The goal of chronic hepatitis B (CHB) treatment is to achieve seroclearance of HBsAg. Nucleos(t)ide analogues (NAs) are one of the first-line treatments for CHB. NAs produce a potent suppression of viral replication but are associated with a low rate of HBsAg seroclearance and a high risk of virological relapse after discontinuation. Because of these reasons, long-term treatment is needed. They are well-tolerated oral drugs, and it seems they do not produce important side-effects in long-term administration. The duration of NA treatment remains unclear, nevertheless, in some patients NAs can be stopped with a low rate of relapse. HBeAg-positive patients could discontinue NA therapy if they achieved HBeAg seroclearance and maintain undetectable HBV DNA. For HBeAg-negative patients, to stop NA treatment is not recommended. In addition to other factors, serum HBsAg titres during treatment have recently been proposed to guide NA-based therapy duration in selected patients. All patients could be stopped from taking treatment if they achieve HBsAg loss.
Hepatitis B virus (HBV) infection is an important public health problem. Approximately one-third of the world's population has serological evidence of past or present HBV infection, and 350–400 million people are chronic HBsAg carriers. The spectrum of disease and natural history of chronic HBV infection are diverse and variable, ranging from an inactive carrier state to progressive chronic hepatitis B (CHB), which may lead to the development of cirrhosis or hepatocellular carcinoma (HCC) [1-5]. Patients with CHB can be HBeAg-positive or HBeAg-negative . Serum HBsAg appears to correlate with the presence of covalently closed circular DNA (cccDNA) and is considered a surrogate marker of infected cells [4, 6]. Seroclearance of HBsAg is the closest event to a cure of HBV infection and it is the main goal of CHB therapy . However, the time to HBsAg loss is important. A loss of HBsAg before the onset of cirrhosis is associated with a more favourable outcome of the infection; that is, a lower risk of cirrhosis, decompensation, and HCC .
The currently available therapies for CHB are interferon (IFN) [standard or pegylated (PEG)] and nucleos(t)ide analogues (NAs) [1, 2, 4], including lamivudine (LAM), adefovir (ADV), telbivudine (TBV), tenofovir (TFV), and entecavir (ETV). TFV and ETV are considered first-line treatments because of their potent antiviral activity and high barrier to resistance [1, 2, 4]. However, most of the studies investigating long-term treatment have been performed with LAM and ADV, as they were the first approved drugs. IFN-based therapy has been found to induce HBsAg seroclearance in HBeAg-positive and HBeAg-negative patients  and it is a finite therapy . However, NA treatment must be long term, as the annual rate of HBsAg loss is very low and virological relapse is common after discontinuation of treatment [5, 8, 9].
This article reviews the factors that can predict HBsAg loss, and discusses the importance of HBsAg loss and other variables that can help identify patients who are candidates for discontinuing antiviral therapy with NAs.
Importance of HBsAg loss
HBsAg seroclearance is considered to be the closest event to a cure of CHB. Patients who achieve HBsAg loss have a more favourable prognosis: survival is greater, there is a lower risk of liver disease-related decompensation and in some cases, HCC, and liver fibrosis may regress .
Closed circular DNA is the transcriptional template of HBV. It exists in the cell nucleus as a viral minichromosome and serves as the intrahepatic reservoir for HBV . HBsAg is one of the subviral replication products of cccDNA, so HBsAg level reflects the transcriptional activity of cccDNA .
Clearance of intrahepatic cccDNA is the main challenge of antiviral therapy for CHB . However, in clinical practice, it is difficult to obtain liver tissue from patients to study cccDNA levels, and serum HBsAg level is used as a partial surrogate marker of this molecule .
Spontaneous HBsAg seroclearance, defined as serum HBsAg loss on two analyses at least 6 months apart and persisting to the last visit, is a rare event in the natural history of CHB infection, occurring in 0.12–2.38% of patients per year in cohorts from Asian countries and in 0.54–1.98% of patients in cohorts from western countries [7, 10]. The cumulative HBsAg seroclearance rate is 8.1% at 10 years, 24.9% at 20 years and 44.7% at 25 years of follow-up [7, 10].
