癌症免疫循环中的断点与可预测应答的HNSCC免疫微环境表型相关

Introduction Head and neck squamous cell carcinoma (HNSCC) displays substantial immune heterogeneity that shapes response and prognosis to immune checkpoint inhibitor (ICI) therapy. Unlike biomarkers for response to ICI therapy, there is still no understanding of resistance mechanisms. To address this need, we performed baseline pathway enrichment analyses across TME subtypes and clinical outcomes. Using pairwise comparisons, we identified reproducible “breakpoints” within the Cancer Immunity Cycle (CIC) that inform future precision-therapeutic strategies. Methods Bulk RNA-sequences from TCGA HNSCCs HPV+ (n=94) and HPV- (n=415) and an our neoadjuvant ICI cohorts HPV+ (n=53) and HPV- (n=29) were assigned TME subtypes: Immune Enriched (IE), Immune Enriched Fibrotic (IEF), Fibrotic (F) and Depleted (D) using published signatures. Unbiased and supervised GSEA were performed, contrasting TME subtype pairs to identify enriched pathways mapping to CIC steps. Step activity was defined as Bioprocess Count × mean NES. Connectivity of CIRCOS plot chords, denoting interplay between steps, was quantified by shared processes × mean STRING signal for protein-protein interaction. CIRCOS plots were generated for TCGA and trial cohorts. Survival analyses used Cox and Kaplan-Meier models (p<0.05). Results HPV+ tumors exhibited a global reduction of stepwise CIC activities compared with HPV-tumors. As expected, CIC mapping revealed consistent low activity in antigen release (step 1) and antigen presentation (step 2), with variable loss of tumor cell killing (step 7) in F/D tumors. IE tumors maintained near-intact steps, IEF showed partial attenuation, whereas F showed breaks in steps 1-2, and D activity was minimal. CIC analysis revealed novel subpopulations within a given TME subtype, differentiated by survival. TCGA survival analyses showed statistically significant higher adaptive immune/TCR activity improved outcomes in IE tumors. In HPV- F tumors, a senescence/autophagy gene set linked to CIC predicted worse progression-free survival (HR=7.06, p=0.03). In contrast, HPV+ D tumors exhibited activity of a senescence gene set strongly associated with improved OS (HR=0.209, p<0.001). Phenotype-level survival (HPV-agnostic) showed significantly worse outcomes for F vs IE tumors. These patterns were reproducible in our clinical cohort. Future pathologic response to ICI mirrored CIC integrity, with higher response rates in IE/IEF, but did not reach statistical significance. Conclusion CIC modeling across independent cohorts reveals recurrent immune breakpoints in HNSCC, particularly in antigen release and presentation. This novel approach to parsing the TME and CIC activity may be valuable for predictive biomarker discovery, pertaining to response to ICI therapy.