Distinct physiological, transcriptomic, and imaging characteristics of asthma-COPD overlap compared to asthma and COPD in smokers

Abstract

Background: The clinical and pathological features of asthma and chronic obstructive pulmonary disease (COPD) can converge in smokers and elderly individuals as asthma-COPD overlap (ACO). This overlap challenges the diagnosis and treatment of the distinct aetiologies underlying these conditions.

Methods: We analysed 2453 smokers (≥10 pack-years), aged 45-80 years, from the Genetic Epidemiology of COPD (COPDGene) Study, stratified as Control, Asthma, COPD, and ACO based on Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria. A comprehensive assessment was performed, encompassing symptomatology, pulmonary function tests (PFTs), complete blood counts (CBCs), bulk RNA sequencing (RNA-seq), and high-resolution quantitative computed tomography (QCT) imaging to evaluate clinical impact, lung function, systemic inflammation, and structural alterations contributing to disease progression across respiratory phenotypes. Differential expression (DE) analysis was performed using whole blood RNA-seq (BH-corrected FDR < 0.01), followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. Group differences were assessed using the Mann-Whitney U-test (MWU) or Chi-squared test (χ²), with Bonferroni correction applied for multiple comparisons. Multivariate linear regression models were used to adjust the associations between disease status and specific clinical outcomes for confounders, with one-way ANOVA and Tukey's Honest Significant Difference (HSD) post-hoc test applied for pairwise comparisons. Our analysis aimed to delineate the extent and variability of clinical features among disease phenotypes to guide targeted therapeutic strategies.

Findings: Our study highlights distinct yet overlapping profiles across ACO, asthma, and COPD. We effectively isolated disease-specific mechanisms by comparing each phenotype to smoking controls (GOLD 0) while accounting for baseline smoking-related inflammation. ACO exhibited the most severe symptom burden, with significantly higher COPD Assessment Test (CAT) score (18.32, 95% CI: [17.02, 19.63], P < 0.0001) and Modified Medical Research Council (mMRC) Dyspnea score (2.14, 95% CI: [1.92, 2.35], P < 0.0001) compared to COPD and asthma. ACO also displayed reduced lung capacity (forced expiratory volume in 1 s [FEV₁]: 52.5%, 95% CI: [50.08, 54.93], P < 0.0001) and airflow limitation (FEV₁/forced vital capacity [FVC]: 0.55, 95% CI: [0.5471, 0.5546], P < 0.0001), closely resembling COPD but significantly worse than asthma. The inflammatory profile of ACO exhibited a mixed response, featuring elevated neutrophil counts (4.57 K/μL, 95% CI: [4.28, 4.86], P < 0.0001) and eosinophil levels (0.22 K/μL, 95% CI: [0.20, 0.25], P < 0.01), contrasting with the predominantly neutrophilic inflammation in COPD and the absence of systemic inflammation in asthma. Structurally, ACO demonstrated significant airway remodelling (Pi10: 2.87, 95% CI: [2.83, 2.91], P < 0.0001), intermediate emphysema (5.66%, 95% CI: [4.72, 6.60], P < 0.0001), and moderate small airway disease (parametric response mapping for functional small airway disease [PRM^fSAD]: 22.94%, 95% CI: [21.53, 24.34], P < 0.0001), reflecting features of both asthma and COPD. COPD was characterised by more extensive emphysema (8.9%, 95% CI: [8.34, 9.45], P < 0.0001), small airway disease (PRM^fSAD: 27.09%, 95% CI: [26.51, 27.66], P < 0.0001), and gas trapping (37.34%, 95% CI: [36.33, 38.35], P < 0.0001), alongside moderate airway remodelling. At a molecular level, DE analysis revealed enrichment of the Hypoxia-Inducible Factor 1 (HIF-1) pathway in ACO, highlighting unique hypoxia-driven metabolic adaptations, while COPD was associated with neutrophil extracellular trap (NET) formation and necroptosis. In contrast, asthma exhibited significant airway remodelling (Pi10: 2.09, 95% CI: [2.05, 2.13], P < 0.0001), minimal parenchymal damage, and no systemic gene expression changes.

Interpretation: Collectively, our findings underscore the lung function impairments, systemic inflammation, molecular mechanisms, and structural correlates distinguishing ACO from COPD and asthma, emphasising the need for precise clinical management and the potential for novel therapeutic interventions.

Funding: This work was supported by National Heart, Lung, and Blood Institute (NHLBI) grants U01 HL089897 and U01 HL089856, as well as by National Institutes of Health (NIH) contract 75N92023D00011. Additional support was provided by grants R01 HL166231 (C.P.H.) and K01 HL157613 (A.S.).

Keywords: Asthma; Asthma-COPD overlap; COPD; COPDGene; Phenotypic characterisation; Pulmonary function test (PFT); Quantitative computed tomography (QCT).

Fig. 1

Comprehensive characterisation of symptomatology and exacerbation patterns in the COPDGene Cohort. (a) Distribution of total participants across control (N = 1365), asthma (N = 283, including those with overlapping COPD), COPD (N = 981, including those with overlapping asthma), and ACO (N = 176) clinical phenotypes (top). The left donut chart displays the proportion of participants classified within distinct clinical phenotypes, while the Venn diagram on the right depicts the overlap between asthma and COPD diagnoses, delineating the ACO group. (b) Clinical disease classification criteria based on GOLD stages and asthma diagnosis history, used to differentiate the phenotypes. (c–h) Comparative analysis of symptom severity and exacerbation patterns across clinical phenotypes. Bar plots displaying symptom severity as measured by (c) CAT scores and (d) mMRC Dyspnea scores across different participant groups. Bar plots showing the proportion of participants experiencing (e) chronic bronchitis, (f) shortness of breath attacks, (g) frequent exacerbations (requiring steroids and/or antibiotics per year), and (h) severe exacerbations (requiring emergency room visits or hospitalisations). Sample sizes for each group are indicated above the bars, and error bars represent 95% confidence intervals. Statistical analyses were conducted using the Mann–Whitney U-test for (c)and(d) and Chi-squared test for (e–h), with Bonferroni correction applied for multiple comparisons. Statistical significance is annotated as: ns – not significant, ∗ $P\le 0.05$, ∗∗ $P\le {10}^{-2}$, ∗∗∗ $P\le {10}^{-3}$, ∗∗∗∗ $P\le {10}^{-4}$.

