Serum HBV DNA as a marker of efficacy during therapy for chronic HBV infection: Analysis and review of the literature

VIRAL HEPATITIS

Serum HBV DNA as a Marker of Efficacy During Therapy for Chronic HBV Infection: Analysis and Review of the Literature Herve Mommeja-Marin, Elsa Mondou, M. Robert Blum, and Franck Rousseau Currently, compounds under evaluation for treatment of chronic hepatitis B virus (HBV) infection are evaluated with liver histology as the primary end point for efficacy. However, because of practical limitations in serial liver biopsies, there is a need for alternate markers to assess efficacy over shorter periods of time. Considering the direct correlation between viral replication and disease progression during human immunodeficiency virus and hepatitis C virus infection, we explored whether such a correlation exists for HBV infection. We reviewed the literature and conducted an analysis to investigate the relationship between absolute or treatment-induced changes in HBV DNA levels and other accepted markers of disease activity. A total of 26 prospective studies met our selection criteria, including 33 evaluable treatment arms. The study treatments consisted of nucleosides and/or interferon regimens and control arms. We found statistically significant and consistent correlations between viral load level or change and histologic grading and biochemical and serologic response. Our analysis suggests that a treatment-induced reduction in HBV DNA level can be used for assessing efficacy of treatment regimens. Further, we observed that quantitative HBV DNA has a broader dynamic range than histology, allowing demonstration of differences between 2 active treatments of unequal potency. The analysis showed stronger results in studies using nucleoside regimens and in hepatitis B e antigen (HBeAg)-positive patients. In conclusion, the goal of anti-HBV therapy should be profound and durable viral suppression, as defined by very sensitive assays. Additional prospective studies are needed to precisely determine the desirable level of viremia to attain. (HEPATOLOGY 2003;37:1309-1319.)

C

hronic hepatitis B virus (HBV) infection often leads to severe complications and death after decades1; therefore, the evaluation of therapeutic regimens is based on surrogate markers such as histology and hepatitis B e antigen (HBeAg) seroconversion.2 Although the treatment of chronic HBV infection has advanced in the past decade through development of immune modulators and nucleoside/nucleotide analogues, there is a significant need for improving therapeutic responses. Currently, compounds under development are evaluated with liver histology as the primary end point for efficacy. Hepatic histology is invaluable for assessment of Abbreviations: HBV, hepatitis B virus; HBeAg, hepatitis B e antigen; ALT, alanine aminotransferase; HBsAg, hepatitis B surface antigen. From Triangle Pharmaceuticals, Inc., Durham, NC. Received June 17, 2002; accepted March 3, 2003. Address reprint requests to: Herve Mommeja-Marin, M.D., Triangle Pharmaceuticals, Inc., 4611 University Dr., P.O. Box 50530, Durham, NC 27717. E-mail: [email protected]; fax: 919-402-5528. Copyright © 2003 by the American Association for the Study of Liver Diseases. 0270-9139/03/3706-0015$30.00/0 doi:10.1053/jhep.2003.50208

the etiology and staging of liver diseases.3 However, serial liver biopsies are not part of the usual management of chronically HBV-infected patients because they are invasive, are associated with a certain level of complications, and may not be performed more frequently than yearly.4 Thus, there is a need for alternate markers that are available in clinical practice and suitably reliable to assess the efficacy of therapeutic interventions over shorter periods of time. Considering the established direct correlation between viral replication and disease progression during human immunodeficiency virus and hepatitis C virus infection, we were interested in exploring whether such a correlation exists for chronic HBV infection. Both cirrhosis and cancer, which are sequelae of chronic HBV infection in humans,5,6 are related to persistent and uncontrolled replication of the virus in the liver. Improvement and/or prevention of these conditions has been associated with the control and prevention of HBV replication.1,7,8 We hypothesized that serum HBV DNA level is a direct reflection of the level of intrahepatic viral replication. 1309

1310

MOMMEJA-MARIN ET AL.

HEPATOLOGY, June 2003

Materials and Methods

In assessing the correlation between viral load and histologic or biochemical response, we used the same point in time (i.e., baseline or end of therapy). In assessing the correlation between aminotransferase normality and serum HBV DNA absolute level, both baseline and end-oftreatment results were used (inclusive of placebo/control arm). In assessing the correlation between absolute level or change in viral load from baseline and serologic response, we used the viral load level achieved at the end of treatment and the incidence of HBeAg seroconversion, defined as the number of hepatitis B e antigen seroconversions per 100 patient-months of observation in the study. Statistical Analysis. Categorical variables were compared using a 2-sided ␹2 test. In all figures, each study arm was plotted as one point; the size of each point is proportional to the population size of the arm. Pearson correlation coefficients were calculated and simple linear regression with estimation via ordinary least squares was used to examine bivariate associations. A t test was used to calculate the P values. For nonlinear regressions (exponential), the Pearson correlation coefficient was calculated after logarithmic transformation of the variables. For all bivariate associations, Pearson correlation coefficients were compared between exponential and linear regression; the regression leading to the highest coefficient was used in the analysis. No weighting was used with respect to the number of patients in each study arm, because the use of the median in most of the reports reviewed prevented such a weighting. The change in median for histology or HBV DNA level is the difference between the observed median at the commencement of treatment and the median at the end of treatment unless the median change from baseline was specified in the original report.

