Document resource
Background Immune checkpoint inhibitors such as pembrolizumab offer a therapeutic option for head and neck squamous cell carcinoma (HNSCC), yet only a minority of patients benefit. Although PD-L1 expression and the combined positive score (CPS) are used as biomarkers for patient selection, they poorly predict clinical response. There is a critical need to identify more robust, spatially-informed biomarkers that better capture the complexity of the tumor immune microenvironment (TME).Methods We analyzed tumor samples from 48 HNSCC patients prior to pembrolizumab treatment using FAST (fast cyclic immunofluorescence), a rapid multiplexed imaging platform. Whole-tissue sections were profiled for 21 protein markers across 6,316 fields of view, capturing nearly 3 million single cells. Computational pipelines were applied to segment, classify, and spatially map immune and tumor cell types. We quantified PD-L1 expression, cellular densities, and spatial interactions, and developed multiparametric spatial scores (MPSSs) to model predictors of treatment response and survival.Results PD-L1 expression, whether assessed by clinical CPS or enhanced single-cell profiling, showed limited correlation with treatment response. Spatial analysis revealed that the abundance and clustering of CCR7 +dendritic cells (DCs)—particularly in the tumor periphery—were significantly associated with therapeutic response and overall survival. These CCR7+ DCs often colocalized with T cells and macrophages, forming immune niches. Modeling revealed that MPSSs incorporating spatial metrics of CCR7+ DC density, clustering, and proximity to T cells outperformed single-feature metrics and CPS in predicting patient outcomes.Conclusions This study demonstrates that spatial organization of the TME, particularly CCR7 + DC niches, is a promising predictor of immunotherapy benefit in HNSCC. These findings highlight the value of high-resolution spatial profiling for biomarker discovery and suggest that CCR7+ DC clusters may serve as actionable biomarkers for response stratification in HNSCC and potentially other cancers.