Biomarkers Predictive of Exacerbations in the SPIROMICS and COPDGene Cohorts
Rationale: Chronic obstructive pulmonary disease exacerbations are associated with disease progression, higher healthcare cost, and increased mortality. Published predictors of future exacerbations include previous exacerbation, airflow obstruction, poor overall health, home oxygen use, and gastroesophageal reflux.Objectives: To determine the value of adding blood biomarkers to clinical variables to predict exacerbations.Methods: Subjects from the SPIROMICS (Subpopulations and Intermediate Outcomes Measures in COPD Study) (n = 1,544) and COPDGene (Genetic Epidemiology of COPD) (n = 602) cohorts had 90 plasma or serum candidate proteins measured on study entry using Myriad-RBM multiplex panels. We defined total exacerbations as subject-reported worsening in respiratory health requiring therapy with corticosteroids and/or antibiotics, and severe exacerbations as those leading to hospitalizations or emergency room visits. We assessed retrospective exacerbations during the 12 months before enrollment and then documented prospective exacerbations in each cohort. Exacerbations were modeled for biomarker associations with negative binomial regression including clinical covariates (age, sex, percent predicted FEV1, self-reported gastroesophageal reflux, St. George's Respiratory Questionnaire score, smoking status). We used the Stouffer-Liptak test to combine P values for metaanalysis.Measurements and Main Results: Between the two cohorts, 3,471 total exacerbations (1,044 severe) were reported. We identified biomarkers within each cohort that were significantly associated with a history of exacerbation and with a future exacerbation, but there was minimal replication between the cohorts. Although established clinical features were predictive of exacerbations, of the blood biomarkers only decorin and α2-macroglobulin increased predictive value for future severe exacerbations.Conclusions: Blood biomarkers were significantly associated with the occurrence of exacerbations but were not robust between cohorts and added little to the predictive value of clinical covariates for exacerbations.