Patients with hepatitis B virus (HBV) were invited to receive ultrasound and alpha-protein examination directly in China. However, not all HBV carriers need to be subjected to further tests. This study aimed to develop a feasible primary screening method to narrow down potential high-risk individuals of liver cancer among populations with HBV.
Based on a prospective community-based cohort, potential risk factors were selected as the predictors, including age, sex, smoking, alcohol consumption, diabetes, liver cancer family history, liver diseases in mothers, source of water, body mass index (BMI), and psychological trauma. Cox proportional regression model was applied to predict the 3-year absolute risk of liver cancer and derive risk scores. The area under receiver operating characteristic curve (AUROC) and calibration plot were used to assess the performance of the model. Bootstrap resampling was used for internal validation.
Age, sex, BMI, alcohol consumption, liver diseases in mothers, and psychological trauma were independent risks of liver cancer. The 1- to 3-year AUROC of the prediction model was 71.15% (95% CI, 66.88–75.42), 71.16% (95% CI, 67.42–74.90), and 72.95% (95% CI, 64.20–81.70), respectively. The predicted risk was calibrated well with the observed liver cancer risk. Bootstrap resampling showed that C-index was 0.70 (0.67–0.74). A 32-point risk score was also developed and a score over 5 was identified for patients at extremely high risk.
A user-friendly primary screening method was created that could estimate the 3-year absolute risk of liver cancer and identify extremely high-risk individuals among the population with HBV.