Risk of hepatocellular carcinoma development in long-term nucles(t)ide analog suppressed patients with chronic hepatitis B

INTRODUCTION

Worldwide hepatocellular carcinoma (HCC) remains a leading cause of morbidity and mortality in patients with chronic hepatitis B (CHB), despite the availability of safe and effective therapies, with an estimated 350,000 deaths per year^[1,2]. Currently, due to their high potency and barrier to resistance development, Entecavir (ETV), Tenofovir dispoproxil fumarate (TDF), and Tenofovir alafenamide (TAF) are the third-generation nucleos(t)ide analogs (NAs) recommended as first-line monotherapy by all international guidelines for treating CHB^[3-5]. These drugs achieve sustained on-treatment viral suppression in more than 95% of patients, resulting in several clinical benefits such as prevention of cirrhosis and liver decompensation, improvement of liver histology, and often reversion of histologic cirrhosis. However, HCC can still develop despite sustained virological response, and it is currently the major complication in the long-term management of CHB patients. Literature data show that in non-cirrhotic patients treated with ETV or TDF, annual HCC incidence ranges from 0% to 1.4% in Asian patients and from 0.1% to 1.0% in predominantly Caucasian populations. In cirrhotic patients treated with ETV or TDF, hepatocellular carcinoma rates are around 4 to 5 times higher, ranging from 0.9% to 5.4% in Asians and from 1.5% to 5.2% in Caucasians^[6]. Limited data are available on the incidence and predictive factors of HCC development beyond the first 5 years of NA therapy. A multicenter European cohort study showed that the HCC risk decreased after 5 years of ETV/TDF therapy^[7].

Current guidelines identify patients with cirrhosis as a high-risk group requiring close HCC surveillance. Regular abdominal ultrasonography (US), with or without the assay of serum alpha-fetoprotein (AFP) levels, is recommended for early diagnosis of HCC in order to improve the applicability of curative therapies and ultimately the patient’s prognosis^[8,9].

Sundry predictive factors have been associated with an increased risk of HCC appearance in patients with chronic hepatitis B. These include host factors (male sex, older age), virological factors (high serum HBV DNA levels, HBeAg/anti-HBe status, genotype C, pre-core and core promoter mutations), and disease status-related factors (serum alanine aminotransferase levels, presence of cirrhosis or advanced fibrosis, comorbidities as diabetes and overweight)^[10-13].

Up to now, several HCC risk prediction scores, including different baseline parameters, have been elaborated to estimate the mild-long term risk prediction of HCC development in CHB patients^[14,15]. In 2016, an HCC risk score (named PAGE-B score) was developed and validated by Papatheodoridis et al.^[16]. This score, including only the baseline patient age, gender, and platelet levels, stands for a simple and trusty score predicting the 5-year HCC risk in Caucasian patients treated with ETV/TDF therapy. The authors showed that PAGE-B scores ≤ 9 imply no or minimal 5-year HCC risk, while PAGE-B scores ≥ 10 and particularly ≥ 18 indicate increased HCC risk requiring constant and careful surveillance. In particular, the cut-off point of 10 in the PAGE-B score offered 100% sensitivity and negative predictive value (NPV) for HCC prediction in both the derivation and validation datasets.

Our study aimed to assess the incidence and predictive risk factors of late HCC occurrence in a real-life cohort of CHB patients treated with NAs for at least 5 years and to verify the exactness of the PAGE B-risk score.

METHODS

RESULTS

Overall, 343 CHB patients consecutively treated with NAs for at least five years, predominantly males (76%), mean age (± SD) 48 (± 12) years, were enrolled in the study [Table 1].

A co-infection with other hepatitis viruses was present in 31 patients (9%), in particular, a co-infection with HDV and HCV was present in 21 and 6 patients, respectively; a triple infection HBV/HDV/HCV was present in 4 cases. At baseline before starting NA therapy, the severity of liver disease was classified according to findings from liver biopsies in 286/343 patients (83.4%). In these patients, histological data excluded not-HBV-related causes of liver damage. A diagnosis of cirrhosis (clinical or histological) was present in 105 patients (31%).

Alcohol use was reported as none/mild in 295/343 patients (86%). A history of previous drug addiction was reported only in 11/343 patients (3%). Most patients started therapy with LAM (197 patients, 57%). A family history of HCC was reported in 50 cases (14%).

PAGE-B score

A binomial logistic regression model was calculated to assess the predictive value of the PAGE B-risk score for developing HCC, by correlating the incidence of HCC with the PAGE-B values. PAGE B score showed an association with HCC occurrence (R² = 0.225, P < .001). The good prognostic capacity of the score was observed [Accuracy: 0.936, Sensitivity: 0.997, Area Under the ROC Curve (AUC) 0.863]. No significant differences were carried out when comparing baseline PageB score with Page B at 5 years of treatment.

