Genes, Ethnicity, and Efavirenz Response: Clinical Pharmacology Update from the 11th CROI
By Adriana Andrade, M.D., M.P.H. and Charles Flexner, M.D.
The HIV Hopkins Report http://www.hopkins-aids.edu
--Race, Ethnicity and Pharmacokinetics
--The Impact of Sex, Weight and Race on the Pharmacokinetics of Antiretrovirals Agents
--Generic Antiretrovirals: How Good Are They?
--Effect of Hepatic Impairment on Tenofovir Metabolism
--Lopinavir/Ritonavir and Fosamprenavir
--Fact or Fiction: Transformation of AZT to D4T-TP
Race, Ethnicity and Pharmacokinetics
Pharmacokinetics, therapeutic response, and side effects have been shown to vary in HIV-infected patients from distinct ethnic backgrounds [Barrett JS, et al. Int J Clin Pharmacol Ther 2002;40:507; Pfister M, et al. Antimicrob Agents Chemother 2003; 47:130]. A recent study found a statistically significant association between poly-morphism in the human multidrug resistance-1 (mdr1) gene, efavirenz (EFV) plasma concentrations, and CD4 changes during treatment [Fellay J, et al. Lancet 2002 5;359:30] suggesting a role for host genomic diversity in explaining these differences. However, the results of this study are controversial, as several later studies failed to confirm this association [Flexner C, Topics HIV Med 2003;11:40]. During the 11th CROI in San Francisco, two studies investigated a possible role for genetic differences as the basis for developing EFV toxicity. Both abstracts presented data from Adult AIDS Clinical Trials Group (AACTG) Protocols 5095/5097, in which HIV-infected antiretroviral-naïve subjects were randomized to receive either efavirenz (EFV) plus zidovudine/lamivudine/abacavir (Trizivir) or Trizivir alone.
In the first study, Heather Ribaudo from Harvard discussed the findings from ACTG 5097, a sub-study of the ACTG 5095 protocol which investigated the relationship between EFV pharmacokinetic parameters, CNS side effects, weight, race, virologic response and treatment discontinuation [Abstract 132]. From the 202 subjects randomized to take an EFV-containing regimen, 81% were males (53% white non-Hispanic, 32% black-non-Hispanics, 12% Hispanics, and 3% other). The investigators found significant associations between drug clearance and weight, and between drug clearance and race. EFV clearance was 24% lower in blacks and Hispanics (9.4 L/hr) compared to whites (12.4 L/hr), while EFV area under the concentration-time curve (AUC) was 24% higher in black and Hispanics (64 mg x h/L) compared to whites (48 mg x h/L). There was a trend towards an increased rate of EFV discontinuation with decreasing EFV clearance and increasing EFV concentration, but no apparent association between EFV clearance and CNS toxicity. Analysis of virologic response is underway.
The second study evaluated the relationship between genetic variants of the cytochrome P450 2B6 (CYP 450 2B6), CYP450 3A4/5 and MDR-1 genes and EFV pharmacokinetics, CNS toxicity and therapeutic effect. EFV is primarily metabolized by the CYP2B6 and 3A4/5 pathways, and genetic polymorphisms in these genes have been described. Using real-time PCR, Haas and colleagues evaluated six allelic variants from patient DNA samples obtained from the AACTG DNA repository: CYP450 2D6 (G516T, C1459T), CYP450 3A4 (A-392G), CYP450 3A4/5 (A6989G), and mdr1 (G2677T, C3435T) [Abstract 133]. Pharmacokinetic sampling of EFV and assessment of CNS side effects were done at weeks 1, 4, 12, and 24. Of the 157 subjects included in the final data analysis, 57% were white, 32% were black and 10% were Hispanic. Median EFV AUC was signifi-cantly greater in blacks (58 µg.hr.mL-1) and Hispanics (66 µg.hr.mL-1) than in European Americans (46 µg.hr.mL-1). All 6 identified allelic variants were significantly associated with EFV plasma concentrations among all subjects. Twenty percent of the African Americans were T/T homozygous at the CYP450 2B6 516 position compared to only 3% of European Americans. The median AUC was 3-times higher with the 516 T/T genotype relative to G/G. Overall, G/G and T/T homozygotes were associated with lower and higher EFV plasma concentrations, respectively, while those who were heter-ozygous (G/T) at this locus had intermediate levels. CYP450 2B6 G516T and CYP450 3A45 A6986G were significantly associated with EFV clearance. No apparent association was found between race and clearance after adjusting for these allelic variants.
