Validation of Hepatitis B Virus-Related Hepatocellular Carcinoma Prediction Models in the Era of Antiviral Therapy

Several risk prediction models have been created to predict hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC) occurrence, with promising results. However, their prognostic performances need to be validated in the era of antiviral therapy. From 2006 to 2011, patients with chronic HBV infection were recruited and those with a history of HCC or hepatic decompensation were excluded. At enrollment, liver stiffness (LS) was measured using transient elastography. We assessed the performances of conventional HCC prediction models (CU-HCC, GAG-HCC, REACH-B, and LSM-HCC scores) and the modified REACH-B (mREACH-B) score where LS values were incorporated into REACH-B score instead of serum HBV-DNA levels. Of 1,308 subjects analyzed, the median age was 50.0 years (883 men). During the follow-up (median, 75.3 months), HCC developed in 125 (9.6%) patients. mREACH-B score had the highest areas under the receiver operating characteristic curves (AUROCs) for the prediction of HCC development at 3/5 years (0.828/0.806), compared with LSM-HCC (0.777/0.759), GAG-HCC (0.751/0.757), REACH-B (0.717/0.699), and CU-HCC (0.698/0.700) scores, respectively, with statistical significances (all P values <0.05 vs. mREACH-B). When serum HBV-DNA levels were excluded from the formula for REACH-B score, AUROCs for HCC development at 3/5 years improved paradoxically (from 0.717/0.699 to 0.757/0.732, respectively). In patients with antiviral therapy (n = 848), mREACH-B score had the better prognostic performances for HCC development at 3/5 years, compared to other prediction models. However, in patients without antiviral therapy (n = 460), it had the prognostic performances comparable to those of other prediction models. Conclusions: Prognostic performances of mREACH-B score seemed better compared to conventional models. In the era of antiviral therapy, incorporation of serum HBV-DNA level should be applied cautiously and individual risks should be assessed effectively based on the fibrotic burden.