Skip to main content
NCBI home page
Springer logoLink to Springer
. 2026 Aug 22;22(5):142. doi: 10.1007/s11306-026-02508-3

Metabolomic profiling reveals distinct and shared biology in COPD and IPF

Aldric Rosario 1,2,#, Nicole Prince 1,2,#, Shariq Madha-Krause 3, Sharon M Lutz 4, Julian E Hecker 1,2, Jeong H Yun 1,2,5, Michael H Cho 1,2,5, Arda Halu 1,2, George R Washko 2,6, Frank C Sciurba 7, Lucas Barwick 8, Andrew H Limper 9, Kevin R Flaherty 10, Gerard J Criner 11, Kevin K Brown 12, Robert Wise 13, Fernando J Martinez 14, Russell P Bowler 15, Robert E Gerszten 16,17,18, Clary B Clish 18, Edwin K Silverman 1,2,5,#, Rachel S Kelly 1,2,✉,#
PMCID: PMC13499876  PMID: 42631875

Abstract

Introduction

Chronic Obstructive Pulmonary disease (COPD) and Idiopathic Pulmonary Fibrosis (IPF) are chronic pulmonary disorders with distinct pathologies but shared risk factors. Metabolomics may provide insights into mechanisms.

Objectives

To identify metabolites associated with COPD and IPF, and to characterize shared and disease-specific signatures.

Methods

Plasma metabolomic profiling was conducted in the Lung Tissue Research Consortium (LTRC). Logistic regression identified metabolites associated with COPD and IPF, and results were replicated in an external cohort, COPDGene. We applied Weighted Gene Co-expression Network Analysis (WGCNA) to explore disease-associated metabolite modules. We further evaluated relationships between significant metabolites and risk genes of interest.

Results

Of 1131 metabolites in LTRC, 246 (21.8%) differed between COPD and controls, and 136 (12.0%) between IPF and controls (FDR < 0.05). Among 80 shared significant metabolites in COPD and IPF, 77 showed concordant directions of effect. Shared metabolomic changes included reduced levels of steroids, triglycerides, diglycerides, phosphatidylcholines, and increased levels of carnitines. In contrast, polyunsaturated fatty acids, nicotine, and thyroxine metabolites differed between COPD and IPF; these findings were further explored by the WGCNA. External replication was performed for 120 metabolites measured in both cohorts, of which 49 (40.8%) replicated in COPD vs. control. In exploratory analyses leveraging quantitative imaging abnormalities (QIA) as a surrogate for IPF; only 4 (3.3%) metabolites replicated in the QIA vs. control model, and only 9 (7.5%) for COPD vs. QIA.

Conclusion

Alterations in metabolomic profiles of COPD and IPF suggested shared dysregulation of several lipids. However, some metabolites pointed to disease-specific differences.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11306-026-02508-3.

Keywords: COPD, IPF, Metabolomics, Lipid dysregulation, PUFA metabolism

Introduction

Chronic obstructive pulmonary disease (COPD) and Idiopathic Pulmonary Fibrosis (IPF) are chronic lung diseases that impart significant global public health burdens (Maher et al., 2021; Momtazmanesh et al., 2023). COPD is a common and heterogeneous syndrome that includes small airway disease and emphysema. IPF, though less common, is the most prevalent form of interstitial lung disease (ILD) and is marked by a restrictive ventilatory defect and extensive lung fibrosis (Chilosi et al., 2012). Despite their widely divergent pathological manifestations, these conditions share some clinical and physiological features, including progressive decline in lung function, resulting in reduced oxygen delivery to proximal tissue (Beghé et al., 2021), and several overlapping risk factors, including aging and smoking (Beghé et al., 2021). However, less is known about the underlying mechanisms contributing to the shared and distinct etiologies and pathobiology of COPD and IPF.

Metabolomics, the large-scale study of small molecules in a biological system, offers a powerful approach to capture biochemical phenomena underlying diseases. Measuring circulating metabolites could reveal novel biology to improve understanding of COPD and IPF pathogenesis (Clish, 2015) as the metabolome captures the downstream products of genetic, transcriptomic, and proteomic processes in addition to exogenous metabolites reflective of diet, medication, microbiome, lifestyle, and environmental exposures. Metabolomic profiling therefore provides a unique opportunity to assess the contributions of the shared genetic and environmental risk factors to COPD and IPF (Berry et al., 2012; Summer et al., 2024). Of specific interest are five genetic loci— located near, CRHR1, DSP, FAM13A, STN1, and ZKSCAN1—implicated in both diseases by genome-wide association studies, but demonstrating discordant directions of effect (Cho et al., 2022; Moss & Rosas, 2023; Sakornsakolpat et al., 2019). While previous studies have explored metabolomic profiles of patients with COPD and IPF (Nambiar et al., 2021; Summer et al., 2024; Tirelli et al., 2024; Yan et al., 2017), no comprehensive untargeted metabolomics study to date has directly compared metabolomic profiles of patients with COPD and IPF to each other and to controls.

In this study, we compared plasma metabolomic profiles of adults with COPD, IPF, and controls within the Lung Tissue Research Consortium (LTRC) (Berry et al., 2012) as part of the Trans-Omics for Precision Medicine (TOPMed) initiative, aiming to identify shared and distinct metabolomic pathways associated with the diseases. We attempted replication in an independent population, the COPDGene cohort (Regan et al., 2011). We further leveraged available DNA sequencing data to assess whether metabolomics could elucidate the discordant effects of five genetic loci of interest.

Methods

Lung Tissue Research Consortium (LTRC)

Blood samples in LTRC were collected from well-phenotyped patients prior to thoracic surgery in one of four clinical centers (Berry et al., 2012; Yang et al., 2014). Phenotypic data was collected through clinical information, questionnaires, and lung function tests. COPD cases were defined by moderate-to-severe airflow limitation (Forced expiratory volume [FEV1] < 80% predicted and FEV1/forced vital capacity [FVC] < 0.7), and IPF was defined by an expert summative clinical diagnosis. Controls were defined by having normal spirometry (FEV1 ≥ 80% predicted and FEV1/FVC ≥ 0.7), without a clinical diagnosis of IPF or sarcoidosis or ILD (See Supplementary Methods). IRB approval and written consent were obtained (2018P000186 [Network Analysis of Multiple Omics Data in the Lung Tissue Research Consortium])

Plasma metabolomic profiling was conducted using four complementary liquid chromatography tandem mass spectrometry (LC-MS) platforms by the Broad Institute as part of the TOPMed initiative (Taliun et al., 2021). Metabolites with percentage coefficient of variation (CV%) > 25% in the pooled QC samples or those missing in greater than 75% of samples were excluded from analysis. Remaining missing values were imputed using half the minimum observed value. All metabolites were log-10 transformed, and metabolites from the untargeted platforms were pareto-scaled. Metabolites missing confident identification were removed. Further details of metabolomics data acquisition and processing can be found in the Supplementary Methods.

COPDGene

COPDGene is a prospective observational study of > 10,000 former and current smokers aged 45–80 years with at least 10 pack-year smoking history (Regan et al., 2011). We used the questionnaires, lung function tests, and blood samples collected at the five-year follow-up visit. Controls were defined by the absence of airflow obstruction, emphysema, and quantitative interstitial abnormalities, QIA (Choi et al., 2023). COPD was defined using post-bronchodilator ratio of FEV1/FVC < 0.7. Patients with IPF were excluded from COPDGene, therefore evidence of QIA was used as a proxy for IPF. QIA are parenchymal changes detected via chest computed tomography (CT) and can be considered indicators of mild or early pulmonary fibrosis, however we note this is a suboptimal proxy for IPF, and as such replication leveraging QIA should be considered exploratory. Informed consent was obtained from all individuals participating in this study. Plasma metabolomic data was generated by Metabolon, Inc., as described previously (Gillenwater et al., 2020) (Supplementary Methods); metabolomic data underwent the same QC procedures as LTRC metabolomic data. IRB approval and written consent were obtained (COPDGene: 2007P000554 [Genetic Epidemiology of COPD])

Statistical analysis

Differences in baseline characteristics between COPD, IPF, and control participants were assessed using Kruskal-Wallis or chi-squared tests, as appropriate. Logistic regression models estimated associations between metabolite levels and disease status after adjusting for age in years, sex, BMI, and self-reported smoking status (current vs. never/former). Correction for False Discovery Rate (FDR) was performed using the Benjamini-Hochberg procedure.

We explored the features of disease-associated lipid metabolites, specifically with respect to carbon chain length and degree of unsaturation (i.e., number of double bonds). Metabolites that were putatively annotated to two possible compounds were excluded from carbon chain length/saturation analysis. We estimated associations between metabolites and SNPs mapping to five genes previously shown to have opposite effects in COPD and IPF (FAM13, DSP, ZKSCAN1, STN1, CRHR1) (See Supplementary Methods).