With regard to EFV-associated CNS toxicity, the CYP450 2B6 position 516 TT genotype was significantly associated with risk of CNS effects only at week 1 (P=0.036). No associations were observed between the allelic variants and immuno-logic and virologic response.
The findings from these two studies corroborate the notion that drug metabolism may be affected by racial background, and suggest that CYP450 2B6 polymorphisms may explain some of the reported differences in EFV exposure and therapeutic effect. The 516 T/T genotype, more frequently found in African Americans, was associated with higher plasma EFV concentrations, slower clearance, and increased CNS toxicity at week 1. However, the association with CNS toxicity was no longer evident by week 4, so the clinical significance of this association is unclear. Furthermore, while the associations described are statistically significant, drug concentrations overlapped substantially amongst the two genotypes. It remains to be seen whether similar associations occur with other antiretroviral agents. Though these observations need to be confirmed in other large databases, they will renew interest in the potential role of the individual genome and treatment of populations with diverse racial/ethnic backgrounds. The extent to which polymorphisms will impact tolerability and virologic suppression is still unclear.
The Impact of Sex, Weight and Race on the Pharmacokinetics of Antiretrovirals Agents
Much effort has been spent characterizing the effects of sex, weight and race on the pharmacokinetics of anti-retroviral agents. There have been reports suggesting that drug metabolism may differ between sexes and among racial groups [Flexner C, HHR 2003;15(3):7] and that body weight could influence the pharmacokinetic parameters of antiretroviral drugs [Keiser P, et al. Abstract 927, 10th CROI, 2003, Boston]. These are potentially important observations since significant variations in the pharmacokinetics of antiretroviral drugs could affect virologic response and/or increase the risk of drug toxicity.
Hitti and colleagues explored these topics in a retrospective analysis of pharmacokinetic data from six studies from the AACTG (ACTG 368, 372, 384, 388, 389 and 5055) [Abstract 604]. These investigators assessed individual pharmaco-kinetic data for EFV, indinavir (IDV), nelfinavir (NFV) and M8 (the primary active metabolite of NVF) in plasma samples from 38 females and 233 males from diverse racial backgrounds. Body weight and body mass index (BMI) were also included in the multivariate logistic regression model. The authors found that female subjects (15 samples) had a 25% higher mean EFV AUC than their male counterparts (82 samples) [see Table below]. This association remained statistically significant after adjusting for weight, racial group and co-administered medications (amprenavir [APV]; ritonavir [RTV]; and any NNRTIs). No significant differences in systemic exposure to IDV, NFV or M8 were found between males and females. Of note, women were more likely than men to belong to a racial minority group, which could have confounded the study results.
HITTI ABSTRACT FROM 11TH CROI (reported by Jules Levin)
Sex and Weight as Covariates in the Pharmacokinetics of Efavirenz, Indinavir, and Nelfinavir (program abstract)
J Hitti*1, S Rosenkranz2, S Cohn3, L Kamemoto4, A Kashuba5, E Caten2, and F Aweeka6
1Univ. of Washington, Seattle, USA; 2Harvard Sch. of Publ. Hlth., Boston, MA, USA; 3Univ. of Rochester, NY, USA; 4Univ. of Hawaii, Honolulu, USA; 5Univ. of North Carolina at Chapel Hill, USA; and 6Univ. of California, San Francisco, USA
Background: Recent data suggest that women may have decreased clearance of saquinavir compared to men. Our objective was to examine sex-specific systemic exposures for other selected antiretroviral medications and to determine whether any observed sex differences could be explained by weight.
