| |
Pharmacogenetics of Nevirapine-Associated Hepatotoxicity: An Adult AIDS Clinical Trials Group Collaboration
|
| |
| |
Clinical Infectious Diseases Sept 15, 2006;43:783-786
David W. Haas,1 John A. Bartlett,2 Janet W. Andersen,4 Ian Sanne,6 Grant R. Wilkinson,1 John Hinkle,3 Franck Rousseau,3 Christiana D. Ingram,1 Audrey Shaw,3 Michael M. Lederman,5 Richard B. Kim,1 and the Adult AIDS Clinical Trials Group
1Vanderbilt University School of Medicine, Nashville, Tennessee; 2Duke University School of Medicine, and 3Gilead Sciences, Durham, North Carolina; 4Harvard School of Public Health, Boston, Massachusetts; 5Case Western Reserve University School of Medicine, Cleveland, Ohio; and 6University of Witwatersrand, Johannesburg, South Africa
(See the article by Ritchie et al.)
Associations have been reported between an MDR1 variant and responses to nonnucleoside reverse-transcriptase inhibitors. We explored associations between MDR1, CYP2B6, and CYP3A polymorphisms and nevirapine hepatotoxicity. Among participants in a randomized study in South Africa (FTC-302), MDR1 3435C T was significantly associated with decreased risk of hepatotoxicity (risk ratio, 0.30; P = .016).
The most frequently prescribed initial regimen for HIV-1 infection is a nonnucleoside reverse-transcriptase inhibitor (NNRTI; either nevirapine or efavirenz) plus 2 nucleoside analogues. Unfortunately, severe hepatotoxicity during the initial weeks of therapy with nevirapine, particularly when initiated in women with CD4+ counts >250 cells/mm3, have prompted changes in prescribing guidelines. Reliable genetic predictors of nevirapine toxicity might allow HIV therapy that includes nevirapine to be initiated for persons with higher CD4+ counts.
Cellular transport of HIV protease inhibitors is affected by P-glycoprotein, the multidrug efflux pump encoded by MDR1 (also called ABCB1), but it is uncertain whether nevirapine or efavirenz are P-glycoprotein substrates [1]. Therefore, it is difficlt to explain the reported associations between MDR1 polymorphisms and NNRTI treatment response. One report [2] suggests an association between MDR1 3435C T and better virologic responses to efavirenz-containing regimens, despite lower plasma drug concentrations. Another study also found this association between MDR1 3435 TT genotype and better virologic responses, although not with CD4+ change or plasma efavirenz exposure [3]. Mechanisms underlying genotype-phenotype associations for MDR1 3435C T are uncertain, because they do not alter the amino acid sequence. This polymorphism may reduce MDR1 mRNA stability [4] or may be linked to other functional polymorphisms [5]. However, in an analysis involving 39 MDR1 polymorphisms, only 3435C T predicted cellular disposition of nelfinavir [6].
Both nevirapine and efavirenz are metabolized primarily by hepatic cytochrome P450 (CYP) 2B6, and associations have been identified between a frequent CYP2B6 variant (516GT) and NNRTI pharmacokinetics. In analyses involving 494 participants from prospective clinical trials, greater plasma efavirenz exposure was predicted by CYP2B6 516G T [3, 7]. Recent data suggest that 516G T also predicts nevirapine exposure [8].
We focused on polymorphisms previously associated with functional effects [3, 5], exploring associations between genetic variants in MDR1, CYP2B6, CYP3A4, and CYP3A5 and nevirapine-associated hepatotoxicity among participants in a prospective, randomized clinical trial in South Africa.
