Objective To develop a prognostic risk model based on immune features for precise stratification of gastric cancer(GC),with the aim of improving combined immunotherapy and re-vealing molecular features of tumor-immune interactions.Methods The immune landscape of GC was quantitatively evaluated using single-sample gene set enrichment analysis(ssGSEA),and pa-tients were classified into clusters with different immune infiltration abundances.The least abso-lute shrinkage and selection operator(LASSO)Cox regression analysis was then utilized to per-form stepwise regression on candidate genes to develop a risk model and conduct external valida-tion.Kaplan Meier analysis,receiver operating characteristic(ROC)curve,nomogram,calibration curve,etc.were performed to evaluate the predictive ability of the model.In addition,the IMvig-or210 cohorts and"oncoPredict"packet were used to predict immunotherapy responses and drug sensitivity.Results The study identified and validated two functional clusters with different prog-nosis and immune infiltration patterns based on 28 immune gene sets,and verified their heteroge-neous immune landscapes.Subsequently,a signature composed of six immune-related genes was determined and constructed,and further divided into risk subgroups with different clinical and pathological features,prognosis,and immune landscape.The high-risk subgroup was characterized by immunosuppression,high levels of static cell subsets infiltration,low-frequency gene muta-tions,and lower immune checkpoint molecule expression levels,which is closely related to im-mune escape and poor prognosis.There are significant differences in half maximal inhibitory con-centration(IC50)in risk groups,which can effectively distinguish"hot tumors"from"cold tumors",which is of great significance for guiding the selection of immunotherapy drugs.Conclu-sion The study established and validated a new classification based on immune characteristics that shows unique value in predicting GC prognosis and guiding the screening of potential drugs.
维普
万方数据