Accurate prediction of HCC risk after SVR in patients with hepatitis C cirrhosis based on longitudinal data

Abstract

Background: Most existing predictive models of hepatocellular carcinoma (HCC) risk after sustained virologic response (SVR) are built on data collected at baseline and therefore have limited accuracy. The current study aimed to construct an accurate predictive model incorporating longitudinal data using a novel modeling strategy. The predictive performance of the longitudinal model was also compared with a baseline model.

Methods: A total of 400 patients with HCV-related cirrhosis who achieved SVR with direct-acting antivirals (DAA) were enrolled in the study. Patients were randomly divided into a training set (70%) and a validation set (30%). Informative features were extracted from the longitudinal variables and then put into the random survival forest (RSF) to develop the longitudinal model. A baseline model including the same variables was built for comparison.

Results: During a median follow-up time of approximately 5 years, 25 patients (8.9%) in the training set and 11 patients (9.2%) in the validation set developed HCC. The areas under the receiver-operating characteristics curves (AUROC) for the longitudinal model were 0.9507 (0.8838-0.9997), 0.8767 (0.6972,0.9918), and 0.8307 (0.6941,0.9993) for 1-, 2- and 3-year risk prediction, respectively. The brier scores of the longitudinal model were also relatively low for the 1-, 2- and 3-year risk prediction (0.0283, 0.0561, and 0.0501, respectively). In contrast, the baseline model only achieved mediocre AUROCs of around 0.6 (0.6113, 0.6213, and 0.6480, respectively).

Conclusions: Our longitudinal model yielded accurate predictions of HCC risk in patients with HCV-relate cirrhosis, outperforming the baseline model. Our model can provide patients with valuable prognosis information and guide the intensity of surveillance in clinical practice.

Keywords: Direct-acting antivirals; Hepatocellular carcinoma; Longitudinal study; Machine learning; Predictive models.

Fig. 1

Trajectories of 10 longitudinal predictors in patients with HCC and without from the entire cohort. The longitudinal predictors were AFP (a), ALT (b), albumin (c), ALP (d), AST (e), cholinesterase (f), direct bilirubin(g), GGT (h), total bilirubin (i), and total protein (j). The grey lines represent individual trajectories of each patient, the blue lines are the averaged trajectories estimated using linear mixed-effects models and the red lines are the averaged trajectories estimated using mixed-effects models that includes natural cubic splines with 2 degrees of freedom. The values of all predictor variables are on a log scale. Abbreviations: AFP, alpha-fetoprotein; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ALP, alkaline phosphatase; GGT, gamma-glutamyl transferase; HCC, hepatocellular carcinoma

Fig. 2

Individual-level prediction of HCC-free probabilities for two patients from the validation set. Survival curves were smoothed with local polynomial regression. The blue lines represent the HCC-free probabilities predicted by the baseline RSF model, and the orange lines represent the HCC-free probabilities predicted by the longitudinal model. Survival curves were overlaid in the final column. Abbreviations: HCC, hepatocellular carcinoma; RSF, random survival forest

Fig. 3

Area under the receiver operating characteristic curves value and brier score of the baseline RSF model and longitudinal model for predictions made 1, 2, and 3 years from Year 3. Predictions were made at Year 3 for HCC occurrence 1, 2, and 3 years from Year 3, which equals 4, 5, and 6 years from baseline. Abbreviations: HCC, hepatocellular carcinoma; AUROC, area under the receiver-operating characteristic curve; RSF, random survival forest