Single-cell hdWGCNA identifies checkpoint inhibitor pneumonitis-linked CD4 + T-cell subsets driving immunotherapy response and prognosis in lung cancer

Background Immune checkpoint inhibitors (ICIs) have achieved promising efficacy in lung cancer patients, yet checkpoint inhibitor-related pneumonitis (CIP) remains a severe and incompletely understood complication. The pathogenesis of CIP and its impact on immunotherapy outcomes remain elusive, especially the role of specific immune cell subsets in mediating these processes. Methods Single-cell RNA sequencing (scRNA-seq) data from CIP patients were analyzed to delineate the distinct immune microenvironment landscape. High-dimensional weighted gene co-expression network analysis (hdWGCNA) was performed to identify CIP-related CD4 + T cell gene modules. Key results were validated in an independent lung adenocarcinoma (LUAD) cohort, and a prognostic signature was constructed. Immunofluorescence staining was applied to verify the existence of the identified T-cell subsets in CIP-lung tissues. Results Our analysis uncovered prominent CD4 + T cell infiltration in the CIP immune microenvironment. hdWGCNA identified key CIP-associated CD4 + T cell genes enriched in pathways including viral response, cytokine production and T cell differentiation. Validation in the LUAD cohort verified robust associations between these CD4 + T cell subsets and both patient prognosis and immunotherapy response. The constructed prognostic signature exhibited favorable predictive performance, with BIRC3 identified as the leading risk gene. Immunofluorescence staining confirmed the spatial localization of BIRC3 + CD4+ T cells in CIP patients. Conclusion This study identifies critical CD4 + T cell subpopulations mediating CIP pathogenesis and establishes a robust gene signature for predicting survival and ICI response in LUAD patients. Our findings provide mechanistic insights into CIP progression and offer potential therapeutic targets for the clinical management of this life-threatening immunotherapy complication.