.2) SSc, diffuse 99 (12.3) SSc, unknown subtype 102 (12.7) All non-SSc-CTD-APAH 304 (37.8) Systemic lupus erythematosus 127 (15.8) Mixed CTD 71 (8.8) Rheumatoid arthritis 42 (5.2) Sjogren syndrome 15 (1.9) Dermatomyositis/polymyositis 8 (1.0) Undifferentiated CTD 12 (1.5) Overlap syndrome 15 (1.9) Other 4 (0.5) Unknown 10 (1.2) APAH = associated with pulmonary arterial hypertension; CTD = connective tissue disease; non-SSc-CTD = connective tissue disease other than systemic sclerosis; SSc = systemic sclerosis.

e 15 (1.9) Dermatomyositis/polymyositis 8 (1.0) Undifferentiated CTD 12 (1.5) Overlap syndrome 15 (1.9) Other 4 (0.5) Unknown 10 (1.2) APAH = associated with pulmonary arterial hypertension; CTD = connective tissue disease; non-SSc-CTD = connective tissue disease other than systemic sclerosis; SSc = systemic sclerosis. TABLE 2 ]  Characteristics, Hemodynamics, and Cardiac and Pulmonary Function at Enrollment

e 15 (1.9) Dermatomyositis/polymyositis 8 (1.0) Undifferentiated CTD 12 (1.5) Overlap syndrome 15 (1.9) Other 4 (0.5) Unknown 10 (1.2) APAH = associated with pulmonary arterial hypertension; CTD = connective tissue disease; non-SSc-CTD = connective tissue disease other than systemic sclerosis; SSc = systemic sclerosis. TABLE 2 ]  Characteristics, Hemodynamics, and Cardiac and Pulmonary Function at Enrollment Characteristic SSc-APAH (n = 500) Non-SSc-CTD-APAH (n = 304) P Value Age at baseline,a y No. 500 304 … Mean ± SD 61.65 ± 11.25 49.88 ± 14.38 < .001 Male sex, No. (%) 63 (12.6) 28 (9.2) .14 Time from diagnostic RHC to enrollment, mo No. 500 304 … Mean ± SD 19.33 ± 23.11 26.72 ± 35.66 < .001 Newly diagnosed, No. (%) 166 (33.2) 88 (28.9) 0.21 NYHA FC, No. (%) < .0001 I 15 (3.4) 25 (9.2) II 121 (27.8) 105 (38.7) III 256 (58.9) 127 (46.9) IV 43 (9.9) 14 (5.2) 6MWD, m No. 380 248 … Mean ± SD 294.01 ± 114.6 360.21 ± 122.2 < .001 Heart rate, bpm No. 471 287 … Mean ± SD 84.29 ± 14.94 83.64 ± 14.41 .55 Systolic BP, mm Hg No. 477 287 … Mean ± SD 118.71 ± 18.97 119.28 ± 19.56 .69 Borg dyspnea index No. 327 220 … Mean ± SD 3.67 ± 2.07 3.15 ± 2.28 .005 Renal insufficiency, No. (%) 41 (8.4) 9 (3.0) .0024 mRAP, mm Hg No. 449 276 … Mean ± SD 9.04 ± 5.77 8.21 ± 5.06 .052 mPAP at rest, mm Hg No. 500 304 … Mean ± SD 44.59 ± 11.43 45.48 ± 10.67 .27 PCWP at rest, mm Hg No. 500 304 … Mean ± SD 9.11 ± 3.48 8.85 ± 3.48 .29 Cardiac output,b L/min No. 499 303 … Mean ± SD 4.42 ± 1.45 4.28 ± 1.35 .20 Cardiac index, L/min/m2 No. 391 237 … Mean ± SD 2.50 ± 0.81 2.40 ± 0.75 .11 PVR,c Wood units No. 499 303 … Mean ± SD 9.31 ± 5.24 9.79 ± 5.34 .21 PVR index,c Wood units × m2 No. 391 237 … Mean ± SD 16.37 ± 9.05 17.36 ± 9.46 .19 FEV1,d % predicted No. 350 179 … Mean ± SD 71.93 ± 18.43 73.90 ± 19.20 .25 FVC,d % predicted No. 352 181 … Mean ± SD 74.08 ± 19.22 76.93 ± 20.12 .11 FEV1/FVC ratioe No. 374 200 .068 Mean ± SD 0.76 ± 0.09 0.77 ± 0.10 … Dlco,d % predicted No. 344 186 … Mean ± SD 40.83 ± 16.27 50.36 ± 1 9.00 < .001 Pericardial effusion, No. (%) None 222 (57.1) 159 (68.2) .0090 Mild 121 (31.1) 62 (26.6) … Moderate 36 (9.3) 12 (5.2) … Moderate-severe 5 (1.3) 0 (0.0) … Severe 5 (1.3) 0 (0.0) … BNP, pg/mL No. 223 154 … Mean ± SD 562.38 ± 929.9 313.49 ± 685.4 .005 N-terminal BNP, pg/mL No. 65 26 … Mean ± SD 3192.37 ± 4687 932.73 ± 1345 .018 PAH medications at enrollment, No. (%) Prostacyclin 154 (31.8) 96 (32.8) .77 ERA 217 (44.7) 120 (41.0) .30 PDE-5 inhibitor 223 (46.0) 137 (46.8) .83 CCB for PAH 42 (8.7) 27 (9.2) .79 PAH medications, No.

SD 562.38 ± 929.9 313.49 ± 685.4 .005 N-terminal BNP, pg/mL No. 65 26 … Mean ± SD 3192.37 ± 4687 932.73 ± 1345 .018 PAH medications at enrollment, No. (%) Prostacyclin 154 (31.8) 96 (32.8) .77 ERA 217 (44.7) 120 (41.0) .30 PDE-5 inhibitor 223 (46.0) 137 (46.8) .83 CCB for PAH 42 (8.7) 27 (9.2) .79 PAH medications, No. (%) 0 90 (18.6) 49 (16.7) .47 1 231 (47.6) 149 (50.9) … 2 129 (26.6) 81 (27.6) … 3 35 (7.2) 14 (4.8) … On combination PAH medications, No. (%) 164 (33.8) 95 (32.4) .69 P value calculation uses χ2 test for categorical data or Fisher exact test for categorical data with small cell counts (≤ 5%), and Student t test for continuous data. 6MWD = 6-min walk distance; BNP = brain natriuretic peptide; bpm = beats per min; CCB = calcium channel blocker; Dlco = diffusion capacity of the lung for carbon monoxide; ERA = endothelin receptor agonist; FC = functional class; FCO = Fick cardiac output; mPAP = mean pulmonary arterial pressure; mRAP = mean right atrial pressure; NYHA = New York Heart Association; PAH = pulmonary arterial hypertension; PCWP = pulmonary capillary wedge pressure; PDE-5 = phosphodiesterase type-5; PVR = pulmonary vascular resistance; RHC = right-sided heart catheterization. See Table 1 legend for expansion of other abbreviations. a Age = (date of informed consent − date of birth)/365.25. b Cardiac output = FCO, or, if FCO is missing, then cardiac output = thermodilution cardiac output.

