Risk of liver cirrhosis in HBV/HCV-infected individuals with first-degree relatives who have liver cancer: development and validation of a simple model.

Identification of high-risk population among HBV/HCV-infected individuals with first-degree relatives (FDRs) who have liver cancer is important to implement precise intervention. A cross-sectional study was conducted under the framework of a population-based Cancer Screening Program in Urban China (CanSPUC), aimed to develop and validate a simple non-invasive model that could assess and stratify cirrhosis risk, in HBV/HCV-infected individuals with FDRs who have liver cancer. People who participated in liver cancer screening in Henan province were enrolled. Using the dataset consisting of participants admitted from October 1,2013 to December 31, 2016, a 24-point scale risk score model was developed through logistic regression, based on education background, dietary habit, smoking index, cooking oil fume exposure, history of severe trauma, HBV/HCV infection status, history of diabetes, history of hyperlipidaemia, and parent history of liver cancer. The model showed excellent discrimination with AUROC of 0.875 (95%CI: 0.853-0.896) and fair calibration with a Hosmer-Lemeshow test P=0.106. The prevalence rates in the medium- and high-risk groups were 2.87 (95%CI: 1.94-4.25) and 47.57 (95%CI: 31.59-71.63) times of low-risk group, respectively. After internal validation, bias-corrected AUROC was 0.874 (95%CI: 0.873-0.875). In the external validation dataset consisting of participants admitted from January 1,2017 to October 31, 2018, the model had achieved similar discrimination, calibration and risk stratification ability. In conclusion, this risk score model we developed can be a practical tool for the screening and prevention of liver cirrhosis among HBV/HCV-infected individuals with FDRs who have liver cancer.