Several viral and host factors have a significant association with HBsAg seroclearance. Older age, normal ALT levels, the presence of cirrhosis or fatty liver, negative status for HBeAg or HBV DNA, genotypes A and B, adr HBsAg serotype, and having an acute viral superinfection are the factors associated with higher rates of HBsAg loss .
Up to 80% of patients achieve detectable anti-HBs (antibody against HBsAg) after HBsAg seroclearance, but the percentage varies during follow-up: only 17% of patients have detectable anti-HBs within 1 year after HBsAg seroclearance, 56% are positive carriers for anti-HBs after the fifth year, and 76% are positive after the tenth year .
At the time of HBsAg seroclearance, most patients test negative for HBV DNA. Despite the extremely low viremic state, 5–18% of carriers have abnormal ALT levels after HBsAg seroclearance. However, in these cases, non-HBV-related etiologies of abnormal ALT levels can be identified in 75–100% of patients [7, 10].
Liver function can improve or remain stable at the time of seroclearance in carriers with cirrhosis who have no evidence of viral superinfection, and hepatic decompensation rarely occurs. However, HCC can still develop, although at a very low rate, especially when cirrhosis is established before HBsAg seroclearance or when HBsAg seroclearance occurs at an older age [4, 7, 10].
Recent studies have shown that serum HBsAg titres strongly correlate with intrahepatic cccDNA levels [4, 6, 10, 11], and several authors have proposed that HBsAg quantification may be a useful marker to monitor IFN treatment . Serum HBsAg levels tend to be higher in HBeAg-positive than in HBeAg-negative patients and also differ according to HBV genotype. Genotype A is associated with the highest HBsAg levels, followed by genotype B and then further behind genotype C, whereas levels are lowest in genotype D . Furthermore, HBsAg levels vary during the natural course of HBV infection and can reflect the stage of the disease. HBsAg is highest in the immune-tolerant phase (4.5–5.0 log10 IU/ml), begins to decline during the immune-clearance phase (3.0–4.5 log10 IU/ml), and decreases slowly and progressively after HBeAg seroconversion, with no differences compared with HBeAg-positive patients who have persistently active disease [4, 11].
Spontaneous HBeAg seroconversion occurs in 2–15% of patients per year, and depends on factors such as age, ALT levels and HBV genotype . Seroconversion from HBeAg to anti-HBe (antibody against HBeAg) leads to an ‘inactive HBV carrier state’ characterized by very low or undetectable serum HBV DNA levels (<2000 IU/ml) and normal serum aminotransferases, and is associated with a favourable long-term outcome and very low risk of cirrhosis or HCC . However, in 2–3% of patients per year, relapse occurs as a result of HBeAg seroreversion or HBeAg-negative CHB, characterized by fluctuating HBV DNA levels and aminotransferases, and active hepatitis, with a high risk of progression to cirrhosis and development of HCC .
Following international guidelines
The ideal endpoint of therapy is HBsAg loss, but this is often not attained with the current anti-HBV therapy. There are two treatment strategies for HBeAg-positive and -negative CHB: IFN (conventional or PEG-IFN) or NAs .
The recommended duration of IFN-based therapy is finite, regardless of the response to treatment. Compared with conventional IFN, PEG-IFN therapy has the advantage of once-weekly dosing. The currently recommended duration of PEG-IFN-based therapy is 48 weeks for both HBeAg-positive and -negative patients .
A finite course of treatment with NAs is feasible in HBeAg-positive patients who seroconvert to anti-HBe on treatment and have undetectable serum HBV DNA levels. Once anti-HBe seroconversion has occurred, treatment should be maintained for an additional 6 months, according to the American Association for the Study of Liver Diseases (AASLD) guidelines, and 12 months according to the European Association for the Study of the Liver (EASL) guidelines and the Asian Pacific Association for the Study of the Liver (APASL) guidelines, to reduce relapses [1, 2, 4] (Table 1).
Table 1. Recommendations of international guidelines for discontinuation of NA treatment
AASLD, American Association for the Study of Liver Diseases; APASL, Asian Pacific Association for the Study of the Liver; EASL, European Association for the Study of the Liver NA, nucleos(t)ide analogue.
