A risk score combining co-expression modules related to myeloid cells and alternative splicing associates with response to PD-1/PD-L1 blockade in non-small cell lung cancer

Background Comprehensive analysis of transcriptomic profiles non-small cell lung cancer (NSCLC) may provide novel evidence for biomarkers associated with response to PD-1/PD-L1 immune checkpoint blockade (ICB). Methods We utilized weighted gene co-expression network (WGCNA) analyze data from two NSCLC datasets Gene Expression Omnibus (GSE135222 and GSE126044) that involved patients received ICB treatment. evaluated the correlation modules responsiveness functionally annotated ICB-related using pathway enrichment analysis, single-cell RNA sequencing, flow cytometry alternative splicing analysis. built a risk score Lasso-COX regression based on hub genes modules. investigated alteration tumor microenvironment between high- low- groups association previously established predictive biomarkers. Results Our results identified black positive blue module negative responsiveness. The was enriched in T activation antigen processing presentation, assigned it were consistently expressed myeloid cells. observed decreased events samples high signature scores module. screened out three ( EVI2B , DHX9 HNRNPM ) constructed validated value poor therapy an in-house cohort pan-cancer KM-plotter database. low-risk group had more immune-infiltrated microenvironment, higher frequencies precursor exhausted CD8 + cells, tissue-resident plasmacytoid dendritic cells type 1 conventional lower frequency terminal which explain its superior therapy. significant tertiary lymphoid structure also implicated possible mechanism this biomarker. Conclusions study related developed accordingly, could potentially serve as biomarker response.