We additionally performed sensitivity analyses of metabolites associated with disease using a binary variable for current smoking status derived from plasma cotinine levels to account for participants with missing smoking status. Post-QC cotinine values were binarized using a threshold of 1 (based in Supplementary Fig. 1), with values ≥ 1 classified as current smokers and < 1 classified as non-smokers. We further explored metabolites associated with cotinine-based smoking status. We used binarized cotinine values as the outcome and metabolites as predictors, adjusting for age, sex, and BMI; these models were evaluated both with and without adjustment for disease status. Finally, to determine whether associations between caffeine/xanthine metabolites and COPD were driven by theophylline medication, we performed a sensitivity analysis excluded individuals with self-reported theophylline use.

We attempted replication in COPDGene (QIA was considered a proxy for IPF) for metabolites that met significance in LTRC, and which were measured in COPDGene. Metabolites in LTRC and COPDGene were matched either by an exact biochemical name or HMDB ID. Metabolites that met significance thresholds (p < 0.05) and displayed consistent directions of effects were considered replicated. In a secondary analysis, we repeated the comparison using a more restrictive definition of COPD in COPDGene, matching the LTRC criteria (FEV1 < 80% predicted and FEV1/FVC < 0.7), while the control group remained unchanged.

To assess the association of validated disease metabolites with disease severity, we fit covariate‑adjusted linear regression models in both LTRC and COPDGene, modeling lung‑function measures (FEV1% predicted [FEV1pp] and FVC% predicted [FVCpp]) as a function of each replicated metabolite. We adjusted for age, sex, BMI, and smoking status, and compared the resulting β‑coefficients across the two cohorts.

All analyses were performed using R version 4.3.0.

Weighted gene correlation network analysis (WGCNA)

WGCNA was used to identify modules of highly correlated metabolites within the LTRC. Metabolite modules were merged using a cut height (i.e., Euclidean distance between clusters) of 0.3 based on an iterative process to optimize module formation for analysis. WGCNA computes a module membership value and associated p-value for each feature within a module, indicating a strength of correlation between metabolites within a module. Modules were summarized by an eigenvector based on the first principal component, representing the overall expression or abundance profile of each module. We then computed Pearson correlation coefficients between the module eigenvectors and disease status. Disease status was encoded as a binary variable for each comparison, assigning a value of 1 to individuals in the disease group of interest and 0 to the other group. The WGCNA package from R 4.2.0 was used for this analysis.

Results

LTRC participants were predominantly white (~ 90%), with an age range of 21–91 years (Table 1). The COPD (53.1%) and IPF (67.9%) groups were predominantly male, while the control group had a lower proportion of males (40%). The COPD group had the lowest average BMI (26.3 kg/m2), while the IPF group showed the highest (29.9 kg/m2). The proportion of never smokers was higher in the control (33.7%) and IPF (35.8%) groups than in the COPD group (4.9%).

Table 1.

Demographics and clinical characteristics of the lung tissue research consortium (LTRC) metabolomics subcohort

Control COPD IPF p†
N 361 480 218
Age, mean (SD) 61.44 (12.60) 63.45 (9.32) 63.73 (8.37) 0.008
Sex (%) Female 219 (60.7)

225

(46.9)

70

(32.1)

 < 0.001
Male 142 (39.3)

255

(53.1)

 148 (67.9)
BMI, mean (SD) 28.94 (6.07) 26.31 (5.25) 29.93 (5.51) < 0.001
Race (%) White 326 (90.3)

434

(90.4)

 196 (89.9) 0.023
Asian 0 (0.0) 0 (0.0) 1 (0.5)
Black 22 (6.1) 30 (6.2) 8 (3.7)
Hispanic 11 (3.0) 12 (2.5) 5 (2.3)
Other 2 (0.6) 4 (0.8) 8 (3.7)
Smoking status (%) Never 110 (30.5)

22

(4.6)

72

(33.0)

 < 0.001
Former 201 (55.7) 389 (81.0) 128 (58.7)
Current

15

(4.2)

35

(7.3)

1

(0.5)
Missing

35

(9.7)

34

(7.1)

17

(7.8)
Pack-years, mean (SD) 20.26 (27.86) 46.89 (31.43) 17.90 (23.33) < 0.001
Post-BD FEV1% predicted, mean (SD) 98.47 (12.36) 46.64 (21.19) 67.45 (18.70) < 0.001
Post-BD FVC % predicted, mean (SD) 96.66 (12.40) 74.15 (18.86) 61.20 (16.93) < 0.001
Post-BD FEV1/FVC, mean (SD) 0.78 (0.05) 0.46 (0.15) 0.83 (0.07) < 0.001
Open in a new tab

†Kruskal-Wallis test was used for continuous variables and Chi-squared for categorical variables. Reported p-values (p#) indicate whether a significant difference exists among any of the groups

BMIbody mass index (units = kg/m2), FEV1 forced expiratory volume in 1 s, FVC forced vital capacity, and SD standard deviation

Characteristics of the COPDGene replication cohort were comparable with a few notable differences. Compared to LTRC, COPDGene was more racially heterogeneous (73% white and 27% black, Supplementary Table 1). While sex distribution patterns for control and COPD groups were similar between the two cohorts, the QIA group in COPDGene had a higher proportion of females (64.2%) than the LTRC IPF group (32.1%). Notably, the QIA phenotype differs from the definition of IPF in LTRC. The proportion of current smokers was higher in all three groups in COPDGene compared to LTRC.

Metabolites associated with disease status versus controls

After QC and data processing, 1131 metabolites were included in the LTRC analysis. Following FDR correction (FDR < 0.05), 246 (21.8%) were significantly associated with COPD status compared to controls, and 136 (12.0%) with IPF status compared to controls (Fig. 1).

Fig. 1.

Fig. 1

Open in a new tab

Volcano plots showing the −log10(FDR-corrected p-values) and beta estimates (natural log of the odds ratio) for metabolites associated with (A) COPD and (B) IPF in LTRC plasma samples relative to controls. In (A), 246 metabolites met the significance threshold (FDR < 0.05, gray dotted line) while in (B) 136 metabolites were significant with the top 20 most significant metabolites in each comparison labeled. Color legend: Blue labels/dots represent the 166 “COPD Only” metabolites that were significant in COPD vs. control but not in IPF vs. control, red labels/dots represent the 56 “IPF Only” metabolites that were significant in IPF vs. control but not in COPD vs. control, orange labels/dots represent the 77 “concordant” metabolites that were significant and demonstrated same direction of effect in both diseases relative to controls, green color dots represent the 3 “discordant” metabolites that were significant and demonstrated opposite direction of effect in the two diseases relative to controls. (C) Venn diagram depicting the overlap in significantly altered metabolites between COPD and IPF relative to controls. D Forest plot showing the odds ratios (ORs) and 95% confidence intervals (CI) for the 3 discordant metabolites identified across disease comparisons relative to controls

Eighty metabolites were commonly associated with COPD and IPF compared to controls, of which 77 displayed concordant directions of effect, including reductions in levels of steroids, and lipid metabolites belonging to diglycerides (DG), triglycerides (TG), lysophosphatidylcholine (LPC), phosphatidylcholine (PC), and phosphatidylethanolamine (PE) classes. Several lipids in the carnitines (CAR) class were elevated in both diseases (Supplementary Table 2). Notably, three metabolites were associated with both outcomes but showed discordant directions of effect between COPD and IPF, all demonstrating higher levels in individuals with COPD and lower levels in individuals with IPF compared to controls: T4/thyroxine (COPD OR = 2.61 [1.41,4.92], FDR = 2.04 × 10− 2; IPF OR = 0.28 [0.12,0.66], FDR = 3.92 × 10− 2), cotinine (COPD: OR = 1.28 [95% CI: 1.09,1.52], FDR = 2.4 × 10− 2; IPF OR = 0.65 [0.47,0.86], FDR = 3.92 × 10− 2), and trans-3-Hydroxycotinine (COPD OR = 1.25 [1.08,1.46], FDR = 2.53 × 10− 2; IPF OR = 0.72 [0.57,0.89], FDR = 3.93 × 10− 2).

A total of 166 metabolites were associated with COPD but not IPF, including caffeine metabolites, glycosphingolipids (GlcCer and HexCer), and sphingomyelins (SM), which were elevated in individuals with COPD. Several lysophosphatidylethanolamine (LPE) species, and n-3 polyunsaturated fatty acid (PUFA) metabolites, such as DHA and EPA, were reduced in individuals with COPD relative to controls. Glycerophosphoethanolamines species PE P– and PE O– were associated with higher levels in COPD while LPE and PE species showed lower levels in COPD compared to controls.

