Tumor Immune Contexture Model Predicts Prognosis and Immunotherapy Response in Gastric Cancer.

backgroundGastric cancer (GC) is highly heterogeneous, and current prognostic models fail to fully capture tumor immune characteristics, limiting personalized treatment. This study introduces the Tumor Immune Environment Score (TIES), an immune-based prognostic model designed to enhance risk stratification and predict response to immunotherapy.

methodsTranscriptomic and clinical data from seven GC cohorts, comprising a total of 1487 patients, were analyzed. The training cohort included GSE15459, GSE62254, GSE84433, and GSE13861, while the validation cohort comprised GSE26899, GSE26901, and TCGA-STAD. Immune-related gene signatures were quantified using ssGSEA and subsequently analyzed via LASSO and Cox regression. Clinical applicability was quantified using decision-curve analysis (DCA) and reclassification metrics (category-free NRI/IDI) for TIES versus TIES + pathological stage. Immunotherapy response was assessed using data from GSE183924, and its associations with immune characteristics, molecular alterations, and drug sensitivity were investigated. For experimental confirmation, we established a 12-gene RT-qPCR panel in 30 institutional GC specimens and computed a weighted qPCR surrogate (qTIES).

resultsA high TIES was associated with worse OS (HR = 2.44, p < 0.001) and DFS (HR = 2.25, p < 0.001). TIES was an independent prognostic factor and demonstrated superior predictive accuracy compared to TNM staging (AUC = 0.75-0.79 vs. 0.61-0.64, p < 0.001). Integrating clinical-pathological features with TIES significantly enhanced predictive performance (AUC = 0.75-0.78 vs. 0.61-0.63, p < 0.01). A low TIES was indicative of an immune-inflamed subtype, characterized by a higher TMB (p < 0.01) and a greater likelihood of response to immunotherapy (HR = 0.24, p = 0.014). TIES outperformed TIDE in predicting immunotherapy outcomes, achieving a higher OS C-index (0.780 vs. 0.743) and DFS C-index (0.664 vs. 0.610). At 3 years, TIES + stage yielded a greater net benefit than TIES alone across clinically relevant thresholds (DCA) and improved NRI/IDI.The qPCR panel reproduced axis biology (CSR/TGF-βincreased with NK/Th2 decreased), correlated with pathological stage, stratified OS/DFS by Kaplan-Meier, and showed 3-year discrimination.

conclusionTIES represents a robust immune-based prognostic tool for GC, facilitating improved risk stratification and immunotherapy response prediction. Its integration into clinical practice, particularly in combination with clinical-pathological features, may enhance personalized treatment strategies and improve patient outcomes.