Objectives of the Analysis. Our objective was to investigate the relationship between viral load level or suppression and treatment response using a summative analysis of the literature. To first assess the variability of viral load level to determine the minimal true change in viral load. We subsequently investigated the relationship between HBV DNA absolute level or treatment-induced changes in serum HBV DNA level to other markers of disease activity. Selection of the Studies. We conducted a systematic search on MEDLINE using “hepatitis B,” “HBV,” and “HBV DNA” as keywords and covering the period from 1996 to 2002. Our preliminary criteria for selecting reports were that they were published in English, specified HBV DNA levels and other end points (alanine aminotransferase [ALT], serology, and histology), and reported medians, means, or sufficient data for estimation of the medians. A total of 279 reports met these criteria. We then further restricted the selection to prospective studies involving more than 25 chronically infected, nonimmunodepressed patients and reporting viral load values at the commencement and/or the end of treatment (including placebo, if applicable). Studies performed after liver transplantation or in patients coinfected with human immunodeficiency virus or hepatitis C virus were excluded. Studies in children and patients with cirrhosis were included in the analysis. Abstraction of the Data. Two separate reviewers abstracted the data from the report in a blinded fashion. To analyze the data consistently, we applied several transformations of the data when needed. Viral loads presented in the literature in picograms per milliliter were converted into copies per milliliter using the following conversion rate: 1 pg/mL ⫽ 3 ⫻ 105 copies/mL.3 When the median viral load level was not stated in the original report, we used the mean. When neither the median nor the mean of the viral load level was available, we estimated the median to be the lower limit of detection of the assay when 50% (or more) of the patients had undetectable HBV DNA levels. Because various assays were used to detect HBV DNA levels in the studies, we summarized performance characteristics of the assays used. Consistent with recent guidelines,3 histologic response to antiviral therapy was defined as improvement of at least 2 points in the Knodell histologic activity index (without fibrosis). Biochemical response was defined as a normal ALT level (within the limit of the normal range, independently of the baseline level) and serologic response as HBeAg seroconversion (disappearance of HBeAg and presence of hepatitis B e antibody).

Results Twenty-six unique studies met the final selection criteria.2,7,9-33 A brief description of the trials included in this analysis is presented in Table 1. These 26 studies involved 3,428 patients (in 33 evaluable treatment arms), of whom 2,524 were HBeAg positive at baseline. The partition of these studies in the various analyses is presented in Table 2. Overall, the study population involved mostly men (77%), the mean age of the population was 35.8 years, 26.4% were of Asian descent (most were white), 33.8% had normal ALT levels at baseline, and 16.8% had evidence of cirrhosis. Assessment of Viral Load. Previous reports suggested a low reliability associated with HBV DNA assays34; however, recent studies have shown an improvement in the accuracy and precision of these tests.9 Tables 3 and 4 summarize the precision of the assays used to assess HBV

HEPATOLOGY, Vol. 37, No. 6, 2003

MOMMEJA-MARIN ET AL.

1311

Table 1. Prospective Studies Included in the Analysis N/Population

Study Design

417/HBeAg positive

Controlled, double-blinded

404/HBeAg negative

Cohort study

164/HBeAg negative, 59 with cirrhosis 118/HBeAg positive

Cohort study

42/HBeAg negative

Randomized

103/HBeAg positive

Follow-up of randomized studies Open-label

Medication Regimen

Reference

11

Liquid hybridization Digene

15

Liquid hybridization

20

3y

Dot-blot

21

7y

bDNA

3TC 100 mg QD

3 y, 2 mo

Double-blinded, randomized Open-label

3TC 100 mg QD vs. 25 mg QD vs. placebo (n ⫽ 68) 3TC 100 mg QD

1y

3TC 100 mg QD vs. placebo (n ⫽ 2)

16 mo

160/36 HBeAg positive 29/HBeAg negative

Controlled, randomized, blinded Cohort Open-label

bDNA and PCR Liquid hybridization Liquid hybridization Liquid hybridization

85/HBeAg negative 41/HBeAg positive

Cohort Randomized

None IFN 3 MU TIW, control (n ⫽ 20)

3.2 y 16 mo

151/HBeAg positive

Randomized

20 mo

226/221 HBeAg positive

Randomized

103/HBeAg positive

Prospective analysis

46/Cirrhosis, 27 HBeAg positive

Cohort, historic controls

3TC 100 mg QD or 3TC plus IFN 9 MU TIW 3TC 100 mg for 1 y vs. IFN 10 MU TIW for 16 weeks vs. 3TC 100 mg for 8 weeks followed by 16 weeks with IFN 10 MU TIW Variable, 103 treated patients and 53 controls 3TC 150 mg QD

45/31 HBeAg positive 159/HBeAg positive

Open-label Controlled studies then open-label Open-label

IFN 6 MU TIW 3TC 25-100 mg QD

3y Median of 29.6 mo 9 mo

27/17 HBeAg positive 358/HBeAg positive 58/HBeAg positive 137/HBeAg positive

98/HBeAg positive 176/HBeAg positive 42/38 HBeAg positive 57/HBeAg positive, pretreated with IFN 101/HBeAg positive 28/HBeAg positive

Cohort Randomized, placebo-controlled Randomized Randomized, oral blinded Open-label

None IFN 6 MU TIW and 3TC 100 mg QD

IFN 10 MU daily for 1 week, then 10 MU TIW Miscellaneous Entecavir, dose escalating IFN 9 MU TIW vs. no treatment (n ⫽ 30) Placebo (n ⫽ 34) vs. placebo and IFN vs. prednisone and IFN 3TC 100 mg QD

18 mo

Assay

bDNA

Randomized

Famciclovir 500 mg TID vs. famciclovir 1.5 g QD vs. placebo (n ⫽ 36) 209 patients treated with IFN, 152 untreated controls 61 control patients, 103 patients received IFN Comparing retreatment with IFN 10 MU TIW and no additive treatment (n ⫽ 61) IFN 6 MU TIW vs. no treatment (n ⫽ 21) IFN at variable doses

Duration of Observation

6 y, 9 mo 6y 12 mo

3y

None 2y

1y

4 y, 2 mo 13 mo

PCR Liquid hybridization bDNA, PCR Liquid hybridization Liquid hybridization Liquid hybridization

PCR

16

2

22 23 24 25

26 27 28 30 32 33

7

bDNA or hybrid capture PCR bDNA

31

PCR

12

bDNA PCR

13 14 17

7 y, 4 mo

Liquid hybridization Slot-blot

2y

PCR

29

Variable 7 mo 12 mo

9 10

19

Abbreviations: TID, 3 times daily; QD, every day; bDNA, branched DNA; IFN, interferon; TIW, 3 times a week; 3TC, lamivudine; PCR, polymerase chain reaction.

