Long-term therapy with oral antiviral agents is an effective strategy for patients with chronic hepatitis B virus (HBV) infection. Yet, the possible disadvantages must be considered as well. Although nucleos(t)ide analogs seem to have few side effects, they still have to prove their safety in the long term. Long-term treatment also results in a considerable financial burden on our healthcare systems, and in many countries, patients are not fully reimbursed for the costs of treatment with nucleos(t)ide analogs. Whether nucleos(t)ide analogs are able to induce a sustained off-treatment response is therefore an important focus of research. Unfortunately, it is still unclear for how long treatment using nucleos(t)ide analogs should be continued and what—if any—criteria can be used to stop therapy.
Nucleos(t)ide analogs effectively inhibit viral replication, yet complete eradication of HBV is rarely achieved. HBV covalently closed circular DNA (cccDNA) plays a major role in viral persistence and reduction or clearance of cccDNA is believed to be a mainly immune-mediated process.[1, 2] Previous studies demonstrated that intrahepatic cccDNA is a strong predictor of sustained off-treatment response. In contrast to pegylated interferon, nucleos(t)ide analogs have no direct immunomodulatory activity and only result in a transient, modest improvement of immune reactivity. At present, it appears therefore, from a theoretical viewpoint, unlikely that finite treatment with nucleos(t)ide analogs is able to induce a sustained off-treatment response.
In hepatitis B e antigen (HBeAg)-positive chronic HBV patients, current international guidelines suggest that finite duration of treatment with nucleos(t)ide analogs is a reasonable option, and it is recommended that treatment can be stopped after HBeAg seroconversion and an additional 6-12 months of consolidation therapy.[5, 6] Initial studies, mainly performed in Western countries, reported HBeAg seroconversion achieved during nucleos(t)ide analog therapy to be durable in 80%-90% of cases. However, these studies had major drawbacks, such as inferior endpoints and potential selection bias. Indeed, subsequent studies, mainly performed in Asian countries, demonstrated disappointing results, with virologic relapse rates in up to 70% of patients.[8, 9] Therefore, induction of HBeAg seroconversion by nucleos(t)ide analogs does not appear to result in a sustained off-treatment immune control over HBV in most patients.
In HBeAg-negative HBV patients, less studies have been published on sustained off-treatment response after a finite duration of nucleos(t)ide analog therapy, yet the results appear to be even more disappointing. Withdrawal of lamivudine therapy was associated with a relapse rate of approximately 90% after 6 months of treatment discontinuation when lamivudine was given for 1 year. Longer treatment and the usage of more-stringent cessation criteria improved the sustained response rates; yet, still 50% of patients experienced virologic relapse after 12 months of post-treatment follow-up.[12, 13] Although entecavir and tenofovir are able to maintain virological response better than lamivudine and adefovir during long-term treatment as a result of lower resistance rates, this does not necessarily mean that one could expect any improvement in sustained off-treatment response rates.
Therefore, guidelines of the American Association for the Study of Liver Disease and European Association for the Study of the Liver recommend long-term treatment with nucleos(t)ide analogs until hepatitis B surface antigen (HBsAg) loss or seroconversion is achieved.[5, 6] In most instances, this means indefinite therapy because HBsAg loss is an uncommon event in nucleos(t)ide analog-treated, HBeAg-negative patients. In contrast, the Asian Pacific Association for the Study of the Liver (APASL) guideline suggests that cessation of nucleos(t)ide analog therapy can be considered if undetectable HBV DNA by real-time polymerase chain reaction is documented on three separate occasions at least 6 months apart. Indeed, there are some studies that suggest that cessation of treatment might be possible before HBsAg loss in HBeAg-negative HBV patients.
A recent publication by Hadziyannis et al. studied finite treatment with nucleos(t)ide analogs in HBeAg-negative HBV patients. In a prospective cohort study, 33 HBeAg-negative HBV patients discontinued therapy after 4-5 years of adefovir monotherapy, during which long-term complete viral suppression was achieved. All patients experienced reappearance of serum HBV DNA soon after treatment cessation. Nevertheless, during longer follow-up, 18 of 33 (55%) patients achieved sustained response (persistently HBV DNA <2.000 IU/mL combined with alanine aminotransferase [ALT] <40 U/L), of whom 72% eventually experienced HBsAg loss as well. Quantitative HBsAg levels at the end of treatment were significantly associated with HBsAg loss during longer follow-up. Another small, retrospective cohort study from Asia suggested that a sustained off-treatment response is, in fact, possible in HBeAg-negative HBV patients treated with lamivudine monotherapy, if a marked reduction of HBsAg levels was achieved. These promising results require further investigation and confirmation in larger, prospective studies.
