| |
Noninvasive Markers of Fibrosis for Longitudinal Assessment of Fibrosis in Chronic Liver Disease: Are They Ready for Prime Time? EDITORIAL
|
| |
| |
The American Journal of Gastroenterology
September 2005
Paul J. Thuluvath, M.D., F.R.C.P.1, and Karen L. Krok, M.D.1
1Department of Medicine, The Johns Hopkins University School of Medicine, Baltimore, Maryland
..... the current published evidence suggests that FibroTest is not yet ready for prime time either to diagnose the severity of fibrosis or for longitudinal assessment of fibrosis since 15-20% are likely to be misdiagnosed by these markers....
The study published that this Editorial refers to follows the Editorial below.
SUMMARY
Over the past decade, there has been a renewed enthusiasm to develop noninvasive serum markers or tests to assess the presence and severity of fibrosis in chronic liver disease. Although a single marker or test has lacked the necessary accuracy to predict fibrosis, different combinations of these markers or tests have shown encouraging results. However, interlaboratory variability and inconsistent results with liver diseases of varying etiologies have made it difficult to assess the reliability of these markers in clinical practice. In this issue of the Journal, Poynard and colleagues describe the "histological" response to lamivudine in patients with chronic HBV over a 24-month period using surrogate serum biomarkers (FibroTest-ActiTest) without corroborating histological data. Investigators found improvement in fibrosis and inflammation in 85% and 91%, respectively, despite the emergence of YMDD mutation in 41.5% of patients. The higher improvement rates reported in this study should be interpreted with caution for a number of reasons including the absence of data on virological response rates, corroboratory histological data, and data on the validity of FibroTest to evaluate fibrosis in a longitudinal manner. Although FibroTest has been studied extensively by the authors of the current study, to date there are only few independent studies. In addition to significant interlaboratory variations, these studies have shown that significant fibrosis could be missed, or conversely significant fibrosis diagnosed in the absence of minimal or no fibrosis in about 15-20% of patients. We may be approaching a time when serum biomarkers may become an integral part of the assessment of patients with chronic liver disease, but published evidence suggests that these markers are not yet ready for prime time.
ARTICLE TEXT
The assessment of the presence and severity of liver fibrosis is of paramount importance in determining treatment strategies, response to treatment, prognosis, and the potential risk for complications in patients with chronic liver disease. Liver biopsy, the gold standard for assessing the severity of necroinflammatory activity and fibrosis, is invasive, and even in expert hands, it is associated with rare but serious complications including bleeding, pneumothorax, and perforation of colon or gallbladder (1, 2). It is also not practical or cost-effective to perform serial biopsies within a short time interval to assess treatment-related response. In addition to the sampling error that is inherently associated with percutaneous liver biopsies, there is an intra- and interobserver variability that may range from 15% to 33% in determining the fibrosis staging (3-5). Nevertheless, liver biopsy has remained the "gold standard" to assess the severity of inflammation and fibrosis in patients with chronic liver disease.
In the past decade, many investigators have proposed noninvasive tests to replace liver biopsy using either a single biochemical marker or a combination of tests. An ideal noninvasive marker for the evaluation of liver fibrosis should accurately predict the presence or absence of significant fibrosis (high sensitivity, specificity, positive, and negative predictive value [NPV]). In addition, it should be readily available and reproducible with low interlaboratory variability and with applicability in liver disease of various etiologies. Although liver biopsy does not fulfill many of these criteria, it has remained the "gold standard" mainly because of the absence of better alternatives. Recently, there has been a renewed interest in noninvasive markers of fibrosis because of evolving novel therapies for hepatitis C and B (6-9). While some investigators have focused on a combination of laboratory tests such as reversal of aspartate aminotransferase (AST)/alanine aminotransferase (ALT) ratio, or AST/platelet ratio index (APRI), others have been searching for more novel markers of fibrosis and inflammation (8-12). Despite the common pathway for fibrosis, an imbalance between synthesis, and degradation of extra cellular matrix by activated stellate cells, there are currently no liver-specific markers that could accurately reflect liver fibrosis in liver disease of different etiologies. Many serum markers of matrix degradation products have been studied previously, either as a single marker or in combination, including N-terminal propeptide of type III collagen (PIIINP), collagen type-IV, hyaluronic acid, prolyl hydroxylase, laminin, matrix metalloproteinase-1 (MMP-1), and tissue inhibitors of metalloproteinase-1 (TIMP-1) and metalloproteinase-2 (TIMP-2). Many of these tests are not widely available, and when critically evaluated, they have shown disease-specific variations and inconsistencies suggesting that they are nonspecific. Many of these markers are nonspecifically elevated in the presence of inflammation or expressed in others tissues including skin, blood vessels, bone, and kidney. Other markers that have been studied include alpha2-macroglobulin, haptoglobins, and apolipoprotein A1. Increased serum levels of alpha2-macroglobulin, a protease inhibitor expressed by activated stellate cells, may relate to increased fibrosis. Haptoglobin levels may be decreased in liver fibrosis because of its association with profibrotic cytokine TGF-beta1, and apolipoprotein A1 levels may be decreased because of its binding affinity with extracellular matrix. Other tests that merits further independent corroboration include YKL-40 (a growth factor for fibroblasts that is expressed in the human liver), transient elastography, 13C-caffeine breath test, and DNA sequenced-based serum protein glycomics (10-13). The recent trend has been to use a combination of these tests to improve accuracy of predicting fibrosis in a consistent and reproducible manner (6-9).
