Lamivudine monotherapy in HBeAg-negative chronic hepatitis B: prediction of response-breakthrough and long-term clinical outcome
"....The current study represents a comprehensive analysis of the factors associated with initial response and breakthrough phenomenon to lamivudine in HBeAg-negative CHB..."
Alimentary Pharmacology & Therapeutics
Volume 23 Page 787 - March 2006
".....The only factor found to be associated with initial virological and biochemical response was necroinflammation in liver biopsy - HAI score.... patients with HAI >7 at baseline compared with those with HAI ≦7 were two (1, 3) and two (1, 3) times more likely to develop initial virologic and biochemical response, respectively....."
".....Patients with pre-treatment serum hepatitis B virus DNA concentrations of more than 106 copies/mL (1 million) were three times more likely to develop virologic breakthrough (VBR)....
..... Kaplan-Meier estimates for time to first detection of VBR were 10%, 28%, 41%, 52% and 65% at 6, 12, 18, 24 and 36 months, respectively, after the initial virological response..."
.... The median time to VBR development was 24 months after initial virologic response.
On lamivudine... "...reversion from cirrhosis was observed in six patients after a median time of 12 months of therapy...."
"....Of the 34 patients with VBR, 15 experienced BBR (biochemical breakthrough) at a median of 6 (0-18) months after onset of VBR. Ten of the 15 patients developed BBR within 6 months of VBR. Kaplan-Meier estimates for time to first detection of BBR after initial biochemical response were 5%, 8%, 14%, 20% and 27% at 6,12, 18, 24 and 36 months, respectively...."
Authors- S. MANOLAKOPOULOS*, S. BETHANIS*, J. ELEFSINIOTIS , S. KARATAPANIS , C. TRIANTOS , G. SOURVINOS , G. TOULOUMI**, M. ECONOMOU*, J. VLACHOGIANNAKOS , D. SPANDIDOS , A. AVGERINOS , & D. TZOURMAKLIOTIS*
*Department of Gastroenterology, Polyclinic General Hospital; Hepatogastroenterology Section, Elena Venizelou General Hospital; Department of Internal Medicine, Elpis General Hospital; 2nd Department of Gastroenterology, Evangelismos General Hospital, Athens; Department of Virology, Faculty of Medicine, University of Crete, Heraklion; **Department of Hygiene and Epidemiology, University of Athens Medical School, Athens, Greece Deceased.
Chronic HBeAg-negative hepatitis B predominates in the Mediterranean area and Asia with an increasing frequency in Western Europe and North America.13 The HBeAg-negative chronic hepatitis B (CHB) often runs a more aggressive course resulting frequently in cirrhosis, hepatocellular failure and hepatocellular carcinoma (HCC).4, 5
The approval of lamivudine and more recently of adefovir dipivoxil (ADV) has revolutionized the treatment of CHB. Lamivudine is the first oral nucleoside analogue that has been licensed to treat patients with CHB. The profound suppressive effect on hepatitis B virus (HBV) and the excellent safety profile of the drug are the main reasons for its widespread use.69 However, long-term lamivudine treatment has been associated with an increasing rate of virological (VBR) and biochemical breakthrough (BBR) because of drug-resistant mutations.912 Data on clinical significance of breakthrough in patients with HBeAg-negative CHB receiving lamivudine monotherapy have shown that breakthroughs may be associated with worsening of liver histology and clinical deterioration in patients with pre-existing cirrhosis or decompensated liver disease.13 ,14 The recently licensed agent, ADV, has been shown to be effective against wild, precore and lamivudine-resistant forms of CHB.1517 In patients with lamivudine-resistance, biochemical and virological improvement have been shown when ADV is added to on going lamivudine therapy or is given as monotherapy; however, cases with progression to liver failure despite treatment with ADV have been reported.1719 Therefore, the optimal management of patients under long-term lamivudine monotherapy remains an open issue. Clinicians should choose between close follow-up for lamivudine-resistance and prompt initiation of ADV following the breakthrough response vs. pre-emptive administration of both agents. So, prediction of breakthrough during long-term lamivudine therapy is of great importance.
The aim of this study was to determine the factors and characteristics associated with initial response and VBR and BBR in patients with HBeAg-negative CHB receiving long-term lamivudine. We also assessed the frequency of hepatic flares, decompensation and deaths during treatment period in order to evaluate the safety of long-term therapy with lamivudine.
Factors that predict response and breakthrough phenomenon to lamivudine monotherapy in patients with HBeAg-negative chronic hepatitis B have not been well defined.
