Stromal remodeling and neoantigen clonality as used to evaluate benefit from SIRT–nivolumab in HCC.

e16285 Background: The combination of selective internal radiation therapy (SIRT) and nivolumab has shown encouraging clinical activity against hepatocellular carcinoma (HCC). However, the molecular determinants of clinical benefit (CB) and the mechanisms underlying therapeutic resistance remain poorly defined. This study aimed to characterize the genomic, transcriptomic, and spatial features that govern response to this combination. Methods: In the NASIR-HCC phase II trial, patients with unresectable HCC received SIRT followed by nivolumab. Baseline tumor biopsies were analyzed using whole-exome sequencing, bulk RNA sequencing, multiplex immunofluorescence, and spatial proteomics. T-cell receptor (TCR) dynamics were monitored in both tumor tissue and peripheral blood. Results: Unlike monotherapy with immune checkpoint inhibitors, pre-existing immune infiltrates and inflammatory signatures did not predict CB. Instead, non-responders exhibited a significant enrichment in fibroblast-related programs, including epithelial-to-mesenchymal transition (EMT) and extracellular matrix remodeling. Spatial proteomics revealed that activated lymphocytes in non-responders were confined within fibrotic stromal regions, suggesting a physical barrier to effective anti-tumor immunity. Transcriptomic integration identified oncofetal cancer-associated fibroblasts (CAFs) and macrophage signatures as hallmarks of resistance. Conversely, patients experiencing CB showed enrichment in oxidative phosphorylation pathways and specific neoantigen (NeoAg) profiles. While total tumor mutational burden (TMB) was uninformative, CB correlated with a higher load of expressed, high-affinity Class II NeoAgs and a more dominant clonal architecture (higher Gini and Clonality indices). Furthermore, SIRT induced diversification of the peripheral TCR repertoire and the maintenance of this effect after nivolumab associated with CB and longer progression-free and overall survival. Conclusions: The efficacy of SIRT plus nivolumab in HCC does not appear to be primarily determined by baseline T cell infiltration and may instead be influenced by fibroblast-driven stromal features and oncofetal CAF-associated immunosuppression. Better outcomes are expected in the presence of high-affinity Class II clonal NeoAgs and with sustained systemic T cell diversification. These findings indicate that stromal remodeling and fibroblast reprogramming could represent promising avenues to enhance responses to SIRT–immunotherapy combinations, although further validation is required.