Methods: We conducted a retrospective analysis of the pharmacokinetics of efavirenz, indinavir and nelfinavir and its metabolite M8 in women and men from 6 AIDS Clinical Trials Group studies with intensive sampling (>=5 data points). Data were available from 15 women and 82 men on efavirenz (600 mg once daily), 14 women and 63 men on indinavir (dose-normalized to 1200 mg twice daily) and 20 women and 133 men on nelfinavir (1250 mg twice daily). Subjects were on stable antiretroviral therapy for >=7 days. Area under the concentration-time curve (AUC, ng*h/mL) was calculated from 0 to 24 hours for efavirenz, from 0 to 12 hours for indinavir, and from 0 to 6 hours for nelfinavir and M8 (for a truncated AUC). T-tests, ANOVA and regression models were used to test the univariate effects of sex, co-administered antiretroviral medications, weight and body mass index on log-transformed AUC. The final analysis of covariance models included sex and weight but not their interaction, and were used to predict the AUCs for women and men at an average weight of 70 kg and an 94th percentile weight of 100 kg.
Results: Women had significantly lower efavirenz AUC compared with men, even after adjustment for weight (p <0.05). There were no significant sex differences in the AUC of indinavir, nelfinavir, or M8. Increased weight was significantly associated with lower efavirenz and indinavir AUC (p <0.05). The presence or absence of co-administered antiretroviral medications did not modify these observed associations.
Table. Predicted AUC (ng*h/mL) for women and Men at 2 Body weights
|Drug ||70 kg ||70 kg ||100 kg ||100 kg ||Sex ||Weight |
| ||Women ||men ||women ||men ||p ||p |
|EFV ||41)(382 ||53)(890 ||33)(654 ||43)(827 ||<0.05 ||<0.05 |
|IDV ||37)(143 ||38)(910 ||19)(528 ||20)(457 ||ns ||<0.05 |
|NFV ||13)(904 ||13)(761 ||14)(831 ||14)(678 ||ns ||ns |
|M8 ||2)(276 ||2)(274 ||2)(510 ||2)(508 ||ns ||ns |
Conclusions: Female sex was associated with lower efavirenz AUC regardless of weight. Greater weight was associated with lower AUC for efavirenz and indinavir in both sexes. These data suggest differences in the absorption or elimination of efavirenz between men and women. Further evaluations of exposure-efficacy/toxicity relationships are warranted with these medications.
With respect to race, blacks had significantly higher IDV AUCs and lower NFV and M8 AUCs. In contrast to the studies by Ribaudo and Haas, EFV AUC was indistinguishable between blacks and subjects from other racial backgrounds,. These results were independent of sex, weight and BMI.
Finally, body weight was a significant predictor of AUC for EFV and IDV but not for NFV or M8. Weight remained a statistically significant determinant of this pharmacokinetic parameter even after adjusting for sex and concurrent medications. It is important to note that body weight was not statistically different between women and men.
These study findings underscore the complex influences of race, gender and weight on the pharmacokinetics of individual antiretroviral agents. A number of studies have described differences in treatment efficacy and antiretroviral-induced toxicity between women and men that may now be explained by sex-based differences in drug concentrations. The positive association between female sex and higher plasma concentrations of some antiretroviral drugs has been previously described [see Flexner C, HHR 2003;15(3):7]. One possibility is that sex-based differences in P-glycoprotein expression may explain this association. P-glycoprotein is the product of the mdr1 gene and is a membrane drug efflux transporter that pumps out of the cell some antiretroviral agents [see Andrade and Flexner, HHR 2001;13(2):12]. In some studies, females were shown to express less P-glycoprotein than males [Cummins CL, et al. Clin Pharmacol Ther 2002;72:474] which could theoretically affect the plasma concentrations of antiretroviral agents that are P-glycoprotein substrates. However, other potential mechanisms may figure into this relationship, since plasma concentrations of IDV, a P-glycoprotein substrate, were equivalent in male and females in this and one previous study [Fletcher CV, et al. Abstract 128, 2nd IAS, 2003, Paris].
