Identification of Serum Circulating MicroRNAs as Novel Diagnostic Biomarkers of Gastric Cancer

Abstract

Gastric cancer (GC) is one of the leading causes of cancer-associated deaths worldwide. Due to the lack of typical symptoms and effective biomarkers for non-invasive screening, most patients develop advanced-stage GC by the time of diagnosis. Circulating microRNA (miRNA)-based panels have been reported as a promising tool for the screening of certain types of cancers. In this study, we performed differential expression analysis of miRNA profiles of plasma samples obtained from gastric cancer and non-cancer patients using two independent Gene Expression Omnibus (GEO) datasets: GSE113486 and GSE124158. We identified three miRNAs, hsa-miR-320a, hsa-miR-1260b, and hsa-miR-6515-5p, to distinguish gastric cancer cases from non-cancer controls. The three miRNAs showed an area under the curve (AUC) over 0.95 with high specificity (>93.0%) and sensitivity (>85.0%) in both the discovery datasets. In addition, we further validated these three miRNAs in two external datasets: GSE106817 [sensitivity: hsa-miR-320a (99.1%), hsa-miR-1260b (97.4%), and hsa-miR-6515-5p (92.2%); specificity: hsa-miR-320a (88.8%), hsa-miR-1260b (89.6%), and hsa-miR-6515-5p (88.7%); and AUC: hsa-miR-320a (96.3%), hsa-miR-1260b (97.4%), and hsa-miR-6515-5p (94.6%)] and GSE112264 [sensitivity: hsa-miR-320a (100.0%), hsa-miR-1260b (98.0%), and hsa-miR-6515.5p (98.0%); specificity: hsa-miR-320a (100.0%), hsa-miR-1260b (100.0%), and hsa-miR-6515.5p (92.7%); and AUC: hsa-miR-320a (1.000), hsa-miR-1260b (1.000), and hsa-miR-6515-5p (0.988)]. On the basis of these findings, the three miRNAs can be used as potential biomarkers for gastric cancer screening, which can provide patients with a higher chance of curative resection and longer survival.

Keywords: biomarkers; differential expression analysis; gastric cancer; gene expression omnibus datasets; microRNAs.

FIGURE 1

Study design. The schematic diagram represents the strategy for in silico discovery and validation of the miRNA signatures. AUC, area under the curve.

FIGURE 2

Identification of 16 candidate serum miRNA signatures and their characterization in the discovery datasets. (A) Venn diagram showing significant differentially expressed (DE) miRNAs in the discovery dataset GSE113458 and GSE124158. The 16 miRNAs are listed in the table on the right along with the adjusted p-value and fold change. (B) Box plots showing higher expression levels of the 16 identified miRNAs in gastric cancer patients compared with the non-cancer control patients in GSE113486 and GSE124158. (C) Heatmap showing a promising result of the hierarchical clustering analysis using the 16 identified miRNAs to distinguish different samples in GSE113486 and GSE124158. NC, non-cancer control patients.

FIGURE 3

Diagnostic performance of the 16 candidate serum miRNA signatures in the discovery dataset GSE113486. The expression levels of 16 candidate miRNAs were individually analyzed using the receiver operating characteristic curve analysis in the GSE113486 cohort. The area under the curve (AUC), specificity, and sensitivity are calculated and displayed for the evaluation of the candidate miRNAs in the diagnosis of gastric cancer.

FIGURE 4

Diagnostic performance of the 16 candidate serum miRNA signatures in the discovery dataset GSE124158. (A) Shows the diagnostic performance of different miRNAs, wherein the threshold is the cut-off that decides whether a sample is cancerous or not. (B) Displays the overlaid receiver operating characteristic curves (ROC) of 16 individual miRNAs.

FIGURE 5

Validation of the diagnostic performance of the three selected miRNA signatures in two independent external datasets. Receiver operating characteristic curve (ROC) analysis was conducted on the two validation datasets: GSE106817 and GSE112264. Area under the curve (AUC), specificity, and sensitivity are calculated and displayed for each dataset.