Therapeutic vulnerability shaped by the microenvironment: multi-omics and AI biomarkers for precision surgical planning in gastrointestinal tumors

Background Therapeutic vulnerability in gastric cancer is profoundly influenced by the tumor microenvironment (TME), yet reliable and clinically actionable preoperative indicators remain insufficient. Methods We developed and validated an artificial intelligence–driven multi-omics TME score (DLRS/TMEscore) by integrating CT-derived imaging features with transcriptomic, immunohistochemical, and molecular profiling. The score was evaluated for its associations with survival outcomes, benefit from adjuvant chemotherapy, and response to anti–PD-1 therapy. Results The DLRS/TMEscore reproducibly stratified disease-free and overall survival across independent cohorts. Patients in the low-risk subgroup derived substantial benefit from adjuvant chemotherapy, whereas those in the high-risk subgroup demonstrated attenuated benefit. Among individuals receiving immunotherapy, the score enriched objective responders and predicted more durable clinical outcomes, outperforming established biomarkers including PD-L1 combined positive score (CPS) and microsatellite instability (MSI). In addition, DLRS/TMEscore correlated with multiple surgical parameters, such as operative complexity, resection margin status, nodal involvement, and postoperative recovery, indicating relevance in perioperative risk assessment. Conclusion This AI-enabled multi-omics framework offers a robust and interpretable approach for characterizing microenvironment-defined therapeutic vulnerability, supporting preoperative risk stratification and individualized systemic treatment strategies in gastric cancer.