Multi-dimension metabolic prognostic model for gastric cancer.

Abstract

Metabolic reprogramming is involved in different stages of tumorigenesis. There are six widely recognized tumor-associated metabolic pathways, including cholesterol catabolism process, fatty acid metabolism, glutamine metabolic process, glycolysis, one carbon metabolic process, and pentose phosphate process. This study aimed to classify gastric cancer patients into different metabolic bio-similar clusters. We analyzed six tumor-associated metabolic pathways and calculated the metabolic pathway score through RNA-seq data using single sample gene set enrichment analysis. The consensus clustering analysis was performed to classify patients into different bio-similar clusters by multi-dimensional scaling. Kaplan-Meier curves were presented between different metabolic bio-similar groups for OS analysis. A training set of 370 patients from the Cancer Genome Atlas database with primary gastric cancer was chosen. Patients were classified into four metabolic bio-similar clusters, which were identified as metabolic non-specificity, metabolic-active, cholesterol-silence, and metabolic-silence clusters. Survival analysis showed that patients in metabolic-active cluster and metabolic-silence cluster have significantly poor prognosis than other patients (p=0.031). Patients in metabolic-active cluster and metabolic-silence cluster had significantly higher intra-tumor heterogeneity than other patients (p=0.032). Further analysis was performed in metabolic-active cluster and cholesterol-silence cluster. Three cell-cycle-related pathways, including G2M checkpoints, E2F targets, and MYC targets, were significantly upregulated in metabolic-active cluster than in cholesterol-silence cluster. A validation set of 192 gastric cancer patients from the Gene Expression Omnibus data portal verified that metabolic bio-similar cluster can predict prognosis in gastric cancer. Our study established a multi-dimension metabolic prognostic model in gastric cancer, which may be feasible for predicting clinical outcome.