Fifty-six metabolites were associated with IPF but not COPD. Among these were higher levels of n-6 PUFA metabolites, such as adrenic acid and eicosadienoic acid, and lower levels of retinol in IPF relative to controls. Interestingly, PUFA metabolites exhibited opposite directions of effect in the two diseases with reduced levels in COPD but elevated levels in IPF compared to controls despite the lack of overlap on an individual metabolite level.

In a secondary analysis using binarized cotinine values in place of self-reported smoking status, we identified 80 additional metabolites significantly associated with COPD compared to controls, while 5 metabolites that were previously significant using questionnaire data were no longer significant (Supplementary Fig. 2, Supplementary Table 4), including the nicotine metabolite, trans-3-hydroxycotinine. To contextualize these findings, we conducted complementary analysis to identify metabolites associated with smoking (Supplementary Table 5). Of the additional 80 metabolites detected in cotinine-adjusted COPD vs. control comparison, 14 metabolites overlapped with smoking-associated metabolites in the analysis with adjustment for disease status: SM 35:1;O2, SM 36:0;O2, DG 38:4, PC P-44:5 or PC O-44:6_A, TG 44:0, TG 48:0, TG 50:0, TG 50:1, TG 50:2, TG 52:0, TG 52:1, TG 54:2, CE 18:0. In addition to these 14 metabolites 12 more metabolites- Bilirubin, LPE 22:6/0:0, PC P-32:0, DG 40:5, Coenzyme Q10, TG 48:3, TG 54:0, CE 16:0, CE 16:1, CE 18:2, Ribothymidine, PC P-38:5 or PC O-38:6, and PE 35:2- overlapped in the analysis without adjustment for disease status. Notably, 80–90% of these belonged to structural or storage lipid classes. This suggests metabolites related to PC, TG, DG, SM, and CE classes may reflect the intertwined effects of smoking related metabolic perturbations and disease related dysregulation in people with COPD. The remaining metabolites not associated with smoking, may provide insights into disease associated biology, but further work is required.

We identified 28 additional significant metabolites in comparisons of IPF vs. control when adjusting for cotinine-based smoking rather than questionnaire-based smoking, and 12 metabolites lost significance in the cotinine adjusted models. The directions of effect were consistent across models for all metabolites, including those that lost significance (Supplementary Table 4). Of the additional 28 metabolites detected in cotinine-adjusted IPF vs. control comparison, 3 metabolites were associated with cotinine-based smoking status in analyses adjusted for disease status: Kynurenic acid, PC O-32:0, and SM 36:2;O2. In addition to these 3, N-Acetyl-Taurine, and SM 36:1;O2, metabolites were significant when not adjusting for disease status (Supplementary Table 5). Although smoking is a smaller confounder in IPF than in COPD, these metabolites may still reflect residual smoke exposure rather than disease biology.

Metabolites associated with COPD vs. IPF

Levels of 47 metabolites differed significantly between COPD and IPF (Supplementary Table 2). Twenty-three showed higher levels in individuals with COPD relative to IPF (OR range = 1.35–13.12 and FDR = 1.38 × 10− 6 − 4.38 × 10− 2) (Fig. 2A), including all 3 metabolites that displayed discordant directions of effect in the two diseases relative to controls: thyroxine (OR = 13.12 [5.83,30.75], FDR = 1.38 × 10− 6), cotinine (OR = 2.36 [1.75,3.36], FDR = 4.96 × 10− 5), and trans-3-hydroxycotinine (OR = 1.90 [1.52,2.41], FDR = 1.47 × 10− 5). There were 24 metabolites with decreased levels in COPD relative to IPF (OR range = 0.21–0.93 and FDR = 1.47 × 10− 5 − 4.56 × 10− 2), including phenylalanine, tyrosine, and DHA. Detailed regression outputs for all metabolites and all LTRC comparisons are available in Supplementary Table 2.

Fig. 2.

Fig. 2

Open in a new tab

Forest plot of the 47 FDR significant metabolites in COPD vs. IPF comparison: A Odds ratio (OR) with 95% CI and FDR p-values of 23 metabolites that showed higher levels in COPD relative to IPF. B Odds ratio (OR) with 95% CI and FDR p-values of 24 metabolites that showed lower levels in COPD compared to IPF. Metabolites that were significant in COPD vs. control but not in IPF vs. control are labeled in blue (“COPD Only”), Metabolites that were significant in IPF vs. control but not in COPD vs. control are labeled in red (“IPF Only”), metabolites that displayed discordant direction of effect in the two diseases relative to controls are labeled in green (“Discordant”) and metabolites that were not significant in the two diseases compared to controls are labeled in black (“Not Significant in diseases vs. control”)

Using the binarized cotinine values instead of the self-reported smoking status resulted in 4 additional metabolites reaching significance in the COPD vs. IPF comparison, while 15 metabolites were no longer significant, including the nicotine metabolite, trans-3-hydroxycotinine. All the significant metabolites identified from the two models had consistent directions of effect (Supplementary Table 4). Of the additional 4 additional metabolites detected in cotinine-adjusted COPD vs. IPF comparison, only 1 metabolite, HexCer 40:1;O2, was associated with cotinine-based smoking status (Supplementary Table 5).

Associations of lipid carbon chain length and number of double bonds with COPD and IPF

A total of 315 metabolites appeared significant in the LTRC disease comparisons, of which a majority (69.8%) belonged to lipid classes. Of the 220 lipid metabolites associated with COPD and IPF in the three analyses depicted in Figs. 2 and 205 were included in a further examination of carbon chain length vs. the degree of saturation. Of these 173, 81, and 11 were significant in the COPD vs. control, IPF vs. control, and COPD vs. IPF comparisons, respectively at FDR < 0.05. Analysis of lipid structural features demonstrated that individuals with COPD and IPF had higher levels medium-chain length CARs, but lower levels of saturated and unsaturated triglycerides and diglycerides compared to controls (Fig. 3A and B). Glycerophosphocholines, including LPC O–, PC O– and PC P– species, with longer chain lengths showed higher levels in COPD relative to controls; however, shorter chain glycerophosphocholines, such as LPCs, showed lower levels (27/31). When directly comparing COPD and IPF, individuals with COPD showed higher levels of sphingomyelins, regardless of chain length and degree of saturation. CARs with carbon chain length 18 or higher displayed lower levels in COPD relative to IPF (Fig. 3C), whereas those with a chain length 7 or lower have higher levels in COPD; these species had double bonds ranging from 0 to 3.

Fig. 3.

Fig. 3

Open in a new tab

Lipids associated with disease status, summarized by carbon chain length and number of double bonds in A COPD vs. control; B IPF vs. Control; C COPD vs. IPF. FDR and nominally significant metabolites are denoted by solid triangles and solid squares respectively. Positive associations (β > 0) are shown in red, and negative associations (β < 0) are shown in teal

Replication of metabolomic associations in COPDGene

Among the 315 metabolites that were significant in LTRC comparisons, only 120 (38.1%) metabolites could be confidently matched in COPDGene, either by an exact biochemical name or HMDB ID. Most unmatched metabolites belonged to lipid classes, which are difficult to confidently identify without targeted analysis approaches and specific lipidomic pipelines. Forty-nine (40.8% of those that could be matched) metabolites were replicated in the COPD vs. control comparison with consistent directions of effect, including caffeine, trans-3-hydroxycotinine, steroid and PUFA metabolites (Fig. 4). A subset of CAR (7 of 11 that could be matched, 63.6%), PC/LPC (4/13, 30.8%), and PE/LPE (5/7, 71.4%) lipid metabolites also replicated. We repeated our analysis using a more restrictive definition of COPD in COPDGene (with the same control group), and 48 of the 49 metabolomic associations were retained.

Fig. 4.

Fig. 4

Open in a new tab

Replication of Metabolomic Associations in LTRC and COPDGene: A Forest plot of ORs of replicated metabolites with higher levels in COPD relative to controls. B Forest plot of ORs of replicated metabolites with lower levels in COPD relative to controls. C IPF vs. Control replication: Forest plot of ORs of matched significant metabolites in the two cohorts. D COPD vs. IPF replication: Forest plot of ORs of matched significant metabolites in the two cohorts. ORs in LTRC and COPDGene are represented in circles and triangles respectively along with the 95% CI. The colors represent the groups the metabolites belong to from Fig. 1. Color coded metabolite groups: Blue = “COPD Only”, red = “IPF Only”, orange = “concordant”, green = “Discordant”, and gray = not significant in disease vs. control. A–D Replicated metabolites: significant in both cohorts and consistent in direction of effect

Using the QIA disease classification in COPDGene, we attempted replication of the IPF associated findings from LTRC. Of 136 metabolites associated with IPF vs. control analysis in LTRC, 40 could be measured; of these, 4 (10%) metabolites showed consistent associations: retinol, N-acetylputrescine, N2,N2-dimethylguanosine, and LPC 20:0. Notably, LPC 20:0 was associated with both diseases compared to control and was replicated in COPDGene for both COPD and IPF/QIA relative to controls. The remaining 36 metabolites did not replicate; these included 12 amino acids, 20 lipids, and 4 xenobiotic metabolites, however, we cannot determine whether these represent false positives in the LTRC analysis or arise from phenotype differences in definitions of IPF and QIA in the two cohorts. None of the replicated metabolites showed significant association with FEV1pp and FVCpp across the two cohorts in subjects with IPF (Supplementary Tables 7 and 9).

