AI-Driven Prediction Unveils SMAD4-Specific HLA-Restricted Epitopes: A Novel Therapeutic Target for Precision Treatment of Peritoneal Metastasis in Colorectal Cancer

Background Peritoneal metastasis (CRC-PM), a lethal feature of advanced colorectal cancer, lacks effective treatments. Neoantigen-based immunotherapy has emerged as a promising therapeutic strategy for metastatic CRC. Methods Multi-omics including TIMER and TISIDB tools, two computational pipelines of NetMHCpan and GenoImmune's artificial intelligence (AI)-driven Neoantigen Techonology (GIANT) were employed to characterize somatic mutations in CRC-PM, focusing on SMAD4-derived neoantigens. Results Genomic analysis of 2,424 CRC and 124 CRC-PM patients revealed KRAS, SMAD4, BRAF, and RNF43 mutations enriched in CRC-PM. KRAS/SMAD4 alterations correlated with higher tumor mutational burden (TMB) and enhanced immune infiltration. Ten novel SMAD4 mutation sites were uniquely identified in CRC-PM, that were associated with lymphocyte populations and immune checkpoint (ICP) gene expression. Neoantigen prediction of SMAD4 mutations identified 20 and 48 MHC-I candidates, using NetMHCpan and GIANT, respectively., Dominant variants (R361C, R361H, D424V, and A118V) of SMAD4 mutations showed strong affinity for HLA-B*07:02, HLA-A*24:02, and HLA-B*40:01. MHC-II epitopes D424V and W524C prioritized HLA-DRB1*15:01/13:02. Conclusions These findings highlight SMAD4 as a recurrent immunogenic driver in CRC-PM, indicating its potential as a neoantigen vaccine target for precision immunotherapy.