Approximately 400 million people worldwide have. Hepatitis B Virus DNA Levels at Week 4 of Lamivudine Treatment Predict the 5-Year Ideal Response

Transcription

1 Hepatitis B Virus DNA Levels at Week 4 of Lamivudine Treatment Predict the 5-Year Ideal Response Man-Fung Yuen, Daniel Yee-Tak Fong, Danny Ka-Ho Wong, John Chi-Hang Yuen, James Fung, and Ching-Lung Lai The best time and hepatitis B virus (HBV) DNA level during an early lamivudine treatment period for predicting the long-term outcome are unknown. We aimed to determine the optimal time and HBV DNA level during an early treatment period for the prediction of the response after a 5-year lamivudine treatment. The HBV DNA levels at the baseline, at weeks 2, 4, 8, 12, 16, 24, and 32, and at yearly intervals until year 5 were measured in 74 hepatitis B e antigen (HBeAg) positive chronic HBV patients receiving lamivudine treatment. Seventeen patients achieved an ideal response [HBV DNA level < 2000 copies/ml (400 IU/ ml), HBeAg seroconversion, normal alanine aminotransferase levels, and absence of tyrosine-methionine-aspartate-aspartate (YMDD) mutations] at year 5. Receiver operating characteristic curves showed good predictions as early as week 4. The areas under the curve for weeks 4 and 16 were 0.89 and 0.94, respectively. Predictive indices revealed 4 and 3.6 log copies/ml (2000 and 800 IU/mL, respectively) to be the best cutoff HBV DNA levels for these 2 times, respectively. All patients with HBV DNA levels lower than these respective cutoff levels at the 2 times achieved an ideal response at year 5. Patients with HBV DNA levels above these cutoff values had 83.8% and 87.7% chances of not achieving an ideal response at year 5, respectively. Conclusion: The measurement of the HBV DNA levels at week 4 of lamivudine treatment should be performed in all patients to predict the long-term outcome. The treatment can be continued for those with HBV DNA levels of less than 4 log copies/ml (2000 IU/mL). The addition of or switch to alternative antiviral agents should be considered for patients who fail to achieve this early target. (HEPATOLOGY 2007;46: ) Approximately 400 million people worldwide have chronic hepatitis B (CHB) infection, of whom 25%-40% will develop long-term sequelae of hepatocellular carcinoma (HCC) and/or cirrhotic complications. 1 The prevention of disease progression is the Abbreviations: ALT, alanine aminotransferase; anti-hbe, antibody to hepatitis B e antigen; AUC, area under the curve; CHB, chronic hepatitis B; CI, confidence interval; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; PCR, polymerase chain reaction; ROC, receiver operating characteristic; SE, standard error. From the Departments of Medicine and Nursing Studies, University of Hong Kong, Queen Mary Hospital, Hong Kong. Received April 30, 2007; accepted July 23, Supported by the Hepatology Research Fund, Division of Gastroenterology and Hepatology, Department of Medicine, University of Hong Kong. Mr. Daniel Yee-Tak Fong (statistician of the Department of Nursing, University of Hong Kong), one of the coauthors, and Dr. Man-Fung Yuen, the principal investigator, performed the statistical analysis for this study. Dr. Man-Fung Yuen had full access to all of the data in this study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Address reprint requests to: Ching-Lung Lai, Department of Medicine, University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China. hrmelcl@hkucc.hku.hk; fax: Copyright 2007 by the American Association for the Study of Liver Diseases. Published online in Wiley InterScience ( DOI /hep Potential conflict of interest: Nothing to report. primary target for the treatment of CHB. Recent evidence shows that a high hepatitis B virus (HBV) DNA level is associated with a higher risk of development of HCC and cirrhosis. 2,3 Effective viral suppression by the nucleoside analogue, lamivudine, reduces the risk of complications in patients with or without advanced disease. 4,5 However, a prolonged lamivudine treatment is associated with increasing incidence of drug-resistant mutations [tyrosine-methionine-aspartate-aspartate (YMDD) mutants], which reaches more than 70% after 5 years of lamivudine treatment. 6 Despite this, lamivudine is still commonly used for several reasons. It has the longest surveillance for safety. Twenty to twenty-five percent of patients do not develop YMDD mutations and continue to have satisfactory viral suppression even after a prolonged treatment. 5 Moreover, it is the only nucleoside/nucleotide analogue proven in randomized controlled studies to reduce disease progression. 4 Therefore, identifying patients who are likely to achieve an ideal response in terms of HBV DNA suppression, hepatitis B e antigen (HBeAg) seroconversion, and normalization of alanine aminotransferase (ALT) levels without the emergence of YMDD mutations with a longterm treatment of lamivudine is important. In a previous 1695

