Machine learning based obesity and aging related signature for predicting the prognosis and immunotherapy benefit in stomach adenocarcinoma

BackgroundStomach adenocarcinoma (STAD) is one of most common cancers with high invasiveness and poor prognosis. Obesity and aging are correlated with higher risk for cancer development and worse prognosis in certain types of malignancies.MethodsAn integrative approach incorporating 10 machine learning methods was employed to develop an obesity and aging-related signature (ORS) using data from the TCGA, GSE15459, GSE26253, GSE62254, and GSE84437 datasets. To assess the predictive value of ORS for immunotherapy benefits, we utilized several indicating scores and three immunotherapy datasets (GSE91061, GSE78220, and IMvigor210).ResultsThe predictive model developed using the LASSO method achieved the highest average C-index and was identified as the optimal ORS. This ORS served as an independent risk factor for the clinical outcomes of STAD patients, demonstrating robust performance in predicting overall survival rates. In the TCGA cohort, the area under the curve values for the 1-, 3-, and 5-year receiver operator characteristic curves were 0.871, 0.803, and 0.768, respectively. Patients with low ORS score exhibited higher gene set scores for immune-activated cells, increased cytolytic activity, and enhanced T cell co-stimulation. Additionally, low ORS score was associated with a reduced tumor immune dysfunction and exclusion score, decreased immune escape score, elevated PD1 and CTLA4 immunophenoscore, higher tumor mutation burden, improved response rates, and better prognosis in STAD. Conversely, the IC50 values for common chemotherapy and targeted therapy regimens were lower in the high ORS score group.ConclusionThe current study developed an optimal ORS in STAD, which could be used for predicting the prognosis, stratifying risk and guiding treatment for STAD patients.