Integrated multi-omics unraveled a basal cell-dependent signature predictor of outcome and immunotherapy response in LUSC

Lung squamous carcinoma (LUSC) lacks clear etiology and effective biomarkers. Building on the hypothesis that basal epithelium transformation seeds LUSC, and recognizing the impact of immunotherapy, we integrated transcriptomic, proteomic, genomic, epigenetic, and single-cell RNAseq data to derive basal cell–dependent signatures predicting therapy response. In TCGA-LUSC we performed unsupervised clustering using hub basal and immune genes, defining two subtypes. “Basalimmune-high” showed higher immune scores, immune infiltration, and superior predicted response to immunotherapy; “Basalimmune-low” exhibited higher proliferative basal scores and elevated KRT5/6 and TP63. Epigenetically, KRT5/6 methylation inversely correlated with RNA expression in Basalimmune-high, while TP63 methylation was unchanged between groups. Clinically, Basalimmune-high was associated with worse overall survival under standard chemotherapy. A 15-gene prognostic signature emerged and was validated across four independent LUSC cohorts using transcriptomic and proteomic data. From top upregulated genes, we prioritized serum proteins and, in a real-world treatment-naïve cohort (stage I–IV, n=100), CXCL5 and CXCL11 concentrations recapitulated the clusters and identified poorer prognosis in Basalimmune-high (p