Gastrointestinal motility, plasma protein levels, and CYP450 enzyme function and excretion activity are examples of other potential factors that could theoretically contribute to the sex-based variations of antiretroviral pharmacokinetics. Gastric emptying is influenced by sex hormones and is reportedly slower in females [Hutson WR, et al. Gastroenterology 1989;96:11]. A trend was found towards a higher CYP450 3A4 metabolic rate in females while sex-based differences are thought to be the result of weight difference between men and women [Gandhi M, et al. Annu Rev Pharmacol Toxicol 2004;44:499]. Although, sex-determined variations in drug metabolism have been identified for a number of commonly prescribed drugs, it remains unclear whether these differences can significantly affect the efficacy and toxicity of antiretroviral agents. To date, there are no compelling data to justify antiretroviral dosing modifications based on sex-related metabolic and pharmacokinetic differences.
Some have questioned the practice of using fixed antiretroviral doses with no regard to body weight. Results from this and other retrospective studies suggest that weight is a predictor of plasma concentrations for some but not all antiretroviral agents [Keiser P, et al. Abstract 927 10th CROI, 2003, Boston]. However, the isolated effect of weight on pharmacokinetics is difficult to discern because in most studies investigating this association women tended to have lower body weights and BMIs than males. In the current study, weight did not differ between men and women but remained a predictor of AUC for EFV and IDV, and for NFV and M8, even after adjusting for sex. These results conflict with findings from a previous study that found a positive association between lower body weight and therapeutic success in patients treated with NFV-containing regimens [Keiser P, et al. Abstract 927, 10th CROI, 2003, Boston]. The lower proportion of women in the current study and differences in study population could explain these discrepancies. Thus, questions remain about the relevance of body weight to the efficacy of antiretroviral drugs.
Generic Antiretrovirals: How Good Are They?
The production of generic versions of brand name antiretroviral agents continues to increase in countries such as Thailand, Brazil, and India. However, there is ongoing concern about the quality of generic formulations. Penzak and colleagues addressed this issue in a quality control and bioequivalence study of six antiretroviral agents: saquinavir (SQV), IDV, lopinavir/ ritonavir (LPV/r), ritonavir (RTV), amprenavir (APV) and EFV from six manufacturers from various international sources [Abstract 581]. Using the United States Pharmcopeia (USP) Uniformity of Dosage Units Test (which specifies that drug content be between 85% to 115% of label claim) the investigators found that, with the exception of RTV, the active ingredient of each drug was within USP specifications. Of note, RTV was not continuously refrigerated which could explain the lower concentration of the active compounds found in this formulation.
These encouraging findings should not dissuade investigators from conducting further bioequivalence studies to continue monitoring the quality of generic drugs, especially in light of recent reports of counterfeit antiretroviral products in the developing world [Apoola A, et al. Lancet 2001;357:1370].
Extended-Release Stavudine and Tenofovir DF
Stavudine extended-release (d4T XR) and tenofovir DF (TDF) are potential components of once-daily HAART regimens and thus, it is important to establish how concomitant administration of these two agents might affect their pharmacokinetics. Previous investigation revealed that TDF had no significant effect on the pharmacokinetics of d4T XR [Kaul S, et al. Abstract 534-W, 10th CROI, Boston 2003]. In this study the same team of investigators evaluated the effect of d4T XR on TDF steady-state pharmacokinetics [Kaul S, et al. Abstract 602]. In this Phase 1, open-label study, 18 healthy volunteer went through three treatment phases: Period 1 d4T XR 100 mg QD for 1 day
Period 2 TDF 300 mg QD for 7 days
Period 3 d4T XR 100 mg QD + TDF 300 mg QD for 1 day
The authors reported that d4T XR had no significant effects on TDF pharmacokinetics and co-administration of TDF and d4T XR was safe and well tolerated. Although the lack of drug interaction and toxicity in this short term study is reassuring, the main concern continues to be possible long-term toxicity associated with d4T XR. In a separate study presented at this meeting, extended-release d4T appears to be less toxic than standard d4T [see Cofrancesco J, HHR 2004;16(2):12]. This was a post-hoc analysis of two different studies and did not compare the d4T XR safety profiles with that of other NRTIs.
