Background: Reliable prediction models based on blood indices and genetic background for accurately identifying high-risk populations for gastric cancer (GC) are lacking.
Objectives: We aimed to construct a risk assessment tool using commonly used clinical blood indices and evaluate its effectiveness for risk prediction and population stratification in different genetic backgrounds.
Methods: We constructed the blood test score (BTS) via LASSO logistic regression based on GC-associated blood indices. Then, we used the receiver operating characteristic (ROC) curve, calibration curve, and decision curve analyses to evaluate its discrimination, calibration, and clinical utility. The polygenic risk score (PRS) was subsequently constructed and combined with the BTS to evaluate the joint effect and interaction on GC risk.
Results: The BTS demonstrated good discriminative ability in the training (AUC = 0.776, 95% CI: 0.756–0.796) and testing sets (AUC = 0.766, 95% CI: 0.734–0.797). The calibration curves and decision curve analysis of the BTS model demonstrated good calibration and clinical application ability. We observed a significant increase in the relative risk of developing GC in quintiles of the BTS and a similar trend was observed in genetic risk groups. Compared with individuals with a low BTS, individuals with intermediate and high BTS values exhibited a greater risk of developing GC in all three genetic risk subgroups, whereas the interaction between the BTS and PRS with GC risk was not significant.
Conclusions: Our study revealed an association between the BTS, genetic factors, and GC risk. The BTS is valuable for predicting the risk of GC and discriminating high-risk individuals with GC.

Distribution pf BTS and PRS and ROC curve