Similarly, in the COPD vs. IPF findings replication using QIA in COPDGene, 26 metabolites were measured in COPDGene among the 47 that were significant in LTRC. A total of 9 metabolites showed consistent associations, including caffeine metabolites, T4/thyroxine, and DHA. We were unable to replicate 17 metabolites including 10 amino acids, 2 lipids, 2 nucleotide and 3 xenobiotics.

To assess the association of replicated metabolites with disease severity, we ran regression analysis in subjects with COPD across both cohorts with pulmonary function test (PFT) parameters FEV1pp and FVCpp after covariate adjustments. Seven of 49 replicated metabolites, including, PE 36:2, PE 36:1, pregnenolone sulfate, alanine, androsterone-3-glucuronide, myo-inositol, dehydroepiandrosterone sulfate showed significant associations with FEV1pp across COPD subjects in both cohorts (Supplementary Tables 6 and 8). Only Dehydroepiandrosterone sulfate exhibited significant association with FVCpp across both cohorts in individuals with COPD (Supplementary Tables 7 and 9).

WGCNA metabolite modules associated with disease

In a correlation network analysis, WGCNA identified five metabolite modules in LTRC, four modules were associated with COPD and/or IPF (Fig. 5, Supplementary Fig. 3, Supplementary Table 3). The blue module was positively correlated with COPD compared to controls (Pearson’s correlation coefficient, r2 = 0.11, p-value = 3 × 10− 4) and COPD compared to IPF (r2 = 0.12, p-value = 2 × 10− 4). This module contained 79 metabolites, primarily of LPC O–, PC O–, PC P– and SM classes. The green module was positively correlated with both COPD and IPF compared to controls (COPD: r2 = 0.084, p-value = 7 × 10− 3; IPF: r2 = 0.11, p-value = 2 × 10− 4). This module contained 29 metabolites belonging to the classes of CAR and hydroxy acids and derivatives. The brown module was negatively correlated with both diseases compared to controls (COPD: r2 = −0.09, p-value = 3 × 10− 2; IPF r2 = −0.15, p-value = 1 × 10− 6). This module was comprised of 50 metabolites, including LPC, LPE, PC, PE, SM, steroids, and steroid derivatives. The turquoise module was also negatively correlated with COPD (r2 = −0.19, p-value = 1 × 10− 9) and with IPF (r2 = −0.12, p-value = 1 × 10− 4) compared to controls. This module contained 108 metabolites, including TGs, DGs, and PEs, which showed positive model membership values; and cholesteryl esters (CEs) which demonstrated negative coefficients of metabolite module membership based on Pearson’s correlation (MM = −0.84, −0.51, MMP-values = 5.22 × 10− 278−6.42 × 10− 71). We could not directly replicate the WGCNA analysis in COPDGene, as most unmatched metabolites belonged to the lipid classes represented by these modules; these results should therefore be considered hypothesis-generating.

Fig. 5.

Fig. 5

Open in a new tab

A Table displaying the Pearson correlation coefficient, along with the p-value in brackets, between WGCNA metabolomic modules and disease status. Positive and negative values represent positive and negative correlations, respectively. Volcano plots of –log10 (p-value) and the coefficient of met module membership for the B 79 metabolites in the blue module C 29 metabolites in the green module D 50 metabolites in the brown module and E 108 metabolites in the turquoise module. The top 15 metabolites by p-value are labelled in B–D. Top 15 metabolites and metabolites with negative coefficient of met module membership are labeled in E. Metabolites are colored by the groups defined in Figs. 1 and 2

Associations between metabolites and SNPs of interest

The genetic analysis sought to estimate associations between SNPs mapping to five genes with opposite directions of effect in COPD and IPF GAWS from prior literature and levels of the 315 metabolites associated with disease status or those differentiating COPD and IPF (Sakornsakolpat et al., 2019). After correction for multiple testing, none of these five SNPs were significantly associated with metabolite levels in the LTRC metabolomics cohort (data not shown). As we saw no significant associations, we did not attempt replication in COPDGene.

Discussion

COPD and IPF are chronic lung diseases with high morbidity and mortality burdens that remain incompletely understood at the biochemical level. Examining their overlapping and disease-specific biochemical signatures may facilitate early detection, support novel management approaches, and lead to better outcomes (Lamas et al., 2011), and metabolomics offers a means to discover relevant biochemical pathways. Our analysis demonstrated a high degree of overlap in altered metabolites and metabolite classes between COPD and IPF while also highlighting metabolites with discordant directions of effect or that were uniquely associated with each disease. These results underscore both common and distinct biochemical alterations in the two diseases.

Previous metabolomic studies have determined that individuals with chronic lung diseases have an increased energy requirement and a disrupted catabolism and anabolism balance that may be mediated by lipids (Chen & Dai, 2023; Liang et al., 2025; Wu et al., 2023). Our results showed reductions in levels of TGs and DGs with concomitant elevations in CARs and CEs across both COPD and IPF, which may reflect common biochemical disruptions involving these lipid metabolites. These metabolite classes were also associated with smoking exposure highlighting the intertwined effects of smoking related metabolic perturbations and disease related lipid dysregulation. These associations were further explored in our WGCNA, with the green network module, primarily comprised of CARs, showing positive correlations with both diseases, while the turquoise module, primarily comprised of TGs, DGs, and CEs, negatively correlated with both diseases compared to controls. Lower levels of PCs were also observed in both diseases, and these are important precursors to TGs and DGs, providing further evidence of global lipid network dysregulation in both diseases (Kotlyarov, 2025). Decreased levels of PCs have been demonstrated in both COPD and IPF, as well as in QIA, which has been proposed to be a precursor to advanced parenchymal diseases (Choi et al., 2023; Cruickshank-Quinn et al., 2018; Schmidt et al., 2002; Seeliger et al., 2022). These findings may indicate a common upstream lipid metabolic disturbance shared across early and late stages of lung disease.

The carbon chain length and double bond content of lipids can have significant effects on the fluidity of the cell membrane by altering the membrane thickness (Park et al. 2021). Multiple studies suggest short fatty acid carbon chain length and fewer double bonds are associated with higher risk for type 2 diabetes (T2D) and cardiovascular diseases (CVD) (Eichelmann et al. 2022; Huang et al. 2021; Olund Villumsen et al. 2022; Toledo et al. 2017). While lipid unsaturation patterns have been explored in COPD (Titz et al., 2016), our study is the first, to our knowledge, to comprehensively explore the lipid class specific carbon chain length and double bond patterns in subjects with COPD and IPF. The similar directions of effect observed in TG, DG, and CE lipids across both COPD and IPF, regardless of structural features, suggest that alterations may represent a common metabolic signature of chronic lung disease.

Steroids also represent important inflammatory regulators, and our results showed consistent relationships of steroid metabolite levels across COPD and IPF, with external replication. While inhaled corticosteroids are often used clinically in COPD to reduce exacerbations, we observed associations with steroids across multiple biosynthesis pathways in COPD and IPF, including androgens and pregnenolones. Notably, several steroid metabolites were associated with lung function parameters, which may mean they are further valuable for understanding lung disease severity. Steroids were also widely prescribed in IPF, although they are no longer recommended (Gay et al., 1998), it is likely that many IPF subjects in the LTRC cohort received steroid treatment during the period of sample collection. We could not assess the impact of steroid medications in this study, so these findings require follow-up with detailed history of medication use.

Several other notable pathways were uniquely associated with either COPD or IPF, which could further suggest differences in underlying biochemical mechanisms relevant to COPD compared to IPF. Sphingolipids such as SMs and glycosphingolipids are cell membrane lipids that can also participate in signaling processes (Slotte, 2013). Previous studies have reported a lack of regulation of SM metabolites in COPD, leading to lung cell injury and elevated rates of apoptosis. This study demonstrated consistent findings with previous work (Koike et al., 2018; Regan et al., 2019; Uhlig & Gulbins, 2008) and the dysregulation of these metabolites were unique to individuals with COPD. Further, several caffeine-related metabolites were associated with COPD status in both cohorts, a finding that reflects the link between smoking and increased coffee intake (Bjørngaard et al., 2017). Associations with caffeine and xanthine metabolites with COPD individuals in LTRC persisted even when individuals using theophylline were removed from the analysis. Only a small subset of the replicated metabolites (7/49, 14.3%) were associated with PFT parameters as a measure of disease severity in individuals with COPD, including PE 36:2, PE 36:1, pregnenolone sulfate, alanine, androsterone-3-glucuronide, myo-inositol, dehydroepiandrosterone sulfate. Those replicated metabolites that were not associated with severity may reflect other facets of disease pathogenesis, or other confounding factors that we were unable to account for in our study.