Effect of Hepatic Impairment on Tenofovir Metabolism
Hepatic impairment is a common finding in the HIV infected population, especially in the setting of hepatitis B or C co-infection. The pharmacokinetics of TDF, an agent with activity against HIV and hepatitis B, have not been studied in patients with hepatic impairment. Even though TDF is primarily excreted unchanged by the kidneys, 20% to 30% of this drug is eliminated by non-renal mechanisms [Viread Disoproxil Fumarate (Tenofovir) tablets product monography. Gilead Sciences, Inc., Foster City, CA, 2002]. It is unknown whether TDF affects the pharmacokinetics of other drugs used to treat hepatitis B infection (adefovir dipivoxil [ADV]) and hepatitis C (ribavirin [RBV]). To address these questions, a group of investigators from Gilead conducted three pharmacokinetic studies [Kearney PB, et al. Abstract 600]. Using the Child-Pugh-Turcotte Score (CPTS), the first study compared single dose TDF pharmacokinetics (300 mg QD) in subjects with unimpaired hepatic function (8 subjects; CPTS 5.0±0) to those with moderate (7 subjects; CPTS 8.0±.8) and severe (8 subjects; CPTS 10.8±1.0) hepatic impairment. The second and third studies evaluated potential pharmacokinetic interactions between ADV (10 mg QD alone for 1 day)/TDF (300 mg QD for 7 days) and RBV (600 mg QD for 1 day)/TDF (300 mg QD for 21 days). The authors reported that TDF concentrations were not significantly altered by moderate or severe hepatic impairment, and ADV and RBV plasma concentrations were indistinguishable when dosed with or without TDF. These findings are consistent with the fact that TDF is not primarily metabolized by the liver, and thus hepatic impairment should have a limited impact on the pharmacokinetics of this drug.
Efavirenz and Statins
PIs are inhibitors of cytochrome P450 3A4 (CYP450 3A4), the same enzyme responsible for the metabolism of a number of HMG-CoA reductase inhibitors (statins). Consequently, there is a considerable potential for pharmacokinetic drug interactions when these agents are combined. It is known that inhibition of the CYP450 3A4 by ritonavir/saquinavir (RTV/SQV) greatly increases simvastatin and atorvostatin concentrations, which could potentially lead to serious drug toxicity such as rhabdomyolysis [Fichtenbaum CJ, et al. AIDS 2002;16:569]. Conversely, a significant reduction in pravastatin plasma concentrations occurred when it was combined with NFV and EFV, raising concerns about a possible reduction in the effectiveness of this drug [Gerber JG, et al. Abstract 870, 2nd IAS, Paris, 2003].
EFV is a mixed inducer/inhibitor of the CYP450 3A4 and might affect statin metabolism. This issue was addressed in the ACTG 5108 study, in which 28 healthy volunteers were treated 14 days of EFV 600 mg QD and then for 3 days with either simvastatin 40 mg QD (14 subjects) or atorvastatin 10 mg QD (14 subjects) alone [Gerber JG, et al. Abstract 603]. EFV reduced simvastatin and atorvostatin AUC0-24 hrs by 58% and 43%, respectively. Neither simvastatin nor atorvostatin altered the plasma pharmacokinetic parameters of EFV. It remains to be determined whether the reduction observed in simvastatin and atorvastatin concentrations has any clinical significance in the treatment of hyperlipidemia, and whether the recommended doses for these statins will have to be increased in an attempt to offset this pharmacokinetic interaction.
Atazanavir and Saquinavir/Ritonavir
Investigators from the Chelsea and Westminster Hospital in London conducted an open label, single arm, pharmacokinetic study to determine the optimal dosage of concomitant atazanavir (ATV) and SQV-hcg (Invirase)/RTV [Boffito M, et al. Abstract 607]. These investigators reported that addition of ATV 300 mg QD to SQV-hcg/RTV 1600/100 mg QD at steady-state, resulted in a statistically significant increase in SQV geometric mean ratios (GMR) of 112% in Ctrough, -42% in Cmax, and 60% in AUC0-24 hrs. ATV pharmacokinetic parameters were comparable to those observed with RTV boosting (300/100 mg QD) without SQV. No significant changes in trans-aminases, glucose, cholesterol, or triglycerides were observed, although one third of the 20 HIV-infected participants developed clinically relevant hyper-bilirubinemia.