To determine pre-treatment and on treatment variables that predict initial response and breakthrough in patients with HBeAg-negative chronic hepatitis B receiving long-term lamivudine.
Seventy-nine patients, with chronic HBeAg-negative hepatitis B, who received lamivudine for a median of 31 months were included in the study.
Initial virologic and biochemical response was observed in 73 (92%) and 70 (89%) patients, respectively, while 34 (47%) and 15 (21%) patients developed virological and biochemical breakthrough, respectively. High levels of necroinflammation in liver biopsy were associated with a higher probability of initial virological and biochemical response. Patients with pre-treatment serum hepatitis B virus DNA concentrations of more than 106 copies/mL (1 million) were three times more likely to develop virologic breakthrough. Two patients died, one with baseline cirrhosis because of liver failure during biochemical breakthrough while the second death was liver and treatment unrelated.
In HBeAg-negative chronic hepatitis B, initial response to lamivudine therapy is associated with necroinflammation, while baseline serum hepatitis B virus DNA exceeding 106 copies/mL (1 million copies/ml) is a strong predictor for breakthrough because of drug-resistant mutations. Severe complications are uncommon and are associated with biochemical breakthrough (BBR) and pre-existing cirrhosis.
The current study represents a comprehensive analysis of the factors associated with initial response and breakthrough phenomenon to lamivudine in HBeAg-negative CHB. We systematically examined a substantial number of factors and characteristics in a sizeable number of patients with compensated liver disease. Multivariate analysis clearly demonstrated that necro-inflammation in liver histology and serum HBV DNA were significant independent factors of initial response to lamivudine and breakthrough phenomenon because of drug-resistant mutants, respectively.
Our finding that the rate of initial response is associated with higher scores of necroinflammation in liver biopsy is noteworthy. Patients with pre-treatment HAI scores >7 had 64% higher probability of response compared with those with HAI 7 or less. It seems that lamivudine treatment offers an additional additive benefit in a pre-existing active immune response against HBV. In contrast to HBeAg-positive CHB studies 22, we did not find any association between pre-treatment transaminase levels or serum HBV DNA concentrations and response to treatment. The reason for controversy might be the much more complex course of HBeAg-negative CHB. Therefore, our results clearly pointed out that in HBeAg-negative CHB liver histology has higher accuracy in assessing the disease activity and prediction to anti-viral treatment compared with serology.
Another important finding of this study is the association between baseline serum HBV DNA and breakthrough because of lamivudine-resistant mutations. Using rigorous statistical analysis, we showed that among patients with initial response those with serum HBV DNA exceeding 106 copies/mL at baseline had three times higher probability to develop VBR and so the level of 106 copies/mL at baseline can be used as a cut-off level in identifying patients for closer follow-up during the treatment or candidates for combination therapy. It seems that high levels of viral replication may reduce the magnitude of HBV DNA suppression and thus facilitate the emergence of lamivudine resistance.
According to our results, we did not find mutations in the YMDD motif in five of the 34 patients with VBR. Although these patients are not undoubtedly typical lamivudine resistant cases, we included them in the analysis as VBR patients because they fulfilled the definition for such a breakthrough; in addition, further analysis without their inclusion did not change our findings regarding the prediction of response and VBR(data not shown). Poor drug compliance or mutations in other HBV polymerase motif might be possible explanations of the above observation. It is noteworthy that the addition of ADV 10 mg daily at a median time of 9 months after onset of VBR resulted in virologic and biochemical remission in all these five patients.
In our study, we confirmed previous observations that during long-term lamivudine monotherapy a significant number of changes occur in the precore and core region of viral genome.12, 23 The above findings suggest that during VBR, changes in viral genome can occur and may restore the replication fitness of lamivudine-resistant HBV mutants. However, these changes did not appear either to correlate with YMDD mutations or help in identifying patients who would develop BBR.
Whether lamivudine-resistant mutations could result in clinical deterioration or is associated with higher risk for developing HCC remains an open question. In our study, one patient developed clinical deterioration during BBR and eventually died and one developed HCC. Both patients had baseline cirrhosis. In addition, although hepatitis flares were observed in seven of 15 patients with BBR, severe hepatitis flares were observed in only three patients. None of our patients developed hepatitis flares out of BBR. Our results are compatible with those reported by Di Marco et al. who observed that clinical deterioration occurred only among patients with cirrhosis and compensated liver disease at a rate of 5%.24 On the other hand, Lok et al. reported a lower rate of severe complications and deaths in a large number of HBeAg-positive patients receiving lamivudine.25 Interestingly, in the HBeAg-positive study, hepatic deterioration occurred in five out of eight patients without evidence of baseline cirrhosis. The epidemiological, clinical and histological differences between patients with the two forms of CHB could explain the above discrepancy.
