# Latent Class Analysis to Identify Novel Phenotypes in Exacerbations of COPD: A Retrospective, Multicenter Cohort Study

## TL;DR

This study identifies six new COPD exacerbation subgroups using biomarkers, revealing significant differences in patient outcomes and aiding precision health management.

## Contribution

This is the first study to identify six ECOPD phenotypes using latent class analysis and 12 biomarkers in a large multicenter cohort.

## Key findings

- Six distinct ECOPD phenotypes were identified using 133 biomarkers and penalized Cox models.
- Phenotype 6 showed significantly higher all-cause mortality risk compared to Phenotype 1.
- Phenotypes are associated with unique clinical features like respiratory failure and cardiovascular disease.

## Abstract

This study aimed to identify novel phenotypes in patients with exacerbations of chronic obstructive pulmonary disease (ECOPD) to enable precise management, as current phenotypic classifications show limited utility in predicting patient prognosis. By analyzing data from a robust, retrospective multicenter registry (n = 13,449) and leveraging 133 biomarkers with penalized Cox models, we developed a six‐phenotype latent class analysis model. Phenotype 1 is distinguished by elevated direct bilirubin (Dbil) and lactate dehydrogenase (LDH). Phenotype 2 features a higher percentage of lymphocytes (LYMPH_pct) and lower percentage of neutrophils (NEUT_pct). Phenotype 3 is marked by increased generalized cardiovascular disease (gCVD) and reduced NEUT_pct. Phenotype 4 is related to higher NEUT_pct and lower LYMPH_pct. Phenotype 5 is associated with a higher prevalence of gCVD and surgical trauma history. Phenotype 6 stands out for its higher rates of respiratory failure and elevated pulse at admission. Compared with Phenotype 1, patients in Phenotype 6 have a significantly higher risk of all‐cause mortality in both the development and validation sets, with adjusted hazard ratios of 2.06 (95% CI: 1.38–3.08) and 2.51 (95% CI: 1.43–4.04), respectively. These findings reveal novel ECOPD subgroups with significant prognostic differences, providing a crucial framework for implementing precision health management and improving patient outcomes.

•This is the first study to identify six phenotypes with specific characteristics using 12 biomarkers from a large‐scale, multicentre cohort of patients with ECOPD.

•We have identified six different subgroups of patients with ECOPD using an unsupervised latent class analysis model, which were associated with different patterns of clinical manifestations.

•We found that compared with patients in Phenotype 1, those in Phenotype 6 had a higher risk of death perhaps attributed to high prevalence of respiratory failure, pulmonary heart disease, and generalized cardiovascular disease.