The significant boosting effect on SQV/RTV pharmacokinetics may be secondary to additional CYP450 3A4 inhibition by ATV. Additional studies are needed to further explore the long-term safety, tolerability, and virologic effects of this regimen and to determine how it compares to other ATV-dual PI combinations.
Lopinavir/Ritonavir and Fosamprenavir
Systemic concentrations of LPV and APV are markedly reduced when lopinavir/ ritonavir (LPV/r) and fosamprenavir ( FPV) are combined [Kashuba ADM, et al. Abstract H-855a, 43rd ICAAC, Chicago 2004]. Two studies presented at CROI explored strategies to counteract this deleterious drug interaction. Corbett and others presented the results of a prospective, non-blinded, randomized, cross-over healthy volunteer pharmacokinetic study investigating whether temporal separation of FPV and LPV/r by 4 and 12 hours, or addition of an extra 200 mg of RTV, could prevent the reduction of FPV and LPV concentrations [Abstract 611]. Study subjects were randomized into 3 groups:
Treatment A: FPV 700 mg bid + LPV/r 400/100 mg bid for 7 days (simultaneous administration)
Treatment B: FPVr 700/100 mg bid + LPV/r 400/100 mg bid for 7 days (dosed 4 hours after FPV/r)
Treatment C: FPV/r 1400/200 mg QD + LPV/r 800/200 mg QD for 7 days (dosed 12 hours after FPV/r)
The authors reported that staggering administration of FPV and LPV/r did not ameliorate this pharmacokinetic interaction, since only LPV concentrations improved.
Investigators from GlaxoSmithKline also assessed dosing strategies to overcome the pharmacokinetic interaction between LPV/r and FPV [Wire MB, et al. Abstract 612]. In a randomized, open-label, steady-state, cross-over pharmacokinetic study, healthy volunteers were treated with either a regimen containing a higher dose of LPV/r (FPV 1400 mg bid + LPV/r 533/133 mg bid) or with standard doses of FPV and LPV/r with an additional RTV 100 mg dosed twice-daily (FPV 1400 mg bid + LPV/r 400/100 mg bid + RTV 100 mg bid). The authors found higher LPV plasma concentrations in the group treated with increased doses of FPV and LPV/r when compared to those treated with extra RTV. Nevertheless, APV concentrations were still below the levels observed in other boosted APV pharmacokinetic studies. LPV/r concentrations were higher with extra RTV, but variability was also increased. FPV and LPV/r combinations were poorly tolerated by these healthy volunteers leading to a high discontinuation rate (36% and 44% in the first and second regimens, respectively). Because of the high plasma concentration variability and discontinuation rate, the authors were unable to make dosing recommendations for this triple PI regimen.
Fact or Fiction: Transformation of AZT to D4T-TP
Last year, Becher and colleagues reported that lymphocytes taken from HIV infected patients taking AZT frequently contained the intracellular triphosphate of stavudine (d4T-TP), suggesting that human cells contained an enzyme capable of converting AZT to d4TTP [Becher F, et al. AIDS 2003;17:555]. These findings, if true, could provide one explanation for the extensive cross-resistance between AZT and d4T [see Flexner C, HHR 2003;15(5):5].
To verify these results, investigators from the University of Puerto Rico used HPLC to measure d4T-TP concentrations in100 plasma samples and d4T-TP and AZT-TP concentrations in 450 plasma samples from HIV-infected subjects who were on stable d4T and AZT therapy, respectively [Melendez M, et al. Abstract 597]. They also used an in vitro system to detect d4T-TP in CEMss cells treated with high concentrations of AZT (up to 100 µ mol). The authors reported that d4T-TP was not measurable in any of the 450 samples collected from patients taking AZT, nor from the CEMss cells, even though d4T-TP was detected in all samples of patient treated with d4T. These findings suggest that the initial observation reporting intracellular conversion of AZT to d4T-TP could represent a laboratory artifact, and provides some reassurance that exposure to one NRTI may not inadvertently produce exposure to two.