Fewer metabolites overall were uniquely associated with IPF, but several long chain fatty acids (LCFAs) were elevated specifically in LTRC individuals with IPF, consistent with previous reports of increased LCFA levels in IPF lung tissue (Chu et al., 2019). LCFAs have been mechanistically linked to the pro-fibrotic phenotype that differentiates COPD and IPF (Geng et al., 2022; McManus & Knight, 2021). The role of these metabolites in COPD and IPF remains to be fully explored, but our analysis indicated they may represent unique biochemical features of each disease.

Three metabolites in the LTRC cohort demonstrated disparate directions of effect in COPD and IPF. Cotinine metabolites have historically served as useful biomarkers of tobacco exposure (Bradicich and Schuurmans 2020), and both cotinine and trans-3-hydroxycotinine levels were elevated in individuals with COPD compared to controls but reduced in individuals with IPF compared to controls. As such, this may reflect differences in environmental/lifestyle factors, particularly smoking, rather than underlying biological processes directly related to disease pathology. Thyroxine, a thyroid hormone that regulates metabolic functions (Mullur et al. 2014), also showed disparate directions of effect between diseases. A previous study directly comparing COPD and IPF individuals found hypothyroidism, marked by lower thyroxine levels, to be more prevalent in IPF, and associated reduced thyroxine levels with decreased survival (Oldham et al. 2015). In addition, the protective antifibrotic effects of thyroxine have been demonstrated in murine models (Yu et al. 2018). Our analysis was consistent with these literature findings, which may suggest a role for thyroxine or related metabolites in discriminating between COPD and IPF based on underlying biology.

There were additional pathways of interest that showed disparate patterns of effect between COPD and IPF, including PUFA metabolism. PUFA levels in blood are primarily reflective of dietary intake, and they have established roles as inflammatory regulators in both disease (Suryadevara et al., 2020; Titz et al., 2016; van der Does et al., 2019; Yaeger et al., 2025, p. 3). Reduced levels of n-3 omega fatty acids - such as DHA and EPA - were observed COPD while elevated n-6 omega fatty acids - such as adrenic acid and eicosadienoic acid - were observed in IPF, suggesting differential impact of this pathway on the two diseases possibly due to differences in inflammatory mechanisms (Patchen et al., 2023; Schwartz, 2000). Notably, the results for COPD were externally replicated, further supporting the importance of n-3 PUFAs in COPD. While most CAR metabolites displayed similar effects on both diseases, lipid feature based analysis of the COPD vs. IPF suggest differential association of a few. In IPF individuals, long‑chain CAR have been reported to be elevated (Summer et al., 2024) and in COPD, metabolomic analyses have identified significant changes in CAR (Gillenwater et al., 2021). In our study, we observed accumulation of long chain CAR in IPF individuals while simultaneous reduction in COPD individuals. This hints at a differential risk factor for lung injury with dysfunctional fatty acid oxidation in COPD and IPF (Otsubo et al., 2015).

There were several strengths in this study. Our study leveraged two large cohorts with physician-diagnosed disease and global metabolomic profiling that allowed us to broadly capture biochemistry relevant to many different mechanisms. Our study also leveraged substantially larger sample sizes compared to prior studies investigating plasma metabolomic differences in COPD, IPF and controls (Nambiar et al. 2021; Yan et al. 2017). However, several limitations should be considered when interpreting the results of this study. Importantly, two different metabolomic profiling platforms were used across the two cohorts. We were only able to attempt replication of 120 (38.1%) of the significant metabolites in the external COPDGene cohort due to differences in analytic platforms. It is well recognized that different platforms may capture different subsets of metabolites depending on the specific parameters used. The crossover in metabolites reported in this study is within the range of previously-reported studies comparing Metabolon and the Broad metabolomics platforms (Yu et al. 2019). Due to the large proportion of metabolites unable to be confidently matched for replication, we are limited in the conclusions we can draw about the generalizability of our findings. Furthermore, the use of untargeted metabolomics precluded direct comparison of effect sizes between cohorts. While COPD disease status was available in the two cohorts, IPF was compared to QIA in COPDGene, and the lower degree of replication found in the analyses involving QIA may reflect the important clinical differences between these two phenotypes. Those metabolites that did replicate may be involved in more general dysregulated lung health, rather than being specific to IPF, and further work in well phenotyped populations is needed to explore this. The genetic analysis was limited by sample size, which could explain our inability to detect confident associations between SNPs of interest and metabolite levels. Finally, our analysis utilized plasma samples, which likely reflect systemic alterations relevant to multiple organs and tissues. While this limits our ability to make lung-specific inferences, this discovery-based approach could ultimately support the development of cost-effective, minimally invasive tools to differentiate the two diseases in patients. Future work with lung tissue samples could enhance our understanding of localized lung specific processes.

In conclusion, this study highlighted several biochemical pathways consistently altered in individuals with COPD and IPF, including structural lipids, inflammatory lipids, and steroids. In contrast, PUFA, thyroxine, and cotinine metabolites discriminated the two diseases. These metabolites may have future clinical utility as targets for therapeutic intervention or disease-specific biomarkers.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (578.5KB, xlsx)
Supplementary Material 2 (742KB, docx)

Acknowledgements

Molecular data for the Trans-Omics in Precision Medicine (TOPMed) program was supported by the National Heart, Lung and Blood Institute (NHLBI). Metabolomics for “NHLBI TOPMed: LTRC (phs001662) was performed at Broad Institute (HHSN268201600034I). Metabolomics for “NHLBI TOPMed: COPDGene (phs000951) was performed at Northwest Genomics center (HHSN268201600032I). Core support including centralized genomic read mapping and genotype calling, along with variant quality metrics and filtering were provided by the TOPMed Informatics Research Center (3R01HL-117626-02S1; contract HHSN268201800002I). Core support including phenotype harmonization, data management, sample-identity QC, and general program coordination were provided by the TOPMed Data Coordinating Center (R01HL-120393; U01HL-120393; contract HHSN268201800001I). We gratefully acknowledge the studies and participants who provided biological samples and data for TOPMed.

Author contributions

AR, NP, ES and RK contributed the concept and design, data analysis, and manuscript writing and editing; SM, SL, JH, JY, MC, and AH contributed to data analysis, statistical support and manuscript editing, GW, FS, LB, AL, KF, GC, KB, RW, FM, RB, RG and CC contributed to data collection, data analysis and manuscript editing.

Funding

This work was supported by NHLBI grants U01 HL089897 and U01 HL089856 and by NIH contract 75N92023D00011. The COPDGene study (NCT00608764) is supported by grants from the NHLBI (U01HL089897 and U01HL089856), by NIH contract 75N92023D00011, and by the COPD Foundation through contributions made to an Industry Advisory Committee that has included AstraZeneca, Bayer Pharmaceuticals, Boehringer-Ingelheim, Genentech, GlaxoSmithKline, Novartis, Pfizer and Sunovion. This study utilized biological specimens and data provided by the Lung Tissue Research Consortium (LTRC) supported by the National Heart, Lung, and Blood Institute (NHLBI). EKS is supported by P01HL114501, R01HL133135, and R01HL152728. NP is supported by K01HL175261 from NHLBI. RSK is supported by Mass General Brigham Research, and R01HL168199, R01HL155742 and R01HL169300 from NHLBI. JHY is supported by K08HL146972, SNUCMAA Hahn Seung Shin Research Grant. MHC was supported by R01HL168199, R01HL162813, R01HL153248, and R01HL135142. JH was supported by NIH/NHLBI K01HL169756. SML is supported by R01MH129337. AR was supported by NIH T32 HL007427 from NIH/NHLBI.

Data availability

The metabolomics data for LTRC and COPDGene are available through monitored public access in dbGaP.

Declarations

Competing interests

In the past three years, EKS received institutional grant support from Bayer and Northpond Laboratories. MHC has received grant support from Bayer and consulting fees from Apogee Therapeutics and BMS, unrelated to the current work. JHY received institutional grant support from Bayer and consulting fee from Genentech and Bridge Biotherpeutics. JH received a gift from BridgeBio Inc. to conduct research.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research board (LTRC: 2018P000186 [Network Analysis of Multiple Omics Data in the Lung Tissue Research Consortium]; COPDGene: 2007P000554 [Genetic Epidemiology of COPD]) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards; Written informed consent was obtained from all individual participants included in the stud.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Aldric Rosario and Nicole Prince contributed equally to this work.