The prospective multicenter study reported in this issue of American Journal of Gastroenterology is a longitudinal assessment of FibroTest and ActiTest (FT-AT) in 283 patients with chronic hepatitis B being treated with lamivudine (14). The biochemical tests were done in a central laboratory, but liver biopsy was done only at baseline. They found a statistically significant decrease in both fibrosis and inflammation as measured by FT-AT at 24 months irrespective of the presence of YMDD mutation. Interestingly, there was an immediate improvement at 6 month with a plateau between 6-12 months and then further improvement at 24 months. Lamivudine had previously been shown to significantly improve the histological features both for necroinflammatory activity and fibrosis. In the current study, 85% showed improvement in fibrosis and 91% showed improvement in inflammation at 24 months. It is reassuring to note the continued improvement in both fibrosis and inflammatory scores despite the emergence of YMDD mutation in 41.5% of patients. The higher improvement rates reported in this study should be interpreted with caution for a number of reasons including the absence of data on virological response rates, no corroboratory histological data, and a paucity of previously published data on the validity of FibroTest to evaluate fibrosis in a longitudinal manner.
The same investigators have published many previous articles on the validity of this proprietary test score (FibroTest) that uses five biochemical components including alpha2-macroglobulin, haptoglobin, apolipoprotein A1, gamma-glutamyl transpeptidase (GGT), and total bilirubin. ActiTest includes the same components plus serum ALT levels. The score is computed (using a complex formula) by entering patient's age and sex (older age and male sex were shown to be associated with more fibrosis in previous studies) along with the five components into a proprietary program (available for a fee) available at a website (http://www.biopredictive.com) and this composite value is used to determine the presence or absence of significant fibrosis. Poynard and colleagues have previously reported that the FibroTest score has a NPV (absence of fibrosis) of 100% and positive predictive value (presence of significant fibrosis) of 91% in patients with appropriate cut-off scores (6). This would be a remarkable development if it could be verified independently in a consistent fashion in liver disease of varying etiology. However, when the same investigators compared the test scores using laboratory results from eight centers with the reference laboratory, the interlaboratory agreement for fibrosis stage using this test score was very poor with kappa (kappa statistics) ranging from 0.32 to 0.94 with a mean 0.6 (15). In the current study, authors do not comment on the reproducibility of the tests in their reference laboratory over an unspecified period of time. Only a few studies have independently assessed FibroTest outside the current study group (12, 16). In an independent study, Rossi et al. found that 21% of patients who were predicted to have significant fibrosis by the FibroTest scores had only minimal fibrosis by histology and conversely, 18% of patients who were not supposed to have fibrosis by the test score had significant fibrosis (16). When FibroTest was studied in patients with chronic HBV in a previous study by Poynard's group, a FibroTest scores <= 0.20 had a NPV of 92% to exclude F2-F4 fibrosis, but the specificity was only 52% (17). The discrepancy in the published results could be explained by the inadequate length or sampling error of liver biopsy specimens, and inconsistencies in the measurement of biochemical markers (5, 15, 16, 18). Nevertheless, the current published evidence suggests that FibroTest is not yet ready for prime time either to diagnose the severity of fibrosis or for longitudinal assessment of fibrosis since 15-20% are likely to be misdiagnosed by these markers (15, 16, 18, 19).
Readers may find it disturbing to note that the first author of the study has a capital interest in Biopredictive, the company that markets FT-AT at their website for a fee, and another author is an employee of Biopredictive. Appropriately, most publications on the validity of the FT-AT have come from the same authors who have a research interest to develop noninvasive markers of fibrosis. The authors, rightfully so, have declared their conflict of interest. Another disturbing aspect of this article is that the authors are withholding important information on virological response rates perhaps for a later publication. This important piece of information is critical to interpret the results of the study.
REFERENCES
1. James CH, Lindor KD. Outcome of patients admitted with complications after outpatient liver biopsies. Ann Intern Med 1993;118: 96-8.
2. McGill DB, Rakela J, Zinsmeister AR, et al.. A 21-year experience with major hemorrhage after percutaneous liver biopsy. Gastroenterology 1990;99: 1396-400.
3. Westin J, Lagging LM, Wejstal R, et al.. Interobserver study of liver histopathology using the Ishak score in patient with chronic hepatitis C virus infection. Liver 1999;19: 183-7.
4. Bedossa P, Dargere D, Paradis V. Sampling variability of liver fibrosis in chronic hepatitis C. Hepatology 2003;38: 1449-57.
5. Regev A, Berhho M, Jeffers LJ, et al.. Sampling error and intraobserver variation in liver biopsy in patients with chronic HCV infection. Am J Gastroenterol 2002;97: 2614-8.