2 1696 YUEN ET AL. HEPATOLOGY, December 2007 study, patients who had undetectable HBV DNA levels at week 24 of treatment were associated with a significantly lower chance of subsequent development of YMDD mutations. 7 The aim of the present study was to identify the optimal time and cutoff HBV DNA level at even earlier times of treatment to predict a good response to 5 years of lamivudine. Patients and Methods Patients. The present study recruited patients who participated in our previous 3 trials [NUCB 3009, NUCB 3018 (a rollover trial for NUCB 3009), and NUCB 4003] with GlaxoSmithKline Research Laboratories between June 1, 1994 and August 31, ,9 The studies were approved by the institutional review board of the University of Hong Kong (Queen Mary Hospital, Hong Kong). A total of 180 patients receiving different doses and regimens of lamivudine or famciclovir under respective protocols were recruited in these 3 trials. All patients were positive for hepatitis B surface antigen (HBsAg) for at least 6 months and were also positive for HBeAg on entry to the trials. Of the 180 patients, 80 who received 100 mg of lamivudine daily without interruption were assessed for the present trial. The remaining 100 patients who did not receive 100 mg daily without interruption because of randomization in the original study protocols with different dosage regimens were excluded. Of the 80 patients continuously receiving 100 mg of lamivudine daily, 6 patients were excluded because of a limited follow-up time of less than 5 years in 2 patients and insufficient serum for analyses in 4 patients. The remaining 74 patients were recruited for the present study, which was separately approved by the institutional review board. Follow-Up Schedule. All patients were followed up every 2 weeks for the first 4 weeks, every 4 weeks thereafter until week 52, then every 8 weeks up to 104 weeks, and every 16 weeks until 5 years. Serum taken during every follow-up was stored at 20 C. HBV DNA Measurements. The serum at 13 times, including the baseline, weeks 2, 4, 8, 12, 16, 24, and 32, and years 1, 2, 3, 4, and 5, was thawed to measure the HBV DNA levels (962 samples for the HBV DNA measurements in all). These were measured with a branched DNA assay (Versant HBV DNA 3.0 assay, Bayer Health- Care Diagnostic Division, Tarrytown, NY), with a lower limit of detection of 2000 copies/ml (400 IU/mL). Samples in which the HBV DNA levels exceeded the upper limit of the linear range were diluted 100,000-fold for a second measurement. Detection of Reverse Transcriptase Gene Mutations, HBV Genotypes, and Core Promoter/Precore Mutations. The presence of YMDD mutations was determined at the baseline and at yearly intervals until year 5 with a Trugene HBV genotyping assay kit (Bayer Health- Care Diagnostic Division) according to the methodology described. 10 This assay was also capable of detecting other mutations that commonly occur in the reverse transcriptase gene reported in other nucleoside/nucleotide analogues. These included rt A181T/V and N236T for adefovir dipivoxil, rt T184G, S202I, and M205G for entecavir, and rt A194T for tenofovir disoproxil fumarate. In addition to the detection of these mutations over these times, the possibility of the earlier emergence of these mutations was examined by direct sequencing at the 2 best times, that is, weeks 4 and 16 (see the Results section), and at week 24. Briefly, the primers used for the amplification of the reverse transcriptase gene were HBV224s (5 -AGAATCCTCACAATACCGCAG-3 ) and HBV1020a (5 -AGCAAAACCCAAAAGACCCA- 3 ), the 796 base pair amplicon of which covered the reverse transcriptase codons The sequence analysis was performed with an ABI-Prism BigDye terminator kit and an ABI-Prism 3700 DNA analyzer (Applied Biosystems, Foster City, CA). HBV genotypes were determined at the baseline with the same Trugene HBV genotyping kit already described. The core promoter and precore mutations were detected by an Innogenetics line probe (INNO-LiPA HBV precore assay, Innogenetics, Ghent, Belgium). The methodology is described in our previous article. 11 Liver Biochemistry and HBV Serology. The liver biochemistry was determined during every follow-up. HBV serological markers, including HBsAg, the antibody to HBsAg, HBeAg, and the antibody to hepatitis B e antigen (anti-hbe), were determined every 3 months by a microparticle enzyme immunoassay (Abbott Laboratories, Chicago, IL). The HBeAg seroconversion was defined as the loss of HBeAg with the development of anti-hbe on at least 2 consecutive follow-ups and without subsequent HBeAg seroreversion up to 5 years of follow-up. Definition of the Treatment Response at Year 5. Patients with an ideal response are defined as those who had HBV DNA levels of less than 2000 copies/ml (400 IU/mL) for at least 2 consecutive years until the end of 5 years of follow-up with HBeAg seroconversion, with normalization of ALT levels, and without YMDD mutations. Patients with treatment failure are defined as those who had HBV DNA levels greater than 10 5 copies/ml (20,000 IU/mL) and who remained HBeAg-positive with abnormal ALT levels after 5 years of treatment. Patients