We have previously shown that clinical deterioration and death may occur very soon after the development of BBR because of YMDD mutants after long-term lamivudine monotherapy in patients with HBeAg-negative decompensated HBV cirrhosis.14 The results of the current and previous study recommend that in non-cirrhotics patients ADV can be introduced when VBR or BBR are first detected, but in patients with HBeAg-negative HBV cirrhosis pre-emptive LAM + ADV combination therapy might be the most appropriate approach.
Viral suppression and disease remission might reduce the risk of HCC in patients receiving lamivudine. According to our data and those of Papatheodoridis et al., HCC developed in 2% of treated patients during therapy.26 In contrast, Di Marco et al. observed a higher rate of HCC development and a clear association with breakthrough because of YMDD mutations.24 The retrospective design of the studies and the higher proportion of patients with cirrhosis in the Italian study may explain the above difference.
Data on HBV genotypes from the Mediterranean are limited. Three recently published studies have shown that prevalence of infection with genotype D ranged from 99% to 100%.13, 27, 28 In contrast to these studies, we have shown that 19% of studied patients were infected with genotype E, which is more common in Africa. Further large-scale studies are needed in order to determine the geographical distribution and clinical significance of HBV genotypes in the different countries of Mediterranean basin.
In conclusion, our data suggested that pre-treatment necroinflammatory activity is a significant predictor of response to lamivudine and serum HBV DNA concentrations are significantly associated with breakthrough because of drug-resistant mutants. Patients with serum HBV DNA of more than 1 X 106 copies/mL should be monitored more closely or receive combination therapy because they had more than three times higher probability for VBR. Another conclusion of this study is that severe life-threatening complications of liver disease during the long-term lamivudine administration in patients with compensated HBeAg-negative CHB are uncommon and are associated with BBR in pre-existing cirrhosis. Therefore, our results clearly justify that careful pre-treatment assessment of liver disease severity and serum HBV DNA determination appears to be mandatory in the management and outcome of patients with HBeAg-negative CHB receiving long-term lamivudine. Future well-designed trials are necessary in order to determine the optimal management that these patients require.
Baseline characteristics of the study population are presented in Table 1. Up to March of 2004, the median duration of lamivudine monotherapy was 31 (range 6-56 months. Thirteen patients (17%) went on to receive lamivudine for a second year, 25 (32%) for a third year and 31 (39%) patients were on lamivudine for more than 36 months. None discontinued or decreased the dose of lamivudine because of side effects.
Initial virological and biochemical response
None of the patients lost HBsAg or changed HBeAg status. Initial virological response was observed in 73 (92%) patients within a median time of 3 months. All responders experienced initial virologic response within the first 13 months since treatment initiation (Figure 1a). Among the non-responders (n = 6), a median of 3log10 reduction in serum HBV DNA was observed.
Initial biochemical response was observed in 70 (87%) patients after a median time of 6 months. All responders experienced the initial biochemical response within the first 18 months of treatment (Figure 1b). The only factor found to be associated with initial virological and biochemical response was necroinflammation in liver biopsy - HAI score (P = 0.05 and P = 0.03, respectively)(Tables 2 and 3). It was observed that patients with HAI >7 at baseline compared with those with HAI ≦7 were two (1, 3) and two (1, 3) times more likely to develop initial virologic and biochemical response, respectively.
Long-term consequences of virological and biochemical breakthrough
Of the 73 patients with initial virological response, 34 (47%) developed VBR. The median time of follow-up after the development of VBR was 16 (0-43) months. VBRs were persistent in all patients. The median time to VBR development was 24 months after initial virologic response. At the onset of VBR serum, HBV DNA did not exceed the baseline levels in 29 of the 34 patients. Kaplan-Meier estimates for time to first detection of VBR were 10%, 28%, 41%, 52% and 65% at 6, 12, 18, 24 and 36 months, respectively, after the initial virological response (Figure 2).
