Gastric cancer (GC) is the second leading cause of cancer-related death in the world. The optimal treatment regimens for GC depend on tumor stage, histopathological subtype, and other factors. The detection of tumor biomarkers is a quick way to get information of the tumor state. In this study, new biomarkers are detected for GC diagnostic and prognostic purposes. A total of 305 cases of diagnosed gastric adenocarcinoma were enrolled, microRNAs (miRNAs) and their transcriptome sequencing data were obtained from the "The Cancer Genome Atlas." Blood samples were collected from GC patients before surgery and therapy. The miRNA levels and the expression of RNA were detected by real-time RT-PCR. Receiver operating characteristic analysis was used to evaluate the sensitivity and specificity of biomarkers. The combining predictors were established with the logistic regression analysis. Hundreds of miRNA were with higher area under curve (AUC) than 0.5; among them, nine miRNAs were with the highest AUC more than 0.90 and displayed strong diagnostic value. Moreover, the mir-17 level was correlated with tumor stage (p = 0.029), while mir-133b, mir-133a-2, and mir-1-2 levels were significantly correlated with race, tumor pathologic, and tumor stage (p < 0.05). The combination biomarker (mir-181a-1/KAT2B with a sensitivity of 95.83 % and specificity of 94.12 %) could be used as an independent diagnostic indicator for GC patients. For GC patients, mir-17, mir-133b, mir-133a-2, and mir-1-2 appear to be a potential novel predictor of tumor stage and preoperative and intraoperative diagnosis. The combination of miRNA and mRNA such as mir-181a-1/KAT2B (with a sensitivity of 95.83 % and specificity of 94.12 %) showed significant improvement in the diagnostic accuracy.
Keywords: Diagnosis; Gastric cancer; ROC; mRNA; miRNA.