Liver Toxicity in Epidemiological Cohorts

Transcription

1 SUPPLEMENT ARTICLE Liver Toxicity in Epidemiological Cohorts Stephen Becker Pacific Horizon Medical Group, San Francisco, California Hepatotoxicity has been demonstrated to be associated with antiretroviral therapy. Previous studies have included small numbers of patients and, thus, were unable to produce adequate statistical comparisons. I review data analyses from the Amsterdam, CHORUS, ICONA and Target studies (5133 patients), which were conducted by a number of investigators. There were differences between the cohorts with respect to the incidence of viral hepatitis and definitions of hepatotoxicity used. However, in all cohorts, hepatotoxicity in human immunodeficiency virus type 1 infected patients was significantly associated with coinfection with viral hepatitis. In 3 cohorts, elevated baseline alanine aminotransferase levels predicted subsequent hepatotoxicity. Overall, there was a low incidence of long-term hepatotoxicity in these cohorts and no consistent association between a particular drug or drug class. Nevirapine use within the first 12 weeks after initiation of therapy with this drug and ritonavir use are associated with increased risk of antiretroviral-associated hepatotoxicity. Coinfection with viral hepatitis and all antiretroviral drug classes used in the treatment of HIV infection have been associated with asymptomatic elevations of aspartate aminotransferase and/or alanine aminotransferase (AST/ALT) levels. Severe and, in some cases, lifethreatening liver toxicity has been reported [1 10]. In triple-combination antiretroviral therapy (HAART) regimens, it is difficult to assess the hepatotoxic effect of a single agent in HIV-infected patients who have other diseases or conditions that are themselves associated with liver toxicity. Previous studies have examined the association of hepatitis B virus (HBV) and/or hepatitis C virus (HCV) infection with elevated liver enzyme levels and liver toxicity in HIV-infected patients being treated with HAART and have found a significantly greater risk of hepatic events in these coinfected patients [2, 11]. These previous studies, and others, have lacked the large number of patients required to provide robust statistical analyses. Variables such as age, sex, nonprescription drug or alcohol abuse, CD4 cell count nadir or change, plasma HIV-1 RNA levels at baseline, increased baseline liver function test (LFT) values, and treatment history may individually or in concert affect any conclusions that can be drawn. Even the definition of hepatotoxicity has differed among study cohorts. Some investigators have used the measure of increased AST/ALT levels to indicate hepatotoxicity, whereas others have only included data from patients who have presented with clinical events, such as clinical hepatitis. To learn more about the effect of different parameters associated with hepatotoxicity in the various treatment regimens of HIV-infected patients, we have reviewed the data from 4 observational cohort databases: the Amsterdam, Collaborations in HIV Outcomes Research US (CHORUS), Italian Cohort of Naive for Antiretrovirals (ICONA), and Target cohorts. For each of these, data concerning hepatic toxicity have been analyzed by the various authors. The present article, which is not a meta-analysis, summarizes the findings of each of these analyses. AMSTERDAM COHORT Reprints or correspondence: Stephen Becker, Pacific Horizon Medical Group, Inc., 2351 Clay St., Ste. 512, San Francisco, CA (slbecker@mindspring.com). Clinical Infectious Diseases 2004; 38(Suppl 2):S by the Infectious Diseases Society of America. All rights reserved /2004/3805S2-0003$15.00 The Amsterdam cohort is part of a Dutch, multicenter, clinical ATHENA cohort of HIV-infected individuals who have been treated with 1 antiretroviral drug [12]. A study of 560 HIV-infected patients from the Am- Liver Toxicity in Epidemiological Cohorts CID 2004:38 (Suppl 2) S49

