Abstract: Objective To detect the expressions of pepsinogens (PGI and PGII), gastrin-17 (G-17) and sHLA-G in patients with gastric cancer or precancerous lesion, and to evaluate the efficiency of combined diagnostic index of PGR (PGI/II), GPR (G-17/PGI) and sHLA-G in these patients. Methods The levels of PGs were detected in 50 healthy volunteers, 46 patients with atrophic gastritis, 50 patients with gastric intraepithelial neoplasia, 36 patients with early gastric cancer and 74 patients with advanced gastric cancer by chemiluminescent microparticle immunoassay, and the levels of G-17 and sHLA-G were detected by ELISA. Logistic regression and ROC curve were used to establish a new diagnostic panel for gastric cancer using the combined diagnosis biomarkers PGR, GPR and sHLA-G. Results Reduced PGR and increased GPR and sHLA-G were detected in patients with early and advanced gastric cancer, compared with healthy volunteers, patients with gastric intraepithelial neoplasia and atrophic gastritis (all P<0.001). The ROC-AUC of sHLA-G and PGR were 0.846 (95%CI, 0.760-0.910) and 0.835 (95%CI, 0.748-0.902) respectively, which were significantly higher than those of GPR (0.716, 95%CI, 0.617-0.802). The new diagnostic model of gastric cancer was logit (Y)=0.41+0.03×sHLA-G-5.90×PGR, and the ROC-AUC of the new variable Y in thediagnosis of gastric cancer was 0.924 (95%CI, 0.853-0.967), which was higher than that of sHLA-G and PGR (P<0.05). Conclusion The combined detection of sHLA-G and PGR is helpful in the diagnosis of gastric cancer, which can serve as a useful supplement to traditional tumor markers.

Key words: Logistic regression, Stomach neoplasms, Soluble human leukocyte antigen G, Pepsinogens, Gastrins, ROC curve