3 HEPATOLOGY, Vol. 46, No. 6, 2007 YUEN ET AL Table 1. Demographic Data for the Studied Population Total Population (n 74) Patients With an Ideal Response (n 17) Patients Without an Ideal Response (n 57) Sex (male:female) 58:16 14:3 44:13 Age (years) 32.2 ( ) 28.6 ( ) 32.5 ( ) Albumin (g/l) 47 (40 54) 47 (42 50) 47 (40 54) Bilirubin ( mol/l) 10 (3 27) 11 (7 21) 9 (3 27) ALT (U/L) 56 (14 506) 154 (16 419)* 47 (14 506)* HBV DNA (log copies/ml) 8.6 ( ) 8.0 ( ) 8.7 ( ) Genotype (B:C) 18:56 4:13 14:43 Core promoter mutation 29 (39.2%) 12 (70.6%) 17 (29.8%) Precore mutation 7 (9.5%) 4 (23.5%) 3 (5.3%) Continuous variables are expressed as medians (range). ALT indicates alanine aminotransferase; and HBV, hepatitis B virus. *The odds ratio for patients with elevated ALT levels was 19.1 (95% confidence interval , P 0.001). The odds ratio for patients with core promoter mutations was 5.6 (95% confidence interval , P 0.004). The odds ratio for patients with precore mutations was 5.5 (95% confidence interval , P 0.045). with laboratory and serologic parameters outside the ideal-response and treatment-failure criteria are regarded as having an intermediate response. Statistical Analysis. The demographics of patients with and without ideal responses were compared by the Mann-Whitney U test for continuous data and Fisher s exact test for dichotomous data. The predictions of ideal responses by HBV DNA taken at various time epochs were first examined by the construction of the corresponding receiver operating characteristic (ROC) curves, with the overall accuracy assessed by the areas under the curves (AUCs). Then, for each time epoch, an optimal cutoff HBV DNA level (log 10 copies/ml) for the prediction of the ideal response was obtained by the maximization of 2 indices over all possible HBV DNA levels: (1) the Youden index, which is defined as the sensitivity plus the specificity minus 1, and (2) the predictive value (PV) index, which is defined as the positive predictive value plus the negative predictive value minus 1. The maximum value attained was called the optimal value. The Youden index was first proposed by Youden 12 for the simultaneous maximization of sensitivity and specificity when accuracy is of higher concern. The PV index was constructed similarly on the basis of predictive values when utility is of higher concern. A 5% level of significance was used, and a 95% confidence interval accompanied all estimates when it was appropriate. Statistical Analysis System version 8, SPSS for Windows release 11.5, and Stata/SE 9.2 were used to perform the analysis. Results Demographics. The demographics for the whole population (n 74) and patients with an ideal response (n 17) and without an ideal response (n 57) are depicted in Table 1. There were no differences in the baseline parameters between the 2 groups of patients, except that the former had a significantly higher baseline median ALT level and a high proportion of patients with core promoter or precore mutations. HBV DNA Levels, HBeAg Seroconversion, and ALT Levels After a 5-Year Lamivudine Treatment. After a 5-year lamivudine treatment, there were 20 patients (27%) with HBV DNA levels of less than 2,000 copies/ml (400 IU/mL), 8 patients (10.8%) with HBV DNA levels between 2000 and 10 5 copies/ml (400-20,000 IU/mL), and 46 patients (62.2%) with HBV DNA levels greater than 10 5 copies/ml (20,000 IU/mL). Twenty-two patients (29.7%) had HBeAg seroconversion. The normalization of ALT levels occurred in 27 of 42 patients (64.3%) with abnormal ALT levels at the baseline. Lamivudine-Resistant HBV. Forty-six patients (62.2%) had YMDD mutations. Another 8 patients (10.8%) developed YMDD mutations transiently, none of whom had detectable YMDD mutations at year 5. The HBV DNA levels at the end of follow-up were less than 2000 copies/ml (400 IU/mL) in 5 patients, 2695 copies/ml (539 IU/mL) in 1 patient, 7619 copies/ml (1524 IU/mL) in 1 patient, and 28,030 copies/ml (5606 IU/ ml) in 1 patient. Five patients had HBeAg seroconversion. Seven patients had normal ALT levels (one had a very mild ALT elevation of 34 U/L, normal 31 U/L). Four of these 8 patients had ideal responses. The virologic profiles of 2 patients are depicted in Fig. 1. The other 2 common lamivudine-resistant mutations, including rt V173L and rt L180M, were present in 7 (9.5%) and 36 (48.6%) patients, respectively, at 5 years. Use of the HBV DNA Level for the Prediction of the Ideal Response. The ROC curves of different early times (baseline to year 1) of the measurement of HBV DNA levels to predict an ideal response at 5 years are depicted in