2 Table 1. Antiretroviral drug use among patients in the Amsterdam cohort. Drug Patients ever treated, % Cumulative use, years Lamivudine Stavudine Zidovudine Didanosine Ritonavir Saquinavir hard gel Indinavir Nelfinavir Nevirapine sterdam cohort was designed to determine whether the use of any licensed antiretroviral drug was associated with an increased risk of developing grade IV liver enzyme elevations (LEEs) [13]. The cohort consisted of patients from an HIV outpatient clinic and included some clinical trial participants. One-fifth (22%) of the cohort was female. All patients had begun potent combination antiretroviral therapy ( 3 antiretroviral drugs), and 48% had been previously treated with nucleoside reverse-transcriptase inhibitors (NRTIs). The median duration of followup was 3.0 years (interquartile range, years). Baseline AST/ALT levels were obtained within 26 weeks before the start of antiretroviral therapy; most samples (92.3%) were obtained within 4 weeks of beginning therapy. Hepatotoxicity was defined according to the grading system used by AIDS Clinical Trial Group [14] but was modified by taking into account the AST/ALT levels at baseline: grade III/IV, 5 times the upper limit of normal (ULN) and 1100 IU/L above baseline; grade IV, 110 times the ULN and 1200 IU/L above baseline. Patients were considered to have chronic HBV infection if HBV surface antigen (HBsAG) was detected in plasma samples on 2 consecutive occasions at least 6 months apart before the start of antiretroviral therapy. A patient was considered to have HCV infection if HCV antibodies were detected in serum. The Table 2. Characteristic charts of all patients with grade IV hepatic events were reviewed, and non drug-related events were excluded. Of importance, certain antiretroviral drugs (zalcitabine, abacavir, saquinavir soft gel capsules, amprenavir, lopinavir, efavirenz, and hydroxyurea) were used by!10% of the study patients for a variety of reasons for example, lopinavir and efavirenz were not extensively used for policy reasons. This cohort analysis, therefore, had limited power to detect hepatotoxicity associated with these drugs. Results. Table 1 depicts the extent and duration of antiretroviral drug use in the Amsterdam cohort. Forty-four patients developed grade IV LEEs, 9 of which were identified as being not drug related on the basis of the patients medical records. Six of 35 drug-related LEEs were symptomatic, and 5 of these occurred in patients who were coinfected with either HBV or HCV. Of these 35 patients, 23 continued antiretroviral therapy. Nineteen patients were receiving their first ever HAART regimen. LEEs resolved to grade II or less in all cases during a median period of 8.9 weeks. The baseline characteristics of the 35 patients who developed grade IV LEEs and of the 525 patients who did not are shown in table 2. The patients who experienced grade IV LEEs were more likely to be female, with either HBV or HCV coinfection, and to have higher pretreatment (baseline) AST/ALT values. Neither plasma HIV-1 RNA levels nor CD4 cell counts were significantly associated with grade IV LEE. A multivariate analysis of the data from the Amsterdam cohort produced the results summarized in table 3. HBV or HCV coinfections, recent use (within 12 weeks of initiation) of nevirapine or ritonavir (!400 mg/day) use, full-dose ritonavir at any time, and HBV rebound after lamivudine discontinuation were all significantly associated with grade III or IV LEEs. First HAART use and female sex were significantly associated with grade IV LEEs. Conclusions. Drug-related grade IV LEE occurred in 6.3% of patients in the Amsterdam cohort. Factors associated with hepatotoxicity included HCV or HBV coinfection, elevated baseline ALT levels, first receipt of HAART regimen, female sex, Baseline characteristics of patients in the Amsterdam cohort. Grade IV (n p 35) No grade IV (n p 525) P CD4 cell count, cells/mm HIV RNA load, log 10 copies/ml AST level, IU/L ALT level, IU/L HBV +, % of patients !.0001 HCV +, % of patients !.0001 Female sex, % of patients NOTE. ALT, alanine aminotransferase; AST, aspartate aminotransferase; HBV, hepatitis B virus; HCV, hepatitis C virus. S50 CID 2004:38 (Suppl 2) Becker

3 Table 3. Results of multivariate analysis for the Amsterdam cohort. Hazard ratio (95% CI) Parameter Grade IV Grade III/IV Female sex 2.8 ( ) NS ALT level, 10 U/L change 1.05 ( ) 1.03 ( ) HCV infection 5.0 ( ) 3.2 ( ) HBV infection 9.2 ( ) 4.6 ( ) Recent nevirapine therapy 9.6 ( ) 2.8 ( ) Recent ritonavir therapy for!12 weeks 4.9 ( ) Ritonavir use at any time 1.9 ( ) First receipt of HAART regimen 2.8 ( ) Recent discontinuation of lamivudine therapy 6.8 ( ) 3.5 ( ) NOTE. receipt of first 12 weeks of nevirapine or low-dose ritonavir therapy, receipt of full-dose ritonavir therapy, and the discontinuation of lamivudine therapy in HBV-infected patients. Further research is required to determine the reasons why these risk factors are important, especially the apparent increased vulnerability of women to the hepatotoxic effects of antiretroviral drugs. CHORUS COHORT The CHORUS database was started in the fall of 1997 and now contains