Identification of Novel Prognostic Biomarkers to Predict Pembrolizumab Response in Gastric Cancer by Bioinformatics Analysis

Abstract

Immune checkpoint inhibitors (ICIs), especially pembrolizumab, have improved outcomes in patients with advanced or metastatic gastric cancer (GC). However, there are no reliable markers for the evaluation of its efficacy. GC samples from the UCSC Xena Browser and Gene Expression Omnibus (GEO) databases were performed by bioinformatics analysis. The single sample gene set enrichment (ssGSEA) algorithm was used to calculate the prognostic genes expression differences score (ssGSEA score) for each sample. Weighted gene co-expression network analysis (WGCNA) was applied to identify the score-related key model. The hub gene was identified based on the intersection between tumor-related transcription factors (TFs) and the key module. The predictive ability for pembrolizumab response of the hub gene was evaluated by the area under the receiver operating characteristic (ROC) curve (AUC). The correlations between the hub gene and immune infiltration in GC were investigated by the CIBERSORT algorithm. According to the ssGSEA score, patients with GC were divided into the high- and low-score group. RUNX1T1 was the hub gene of the ICIs treatment in GC. Its high expression was related to the low-score group, which indicated low sensitivity to pembrolizumab and poor prognosis. RUNX1T1 showed good predictive power for pembrolizumab response (AUC = 0.742). The CIBERSORT analysis showed that a high expression for RUNX1T1 weakened the anti-tumor immune response by affecting the immune cell infiltration.Runx1T1 may be a potential predictor of pembrolizumab efficacy in GC and be a prognostic marker for GC.