4 1698 YUEN ET AL. HEPATOLOGY, December 2007 Fig. 1. Virologic profiles of the 2 patients with transient emergence of YMDD mutations during lamivudine treatment. Fig. 2. In general, there was a stepwise improvement in the accuracy from the baseline through year 1. The exact values of AUC at each time from the baseline up to year 5 are listed in Table 2 and illustrated in Fig. 3. The prediction was already of high accuracy if the assessment was done as early as week 4 with an AUC of 0.89 and was most accurate for times in the first year at week 16 with an AUC of Compared to week 16, week 4 had the advantage of allowing for a very early assessment, with no statistical difference in AUC (P 0.169). The optimal cutoff HBV DNA levels were obtained by the adoption of the Youden and PV indices (Fig. 4). The Youden index provides a more optimal index if the emphasis is on sensitivity and specificity. The PV index, on the other hand, is a better model for positive and negative predictive values. For the treatment strategy of CHB patients receiving lamivudine, the prediction of achievement of an ideal response is more practical and useful. The optimal cutoff HBV DNA levels were 4.0 log copies/ml (10,000 copies/ml or 2000 IU/mL) for week 4 and 3.6 log copies/ml (4000 copies/ml or 800 IU/mL) for week 16, corresponding to the PV indices of 0.84 and 0.88, respectively. The corresponding sensitivity, specificity, positive likelihood ratios, negative likelihood ratios, positive predictive values, and negative predictive values are listed in Table 3. Patients who have HBV DNA levels of less than 4.0 log copies/ml (2000 IU/mL) at week 4 have a 100% chance of achieving an ideal response at year 5 if the treatment is continued. Conversely, patients who have HBV DNA levels greater than or equal to 4.0 log copies/ml (2000 IU/mL) at week 4 have an 83.8% chance of failure to achieve an ideal response. This pre- Fig. 2. ROC using the HBV DNA level (diagram on the left) and HBV DNA reduction from the baseline (diagram on the right) at different times during the first year of lamivudine treatment to predict the ideal response at 5 years.

5 HEPATOLOGY, Vol. 46, No. 6, 2007 YUEN ET AL Table 2. Area Under the ROC Curve at Each Time by the Use of the HBV DNA Levels (log 10 Copies/mL) or HBV DNA Reduction from the Baseline To Predict a Good Response at 5 Years HBV DNA Level HBV DNA Reduction from the Baseline Time AUC SE P* 95% CI AUC SE P* 95% CI Week , 0.77 Week , , 0.82 Week , , 0.90 Week , , 0.90 Week , , 0.89 Week , , 0.91 Week , , 0.88 Week , , 0.92 Year , , 0.90 Year , , 0.95 Year , , 0.96 Year , , 0.98 Year , , 0.97 AUC indicates area under the curve; CI, confidence interval; HBV, hepatitis B virus; ROC, receiver operating characteristic; and SE, standard error. *Statistical significance implies that the corresponding AUC was significantly different from 0.5. diction for not achieving an ideal response increases to 87.7% if the HBV DNA level is greater than or equal to 3.6 log copies/ml (800 IU/mL) at week 16. Mutations at the Reverse Transcriptase Gene at Weeks 4, 16, and 24. In the samples of all 74 patients, lamivudine-resistant (rt M204I/V), adefovir-resistant (rt A181T/V and N236T), entecavir-resistant (rt T184G, S202I, and M205G), and tenofovir-resistant (rt A194T) viruses were not detectable at the baseline, week 4, or week 16. However, 3 patients had the rt M204I lamivudine-resistant virus, and 2 patients had the rt A181T adefovir-resistant virus detected at week 24. Use of HBV DNA Reduction from the Baseline for the Prediction of the Ideal Response. The ROC curves of different early times, with HBV DNA reduction from the baseline used to predict the ideal response, are depicted in Fig. 2. Similarly to curves obtained with absolute HBV DNA levels, there was a stepwise improvement in accuracy from week 2 through year 1. However, the accuracy was inferior to that of absolute HBV DNA levels (Fig. 2). The AUC at each time with HBV DNA reduction was lower than that with the absolute HBV DNA levels at each respective time from week 2 to year 1 (Table 2 and Fig. 3). The Youden and PV indices were 0.51 and 0.78, respectively, at week 4 (compared to 0.67 and 0.84, respectively, with absolute HBV DNA levels at week 4; see Fig. 4). Patients with Treatment Failure at Year 5. There were 23 patients with treatment failure; that is, they had an HBV DNA level greater than 10 5 copies/ml (20,000 IU/mL), were positive for HBeAg, and had abnormal ALT levels at year 5. With the absolute HBV DNA levels at an early treatment period used for the prediction of the chance of treatment failure at 5 years, the accuracy was Fig. 3. AUC at different times during lamivudine treatment to predict the ideal response at 5 years. Fig. 4. Optimal value to define the best cutoff HBV DNA level at different times during lamivudine treatment to predict the ideal response at 5 years.