(%) 0 90 (18.6) 49 (16.7) .47 1 231 (47.6) 149 (50.9) … 2 129 (26.6) 81 (27.6) … 3 35 (7.2) 14 (4.8) … On combination PAH medications, No. (%) 164 (33.8) 95 (32.4) .69 P value calculation uses χ2 test for categorical data or Fisher exact test for categorical data with small cell counts (≤ 5%), and Student t test for continuous data. 6MWD = 6-min walk distance; BNP = brain natriuretic peptide; bpm = beats per min; CCB = calcium channel blocker; Dlco = diffusion capacity of the lung for carbon monoxide; ERA = endothelin receptor agonist; FC = functional class; FCO = Fick cardiac output; mPAP = mean pulmonary arterial pressure; mRAP = mean right atrial pressure; NYHA = New York Heart Association; PAH = pulmonary arterial hypertension; PCWP = pulmonary capillary wedge pressure; PDE-5 = phosphodiesterase type-5; PVR = pulmonary vascular resistance; RHC = right-sided heart catheterization. See Table 1 legend for expansion of other abbreviations. a Age = (date of informed consent − date of birth)/365.25. b Cardiac output = FCO, or, if FCO is missing, then cardiac output = thermodilution cardiac output. c PVR (Wood units) = (mean pulmonary arterial pressure at rest − PCWP at rest)/cardiac output, where cardiac output = FCO, or, if FCO is missing, then cardiac output = thermodilution cardiac output. d Predicted value based on Hankinson et al14 computation. e FEV1/FVC ratio is missing if FVC is zero.

b Cardiac output = FCO, or, if FCO is missing, then cardiac output = thermodilution cardiac output. c PVR (Wood units) = (mean pulmonary arterial pressure at rest − PCWP at rest)/cardiac output, where cardiac output = FCO, or, if FCO is missing, then cardiac output = thermodilution cardiac output. d Predicted value based on Hankinson et al14 computation. e FEV1/FVC ratio is missing if FVC is zero. Poorer Survival in SSc-APAH Compared With Non-SSc-CTD-APAH Three-year survival in the SSc group was worse than in the non-SSc group in both the previously and newly diagnosed populations (61.4% ± 2.7% vs 80.9% ± 2.7% and 51.2% ± 4.0% vs 76.4% ± 4.6%, respectively; P < .001) (Fig 2). Figure 2 – Three-year survival curves in patients with SSc and non-SSc-CTD-APAH. A, Three-year survival from enrollment in the newly diagnosed SSc group was 51.2% ± 4.0% compared with 76.4% ± 4.6% in the non-SSc-CTD group (P < .001). B, Three-year survival from enrollment in the previously diagnosed SSc group was 61.4% ± 2.7% compared with 80.9% ± 2.7% in the non-SSc-CTD group (P < .001). See Figure 1 legend for expansion of abbreviations.

the newly diagnosed SSc group was 51.2% ± 4.0% compared with 76.4% ± 4.6% in the non-SSc-CTD group (P < .001). B, Three-year survival from enrollment in the previously diagnosed SSc group was 61.4% ± 2.7% compared with 80.9% ± 2.7% in the non-SSc-CTD group (P < .001). See Figure 1 legend for expansion of abbreviations. Unique Predictors of Mortality in SSc-APAH Figure 3 shows the univariate analyses of previously identified predictors of mortality from the overall REVEAL Registry cohort in the SSc and non-SSc groups. The following variables were predictive of mortality in both groups: age > 60 years, NYHA FC III or IV status, 6MWD < 165 m, and BNP > 180 pg/mL. 6MWD ≥ 440 m was protective in both groups. Unique predictors of mortality in the SSc group, but not the non-SSc group, included male sex, systolic BP (SBP) ≤ 110 mm Hg, pericardial effusion, Dlco ≤ 32% predicted, mRAP > 20 mm Hg within 1 year, PVR > 32 WU, and newly diagnosed status. BNP levels < 50 pg/mL were protective in patients with SSc (hazard ratio [HR] = 0.34; 95% CI, 0.16-0.72; P = .005) but not in the non-SSc group (HR = 0.68; 95% CI, 0.36-1.29; P = .24). Figure 3 also shows the univariate analyses of additional variables that are relevant to the CTD-APAH population. A higher glomerular filtration rate was protective in both groups. Mild to moderate ILD was the only feature that increased mortality in patients with non-SSc-CTD-APAH but not in patients with SSc-APAH (HR = 2.19; 95% CI, 1.14-4.23; P = .02 vs HR = 0.84; 95% CI, 0.55-1.30; P = .44). When compared with IPAH, mRAP > 20 mm Hg within 1 year, PVR > 32 WU, and newly diagnosed status remained unique predictors of death in the SSc-APAH group.

eased mortality in patients with non-SSc-CTD-APAH but not in patients with SSc-APAH (HR = 2.19; 95% CI, 1.14-4.23; P = .02 vs HR = 0.84; 95% CI, 0.55-1.30; P = .44). When compared with IPAH, mRAP > 20 mm Hg within 1 year, PVR > 32 WU, and newly diagnosed status remained unique predictors of death in the SSc-APAH group. Figure 3 – Predictors of mortality for patients with SSc-APAH and non-SSc-CTD-APAH using univariate Cox regression analyses. Unique predictors of mortality in the SSc group, but not the non-SSc group, included male sex, SBP ≤ 110 mm Hg, pericardial effusion, Dlco ≤ 32% predicted, mRAP > 20 mm Hg within 1 y, PVR > 32 WU, and newly diagnosed status. BNP levels < 50 pg/mL were protective in patients with SSc, but not in the non-SSc group. Higher GFR was protective in both groups. Mild to moderate ILD was the only feature that increased mortality in the non-SSc group but not in patients with SSc. 6MWD = 6-min walk distance; BNP = brain natriuretic peptide; DLCO = diffusion capacity of the lung for carbon monoxide; FC = functional class; GFR = glomerular filtration rate; HR = hazard ratio; mRAP = mean right atrial pressure; NYHA = New York Heart Association; PVR = pulmonary vascular resistance; SBP = systolic BP; WHO = World Health Organization; WU = Wood units. See Figure 1 legend for expansion of other abbreviations.

monoxide; FC = functional class; GFR = glomerular filtration rate; HR = hazard ratio; mRAP = mean right atrial pressure; NYHA = New York Heart Association; PVR = pulmonary vascular resistance; SBP = systolic BP; WHO = World Health Organization; WU = Wood units. See Figure 1 legend for expansion of other abbreviations. In multivariate analyses, the following variables remained predictive of mortality in both the SSc and non-SSc groups: NYHA FC III or IV status and BNP > 180 pg/mL (Table 3). Unique predictors of mortality in the SSc group included men > 60 years, SBP ≤ 110 mm Hg, 6MWD < 165 m, mRAP > 20 mm Hg within 1 year, and PVR > 32 WU. 6MWD ≥ 440 m was protective in the non-SSc group, but not in the SSc group, whereas BNP < 50 pg/mL was protective in the SSc group, but not in the non-SSc group. TABLE 3 ]  Multivariate Model of Predictors of Mortality Risk Score Characteristic HR 95% CI P Value SSc-APAH Men aged > 60 y 2.222 1.421-3.474 < .001 NYHA FC III 1.326 1.002-1.756 .049 NYHA FC IV 2.938 1.921-4.492 < .001 Systolic BP ≥ 110 mm Hg 1.334 1.034-1.723 .027 6MWD < 165 m 2.252 1.614-3.142 < .001 BNP < 50 pg/mL 0.450 0.209-0.966 .040 BNP > 180 pg/mL 2.082 1.617-2.682 < .001 mRAP > 20 mm Hg within 1 y 1.910 1.003-3.637 .049 PVR > 32 Wood units 14.567 3.464-61.262 < .001 Non-SSc-CTD-APAH NYHA FC III 1.679 1.067-2.641 .025 NYHA FC IV 5.427 2.588-11.383 < .001 6MWD ≥ 440 m 0.293 0.118-0.732 .009 BNP > 180 pg/mL 2.466 1.589-3.826 < .001 HR = hazard ratio. See Table 1 and 2 legends for expansion of other abbreviations.

y 1.910 1.003-3.637 .049 PVR > 32 Wood units 14.567 3.464-61.262 < .001 Non-SSc-CTD-APAH NYHA FC III 1.679 1.067-2.641 .025 NYHA FC IV 5.427 2.588-11.383 < .001 6MWD ≥ 440 m 0.293 0.118-0.732 .009 BNP > 180 pg/mL 2.466 1.589-3.826 < .001 HR = hazard ratio. See Table 1 and 2 legends for expansion of other abbreviations. Discussion Our study provides further evidence that patients with SSc-APAH experience higher mortality rates than do patients with other CTD-APAH in both incident and prevalent populations. Our results validate the usefulness of the risk score calculator in patients with CTD-APAH, including in patients with SSc-APAH. We identified several baseline risk factors that were significantly associated with mortality in the SSc-APAH population in comparison with the non-SSc-CTD-APAH population, including being an elderly man, having a low SBP, having poor exercise capacity, and having severe hemodynamic indices including elevated mRAP and PVR. Identifying patients with SSc-APAH with high mortality risk based on the presence of these unique predictors of mortality will enable physicians to monitor these patients more closely and escalate therapy when indicated.

ving poor exercise capacity, and having severe hemodynamic indices including elevated mRAP and PVR. Identifying patients with SSc-APAH with high mortality risk based on the presence of these unique predictors of mortality will enable physicians to monitor these patients more closely and escalate therapy when indicated. Three-year survival in the newly diagnosed SSc-APAH population was 51%, which is similar to survival rates found in other cohorts assessed in the modern treatment era.1,5,9,15,16 Other studies have found better survival rates (75%-81%) in patients with SSc-APAH; these rates are similar to the survival rate of 77% that we and others observed in patients with non-SSc-CTD-APAH.3,5,10,17,18 This survival discrepancy could be related to early detection algorithms that have been implemented in these SSc-APAH cohorts, with the goal to initiate PAH-specific therapy when the disease is less severe. Survival in patients with non-SSc-CTD-APAH appears to be more similar to those with IPAH than to those with SSc-APAH, despite similar baseline hemodynamics and PAH-specific therapies.3 Whether initiating aggressive PAH treatment in patients with SSc-APAH with a particular high mortality risk may improve outcomes remains an important question to be answered.

