WCLC 2025 - Posters & ePosters-P1.17.23 Role of Lung Microbiome on Precise Diagnosis of Patients With Checkpoint Inhibitor Pneumonia: Result From a Prospective Cohort Study

P1.17.23 Role of Lung Microbiome on Precise Diagnosis of Patients With Checkpoint Inhibitor Pneumonia: Result From a Prospective Cohort Study

Back to course

Pdf Summary

This prospective cohort study led by Dr. Zhenhua Zhou investigates the role of the lung microbiome in accurately diagnosing Checkpoint Inhibitor Pneumonia (CIP), a serious immune-related adverse event occurring in 3%-10% of patients treated with checkpoint inhibitors and associated with a high mortality rate in severe cases. Differentiating CIP from pulmonary infection (PI) poses a significant clinical challenge.<br /><br />Utilizing metagenomic next-generation sequencing (mNGS) of bronchoalveolar lavage fluid (BALF), the study characterized lung microbiota profiles in two CIP subtypes—pure-type CIP (PT-CIP) and mixed-type CIP (MT-CIP)—as well as PI patients. The cohort included 38 patients with comparable demographic and clinical features, except for smoking history differences.<br /><br />Results revealed distinct microbial signatures associated with CIP and PI. PT-CIP was dominated by Streptococcus, Porphyromonas, Rothia, and Prevotella, whereas PI patients had microbiomes enriched with Candida, Prevotella, Rothia, and Streptococcus. Notably, Candida abundance correlated with systemic immune-inflammatory markers, suggesting an interplay between microbial composition and host immunity. Functional enrichment analyses indicated metabolic pathway differences between CIP and PI groups.<br /><br />Leveraging these data, the researchers developed a machine learning diagnostic model based on lung microbiome profiles. A decision tree algorithm incorporating key microbial features, including Funiculosus and Forsythia, achieved high diagnostic accuracy for CIP with an area under the curve (AUC) of 0.88.<br /><br />In sum, this study is the first to delineate lung microbial landscapes in CIP subtypes and PI, linking them to immune-inflammatory dynamics. The findings underscore the clinical potential of microbiome-based machine learning tools to enhance precision diagnosis of CIP, providing critical insights to improve patient management and understanding of CIP pathophysiology.<br /><br />The study was supported by grants from the National Natural Science Foundation of China and Guangdong Basic and Applied Basic Research Foundation.

Asset Subtitle

Zhenhua Zhou

Meta Tag

Speaker Zhenhua Zhou

Topic Global Health, Health Services, and Health Economics

Keywords

lung microbiome

Checkpoint Inhibitor Pneumonia

CIP diagnosis

metagenomic next-generation sequencing

bronchoalveolar lavage fluid

pure-type CIP

mixed-type CIP

pulmonary infection

machine learning diagnostic model

immune-inflammatory markers