Anti-HBe positive and undetectable DNA for 12 months
Anti-HBe positive and undetectable DNA for 12 months
Anti-HBe positive and undetectable DNA for 6 months
Minimum 2 years with undetectable DNA
In the EASL and AASLD guidelines, long-term treatment with NAs is recommended in HBeAg-positive patients who do not show anti-HBe seroconversion and for HBeAg-negative patients . On the other hand, the APASL guidelines suggest that treatment discontinuation can be considered in HBeAg-negative patients if the patients have been treated for at least 2 years and undetectable HBV DNA has been documented on three separate occasions, 6 months apart . In the AASLD guidelines, treatment should be continued in HBeAg-negative patients until HBsAg clearance is achieved .
Response to and discontinuation of IFN therapy
Conventional and PEG-IFN have both proven effective for treating CHB by their antiviral and immunomodulator activity. The immunomodulator effect can prevent infection in previously uninfected hepatocytes and clear infected hepatocytes . It has been shown that PEG-IFN can improve the prognosis and reduce the risk of HCC . In addition, PEG-IFN-based treatment reduces serum HBsAg levels .
Half of responders to IFN treatment, defined as patients with HBeAg loss during 12 months after the end of treatment, achieve HBsAg clearance compared with 9% of non-responders (P < 0.001) .
Interferon is associated with an approximately three-fold increased probability of HBsAg seroclearance in western countries and six-fold in Asian countries compared with untreated patients. The mean annual rate of HBsAg seroclearance in conventional IFN-treated patients is 2.6–4.4%, but it is much lower in Asian patients .
HBeAg seroconversion is achieved in 20–40% of cases, and delayed HBeAg seroconversion occurs in 10–15% of patients at 1 or 2 years post-treatment. The 15-year cumulative incidence of HBeAg seroconversion is up to 75% .
A sustained virological response (SVR), in which HBV DNA is undetectable during follow-up, is seen in 22–30% of IFN-treated patients. In a study with IFN therapy extended to 24 months, the reported rate was 18% of HBsAg loss at the 6-year follow-up .
HBsAg loss is reached in 3–7% of patients receiving PEG-IFN after long-term follow-up . Viral response to PEG-IFN-based therapy in HBeAg-positive patients, defined as HBeAg loss and HBV DNA <10 000 copies/ml at 6 months post-treatment, is achieved in approximately 25% of patients [4, 15].
A decline in HBsAg serum titres at weeks 12 and 24 during PEG-IFN treatment can be used as a predictor of SVR in patients with HBeAg-positive CHB . It has been reported that low HBsAg levels or a greater HBsAg decline earlier during treatment is associated with higher HBeAg seroconversion rates and HBV DNA suppression 6 months post-treatment .
Chan et al. reported that patients with an HBsAg decline >1 log10 and serum level ≤300 IU/ml at month 6 during treatment achieved a SVR rate of 75% compared with 15% in those who did not have these criteria, with a negative predictive value (NPV) of 85% [11, 16]. In the NEPTUNE study, patients with HBsAg titres >20 000 IU/ml at weeks 12 or 24 of PEG-IFN-α-2a were considered not to have achieved a post-treatment response, and this would be used as a stopping rule. Patients with HBsAg levels <1500 IU/ml at week 12 of PEG-IFN attained an HBsAg clearance rate of 17.6% at 6 months post-treatment  (Fig. 1).
Sonneveld et al.  proposed that any decline in serum HBsAg titres at week 12 of PEG-IFN treatment would be a good predictor of response, with a NPV of 97%, and that patients who did not show a decline should be considered for treatment discontinuation. However, Gane et al.  estimated an 18–29% probability of SVR in patients with no HBsAg decline from baseline to week 12, and suggested that applying this stopping rule would prematurely stop treatment in some patients. The on-treatment decline in HBsAg titres can vary according to HBV genotype (genotypes A and B have steeper HBsAg declines than genotypes C and D) ; thus, further study based on genotype is needed to identify and validate a stopping rule.
