Prediction model for hepatocellular carcinoma occurrence in patients with hepatitis C in the era of direct-acting anti-virals.

BACKGROUND Several factors associated with hepatocellular carcinoma (HCC) occurrence after sustained virological response (SVR) in patients with hepatitis C have been reported. However, few validation studies have been performed in the era of direct-acting anti-virals (DAAs). AIMS To develop a prediction model for HCC occurrence after DAA-mediated SVR and validate its usefulness. METHODS We analysed 2209 patients with SVR and without a history of HCC who initiated DAA treatment at 24 Japanese hospitals. These patients were divided into a training set (1473 patients) and a validation set (736 patients). RESULTS In the training set, multivariate Cox proportional hazards analysis showed that the baseline BMI (≥25.0 kg/m2 , P = 0.024), baseline fibrosis-4 (FIB-4) index (≥3.25, P = 0.001), albumin level at SVR (<4.0 g/dL, P = 0.010) and alpha-foetoprotein level at SVR (≥5.0 ng/mL, P = 0.006) were significantly associated with HCC occurrence. We constructed a prediction model for HCC occurrence with these four factors (2 points were added for the FIB-4 index, and 1 point was added for each of the other three factors). Receiver operating characteristics curve analysis identified a score of 2 as the optimal cut-off value for the prediction model (divided into 0-1 and 2-5). In the validation set, the sensitivity and negative predictive value for HCC occurrence were 87.5% and 99.7%, respectively, at 2 years and 71.4% and 98.0%, respectively, at 3 years. CONCLUSION A prediction model combining these four factors contributes to an efficient surveillance strategy for HCC occurrence after DAA-mediated SVR.