Prediction of hepatocellular carcinoma after sustained virologic response in patients with compensated advanced chronic liver disease

Background & Aims Hepatocellular carcinoma (HCC) is a main cause of morbidity and mortality in patients with advanced chronic liver disease (ACLD) due to chronic hepatitis C (CHC) who have achieved sustained virologic response (SVR).We aimed to elaborate the optimal risk stratification algorithms for de-novo-HCC-development after SVR and to validate them in an independent cohort.

Methods Derivation cohort: 527 patients with pre-treatment advanced chronic liver disease (ACLD) receiving interferon (IFN)-free therapy for CHC were evaluated for de-novo-HCC-development post-treatment. Among other potential risk factors, non-invasive surrogates of portal hypertension including liver-stiffness measurement (LSM) and von Willebrand factor, as well as levels of alpha-fetoprotein (AFP) were assessed pre- and post-treatment. Validation cohort: 1500 patients with compensated ACLD (cACLD) from other European centers.

Results During a median follow-up (FU) of 41 months, 22/475 cACLD patients (4.6 %) developed HCC (1.45/100 patient-years) vs. 12/52 decompensated patients (23.1 %, 7.00/100 patient-years, p < 0.001). Since decompensated patients were at substantial HCC-risk, we focused on cACLD for all further analyses.In cACLD, post-treatment-values showed a higher accuracy for predicting HCC than pre-treatment-values or absolute/relative changes. A model based on post-treatment age ≥ 59years - 2 points, albumin < 42gxL^-1 -1 point, LSM≥19.0kPa - 1 point, and AFP≥4.6ngxmL^-1 - 3 points most-accurately predicted de-novo-HCC-risk during FU (bootstrapped Harrel’s C: 0.874). Importantly, these parameters also provided independent prognostic information in competing risk analysis and accurately stratified patients into low-(0-3 points; ≈2/3 of patients) and high-risk (4-7 points; ≈1/3) groups in the derivation (HCC at 4 years: 0.5 % vs. 16.7 %) and validation cohort (3.2 % vs. 19.1 %). An alternative approach based on age/FU-albumin/FU-LSM (i.e., without FU-AFP) also showed a robust performance.

Conclusions Simple algorithms based on post-treatment age/albumin/LSM, and optionally, AFP, accurately stratified de-novo-HCC-risk in cACLD patients with SVR. Approximately 2/3 were identified to have an HCC-risk < 1 %/y, thereby clearly falling below the cost-effectiveness threshold for HCC-surveillance.

Publication History

Article published online:
01 September 2021