APSR2026-925

 
Late-Breaking
Submitted
Oral Presentation
AO34-002
 
 
https://storage.unitedwebnetwork.com/files/1285/1241178-46903-AO34-002 Yanping Yang_440026.pptx
Format: PPTX Only
 
Format: PDF Only
08. Lung Cancer
4

No more than 15 authors can be listed.
Please ensure the authors are listed in the right order.
If there are multiple affiliations, please use the "+" button next to "Select Country/Region" to add additional affiliation fields.

Yanping Yang * Fudan University Shanghai Institute of Infectious Disease and Biosecurity Shanghai China Zhongshan Hospital, Fudan University Department of Pulmonary and Critical Care Medicine Shanghai China
Shujin Li - Zhongshan Hospital, Fudan University Department of Pulmonary and Critical Care Medicine Shanghai China
Nuo Xu - Zhongshan Hospital, Fudan University Department of Pulmonary and Critical Care Medicine Shanghai China
Yuanlin Song - Fudan University Shanghai Institute of Infectious Disease and Biosecurity Shanghai China Zhongshan Hospital, Fudan University Department of Pulmonary and Critical Care Medicine Shanghai China
 
 
 
 
 
 
 
 
 
 
 
Presenting Author Details
Yanping
Yang
yyp1ngg@163.com
+8613647156949
Under 30
China
Original Research
Alterations of BALF Microbiota and Machine Learning-Based Prediction of Checkpoint Inhibitor Pneumonitis in Lung Cancer Patients
Maximum 25 words
Abstract Content
Checkpoint inhibitor pneumonitis (CIP) is a clinically challenging immune-related adverse event associated with immune checkpoint inhibitor therapy. The role of the pulmonary microbiota in CIP pathogenesis remains poorly understood. This study aimed to characterize the bronchoalveolar lavage fluid (BALF) microbiota composition and assembly in CIP patients, explore its functional alterations, and develop machine learning models for CIP prediction.
A total of 103 patients were enrolled and divided into CIP and Non-CIP groups. BALF samples were collected and subjected to next-generation sequencing (NGS). Microbial diversity, community assembly, taxonomic composition, and functional profiles (GO enrichment) were comprehensively analyzed. Seven machine learning algorithms were evaluated for their ability to distinguish CIP from Non-CIP patients based on microbial features.
The CIP group exhibited significantly lower alpha diversity and altered beta diversity compared to the Non-CIP group. Community assembly was governed predominantly by deterministic rather than stochastic processes. Taxonomically, CIP patients showed reduced abundances of Bacteroidota, Fusobacteriota, Prevotella, Streptococcus, Herbaspirillum, and Ralstonia, with depletion of species such as Prevotella melaninogenica and Ralstonia pickettii. Conversely, enrichment of Actinomyces and Mycolicibacter terrae was observed in CIP. Functional analysis revealed downregulation of key GO terms related to transcriptional regulation (GO:0006351), fatty acid oxidation (GO:0001681), pyruvate dehydrogenase phosphatase activity (GO:0016462), and negative regulation of apoptosis (GO:0043066) in CIP patients. Among the machine learning models, xGboost achieved the best predictive performance, with a mean accuracy of 0.770, AUC of 0.781, and F1 score of 0.764 at the species level combined with gene features.
CIP is associated with distinct alterations in BALF microbiota, characterized by reduced diversity, deterministic community assembly, and functional reprogramming involving transcriptional, metabolic, and apoptotic pathways. Machine learning models, particularly xGboost, integrated with BALF microbial features hold promise for non-invasive CIP risk assessment.
 
 
 
 
 
No
Yes
https://storage.unitedwebnetwork.com/files/1285/1241178-37089-Fig_397419.png
JPG or JPEG or PNG format ONLY
checkpoint inhibitor pneumonitis, machine Learning, microbial signatures
maximum 3 keywords and no abbreviations, separated by commas
The authors declare that they have no conflicts of interests.
292