Predictive measures are graphically reported in the Receiver Operator Characteristic (ROC) curve in Figure 1.

Figure 1. Receiver Operator Characteristic (ROC) curve representing the predictive measures of PAGE-B score toward the incidence of HCC among CHB patients in the study cohort. The cut-off value was set at 0.5.

DISCUSSION

This study assessed in a real-life cohort of virologically suppressed CHB patients treated with long-term NAs the risk and predictors of late HCC development.

Firstly, our results confirm the favourable impact of viral suppression on the natural history of HBV-related liver disease. In fact, no patient with chronic hepatitis progressed to cirrhosis, and no patient with portal hypertension diagnosis presented with bleeding esophageal varices. Only one patient with HBV/HDV-related cirrhosis developed liver decompensation due to the active Delta infection, resulting in the need for liver transplantation.

Moreover, long-term NA treatment in a few patients achieved a functional cure (HBsAg loss), the primary endpoint of future therapeutic strategies for chronic hepatitis B.

Nevertheless, in our cohort of patients, the persistent viral replication suppression does not abolish the risk of HCC, especially in those with advanced liver disease. In fact, a late HCC development was still diagnosed in 21 patients (6%), regardless of successful long-term treatment. Most patients with HCC (16/21) received TDF as rescue therapy after a virological breakthrough due to LAM resistance. The remaining five patients received TDF (2 cases) or ETV (3 cases) from the start.

Limited data are available on the incidence and predictive factors of late HCC occurrence in CHB patients beyond the first 5 years of NA therapy. Papatheodoridis et al. showed that after 5 years of ETV/TDF therapy, the HCC risk decreased. They reported 1,951 Caucasian patients under NA treatment for a median period of 6.8 years, an overall yearly HCC incidence rate of 1.22% within and 0.73% after the first 5 years (P = 0.050)^[7].

In this study, the overall incidence of HCC at 5-15 years (6%) was higher than that reported by Papatheodoridis et al. at 5-10 years (1.4%). This result could be explained by the difference in baseline patient characteristics and risk factors for HCC development between the two study populations. In fact, our study included a higher number of patients with cirrhosis, with a family history of HCC, and with HDV-HCV co-infections. In multivariate analysis, the diagnosis of cirrhosis at baseline and a family history of HCC were significantly associated with HCC occurrence.

Noteworthy, there were no female patients with HCC in our study, suggesting that male gender is a risk factor for HCC development.

As already known, more than half of our HCC patients HCC showed normal serum levels of alpha-fetoprotein at diagnosis, suggesting a limited value of this serum cancer marker and a great need for better tumor markers. Moreover, in 24% of patients, HCC showed characteristics of multifocal lesions already at diagnosis, representing one of the main reasons for the exclusion from the liver transplant list.

Considering these results, having a simple score predicting the development of HCC is crucial. Until March 2019, seven scores were investigated in Asian (n = 6) or Caucasian (n = 1) CHB-treated patients. All scores performed well for HCC development prediction in the derivation and validation cohorts (c-statistic: 0.76-0.95) and generally classified patients into low, medium, and high HCC risk groups^[16-22].

Recently, Shen et al. published a meta-analysis including 21,561 patients and 14 predictive scores^[14]. The authors showed that in CHB-treated patients, the HCC risk scores exhibit good performance in predicting the mild-long term HCC risk development. Serum HBV DNA levels are an unimportant factor in these scores; in contrast, the presence of cirrhosis or surrogates of advanced fibrosis (such as APRI index, Liver stiffness, and serum albumin levels) retains an important role.

An ideal HCC risk score should be uncomplicated, inexpensive, and not difficult to calculate; thus, it should include commonly available parameters. Papatheodoridis et al. have developed and validated an HCC risk score (named PAGE-B score) in Caucasian CHB patients treated with ETV or TDF, based only on the patient’s age, gender, and platelet count without the need for mathematical calculation. In our study, the PAGE-B score was strongly associated with the incidence of new HCC cases, as all HCC patients, except two, had a baseline score greater than 18. The test was also strongly predictive of HCC development, showing excellent accuracy and sensitivity.

Our study has some limitations: the retrospective nature of the study, the diagnosis of HCC was based on ultrasonographic findings performed by different radiologists, and all of our patients were from the same geographic area. Nevertheless, the long median duration of follow-up in our patients makes the results of considerable significance.

In conclusion, our results confirm that in a real-life setting, HCC can still occur despite long-term virological suppression. The PAGE-B score could be a helpful tool for identifying high-risk subjects, thus supporting clinicians in the prediction and decisions on HCC surveillance in CHB patients.