Edwin K. Silverman and Rachel S. Kelly contributed equally to this work.

References

  1. Beghé, B., Cerri, S., Fabbri, L. M., & Marchioni, A. (2021). COPD, pulmonary fibrosis and ILAs in aging smokers: The paradox of striking different responses to the major risk factors. International Journal of Molecular Sciences,22(17), Article 9292. 10.3390/ijms22179292 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Berry, C. E., Drummond, M. B., Han, M. K., Li, D., Fuller, C., Limper, A. H., Martinez, F. J., Schwarz, M. I., Sciurba, F. C., & Wise, R. A. (2012). Relationship between lung function impairment and health-related quality of life in COPD and interstitial lung disease. Chest, 142(3), 704–711. 10.1378/chest.11-1332 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bjørngaard, J. H., Nordestgaard, A. T., Taylor, A. E., Treur, J. L., Gabrielsen, M. E., Munafò, M. R., Nordestgaard, B. G., Åsvold, B. O., Romundstad, P., & Davey Smith G. (2017). Heavier smoking increases coffee consumption: Findings from a Mendelian randomization analysis. International Journal of Epidemiology,46(6), 1958–1967. 10.1093/ije/dyx147 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bradicich, M., & Schuurmans, M. M. (2020). Smoking status and second-hand smoke biomarkers in COPD, asthma and healthy controls. ERJ Open Research,6(2), 00192–02019. 10.1183/23120541.00192-2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chen, R., & Dai, J. (2023). Lipid metabolism in idiopathic pulmonary fibrosis: From pathogenesis to therapy. Journal of Molecular Medicine,101(8), 905–915. 10.1007/s00109-023-02336-1 [DOI] [PubMed] [Google Scholar]
  6. Chilosi, M., Poletti, V., & Rossi, A. (2012). The pathogenesis of COPD. Distinct horns of the same devil?. [DOI] [PMC free article] [PubMed]
  7. Cho, M. H., Hobbs, B. D., & Silverman, E. K. (2022). Genetics of chronic obstructive pulmonary disease: Understanding the pathobiology and heterogeneity of a complex disorder. The Lancet Respiratory Medicine,10(5), 485–496. 10.1016/S2213-2600(21)00510-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Choi, B., San José Estépar, R., Godbole, S., Curtis, J. L., Wang, J. M., San José Estépar, R., Rosas, I. O., Mayers, J. R., Hobbs, B. D., Hersh, C. P., Ash, S. Y., Han, M. K., Bowler, R. P., Stringer, K. A., Washko, G. R., & Labaki, W. W. (2023). Plasma metabolomics and quantitative interstitial abnormalities in ever-smokers. Respiratory Research,24(1), 265. 10.1186/s12931-023-02576-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chu, S. G., Villalba, J. A., Liang, X., Xiong, K., Tsoyi, K., Ith, B., Ayaub, E. A., Tatituri, R. V., Byers, D. E., Hsu, F. F., El-Chemaly, S., Kim, E. Y., Shi, Y., & Rosas, I. O. (2019). Palmitic acid–rich high-fat diet exacerbates experimental pulmonary fibrosis by modulating endoplasmic reticulum stress. American Journal of Respiratory Cell and Molecular Biology, 61(6), 737–746. 10.1165/rcmb.2018-0324OC [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Clish, C. B. (2015). Metabolomics: An emerging but powerful tool for precision medicine. Molecular Case Studies,1(1), Article a000588. 10.1101/mcs.a000588 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cruickshank-Quinn, C. I., Jacobson, S., Hughes, G., Powell, R. L., Petrache, I., Kechris, K., Bowler, R., & Reisdorph, N. (2018). Metabolomics and transcriptomics pathway approach reveals outcome-specific perturbations in COPD. Scientific Reports, 8(1), 17132. 10.1038/s41598-018-35372-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Eichelmann, F., Sellem, L., Wittenbecher, C., Jäger, S., Kuxhaus, O., Prada, M., Cuadrat, R., Jackson, K. G., Lovegrove, J. A., & Schulze, M. B. (2022). Deep lipidomics in human plasma: Cardiometabolic disease risk and effect of dietary fat modulation. Circulation,146(1), 21–35. 10.1161/CIRCULATIONAHA.121.056805 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gay, S. E., Kazerooni, E. A., Toews, G. B., Lynch, J. P., 3rd., Gross, B. H., Cascade, P. N., Spizarny, D. L., Flint, A., Schork, M. A., Whyte, R. I., Popovich, J., Hyzy, R., & Martinez, F. J. (1998). Idiopathic pulmonary fibrosis: Predicting response to therapy and survival. American Journal of Respiratory and Critical Care Medicine, 157(4), 1063–1072. 10.1164/ajrccm.157.4.9703022 [DOI] [PubMed] [Google Scholar]
  14. Geng, J., Liu, Y., Dai, H., & Wang, C. (2022). Fatty acid metabolism and idiopathic pulmonary fibrosis. Frontiers in Physiology,12, Article 794629. 10.3389/fphys.2021.794629 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gillenwater, L. A., Kechris, K. J., Pratte, K. A., Reisdorph, N., Petrache, I., Labaki, W. W., O’Neal, W., Krishnan, J. A., Ortega, V. E., DeMeo, D. L., & Bowler, R. P. (2021). Metabolomic profiling reveals sex specific associations with chronic obstructive pulmonary disease and emphysema. Metabolites, 11(3), 161. 10.3390/metabo11030161 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gillenwater, L. A., Pratte, K. A., Hobbs, B. D., Cho, M. H., Zhuang, Y., Halper-Stromberg, E., Cruickshank-Quinn, C., Reisdorph, N., Petrache, I., Labaki, W. W., O'Neal, W. K., Ortega, V. E., Jones, D. P., Uppal, K., Jacobson, S., Michelotti, G., Wendt, C. H., Kechris, K. J., & Bowler, R. P. (2020). Plasma metabolomic signatures of chronic obstructive pulmonary disease and the impact of genetic variants on phenotype-driven modules. Network and Systems Medicine,3(1), 159–181. 10.1089/nsm.2020.0009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Huang, H., Ye, G., Lai, S. Q., Zou, H. X., Yuan, B., Wu, Q. C., Wan, L., Wang, Q., Zhou, X. L., Wang, W. J., Cao, Y. P., Huang, J. F., Chen, S. L., Yang, B. C., & Liu, J. C. (2021). Plasma lipidomics identifies unique lipid signatures and potential biomarkers for patients with aortic dissection. Frontiers in Cardiovascular Medicine,8, 757022. 10.3389/fcvm.2021.757022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Koike, K., Berdyshev, E. V., Bowler, R. P., Scruggs, A. K., Cao, D., Schweitzer, K. S., Serban, K. A., & Petrache, I. (2018). Bioactive sphingolipids in the pathogenesis of chronic obstructive pulmonary disease. Annals of the American Thoracic Society,15(Supplement_4), S249–S252. 10.1513/AnnalsATS.201809-592MG [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kotlyarov, S. (2025). Linking lipid metabolism and immune function: new insights into chronic respiratory diseases. Pathophysiology : The Official Journal Of The International Society For Pathophysiology / Isp,32(2), 26. 10.3390/pathophysiology32020026 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Lamas, D. J., Kawut, S. M., Bagiella, E., Philip, N., Arcasoy, S. M., & Lederer, D. J. (2011). Delayed access and survival in idiopathic pulmonary fibrosis: A cohort study. American Journal of Respiratory and Critical Care Medicine,184(7), 842–847. 10.1164/rccm.201104-0668OC [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Liang, Q., Wang, Y., & Li, Z. (2025). Lipid metabolism reprogramming in chronic obstructive pulmonary disease. [DOI] [PMC free article] [PubMed]