DNA levels in the studies included in this analysis. Although there is no strict correspondence between the results of the different assays, several investigators recently compared different assays and reported sufficient intra-

assay precision to allow the follow-up of a given patient over time with the same assay, thereby facilitating the assessment of change from baseline across different assays.9,37 The assays provide reproducible results that are

1312

MOMMEJA-MARIN ET AL.

HEPATOLOGY, June 2003

Table 2. Summary of Studies Used for the Analysis by Parameter Compared With HBV DNA Parameters for Correlation

No. of Patients

No. of Studies

No. of Study Arms

Treatment Arms

References

Absolute HBV DNA and histology Absolute HBV DNA and histology

1,709 (1,046 HBeAg⫹) 1,452 (1,276 HBeAg⫹)

7

14*

All untreated

2,11,17,22,23,28,32

8

17

11,17,21-23,25,26,32

HBV DNA change from baseline and histology Absolute HBV DNA and proportion with normal ALT

1,090 (1,080 HBeAg⫹)

5

11

2,633 (2,033 HBeAg⫹)

16

28

HBV DNA change from baseline and seroconversion to anti-HBe Absolute HBV DNA and seroconversion to anti-HBe

1,743 (all HBeAg⫹)

10

20

931 (all HBeAg⫹)

8

13

Absolute HBV DNA and seroconversion to anti-HBs

427 (192 HBeAg⫹)

6

9

7 3 1 6 7 1 3 9 7 3 7 2 9 4 2 5 4 5 1 3 1 4 4

Nuc IFN IFN ⫹ nuc PBO Nuc IFN ⫹ nuc PBO Nuc IFN IFN ⫹ nuc PBO None/miscellaneous Nuc IFN IFN ⫹ nuc PBO Nuc IFN IFN ⫹ nuc PBO Nuc IFN PBO

11,22,23,25,32

9,11,13-15,17,20-24,27,28,32,33

10-12,17,20,23,25,31-33

7,17,18,23,30-32

7,9,21,22,28,30

Abbreviations: HBeAg⫹, HBeAg positive at baseline; IFN, interferon; PBO, placebo; nuc, nucleoside analogue; anti-HBe, antibody to hepatitis B e antigen; anti-HBs, antibody to hepatitis B surface antigen. *Includes baseline data.

internally consistent and broadly correlated across methodologies. Of note, the lower limit of detection of the liquid hybridization assay has been underestimated initially; therefore, we have used in the analyses the level stated in the report and conducted a secondary analysis after correction of this limit of detection to 106 copies/ mL.3 Recently, the first international standard (“97/746”) for HBV DNA nucleic acid amplification techniques was made available by the World Health Organization, which will likely lead to further consistency in the results of these assays.44 As presented in Tables 3 and 4, the highest coefficient of variation in HBV DNA measure is less than 100% (i.e., ⫾0.35 log10 copies/mL). The intraindividual variation of HBV DNA level in placebo arms of clinical trials over 1 to 2 months was reported to be as low as 2% to 4%, or 0.012 (⫾0.13) log10 copies/mL.11,14,25 Thus, within these stringent conditions, the overall precision of the quantitative measure suggests that more than 1 log10 variation from one sample to another in a given patient is unlikely to be due to an assay or intrasubject variability in the patient population enrolled in a clinical trial and at a group level. Correlation Between HBV DNA Level or Change and Histologic Grading. We used baseline information to compare histologic activity index and HBV DNA level

among untreated patients. As shown in Fig. 1, we observed a correlation between viral load levels and histologic grading (r ⫽ 0.78; P ⫽ .0001). However, this correlation was essentially influenced by one study in patients with low viral load levels. This study enrolled 58 inactive HBV carriers with histologic assessment.28 Subpopulation analyses showed statistically significant correlations in HBeAg-positive and -negative patients (Table 5). The small number of arms involving cirrhotic patients prevented this analysis in this population. As shown in Fig. 2, there was a statistically significant correlation between HBV DNA levels and histologic grading at the end of treatment (r ⫽ 0.71; P ⫽ .003). This correlation was present in the subset analyses of HBeAgpositive patients and in those treated with nucleoside analogues (Table 5). Conversely, no statistically significant correlations were shown in HBeAg-negative patient arms (n ⫽ 4 study arms) and in those studying interferon therapy (n ⫽ 8 study arms). Figure 3 shows the change from baseline to the end of treatment in median histologic grading versus change in median HBV DNA levels. The data show a strong exponential correlation between serum viral suppression and histologic improvement (r ⫽ 0.96; P ⫽ 10⫺6). The removal of the 2 highest points in the curve does not modify

HEPATOLOGY, Vol. 37, No. 6, 2003

MOMMEJA-MARIN ET AL.

1313

Table 3. Characteristics of Non-Polymerase Chain Reaction HBV DNA Assays

Assay

Detection Limit

Linearity

Specificity

Genostix; Abbott Laboratories, Chicago, IL, liquid hybridization

4 to 5 ⫻ 105 c/mL or 106 c/mL*

Genotype dependent

Genotype dependent

Digene HBV test Hybrid Capture II (standard); Digene Corp., Gaithersburg, MD Digene HBV test Hybrid Capture II (ultrasensitive); Digene Corp., Gaithersburg, MD Quantiplex; HBV DNA assay (branched DNA); Chiron Corp., Emeryville, CA

1.4 ⫻ 105 c/mL

1.4 ⫻ 105 to 1.7 ⫻ 109 c/mL

97.7%-100%

4.7 ⫻ 103 c/mL

4.7 ⫻ 103 to 3 ⫻ 108 c/mL

7 ⫻ 105 c/mL

7 ⫻ 105 to 5.7 ⫻ 109 c/mL overreporting 0.5 log10 c/mL when compared with Amplicor and Digene Hybrid Capture

Within-Run Coefficient of Variation

Between-Run Coefficient of Variation

References

Less precise than the Chiron Quantiplex and Digene Hybrid Capture 5%-21% ⫾ 0.025 log10 c/mL