Fig. 2

Comparative analysis of pulmonary function tests and complete blood counts in COPDGene Cohort. a–d Boxplot representing post-bronchodilator (post-BD) lung function parameters: (a) Percentage of predicted FEV₁, (b) FEV₁/FVC ratio (adjusted for confounders), (c) Percentage of predicted TLC, and (d) Percentage of predicted DL_CO. Panels e–h illustrate blood cell counts: (e) WBC, (f) Neutrophils, (g) Eosinophils, and (h) Monocytes. Each boxplot denotes the interquartile range (IQR), with the median value represented by the horizontal line within the box. Whiskers extend to the furthest data points that are not considered outliers, and individual outliers are represented as dots. Mann–Whitney U test was used for panels (a), (c), (d), and (e–h), with Bonferroni correction applied for multiple comparisons. For the FEV₁/FVC ratio, adjusted for confounders, in panel (b), pairwise comparisons were performed using ANOVA followed by Tukey's HSD test. Sample sizes for each group are indicated above the boxplots, with significance levels annotated as follows: ns for non-significant, ∗ $P\le 0.05$, ∗∗ $P\le {10}^{-2}$, ∗∗∗ $P\le {10}^{-3}$, ∗∗∗∗ $P\le {10}^{-4}$.

Fig. 3

Functional enrichment and Gene Expression Profiling in COPD and ACO. (a) Tree plot showing Gene Ontology (GO) enrichment analysis of biological process annotations in COPD, displaying the top 10 most statistically significant GO terms, clustered into three primary categories. The size of dots represents the gene count of differentially expressed genes (DEGs) associated with each GO term, while the colour indicates -log₁₀ FDR value. (b) Tree plot of GO term enrichment analysis for ACO, displaying the top 10 significant categories of GO terms, organised into three distinct clusters. (c) Circos plot showing the association between upregulated DEGs in COPD and top 5 GO biological processes. The colour gradient indicates the log₂ fold change for each gene. (d) Circos plot illustrating the mapping of upregulated DEGs in ACO to the top 5 GO categories. In panels (a–d), top categories in COPD and ACO are coloured by the rank based on statistical significance rather than nomenclature, as they may correspond to different gene sets. (e) Box plots showing the expression levels of upregulated genes involved in cytokine production in COPD and ACO. (f) Box plots depicting the expression levels of upregulated genes involved in T-cell activation in COPD and ACO. (g) Box plots highlighting the differential expression of upregulated inflammatory and immune response genes in COPD and ACO. Statistical significance is denoted by asterisks (ns = not significant, ∗ $P\le 0.05$, ∗∗ $P\le {10}^{-2}$, ∗∗∗ $P\le {10}^{-3}$, ∗∗∗∗ $P\le {10}^{-4}$). This figure illustrates the distinct molecular signatures of immune response in COPD and ACO.

Fig. 4

Pathway activation in ACO and COPD identified through KEGGpathway enrichmentanalysis. Pathway maps displaying (a) The HIF-1 signalling pathway enriched in ACO. (b) Necroptosis pathway enriched in COPD. In (a and b), upregulated pathway components are marked in pink, while downregulated components are marked in green. These diagrams depict the molecular mechanisms underlying hypoxic adaptation in ACO and cell death regulation in COPD, providing insights into the distinct and overlapping biological processes that contribute to the pathophysiology of these conditions.

Fig. 5

Comparative analysis of quantitative CT scan-derived pulmonary metrics across different respiratory conditions. Boxplots displaying (a) percentage of emphysema determined by low attenuation areas (<−950 HU), (b) lung density measured at the 15th percentile (Perc15), (c) Parametric Response Mapping for small airways disease (PRM^fSAD), (d) gas trapping quantified by low attenuation at −856 HU, (e) Pi10 measurement, reflecting standardised airway wall thickness, (f) airway wall area percentage as an indicator of airway remodelling. Each boxplot represents the interquartile range, with median values marked by horizontal lines. Whiskers extend to the most extreme data points not considered outliers, and outliers are plotted as distinct points. The sample size for each disease phenotype is noted above the respective boxplots. Statistical significance was determined using ANOVA followed by Tukey's HSD test, with significant differences between disease groups annotated as follows: ns = not significant, ∗ $P\le 0.05$, ∗∗ $P\le {10}^{-2}$, ∗∗∗ $P\le {10}^{-3}$, ∗∗∗∗ $P\le {10}^{-4}$.

Fig. 6

Schematic overview of respiratory pathologies in asthma, COPD, and ACO,depictingthelung regionsaffectedin each condition. The diagram illustrates the primary areas impacted by asthma (left), characterised by bronchial hyperresponsiveness and inflammation; COPD (right), marked by alveolar damage and airway obstruction; and ACO (centre), which exhibits features of both asthma and COPD, including airway inflammation and parenchymal destruction. This visual representation underscores the distinct and overlapping pathophysiological features of these chronic respiratory conditions. The schematic was generated using Biorender.com.