Model including hepatitis B viral load accurately predicts HCC risk in untreated patients

A version of the REACH-B (Risk Estimation for Hepatocellular Carcinoma in Chronic Hepatitis B) model that included data on hepatitis B viral (HBV) load was effective in predicting hepatocellular carcinoma (HCC) risk in untreated patients, a recent study found.

Researchers developed and externally validated the Revised REACH-B model in noncirrhotic adult patients with chronic hepatitis B and no notable elevations in alanine aminotransferase (ALT) level who would not meet current criteria for antiviral treatment. The development cohort included 6,949 patients with chronic hepatitis B from Korea, and the validation cohort included 7,429 patients from Taiwan, Korea, and Hong Kong. The study results were published Sept. 17 by Annals of Internal Medicine.

Mean age in the derivation cohort was 45 years, 29.9% were positive for hepatitis B e antigen (HBeAg), median HBV DNA levels were 3.1 log₁₀ IU/mL, and median ALT level was 25 U/L. In the validation cohort, corresponding values were 46 years, 21.0%, 3.4 log₁₀ IU/mL, and 20 U/L. The derivation cohorts and validation cohorts included 435 and 467 incident cases of HCC over median follow-up periods of 10.0 and 12.2 years, respectively.

HBV DNA level at baseline was one of the strongest predictors of HCC, with moderate viral loads (approximately 6 log₁₀ IU/mL) indicating the highest HCC risk. Age, sex, platelet count, ALT levels, and positive HbeAg result were other variables included in the model. The model had satisfactory discrimination and calibration (c-statistics, 0.844 in the derivation cohort and 0.813 in the validation cohort with multiple imputation) and a greater positive net benefit versus other strategies in the threshold probability range of 0% to 18%.

The authors noted that their study was observational and that the model was developed and validated using untreated Asian patients, among other limitations. “In conclusion, we have developed and validated a novel HCC risk prediction model that incorporates the nonlinear parabolic association between baseline serum HBV viral load and HCC risk in treatment-naive, noncirrhotic, adult patients with CHB [chronic hepatitis B] without notable ALT level elevation,” they wrote, adding that the Revised REACH-B model used six variables to reliably predict HCC risk with fair clinical utility.

“By providing more comprehensive risk stratification, this model has the potential to inform optimal medical management strategies for patients with CHB who do not meet the current criteria for antiviral treatment,” the authors concluded.

An accompanying editorial said that because the new model is easy to use and highly accurate, it will be useful for patient counseling, monitoring, treatment planning, and prognostication. However, the editorialists agreed that validation in additional groups of patients is critically necessary for generalizability, since route of transmission, infection duration, host genetic susceptibility, and HBV vary globally.

“Finally, noninvasive assessment of hepatic fibrosis staging is routinely done at baseline in most patients with CHB,” they wrote. “Because fibrosis stage affects treatment decisions and is a potent predictor of liver outcomes, including HCC development, the utility of adding fibrosis stage to the Revised REACH-B model requires further study.”