Materials and methods. We analyzed data and specimens from participants who initiated nevirapine during the Gilead study protocol FTC-302. This randomized, double-blind trial enrolled antiretroviral-naive participants in South Africa in 1999 and 2000 to compare the efficacy and safety of emtricitabine and lamivudine [9]. Eligible participants had plasma HIV-1 RNA levels of >5000 copies/mL and CD4+ cell counts >200 cells/mm3. Participants were stratified to receive either nevirapine (200 mg once daily for 2 weeks, then 200 mg twice daily) or efavirenz (600 mg once daily). Because of limited data regarding nevirapine efficacy in participants with plasma HIV-1 RNA levels >100,000 copies/mL, participants with <100,000 copies/mL were stratified to receive nevirapine, whereas participants with higher viral loads received efavirenz. All participants were also randomized 1 : 1 to receive emtricitabine (200 mg once daily) or lamivudine (150 mg twice daily), and all received stavudine (30 or 40 mg twice daily, depending on body weight). Only nevirapine recipients were included in the present analysis. Hepatotoxicity was defined as a treatment-emergent grade 3 (5-10 times the upper limit of the normal range) or grade 4 (>10 times the upper limit of the normal range) elevation in the level of serum aminotransferase (alanine aminotransferase or aspartate aminotransferase), with or without associated symptoms. To preserve anonymity, the study sponsor matched each case patient with 2 controls for race (black vs. other), sex, age, baseline CD4+ cell count, and baseline HIV-1 RNA level. Fever and rash were not exclusionary. Drug assays were not performed. This study complied with the Helsinki Declaration. Institutional review boards at Vanderbilt University, Duke University, and the University of Witwatersrand approved the use of DNA.
Cryopreserved PBMCs were subjected to robotic DNA extraction. Whole genome amplification (OmniPlex; Rubicon Genomics) was performed on specimens with scant yields. Genotyping was performed by oligonucleotide ligation (CYP2B6 516G T, CYP2B6 1459C T, CYP3A4 -392A G, CYP3A5 14690G A, MDR1 2677G T/A, and MDR1 3435C T) or real-time PCR (CYP3A5 6986A G). Genotyping methods are available upon request.
Conditional logistic regression was used to investigate the association of genotype with hepatotoxicity using the SAS procedure PHREG, which allows for 1 : M matching and uses information on incomplete case patient and control groups. Genotypes were coded as ordered categorical variables. The strength of association is expressed as the risk (hazard) ratio. There was no adjustment for multiple comparisons. Inferred haplotypes of MDR1 and CYP3A5 were reconstructed by Bayesian and likelihood methods. Only data collected while participants were receiving nevirapine were included.
Results. Among 385 nevirapine recipients who were participating in Gilead study protocol FTC-302, 66 (17%) experienced hepatotoxicity, which occurred within the first 12 weeks of therapy in 80% of patients [9]. Of these 66 case patients, 63 had PBMCs available, of whom 53 (80%) had adequate DNA available for analysis after chromosomal amplification. A total of 126 controls were identified. Of these, 108 controls (86%) had specimens with adequate DNA for analysis after amplification. Baseline characteristics of case patients and controls and frequencies of genetic variants are shown in table 1. Small differences in demographic characteristics between case patients and controls are the result of incomplete matching triads.
As expected, there was a strong link between CYP3A4-392A G and CYP3A5 6986G A and between MDR1 2677G T and 3435C T [3]. By univariate analysis, participants who experienced hepatotoxicity were less likely to have at least one MDR1 T allele at position 3435CT (P = .016; risk ratio, 0.3). Genotypes at MDR1 position 3435 in case patients and controls were CC in 83% and 67%, CT in 13% and 28%, and TT in 4% and 6%, respectively. Case patients with hepatotoxicity also tended to have at least 1 CYP3A5 G allele at position 6986A G (risk ratio, 1.64; P = .11). By step-up/step-down multicovariate analysis that included all 7 polymorphisms, only MDR1 3435C T was significantly associated with reduced risk of nevirapine hepatotoxicity (P = .016). No other MDR1, CYP2B6, or CYP3A polymorphism was significantly associated with risk of hepatotoxicity, when MDR1 3435C T was adjusted for. Additionally, neither MDR1 nor CYP3A5 haplotypes were predictive after adjustment for MDR1 3435C T.