At least 0.6 log10 c/mL

35,36

4.4% to 39.1% ⫾ 0.194 log10 c/mL

37,38

96.8%-100%

7%-22%

3%-33%

36-38

97.7%-100%

12.3% ⫾ 0.029 log10 c/mL

5%-39.5% at worst 0.7 log10 c/mL

35,37

Abbreviation: c/mL, copies/mL. *Corrected limit of detection.3

the correlation coefficient (r ⫽ 0.97; P ⫽ .00001). All but one study arm enrolled HBeAg-positive patients, and one arm represents patients treated with an interferon-containing regimen. This result suggests that histologic grading improves with a modest decrease in viral load level. For example, a trial comparing famciclovir with placebo showed a significant difference in favor of the drug arm based on the percentage of patients with histologic response but a modest 0.6 log10 copies/mL decrease in median HBV DNA.11 The spontaneous variation of HBV DNA over 1 year was 0.2 to 0.4 log10 copies/mL in the placebo or control arms of the studies.11,17,23,25

Correlation Between HBV DNA Level and Abnormal Aminotransferase Levels. As shown in Fig. 4, we observed a correlation between level of viral load and percentages of patients with normal aminotransferase levels (r ⫽ 0.62; P ⫽ .0004). The removal of the 2 highest points of the curve (e.g., with the lower viral load) did not modify the correlation. Furthermore, the percentage of patients with normal aminotransferase levels was 76% (813 of 1,070) (17 treatment arms) in a population of patients with a median HBV DNA less than 106 copies/mL versus 18% (333 of 1,826) in a population of patients with a median HBV DNA greater than 106 cop-

Table 4. Characteristics of Polymerase Chain Reaction HBV DNA Assays

Assay

Microwell plate Amplicor HBV Monitor; Roche Diagnostics, Branchburg, NJ Cross-linking naxcor PCR Real-time PCR (e.g., TaqMan) Cobas Amplicor HBV Monitor; Roche Diagnostics, Branchburg, NJ PCR beacon molecular detection TMA-HPA

Detection Limit

103

c/mL

5 ⫻ 105 c/mL 300 c/mL 4 ⫻ 102 c/mL

Linearity

103

to 3.2 ⫻

105

c/mL

100 c/mL

5 ⫻ 105 to 3 ⫻ 109 c/mL 10 to 109 c/mL 4 ⫻ 102 to 4 ⫻ 105 c/mL, underestimation at higher levels 100 to 109 c/mL

5 ⫻ 103 c/mL

5 ⫻ 103 to 5 ⫻ 108 c/mL

Specificity

94%-100%

96.1%-100% 99%

Within-Run Coefficient of Variation

Between-Run Coefficient of Variation

References

0%-56.8% ⫾ 0.093 log10 c/mL

8.5%-94% (average, 38.9%)

37,39

4.3%-7.3% 0.65%-23% 8.8%-53.5% ⫾ 0.093 log10 c/mL

4% 1.32%-22% 17.8%-96.1% ⫾ 0.356 log10 c/mL

40 41,42 39,43

0.35 ⫾ 0.08

38

0.7%-1.4%

13

0.9%-2.1%

Abbreviations: c/mL, copies/mL; PCR, polymerase chain reaction; TMA-HPA, transcription-mediated amplification and hybridization protection assay.

1314

MOMMEJA-MARIN ET AL.

HEPATOLOGY, June 2003

Fig. 1. Correlation between histologic necroinflammatory activity and HBV DNA level among untreated patients (r ⫽ 0.78; P ⫽ .0001). Prospective studies are reported as one point per arm; the size of the point is proportional to the arm size.

Fig. 2. Correlation between histologic activity index and HBV DNA level at the end of the treatment (r ⫽ 0.71; P ⫽ .003). Each treatment arm is represented as one point; the size of the point is proportional to the arm size.

ies/mL (P ⬍ .0001) (11 treatment arms). The correlation between absolute level of HBV DNA and ALT normality was observed among studies enrolling HBeAg-positive, HBeAg-negative, and cirrhotic patients (Table 5). Overall, we did not observe a significant correlation between HBV DNA change from baseline and ALT normality in treated patients. However, after separating trials with nucleosides and interferon, we saw in studies using nucleosides a correlation between the proportion of patients with normal aminotransferase levels and a change in HBV DNA level from baseline. In studies with interferon-containing regimens, a weak correlation was shown between the absolute levels and ALT normality. Absolute Level or Change From Baseline in HBV DNA and HBeAg Serologic Response. As shown in Fig. 5, there was a statistically significant correlation between

change in viral load and incidence of HBeAg seroconversion (r ⫽ 0.72; P ⫽ .0002). A similar correlation was observed between absolute level of viral load and incidence of seroconversion (Table 5). These correlation coefficients were especially high among studies with nucleosides. Among reports studying interferon therapy, a correlation was shown only with the change from baseline in median serum HBV DNA levels. The number of studies available among cirrhotic patients was too small to perform this subanalysis. Separate reports (partly included in this analysis) identified various levels of HBV DNA to be reached 4 to 16 weeks after the initiation of therapy and associated with a higher probability of seroconversion. The exact level of viremia to be achieved was not clearly defined in most of these studies because they used insufficiently sensitive assays; thus, this level ranged

Table 5. Coefficient of Correlation and Statistical Significance Between HBV DNA and Specific End Points by Subgroups HBV DNA Versus End Points

Absolute level HAI score at baseline HAI score at the end of treatment Proportion of patients with normal ALT levels Incidence of HBe seroconversion Change in median from baseline HAI score, median change from baseline Proportion of patients with normal ALT levels Incidence of HBe seroconversion

Statistics

Overall

HBeAg Positive

HBeAg Negative

Cirrhosis

Nucleoside Analogues

Interferon

Liquid Hybridization Assay LOD Corrected

r P r P r P r P

0.78 .0001 0.71 .003 0.62 .004 0.72 .002

0.61 .05 0.71 .01 0.66 .0008 NA NA

0.95 .0004 0.66 .3 0.71 ⬍.05 NA NA

ND ND ND ND 0.8 .003 ND ND

NA NA 0.76 .006 0.82 .0001 0.92 .0001

NA NA 0.65 .08 0.55 .02 ND ND

0.65 .002 0.67 .01 0.61 .0005 0.55 .03

r P r P r P

0.96 10⫺6 0.5 .06 0.72 .0002

0.98 6 ⫻ 10⫺6 ND ND NA NA

ND ND ND ND NA NA

ND ND ND ND 0.87 .00005

0.96 .00003 0.76 .003 0.7 .05

ND ND ND ND ND ND

NA NA NA NA NA NA

Abbreviations: HAI, histologic activity index; LOD, lower limit of detection; ND, not done; NA, not applicable; HBe, hepatitis B e.

