Peritumoral adipose tissue as a prognostic imaging biomarker for immunotherapy response in HNSCC

Introduction: Obesity is a well-known risk factor for many cancers, including head and neck squamous cell carcinoma (HNSCC), yet greater systemic adiposity has been linked to improved responses to immune checkpoint inhibitors (ICIs). Increased adipose tissue modulates the tumor microenvironment via immune cell infiltration and PD-1 signaling. While prior work has focused on systemic obesity, the role of peritumoral adipose tissue remains unexplored. How peritumoral adiposity impacts ICI response may reveal mechanisms driving immunotherapy efficacy. Methods: We retrospectively analyzed 36 HNSCC patients treated with neoadjuvant and adjuvant pembrolizumab concurrent with adjuvant radiation +/- chemotherapy. Tumors were segmented on pre-treatment CT scans using 3D Slicer, and a 1 cm peritumoral shell was generated in 24 patients with available imaging. Adipose volumes were quantified using patient specific Hounsfield unit thresholds: peritumoral adipose fraction (peritumoral adipose volume / peritumoral volume) and adipose:tumor ratio. Pathologic response was determined histologically. Group differences were tested using Wilcoxon rank-sum and Fisher's exact tests. Survival was censored at 2 years with Kaplan-Meier and log-rank testing. Adipose-volume correlations used Spearman's rho. In a subset of patients with paired tissue samples pre- and post-pembrolizumab, RNA-seq assessed differential pathway enrichment. Gene set enrichment analysis (GSEA) was performed using curated Reactome and KEGG pathways. Results: Among 24 patients, median peritumoral adipose fraction was 0.017 (IQR 0.026), median adipose:tumor ratio was 0.096 (IQR 0.194), and the median tumor volume was 8.5 cm 3 (IQR 23.3). Pathologic responders had higher peritumoral adiposity compared to non-responders (median 0.0314 vs 0.0076; Wilcoxon rank-sum, p = 0.026). When dichotomized by the median, higher peritumoral adiposity was associated with pathologic response (Fisher's exact test, OR = 8.9, 95% CI 1.1-132.4, p = 0.036). Kaplan-Meier analysis showed a significantly improved survival in the high-adipose group (2-year OS 100% vs 58%; log-rank p = 0.0098). Transcriptomic analysis identified ~60 genes changed significantly pre- to post-pembrolizumab in responders (adjusted p<0.1). GSEA demonstrated that responders had significant activation of pathways involving TGF-beta receptor signaling, regulation of gene expression by HIF1A, TRAF6-mediated cytokine induction, and innate and adaptive immune signaling. Conclusion: Greater peritumoral adiposity was associated with improved pathologic response and 2-year survival in HNSCC patients treated with pembrolizumab. Transcriptomic analysis reveals activation of TGF-beta, HIF1A, and immune signaling pathways in responders, suggesting peritumoral adipose tissue may shape a treatment-permissive microenvironment that enhances response to PD-1 blockade.