6. Imbert-Bismut F, Ratziu V, Pieroni L, et al.. Biochemical markers of liver fibrosis in patients with hepatitis C virus infection. Lancet 2001;357: 1069-75.
7. Forns X, Ampurdanes S, Llovet JM, et al.. Identification of chronic hepatitis C patients without hepatic fibrosis by simple predictive model. Hepatology 2002;36: 986-92.
8. Wai CT, Greenson JK, Fontana RJ, et al.. A simple noninvasive index can predict both significant fibrosis and cirrhosis in patients with chronic hepatitis C. Hepatology 2003;38: 518-26.
9. Rosenberg WM, Voelker M, Thiel R, et al.. Serum markers detect the presence of liver fibrosis: A cohort study. Gastroenterology 2004;127: 1704-13.
10. Saitou Y, Shiraki K, Yamanaka Y, et al.. Noninvasive estimation of liver fibrosis and response to interferon therapy by a serum fibrinogenesis marker, YKL-40, in patients with HCV-associated liver disease. World J Gastroenterol 2005;11: 476-81.
11. Callewaert N, Van Vlierberghe H, Van Hecke A, et al. Noninvasive diagnosis of liver cirrhosis using DNA sequencer-based total serum protein glycomics. Nat Med 2004;10: 429-34.
12. Caster L, Vergniol J, Foucher J, et al.. Prospective comparison of transient elastography, Fibrotest, APRI, and liver biopsy for the assessment of fibrosis in chronic hepatitis C. Gastroenterology 2005;l128: 343-50.
13. Park GJ-H, Katelaris PH, Jones DB, et al.. Validity of C-caffeine breath test as a non-invasive, quantitative test of liver function. Hepatology 2003;38: 1227-36.
14. Poynard T, Zoulim F, Ratziu V, et al.. Longitudinal assessment of histology surrogate markers (Fibrotest-Actitest) during lamivudine therapy in patients with chronic hepatitis B infection. Am J Gastroenterol 2005;100: 1970-80.
15. Halfon P, Imbert-Bismut F, Messous D, et al.. A prospective assessment of inter-laboratory variability markers of fibrosis (FibroTest) and activity (ActiTest) in patients with chronic liver disease. Comp Histol 2002;1: 3.
16. Rossi E, Adams L, Prins A, et al.. Validation of Fibro Test biochemical markers score in assessing liver fibrosis in hepatitis C patients. Clin Chem 2003;49: 450-4.
17. Myers RP, Tainturier MH, Ratziu V, et al.. Prediction of liver histological lesions with biochemical markers in patients with chronic hepatitis B. J Hepatol 2003;39: 222-30.
18. Rosenthall-Allieri MA, Peritore ML, Tran A, et al.. Analytical variability of the Fibrotest proteins. Clin Biochem 2005;38: 473-8.
19. Bissell DM. Assessing fibrosis without a liver biopsy: Are we there yet? Gastroenterology 2004;127: 1847-9.
Longitudinal Assessment of Histology Surrogate Markers (FibroTest-ActiTest) During Lamivudine Therapy in Patients with Chronic Hepatitis B Infection
The American Journal of Gastroenterology
Volume 100 Issue 9 Page 1970 - September 2005
Thierry Poynard, M.D., Ph.D.1, Fabien Zoulim, M.D., Ph.D.2, Vlad Ratziu, M.D., Ph.D.1, Françoise Degos, M.D., Ph.D.3, Francoise Imbert-Bismut, Ph.D.4, Paul Deny, M.D.5, Paul Landais, M.D.6, Abdelkader El Hasnaoui, M.D.7, Alain Slama, M.D.7, Patrick Blin, Ph.D.8, Vincent Thibault, M.D.9, Parviz Parvaz, M.D.10, Mona Munteanu, M.D.11, and Christian Trepo, M.D.2
ABSTRACT
OBJECTIVES: The noninvasive serum markers, FibroTest-ActiTest (FT-AT), are an alternative to liver biopsy in patients with chronic hepatitis C and B. The aim was to use these markers in a prospective study of patients treated with lamivudine in order to assess the impact of treatment, as well as the factors associated with fibrosis progression.
METHODS: Two hundred and ninety-eight patients were included in a prospective longitudinal study in 50 hospitals across France. FT-AT were measured at baseline, and then after 6, 12, and 24 months of lamivudine 100-mg treatment. Epidemiological, clinical, and virologic characteristics were analyzed by univariate and multivariate analysis.
RESULTS: Two hundred and eighty-three patients were included for analysis. The accuracy of FT-AT versus biopsy was validated with the area under the ROC curve, 0.77 (SE = 0.03) for bridging fibrosis and 0.75 (SE = 0.06) for severe activity (A3). At baseline, bridging fibrosis (METAVIR stages F2-F3-F4) was highly associated (p< 0.001) in multivariate analysis with male gender and age and marginally associated with anti-HBe presence (p= 0.05) and non-Asian ethnic origin (p= 0.046). Lamivudine treatment had a very significant impact overall. FT decreased significantly from 0.51 at baseline to 0.37 at 24 months (p< 0.001), and 85% of patients had improvement at 24 months. AT also decreased significantly from 0.56 to 0.13 (p< 0.0001), and 91% of patients had improvement at 24 months. A three-phase kinetics was observed for both fibrosis and activity; there was a marked improvement during the first 6 months, followed by a plateau between 6 and 12 months, and another improvement between 12 and 24 months. The occurrence of a YMDD variant does not entirely explain these three-phase variations. The first phase impact on fibrosis rates was higher in Asian patients (p= 0.01) and in patients younger than 40 yr (p< 0.001).
