Deciphering the Heterogeneous Microenvironment of Head and Neck Squamous Cell Carcinoma Through an Integrated Immune Inflammation Framework.

Background: Head and neck squamous cell carcinoma (HNSC) exhibits substantial prognostic and microenvironmental heterogeneity. However, the integrated prognostic relevance of synergistic immune and inflammatory signatures in HNSC remains fully elucidated. Methods: Weighted gene coexpression network analysis (WGCNA) was integrated with curated immune- and inflammation-related gene sets to identify key tumor-associated candidate genes. A crucial phenotypic module exhibiting the strongest positive correlation with tumor status was prioritized, yielding six overlapping candidate genes. Utilizing the TCGA-HNSC, GSE65858, and GSE41613 cohorts, we systematically compared multiple machine learning algorithms to construct a robust immune-inflammation score (IIS), subsequently evaluating its prognostic efficacy and biological relevance. Results: The random survival forest model outperformed other algorithms and was utilized to establish the IIS. An elevated IIS was consistently predictive of inferior survival and served as an independent prognostic indicator. Furthermore, the IIS significantly correlated with specific immune infiltration patterns, immune checkpoint expressions, TIDE-related features, tumor microenvironment scores, and distinct genomic mutation profiles, including tumor mutation burden. Notably, CSF2, IL1R2, and IL20RB were identified as pivotal model constituents, displaying cell type-specific and spatially discrete expression trajectories. Conclusions: The proposed IIS constitutes a robust, clinically relevant prognostic biomarker for HNSC, capturing the profound immune and genomic heterogeneity inherent in the disease.