  22. Maher, T. M., Bendstrup, E., Dron, L., Langley, J., Smith, G., Khalid, J. M., Patel, H., & Kreuter, M. (2021). Global incidence and prevalence of idiopathic pulmonary fibrosis. Respiratory Research. 10.1186/s12931-021-01791-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. McManus, B., & Knight, D. (2021). Long-chain fatty acids are bad in IPF, or are they? Respirology, 26(3), 220–221. 10.1111/resp.13981 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. GBD 2019 Chronic Respiratory Diseases Collaborators. (2023). Global burden of chronic respiratory diseases and risk factors, 1990–2019: An update from the Global Burden of Disease Study 2019. eClinicalMedicine, 59, 101936. 10.1016/j.eclinm.2023.101936 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Moss, B. J., & Rosas, I. O. (2023). Defining the genetic landscape of idiopathic pulmonary fibrosis: Role of common and rare variants. American Journal of Respiratory and Critical Care Medicine,207(9), 1118–1120. 10.1164/rccm.202301-0177ED [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Mullur, R., Liu, Y. Y., & Brent, G. A. (2014). Thyroid hormone regulation of metabolism. Physiological Reviews,94(2), 355–382. 10.1152/physrev.00030.2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Nambiar, S., Tan, D. B. A., Clynick, B., Bong, S. H., Rawlinson, C., Gummer, J., Corte, T. J., Glaspole, I., Moodley, Y. P., & Trengove, R. (2021). Untargeted metabolomics of human plasma reveal lipid markers unique to chronic obstructive pulmonary disease and idiopathic pulmonary fibrosis. Proteomics – Clinical Applications, 15(2–3), 2000039. 10.1002/prca.202000039 [DOI] [PubMed] [Google Scholar]
  28. Oldham, J. M., Kumar, D., Lee, C., Patel, S. B., Takahashi-Manns, S., Demchuk, C., Strek, M. E. & Noth, I. (2015). Thyroid disease is prevalent and predicts survival in patients with idiopathic pulmonary fibrosis. Chest, 148(3), 692–700. 10.1378/chest.14-2714 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Olund Villumsen, S., Benfeitas, R., Knudsen, A. D., Gelpi, M., Høgh, J., Thomsen, M. T., Murray, D., Ullum, H., Neogi, U., & Nielsen, S. D. (2022). Integrative Lipidomicsand Metabolomics for System-Level Understanding of the Metabolic Syndrome in Long-Term Treated HIV-Infected Individuals. Frontiers in Immunology12, 742736. 10.3389/fimmu.2021.742736 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Otsubo, C., Bharathi, S., Uppala, R., Ilkayeva, O. R., Wang, D., McHugh, K., Zou, Y., Wang, J., Alcorn, J. F., Zuo, Y. Y., Hirschey, M. D., & Goetzman, E. S. (2015). Long-chain acylcarnitines reduce lung function by inhibiting pulmonary surfactant. Journal of Biological Chemistry, 290(39), 23897–23904. 10.1074/jbc.M115.655837 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Park, J., Choi, J., Kim, D. D., Lee, S., Lee, B., Lee, Y., Kim, S., Kwon, S., Noh, M., Lee, M. O., Le, Q. V., & Oh, Y. K. (2021). Bioactive lipids and their derivatives in biomedical applications. Biomolecules & Therapeutics,29(5), 465–482. 10.4062/biomolther.2021.107 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Patchen, B. K., Balte, P., Bartz, T. M., Barr, R. G., Fornage, M., Graff, M., Jacobs, D. R., Jr., Kalhan, R., Lemaitre, R. N., O’Connor, G., Psaty, B., Seo, J., Tsai, M. Y., Wood, A. C., Xu, H., Zhang, J., Gharib, S. A., Manichaikul, A., North, K., Steffen, L. M., Dupuis, J., Oelsner, E., Hancock, D. B., & Cassano, P. A. (2023). Investigating associations of omega-3 fatty acids, lung function decline, and airway obstruction. American Journal of Respiratory and Critical Care Medicine, 208(8), 846–857. 10.1164/rccm.202301-0074OC [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Regan, E. A., Hersh, C. P., Castaldi, P. J., DeMeo, D. L., Silverman, E. K., Crapo, J. D., & Bowler, R. P. (2019). Omics and the search for blood biomarkers in chronic obstructive pulmonary disease. Insights from COPDGene. American Journal of Respiratory Cell and Molecular Biology,61(2), 143–149. 10.1165/rcmb.2018-0245PS [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Regan, E. A., Hokanson, J. E., Murphy, J. R., Make, B., Lynch, D. A., Beaty, T. H., Curran-Everett, D., Silverman, E. K., & Crapo, J. D. (2011). Genetic epidemiology of COPD (COPDGene) study design. COPD: Journal of Chronic Obstructive Pulmonary Disease,7(1), 32–43. 10.3109/15412550903499522 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Sakornsakolpat, P., Prokopenko, D., Lamontagne, M., Reeve, N. F., Guyatt, A. L., Jackson, V. E., Shrine, N., Qiao, D., Bartz, T. M., Kim, D. K., Lee, M. K., Latourelle, J. C., Li, X., Morrow, J. D., Obeidat, M., Wyss, A. B., Bakke, P., Barr, R. G., Beaty, T. H., Belinsky, S. A., Brusselle, G. G., Crapo, J. D., de Jong, K., DeMeo, D. L., Fingerlin, T. E., Gharib, S. A., Gulsvik, A., Hall, I. P., Hokanson, J. E., Kim, W. J., Lomas, D. A., London, S. J., Meyers, D. A., O’Connor, G. T., Rennard, S. I., Schwartz, D. A., Sliwinski, P., Sparrow, D., Strachan, D. P., Tal-Singer, R., Tesfaigzi, Y., Vestbo, J., Vonk, J. M., Yim, J. J., Zhou, X., Bossé, Y., Manichaikul, A., Lahousse, L., Silverman, E. K., Boezen, H. M., Wain, L. V., Tobin, M. D., Hobbs, B. D., Cho, M. H.; SpiroMeta Consortium; International COPD Genetics Consortium. (2019). Genetic landscape of chronic obstructive pulmonary disease identifies heterogeneous cell-type and phenotype associations. Nature Genetics,51(3), 494–505. 10.1038/s41588-018-0342-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Schmidt, R., Meier, U., Markart, P., Grimminger, F., Velcovsky, H. G., Morr, H., Seeger, W., & Günther, A. (2002). Altered fatty acid composition of lung surfactant phospholipids in interstitial lung disease. American Journal of Physiology-Lung Cellular and Molecular Physiology,283(5), L1079–L1085. 10.1152/ajplung.00484.2001 [DOI] [PubMed] [Google Scholar]
  37. Schwartz, J. (2000). Role of polyunsaturated fatty acids in lung disease. The American Journal of Clinical Nutrition, 71(1), 393S–396S. 10.1093/ajcn/71.1.393S [DOI] [PubMed] [Google Scholar]
  38. Seeliger, B., Carleo, A., Wendel-Garcia, P. D., Fuge, J., Montes-Warboys, A., Schuchardt, S., Molina-Molina, M., & Prasse, A. (2022). Changes in serum metabolomics in idiopathic pulmonary fibrosis and effect of approved antifibrotic medication. Frontiers in Pharmacology,13, 837680. 10.3389/fphar.2022.837680 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Slotte, J. P. (2013). Biological functions of sphingomyelins. Progress in Lipid Research, 52(4), 424–437. 10.1016/j.plipres.2013.05.001 [DOI] [PubMed] [Google Scholar]
  40. Summer, R., Todd, J. L., Neely, M. L., Lobo, L. J., Namen, A., Newby, L. K., Shafazand, S., Suliman, S., Hesslinger, C., Keller, S., Leonard, T. B., Palmer, S. M., Ilkayeva, O., Muehlbauer, M. J., Newgard, C. B., & Roman, J. (2024). Circulating metabolic profile in idiopathic pulmonary fibrosis: Data from the IPF-PRO registry. Respiratory Research,25(1), 58. 10.1186/s12931-023-02644-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Suryadevara, V., Ramchandran, R., Kamp, D. W., & Natarajan, V. (2020). Lipid mediators regulate pulmonary fibrosis: Potential mechanisms and signaling pathways. International Journal of Molecular Sciences,21(12), Article 4257. 