CONCLUSIONS: In patients with chronic hepatitis B, a 24-month course of lamivudine treatment leads to a significant decrease in necroinflammatory grades and fibrosis stages as assessed by noninvasive markers, with the occurrence of a three-phase kinetics. FT-AT should be useful in the noninvasive follow-up of lamivudine treatment.
AUTHOR DISCUSSION
In this large multicenter cohort study of patients with chronic hepatitis B starting lamivudine therapy, we used noninvasive markers to estimate the histological impact of the first 24 months of treatment.
The use of biomarkers has many advantages and few disadvantages. The main advantage is the simplicity of the estimation procedure. Such simplicity would be unachievable using the classical estimate of four liver biopsies in 24 months. The detailed study by Dienstag et al. was possible only on a small sample of 63 patients who had already been included in a randomized trial, and used only three estimates (35). Another advantage of biomarkers is the reduced variability related to time differences. There is much more time variability for biopsy estimates than for biomarker estimates. In previous studies, which described 3 yr of treatment between the baseline biopsy and the end of follow-up biopsy, the exact interval was often actually 24 wk longer, which, therefore, included time without treatment (20, 35). This time variability is probably more significant for necroinflammatory activity than for fibrosis, as the variation in grades is generally more rapid than the variation in stages.
One disadvantage of biomarkers is that they are indirect markers of histological features and have 20-30% of discordance with biopsy. However, prospective studies in chronic hepatitis C have demonstrated that many of these discordances were in fact due to sampling error from biopsies that were too small (17, 20, 25). The diagnostic values of FT-AT observed in the present multicenter study were similar (AUROC = 0.74-0.77) to those observed in patients contaminated with HCV (10-26) and in the first validation study in patients contaminated with HBV (27).
The present study with biomarkers confirms previous findings made with biopsy on a greater number of patients but also presents new findings. In patients treated with lamivudine for 2 yr, there was an improvement of histological features both for necroinflammatory activity and fibrosis. Compared with Dienstag et al.'s study (35) and using a definition of improvement as a decrease in 0.01 units for FT or AT, we observed a higher improvement rate, probably because biomarkers are more sensitive than biopsy. If results were expressed in worsening percentages, the overall results for activity were similar: 11% for Dienstag at 3 yr versus 9% at 2 yr in the present study (35). The results were also similar for fibrosis impact. In patients with baseline bridging fibrosis, Dienstag et al. observed a 30% improvement rate (12 out of 19 patients with a drop of HAI score from 3 to 0 or 1) (35). Here, we observed a 94% FT improvement rate, including 31 patients (32%) with a drop of FT values of 0.30, which is equivalent to a 2 histological stage improvement. In cirrhotic patients, significant fibrosis regression was observed in 32%, confirming the reversibility of cirrhosis. This confirms previous studies using liver biopsy in a smaller group of HBV patients (35), and FT in HCV treated patients, without the risk of biopsy sampling error (20).
It was also confirmed that there had been an overall clear and significant histological benefit in patients with YMDD-variant HBV at 2 yr, without any case of dramatic worsening. However, the histological benefit was reduced after prolonged YMDD-variant HBV. We observed that the occurrence of a YMDD-variant significantly increased the mean AT compared to patients without a YMDD-variant; this occurred as early as the 6 month follow-up and during 2 yr of treatment. This early difference in AT at 6 months was not expected, as only 8% of patients already had a detectable YMDD-variant at this date. AT is, therefore, a sensitive marker which could be useful for the management of patients treated by lamivudine. FT values were not significantly higher in patients with a YMDD-variant, but results were close to those observed in Dienstag et al.'s study (35).
The new findings from this study were first, the kinetics of the histological impact of lamivudine. There was, both for fibrosis and activity estimates, a first phase of rapid improvement in the first 6 months of treatment, followed by a second phase with a plateau between 6 and 12 months and a third phase with another improvement between 12 and 24 months.
The first phase was particularly marked in patients with baseline bridging fibrosis who had two significant independent favorable prognostic factors: younger age and Asian ethnicity. This observation suggests that long-term management (duration of treatment and follow-up) of patients with HBV could be different in patients according to age and ethnic origin.
Contrary to expectations, the plateau observed in the second phase could not be entirely explained by the occurrence of a YMDD-variant, as the plateau was also observed in patients without a YMDD-variant. The only differences were a small increase of FT-AT between 6 and 12 months in patients with a YMDD-variant, which was not observed in patients without a YMDD-variant. We have no clear explanation for this plateau in patients without a YMDD-variant. The presence of an occult variant could be one explanation, as could be an immune response.