data on patients [15]. From this database, a total of 1120 patients with all baseline data available were identified as receiving their initial regimen of HAART. HBV infection was defined by the presence of HBsAg or the presence of HBV DNA. HCV infection was defined by the presence of HCV antibody or the presence of HCV RNA. Onehalf of the patients (49.9%) were receiving a protease inhibitor (PI) with an NRTI. One-fourth (25.4%) were receiving NRTIs plus a nonnucleoside reverse-transcriptase inhibitor (NNRTI); a smaller proportion of patients were receiving 3 NRTIs ALT, alanine aminotransferase; HBV, hepatitis B virus; HCV, hepatitis C virus. (17.1%) or triple-class (PI, NRTI, and NNRTI) therapy (7.5%). In this analysis, the incidence of grade II IV AST/ALT elevations was used as the subclinical event marker. Clinical events, including the discontinuation of HAART, were also noted. Data on alcohol consumption were only collected at one time point in the study, so conclusions about this risk factor are limited. Further studies on the effect of alcohol consumption on AST/ ALT elevations and hepatic events would be informative. Results. Figure 1 illustrates the incidence of hepatotoxicity per 100 person-years in the CHORUS cohort. The incidence of hepatotoxicity was approximately the same for all antiretroviral regimens. Patients in the NNRTI/NRTI group experienced a lower incidence of grade III/IV AST/ALT elevations, but this was not statistically significant. The overall rate of grade III/IV hepatotoxicity was 5.5%. No patient received a diagnosis of clinical hepatitis. The low rate of incidence of hepatotoxicity precluded modeling with control for multiple risk factors; thus, the definition of hepatotoxicity was modified to include all grade II (12.5 times ULN) to grade IV elevations in AST/ALT levels. Table 4 shows the predictors of grade II IV LEEs in patients in this database. Increased LFT results at baseline, HBV Figure 1. Incidence of hepatotoxicity per 100 person-years (%) in the CHORUS cohort. NNRTI, nonnucleoside reverse-transcriptase inhibitor; NRTI, nucleoside reverse-transcriptase inhibitor; PI, protease inhibitor. Liver Toxicity in Epidemiological Cohorts CID 2004:38 (Suppl 2) S51

4 Table 4. Predictors of grade II IV aspartate aminotransferase or alanine aminotransferase elevations in the CHORUS cohort. Predictor OR 95% CI P Type of HAART Baseline increased LFTs !.0001 HBV or HCV coinfection Concurrent hepatitis !.0001 NOTE. tests. or HCV coinfection, and concurrent hepatitis were all statistically significantly positive predictors of elevations in LFT results. No class or specific antiretroviral drug was associated with hepatotoxicity in a statistically significant manner in the present analysis. Conclusions. Laboratory hepatotoxicity occurred at a low rate with all of the regimens used by the patients initiating HAART in the CHORUS cohort. The incidence of hepatotoxicity was similar for all antiretroviral regimens. Concurrent hepatitis, coinfection with HBV or HCV, and baseline increased LFT results at baseline were significant predictors of AST elevations in HIV-infected patients in this cohort. ICONA COHORT The ICONA cohort consisted of 4656 patients; the prevalence of HCV coinfection was 17%. Data from 1255 patients who had started a first HAART regimen including at least 2 NRTIs and 1 PI or 1 NNRTI were analyzed [5]. Inclusion criteria required an ALT level of!200 IU/L at baseline, at least 1 measurement of the ALT level during therapy, and known HBV and HCV status. In this group of patients, there were very few who received NNRTIs (0.4%). Almost all patients had received PIs (99.6%). The end points were ALT values of 200 IU/L and the time to discontinuation of HAART because of hepatotoxicity. Results. In this study, 61 patients (4.9%) developed ALT levels of 200 IU/L and discontinued HAART because of hepatotoxicity. The overall probability of reaching an end point at 24 months was 7.9%. If the patient was HBV and HCV negative, Table 5. HBV, hepatitis B virus; HCV, hepatitis C virus; LFTs, liver function Risk factors for hepatotoxicity in the ICONA cohort. Parameter Hazard ratio 95% CI P Elevated baseline ALT level !.01 HCV infection !.01 HBV infection d4t- vs. ZDV-containing regimen !.01 the probability was 1.5%. If the patient was HBV positive, the probability rose to 8.0%; if the patient was HCV positive, the probability was 14.5%. If the patient was coinfected with both HBV and HCV, the probability of reaching an end point rose to 23%. Patients who were negative for both HBV and HCV