| |
Serum HBV DNA as a marker of efficacy during therapy for chronic HBV infection: Analysis and review of the literature
|
| |
| |
Hepatology, June 2003, Volume 37, Number 6
Herve, Mommeja-Marin
From Triangle Pharmaceuticals, Inc., Durham, NC.
"...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.....Our analysis, almost exclusively based on HBeAg-positive patients treated with nucleoside analogues, comparing 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.... these studies could also help to define a suboptimal response early in the course of therapy...."
Chronic hepatitis B virus (HBV) infection often leads to severe complications and death after decades; therefore, the evaluation of therapeutic regimens is based on surrogate markers such as histology and hepatitis B e antigen (HBeAg) seroconversion. 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 the etiology and staging of liver diseases. 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. 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, 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. We hypothesized that serum HBV DNA level is a direct reflection of the level of intrahepatic viral replication.
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.
We used baseline information to compare histologic activity index and HBV DNA level among untreated patients. We observed a correlation between viral load levels and histologic grading. 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. Subpopulation analyses showed statistically significant correlations in HBeAg-positive and -negative patients.
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 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. 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.
A similar correlation was observed between absolute level of viral load and incidence of seroconversion. 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 from less than 103 copies/mL to less than 3 x 106 copies/mL. Conversely, treatments without antiviral efficacy, as measured by viral load evolution, had no significant differences in HBeAg seroconversion rates compared with placebo.
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 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.
Discussion by authors
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 interferon-studying trials. However, 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. Studies used in
these analyses differ in size and quality; however, this heterogeneity also reflects the variability of measurements used in practice 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. 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 HBeAg-positive 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. 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. The data in our analysis are consistent with this definition.
Our analysis, almost exclusively based on HBeAg-positive patients treated with nucleoside analogues, comparing 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. 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 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.
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. 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 morbidity 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.
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 correlations 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.
|
|
| |
| |
|
|
|