Weaknesses of the present study were the lack of intermediate estimates and a longer follow-up. We have assessed biomarkers at four different times only, baseline, 6, 12 and 24 months. We speculate that there is perhaps a plateau between 6 and 12 months in the antifibrotic efficacy of lamivudine. Other scenario are, however, possible such as rapid early improvement, followed by more steady improvement in inflammation and, particularly, fibrosis scores. The three phase pattern should be validated in another study to eliminate an artefact. However, these findings suggest that lamivudine treatment should not be stopped too early in patients with high risk factors such as non-Asian patients or those older than 40 yr.
Another original finding is the difference in the lamivudine impact according to baseline liver injury. In patients with baseline bridging fibrosis, the impact of treatment was already significant at 6 months, a point that encourages starting treatment very early in these patients. In patients without bridging fibrosis at baseline, the effect on fibrosis was much slower, with a significant decrease observed only at 24 months. For necroinflammatory activity, the impact of treatment was also very rapid for patients with moderate or severe activity at baseline. In patients without activity or with minimal activity, the impact was much slower, with a significant decrease only observed at 24 months. This observation is interesting, as it does not support the classical view of the nonefficacy of treatment in patients with low ALT, which are considered to lack a strong spontaneous immune response. Although delayed, an improvement of necroinflammatory features was observed. However, the risk-benefit of treatment in patients without significant activity at baseline should be discussed, as the benefit in terms of necroinflammatory features was only observed in patients without a YMDD-variant.
Factors Associated with Baseline Fibrosis
This study has demonstrated for the first time in a large number of subjects that sophisticated viral characteristics, such as HBV genotype or precore mutations, were marginally associated with bridging fibrosis compared with gender and age. Most of the significance observed in univariate analysis was no longer seen when adjusted for age and sex. It is, therefore, mandatory that studies assessing the prognostic value of molecular characteristics take into account age and gender. The present study was not perfect, as it is very difficult to assess the duration of contamination in our population with mixed ethnic origin. One advantage was that in contrast with the Phase III trials, which imposed rigorous inclusion and exclusion criteria, our study recruited a consecutive sample of patients with hepatitis B infections fulfilling broad inclusion criteria, thus representative of patients presenting at specialist hepatology departments in France.
Symptoms Observed During Treatment
In terms of clinical outcome, all patients responded to lamivudine treatment with a progressive decrease in symptoms of liver disease and ALT levels during the study period, HBe-seroconversion, and reduction of HBV DNA viral load detected by PCR. This desirable therapeutic response was observed in both patients with and without HBV polymerase mutations. There were no differences in the rate of disappearance of symptoms or in the incidence of new symptoms according to the presence or absence of severe activity or bridging necrosis estimated by biopsy or FT-AT (data not shown).
ALT flares accompanied by hepatic decompensation (i.e., jaundice, liver failure) have been observed at the time of lamivudine resistance and viral breakthrough (43). These cases were mainly described in patients with severe liver disease at baseline and/or with a precore mutant infection. In our cohort study, no cases of acute exacerbation and liver disease decompensation were observed, at least during the study period.
In conclusion, our cohort study provides new information on the histological impact of lamivudine in patients with chronic hepatitis B. This began with a rapid improvement in patients with baseline bridging fibrosis during the first 6 months, particularly in Asian and young patients, followed by a plateau between the 6th and 12th month, and another phase of improvement during the second year of treatment. The emergence of a YMDD-variant was associated with higher necroinflammatory activity at 2 yr. The use of FT-AT every 6 months could be a considerable help for making treatment decisions in patients treated with lamivudine. Future protocols assessing the histological impact of other drugs, alone or in combination, should make use of these noninvasive markers.
INTRODUCTION
Chronic hepatitis B virus (HBV) infection affects 350 million individuals globally. Approximately 15-40% may develop serious complications, including end-stage liver disease and hepatocellular carcinoma (1). Patients with significant hepatic inflammation and fibrosis are at the highest risk of these complications (2). Prior to considering antiviral treatment, current guidelines recommend liver biopsy (3). This procedure provides important information regarding the severity of necroinflammatory activity and fibrosis, features potentially useful for predicting treatment response and prognosis. Unfortunately, liver biopsy is invasive (4), costly, and limited by sampling error and poor intra- and interobserver concordance (5-7).
Considering these limitations and patient reluctance to undergo liver biopsy, noninvasive predictors of histology have been studied in the last 4 yr, particularly in patients with chronic hepatitis C (8). Aspartate (AST) and alanine amino transferase (ALT) are widely used for assessing hepatitis activity, but the ideal cut-offs are unclear (9). We developed a panel of biochemical markers, FibroTest-ActiTest (FT-AT), which have repeatedly demonstrated high predictive values for fibrosis and necroinflammatory histological activity in patients with chronic hepatitis C (10-26). FT-AT was also validated in patients with chronic hepatitis B in a single study (27).