6 1700 YUEN ET AL. HEPATOLOGY, December 2007 Table 3. Prediction of a Good Response at 5 Years by HBV DNA Levels (log 10 Copies/mL) Measured At Weeks 4 and 16 Week 4 Week 16 Youden Index PV Index Youden Index PV Index Optimal cutoff HBV DNA level (log copies/ml) Positive predictive value 55.6% (35.3%,74.5%) 100% (54.1%, 100%) 77.8% (52.4%,93.6%) 100% (66.4%, 100%) Negative predictive value 95.7% (85.5%,99.5%) 83.8% (72.9%,91.6%) 94.6% (85.1%,98.9%) 87.7% (77.2%,94.5%) Sensitivity 88.2% (63.6%,98.5%) 35.3% (14.2%,61.7%) 82.4% (56.6%,96.2%) 52.9% (27.8%,77.0%) Specificity 78.9% (66.1%,88.6%) 100% (93.7%, 100%) 93.0% (83.0%,98.1%) 100% (93.7%, 100%) Positive likelihood ratio 4.19 (2.46, 7.13) 41.9 (2.48, 708) 11.7 (4.45, 31) 61.2 (3.74, 1001) Negative likelihood ratio 0.15 (0.04, 0.55) 0.64 (0.46, 0.91) 0.19 (0.07, 0.53) 0.48, (0.29, 0.78) The values in parentheses represent the 95% confidence intervals. Table 4. Detailed Parameters at Year 5 for the 34 Patients with an Intermediate Response Number of Patients (n 34) HBV DNA levels 2000 copies/ml (400 IU/mL) 3 (8.8%) copies/ml (400-20,000 IU/mL) 8 (23.5%) 10 5 copies/ml (20,000 IU/mL) 23 (67.6%) HBeAg seroconversion 5 (14.7%) Normal ALT level 30 (88.2%) YMDD mutations 23 (67.6%) ALT indicates alanine aminotransferase; HBeAg, hepatitis B e antigen; and HBV, hepatitis B virus. suboptimal, with the best AUC being only 0.67 at week 24 (0.61 and 0.66 at weeks 4 and 16). The Youden and PV indices at week 24 were 0.31 and 0.7, respectively. This indicates that HBV DNA levels at an early treatment period are not good predictors of treatment failure at year 5. Similar to the findings for the use of HBV DNA reduction from the baseline for an ideal response, the predictive values and accuracy for predicting treatment failure were inferior to those from absolute HBV DNA levels (data not shown). Patients with an Intermediate Response at Year 5. Thirty-four patients had laboratory and virologic parameters outside the defined criteria of an ideal response or treatment failure. The detailed parameters at year 5 for these patients are listed in Table 4. As in the patients with treatment failure, the AUC for predicting patients with an intermediate response were suboptimal. For HBV DNA levels, the AUCs for predicting an intermediate response at year 5 were 0.48 at week 4, 0.53 at week 16, and 0.56 at week 24. The AUCs for HBV DNA reduction were even lower than those from absolute HBV DNA levels at these 3 times (data not shown). The HBV DNA levels during the early treatment period were not good predictors of an intermediate response at year 5. Predictive Values of Undetectable HBV DNA Levels at Years 1 and 2. In our previous study, achieving undetectable HBV DNA levels at week 24 was of predictive value for the development of YMDD mutations on longer follow-up. 7 To determine whether future viral breakthrough is likely to occur for patients requiring a longer time to achieve undetectable HBV DNA levels, the chances of YMDD mutations and HBV DNA breakthroughs (defined as the reappearance of HBV DNA) at year 5 were studied with respect to patients with undetectable HBV DNA levels at years 1 and 2. There were 17 patients with undetectable HBV DNA levels at year 1; 8 more patients with detectable HBV DNA at year 1 had undetectable HBV DNA levels at year 2. YMDD mutations at year 5 were detected in 6 (35.3%) and 4 (50%), respectively. HBV DNA breakthrough at year 5 occurred in 5 (29.4%) and 3 (37.5%), respectively. The number of patients in both groups was too small for meaningful statistical analysis, although the proportion of patients who had undetectable HBV DNA during the second year of therapy appeared to be higher than the proportion of patients who had undetectable HBV DNA by year 1 for the development of YMDD mutations and HBV DNA breakthroughs at year 5. Discussion To date, there have been several studies addressing the significance of early viral suppression by nucleotide/nucleoside analogues. We first demonstrated that lamivudine-treated patients with HBV DNA levels of less than 10 3 copies/ml (200 IU/mL) at week 24 were associated with a 13% chance of development of YMDD mutations at 29 months versus 63% for patients with HBV DNA levels greater than 10 3 copies/ml (200 IU/mL; P 0.001). 7 Subsequently, Locarnini and his colleagues 13 also showed that HBV DNA levels at year 1 were predictive of the chance of adefovir resistance at 3 years [4%, 26%, and 67% for patients with HBV DNA levels of less than 10 3 copies/ml (200 IU/mL), of copies/ml( ,000 IU/mL), and of more than 10 6 copies/ml (200,000 IU/mL), respectively]. For the significance for viral suppression, we have shown in a phase II telbivudine

