External validation of the PAGE-B score for HCC risk prediction in people living with HIV/HBV coinfection.

BACKGROUND & AIMS

Hepatitis B virus (HBV) coinfection is common among people living with HIV (PLWH) and the most important cause of hepatocellular carcinoma (HCC). Whereas risk prediction tools for HCC exist for patients with HBV monoinfection, they have not been evaluated in PLWH. We performed an external validation of PAGE-B in people with HIV/HBV coinfection.

METHODS

We included PLWH with a positive HBsAg and without HCC before starting tenofovir from four European cohorts, and estimated the predictive performance of PAGE-B on HCC occurrence over 15 years of tenofovir-containing antiretroviral therapy (ART). Model discrimination was assessed after multiple imputation using Cox regression with the prognostic index as covariate, and by calculating Harrell's c-index. Calibration was assessed by comparing cumulative incidences with the PAGE-B derivation study using Kaplan-Meier curves.

RESULTS

In total, 2'963 individuals with HIV/HBV coinfection on tenofovir-containing ART were included. PAGE-B was <10 in 26.5%, 10-17 in 57.7%, and ≥18 in 15.7% of patients. Within a median follow-up of 9.6 years, HCC occurred in 68 individuals (2.58/1000 patient-years, 95% confidence interval [CI] 2.03-3.27). The regression slope of the prognostic index for developing HCC within 15 years was 0.93 (95% CI 0.61-1.25), and the pooled c-index was 0.77 (range 0.73-0.80), both indicating good model discrimination. Cumulative incidence of HCC was lower in our study compared to the derivation study. A PAGE-B cut-off of <10 had a negative predictive value for developing HCC within 5 years of 99.4%. Restricting efforts to individuals with a PAGE-B of ≥10 would spare HCC screening in 27% of individuals.

CONCLUSIONS

For individuals with HIV/HBV coinfection, PAGE-B is a valid tool to determine the need for HCC screening.

IMPACT AND IMPLICATIONS

Chronic hepatitis B virus (HBV) infection is the most important cause of hepatocellular carcinoma (HCC) among people living with HIV, and valid risk prediction may guide HCC screening efforts to high-risk individuals. We aimed at validating PAGE-B, a risk prediction tool that is based on age, gender, and platelets, among 2963 individuals with HIV/HBV coinfection who received tenofovir-containing antiretroviral therapy. In the present study, PAGE-B showed good discrimination, adequate calibration, and a cut-off of less than 10 had a negative predictive value for developing HCC within 5 years of 99.4%. These results indicate that PAGE-B is a simple and valid risk prediction tool to determine the need for HCC screening among people living with HIV and HBV.