The introduction of lamivudine in 1998 represented a significant advance in the treatment of chronic hepatitis B infections (28). This drug is a nucleoside analog that prevents replication of the HBV by inhibiting the viral RNA-dependent DNA polymerase enzyme. Lamivudine has been shown to stimulate viral clearance and improve clinical status in several randomized clinical trials and to provide sustained benefit over periods up to 4 yr (1, 28, 29). The efficacy of lamivudine in the treatment of hepatitis B is, however, compromised by the development of viral resistance (30). This is due to the selection of HBV mutants containing mutations in the YMDD motif of the hepatitis B polymerase. These viral strains are present as minor species in the pretreatment viral quasi-species and thus become the dominant species due to selection pressure during drug treatment. However, these polymerase mutants have reduced viability due to impaired catalytic activity and are replaced by the wild-type virus if lamivudine treatment is stopped (31).
The histological consequences of YMDD mutations are not well established. Only a few prospective studies have been undertaken using liver biopsy to determine the progression of fibrosis and activity in patients treated with lamivudine mostly at 1 year (28, 32-34). Two studies, including one study on 63 patients (35), have investigated treatment longer than 1 year, being the histological consequences of HBV mutants resistant to lamivudine by repeated estimates of histological features during 3 yr of treatment (35, 36).
The aim of this study was to use noninvasive markers, instead of biopsy, in a prospective study of patients treated with lamivudine in order to assess the dynamic impact of treatment on fibrosis and necroinflammatory histological activity while taking risk factors, particularly the occurrence of YMDD mutations, into account.
METHODS
Subject Selection
This prospective longitudinal study was carried out in 50 hospital hepato-gastroenterology or internal medicine departments across France.
Patients were to be included in the study if they were over the age of 18 yr and presented with active chronic hepatitis B infection that the investigator decided to treat with lamivudine. Exclusion criteria included previous exposure to lamivudine, dialysis, HIV infection, immunosuppressant chemotherapy, or organ graft recipients. In addition, lamivudine was only provided if the conditions set out in the official prescribing information for the drug were fulfilled, notably with respect to pregnancy.
Study Design
Following initiation of treatment with lamivudine, patients were followed up for a total of 24 months. A blood sample was taken for virologic characterization and for biochemical markers at baseline, 6, 12, and 24 months.
At the baseline visit, data were recorded on the sociodemographic characteristics of the patient, alcohol consumption, risk factors for hepatitis B infections (intravenous drug use, at-risk sexual practices, travel to countries where hepatitis B is endemic, familial or perinatal contamination), other viral infections including hepatitis C and D, duration of hepatitis B infection, symptom presentation, and results of previous serological tests of liver biopsies and previous treatments for hepatitis B.
Estimates of Liver Injury
We used the previously validated FT-AT (Biopredictive, Paris, France; FibroSURE LabCorp, Burlington, NC) (10-26). FT-AT is a noninvasive blood test that combines the quantitative results of six serum biochemical markers (alpha2-macroglobulin, haptoglobin, gamma glutamyl transpeptidase (GGT), total bilirubin, apolipoprotein A1, and ALT) with patients' age and gender in a patented artificial intelligence algorithm (USPTO 6631330) in order to generate a measure of fibrosis and necroinflammatory activity in the liver. FT-AT is a continuous linear biochemical assessment of fibrosis stage and necroinflammatory activity grade. It provides a numerical quantitative estimate of liver fibrosis, ranging from 0.00 to 1.00, corresponding to the well-established METAVIR scoring system (37) of stages F0-F4 and of grades A0-A3. The same conversions that had been established in HCV patients were used. Corresponding stages and grades were calculated from median scores and 95% confidence intervals (CIs) were observed in 1,270 HCV patients and 300 healthy blood donors (22, 24, 25). Among the 300 controls, the median FT value (±SE) was 0.08 ± 0.004 (95%, 0.23) and the median AT value was 0.07 ± 0.004 (95%, 0.26). Among the 1270 HCV-infected patients, the FT conversion was 0.000-0.2100 for F0; 0.2101-0.2700 for F0-F1; 0.2701-0.3100 for F1; 0.3101-0.4800 for F1-F2; 0.4801-0.5800 for F2; 0.5801-0.7200 for F3; 0.7201-0.7400 for F3-F4; and 0.7401-1.00 for F4. The AT conversion was 0.00-0.1700 for A0; 0.1701-0.2900 for A0-A1; 0.2901-0.3600 for A1; 0.3601-0.5200 for A1-A2; 0.5201-0.6000 for A2; 0.6001-0.6200 for A2-A3; and 0.6201-1.00 for A3.
Samples stored at-80°C were centralized in the reference laboratory at Pitié Salpêtrière Hospital. GGT, ALT, and total bilirubin were measured by Hitachi 917 Analyzer and Roche Diagnostics reagents (both Mannheim, Germany). alpha2-Macroglobulin, apolipoprotein A1, and haptoglobin were measured using a Modular analyzer (BNII, Dade Behring; Marburg, Germany). All coefficients of variation assays were lower than 6%.