In HBeAg-positive patients, the baseline factors that best predict SVR (HBeAg loss and HBV DNA level <2000 IU/ml 6 months after treatment) are HBV genotype A and B, high ALT levels (≥two-fold the upper limit of normal), low HBV DNA levels (<2.0 × 108 IU/ml), female gender, older age and absence of previous IFN therapy . The best candidates for a sustained response to PEG-IFN α are genotype A patients with high ALT or low HBV DNA levels, and genotype B and C patients who have both high ALT and low HBV DNA. Genotype D patients have a small chance of sustained response (8% of genotype D patients) and are not good candidates for PEG-IFN therapy .
In HBeAg-negative virological responders treated with PEG-IFN, HBsAg loss occurs at a rate of approximately 10% per year . In a long-term follow-up study, HBsAg loss was achieved in 44% patients who maintained sustained HBV DNA suppression to undetectable levels 3 years after treatment .
On-treatment quantification of HBsAg in HBeAg-negative patients treated with PEG-IFN identifies those likely to achieve HBsAg loss. An HBsAg level <10 IU/ml at week 48 and on-treatment decline of >1 log10 IU/ml are significantly associated with sustained HBsAg clearance 3 years after treatment . The likelihood of HBsAg loss in patients who achieve HBsAg levels <10 IU/ml at the end of treatment is 53% at 3 years .
Viral response to PEG-IFN therapy, defined as HBV DNA <2000 IU/ml and normal ALT 24 weeks post-treatment is achieved in 25% of HBeAg-negative patients .
A cut-off of 1 log10 IU/ml decrease in serum HBsAg at week 24 of therapy has a 97% NPV of virological response. Thus, this criterion would allow physicians to stop PEG-IFN and switch to another antiviral therapy .
In one study among HBeAg-negative patients predominantly infected with HBV genotype D and treated with PEG-IFN ± RBV, patients who failed to achieve a decline in HBsAg levels and an HBV DNA drop of >2 log10 IU/ml at week 12 of PEG-IFN-α-2a did not achieve virological response. Thus, this criterion was validated as a stopping rule to optimize the cost-effectiveness of PEG-IFN by avoiding unnecessary treatment [1, 11, 22] (Fig. 2).
Baseline factors that have been associated with a SVR in HBeAg-negative patients (defined as ALT normalization and HBV DNA level of <20 000 copies/ml 24 weeks post-treatment) treated with PEG-IFN-α-2a are younger age, female gender, high baseline ALT, low baseline HBV DNA, and HBV genotypes B and C, with genotype D being associated with a poorer response to treatment than the other genotypes .
Response to and discontinuation of NA therapy
Nucleos(t)ide analogues, particularly TFV and ETV, are the most commonly used drugs for the treatment of CHB, because of their potent antiviral activity and minimal or null associated drug resistance [1, 2, 4]. Because of the high antiviral potency of these drugs, most patients achieve undetectable HBV DNA in long-term treatment, even those with a high viral load. In addition, several studies have shown that NAs can improve liver fibrosis, and lead to the regression of cirrhosis and carcinogenesis . NAs are orally administered drugs with minimal side effects and can be used in all stages of the disease, including advanced and decompensated cirrhosis [1, 8].
Nucleos(t)ide analogues produce a rapid reduction in HBV DNA and normalization of serum ALT levels . However they have little effect on HBV cccDNA, that maintains viral replication and is responsible for relapses [7, 25]. Thus, HBsAg seroclearance is uncommon, usually <5% at 5 years in HBeAg-positive patients [7-9]. The potential limitation of NAs is that the treatment should be long term as the percentage of HBsAg loss is very low and there is often virological relapse after treatment discontinuation, even when HBeAg loss has been achieved [5, 8, 9, 26].
Lamivudine, which was introduced as HBV therapy in the early 1990s, revolutionized the treatment for this condition as it was the first drug to reduce the incidence of hepatic decompensation and the risk of HCC [2, 27]. However, it is now known that a high rate of mutations develop under LAM, with emergence of resistance to the treatment; thus, this drug is not currently recommended as a first-line therapy for HBV infection . In addition, LAM resistance is an important factor in the virological relapse to other NAs .
