Multi-omics identification of MBNL2 associated with poor pathological response to neoadjuvant chemoimmunotherapy in...

Neoadjuvant chemoimmunotherapy (NACI) improves outcomes in resectable lung squamous cell carcinoma (LUSC), yet response varies widely and current biomarkers lack precision. Novel correlates of immunotherapy sensitivity tailored to the LUSC tumor microenvironment (TME) are urgently needed.

Using TCGA-LUSC transcriptomic data, we constructed a 25-gene prognostic model and applied three machine learning algorithms in combination with the Tumor Immune Dysfunction and Exclusion (TIDE) algorithm to identify core genes linked to prognosis and immunotherapy response. Immune infiltration and enrichment analyses were performed to characterize the TME. An independent pre-NACI biopsy cohort (n=36) was used for histopathological validation to explore correlations with pathological response, while single-cell RNA-seq (GSE207422) and CellChat were used to infer tumor-stromal crosstalk and explore underlying mechanisms.

The risk score independently stratified prognosis. Among three core genes, high MBNL2 expression was associated with higher TIDE scores, lower TIDE-predicted response rates, and elevated cancer-associated fibroblast (CAF) scores. scRNA-seq revealed systematically enhanced communication between MBNL2-high tumor cells and FAP⁺ CAFs, with unique ligand-receptor pairs enriched in WNT and EGF pathways; FAP⁺ CAFs interacted with regulatory T cells via ECM remodeling and the MDK-NCL axis. Histopathological validation confirmed that low tumor-cell MBNL2 expression correlated with higher pathological response and pCR rates, reduced FAP, and decreased FOXP3 expression.

This study establishes a 25-gene prognostic model for LUSC and identifies MBNL2 as a novel correlate of poor pathological response to NACI. Elevated MBNL2 expression in tumor cells is associated with enhanced tumor-CAF crosstalk, CAF activation, and an immunosuppressive TME, laying a foundation for future mechanistic investigation.