HEPATOLOGY, Vol. 37, No. 6, 2003

Fig. 3. Correlation between change in median viral load level and change in median histologic activity index from beginning to end of the treatment (r ⫽ 0.96; P ⬍ 3 ⫻ 10⫺6). Prospective studies are reported as one point per arm; the size of the point is proportional to the arm size.

from less than 103 copies/mL to less than 3 ⫻ 106 copies/ mL.9,12,18,20,29,47-50 Conversely, treatments without antiviral efficacy, as measured by viral load evolution, had no significant differences in HBeAg seroconversion rates compared with placebo.11 Correlation Between HBV DNA Absolute Level and Loss of Hepatitis B Surface Antigen. The correlation between loss of hepatitis B surface antigen (HBsAg) and HBV DNA change from baseline was not assessable in this analysis; however, we were able to conduct the analysis correlating HBV DNA absolute levels and loss of HBsAg. Our analysis suggests that HBsAg clearance occurs at very low viral replication levels, close to 1,000 copies/mL or less. Two of 367 patients (0.5%) with viral load levels greater than 1,000 copies/mL lost HBsAg versus 25 of 74 (33.8%) with viral load levels less than 1,000

Fig. 4. Correlation between viral load level (log10 HBV DNA) and percentage ⫾ 95% confidence interval of patients with normal ALT levels (r ⫽ 0.62; P ⫽ .0004). Prospective studies are represented as one point per arm, studies without interferon (grey), and studies with interferon (black).

MOMMEJA-MARIN ET AL.

1315

Fig. 5. Correlation between change in median HBV DNA log10 level from baseline to the end of treatment and incidence of HBeAg seroconversion ⫾ 95% confidence interval (per 100 patient-months) (r ⫽ 0.72; P ⬍ .0002). Prospective studies are represented as one point per arm, studies without interferon (grey), and studies with interferon (black).

copies/mL (P ⬍ .0001). However, this analysis almost exclusively relies on interferon trials, and these findings may not be directly applicable to nucleosides. The predominance of interferon studies in this analysis may merely reflect the late occurrence of HBsAg loss (median of 14 months after the end of treatment), because most reports of nucleoside analogue studies have not yet described this duration of follow-up.7

Discussion The results presented herein provide information useful for the determination of the goal of anti-HBV therapy. However, because the analyses rely on published clinical trials, they can only be extended to the population of such trials. For instance, most trials were conducted in patients with detectable levels of viral load (using non–polymerase chain reaction assays) at entry and abnormal aminotransferase levels. Only a subset of patients had cirrhosis, and most had a mild compensated liver disease. The assays used to measure viral load are frequently nonsensitive, especially in older studies, and therefore in interferonstudying trials. However, as shown in Table 5, most of the correlations were robust to the correction of the liquid hybridization assay to 106 copies/mL. Conversely, we believe that treatment effect on viral replication as measured by the serum HBV DNA level changes from baseline is less susceptible to those differences in measurement and yielded comparable results. This assumption is consistent with recent guidelines for the use of serum HBV DNA measure in clinical practice.9,37 Studies used in these analyses differ in size and quality; however, this heterogeneity also reflects the variability of measurements used in prac-

1316

MOMMEJA-MARIN ET AL.

tice and may provide information useful to daily practice. These results provide a supportive data package for the use of HBV DNA as a surrogate marker of treatment effect and need to be confirmed by other data sets using individual patient data. We found a weak correlation between serum HBV DNA levels and histologic scoring at baseline. This finding is probably related to the skewed distribution of median HBV DNA levels in patients enrolled in clinical trials. When viral replication is greater than 106 copies/ mL, the median histologic grading is high in general. The paucity of studies with sensitive measurement of the HBV DNA levels leads to an overestimation of the median viral load level. Our findings are consistent with other studies reporting a lack of correlation between histology and HBV DNA levels assessed with nonsensitive assays in HBeAg-positive patients (known to have high viral load), whereas studies conducted in HBeAg-negative patients with lower viral load and with polymerase chain reaction assays showed a correlation.45 When analyzing the same correlation at the end of therapy, our findings confirm the trend seen in untreated patients and reinforce our hypothesis that the lack of correlation between HBV DNA level and histologic grading in previous reports is likely attributable to the limited dynamic range of the assays used in HBeAg-positive patients. Subset analyses showed that this correlation is statistically significant only in HBeAgpositive patients and in those treated with nucleoside analogues. The lack of correlation in HBeAg-negative patients is possibly related to the small number of studies available; however, we cannot rule out a different pathophysiology of the disease in the HBeAg-positive and -negative patient population, which could explain this lack of correlation. With respect to the interferon trials, it is likely that the use of non–polymerase chain reaction assays in all but one study has impaired the ability to adequately measure the viral load levels, therefore biasing the results. Because the mechanism of action of interferon is different from that of nucleoside analogues, it is also possible that the viral suppression is not, with this compound, the primary mediator of histologic change. In several studies, a 105 to 106 copies/mL HBV DNA level allowed differentiation between patients with active chronic HBV infection and asymptomatic chronic carriers.9 Although recognizing that there is no strict scientific evidence of the relevance of this threshold, a recent National Institutes of Health workshop established 105 copies/mL as an upper limit defining inactive HBsAg carriers.3 The data in our analysis are consistent with this definition. Our analysis, almost exclusively based on HBeAg-positive patients treated with nucleoside analogues, compar-