Of the 34 patients with VBR, 15 experienced BBR at a median of 6 (0-18) months after onset of VBR. Ten of the 15 patients developed BBR within 6 months of VBR. Kaplan-Meier estimates for time to first detection of BBR after initial biochemical response were 5%, 8%, 14%, 20% and 27% at 6,12, 18, 24 and 36 months, respectively (Figure 3). BBR was persistent in all cases with a fluctuating pattern of transaminase levels. Twelve of the 15 patients developed the highest transaminase activity at the onset of BBR, two patients 3 months and one patient 9 months later. Forty-six per cent of the patients with BBR had hepatitis flares after a median time of 18 months from initial biochemical response; in three patients, transaminase levels reached acute hepatitis levels (transaminase levels >8 ULN). Two hepatitis flares were observed during first year, two during second, one during third and two during fourth year of treatment (Figure 4). Hepatic decompensation was observed in only one patient who eventually died. Bilirubin levels higher than 3 mg/dL were observed in three of seven patients with hepatic flare and prolongation of prothrombin time more than 3 s in one individual. Among patients with hepatic flares, two had pre-existing cirrhosis. Patients with flares did not appear to have higher serum HBV DNA concentrations at or about the time of the peak of transaminase elevations (5.46 log10 vs. 5.23 log10, P = 0.89). No difference in serum HBV DNA concentrations at the onset of VBR between patients who developed BBR more than 6 months after VBR or less than 6 months was observed (4.9 log10 vs. 5.43log10, P = 0.39). All patients with VBR who did not develop BBR were asymptomatic. Results from Cox regression analysis showed that baseline serum HBV DNA concentrations were significant predictors of VBR (Table 4). Among patients with initial virologic response, those with serum HBV DNA concentrations higher than 1 X 106 copies/mL (median value) were three (1-7, P = 0.003) times more likely to develop VBR compared with those with HBV DNA <1 X 106 copies/mL. A ROC analysis confirmed that the baseline HBV DNA levels of 1 X 106 copies/mL was the best cut-off point for identifying subjects at greater risk for developing VBR at 12 months since initial virological response.
At the onset of VBR, mutations involving the YMDD motif were detected in 29 of the 34 patients. L528M + M552V mutants were detected in 11 patients; L528M + M552I in nine; M552I in eight and the M552V mutant strain in one patient. In two of the five patients without detectable YMDD mutants, a reversion to tryptofan at codon 28 was observed at the onset of VBR. In addition, no difference in serum HBV DNA concentrations among these five patients without detectable YMDD mutant strains and those with YMDD mutations was observed (P = 0.26).
Changes in precore and core variants in association with emergence of lamivudine-resistant mutations
The precore and core-promoter region of HBV genome at baseline and during VBR were examined in 33 patients. Before the initiation of lamivudine treatment, four patients harboured the classic precore stop codon mutation at nucleotide 1896 and 20 patients had a mixture of wild-type and precore mutants. The dual core promoter variants were observed in 16 patients. Changes to precore sequence were observed in three patients (10%) with a reversion to tryptofan at codon 28, while changes in the core promoter sequence were observed in 13 patients. We did not observe any association between mutations in the precore/core region and serum HBV DNA, time for BBR development and mutations in the YMDD motif of polymerase gene.
In the 44 patients with paired biopsies, median time between pre-treatment and follow-up biopsy was 12 months (12-48). Lamivudine induced a significant reduction in necroinflammation and fibrosis among the 41 patients who were on virologic and biochemical remission. The median HAI score was reduced from 7 (2-14) to 4 (1-12) (P < 0.001) and the median staging score improved by one point from 3 (0-6) to 2 (0-6) (P = 0.02). None of the 28 out of 41 patients without cirrhosis at baseline progressed to cirrhosis, while reversion from cirrhosis was observed in six patients after a median time of 12 months of therapy. On the contrary, no difference was observed in grading median: at baseline 9 (5-10); at follow-up: 9 (7-10), P = 0.18 and staging median: at baseline 3 (1-6); at follow-up 2 (1-6), P = 0.16 among patients with BBR.
At the time of analysis two (3%) patients had died; the first patient died at the 42nd month of treatment as a result of liver failure after BBR because of lamivudine-resistance. The patient had cirrhosis before starting lamivudine. BBR was observed at the beginning of third year of treatment and ADV was initiated 6 months after decompensation. The second patient died at 36th month of treatment from carcinoma of the pharynx. One patient developed HCC at 14 months after treatment initiation. There were no hepatic flares in patients without lamivudine-resistance. ADV was initiated in 20 patients by March 2004 after a median time of 12 (0-24) months from the onset of VBR. Fifteen of above 20 had VBR and BBR, while five VBR. Lack of drug availability is the reason for the delay in starting ADV administration.
