Abstract

Background:

The emergence of direct-acting anti-HCV agents has led to a significant increase of patients who achieve sustained virologic response (SVR) but remain at risk of HCC development over a decade. HCC risk stratification post SVR is urgently needed.

Methods:

A hepatic transcriptome-based Prognostic Liver Signature for SVR (PLS-SVR) was identified in 85 patients who had curative resection/ablation for post-SVR HCC (derivation set), and independently validated in 39 HCC-naïve post-SVR patients (tissue validation set). Subsequently, a blood-based surrogate of PLS-SVR, Prognostic Liver Secretome signature for SVR (PLSec-SVR), was defined using our computational pipeline (TexSEC) for non-invasive HCC risk assessment. PLSec-SVR was validated in two independent cohorts of HCC-naïve (serum validation set 1; matched 41 cases and 123 controls) and HCC-experienced (serum validation set 2; cohort of 146 subjects) patients. In the serum validation sets, we assessed clinical utility of PLSec-SVR as an etiology-specific “plug-in” to refine HCC risk prediction with our previously reported etiology-agnostic PLSec-AFP score.

Results:

We defined a 170-gene PLS-SVR, including 133 high- and 37 low-risk genes, in the derivation set, where the 5-year de novo HCC recurrence rates were 66.1% and 6.3% in high- and low-risk patients, respectively (adjusted hazard ratio [aHR], 48.9; 95% confidence interval [CI], 11.3-194.8). In the tissue validation set, 25 (64%) patients were classified as high risk, which was significantly associated with incident HCC (aHR, 9.99; 95% CI, 1.11-1319.4). PLS-SVR was converted to a 6-protein PLSec-SVR. In serum validation set 1 and 2, PLSec-SVR identified 59 (36%) and 43 (29%) high-risk patients, respectively, and was independently associated with HCC risk (adjusted OR [aOR], 3.08; 95% CI 1.42-6.65 and aHR, 2.59; 95% CI, 1.55-4.32 in serum validation sets 1 and 2, respectively). Integration of PLSec-SVR with the etiology-agnostic PLSec-AFP improved prognostic capability as a “plug-in” module (aOR, 13.9; 95% CI, 3.82-50.5 and aHR, 14.2; 95% CI, 6.17-32.6 in serum validation sets 1 and 2, respectively).

Conclusion:

PLSec-SVR enables prediction of long-term risk of post-SVR HCC development and substantially improves the performance of the etiology-agnostic PLSec-AFP score to inform necessity of regular HCC screening.