7 HEPATOLOGY, Vol. 46, No. 6, 2007 YUEN ET AL study that 100% of patients with undetectable HBV DNA by a polymerase chain reaction (PCR) at week 24 will remain PCR-negative at week Sixty-two percent of patients with HBV DNA levels of less than 10 3 copies/ml (200 IU/mL) at week 24 will have undetectable HBV DNA by PCR at week 52 versus 24% and 7% of patients with HBV DNA levels of copies/ml ( IU/mL) and with HBV DNA levels greater than 10 4 copies/ml (2000 IU/mL) at week 24, respectively. Similar findings have been observed in a phase III telbivudine study showing that HBV DNA levels of less than 300 copies/ml (60 IU/mL) or less than 10 3 copies/ml (200 IU/mL) at week 24 are associated with a significantly lower chance of virologic breakthrough and a higher chance of loss of HBeAg, normalization of the ALT level, and undetectable HBV DNA by a PCR assay at week All these studies demonstrate that early and maximal viral suppression is the key to success for the treatment of CHB by antiviral agents. However, there are certain aspects concerning these studies that require further investigation. First, measurements of the HBV DNA levels are performed only once or very few times during an early treatment period in all of these studies. The best early time for the prediction of a good outcome has yet to be defined. Second, the predictions are calculated with reference to a relatively short treatment period of 1-3 years. In the context of long-term viral suppression to achieve a reduction of the development of complications from cirrhosis or HCC, factors that are of predictive value for more long-term treatment are more important. Third, the HBV DNA levels are used mostly to determine a single outcome, either the emergence of drug-resistant mutations or the eventual degree of HBV DNA suppression. The present study is designed to answer these unsettled issues. We measured the HBV DNA levels at frequent times during the first year of therapy, starting at week 2 after the initiation of lamivudine. A comparison of the accuracy of prediction with respect to the best time and best cutoff HBV DNA level was carried out at multiple early times to reach the best predictive value. In addition, a composite endpoint for an ideal response with HBV DNA levels of less than 2000 copies/ml (400 IU/mL), HBeAg seroconversion, ALT normalization, and an absence of YMDD mutations after 5 years of treatment was adopted. Although the target HBV DNA level according to various treatment guidelines for HBeAg-positive patients is less than 10 5 copies/ml (20,000 IU/mL), there is good evidence showing that the risk of development of HCC is significantly less only for patients with HBV DNA levels of less than 10 4 copies/ml (2000 IU/mL). Our very strict composite endpoint for the ideal response was aimed at identifying the subgroup of patients who will have a very good outcome with lamivudine in the context of the availability of newer and more potent nucleoside/nucleotide analogues. The present study, by exploring a model that identifies the best early time and HBV DNA level for predicting an ideal treatment response, can also be applied to therapy by other nucleoside/nucleotide analogues in future studies. According to the present study, the absolute HBV DNA levels after treatment were of better predictive value for an ideal response than the reduction of HBV DNA levels from the baseline. This suggests that the response to lamivudine treatment is not totally dependent on the baseline HBV DNA levels. Patients with a very high viral load at the baseline may still have an excellent long-term response to lamivudine treatment. One of the possible explanations is the intrinsically different responsiveness of the virus to specific antiviral agents. This may be related to the different naturally occurring primary sequences in the reverse transcriptase domain of the HBV polymerase gene in which the nucleotide/nucleoside analogues act. It has been shown that patients with the polymorphic sequence of rt256c have a poor response to lamivudine. 16,17 Future studies should concentrate on correlating the different sequences in the whole reverse transcriptase domain with the treatment response to different nucleoside/nucleotide analogues. This study provides a very practical guideline for clinicians and patients for the consideration of the long-term use of lamivudine. If patients achieve HBV DNA levels of less than 4 log copies/ml (2000 IU/mL) at week 4, lamivudine can be continued for its long-term benefit 4,5 without the risk of the development of resistance. On the contrary, the continuation of lamivudine beyond 4 weeks in patients who still have HBV DNA levels greater than or equal to 4 log copies/ml (2000 IU/mL) is not advisable. These patients have an 83.8% chance of failure to achieve an ideal response at 5 years; that is, 83.8% of these patients would have treatment failure or an intermediate response. The early termination of lamivudine should be considered in order to minimize the chance of developing YMDD mutations in these patients because the response to either adefovir dipivoxil or entecavir is reduced in patients with lamivudine-resistant HBV Alternatively, to decide whether lamivudine should be discontinued, one can allow a later second HBV DNA measurement at week 16. For patients who still have HBV DNA levels greater than or equal to 3.6 log copies/ml (800 IU/mL), there is an 87.7% chance of failure to achieve an ideal response at year 5. Delaying this decision from week 4 to week 16 (with the gain of 3.9% in the negative predictive value) does not seem to increase the

