Identification and Validation of Serum MicroRNA Biomarker for Non-invasive Detection of Gastric Cancer (14273)
Background: Gastric cancer (GC) is the 2nd leading cause of cancer deaths worldwide. Majority of GC are diagnosed at late stage, leading to poorer outcomes in general. Currently, endoscopy is the only reliable method for early diagnosis but limited as a screening test due to high cost and risk. A simple, less-invasive screening test for GC is highly desirable. Aberrant miRNA expressions are involved in the cancer pathogenesis and tumor-specific circulating miRNAs were recently discovered. MiRNAs are exceptionally stable in body fluids and can be readily measured using existing clinical lab workflows, making them excellent candidate disease biomarkers.
Aim: To identify serum miRNA biomarker to detect GC in a Chinese high risk population in Singapore.
Methods: A total of236 Chinese with newly diagnosed stage 1-4 GC and 236 age, gender and race matched high risk controls (patients with chronic gastritis and other gastric illnesses) were selected from the Singapore Gastric Cancer Cohort Study. The subjects were divided into discovery and validation set with 1:1 ratio. Absolute expressions of 578 candidate miRNAs were quantified in sera from all subjects by high-throughput qPCR using a proprietary, wet-lab validated assay system with multi-layered technical controls. MiRNA biomarkers were identified using support vector machine in the discovery set and tested for its predictive performance in the validation set. Concurrently, these candidate miRNAs were profiled using the same assays in two other cancer studies, breast cancer (N=450) and ovarian cancer (N=80). Cancer specificity of the identified biomarkers were tested by analyzing data from all three studies.
Results: Seventy-five informative miRNAs were significantly altered in GC (FDR corrected p-values<0.01). Multivariate miRNA biomarker panels showed high diagnostic accuracy (AUC of 0.85-0.93) in validation set. Independent statistical analysis using forward logistic regression confirmed the performance of these biomarkers. Importantly, majority of these serum miRNA biomarkers are specifically regulated in gastric cancer, but not breast and ovarian cancers, indicating cancer specificity.
Conclusion and Discussion: A multi-variant serum miRNA biomarker panel with high predictive performance was identified in the most extensive GC-miRNA study performed to date. The panel is currently being prototyped as an IVD test kit and validated in additional retrospective cohorts from Korea, Japan and South America as well as a large prospective validation study (N=5000) designed according to NIH biomarker guidelines. The cost-effectiveness of such technology is being evaluated by Health Technology Assessment.