infection had a much lower probability of severe hepatotoxicity than those who were coinfected with HBV, HCV, or both ( P!.01). Table 5 summarizes the risks for hepatotoxicity as identified in the ICONA study. Increased hazard ratios were demonstrated for patients with elevated baseline ALT levels or those infected with HBV or HCV. The presence of stavudine instead of zidovudine in a regimen reduced the risk of hepatotoxicity; however, the study was not designed to address this issue, and no conclusions about the backbone NRTI drugs could be drawn. All hazard ratios were statistically significant, except for the presence of HBV. Conclusions. Severe hepatotoxicity occurred rarely in the ICONA cohort. Coexisting HCV, but not HBV, infection and elevated baseline ALT levels were important predictors of hepatotoxicity. TARGET COHORT The Target cohort included data obtained from 10 clinical practice sites using an integrated electronic medical record system [16]. At the time of analysis, the cohort included 2198 HAARTtreated patients without distinction as to whether these patients were receiving first or subsequent HAART regimens [17]. Base- NOTE. ALT, alanine aminotransferase; HBV, hepatitis B virus; HCV, hepatitis C virus; d4t, stavudine; ZDV, zidovudine. S52 CID 2004:38 (Suppl 2) Becker

5 Table 6. Patient characteristics of the Target cohort ( N p 2198). Characteristic Percentage of patients Sex Male 90 Female 10 Age, years! HBV or HBC infection 23 CD4 count of 350 cells/mm 3 43 Race Black 15 White 21 Hispanic 28 Other 36 NOTE1 virus. HBV, hepatitis B virus; HBC, hepatitis C line AST/ALT levels were measured within 12 months of treatment initiation. Coinfection with HCV was established by the presence of HCV antibody. Similarly, a diagnosis of HBV infection was made if HBV antigen was detected. Patient demographics are shown in table 6. The majority of patients in this cohort were male (90%); Figure 2. Kaplan-Meier plot of elevated liver function test (LFT) values in patients receiving HAART in the Target cohort. NNRTI, nonnucleoside reverse-transcriptase inhibitor; NRTI, nucleoside reverse-transcriptase inhibitor; PI, protease inhibitor; RR, relative risk. 46% were aged years; and 23% were coinfected with either HBV or HCV. Slightly more than one-half (57%) had a CD4 cell count of!350 cells/mm 3. Results. Using LEEs 15 times the ULN as an end point, the Kaplan-Meier plot in figure 2 presents the time to elevated LFT results over a 2-year period in patients who received various HAART regimens. With PIs representing a relative risk (RR) of 1, NRTIs had an RR of 1.11 (95% CI, ), and NNRTIs had an RR of 1.46 (95% CI, ). These differences did not achieve statistical significance. Figure 3 presents the relative risk of elevated LFT results, defined as levels of 10 times the ULN. Ritonavir was the only Figure 3. Relative risk of elevated liver function test (LFT) values ( 10 times the upper limit of normal [ULN]) in the Target cohort. 3TC, lamivudine; ABV, abacavir; APV, amprenavir; d4t, stavudine; ddi, didanosine; EFV, efavirenz; HBV, hepatitis B virus; HCV, hepatitis C virus; IDV, indinavir; NFV, nelfinavir; NNRTI, nonnucleoside reverse-transcriptase inhibitor; NRTI, nucleoside reverse-transcriptase inhibitor; NVP, neviparine; PI, protease inhibitor; RTV, ritonavir; SQV, saquinavir; ZDV, zidovudine. Liver Toxicity in Epidemiological Cohorts CID 2004:38 (Suppl 2) S53

6 Table 7. Comparison of cohorts. Cohort No. of patients Time period Length of follow-up, months Hepatitis coinfection, % of patients HBV HCV Incidence of LEE Hazard Amsterdam % grade IV Female, HBV or HCV coinfection, elevated baseline ALT level CHORUS % grade III or IV ICONA % ALT level 1200 IU/L Coinfection, elevated baseline ALT level HCV coinfection, elevated baseline ALT level Target ND Age 160 years, coinfection NOTE. ALT, alanine aminotransferase; grade III liver enzyme elevation (LEE), 15 times the upper limit of normal (ULN); grade IV, 110 times the ULN; HBV, hepatitis B virus; HCV, hepatitis C virus; ND, not done. drug to have an RR of Other factors with a greater RR included age of 160 years and coinfection with HBV or HCV. The relatively infrequency of grade III IV events meant that subgroup analyses to detect subtle differences between patient groups were not possible. Conclusions. The analysis of data from the Target cohort failed to demonstrate a clear drug class effect or statistically significant differences between individual drugs within the NRTI, NNRTI, or PI classes. Age of 160 years and coinfection with HBV or HCV emerged as the most