8 1702 YUEN ET AL. HEPATOLOGY, December 2007 chance of getting a more drug-resistant virus as they all were not detectable both times. On the other hand, extending the decision-making to week 24 is not advisable. Although the AUC of week 24 was similar to that of week 4 (0.9 versus 0.89, respectively), it was inferior to that of week 16 (AUC 0.94; Table 2). In addition, 3 patients had already developed the lamivudine-resistant viruses and 2 patients had developed adefovir-resistant viruses (rt A181T) at week 24. According to our previous study, 5% of patients will develop lamivudine-resistant HBV after 24 weeks of treatment. 17 Though rt A181T is located in domain B of the reverse transcriptase gene of HBV, it is a cross-resistant mutant virus that can be selected by either lamivudine or adefovir treatment. It has been shown in woodchuck HBV and human HBV studies that lamivudine per se can also select for this mutation. 22,23 The use of HBV DNA measurement at early times, such as weeks 4, 16, and 24, is not of good predictive value for either treatment failure [HBV DNA 10 5 copies/ml (20,000 IU/mL), HBeAg positivity, and abnormal ALT levels] or an intermediate response at year 5. There are several other important findings generated from the present study. First, high pretreatment ALT is associated with an ideal response in the short-term 24 and up to 5 years of treatment. Because high ALT levels indicate a more intense immunologic response to HBV through cytolytic T cells on infected hepatocytes, these patients could be expected to have a better response to lamivudine. Second, we have confirmed in the present study with a larger number of patients that core promoter and precore mutations are associated with a higher chance of achieving an ideal response to lamivudine (Table 1), a finding also observed in other studies. 25,26 Although the exact reasons for these phenomena have not been fully elucidated, both core promoter and precore mutant viruses may be more responsive to lamivudine treatment. This is indirectly supported by the finding from other studies that core promoter and precore mutants revert to the wild type with lamivudine treatment. 17,27 Third, some patients with YMDD mutations still had a good outcome because the YMDD mutations might have been detected only transiently during the course of treatment, and this suggests that certain YMDD mutations may remain susceptible to lamivudine. Nearly all the patients with transient YMDD mutations had low HBV DNA levels and normal ALT levels (5 out of 8 had HBeAg seroconversion) at the end of follow-up. In conclusion, the measurement of the HBV DNA level at week 4 of lamivudine treatment should be performed in all patients to predict those who will achieve a long-term ideal response. For patients with HBV DNA levels of less than 4 log copies/ml (2000 IU/mL) at week 4, all will have HBV DNA levels of less than 2000 copies/ml (400 IU/mL), HBeAg seroconversion, normal ALT levels, and no YMDD mutations at year 5. The addition of or switch to an alternative antiviral should be considered in patients with HBV DNA levels greater than or equal to 4 log copies/ml (2000 IU/mL) at week 4 because 83.8% of these patients would not achieve a longterm ideal response. Acknowledgment: We thank Dr. David A. Hendricks and Bayer HealthCare (Berkeley, CA) for the arrangement and provision of assays for the measurement of the HBV DNA levels and for the determination of the genotypes and mutations of the reverse transcriptase gene of HBV for this study. References 1. Lai CL, Ratziu V, Yuen MF, Poynard T. Viral hepatitis B. Lancet 2003; 362: Chen CJ, Yang HI, Su J, Jen CL, You SL, Lu SN, et al. Risk of hepatocellular carcinoma across a biological gradient of serum hepatitis B virus DNA level. JAMA 2006;295: Iloeje UH, Yang HI, Su J, Jen CL, You SL, Chen CJ, et al. Predicting cirrhosis risk based on the level of circulating hepatitis B viral load. Gastroenterology 2006;130: Liaw YF, Sung JJ, Chow WC, Farrell G, Lee CZ, Yuen H, et al. Lamivudine for patients with chronic hepatitis B and advanced liver disease. N Engl J Med 2004;351: Yuen MF, Seto WK, Chow DHF, Tsui