HEPATOLOGY, June 2003

ing median change in histologic score with median change in log10 serum HBV DNA level suggests that a greater than 1 log10 copies/mL change in median serum HBV DNA level will translate into a 2-point change in median histologic grading, thus meeting the current criterion for histologic response. Interestingly, our analysis indicates that it is difficult to show more than a 4-point improvement in the median histologic activity index score, and this narrow range of assessable change in median histologic activity index score is responsible for the asymptotic nature of the curve. Therefore, considering the ongoing effort to develop new antiviral drugs for the treatment of chronic HBV infection, it is possible that histologic scoring will not be sensitive enough to distinguish between 2 active antiviral regimens. Histology is adequate to evaluate the efficacy of a treatment versus placebo. However, histologic grading lacks the ability to rapidly discriminate differences in potency at multiple time points in clinical trials comparing active drugs or combination therapy. This impracticability might be of concern in the light of other chronic viral diseases, showing a direct relationship between incomplete viral suppression and emergence of resistant strains. For instance, 2 treatments producing a 2 or 4 log10 copies/mL decrease in HBV DNA, respectively, would translate into a similar histologic response related to the small number of possibility in using this semiquantitative measure. By contrast, the difference in serum HBV DNA suppression would be distinguishing the 2 therapies. Our findings support the view that viral replication is the disease and liver histologic damage is the consequence of the disease, similar to human immunodeficiency virus infection, in which viral replication is the disease and low CD4⫹ cell count is the consequence. Analyzing the correlation between proportion of patients with normal ALT levels and serum HBV DNA levels, we found variable levels of statistically significant correlations in all of the populations tested (including HBeAg-negative patients and those with cirrhosis). The correlation was weak in patients treated with interferon. The evidence of a correlation only to the absolute level of serum HBV DNA seems to argue in favor of a threshold effect of viral load on elevation of aminotransferases. This threshold seems to be close to 106 copies/mL, in agreement with recent guidelines.3 In HBeAg-negative patients, the high statistical significance in the correlation between ALT normality and HBV DNA levels supports our hypothesis that the lack of correlation between serum HBV DNA levels and histology shown previously was due to the small number of trials available for this analysis. The correlation between change in median serum HBV DNA level and ALT normality was only found in the

HEPATOLOGY, Vol. 37, No. 6, 2003

subset of patients treated with nucleoside analogues. This is possibly due to the limitations related to the use of nonsensitive assays in the interferon trials. The difference shown when comparing interferon with nucleoside analogues may also reflect a difference in the mechanism of action of the drugs. With interferon, a transient hepatic flare is commonly seen during treatment and has been associated with serologic response and long-term normalization of transaminases.7,46 Consistent with previous reports, our analysis confirms a relationship between the decrease of viral load and the incidence of HBeAg seroconversion. At a similar magnitude of viral suppression, the incidence of seroconversion was higher with interferon than with nucleoside therapy, as shown in Fig. 5. This is possibly related to the immunologic activity of interferon. The strength of the correlation was robust to this difference because it was even higher among studies with nucleoside-containing regimens alone. The slightly lower correlation shown between HBV DNA absolute level and incidence of seroconversion was also stronger in the subset of studies involving only nucleoside-containing regimens. Again, among the interferon-studying trials, the correlation was weak and only significant when comparing the change from baseline and the incidence of seroconversion. HBeAg seroconversion has already been shown to correlate with improvement in survival and morbidity7,8 and is probably a reliable surrogate marker for prognosis of chronic HBV infection. Several studies have shown an improvement in survival, progression to cirrhosis, liver disease–associated complications, and hepatocellular carcinoma in patients experiencing an HBeAg seroconversion spontaneously or induced by treatment.2,19 Our findings show that HBeAg seroconversion is correlated with a change or decrease in viral replication. These findings support the conclusion that the goal of anti-HBV therapy should be a profound and durable viral suppression, optimally leading to loss of HBsAg. Considering the limitations of our approach (i.e., the summative methodology using summary statistics for point estimation of response per treatment arm instead of individual data in a selected patient population), we were surprised to find statistically significant and consistent correlations between viral load change or viral load level and various accepted markers of disease activity (histologic grading and biochemical and serologic response). Although not as strong, a relationship was also observed between viral load level and loss of HBsAg. However, these findings are essentially applicable to the patient population enrolled in clinical trials, HBeAg positive at baseline, and treated with nucleoside analogues. Further studies are warranted to determine whether such correla-

MOMMEJA-MARIN ET AL.

1317

tions are true in the course of the HBeAg-negative disease. Our analysis suggests that treatment-induced reduction in HBV DNA level of greater than 1 log10 copies/mL could translate into histologic response at a population level. We observed that quantitative HBV DNA has a broader dynamic range than semiquantitative histologic grading and therefore would allow demonstration of a difference between 2 active treatments of unequal potency. As is the case for the management of hepatitis C virus and human immunodeficiency virus infection, these findings support the conclusion that the goal of anti-HBV therapy should be a profound and durable viral suppression, as defined by very sensitive assays. Additional prospective studies are needed to confirm these preliminary results and precisely determine the desirable level of viremia to attain. Furthermore, these studies could also help to define a suboptimal response early in the course of therapy. Such definition of treatment failure would allow modification of therapy and limit unnecessary toxicity and/or emergence of resistance. Acknowledgment: The authors thank Jeff Sorbel, Charles Wakeford, and Joseph Quinn for their constructive criticism and comments as well as Patience Vanderbush for her contribution to the manuscript.