Materials and methods
This retrospective study included 79 consecutive adult patients, with HBeAg-negative CHB, who started lamivudine monotherapy (GSK, Athens, Greece) 100 mg daily between March 1999 and September 2003 at four centres in Greece. They were followed up until March 2004. Criteria for inclusion in the analysis were (a) hepatitis B surface antigen positive and hepatitis B e antigen negative for at least 6 months, (b) elevated alanine (ALT) or aspartate (AST) aminotransferase >1.5 upper limit of normal (ULN) on at least two occasions during the 6 months preceding therapy, (c) detectable serum HBV DNA within the last month before starting therapy and (d) histological findings of chronic hepatitis or cirrhosis (Child-Pugh A) within 12 months before starting the therapy. We also included five patients where the diagnosis of cirrhosis was based on the clinical and imaging data. No patient had decompensated liver disease, detectable antibodies against hepatitis C, D or human immunodeficiency viruses; none had evidence of HCC or had received any anti-viral drug other than interferon-ƒ¿. All patients gave written informed consent and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki.
All patients were followed up in the out-patient clinics and had blood drawn for liver chemistry, serum amylase, creatinine, full blood count and prothrombin time at months 1, 3 and then every 3 months thereafter. Serum HBV DNA was evaluated at least every 6 months.
Virological response was considered the clearance of serum HBV DNA by polymerase chain reaction (PCR) assay and biochemical response as the decline of transaminase levels within normal range. VBR was considered as the reappearance of detectable serum HBV DNA by PCR after initial response and BBR as the increase of AST/ALT levels to >1.5 ULN after initial biochemical response. Hepatic flares were defined as increases in transaminase levels to >3 ULN. Hepatic decompensation was defined as an increase of Child-Turcotte-Pugh score >8 or the occurrence of 1 or more of the following events: ascites, portal hypertensive bleeding or hepatic encethalopathy during the treatment period.
Serum HBV DNA concentrations were determined by a commercially available quantitative assay (AMPLICOR HBV MONITOR assay, Roche Diagnostics, GmbH Mannheim, Germany). The sensitivity of the assay is approximately 400 copies/mL.20 Mutations of amino acid located within the C-domain of the polymerase gene of HBV, PC(G1896A) stop codon and CP(A1762T,G1764A) variants and HBV genotyping of tested serum samples were determined by a line probe assay (Inno-Lipa HBV, Innogenetics, Belgium). Apart from the primers, DNA extraction, subsequent HBV polymerase region amplification, detection and readout were similar and performed according to the manufacturer's instructions.
Where possible, liver biopsies were performed during the treatment period. Indeed, 44 of 79 patients underwent a follow-up liver biopsy (thirty-five at the end of first year of treatment and nine between 2nd and 4th year). Forty-one biopsies were performed during the period of virological and biochemical remission and three during an episode of BBR of more than 6 months (6-12 months). Liver biopsies were evaluated by two experienced pathologists who were unaware of the clinical and virological status of most specimens and did not participate in the study. The histologic lesions were analysed according to Ishak classification system.21
Analyses were performed in stata 8.0 (Stata Corporation, College Station, TX, USA). The clinical, biochemical and virologic data are presented as median with range. Serum HBV DNA concentrations below detection limit (<400 copies/mL) were imputed to 399 copies/mL for the analyses. All HBV DNA values were log-transformed for normality. Continuous variables were compared using the Wilcoxon's rank-sum and Mann-Whitney tests. Categorical variables were compared using chi-squared or Fisher's exact tests as appropriate. Receiver operator characteristic (ROC) analysis was used to determine the cut-off levels of the quantitative factors that best predict response rates. The main part of our work was based on the survival analysis. Time zero was considered the first day of lamivudine treatment. Patients' follow-up was ended at the time of death or administration of another anti-viral agent or last visit if lost to follow-up or at the study termination (March 2004). Time from the first day of treatment to initial virological (or biochemical) response and time from initial virological (or biochemical) response to VBR (or BBR, respectively) were explored with Kaplan-Meier analyses. Cox proportional hazards model was used to evaluate the association of several parameters with the outcomes analysed. The parameters examined included: gender, age, past history of interferon-ƒ¿ treatment, transaminase levels, serum HBV DNA, body mass index (BMI), necroinflammation and fibrosis in liver biopsy, genotype, changes in precore and core variants and serum HBV DNA at 6 months of therapy.