Therapy with LAM is associated with low HBsAg seroclearance (0–1% at 12 months of treatment in HBeAg-positive patients and 0% at 12 months of treatment in HBeAg-negative) and the rate of HBeAg seroconversion after a year of treatment is 16–18% . At 9 years of treatment, the rate of HBeAg and HBsAg loss or seroconversion is 95% and 7% respectively . Loss of HBeAg progressively increases during treatment. Virological relapse is frequent and rapid (over 25%, often in the first year after treatment discontinuation) [8, 27]. Nevertheless, some studies have shown that virological relapse may be low in patients who achieve low HBV DNA levels (<50 copies/ml) . HBsAg seroclearance occurs less frequently in patients treated with LAM than in those receiving PEG-IFN .
Adefovir, another nucleotide analogue, has been evaluated as primary monotherapy for patients with CHB, and in patients who develop resistance to LAM . However, it is more expensive and less effective than TFV and can result in high rates of resistance .
At 12 months of treatment, 0% of both HBeAg-positive and HBeAg-negative patients develop HBsAg seroclearance, and the rate of HBeAg seroconversion is 12–18% . However, in a study performed in HBeAg-negative patients who received long-term treatment with ADV, approximately 5% of cases achieved HBsAg seroclearance . After 5 years of treatment in HBeAg-positive patients, 58% attained HBeAg loss, 48% HBeAg seroclearance, and 2% HBsAg loss . In 2012, Hadziyannis et al. reported a follow-up study with HBeAg-negative patients previously treated with ADV for 4–5 years. All patients had undetectable viral load after this period; thus, antiviral treatment was discontinued and patients were monitored for almost 6 years. At the end of the study, 55% of patients had achieved a SVR and 39% of patients lost HBsAg . However, HBV DNA became detectable in all cases, in most within 2 months post-treatment. Nonetheless, this increase was transient and low in most of the patients, and HBsAg loss gradually increased up to the end of follow-up. Of note, 45% of patients reinitiated antiviral therapy during follow-up.
Telbivudine is a potent inhibitor of HBV replication, however, it has a lower barrier to resistance and selects for the same resistant mutants as LAM. In addition, it is a more costly option . The efficacy of TBV is poorer than that of ETV at 2 years of treatment and resistance is higher . Thus, its role as primary therapy is limited.
A 12 months of treatment, 0.5% of HBeAg-positive and 0% of HBeAg-negative patients achieved HBsAg seroclearance. The HBeAg seroconversion rate was 22% after 1 year of treatment  and 46% after 2 years of treatment . Recent studies with TBV have shown a rapid decline in HBsAg, with HBsAg seroclearance and SVR . In a study in HBeAg-positive and -negative patients, Xuefen Li observed a remarkable decline in HBsAg levels in the first 12 weeks of TBV treatment. Baseline HBsAg titres were lower in responders, as was the baseline HBV DNA, which presented a significant decrease after 52 weeks of treatment. However, none of the patients studied achieved HBsAg seroconversion after 52 weeks of treatment with TBV .
Tenofovir is a potent, selective inhibitor of viral activity that is effective against LAM-resistant HBV and superior to ADV. At 48 weeks of treatment in HBeAg-positive patients, 3.2% treated with TFV achieved HBsAg loss compared with 0% treated with ADV. In contrast, at week 48 in HBeAg-negative patients, 0% of cases achieved HBsAg seroclearance with either TFV or ADV .
At 3 years of treatment with TFV, 8% of HBeAg-positive patients showed HBsAg loss, and 6–7% achieved HBsAg seroconversion. None of HBeAg-negative patients developed HBsAg seroclearance during this treatment period  (Table 2). At 5 years of treatment, 10% of HBeAg-positive patients achieved HBsAg seroclearance and 8% showed seroconversion. Only one HBeAg-negative patient had HBsAg loss in the same treatment period.
Table 2. HBsAg loss with first-line NAs at 1, 3 and 5 years of treatment in HBeAg-positive and -negative patients
NAs, nucleos(t)ide analogues.