K, Wong DKH, Chan P, et al. Long-term beneficial outcome of Chinese asymptomatic patients with HBeAg-positive chronic hepatitis B on continuous lamivudine therapy: 7-year experience. [abstract] HEPATOLOGY 2005;42(4 suppl 1):583A. 6. Lai CL, Dienstag J, Schiff E, Leung NW, Atkins M, Hunt C, et al. Prevalence and clinical correlates of YMDD variants during lamivudine therapy for patients with chronic hepatitis B. Clin Infect Dis 2003;36: Yuen MF, Sablon E, Hui CK, Yuan HJ, Decraemer H, Lai CL. Factors associated with hepatitis B virus DNA breakthrough in patients receiving prolonged lamivudine therapy. HEPATOLOGY 2001;34(4 pt 1): Lai CL, Chien RN, Leung NW, Chang TT, Guan R, Tai DI, et al., for the Asia Hepatitis Lamivudine Study Group. A one-year trial of lamivudine for chronic hepatitis B. N Engl J Med 1998;339: Lai CL, Yuen MF, Hui CK, Garrido-Lestache S, Cheng CT, Lai YP. Comparison of the efficacy of lamivudine and famciclovir in Asian patients with chronic hepatitis B: results of 24 weeks of therapy. J Med Virol 2002;67: Gintowt AA, Germer JJ, Mitchell PS, Yao JD. Evaluation of the MagNA Pure LC used with the TRUGENE HBV genotyping kit. J Clin Virol 2005;34: Yuen MF, Sablon E, Yuan HJ, Hui CK, Wong DK, Doutreloigne J, et al. Relationship between the development of precore and core promoter mutations and hepatitis B e antigen seroconversion in patients with chronic hepatitis B virus. J Infect Dis 2002;186: Youden WJ. Index for rating diagnostic test. Cancer 1950;3: Locarnini S, Qi X, Arterburn S, Snow A, Brosgart CL, Currie G, et al. Incidence and predictors of emergence of adefovir resistant HBV during four years of adefovir dipivoxil (ADV) therapy for patients with chronic hepatitis B (CHB). J Hepatol 2005;42(suppl 2): Lai CL, Leung N, Teo EK, Tong M, Wong F, Hann HW, et al. A 1-year trial of telbivudine, lamivudine, and the combination in patients with

9 HEPATOLOGY, Vol. 46, No. 6, 2007 YUEN ET AL hepatitis B e antigen-positive chronic hepatitis B. Gastroenterology 2005; 129: Lai CL, Gane E, Liaw YF, Thongsawat S, Wang Y, Chen Y, et al. Maximal early HBV suppression is predictive of optimal virologic and clinical efficacy in nucleoside-treated hepatitis B patients: scientific observations from a large multinational trial (the Globe Study). [abstract] HEPATOLOGY 2005;42(suppl 1):232A. 16. Ciancio A, Smedile A, Rizzetto M, Lagget M, Gerin J, Korba B. Identification of HBV DNA sequences that are predictive of response to lamivudine therapy. HEPATOLOGY 2004;39: Yuen MF, Sablon E, Libbrecht E, De Velde HV, Wong DKH, Fung J, et al. Significance of HBV viral load, core promoter/precore mutations and specific sequences of polymerase gene in HBV-infected patients on 3-year lamivudine treatment. Antivir Ther 2006;11: Fung J, Yuen JCH, Wong DKH, Tanaka Y, Mizokami M, Lai CL, et al. Adefovir dipivoxil monotherapy and combination therapy with lamivudine in the treatment of chronic hepatitis B in an Asian population. Antivir Ther 2007;12: Lee YS, Suh DJ, Lim YS, Jung SW, Kim KM, Lee HC, et al. Increased risk of adefovir resistance in patients with lamivudine-resistant chronic hepatitis B after 48 weeks of adefovir dipivoxil monotherapy. HEPATOLOGY 2006;43: Sherman M, Yurdaydin C, Sollano J, Silva M, Liaw YF, Cianciara J, et al. Entecavir for treatment of lamivudine-refractory, HBeAg-positive chronic hepatitis B. Gastroenterology 2006;130: Lai CL, Shouval D, Lok AS, Chang TT, Cheinquer H, Goodman Z, et al. Entecavir versus lamivudine for patients with HBeAg-negative chronic hepatitis B. N Engl J Med 2006;354: Tatti KM, Korba BE, Stang HL, Peek S, Gerin JL, Tennant BC, et al. Mutations in the conserved woodchuck hepatitis virus polymerase FLLA and YMDD regions conferring resistance to lamivudine. Antiviral Res 2002;55: Yeh CT, Chien RN, Chu CM, Liaw YF. Clearance of the original hepatitis B virus YMDD-motif mutants with emergence of distinct lamivudineresistant mutants during prolonged lamivudine therapy. HEPATOLOGY 2000;31: Chien RN, Liaw YF, Atkins M, for the Asian Hepatitis Lamivudine Trial Group. Pretherapy alanine transaminase level as a determinant or hepatitis B e antigen seroconversion during lamivudine therapy in patients with chronic hepatitis B. HEPATOLOGY 1999;30: Asahina Y, Izumi N, Uchihara M, Noguchi O, Nishimura Y, Inoue K, et al. Core promoter/pre-core mutations are associated with lamivudine-induced HBeAg loss in chronic hepatitis B with genotype C. J Hepatol 2003;39: Shin JW, Chung YH, Choi MH, Kim JA, Ryu SH, Jang MK, et al. Precore stop codon mutation of hepatitis B virus is associated with low breakthrough rate following long-term lamivudine therapy. J Gastroenterol Hepatol 2005;20: Cho SW, Hahm B, Kim JH. Reversion from precore/core promoter mutants to wild-type hepatitis B virus during the course of lamivudine therapy. HEPATOLOGY 2000;32: