Less can be more: A finite treatment approach for HBeAg‐negative chronic hepatitis B

Potential conflict of interest: Dr. Lampertico advises and is on the speakers' bureau for Gilead, Roche, Bristol‐Myers Squibb, GlaxoSmithKline, MSD, Arrowhead and Alnylam. Dr. Berg consults, advises, is on the speakers' bureau and received grants from Gilead and Bristol‐Myers Squibb. SEE ARTICLES ON PAGES 415 and 425 Long‐term administration of potent nucleos (t)ide analogs (NAs) is the cornerstone of antiviral therapy for hepatitis B e antigen (HBeAg)‐negative chronic hepatitis B, as it reduces liver‐related complications by efficiently suppressing viral replication.1 Although there is agreement that the best and safest stopping rule in this setting is hepatitis B surface antigen (HBsAg) loss, it remains highly debated whether NA discontinuation before HBsAg loss can be recommended, when it should be recommended, and with what aim.1 In this issue of Hepatology, two publications address relevant questions concerning the finite NA approach in HBeAg‐negative patient populations.4 The first study, a collaboration between two expert centers in Greece and Taiwan, aimed to better characterize the patterns of relapse after stopping NA therapy and to elucidate whether an early relapse pattern might be predictive for long‐term outcome and need for retreatment.4 The study included 130 HBeAg‐negative patients without cirrhosis who had on‐treatment undetectable hepatitis B virus (HBV) DNA levels for at least 24 months. Fourteen different relapse patterns were predefined, ranging from very stringent criteria for virologic relapses to more clinically relevant combined virologic and biochemical relapses. The posttreatment cumulative rates at 6, 12, and 24 months were approximately 70%, 80%, and 90% for HBV DNA >200 IU/mL and approximately 55%, 70%, and 70% for HBV DNA >2000 IU/mL, respectively, By using a relapse definition of HBV DNA >2000 IU/mL in combination with any elevation of alanine aminotransferase (ALT) levels, 35%, 50%, and 55% of patients would have been considered to have a clinically relevant relapse by 6, 12, and 24 months, respectively. No liver decompensation or death occurred. Cumulative retreatment rates, however, did not mirror the relapse rates and were only 15%, 22%, and 40% after 6, 12, and 24 months, respectively, after NA discontinuation. Although no strong and independent predictors of virologic relapse were identified, the probability of relapse defined as HBV DNA >2000 IU/mL and ALT above the upper limit of normal was found to be significantly associated with the use of tenofovir compared with entecavir therapy in multivariate analyses (P = 0.029). Likewise, the need for retreatment could not be independently predicted by any of the available markers in multivariate analyses. However, HBsAg levels at baseline or at NA cessation, a parameter that has been shown to predict postcessation outcomes in other studies, were not available in this study, limiting the conclusions regarding relapse and retreatment prediction. The second study, by Jeng et al., reviewed the incidence and predictors of HBsAg seroclearance after NA cessation in 691 Taiwanese patients with HBeAg‐negative chronic hepatitis B who discontinued oral therapy according to the Asian Pacific Association for the Study of the Liver guidelines.5 In this study, all patients received tenofovir or entecavir, and 308 patients (45%) had cirrhosis. During the off‐therapy follow‐up of approximately 3 years, the cumulative virologic and clinical relapse rates were 79% and 61%, respectively. Forty‐two patients had confirmed HBsAg clearance, with a 6‐year cumulative incidence of 13%, 9% in patients with cirrhosis and 16% in patients without cirrhosis. End‐of‐treatment (EOT) HBsAg levels strongly predicted HBsAg loss. The 4‐year probability of HBsAg loss was approximately 30% in patients with EOT HBsAg <100 IU/mL compared with 0% in patients with >100 IU/mL. Additionally, patients with a clinical relapse who remained untreated had a 7.4‐fold greater incidence of HBsAg clearance compared with those who were retreated. Of note, the 5‐year cumulative incidence of decompensation was approximately 3% in patients with cirrhosis, and three patients died despite prompt retreatment for HBV. In contrast, none of the patients without cirrhosis decompensated. The 6‐year cumulative incidence of hepatocellular carcinoma (HCC) among patients with cirrhosis was 9%, an incidence that parallels the expected HCC rates among patients with cirrhosis on long‐term NA treatment. What can we learn from these studies? First, when stopping NA treatment, a virologic relapse is nearly universal, even after prolonged periods of HBV DNA suppression. The dimension of the relapse, however, is highly variable and can range from only mild HBV DNA rebounds to clinically significant ALT flares, which normally follow a marked HBV DNA rebound (Fig. 1). A flare‐associated decompensation of liver disease is rare and is observed only in patients with cirrhosis. Although the cumulative decompensation rate was low (3% among patients with cirrhosis), three patients (1%) died, a finding that in our view should be taken as a strong argument not to stop NA treatment in patients with cirrhosis. Whether an increase in viral replication after NA cessation may also affect the HCC risk has to be further proven in long‐term follow‐up studies, but the findings by Jeng et al. do not support a major concern in this respect.Figure 1: Illustration of the typical courses observed in patients with HBeAg‐negative chronic hepatitis B after stopping long‐term NA treatment, which can be divided into four different phases. First, in the lag phase after NA cessation, HBV DNA is still