A comprehensive investigation to oxidative stress balance in human stomach adenocarcinoma

Abstract OBJECTIVES : High level of pro-oxidative stress was hallmark of stomach adenocarcinoma (STAD), which depended on the imbalance between the production and elimination of pro-oxidative and antioxidative factors. The aims of this study was to investigate the crosstalk between oxidative stress balance and STAD. DESIGN AND METHODS : Pro-oxidative stress and antioxidant gene set score was evaluated by GSVA. Subsequently, qPCR and HPA database was used to validated differential expressed oxidative gene at the level of mRNA and protein, respectively. Furthermore, oxidative stress gene set were used to classify STAD patients into four subtypes with the help of consensus clustering method, and we used 28 immune cell gene set to identify immune infiltering levels across four oxidative subtypes based on GSVA method. WGCNA methods and Metascape database were performed to detect gene module and corresponding pathway associated oxidative subtype, respectively. Finally, we construct oxidative stress associated with risk score system based on oxidative stress gen set by machine learning. C index, 30 1-3 years AUC curve and Kaplan-Meier curve (K-M) analysis were used to evaluated model predictive ability and application. Besides, the differential score of 50 hallmark pathway and multiple immune inflating methods (TIMER, quanTIseq, MCP-counter, xcell, and EPIC) was identified between high risk and low risk group. RESULTS: Differential GSVA and GSEA from pan-cancer showed that pro-oxidative stress score was significantly increased and pro-oxidative stress (ECT2), antioxidant (ASAH2, MSRB3, and GPX3) also was elevated in STAD, especially MSI subtype (TCGA STAD subtype). Furthermore, STAD was classified into four oxidative stress subtypes, including pro-oxidative stress dominant, antioxidant dominant, mix, and quiescent. Compared with quiescent subtype, mix subtype remain low immune inhibitor cell (MDSC and Regulatory T cell), high immune active cell (CD8+) and significantly positively correlated with DNA metabolic process, and DNA IR-damage. Moreover, 6-gene oxidative stress signature, which showed a good predictive ability, was used to constructed risk predictive model by Lasso, RSF and stepping Cox regression. The group with high-risk score showed poor prognosis and high immune inhibitor cells infiltration in contrast with low-risk group. CONCLUSION: Therefore, targeting oxidative stress would be a potential target for STAD therapy.