Liver biopsies were fixed, paraffin-embedded, and stained with at least hematoxylin-eosin-safran, and Masson's trichrome or picrosirius red for collagen. At each center pathologists analyzed in a blinded fashion the biopsies using the METAVIR classification (37). Fibrosis was staged from F0 to F4: F0, no fibrosis; F1, portal fibrosis without septa; F2, portal fibrosis with few septa; F3, numerous septa without cirrhosis; and F4, cirrhosis. Necroinflammatory activity was graded from A0 to A3: A0, no activity; A1, mild activity; A2, moderate activity; and A3, severe activity.
Virology
Blood samples taken at each study visit were subjected to virological analysis in a single center (INSERM 271, Lyon) without knowledge of the clinical data. Blood samples were analyzed for viral genome centrally in a blinded fashion with respect to the clinical data. Viral DNA was extracted from serum using the QIAmp DNA Blood Mini Kit (QIAGEN, Courtaboeuf, France). Baseline HBV DNA was quantified using the Versant branched DNA kit version 3.0 (Bayer, France) with a lower limit of detection of 357 IU/ml (1 IU/ml represents 5.6 copies/ml). The HBV polymerase gene was sequenced following amplification by polymerase chain reaction (PCR) using appropriate oligonucleotide primers. The lower limit of detection of the PCR assay was 70 IU/ml (data not shown). The sequencing reaction was performed using labeled nested primers (CY5.5-POL3M and CY5.0-P4M), CY5/CY5.5-Dye Primer Kit, Long-Read TOWER sequencer, and Opengene System software (Visible Genetics, Evry, France). INNO-LiPA genotyping and Precore mutant detection kits were used, and the blot reactions were performed using AUTOLiPA kits on nitrocellulose strips.
Statistical Analysis
We performed statistical analyses on all patients with at least one sample and in patients with all four samples (sensitivity analysis). Categorical variables were compared with the chi2 test or Fisher's exact test (in cases with less than five theoretical subjects per cell), quantitative variables with the Mann-Whitney test, and ANOVA for repeated measures used Bonferroni and Tukey-Kramer multiple comparison tests (38). Multivariate analysis used logistic regression analysis. Diagnostic value was assessed by the area under the receiver operating characteristics curve (AUROC). A sensitivity analysis was performed according to the duration between biopsy and baseline serum, using the median as a cutoff. Three methods were used in the statistical comparison of liver injury dynamics: the difference between FT-AT values, the percentage of patients with FT-AT improvement, and the fibrosis rate per year as previously used for HCV modeling (39-41). A sensitivity analysis included date of birth to calculate the fibrosis progression before treatment in order to exclude bias due to inaccurate date of infection (42). As we had already observed a strong interaction between baseline fibrosis and treatment impact in a previous analysis of treatment impact in patients with chronic hepatitis C, we stratified the analysis according to the baseline fibrosis stage (bridging vs non bridging fibrosis) (20). All data were centrally analyzed using BMDP statistical software (38).
RESULTS
Patients Included
Between October 1999 and March 2001, 298 subjects were included. Fifteen (5.1%) were excluded from the analysis due to the absence of follow-up data in 11 and protocol violation in 4. In all, 274 subjects (96.8% of the analyzed population of 283 subjects and 91.9% of the included population) completed the 24-month course of treatment. A total of 916 FT were assessed in the analyzed population; 258 were done at baseline, 229 at 6 months, 227 at 12 months, and 202 at 24 months (Fig. 1). In the remaining patients (no different from included patients, data not shown), serum was not available for FT assessments.
At the time of inclusion, 55.1% of the population was treatment naive; the rest had received interferon alpha predominantly. Only 20 patients had received previous nucleoside analog antiviral drugs, principally vidarabine (12 patients) and/or famciclovir (10 patients).
At the time when biochemical markers were obtained, liver biopsy results were available in 214 subjects, at a median of 5.2 months before inclusion in the study. At the time of biopsy, 66.7% of patients had bridging fibrosis (F2-F3-F4), including 18.7% cirrhosis F4. At the time of inclusion, as estimated by biochemical markers, 57.8% of patients had bridging fibrosis, including 25.2% cirrhosis (Table 1).
Hepatitis B virus DNA was quantified at baseline in all patients. At inclusion, basic core promoter (BCP) and precore mutants were identified in the majority of subjects (81.0%). There was a slight preponderance of promoter mutations compared to stop codon mutations, with both mutations being observed in 33.6% of subjects. A total of 119 patients (42%) were HBeAg-positive and 164 (58%) were HBeAg-negative. Genotype D was the most frequent genotype, followed by genotype A. Genotypes F and G were very rare.
Validation of Biochemical Markers
Using the previous biopsy as "a gold standard," the AUROC for the diagnosis of bridging fibrosis (METAVIR stages F2-F3-F4) was 0.77 (SE = 0.03); for many septa (F3) or cirrhosis (F4), 0.74 (SE = 0.03); for the diagnosis of moderate or severe activity (METAVIR grades A2, A3), 0.62 (SE = 0.04); and for severe activity (A3), 0.75 (SE = 0.06). When AUROCs were compared according to the duration between biopsy and biomarkers measurements (less vs more than 4 months), there was no significant difference for FT (for F2-F3-F4): 0.76 (0.05) versus 0.79 (0.05) (p= 0.53) and a significance for AT (for A2A3): 0.69 (0.06) versus 0.52 (0.06) p= 0.04.
