Related Experiment Video
Updated: Jul 1, 2026

Advanced Imaging of Lung Homing Human Lymphocytes in an Experimental In Vivo Model of Allergic Inflammation Based on Light-sheet Microscopy
Published on: April 16, 2019
Blood transcriptomic endotyping of COPD identifies a neutrophil-driven inflammatory endotype reflected by the
Youjin Song1,2, WooJin Kim3, Ji Ye Jung4
1Department of Biomedical Informatics, Korea University College of Medicine, Seoul, Republic of Korea.
Background:
Chronic obstructive pulmonary disease (COPD) is molecularly heterogeneous, yet clinically deployable approaches for blood-based endotyping remain limited. We aimed to identify blood transcriptomic COPD endotypes and evaluate whether the neutrophil-to-lymphocyte ratio (NLR), derived from a routine complete blood count, is associated with a neutrophil-driven inflammatory endotype and could serve as a clinically accessible surrogate biomarker.
Methods:
Whole-blood RNA sequencing was performed in a prospective multicentre Korean COPD cohort. After quality control, 492 samples underwent unsupervised consensus clustering. The final analysable cohort comprised 481 participants with valid NLR values and complete spirometry, including 285 with confirmed COPD (post-bronchodilator FEV₁/FVC < 0.70) and 196 symptomatic at-risk individuals (post-bronchodilator FEV₁/FVC ≥ 0.70 with respiratory symptoms). Clinical, radiologic and laboratory features were compared across clusters with Benjamini-Hochberg false discovery rate correction. Associations between NLR and the inflammatory K1 endotype were assessed using multivariable logistic regression with multiple imputation by chained equations and receiver operating characteristic analyses with Harrell optimism-corrected bootstrap resampling. External validation of the K1 signature and the NLR-K1 association was performed in the COPDGene cohort (n = 3,785).
Results:
Four reproducible blood transcriptomic endotypes were identified: K1 (23.5%), a neutrophil-driven inflammatory subtype enriched for IL-6/TLR5-NFkB signalling; K2 (18.9%), a hypoxia-vascular subtype; K3 (47.4%), a metabolically preserved subtype; and K4 (10.2%), a tissue remodelling subtype. NLR strongly identified the inflammatory K1 endotype (adjusted OR 2.89, 95% CI 2.23-3.74; optimism-corrected AUC 0.86, 95% CI 0.82-0.89).Performance was highest in symptomatic at-risk individuals without fixed airflow obstruction (AUC 0.90, 95% CI 0.85-0.95). A two-threshold framework derived from the discovery cohort (NLR < 1.76 for rule-out; NLR ≥ 2.58 for rule-in) showed strong transferability in COPDGene, where NLR achieved an AUC of 0.88 (95% CI 0.87-0.89) for signature-derived K1. Higher NLR was associated with neutrophilic markers including IL1R2 and MMP9, supporting the biological coherence of the NLR-K1 relationship.
Conclusions:
The NLR was strongly associated with a transcriptomically defined neutrophil-driven COPD endotype and may provide a clinically accessible surrogate marker for inflammatory stratification using routine blood testing. These findings support the translational potential of blood-based endotyping but remain hypothesis-generating pending prospective outcome-based validation.