Prediction model for hepatocellular carcinoma risk in treatment-naive chronic hepatitis B patients receiving entecavir/tenofovir.

Abstract

Accurate assessment of hepatocellular carcinoma (HCC) risk in chronic hepatitis B (CHB) patients receiving entecavir (ETV)/tenofovir disoproxil fumarate (TDF) is likely to play a pivotal role in post-treatment follow-up strategy. We aimed to develop a simple and reliable predictive model for HCC risk in these patients. A database of 1242 consecutive treatment-naive CHB patients who initially underwent ETV/TDF between February 2007 and January 2017 at four referral hospitals in South Korea was analyzed. The HCC risk model was constructed on the basis of a multivariable Cox proportional hazards model in the derivation dataset (n=944) and was validated using Harrell's C-statistic in a validation dataset (n=298). The 3/5-year cumulative incidence rates of HCC were 3.9/6.5 and 4.2/11.6% in the derivation and the validation dataset, respectively (P=0.08). In the derivation dataset, we identified four factors associated with HCC, namely, age, albumin, sex, and liver cirrhosis. The AASL (age, albumin, sex, liver cirrhosis)-HCC scoring system was developed on the basis of these factors, and simplified to an integer scoring system. AASL-HCC scores were found to have high discriminating performance for the prediction of HCC development at 5 years in the derivation (C-statistics=0.802, 95% confidence interval: 0.716-0.888) and validation dataset (C-statistics=0.805, 95% confidence interval: 0.671-0.939). When AASL-HCC scores were classified as 5 or less, 6-19, and at least 20 (low-risk, intermediate-risk, and high-risk groups, respectively), the 5-year cumulative incidence rates of HCC were 0, 4.2, and 17.6%, respectively, in the derivation dataset. The AASL-HCC model was simple and reliable for HCC risk prediction in treatment-naive CHB patients receiving ETV/TDF, and is easily applicable in the clinical setting.