suppressed for several weeks or months. The duration of the lag phase likely depends on the duration of viral suppression under NA, and the NA type. Second, in the reactivation phase, most (but not all) patients show some level of HBV DNA rebound, which is often followed by an increase in ALT levels. Severe hepatic flares may occur in this phase and require immediate retreatment. Third, in the consolidation phase, patients can be separated into those with transient flares entering a low‐replicative “carrier” state or those developing some degree of disease activity. The consolidation phase is predictive for what will be the final outcome in the long term after NA cessation. In this latter phase, four different outcome categories can be defined. Some patients can still be in an intermediate state, neither fulfilling criteria for chronic hepatitis B nor for an inactive carrier state, and therefore require further follow‐up for a precise disease activity characterization and to decide whether retreatment should be commenced. Abbreviations: ALT, alanine aminotransferase; ETV, entecavir; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; NA, nucleos(t)ide analog; TDF, tenofovir.In our view, the principal aim of stopping NA is to induce a durable remission of HBV infection, either in the form of a healthy carrier state with undetectable viremia, or ideally a functional cure with HBsAg loss. In this respect, a virologic relapse could represent a prerequisite for achieving this kind of stable retreatment‐free virologic remission, which can then be followed by a functional cure (i.e., HBsAg loss) at later stages (Fig. 1). Indeed, the results reported by Jeng et al. are clearly in line with these assumptions, showing a 6‐year cumulative HBsAg loss incidence of 16% in patients without cirrhosis; however, the authors also found that patients with clinical relapses that remained untreated had a significantly higher incidence of HBsAg clearance compared with those who were retreated. What is needed are better definitions of what type of relapse might associate with a later response but also rules for immediate retreatment indications to prevent harm to our patients. In addition, it seems that starting retreatment too early might counteract the potential relapse‐associated induction of a functional cure. Thus, a better definition of when to start early and when to wait is needed. Second, predicting which patients will develop a clinically significant relapse with a high level of accuracy is impossible at present, although according to the current studies and other reports from Asia, several viral‐, host‐, and treatment‐related parameters show some association with the relapse pattern as highlighted in Figure 1. Unfortunately, none of the current studies identify strong and easily measurable EOT predictors that can identify patients who are more likely to achieve higher rates of functional cure. The Taiwanese study identified low HBsAg levels at EOT (<100 IU/mL) as the strongest predictor of response, but HBsAg levels were not available in the second study. In the 2012 landmark article by Hadziyannis et al.,6 all patients with EOT HBsAg <1000 IU/mL had a sustained virologic response and/or HBsAg loss. In contrast, in the recent FINITE study, which is the only prospective study performed so far, an HBsAg cutoff of 25,000 IU/mL separated the patients with and without HBsAg response.7 The development of new HBV biomarkers, such as HBV RNA, hepatitis B core‐related antigen, and HBsAg fragments, among others, may provide new clues in this respect but have not been prospectively assessed in the setting of patients discontinuing oral therapy before HBsAg loss.3 Another relevant question is how these results may impact the study design of new therapeutic approaches that are designed to increase the rates of functional cure.3 Entry inhibitors, Capsid Assembly Modulators, small interfering RNA, covalently closed circular DNA targeting drugs, release inhibitors, and immunomodulators are under preclinical and early clinical development with the aim to foster HBsAg rates among NA‐treated patients to stop oral therapy. Apart from the challenging situation of developing an effective, safe, and inexpensive combination therapy with two or three new agents with the aim of stopping oral therapy, these new strategies face a new problem. If the significant rates of functional and/or partial cure observed in selected patients who discontinue NA before HBsAg loss are confirmed by additional studies, then any new therapeutic regimen will have to demonstrate cure rates exceeding 20% after 1 year of being off therapy. In conclusion, the evidence is increasing that a finite NA treatment approach leads to higher HBsAg loss rates compared with the current long‐term NA strategy and can be considered a rational strategy to induce a functional cure in selected HBeAg‐negative patients without cirrhosis who are willing to comply with close follow‐up monitoring. To reduce the drawbacks associated with this approach on an individual level in terms of close monitoring, uncertainties when to start retreatment, and the potential risk of severe hepatitis flares, better definitions of the patient populations benefitting from the finite approach are needed. If we were able to exclude the 40%‐50% of patients who restart NA therapy, then the current observed functional cure rates—which will then reach approximately 40%—would be well worth the effort. Ongoing large trials evaluating the finite NA approach prospectively in HBeAg‐negative patients will hopefully answer most of the unanswered questions. This will have a major impact on the study design and the off‐treatment endpoints of studies involving new therapeutic strategies.