References 1. McMahon BJ, Holck P, Bulkow L, Snowball M. Serologic and clinical outcome of 1536 Alaska natives chronically infected with hepatitis B virus. Ann Intern Med 2001;135:759-768. 2. Lau DT-Y, Everhart J, Kleiner DE, Park Y, Vergalla J, Schmid P, Hoofnagle JH. Long term follow up of patients with chronic hepatitis B treated with interferon alfa. Gastroenterology 1997;113:1660-1667. 3. Lok ASF, McMahon BJ. Chronic hepatitis B. AASLD practice guidelines. HEPATOLOGY 2001;34:1225-1241. 4. Bravo AA, Sheth SG, Chopra S. Liver biopsy. N Engl J Med 2001;334: 495-500. 5. Lee WM. Medical progress: hepatitis B virus infection. N Engl J Med 1997;337:1733-1745. 6. Evans AA, O’Connell AP, Pugh JC, Mason WS, Shen FM, Chen GC, Lin WY, et al. Geographic variation in viral load among hepatitis B carriers with differing risk of hepatocellular carcinoma. Cancer Epidemiol Biomarkers Prev 1998;7:559-565. 7. Niederau C, Heintges T, Lange S, Goldmann G, Niederau CM, Mohr L, Haussinger D. Long term follow-up of HBeAg-positive patients treated with interferon alpha for chronic hepatitis B. N Engl J Med 1996;334: 1422-1427. 8. Di Marco V, Lo Iacono O, Camma C, Vaccaro A, Giunta M, Martorana G, Fuschi P, et al. The long term course of chronic hepatitis B. HEPATOLOGY 1999;30:257-264. 9. Gerken G, Gomes J, Lampertico P, Colombo M, Rothaar T, Trippler M, Colucci G. Clinical evaluation and applications of the amplicor HBV Monitor test, a quantitative HBV DNA PCR assay. J Virol Methods 1998; 74:155-165. 10. Yuen MF, Sablon E, Hui CK, Yuan HJ, Decreamer H, Lai CL. Factors associated with hepatitis B virus DNA breakthrough in patients receiving prolonged lamivudine therapy. HEPATOLOGY 2001;34:785-791. 11. De Man RA, Marcellin P, Habal F, Desmond P, Wright T, Rose T, Jurewicz R, et al. A randomized, placebo-controlled study to evaluate the

1318

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

MOMMEJA-MARIN ET AL.

efficacy of 12-month famciclovir treatment in patients with chronic hepatitis B e antigen-positive hepatitis B. HEPATOLOGY 2000;32:413-417. Lindh M, Hannoun C, Horal P, Krogsgaard K, and the interpred study group. Virological response to interferon therapy of chronic hepatitis B as measured by a highly sensitive assay. J Viral Hepat 2001;8: 349-357. Sakugawa H, Nakasone H, Nakayoshi T, Kawakami Y, Yamashiro T, Maeshiro T, Kinjo F, et al. Correlation between serum transaminase activity and virus load among patients with chronic liver disease type B. Hepatol Res 2001;21:159-168. De Man RA, Wolters LMM, Nevens F, Chua D, Sherman M, Lai CL, Gadano A, et al. Safety and efficacy of oral entecavir given for 28 days in patients with chronic hepatitis B virus infection. HEPATOLOGY 2001;34: 578-582. Papatheodoridis GV, Manesis E, Hadziyannis SJ. The long term outcome of interferon-alpha treated and untreated patients with HBeAg-negative chronic hepatitis B. J Hepatol 2001;34:306-313. Brunetto MR, Oliveri F, Coco B, Leandro G, Colombatto P, Gorin JM, Bonino F. Outcome of antiHBe positive chronic hepatitis B in alphainterferon treated and untreated patients: a long term cohort study. J Hepatol 2002;36:263-270. Carreno V, Marcellin P, Hadziyannis S, Salmeron J, Diago M, Kitis GE, Vafiadis I, et al. Retreatment of chronic hepatitis B e antigen-positive patients with recombinant interferon alfa-2a. HEPATOLOGY 1999;29:277282. Hwang SJ, Lu RH, Wang YH, Chu CW, Wu JC, Chang FY, Lee SD. Clinical significance of changes in serum hepatitis B virus DNA titer in patients with chronic hepatitis B treated with interferon. Chung Hua I Hsueh Tsa Chih (Taipei) 2000;63:1-7. Lin S-M, Sheen I-S, Chien R-N, Chu C-M, Liaw Y-F. Long term beneficial effect of interferon therapy in patients with chronic hepatitis B virus infection. HEPATOLOGY 1999;29:971-975. Janssen HL, Gerken G, Carreno V, Marcellin P, Naoumov NV, Craxi A, Ring-Larsen Helmer, et al. Interferon alfa for chronic hepatitis B infection: increased efficacy of prolonged treatment. The European concerted action on viral hepatitis (EUROHEP). HEPATOLOGY 1999;30:238-243. Lampertico P, Del Ninno E, Manzin A, Donato MF, Rumi MG, Lunghi G, Morabito A, et al. A randomized, controlled trial of 24-month course of interferon-alfa 2b in patients with chronic hepatitis B who had hepatitis B DNA without hepatitis B e antigen in serum. HEPATOLOGY 1997;26:16211625. Lau DT-Y, Khokhar MF, Doo E, Ghany MG, Herion D, Park Y, Kleiner DE, et al. Long term therapy of hepatitis B with lamivudine. HEPATOLOGY 2000;32:828-834. Lai C-L, Chien R-N, Leung NWY, Chang T-T, Guan R, Tai D-I, Ng K-Y, et al. A one year trial of lamivudine for chronic hepatitis B. N Engl J Med 1998;339:61-68. Leung NWY, Lai C-L, Chang T-T, Guan R, Lee C-M, Ng K-Y, Lim S-G, et al. Extended lamivudine treatment in patients with chronic hepatitis B enhances hepatitis B e antigen seroconversion rate: results after 3 years of therapy. HEPATOLOGY 2001;33:1527-1532. Dienstag JL, Schiff ER, Wright TL, Perillo RP, Hann H-WL, Goodman Z, Crowther L, et al. Lamivudine as initial treatment for chronic hepatitis B in the United States. N Engl J Med 1999;341:1256-1263. Lindh M, Horal P, Dhillon AP, Norkrans G. Hepatitis B virus DNA levels, precore mutations, genotypes and histological activity in chronic hepatitis B. J Viral Hepat 2000;7:258-267. Tatulli I, Francavilla R, Rizzo GL, Vinciguerra V, Ierardi E, Amoruso A, Panella C, et al. Lamivudine and alpha-interferon in combination long term for precore mutant chronic hepatitis B. J Hepatol 2001;35: 805-810. Martinot-Peignoux M, Boyer N, Colombat M, Akremi R, Pham BN, Ollivier S, Castelnau C, et al. Serum hepatitis B virus DNA levels and liver histology in inactive HBsAg carriers. J Hepatol 2002;36:543546.