In this issue of Hepatology, Jeng et al. describe the off-treatment durability of response in entecavir-treated, HBeAg-negative HBV patients. It is an observational study of 95 patients who had been treated with entecavir monotherapy and monitored for at least 12 months after treatment cessation according to the APASL stopping rule (undetectable HBV DNA on three occasions at least 6 months apart). During follow-up, virologic relapse occurred in the majority (58%) of patients. The 1-year rate of clinical relapse, defined as HBV DNA >2,000 IU/mL and ALT >2× the upper limit of normal, was estimated to be 45%. Of the 39 patients with cirrhosis at baseline, 17 experienced clinical relapse, which resulted in hepatic decompensation in 1 patient. Although this study uses a potent nucleoside analog, it has several limitations. First, it probably underestimates the proportion of relapsers, because a potential selection bias exists in this study. One hundred and ninety-three patients were excluded because their post-treatment follow-up duration was less than 48 weeks. The reasons for a short follow-up are not specified, but one might speculate that many patients already experienced relapse and were subsequently retreated before an off-therapy follow-up duration of 12 months was achieved. Second, only the 1-year results are described in the article. According to the figure presented, the incidence of post-treatment relapse is expected to increase further with longer follow-up. It thus appears that the glass is half empty and may become increasingly empty with continued follow-up. Third, the identified predictors of relapse seem to have only limited discriminatory value, which makes them insufficient to be useful in clinical practice. Overall, baseline HBV DNA >200,000 IU/mL was identified as a predictor of clinical relapse, yet the area under the receiver operating characteristic curve was only 0.611 (P = 0.063). Sensititvity and specificity values were not provided. In the subgroup of patients without cirrhosis, longer consolidation treatment was identified as well. However, approximately one third of patients who received consolidation therapy for more than 64 weeks still experienced clinical relapse. Another interesting observation in this study is that only 9 of 43 (21%) patients experienced spontaneous remission after initial clinical relapse, which is significantly lower, compared to the study of Hadziyannis et al. (55%). A possible explanation might be the shorter period of maintained remission and shorter consolidation therapy. Lower HBsAg levels may play a role as well, although in the Hadziyannis et al. study, quantitative HBsAg levels were only associated with HBsAg loss in the long term.
Based on the study by Jeng et al., the APASL stopping rule results in approximately 50% relapse within 1 year and also resulted in hepatic decompensation in 1 patient with cirrhosis. Therefore, this study, in our opinion, confirms earlier observations in studies performed with lamivudine[12, 13] that finite duration of nucleos(t)ide analog treatment is not really feasible in the majority of HBeAg-negative HBV patients, even not with the usage of more-potent antiviral drugs and stringent cessation criteria. At this moment, long-term, if not lifelong, treatment may still be considered for the vast majority of patients. This recommendation definitely applies to patients who have already progressed to liver cirrhosis, because withdrawal hepatitis flares can result in subsequent liver failure and death. We also think that in future withdrawal studies, patients with cirrhosis should not be included until more data have become available from patients without advanced liver disease.
Despite our concerns, the possibility that treatment cessation is feasible in a selected group of patients should drive further research in this field. Unfortunately, it is currently unclear what criteria and which markers can be helpful to identify those patients in whom it may be safe to stop antiviral therapy with a low risk of relapse of disease. The usage of quantitative HBsAg levels may help to find these patients who prevent the glass from becoming completely empty.
Jurriën G.P. Reijnders, M.D., Ph.D.1
Harry L.A. Janssen, M.D., Ph.D.1,2
1Department of Gastroenterology and Hepatology Erasmus MC University Medical Center Rotterdam Rotterdam, the Netherlands
2Liver Clinic Toronto Western and General Hospital University Health Network Toronto, Ontario, Canada