robust predictors of hepatotoxicity. COHORT COMPARISONS Table 7 compares the characteristics and outcomes of the 4 cohort analyses. All of the studies included data from patients followedup from the mid-1990s to There is a clear discrepancy between the cohorts in terms of the presence of viral hepatitis. The ICONA cohort patients were more likely to be coinfected with HCV than were patients in the other cohorts; HCV and HBV infections were not differentiated in the Target cohort. The CHORUS cohort patients were more likely to be coinfected with HBV than were patients in the other cohorts. These demographic differences aside, the conclusions from each cohort analysis are quite similar. All cohorts demonstrated an increased risk of hepatotoxicity in patients who were coinfected with viral hepatitis. In 3 of the cohorts, an elevated baseline ALT level was a predictor of later elevations of AST/ALT levels. Overall, there was a fairly low incidence of hepatotoxicity in these cohorts, even accounting for the differences in the definitions of hepatotoxicity used. There was no consistent association between a particular drug or drug class and the development of subsequent hepatotoxicity. The data suggest that only ritonavir and the recent use of nevirapine (within 12 weeks of initiating therapy) are associated with an increased risk of antiretroviral-associated hepatotoxicity. References 1. Sulkowski MS, Thomas DL, Chaisson RE, et al. Hepatotoxicity associated with antiretroviral therapy in adults infected with human immunodeficiency virus and the role of hepatitis C or B virus infection. JAMA 2000; 283: den Brinker M, Wit FWNM, Wertheim-van Dillen PME, et al. Hepatitis B and C virus co-infection and the risk for hepatotoxicity of highly active antiretroviral therapy in HIV-1 infection. AIDS 2000; 14: Martinez E, Blanco JL, Arnaiz JA, et al. Hepatotoxicity in HIV-1 infected patients receiving nevirapine-containing antiretroviral therapy. AIDS 2001; 15: Nuñez M, Lana R, Mendoza JL, et al. Risk factors for severe hepatic injury after introduction of highly active antiretroviral therapy. J Acquir Immune Defic Syndr 2001; 27: Monforte AA, Bugarini R, Pezzotti P, et al. Low frequency of severe hepatotoxicity and associate with HCV coinfection in HIV-positive patients treated with HAART. J Acquir Immune Defic Syndr 2001; 28: Pilero PJ, Purdy B. Nevirapine-induced hepatitis: a case series and review of the literature. AIDS Reader 2001; 7: Palmon R, Koo BCA, Shoultz DA, et al. Lack of hepatotoxicity associated with nonnucleoside reverse transcriptase inhibitors. J Acquir Immune Defic Syndr 2002; 29: Sulkowski MS, Thomas DL, Mehta SH, et al. Hepatotoxicity associated with nevirapine or efavirenz-containing antiretroviral therapy: role of hepatitis C and B infections. Hepatology 2002; 35: Stern JO, Love JT, Lanes SF, et al. Hepatic safety of nevirapine: results of the Boehringer Ingelheim Viramune hepatic safety project [abstract LBOr15]. XIV International AIDS Conference (Barcelona, Spain) Reisler R, Liou S, Servoss J, et al. Incidence of hepatotoxicity and mortality in 21 adult antiretroviral treatment trials [abstract 43]. First IAS Conference on HIV Pathogenesis and Treatment (Buenos Aires) Rodriguez-Rosado R, Garcia-Samaniego J, Soriano V. Hepatotoxicityafter introduction of high active antiretroviral therapy. AIDS 1998; 12: Dieleman JP, Jambroes M, Gyssens IC, et al. Determinant of recurrent toxicity-driven stitches of high active antiretroviral therapy: the ATHENA cohort. AIDS 2002; 16: Wit FWNM, Weverling GJ, Weel J, et al. Incidence of and risk factors for severe hepatotoxicity associated with antiretroviral combination therapy. J Infect Dis 2002; 186: AIDS Clinical Trials Group. Table of grading severity of adult adverse experiences. Rockville, MD: Division of AIDS, National Institute of Allergy and Infectious Diseases, Justice AC, Wagner JH, Fusco GP, et al. HIV survival: liver function tests independently predict survival [abstract MoOrB1058]. XIV International AIDS Conference (Barcelona) S54 CID 2004:38 (Suppl 2) Becker

7 16. Dietrich DT, Becker SL, Fusco JS, et al. Low incidence of grade III/IV hepatotoxicity in first HAART: observations from 1100 patients followed for 1 year [abstract TuPeB4534]. XIV International AIDS Conference (Barcelona) Imperiale SM, Lanes SF, Stern JO, et al. TARGET: incidence of elevated ALT/AST with HAART in a large observational cohort [abstract P150]. 6th International Congress on Drug Therapy in HIV Infection (Glasgow) Liver Toxicity in Epidemiological Cohorts CID 2004:38 (Suppl 2) S55