HEPATOLOGY, June 2003

29. Zollner B, Schafer P, Feucht H-H, Schroter M, Petersen J, Laufs R. Correlation of hepatitis B virus load with loss of e antigen and emerging drug-resistant variants during lamivudine therapy. J Med Virol 2001;65: 659-663. 30. Sarin SK, Guptan RC, Thakur V, Malhotra S, Malhotra V, Banerjee K, Khandekar P. Efficacy of low dose alpha interferon therapy in HBV-related chronic liver disease in Asian Indians: randomized controlled trial. J Hepatol 1996;24:391-396. 31. Yao FY, Terrault NA, Freise C, Maslow L, Bass NM. Lamivudine treatment is beneficial in patients with severely decompensated cirrhosis and actively replicating hepatitis B infection awaiting liver transplantation: a comparative study using a matched, untreated cohort. HEPATOLOGY 2001; 34:411-416. 32. Barbaro G, Zechini F, Pellicelli AM, Francavilla R, Scotto G, Bacca D, Bruno M, et al. Long-term efficacy of interferon alpha-2b and lamivudine in combination compared to lamivudine monotherapy in patients with chronic hepatitis B. An Italian multicenter, randomized trial. J Hepatol 2001;35:406-411. 33. Schalm SW, Heathcote J, Cianciara J, Sherman M, Willems B, Dhillon A, Moorat A, et al. Lamivudine and alpha interferon combination treatments of patients with chronic hepatitis B infection: a randomised trial. Gut 2000;46:562-568. 34. Quint WGV, Heijtink RA, Schirm J, Gerlich WH, Niesters HGM. Reliability of methods for hepatitis B virus DNA detection. J Clin Microbiol 1995;33:225-228. 35. Heermann KH, Gerlich WH, Chudy M, Schaefer S, Thomssen R, and the Eurohep Pathobiology Group. Quantitative detection of hepatitis B virus DNA in two international reference plasma preparations. J Clin Microbiol 1999;37:68-73. 36. Niesters HGM, Krajden M, Cork L, de Medina M, Hill M, Fries E, Osterhaus ADME. A multicenter study evaluation of the Digene hybrid capture II signal amplification technique for detection of hepatitis B virus DNA in serum samples and testing of Eurohep standards. J Clin Microbiol 2000;38:2150-2155. 37. Pawlotsky JM, Bastie A, Hezode C, Lonjon I, Darthuy F, Remire J, Dhumeaux D. Routine detection and quantification of hepatitis B virus DNA in clinical laboratories: performance of three commercial assays. J Virol Methods 2000;85:11-21. 38. Yates S, Penning M, Goudsmit J, Frantzen I, van der Weijer B, van Strijp D, van Gemen B. Quantitative detection of hepatitis B virus DNA by real-time nucleic acid sequence-based amplification with molecular beacon detection. J Clin Microbiol 2001;39:3656-3665. 39. Noborg U, Gusdal A, Pisa EK, Hedrum A, Lindh M. Automated quantification analysis of hepatitis B virus DNA by using the cobas amplicor HBV monitor test. J Clin Microbiol 1999;37:2793-2797. 40. Kao JH, Wood M, Chen PJ, Lai MY, Chen DS. Comparison of two methods for quantification of hepatitis B viral DNA. J Gastroenterol Hepatol 1999;14:423-426. 41. Gruber F, Falkner FG, Dorner F, Hammerle T. Quantification of viral DNA by real time PCR applying duplex amplification, internal standardization and two color fluorescence detection. Appl Environ Microbiol 2001;67:2837-2839. 42. Pas SD, Fries E, De Man RA, Osterhaus ADME, Niesters HGM. Development of a quantitative real-time detection assay for hepatitis B virus DNA and comparison with two commercial assays. J Clin Microbiol 2000; 38:2897-2901. 43. Marin IJ, Poljak M, Seme K, Meglic-Volkar J, Maticic M, Lesnicar G, Brinovec V. Comparative evaluation of semiautomated cobas amplicor hepatitis B virus (HBV) monitor test and manual microwell plate-based amplicor HBV monitor test. J Clin Microbiol 2001;39:758-761. 44. Saldanha J, Gerlich W, Lelie N, Dawson P, Heerman K, Heath A, the WHO Collaborative Study. An international collaborative study to establish a world health organization international standard for hepatitis B virus DNA nucleic acid amplification techniques. Vox Sang 2001;80:63-71. 45. Zavaglia C, Mondazzi L, Maggi G, Iamoni G, Gelosa F, Bellati G, Col-

HEPATOLOGY, Vol. 37, No. 6, 2003

loredo G, et al. Are alanine aminotransferase, hepatitis B virus DNA or IgM antibody to hepatitis B core antigen serum levels predictors of histological grading in chronic hepatitis B? Liver 1997;17:83-87. 46. Nair S, Perillo RP. Serum alanine aminotransferase flares during interferon treatment of chronic hepatitis B: is sustained clearance of HBV-DNA dependent on levels of pretreatment viremia? HEPATOLOGY 2001;34;10211026. 47. Nagata I, Colucci G, Gregorio GV, Cheeseman P, Williams R, MieliVergani G, Vergani D. The role of HBV DNA quantitative PCR in monitoring the response to interferon treatment in chronic hepatitis B virus infection. J Hepatol 1999;30:965-969.

MOMMEJA-MARIN ET AL.

1319

48. Lok AS, Ghany MG, Waston G, Ayola B. Predictive value of aminotransferase and hepatitis B virus DNA levels on response to interferon therapy for chronic hepatitis B. J Viral Hepat 1998;5:171-178. 49. Lau GKK, Tsiang M, Hou J, Yuen S-T, Carman WF, Zhang L, Gibbs CS, et al. Combination therapy with lamivudine and famciclovir for chronic hepatitis B-infected Chinese patients: a viral dynamics study. HEPATOLOGY 2000;32:394-399. 50. J Gauthier, Bourne EJ, Lutz MW, Crowther LM, Dienstag JL, Brown NA, Condreay LD. Quantification of hepatitis B viremia and emergence of YMDD variants in patients with chronic hepatitis B treated with lamivudine. J Infect Dis 1999;180:1757-1762.