PAH appears to be more similar to those with IPAH than to those with SSc-APAH, despite similar baseline hemodynamics and PAH-specific therapies.3 Whether initiating aggressive PAH treatment in patients with SSc-APAH with a particular high mortality risk may improve outcomes remains an important question to be answered. Overall, predictors identified in the multivariate model in SSc-APAH were very similar to the core predictors for PAH as a whole, including all subtypes.6 Our results concur with those of other studies on patients with SSc-APAH in that male sex, older age, and FC III and IV status were significant predictors of death.5,9,10,15 Our results confirmed those of a single-center study that identified high PVR as a strong predictor of mortality.19 Unlike these other studies, we did not find that low Dlco or glomerular filtration rate were predictive of mortality in the SSc-APAH group in multivariate analyses, although they were significant in univariate analyses. Lefèvre et al15 identified additional poor prognostic factors in patients with SSc with pulmonary hypertension in a meta-analysis including patients with WHO groups II and III pulmonary hypertension: pericardial effusion, low 6MWD, high mean pulmonary arterial pressure, poor cardiac index, and elevated mRAP were poor prognostic factors. Although pericardial effusion lost its significance in our multivariate analysis of patients with SSc-APAH, poor exercise capacity and elevated mRAP remained significant predictors of death. Interestingly, 6MWD < 165 m was predictive of death only in the SSc group, whereas 6MWD ≥ 440 m was protective only in the non-SSc-CTD-APAH group in multivariate analyses. A potential explanation for these discrepancies is that patients with SSc can suffer from the presence of contractures and tendon friction rubs that can significantly limit mobility (particularly those with diffuse skin disease) in addition to other factors that limit exercise capacity (such as anemia and joint or muscle inflammation) in patients with other CTDs.20,21 However, including all variables in the multivariate model without stepwise selection, 6MWD < 165 m was a significant predictor of death in the non-SSc group (HR = 2.03; 95% CI, 1.01-4.12; P = .05), and 6MWD ≥ 440 m showed a trend toward a protective effect in the SSc group (HR = 0.62; 95% CI, 0.33-1.15; P = .13).

0,21 However, including all variables in the multivariate model without stepwise selection, 6MWD < 165 m was a significant predictor of death in the non-SSc group (HR = 2.03; 95% CI, 1.01-4.12; P = .05), and 6MWD ≥ 440 m showed a trend toward a protective effect in the SSc group (HR = 0.62; 95% CI, 0.33-1.15; P = .13). In addition, when we evaluated the effect of 6MWD on mortality risk in the various cutaneous subgroups of SSc, an increase in distance of 100 m was significantly protective in all three groups (P < .001): diffuse HR = 0.53 (95% CI, 0.38-0.75); limited 0.59 (95% CI, 0.51-0.68); unclassified 0.54 (95% CI, 0.40-0.71).

0,21 However, including all variables in the multivariate model without stepwise selection, 6MWD < 165 m was a significant predictor of death in the non-SSc group (HR = 2.03; 95% CI, 1.01-4.12; P = .05), and 6MWD ≥ 440 m showed a trend toward a protective effect in the SSc group (HR = 0.62; 95% CI, 0.33-1.15; P = .13). In addition, when we evaluated the effect of 6MWD on mortality risk in the various cutaneous subgroups of SSc, an increase in distance of 100 m was significantly protective in all three groups (P < .001): diffuse HR = 0.53 (95% CI, 0.38-0.75); limited 0.59 (95% CI, 0.51-0.68); unclassified 0.54 (95% CI, 0.40-0.71). In our study, BNP > 180 pg/mL increased the risk of death in both the SSc and non-SSc-APAH groups by more than twofold, as has also been shown in patients with IPAH.22 We and others have shown that patients with SSc-APAH have markedly elevated BNP and N-terminal-pro-BNP (NT-pro-BNP) levels compared with patients with IPAH and patients with non-SSc-CTD-APAH.3,23 Williams et al24 found in a UK SSc-APAH cohort that for every order of magnitude increase in baseline NT-pro-BNP level there was a fourfold increased risk of death (P = .002). In addition, several studies have found that NT-pro-BNP is useful in the screening and early detection of PAH in patients with SSc, and this biomarker has been integrated into novel screening algorithms.25‐27 To our knowledge, our study is the first to show that BNP is an independent predictor of mortality in patients with CTD-APAH and SSc-APAH, in particular. Unfortunately NT-pro-BNP levels were not available in 89% of our CTD-APAH cohort, and, therefore, they could not be included in the regression models.

lgorithms.25‐27 To our knowledge, our study is the first to show that BNP is an independent predictor of mortality in patients with CTD-APAH and SSc-APAH, in particular. Unfortunately NT-pro-BNP levels were not available in 89% of our CTD-APAH cohort, and, therefore, they could not be included in the regression models. To our knowledge, this is the first study to identify low baseline SBP ≤ 110 mm Hg as an independent predictor of death in patients with SSc-APAH. Other studies have shown that low SBP, both at peak exercise and upon admission to the hospital for right-sided heart failure, is an independent risk factor for death in PAH.28,29 A potential pathophysiologic explanation for this finding is that the presence of high right ventricular pressure results in a more pronounced effect of low SBP on coronary perfusion. Thus, low SBP can lead to greater right ventricular dysfunction caused by ischemia. In addition, low SBP may be a sign of low cardiac output, reduced stroke volume, and neurohormonal activation.29 Unless complicated by renal disease, patients with SSc have relatively low baseline BP,30 and the mean SBP was 119 ± 19 mm Hg in the patients with SSc-APAH in our study. Given that BP can be monitored easily, identification of low baseline SBP as a risk factor in SSc-APAH is an important finding.

hormonal activation.29 Unless complicated by renal disease, patients with SSc have relatively low baseline BP,30 and the mean SBP was 119 ± 19 mm Hg in the patients with SSc-APAH in our study. Given that BP can be monitored easily, identification of low baseline SBP as a risk factor in SSc-APAH is an important finding. We did not find that mild to moderate ILD was predictive of death in patients with SSc-APAH. Although a significant predictor in the non-SSc-APAH group in univariate analysis, it was no longer significant in multivariate analysis. We attempted to exclude patients with substantial ILD as defined previously but did not have precise measurements regarding the degree of fibrosis on imaging.

patients with SSc-APAH. Although a significant predictor in the non-SSc-APAH group in univariate analysis, it was no longer significant in multivariate analysis. We attempted to exclude patients with substantial ILD as defined previously but did not have precise measurements regarding the degree of fibrosis on imaging. Our study does have some limitations. The SSc-APAH and non-SSc-CTD-APAH cohorts are smaller than the overall cohort. Thus, differences in significant multivariable predictors may be caused by loss of power as opposed to true differences in predictors for different subtypes. In addition, the model does not include therapies. The majority of REVEAL Registry patients, particularly patients who had previous diagnoses, were receiving phosphodiesterase-5 inhibitors, endothelin receptor antagonists, prostacyclins, or a combination. Therefore, the model does not provide insights into prognosis for untreated patients. Although 86% of the patients with CTD-APAH were enrolled at sites that routinely involve a rheumatologist in the diagnosis and care of these patients, misclassification of some patients may have occurred. Finally, the analysis only assessed variables available in the REVEAL Registry database. There may be additional factors particular to patients with CTD-APAH, such as autoantibody status, that could impact the results.

st in the diagnosis and care of these patients, misclassification of some patients may have occurred. Finally, the analysis only assessed variables available in the REVEAL Registry database. There may be additional factors particular to patients with CTD-APAH, such as autoantibody status, that could impact the results. Conclusions In conclusion, patients with SSc-APAH have higher mortality rates than patients with non-SSc-CTD-APAH. Our results validate the usefulness of the PAH risk score in patients with SSc-APAH. We have identified unique predictors of mortality in patients with SSc-APAH, including being an older man, having a low baseline SBP, having poor exercise capacity, and having an elevated mRAP and PVR; these can be used to identify high-risk patients who may benefit from closer monitoring and more aggressive treatment. Supplementary Material Online Supplement Click here for additional data file. Author contributions: L. C. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. L. C., H. W. F., R. B., D. P. M., L. P., P. M. H., M. M., M. R. N., and R. T. Z. contributed to data analysis and interpretation, drafting and critical review of the manuscript, and approval of the final version.

in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. L. C., H. W. F., R. B., D. P. M., L. P., P. M. H., M. M., M. R. N., and R. T. Z. contributed to data analysis and interpretation, drafting and critical review of the manuscript, and approval of the final version. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Chung has received research support funding from Gilead Sciences, Inc; United Therapeutics Corp; Pfizer, Inc; and Actelion Pharmaceuticals Ltd, and has served on the Advisory Board for Gilead Sciences, Inc. Dr Farber has served as a consultant for Gilead Sciences, Inc, Actelion Pharmaceuticals Ltd, Bayer, United Therapeutics Corp, and Bristol-Myers Squibb; has served on the speakers bureau for Actelion Pharmaceuticals Ltd, Gilead Sciences, Inc, and Bayer; and has received grant support from Gilead Sciences, Inc and United Therapeutics Corp. Dr Benza has grant support from Actelion Pharmaceuticals Ltd and is a member of the Steering Committee for the REVEAL Registry. Mr Miller is an employee of ICON Clinical Research, a company that receives funding from Actelion Pharmaceuticals Ltd and acts as a BioStatistical CRO for the REVEAL Registry, as well as received funding from other pharmaceutical companies. Ms Parsons is an employee of ICON Clinical Research, a company that receives funding from Actelion Pharmaceuticals Ltd and acts as a BioStatistical CRO for the REVEAL Registry, as well as received funding from other pharmaceutical companies. Dr Hassoun has received research funding support from Actelion/CoTherix and is on the Advisory Board for Novartis. Dr McGoon has received research funding from Gilead Sciences, Inc and Medtronic, Inc and has served on steering committees for Gilead Sciences, Inc and Lung Rx, LLC and has participated on clinical end-point committees in studies sponsored by Actelion Pharmaceuticals Ltd. He is on a Data Safety Monitoring Board for a study sponsored by Gilead Sciences, Inc and has received honoraria for his service on the REVEAL Registry Steering Committee, which is supported by Actelion Pharmaceuticals Ltd. Dr Zamanian has received research funding support through the Enteligence-Actelion career development research grant and has served as a consultant to United Therapeutics Corporation and Gilead Sciences, Inc.

for his service on the REVEAL Registry Steering Committee, which is supported by Actelion Pharmaceuticals Ltd. Dr Zamanian has received research funding support through the Enteligence-Actelion career development research grant and has served as a consultant to United Therapeutics Corporation and Gilead Sciences, Inc. Dr Nicolls has reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: The sponsor, Actelion Pharmaceuticals US Inc, provided the study design, statistical analysis plan, and management of study registry and participated in data analysis, interpretation, and preparation of manuscript. Other contributions: The authors thank Wolters Kluwer for coordinating feedback among the authors. Additional information: The e-Appendix can be found in the Supplemental Materials section of the online article. FUNDING/SUPPORT: Actelion Pharmaceuticals US Inc is the sponsor of REVEAL Registry and provided funding and support for the analysis presented. ABBREVIATIONS 6MWD6-min walk distance BNPbrain natriuretic peptide CTDconnective tissue disease CTD-APAHpulmonary arterial hypertension associated with connective tissue disease Dlcodiffusion capacity of the lung for carbon monoxide FCfunctional class HRhazard ratio ILDinterstitial lung disease IPAHidiopathic pulmonary arterial hypertension mRAPmean right atrial pressure non-SSc-CTDconnective tissue disease other than systemic sclerosis NT-pro-BNPN-terminal-pro-brain natriuretic peptide NYHANew York Heart Association PAHpulmonary arterial hypertension

Dlcodiffusion capacity of the lung for carbon monoxide FCfunctional class HRhazard ratio ILDinterstitial lung disease IPAHidiopathic pulmonary arterial hypertension mRAPmean right atrial pressure non-SSc-CTDconnective tissue disease other than systemic sclerosis NT-pro-BNPN-terminal-pro-brain natriuretic peptide NYHANew York Heart Association PAHpulmonary arterial hypertension PVRpulmonary vascular resistance REVEAL RegistryRegistry to Evaluate Early and Long-Term PAH Management RHCright-sided heart catheterization SBPsystolic BP SScsystemic sclerosis SSc-APAHpulmonary arterial hypertension associated with systemic sclerosis WHOWorld Health Organization WUWood units

Asthma is the most common inflammatory disease, and its prevalence is increasing throughout the world. Although corticosteroids are the most effective antiinflammatory agents for the treatment of asthma,1 adult patients with asthma who currently smoke have relative steroid resistance.2 Furthermore, their asthma becomes more severe and their lung function decreases more rapidly compared with nonsmoking patients with asthma.3,4 Passive smoking (PS) also worsens asthma symptoms and causes poor asthma control in both adults and children.5,6 Exposure to parental smoking is related to exacerbation of asthma symptoms in children and can be a risk factor for the persistence of asthma in later childhood.5 However, the molecular mechanisms of the effects of PS exposure in childhood are currently unknown.

auses poor asthma control in both adults and children.5,6 Exposure to parental smoking is related to exacerbation of asthma symptoms in children and can be a risk factor for the persistence of asthma in later childhood.5 However, the molecular mechanisms of the effects of PS exposure in childhood are currently unknown. There are several possible mechanisms for corticosteroid resistance in asthma, including overexpression of proinflammatory transcription factors, phosphorylation of glucocorticoid receptors, and increases in the decoy glucocorticoid receptor-β.7 Histone deacetylase (HDAC)-2 (HDAC2) has been shown to be a prerequisite molecule for corticosteroids to switch off activated inflammatory genes. Oxidative stress, such as tobacco smoke, impairs HDAC2 function, leading to corticosteroid insensitivity in vitro and in vivo.8‐10 HDAC2 expression and activity are reduced in the airways of, and alveolar macrophages (AMs) from, adults with severe asthma11‐13 and COPD.14,15 Even more importantly, in patients with asthma who smoke, there is a significantly greater reduction of HDAC activity in bronchial biopsy specimens than in patients with asthma who do not smoke.16 Further analysis revealed that oxidative stress such as tobacco smoke impairs HDAC2 via phosphoinositide-3-kinase (PI3K) δ (PI3Kδ)/Akt activation.9,17 In this study, we tested the hypothesis that passive exposure to tobacco smoke is associated with reduced HDAC2 in AMs in children with severe and refractory asthma.

.16 Further analysis revealed that oxidative stress such as tobacco smoke impairs HDAC2 via phosphoinositide-3-kinase (PI3K) δ (PI3Kδ)/Akt activation.9,17 In this study, we tested the hypothesis that passive exposure to tobacco smoke is associated with reduced HDAC2 in AMs in children with severe and refractory asthma. Materials and Methods Reagents 3-(4,5-dimethylthiazol-2yr)-2-5-diphenyltetrazolium bromide, dimethyl sulfoxide, phorbol 12-myristate 13-acetate (PMA), the rabbit polyclonal HDAC-1 (HDAC1) antibody, and the mouse monoclonal HDAC2 antibody were purchased from Sigma-Aldrich. The rabbit polyclonal antibody to phospho-HDAC2 (Ser394) and the mouse monoclonal antibody to β-actin were obtained from Abcam. Protein A/G plus-agarose immunoprecipitation reagent was obtained from Santa Cruz Biotechnology, Inc. The mouse monoclonal anti-phospho-Akt1/PKBα (Ser473) antibody and the rabbit polyclonal anti-Akt1/PKBα antibody were obtained from Millipore. Recombinant human tumor necrosis factor (TNF)-α was purchased from R&D Systems Europe Ltd.

us-agarose immunoprecipitation reagent was obtained from Santa Cruz Biotechnology, Inc. The mouse monoclonal anti-phospho-Akt1/PKBα (Ser473) antibody and the rabbit polyclonal anti-Akt1/PKBα antibody were obtained from Millipore. Recombinant human tumor necrosis factor (TNF)-α was purchased from R&D Systems Europe Ltd. Patients Nineteen children with severe asthma were recruited for bronchoscopy as part of the workup for severe, therapy-resistant asthma.18 All the children were under regular follow-up at Royal Brompton Hospital. Asthma was diagnosed according to American Thoracic Society criteria, and the severity was defined based on GINA (Global Initiative for Asthma) criteria. All had undergone a detailed evaluation to exclude as far as possible reversible factors such as poor adherence to therapy.19 Subjects were classified into two groups (non-PS and PS). Exposure to PS was assessed on the basis of information reported by parents concerning their smoking habits. Cotinine levels in saliva or urine were measured to support their statements. The study was conducted in accordance with the amended Declaration of Helsinki (http://www.wma.net/en/30publications/ 10policies/b3/) and was approved by the ethics committee of the Royal Brompton and Harefield NHS Trust (Ethics approval number 08/H0708/3). All carers gave written informed consent, with age-appropriate assent from the children.

accordance with the amended Declaration of Helsinki (http://www.wma.net/en/30publications/ 10policies/b3/) and was approved by the ethics committee of the Royal Brompton and Harefield NHS Trust (Ethics approval number 08/H0708/3). All carers gave written informed consent, with age-appropriate assent from the children. Nitric Oxide Measurement Fraction of exhaled nitric oxide (Feno) was measured according to current guidelines.20 A NIOX chemiluminescence analyzer at a flow rate of 50 mL/s was used for analysis of Feno. BAL and Macrophage Processing BAL using fiber-optic bronchoscopy was performed under general anesthetic, as described previously.21 Cells were centrifuged and washed with Hanks’ balanced salt solution. Cytospins were prepared and stained with Diff-Quick for differential cell count. Cell viability was assessed using the Trypan blue exclusion method. BAL macrophages were isolated by plastic adhesion and were incubated in Macrophage Serum Free Medium (Invitrogen Ltd). Cells The human monocytic cell line U937 was purchased from LGC Standards. The cells were differentiated into an adherent macrophage-like morphology by exposure to PMA (50 ng/mL) for 48 h. Cytokine Enzyme-Linked Immunosorbent Assay and Corticosteroid Sensitivity CXCL8 concentrations were determined by sandwich enzyme-linked immunosorbent assay (R&D Systems Europe Ltd). AMs or U937 cells were treated with dexamethasone (10−6 M), followed by TNF-α stimulation (10 ng/mL) for 2 h. The ability of dexamethasone to inhibit TNF-α-induced CXCL8 release was analyzed as a marker of corticosteroid sensitivity.

ere determined by sandwich enzyme-linked immunosorbent assay (R&D Systems Europe Ltd). AMs or U937 cells were treated with dexamethasone (10−6 M), followed by TNF-α stimulation (10 ng/mL) for 2 h. The ability of dexamethasone to inhibit TNF-α-induced CXCL8 release was analyzed as a marker of corticosteroid sensitivity. Thiobarbituric Acid Reactive Substances Assay As a marker of oxidative stress, malondialdehyde (MDA) was measured as thiobarbituric acid reactive substances using a TBARS Assay Kit (Cayman Chemical Company). The levels were calculated using a standard curve. Protein Extraction and Detection Whole cell protein extracts were prepared using a radioimmunoprecipitation assay (RIPA) buffer as described previously.8 Immunoprecipitation was conducted overnight with 2 μg of anti-HDAC2 antibody (Sigma-Aldrich) in RIPA buffer. Cell lysates or immunoprecipitates were analyzed by SDS-PAGE (Invitrogen Ltd) and detected with Western blot analysis by chemiluminescence (ECL Plus; GE Healthcare) as reported previously.22 Total HDAC and HDAC2 Activity To measure in-cell HDAC activity, cells were incubated with Fluor de Lys substrate (200 μM) for 1 h before cell lysis using RIPA buffer. Total HDAC activity was measured using the HDAC Fluorimetric Assay/Drug Discovery Kit (BIOMOL International, Inc). Immunoprecipitated (IP) HDAC2 was resuspended in HDAC assay buffer, and the activity was measured as indicated earlier. HDAC activity was expressed as micromolars of fluorescence standard provided in the kit.

al HDAC activity was measured using the HDAC Fluorimetric Assay/Drug Discovery Kit (BIOMOL International, Inc). Immunoprecipitated (IP) HDAC2 was resuspended in HDAC assay buffer, and the activity was measured as indicated earlier. HDAC activity was expressed as micromolars of fluorescence standard provided in the kit. Immunocytochemistry HDAC2 in BAL macrophage cytospins was incubated with HDAC2 antibody (Santa Cruz Biothechnology, Inc; diluted 1:25). The HDAC2 was visualized using the VECTASTAIN kit (Vector Laboratories) as described previously.8 Statistical Analysis Statistical analysis between the non-PS group and the PS group was conducted with the Mann-Whitney U test. If applicable in the in vitro test, the Student t test was used for the comparison in paired groups. Correlation coefficients were calculated with the use of Spearman’s rank method. The results were expressed as medians (first and third quartiles) or mean ± SEM. GraphPad Prism (GraphPad Software) was used for all statistical analyses, and P values < .05 were considered to be statistically significant.

n paired groups. Correlation coefficients were calculated with the use of Spearman’s rank method. The results were expressed as medians (first and third quartiles) or mean ± SEM. GraphPad Prism (GraphPad Software) was used for all statistical analyses, and P values < .05 were considered to be statistically significant. Results Patient Characterization The characteristics of the patients are summarized in Table 1. All patients were receiving regular inhaled corticosteroids (median dose, 1,600 μg beclomethasone dipropionate equivalent in patients without PS and 1,600 μg in those with PS; P = .78) and a long-acting β2-adrenoceptor agonist. There were no differences in age, sex, FEV1 % predicted, IgE level, atopic status, or use of any other medications between the non-PS and the PS groups. Feno (51.2 parts per billion [ppb] in the non-PS group vs 44.0 ppb in the PS group, P = .60) and the Asthma Control Test (ACT) score (16.0 in the non-PS group vs 11.0 in the PS group, P = .095) were relatively lower in the PS group, but there was no significant difference between the groups. The median number of years of exposure to PS in the PS group was 10.0. Consistent with information reported by the patients’ parents, all subjects in the non-PS group had little or no cotinine detected, whereas the PS group showed relatively higher cotinine levels (> 2 μg/L in urine or > 0.2 μg/L in saliva). Table 1 —Characteristics of Subjects

Results Patient Characterization The characteristics of the patients are summarized in Table 1. All patients were receiving regular inhaled corticosteroids (median dose, 1,600 μg beclomethasone dipropionate equivalent in patients without PS and 1,600 μg in those with PS; P = .78) and a long-acting β2-adrenoceptor agonist. There were no differences in age, sex, FEV1 % predicted, IgE level, atopic status, or use of any other medications between the non-PS and the PS groups. Feno (51.2 parts per billion [ppb] in the non-PS group vs 44.0 ppb in the PS group, P = .60) and the Asthma Control Test (ACT) score (16.0 in the non-PS group vs 11.0 in the PS group, P = .095) were relatively lower in the PS group, but there was no significant difference between the groups. The median number of years of exposure to PS in the PS group was 10.0. Consistent with information reported by the patients’ parents, all subjects in the non-PS group had little or no cotinine detected, whereas the PS group showed relatively higher cotinine levels (> 2 μg/L in urine or > 0.2 μg/L in saliva). Table 1 —Characteristics of Subjects Characteristics Non-Passive Smoking (n = 10) Passive Smoking (n = 9) Age, y 9.5 (8.5-11.5) 10.0 (8.5-13.5) Sex, male (female) 7 (3) 5 (4) Lifetime exposure to passive smoking, y 0 10.0 (8.5-13.5) FEV1 % predicted 71.5 (68.0-78.5) 72.0 (51.5-89.0) Feno,a ppb 51.2 (29.6-73.7) 44.0 (26.9-61.9) ACT score 16.0 (10.5-17.5) 11.0 (8.5-14.5) ≤ 15 5 8 Total IgE, IU/mL 386 (150-1,511) 355 (158-1,462) Atopyb 8c 7 Medication ICS,d μg 1,600 (1,300-2,000) 1,600 (1,200-2,000) LABA 10 9 LTRA 9 6 Theophylline 2 2 Data are presented as median (first and third quartile) or No. unless indicated otherwise. There were no significant differences between the groups. ACT = Asthma Control Test; Feno = fraction of exhaled nitric oxide; ICS = inhaled corticosteroid; IU = International Units; LABA = long-acting β2-adrenoceptor agonist; LTRA = leukotriene receptor antagonist; ppb = parts per billion.

. unless indicated otherwise. There were no significant differences between the groups. ACT = Asthma Control Test; Feno = fraction of exhaled nitric oxide; ICS = inhaled corticosteroid; IU = International Units; LABA = long-acting β2-adrenoceptor agonist; LTRA = leukotriene receptor antagonist; ppb = parts per billion. a Values in non-passive smoking represent medians of nine subjects. b Atopy was defined as at least one positive specific IgE radioallergosorbent assay (≥ 0.34 kU/L). c Values in non-passive smoking represent No. out of nine subjects. d Beclomethasone dipropionate equivalent dose. BAL Analysis There was no difference in total BAL cell counts or in numbers of macrophages and lymphocytes between the PS and the non-PS group. Subjects in the PS group had a relatively higher number of neutrophils than did non-PS group subjects, who had more eosinophils in BAL fluid (BALF) (Table 2). In line with the increase of neutrophils, higher levels of CXCL8 were found in BALF from the subjects in the PS group (Table 2), and furthermore, there was a positive correlation between CXCL8 concentrations and the percentage of neutrophils (r = 0.71, P < .001). Table 2 —Cell and CXCL8 Analysis in BAL

BAL Analysis There was no difference in total BAL cell counts or in numbers of macrophages and lymphocytes between the PS and the non-PS group. Subjects in the PS group had a relatively higher number of neutrophils than did non-PS group subjects, who had more eosinophils in BAL fluid (BALF) (Table 2). In line with the increase of neutrophils, higher levels of CXCL8 were found in BALF from the subjects in the PS group (Table 2), and furthermore, there was a positive correlation between CXCL8 concentrations and the percentage of neutrophils (r = 0.71, P < .001). Table 2 —Cell and CXCL8 Analysis in BAL Readout Non-Passive Smoking (n = 10) Passive Smoking (n = 9) P Value Cell Count, × 103/mL % of Total Cells Cell Count, × 103/mL % of Total Cells Cell Count, × 103/mL % of Total Cells Total 370 (265-553) … 280 (205-403) … .2110 … Macrophage 287 (196-389) 76.2 (70.5-78.0) 236 (326-162) 82.8 (81.3-85.9) .4967 .0030 Neutrophil 13.3 (9.2-23.1) 3.7 (3.0-4.5) 24.1 (12.7-34.4) 8.3 (6.5-9.5) .1333 .0021 Lymphocyte 55.2 (33.3-80.9) 14.7 (10.7-18.1) 10.9 (7.7-38.1) 5.0 (3.7-8.8) .0101 < .001 Eosinophil 11.6 (5.8-45.1) 4.0 (1.2-10.4) 2.9 (1.0-9.6) 1.6 (0.5-3.1) .0435 .1333 CXCL8, pg/μg protein 0.11 (0.05-0.24) … 0.78 (0.27-1.40) … .0021 … Data are presented as median (first and third quartile).

8.3 (6.5-9.5) .1333 .0021 Lymphocyte 55.2 (33.3-80.9) 14.7 (10.7-18.1) 10.9 (7.7-38.1) 5.0 (3.7-8.8) .0101 < .001 Eosinophil 11.6 (5.8-45.1) 4.0 (1.2-10.4) 2.9 (1.0-9.6) 1.6 (0.5-3.1) .0435 .1333 CXCL8, pg/μg protein 0.11 (0.05-0.24) … 0.78 (0.27-1.40) … .0021 … Data are presented as median (first and third quartile). HDAC2 Expression and Activity In AMs, HDAC2 protein level normalized to β actin protein expression was significantly lower in the PS group subjects compared with the non-PS group subjects although HDAC1 protein expression normalized to β actin protein was not different between PS and non-PS groups (Figs 1A, 1B). This reduction of HDAC2 was confirmed by immunocytochemistry (Fig 1C) because positive brown signals in AMs were reduced in the PS group. Although there was no significant difference in total HDAC activities in AMs between the two groups (median [first and third quartile]: 30.6 [10.8-45.0] arbitrary fluorescence units [AFU]/μg in the non-PS group vs 22.4 [13.8-28.6] AFU/μg in the PS group), the activity of IP-HDAC2 was significantly reduced in the PS group compared with the non-PS group (5.8 [4.5-8.8] AFU/μg in the non-PS group vs 3.7 [1.4-5.4] AFU/μg in the PS group) (Figs 1D, 1E). Moreover, there was a positive correlation between HDAC2 expression and IP-HDAC2 activity (r = 0.59, P = .0072) (Fig 1F), suggesting that the reduction in IP-HDAC2 activity was caused by the reduction of HDAC2 protein expression.

.5-8.8] AFU/μg in the non-PS group vs 3.7 [1.4-5.4] AFU/μg in the PS group) (Figs 1D, 1E). Moreover, there was a positive correlation between HDAC2 expression and IP-HDAC2 activity (r = 0.59, P = .0072) (Fig 1F), suggesting that the reduction in IP-HDAC2 activity was caused by the reduction of HDAC2 protein expression. Figure 1. Effects of PS on HDAC2 protein expression and activity in alveolar macrophages from children with severe asthma. A and B, Western blotting analysis of HDAC2 and HDAC1 protein expression, respectively, normalized to β-actin expression. C, HDAC2 protein expression detected by immunocytochemistry (original magnification × 400). Results were representative of at least seven subjects in each group. D, Total in-cell HDAC activity. E, Immunoprecipitated HDAC2 activity. F, Correlation between HDAC2 protein expression and activity. Values in A, B, D, and E represent mean ± SEM of 10 non-PS group subjects or nine PS group subjects. AFU = arbitrary fluorescence units; HDAC = histone deacetylase; PS = passive smoking.

. D, Total in-cell HDAC activity. E, Immunoprecipitated HDAC2 activity. F, Correlation between HDAC2 protein expression and activity. Values in A, B, D, and E represent mean ± SEM of 10 non-PS group subjects or nine PS group subjects. AFU = arbitrary fluorescence units; HDAC = histone deacetylase; PS = passive smoking. Akt1 and HDAC2 Phosphorylation As shown in Figure 2A, HDAC2 was highly phosphorylated in AMs of the PS group compared with the non-PS group. The levels of phosphorylation of Akt1, a surrogate marker of PI3K signaling activation, were also significantly increased by 2.4 fold in AMs of the PS group compared with the non-PS group (Fig 2B) when normalized to total Akt1 protein expression. Furthermore, Akt1 phosphorylation levels were positively correlated with HDAC2 phosphorylation levels (r = 0.71, P < .001) (Fig 2C) and negatively correlated with HDAC2 activity (r = −0.54, P = .018) (Fig 2D). Figure 2. Effects of PS on phosphorylation level of Akt1 and HDAC2 in alveolar macrophages from children with severe asthma. A and B, Phosphorylation levels of HDAC2-Ser394 and Akt1, respectively, normalized to total HDAC2 and Akt1 expression. Values represent means of 10 (in non-PS group) or nine (in PS group) subjects ± SEM; C, Correlation between HDAC2 and Akt1 phosphorylation levels. D, Correlation between immunoprecipitated HDAC2 activity and phosphorylation level of Akt1. See Figure 1 legend for expansion of abbreviations.

al HDAC2 and Akt1 expression. Values represent means of 10 (in non-PS group) or nine (in PS group) subjects ± SEM; C, Correlation between HDAC2 and Akt1 phosphorylation levels. D, Correlation between immunoprecipitated HDAC2 activity and phosphorylation level of Akt1. See Figure 1 legend for expansion of abbreviations. Corticosteroid Sensitivity of CXCL8 Production in AMs Dexamethasone at 10−6 M significantly inhibited TNF-α-induced CXCL8 production by 40% in AMs from subjects in the non-PS group. However, there was no significant inhibition in AMs from subjects in the PS group (Fig 3A). Figure 3. Effects of PS on corticosteroid sensitivity. A, Effects of Dex (10−6 M) on TNF-α-induced CXCL8 release in alveolar macrophages. B-D, Effects of BAL supernatant on (B) Akt1 phosphorylation levels, (C) HDAC2-Ser394 phosphorylation, and (D), immunoprecipitated HDAC2 activity. E, Effects of Dex on TNF-α-induced CXCL8 release. PMA-differentiated U937 cells were preincubated overnight with BAL fluid (BALF) (adjusted to 2 μg/mL of protein). F, MDA levels in BALF from non-PS and PS group subjects. Dex = dexamethasone; MDA = malondialdehyde; PMA = phorbol 12-myristate 13-acetate; TNF = tumor necrosis factor. See Figure 1 legend for expansion of other abbreviations.

937 cells were preincubated overnight with BAL fluid (BALF) (adjusted to 2 μg/mL of protein). F, MDA levels in BALF from non-PS and PS group subjects. Dex = dexamethasone; MDA = malondialdehyde; PMA = phorbol 12-myristate 13-acetate; TNF = tumor necrosis factor. See Figure 1 legend for expansion of other abbreviations. Next, PMA-differentiated macrophage-like U937 cells were exposed to BALF obtained from the PS group or the non-PS group. BALF from the PS group significantly increased Akt1 phosphorylation (Fig 3B) and HDAC2 phosphorylation (Fig 3C), and reduced IP-HDAC2 activity (336.6 [310.1-358.4] AFU/μg in the non-PS group, 299.1 [277.0-316.7] AFU/μg in the PS group) in U937 cells (Fig 3D). Pretreatment of dexamethasone (10−6 M) completely inhibited TNF-α-induced CXCL8 release in U937 exposed to BALF from subjects in the non-PS group, but the inhibitory action was limited in U937 exposed to BALF from PS group subjects (Fig 3E). Furthermore, the level of MDA, a marker of oxidative stress, was found to be significantly higher in BALF in the PS group than in the non-PS group (Fig 3F).

8 release in U937 exposed to BALF from subjects in the non-PS group, but the inhibitory action was limited in U937 exposed to BALF from PS group subjects (Fig 3E). Furthermore, the level of MDA, a marker of oxidative stress, was found to be significantly higher in BALF in the PS group than in the non-PS group (Fig 3F). Discussion This article reports that children with asthma who are passively exposed to tobacco smoke had the same molecular abnormalities leading to in vitro steroid resistance as do adults who actively smoke.8 Children with severe asthma who were exposed to PS had higher CXCL8 levels (Table 2), in agreement with previous reports,14,23 and a higher neutrophil count and percentage in BALF (Table 2), and the neutrophil percentage of total cell counts positively correlated with CXCL8 concentration. Even more importantly, AMs from PS-exposed children with asthma exhibited corticosteroid insensitivity in the in vitro assay, manifest by the reduced effect of dexamethasone in suppressing TNF-α-induced CXCL8 production compared with children not exposed to PS with uncontrolled severe asthma. The prescribed corticosteroid dose was the same in the PS and the non-PS groups; therefore, inhaled corticosteroids were not a potential modifier of this response. To our knowledge, this is the first evidence that PS affects corticosteroid sensitivity in resident lung cells in children with severe asthma in vitro. Because there is no agreed-upon definition of steroid sensitivity in children with asthma, we did not compare the in vitro steroid sensitivity and clinical response to steroids, and, thus, the clinical significance of our findings is speculative. However, we have shown previously that responses to steroids are reduced in children exposed to PS.24

pon definition of steroid sensitivity in children with asthma, we did not compare the in vitro steroid sensitivity and clinical response to steroids, and, thus, the clinical significance of our findings is speculative. However, we have shown previously that responses to steroids are reduced in children exposed to PS.24 A reduction in HDAC activity, especially HDAC2, has been reported to cause corticosteroid insensitivity.8,25 We found that PS is associated with a lower expression of HDAC2, but not HDAC1, in AMs (Figs 1A‐C). AMs in BALF and peripheral lung tissue from patients with COPD showed lower levels of HDAC2 expression with increasing disease severity; however, HDAC1 expression was not altered.15 Furthermore, in AMs from healthy cigarette smokers, HDAC2, but not HDAC1, expression was significantly decreased compared with those from healthy nonsmokers,8 although in bronchial biopsy specimens obtained from adults with mild asthma, the expression levels of both HDAC1 and HDAC2 are slightly reduced.11,12 These findings suggest that HDAC2 may be more sensitive to oxidative stress, especially cigarette smoking, than is HDAC1, as we reported previously.25,26

hy nonsmokers,8 although in bronchial biopsy specimens obtained from adults with mild asthma, the expression levels of both HDAC1 and HDAC2 are slightly reduced.11,12 These findings suggest that HDAC2 may be more sensitive to oxidative stress, especially cigarette smoking, than is HDAC1, as we reported previously.25,26 The molecular mechanism of HDAC2 reduction has been studied extensively. Defects of HDAC2 were induced by oxidative stress, leading to nitration and subsequent ubiquitination of HDAC2,26 or carboxylation/oxidation.27,28 A well-documented mechanism is PI3K-dependent phosphorylation of HDAC2. Akt is phosphorylated through the PI3Kδ pathway during oxidative stress,9,17 and activated Akt-dependent phosphorylation, mainly at Ser394, Ser422, and Ser424, may cause degradation and inactivation of HDAC2.29‐31 In line with these previous reports, we found that PS increased phosphorylation of Akt1 (Fig 2B) and HDAC2-Ser394 (Fig 2A). Akt1 phosphorylation was positively correlated with HDAC2 phosphorylation and negatively correlated with HDAC2 activity in AMs from children with severe asthma (Figs 2C, 2D). These observations seem to be driven by the presence of a single outlier, but when the outlier was removed, pAkt1 was still correlated with pHDAC2 (r = 0.66, P = .0027). Thus, PI3K signaling activation by PS is likely one of the major causes of HDAC2 downregulation in children with severe asthma

e asthma (Figs 2C, 2D). These observations seem to be driven by the presence of a single outlier, but when the outlier was removed, pAkt1 was still correlated with pHDAC2 (r = 0.66, P = .0027). Thus, PI3K signaling activation by PS is likely one of the major causes of HDAC2 downregulation in children with severe asthma Oxidative stress, such as from tobacco smoke, may be an important factor in inducing corticosteroid insensitivity. MDA is a product of lipid oxidation and an oxidative stress biomarker.32 Previous reports showed that MDA was increased in several biologic fluids with exposure to cigarette smoke.27,33,34 In this study, we also found a higher level of MDA in BALF from PS group subjects (Fig 3F). Importantly, when BALF from PS-exposed children was added to the macrophage-type U937 cell line, the cells became corticosteroid insensitive, with increased phosphorylated Akt/HDAC2 and reduced HDAC2 activity. Thus, high levels of oxidative stress and their end products may induce corticosteroid insensitivity with dysfunction of HDAC2. In contrast, the Feno level was relatively lower in the PS group than in the non-PS group, which is consistent with previous findings in adults.35

rylated Akt/HDAC2 and reduced HDAC2 activity. Thus, high levels of oxidative stress and their end products may induce corticosteroid insensitivity with dysfunction of HDAC2. In contrast, the Feno level was relatively lower in the PS group than in the non-PS group, which is consistent with previous findings in adults.35 In our study, more children with severe asthma who were exposed to PS had ACT scores of ≤ 15 when compared with patients with severe asthma who were not exposed to PS, indicating that the asthma of PS-exposed children was relatively poorly controlled, as reported previously.36 However, there were no significant differences in ACT scores between the groups, probably because of the relatively small number of patients. To strengthen our findings, more clinical outcomes are needed (eg, the effect of avoidance of PS in a longitudinal and much bigger group of children with uncontrolled severe asthma). Conclusions We have demonstrated the molecular and cellular basis of an important adverse effect of PS exposure in children with severe asthma. PS exposure impairs HDAC2 function via PI3K activation, which may contribute to a more steroid-resistant phenotype. Clearly, the avoidance of PS exposure is of paramount importance in all children, particularly those with severe asthma.5 This study underscores on a molecular level the harm done to asthmatic children by parents who smoke. Author contributions: Drs Kobayashi and Ito had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Conclusions We have demonstrated the molecular and cellular basis of an important adverse effect of PS exposure in children with severe asthma. PS exposure impairs HDAC2 function via PI3K activation, which may contribute to a more steroid-resistant phenotype. Clearly, the avoidance of PS exposure is of paramount importance in all children, particularly those with severe asthma.5 This study underscores on a molecular level the harm done to asthmatic children by parents who smoke. Author contributions: Drs Kobayashi and Ito had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Dr Kobayashi: contributed to the conception and design of the study, conception of the experiments, analysis and interpretation of the data, drafting of the manuscript, review of the report, and approval of the final version. Dr Bossley: contributed to patient enrollment, conception of the experiments, analysis and interpretation of the data, review of the report, and approval of the final version. Dr Gupta: contributed to patient enrollment, conception of the experiments, analysis and interpretation of the data, review of the report, and approval of the final version. Dr Akashi: contributed to the conception of the experiments, review of the report, and approval of the final version. Dr Tsartsali: contributed to patient enrollment, conception of the experiments, analysis and interpretation of the data, review of the report, and approval of the final version.

Dr Gupta: contributed to patient enrollment, conception of the experiments, analysis and interpretation of the data, review of the report, and approval of the final version. Dr Akashi: contributed to the conception of the experiments, review of the report, and approval of the final version. Dr Tsartsali: contributed to patient enrollment, conception of the experiments, analysis and interpretation of the data, review of the report, and approval of the final version. Dr Mercado: contributed to the conception of the experiments, analysis and interpretation of the data, review of the report, and approval of the final version. Dr Barnes: contributed to the conception and design of the study, drafting of the manuscript, review of the report, and approval of the final version. Dr Bush: contributed to the conception and design of the study, patient enrollment, drafting of the manuscript, review of the report, and approval of the final version. Dr Ito: contributed to the conception and design of the study, conception of the experiments, analysis and interpretation of the data, drafting of the manuscript, review of the report, and approval of the final version.

Dr Bush: contributed to the conception and design of the study, patient enrollment, drafting of the manuscript, review of the report, and approval of the final version. Dr Ito: contributed to the conception and design of the study, conception of the experiments, analysis and interpretation of the data, drafting of the manuscript, review of the report, and approval of the final version. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Barnes has served on scientific advisory boards for AstraZeneca; Boehringer-Ingelheim; Chiesi Pharmaceuticals; Daiichi Sankyo, Inc; GlaxoSmithKline; Novartis; Nycomed; Pfizer Inc; RespiVert; Teva Pharmaceutical Industries, Ltd; and UCB and has received research funding from Aquinox Pharmaceuticals; AstraZeneca; Boehringer-Ingelheim; Chiesi Pharmaceuticals; Daiichi-Sankyo, Inc; GlaxoSmithKline; Novartis; Nycomed; Pfizer Inc; and Prosonix. Dr Ito is currently an employee of RespiVert and has an honorary contract with Imperial College. Drs Kobayashi, Bossley, Gupta, Akashi, Tsartsali, Mercado, and Bush have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: The sponsors had no role in the design of the study, the collection and analysis of the data, or in the preparation of the manuscript.

Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Barnes has served on scientific advisory boards for AstraZeneca; Boehringer-Ingelheim; Chiesi Pharmaceuticals; Daiichi Sankyo, Inc; GlaxoSmithKline; Novartis; Nycomed; Pfizer Inc; RespiVert; Teva Pharmaceutical Industries, Ltd; and UCB and has received research funding from Aquinox Pharmaceuticals; AstraZeneca; Boehringer-Ingelheim; Chiesi Pharmaceuticals; Daiichi-Sankyo, Inc; GlaxoSmithKline; Novartis; Nycomed; Pfizer Inc; and Prosonix. Dr Ito is currently an employee of RespiVert and has an honorary contract with Imperial College. Drs Kobayashi, Bossley, Gupta, Akashi, Tsartsali, Mercado, and Bush have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: The sponsors had no role in the design of the study, the collection and analysis of the data, or in the preparation of the manuscript. Other contributions: We are grateful to all the patients and parents for agreeing to take part in our study. We gratefully acknowledge the following people for their invaluable help performing the bronchoscopies: S. Saglani, MD; M. Rosenthal, MD; I. Balfour-Lynn, MD; C. Hogg, MD; and J. Davies, MD. This study was presented previously at the American Thoracic Society 2011 International Conference, May 13-18, 2011, Denver, CO.

Other contributions: We are grateful to all the patients and parents for agreeing to take part in our study. We gratefully acknowledge the following people for their invaluable help performing the bronchoscopies: S. Saglani, MD; M. Rosenthal, MD; I. Balfour-Lynn, MD; C. Hogg, MD; and J. Davies, MD. This study was presented previously at the American Thoracic Society 2011 International Conference, May 13-18, 2011, Denver, CO. Funding/Support: This study was funded by the Wellcome Trust [WHRD_P31768 to Dr Barnes], London, England, and by the National Institute of Health Research Respiratory Disease Biomedical Research Unit at the Royal Brompton and Harefield NHS Foundation Trust and Imperial College London (to Drs Bossley, Gupta and Bush). Dr Gupta is also the recipient of a British Medical Association, James Trust Fellowship. This is a Wellcome-Trust-compliant open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/). Abbreviations ACTAsthma Control Test AFUarbitrary fluorescence units AMalveolar macrophage BALFBAL fluid Fenofraction of exhaled nitric oxide HDAChistone deacetylase IPimmunoprecipitated MDAmalondialdehyde PI3Kphosphoinositide-3-kinase PMAphorbol 12-myristate 13-acetate ppbparts per billion PSpassive smoking RIPAradioimmunoprecipitation assay TNFtumor necrosis factor

Approximately 30% to 50% of patients with heart failure (HF) have a preserved left ventricular ejection fraction (HFpEF), a condition associated with dismal prognosis1 and for which there are still no proven therapies to improve outcomes.2 Several pathophysiologic mechanisms have been demonstrated in HFpEF, including left ventricular (LV) diastolic dysfunction,3 arterial stiffening, and abnormalities of LV-arterial coupling,4,5 which may compromise LV energetic efficiency. With longstanding pulmonary venous congestion, the majority of patients with HFpEF develop increased pulmonary arterial pressures and pulmonary hypertension (PH). PH has been shown to be a major determinant of mortality in this population and represents a potential novel therapeutic target in HFpEF.6,7 HF- and PH-related diseases are characterized by endothelial dysfunction.8‐12 Preclinical and small clinical studies suggest that deficient nitric oxide-soluble guanylate cyclase-cyclic guanosine monophosphate (NO-sGC-cGMP) signaling is involved in impaired cardiac relaxation/distensibility.13 Furthermore, low cyclic guanosine monophosphate (cGMP) levels in myocardial tissue may underlie abnormal cardiomyocyte function.14 Thus, targeting the NO-sGC-cGMP signaling pathway may be a promising approach for the treatment of HFpEF with PH.

GMP) signaling is involved in impaired cardiac relaxation/distensibility.13 Furthermore, low cyclic guanosine monophosphate (cGMP) levels in myocardial tissue may underlie abnormal cardiomyocyte function.14 Thus, targeting the NO-sGC-cGMP signaling pathway may be a promising approach for the treatment of HFpEF with PH. Riociguat is a novel soluble guanylate cyclase (sGC) stimulator15 with a dual mode of action, sensitizing sGC to endogenous nitric oxide (NO) and directly stimulating sGC independent of NO.16 Riociguat induces vasodilation and has antifibrotic, antiproliferative, and antiinflammatory effects.15‐18 In clinical studies, riociguat has proven efficacy in pulmonary arterial hypertension and chronic thromboembolic PH.19,20 In a randomized, placebo-controlled phase 2b study in patients with HF and PH due to systolic LV dysfunction (Left Ventricular Systolic Dysfunction Associated With Pulmonary Hypertension Riociguat Trial [LEPHT]), riociguat was well tolerated and improved cardiac index, pulmonary (PVR) and systemic vascular resistance (SVR), as well as quality of life, without significantly changing mean pulmonary artery pressure (mPAP, primary end point) or systolic BP.21 In the current study (Acute Hemodynamic Effects of Riociguat in Patients With Pulmonary Hypertension Associated With Diastolic Heart Failure [DILATE-1]), we aimed to characterize the hemodynamic effects, safety, and pharmacokinetics of single oral doses of riociguat in patients with HFpEF and PH.