Abstract Rationale Chronic obstructive pulmonary disease (COPD) is characterized by molecular heterogeneity, involving both transcriptomic and metabolic dysregulation. We previously identified a blood transcriptomic gene signature associated with lung function decline using longitudinal RNA-sequencing data from COPDGene. In this study, we integrated that gene signature with plasma metabolomic profiles to uncover metabolic pathways mediating lung function decline and other COPD-related phenotypes. Methods Phase 2 data from COPDGene participants with both RNA-sequencing and metabolomic profiles (n = 3, 702) were analyzed. Gene expression was normalized using trimmed mean of M-values (TMM) and transformed into a ranked-based single-sample gene scores using singscore, which quantifies coordinated up- or down-regulation of genes within a predefined signature for each participant. Associations between the lung function decline gene signature and 1, 028 plasma metabolites were tested using limma, adjusting for age, sex, race, BMI, smoking status, pack-years, and white blood cell proportions. A weighted metabolite signature score (MetSigWeighted) was constructed from the log fold-change (logFC) values of significant metabolites. Weighted gene co-expression network analysis (WGCNA) was applied to the 501 significant metabolites to identify co-regulated metabolic modules, followed by covariate-adjusted associations with COPD-related phenotypes. Results We identified 501 metabolites significantly associated with the transcriptomic signature (FDR 0. 05). The MetSigWeighted score was significantly associated with FEV1 decline between Phase 2 and Phase 3 (β = 10. 7 ml, P = 0. 038) and showed robust associations with exacerbation frequency (P = 7. 8 × 10−6), percent emphysema (P = 2. 7 × 10−19), airway wall thickness (P = 1. 1 × 10−12), chronic bronchitis (P = 1. 0 × 10−4), and impaired gas exchange (P ≤ 1. 6 × 10−12). WGCNA generated five metabolomic modules, each showing distinct associations with COPD traits. The brown module was significantly enriched for glycerophospholipid metabolism, phospholipid biosynthesis, and sphingolipid metabolism (FDR 0. 004), correlating strongly with emphysema, airway remodeling, and DLCO. Other modules captured complementary biology, including androgenic and xenobiotic metabolism, fatty acid oxidation, and amino acid metabolism, reflecting broader metabolic perturbations in COPD. These findings are consistent with prior metabolomic and lipidomic studies linking phospholipid and sphingolipid dysregulation to COPD pathogenesis and lung function decline, strengthening the biological validity of our results. Conclusions Our integrative gene-to-metabolite-to-phenotype framework identified coherent phospholipid and sphingolipid remodeling networks linking transcriptomic activity to COPD progression across multiple omic layers. Our findings highlight a conserved metabolic axis underlying disease progression and enhance biological specificity compared to direct metabolite-phenotype models. This abstract is funded by: NHLBI grants U01 HL089897 and U01 HL089856 and by NIH contract 75N92023D00011
Elhusseini et al. (Fri,) studied this question.