10.3390/ijms21124257 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Taliun, D., Harris, D. N., Kessler, M. D., Carlson, J., Szpiech, Z. A., Torres, R., Taliun, S. A. G., Corvelo, A., Gogarten, S. M., Kang, H. M., Pitsillides, A. N., LeFaive, J., Lee, S. B., Tian, X., Browning, B. L., Das, S., Emde, A. K., Clarke, W. E., Loesch, D. P., Shetty, A. C., Blackwell, T. W., Smith, A. V., Wong, Q., Liu, X., Conomos, M. P., Bobo, D. M., Aguet, F., Albert, C., Alonso, A., Ardlie, K. G., Arking, D. E., Aslibekyan, S., Auer, P. L., Barnard, J., Barr, R. G., Barwick, L., Becker, L. C., Beer, R. L., Benjamin, E. J., Bielak, L. F., Blangero, J., Boehnke, M., Bowden, D. W., Brody, J. A., Burchard, E. G., Cade, B. E., Casella, J. F., Chalazan, B., Chasman, D. I., Chen, Y. I., Cho, M. H., Choi, S. H., Chung, M. K., Clish, C. B., Correa, A., Curran, J. E., Custer, B., Darbar, D., Daya, M., de Andrade, M., DeMeo, D. L., Dutcher, S. K., Ellinor, P. T., Emery, L. S., Eng, C., Fatkin, D., Fingerlin, T., Forer, L., Fornage, M., Franceschini, N., Fuchsberger, C., Fullerton, S. M., Germer, S., Gladwin, M. T., Gottlieb, D. J., Guo, X., Hall, M. E., He, J., Heard-Costa, N. L., Heckbert, S. R., Irvin, M. R., Johnsen, J. M., Johnson, A. D., Kaplan, R., Kardia, S. L. R., Kelly, T., Kelly, S., Kenny, E. E., Kiel, D. P., Klemmer, R., Konkle, B. A., Kooperberg, C., Köttgen, A., Lange, L. A., Lasky-Su, J., Levy, D., Lin, X., Lin, K. H., Liu, C., Loos, R. J. F., Garman, L., Gerszten, R., Lubitz, S. A., Lunetta, K. L., Mak, A. C. Y., Manichaikul, A., Manning, A. K., Mathias, R. A., McManus, D. D., McGarvey, S. T., Meigs, J. B., Meyers, D. A., Mikulla, J. L., Minear, M. A., Mitchell, B. D., Mohanty, S., Montasser, M. E., Montgomery, C., Morrison, A. C., Murabito, J. M., Natale, A., Natarajan, P., Nelson, S. C., North, K. E., O’Connell, J. R., Palmer, N. D., Pankratz, N., Peloso, G. M., Peyser, P. A., Pleiness, J., Post, W. S., Psaty, B. M., Rao, D. C., Redline, S., Reiner, A. P., Roden, D., Rotter, J. I., Ruczinski, I., Sarnowski, C., Schoenherr, S., Schwartz, D. A., Seo, J., Seshadri, S., Sheehan, V. A., Sheu, W. H., Shoemaker, M. B., Smith, N. L., Smith, J. A., Sotoodehnia, N., Stilp, A. M., Tang, W., Taylor, K. D., Telen, M., Thornton, T. A., Tracy, R. P., Van Den Berg, D. J., Vasan, R. S., Viaud-Martinez, K. A., Vrieze, S., Weeks, D. E., Weir, B. S., Weiss, S. T., Weng, L., Willer, C. J., Zhang, Y., Zhao, X., Arnett, D. K., Ashley-Koch, A. E., Barnes, K. C., Boerwinkle, E., Gabriel, S., Gibbs, R., Rice, K. M., Rich, S. S., Silverman, E. K., Qasba, P., Gan, W. NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium; Papanicolaou, G. J., Nickerson, D. A., Browning, S. R., Zody, M. C., Zöllner, S., Wilson, J. G., Cupples, L. A., Laurie, C. C., Jaquish, C. E., Hernandez, R. D., O’Connor, T. D., & Abecasis, G. R. (2021). Sequencing of 53,831 diverse genomes from the NHLBI TOPMed program. Nature,590(7845), 290–299. 10.1038/s41586-021-03205-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Tirelli, C., Mira, S., Belmonte, L. A., De Filippi, F., De Grassi, M., Italia, M., Maggioni, S., Guido, G., Mondoni, M., Canonica, G. W., & Centanni, S. (2024). Exploring the potential role of metabolomics in COPD: A concise review. Cells,13(6), 475. 10.3390/cells13060475 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Titz, B., Luettich, K., Leroy, P., Boue, S., Vuillaume, G., Vihervaara, T., Ekroos, K., Martin, F., Peitsch, M. C., & Hoeng, J. (2016). Alterations in serum polyunsaturated fatty acids and eicosanoids in patients with mild to moderate chronic obstructive pulmonary disease (COPD). International Journal of Molecular Sciences,17(9), 1583. 10.3390/ijms17091583 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Toledo, E., Wang, D. D., Ruiz-Canela, M., Clish, C. B., Razquin, C., Zheng, Y., Guasch-Ferré, M., Hruby, A., Corella, D., Gómez-Gracia, E., Fiol, M., Estruch, R., Ros, E., Lapetra, J., Fito, M., Aros, F., Serra-Majem, L., Liang, L., Salas-Salvadó, J., Hu, F. B., & Martínez-González, M. A. (2017). Plasma lipidomic profiles and cardiovascularevents in a randomized intervention trial with the Mediterranean diet. The American Journal of Clinical Nutrition, 106(4), 973–983. 10.3945/ajcn.116.151159 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Uhlig, S., & Gulbins, E. (2008). Sphingolipids in the lungs. American Journal of Respiratory and Critical Care Medicine,178(11), 1100–1114. 10.1164/rccm.200804-595SO [DOI] [PubMed] [Google Scholar]
  47. van der Does, A. M., Heijink, M., Mayboroda, O. A., Persson, L. J., Aanerud, M., Bakke, P., Eagan, T. M., Hiemstra, P. S., & Giera, M. (2019). Dynamic differences in dietary polyunsaturated fatty acid metabolism in sputum of COPD patients and controls. Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids,1864(3), 224–233. 10.1016/j.bbalip.2018.11.012 [DOI] [PubMed] [Google Scholar]
  48. Wu, W., Li, Z., Wang, Y., Huang, C., Zhang, T., & Zhao, H. (2023). Advances in metabolomics of chronic obstructive pulmonary disease. Chinese Medical Journal Pulmonary and Critical Care Medicine, 1(4), 223–230. 10.1016/j.pccm.2023.10.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Yaeger, M. J., Leuenberger, L., Shaikh, S. R., & Gowdy, K. M. (2025). Omega-3 fatty acids and chronic lung diseases: A narrative review of impacts from womb to tomb. The Journal of Nutrition,155(2), 453–464. 10.1016/j.tjnut.2024.10.028 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Yan, F., Wen, Z., Wang, R., Luo, W., Du, Y., Wang, W., & Chen, X. (2017). Identification of the lipid biomarkers from plasma in idiopathic pulmonary fibrosis by Lipidomics. BMC Pulmonary Medicine, 17(1), 174. 10.1186/s12890-017-0513-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Yang, I. V., Pedersen, B. S., Rabinovich, E., Hennessy, C. E., Davidson, E. J., Murphy, E., Guardela, B. J., Tedrow, J. R., Zhang, Y., Singh, M. K., Correll, M., Schwarz, M. I., Geraci, M., Sciurba, F. C., Quackenbush, J., Spira, A., Kaminski, N., & Schwartz, D. A. (2014). Relationship of DNA methylation and gene expression in idiopathic pulmonary fibrosis. American Journal of Respiratory and Critical Care Medicine,190(11), 1263–1272. 10.1164/rccm.201408-1452OC [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Yu, B., Zanetti, K. A., Temprosa, M., Albanes, D., Appel, N., Barrera, C. B., Ben-Shlomo, Y., Boerwinkle, E., Casas, J. P., Clish, C., Dale, C., Dehghan, A., Derkach, A., Eliassen, A. H., Elliott, P., Fahy, E., Gieger, C., Gunter, M. J., Harada, S., Harris, T., Herr, D. R., Herrington, D., Hirschhorn, J. N., Hoover, E., Hsing, A. W., Johansson, M., Kelly, R. S., Khoo, C. M., Kivimäki, M., Kristal, B. S., Langenberg, C., Lasky-Su, J., Lawlor, D. A., Lotta, L. A., Mangino, M., Le Marchand, L., Mathé, E., Matthews, C. E., Menni, C., Mucci, L. A., Murphy, R., Oresic, M., Orwoll, E., Ose, J., Pereira, A. C., Playdon, M. C., Poston, L., Price, J., Qi, Q., Rexrode, K., Risch, A., Sampson, J., Seow, W. J., Sesso, H. D., Shah, S. H., Shu, X. O., Smith, G. C. S., Sovio, U., Stevens, V. L., Stolzenberg-Solomon, R., Takebayashi, T., Tillin, T., Travis, R., Tzoulaki, I., Ulrich, C. M., Vasan, R. S., Verma, M., Wang, Y., Wareham, N. J., Wong, A., Younes, N., Zhao, H., Zheng, W., & Moore, S. C. (2019). The Consortium of Metabolomics Studies (COMETS): metabolomics in 47 prospective cohort studies. American Journal of Epidemiology,188(6), 991–1012. 10.1093/aje/kwz028 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Yu, G., Tzouvelekis, A., Wang, R., Herazo-Maya, J. D., Ibarra, G. H., Srivastava, A., de Castro, J. P. W., DeIuliis, G., Ahangari, F., Woolard, T., Aurelien, N., Arrojo e Drigo, R., Gan, Y., Graham, M., Liu, X., Homer, R. J., Scanlan, T. S., Mannam, P., Lee, P. J., Herzog, E. L., Bianco, A. C., & Kaminski, N. (2018). Thyroid hormone inhibits lung fibrosis in mice by improving epithelial mitochondrial function. Nature Medicine,24(1), 39–49. 10.1038/nm.4447 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (578.5KB, xlsx)
Supplementary Material 2 (742KB, docx)

Data Availability Statement

The metabolomics data for LTRC and COPDGene are available through monitored public access in dbGaP.

Articles from Metabolomics are provided here courtesy of Springer

RESOURCES