Discussion. Hepatotoxicity in study protocol FTC-302 and, therefore, in the present analysis, was almost certainly caused by nevirapine, because grade 3 or 4 elevations in hepatic transaminase levels affected 17% of nevirapine recipients but 0 of 83 efavirenz recipients [9]. Here, we report that the MDR1 position 3435 T allele was associated with decreased risk of hepatotoxicity during combination antiretroviral therapy that included nevirapine. In contrast, we did not find an association with CYP2B6 516G T, despite a recent report [8] suggesting its association with plasma nevirapine exposure, nor did we find an association with other candidate polymorphisms included in this analysis.
P-glycoprotein is expressed in many normal human tissues, including the canalicular surface of hepatocytes and the brush border surface of enterocytes. Its tissue localization and directional action suggest that P-glycoprotein acts as a protective barrier to prevent absorption, limit tissue penetration, and promote elimination of potential toxins [5].
Despite extensive previous studies of P-glycoprotein and genetic variants of MDR1, this gene remains enigmatic. An association between MDR1 3435C T and decreased hepatotoxicity of nevirapine could be explained if this polymorphism affected hepatocyte P-glycoprotein activity or first-pass hepatic exposure through altered intestinal P-glycoprotein activity. Results of previous studies that explored associations with MDR1 3435C T and various phenotypes have been contradictory (reviewed in [5]). For example, healthy Japanese subjects with the 3435T allele had increased intestinal MDR1 mRNA expression and lower plasma levels of the P-glycoprotein substrate digoxin after oral administration. Conversely, healthy white persons had reduced intestinal P-glycoprotein expression in association with the 3435T allele, and increased plasma concentrations of digoxin after oral administration.
Of numerous identified MDR1 polymorphisms, 3435C T in exon 26 was the first to be associated with altered protein expression. Linkage disequilibrium between 3435C T and an exon 21 polymorphism suggests that the observed functional differences initially attributed to 3435C T may be caused by associated nonsynonymous polymorphism elsewhere, and multiple MDR1 haplotypes have been confirmed by several groups [5]. However in our study, risk of hepatotoxicity was not associated with MDR1 2677G T, either alone or when combined with 3435CT by haplotype reconstruction. This is important because MDR1 2677G T is nonsynonymous and has been associated with functional effects in some studies [5]. Although nevirapine is not known to be a P-glycoprotein substrate [1], a recent report that nevirapine concentrations within PBMCs correlate inversely with levels of P-glycoprotein expression suggests active transport by P-glycoprotein [10]. Little is known about nevirapine transport in hepatocytes. We hypothesize that altered P-glycoprotein efflux activity in the gut associated with MDR1 variants influences disposition of nevirapine and/or its metabolites in a way that affects intracellular drug concentrations and toxicity.
The matched case-control design used in this study minimizes effects of potential confounders and may increase power to identify genetic associations. Although this prevented us from exploring associations between the matching variables and hepatotoxicity, extensive analyses of possible nongenetic predictors of hepatotoxicity have already been reported for FTC-302. Female sex, body mass index, serum albumin concentration, plasma HIV RNA level, lactate dehydrogenase level, and aspartate aminotransferase level were significantly, independently associated with subsequent risk of hepatotoxicity in nevirapine recipients, whereas hepatitis B or C coinfection, other potentially hepatotoxic drugs, and use of traditional medication were not. The well established association of nevirapine-induced liver toxicity with high CD4+ cell count was not apparent in FTC-302, likely because few nevirapine recipients had low CD4+ cell counts [9].
Additional studies are needed to confirm the association between MDR1 and nevirapine hepatotoxicity and to identify and validate other potential predictors of nevirapine toxicity [11]. We are encouraged by another report that included several investigators from the present study that suggests the same association between MDR1 3435C T and NNRTI hepatotoxicity [12].
|
|
| |
| |
|
|
|