Six patients were at high risk of having false positive results for baseline FT due to possible hemolysis (haptoglobin <0.11 g/l); 1 patient was at risk of having a false negative result due to acute inflammation (A2M 6.22 g/l, haptoglobin 1.95 g/l).
Factors Associated with Fibrosis at Baseline
In univariate analysis, bridging fibrosis (METAVIR stages F2-F3-F4) was associated with age, male gender, BMI, alcohol consumption, baseline activity grade, anti-HBe presence in the serum, and genotypes A, D, E (Table 2).
In multivariate analysis, baseline bridging fibrosis was highly associated (p< 0.001) with male gender (odds ratio (OR) = 0.14, 95% CI 0.06-0.32) and age (OR = 0.93, CI = 0.91-0.96), and marginally associated with anti-HBe presence (OR = 0.53, CI = 0.28-1.00; p= 0.05) and non-Asian ethnic origin (OR = 2.31, CI = 1.01-5.30; p= 0.046).
Evolution of Fibrosis and Histological Activity During Treatment
The values of FT-AT before and during treatment are shown in Table 3. Lamivudine treatment had a very significant impact overall. FT decreased significantly from 0.51 at baseline to 0.37 at 24 months (p< 0.001), and 85% of patients had an improvement at 24 months. AT also decreased significantly from 0.56 to 0.13 (p< 0.0001), and 91% of patients had an improvement at 24 months. The mean fibrosis progression rate was 0.15 FT units per year (0.02; n = 254) before treatment,-0.12 (0.02; n = 217) during the first 6 months, 0.04 (0.02; n = 187) between 6 and 12 months, and-0.09 (0.02; n = 166) between 12 and 24 months (the progression rates before treatment and between 6 and 12 months were significantly different vs all) (Table 4). A total of 44 patients had at baseline a cirrhosis estimated by FT and a second FT assessment at 24 months; 95% (42/44) had an improvement of FT at 24 months, including a significant regression of fibrosis greater than 0.30 in 32% (14/44).
The impact of treatment on fibrosis was greater during the first 6 months, with a plateau between 6 and 12 months and a greater impact between 12 and 24 months. The impact of treatment on necroinflammatory activity was marked at 6 months (87% improved), with a plateau between 6 and 12 months and another impact between 12 and 24 months.
Factors Associated with Lamivudine Impact
The impact was greater in patients with bridging fibrosis at baseline. The significant impact of treatment was observed both in patients with and without an occurrence of YMDD mutation.
The impact on AT was lower at 12 and 24 months in patients with YMDD mutation than in patients without (Table 3).
During the course of the study, YMDD mutations were observed in 114 patients. This corresponds to an incidence rate of 41.5% (95% CI limits 35.6% and 47.3%). The time course for the emergence of mutations was linear, with 17 (6%) occurring in the first 6 months, 47 (17%) between 6 and 12 months, and 51 (18%) between 12 and 24 months. The FT changes were from 0.41 ± 0.07 to 0.33 ± 0.07 (NS) in the group of mutation occurrence in the first 6 months; from 0.59 ± 0.04 to 0.43 ± 0.05 (p< 0.0001) in the group of mutation occurrence between 6 and 12 months; from 0.46 ± 0.04 to 0.36 ± 0.04 (p< 0.0001) in the group of mutation occurrence after 12 months. In 44 patients with baseline cirrhosis, the only 2 patients with worsening FTs were those with YMDD-variant HBV (2/20); there was a significant FT reduction in 25% of these patients (5/20).
The AT changes were from 0.53 ± 0.10 to 0.10 ± 0.03 (p= 0.04) in the group of mutation occurrence in the first 6 months; from 0.59 ± 0.04 to 0.20 ± 0.03 (p< 0.0001) in the group of mutation occurrence between 6 and 12 months; from 0.55 ± 0.04 to 0.19 ± 0.04 (p< 0.0001) in the group of mutation occurrence after 12 months.
The fibrosis progression rates during the three phases according to the baseline fibrosis and YMDD mutation are shown in Table 4. In the group of patients with baseline bridging fibrosis, very significant differences in the kinetics of the first 12 months of treatment were observed in Asian patients, who had a greater decrease in fibrosis rate during the first month compared with non-Asian patients (p= 0.01) (Fig. 2A), and patients younger than 40 yr (p< 0.001) (Fig. 2B), both still significant in multivariate analysis (p= 0.03 and p= 0.04, respectively). Kinetics were not different according to gender or anti-HBe status (data not shown).
Sensitivity Analyses
The same results were obtained when the population of patients with all completed FT or AT at the four follow-up dates was used (Table 3 panel B). When patients were stratified according to stage or grade as estimated by liver biopsy instead of FT-AT, the results were similar (data not shown). When baseline fibrosis progression rates were calculated according to the date of birth instead of the date of contamination, results were similar (data not shown).
|
|
| |
| |
|
|
|
