Promising protein biomarkers for early gastric cancer: clinical performance of combined detection

Abstract Objectives In the early stage of gastric cancer (GC), identifying cancer-specific biomarkers is a key step in the disease screening process. This study aims to explore the clinical value of five novel protein biomarkers and their combination (5 MP) for GC early diagnosis. Methods The candidate biomarkers were mined from TCGA, GTEx, and CPTAC databases. The clinical value of the five biomarkers and 5 MP in the early diagnosis from healthy control, benign gastric disease (BGD), precancerous lesions (PLGC), early GC (EGC), and GC was evaluated by receiver operator characteristic curve (ROC), the area under the curve (AUC), sensitivity, specificity, and accuracy. Results Five candidate biomarkers, COL10A1, GKN1, GKN2, LIPF, and REG4, were mined from TCGA, GTEx, and CPTAC databases. In the training cohort, the five proteins were confirmed to be differentially expressed in the serum of control, BGD, EGC, and GC. COL10A1 has the highest AUC of a single protein in control vs. EGC (0.857). GKN2 has the highest AUC of a single protein in BGD vs. EGC (0.822). 5 MP has an AUC of 0.890 in Control vs. EGC, and 0.854 in BGD vs. EGC. In the validation cohort, 5 MP has an AUC of 0.834 in PLGC vs. GC, and 0.839 in PLGC vs. EGC. Conclusions Our findings suggest that COL10A1, GKN1, GKN2, LIPF, and REG4 are useful non-invasive serum biomarkers for GC diagnosis. Combination detection (5 MP) has enhanced early diagnosis of GC and distinguishing between benign and malignant gastric diseases.