The rate of HBeAg seroconversion after 1 year of treatment with TFV was over 21% , and after 3 years of treatment 34% of patients achieved HBeAg loss and 26% HBeAg seroconversion . At fifth year, HBeAg loss was 49%, with 40% of HBeAg seroconversion .
Entecavir is another first-line treatment for HBV infection. At 12 months of follow-up, 2% of HBeAg-positive and 0% of HBeAg-negative patients achieved HBsAg loss [1, 35] (Table 2). The rate of HBeAg seroclearance at 1 year of treatment was 21% .
In a 96-week study with ETV in HBeAg-positive patients, HBsAg loss was observed in 5% of patients and HBeAg seroconversion in 31% . In a study with 95 HBeAg-negative patients treated with ETV during 2 years and with 1 year of post-treatment follow-up, none of the cases showed HBsAg loss during treatment or at 1 year after discontinuation. Five patients achieved a >1 log10 reduction in HBsAg during treatment, and three patients relapsed afterwards. There was no difference in the extent of HBsAg decline from baseline to 6 months of therapy between relapsers and non-relapsers .
Table 3. Indications for discontinuation of NAs
Indications for discontinuation of NAs
HBV, hepatitis B virus; NAs, nucleos(t)ide analogues.
All patients: HBsAg loss
HBeAg-positive patients: HBeAg loss and undetectable HBV DNA
HBeAg-negative patients: undetectable HBV DNA and very low titres of HBsAg
All other patients
Predictors of HBsAg loss in NA treatment
According to the EASL guidelines, low viral load (HBV DNA <2 × 108 IU/mL), high serum ALT levels, and high activity scores on liver biopsy in the pre-treatment stage are predictors of HBeAg seroconversion .
During treatment, HBsAg is seen to decline more dramatically in HBeAg-positive than HBeAg-negative patients [34, 38], which suggests that HBeAg loss could favour HBsAg seroclearance . The rate of HBV DNA suppression in the first months of treatment is also considered a predictor of HBsAg loss [7, 38].
Unlike the situation with IFN, HBV genotype does not have an impact on the virological response to any NA . However, patients infected with HBV genotype A have a higher tendency to achieve HBsAg loss .
It has been observed that HBsAg titres seem to be related to virological response. In 2011, Heathcote et al. proposed that patients who achieve HBsAg loss had higher median baseline HBsAg levels and a greater median change from baseline compared with patients who do not cleared HBsAg . In contrast, studies published in 2013, show baseline serum HBsAg titres <1000 IU/ml are significantly associated with HBsAg loss , even as an independent factor of seroclearance . In asymptomatic HBeAg-negative patients, with CHB, baseline HBsAg titres can predict HBsAg seroclearance .
It has been proposed that the HBsAg decline during antiviral therapy with NAs reflects an improvement in the degree of host immune control against the virus . An HBsAg reduction rate of >0.166 log IU/ml/year has been suggested as a predictor of HBsAg seroclearance, with a high NPV .
Finally, lower HBsAg levels at completion of treatment are associated with maintained remission after therapy discontinuation and higher levels with a greater risk of virological relapse. Low HBsAg titres at end of treatment could predict a subsequent HBsAg loss  and an HBsAg level ≤2 log10 IU/ml at the end of the treatment has been related with a 9% probability of virological relapse at 6 months and 1 year of follow-up . Therefore, cessation of NAs may be an option in patients with low HBsAg titres [8, 30, 38, 41].
In conclusion, discontinuation of antiviral therapy with NAs should be recommended in all patients who achieve HBsAg seroclearance. In addition, cessation of NAs is an option in HBeAg-positive patients who achieve HBeAg seroclearance and undetectable HBV DNA. NA cessation is controversial in HBeAg-negative patients with chronic hepatitis, a persistently undetectable viral load, and very low HBsAg titres. In patients with cirrhosis, the risk of a hepatitis flare with potential disease decompensation must be considered; thus, in these patients, therapy should not be discontinued unless HBsAg loss occurs [25, 28] (Table 3).
Drs. Pérez-Cameo and Pons do not have any disclosure to report. Dr. Esteban is Advirsory of Gilead, MSD, Novartis and BMS.