|Home | About | Journals | Submit | Contact Us | Français|
Non-nucleoside reverse transcriptase inhibitors (NNRTIs) have been used widely for treating human immunodeficiency virus type 1 (HIV-1) infected patients as a component of highly active antiretroviral therapy (HAART) and for the prevention of mother-to-child transmission (MTCT). Cytochrome P450 (CYP) 2B6 is an important hepatic isoenzyme responsible for the metabolism of NNRTIs including efavirenz and nevirapine. Recent pharmacogenetic studies have shown that CYP2B6 genetic variants alter hepatic CYP2B6 protein expression and function, and the pharmacokinetics of several CYP2B6 substrates. In particular, the CYP2B6-G516T polymorphism in exon 4 affects the pharmacokinetics of efavirenz. Other studies have shown associations of the CYP2B6-G516T genotype with nevirapine pharmacokinetics and central nervous system adverse effects related to efavirenz use. In total, CYP2B6 genetic variants are important determinants of efavirenz and nevirapine pharmacokinetics . Further studies are needed to identify the associations of CYP2B6 genetic variants with the development of NNRTI resistant viruses.
The morbidity and mortality associated with HIV infection in adults [1, 2] and children  have been improved significantly due to durable virologic suppression, immunologic recovery, and clinical improvement achieved by highly active antiretroviral therapy (HAART). Current treatment guidelines for the use of HAART in antiretroviral naïve adults  and children  recommend the use of two nucleoside reverse transcriptase inhibitors (NRTIs) in combination with either a non-nucleoside reverse transcriptase inhibitor (NNRTI) or a protease inhibitor (PI), or in selected cases, three NRTI-based regimens (3 NRTI). NNRTI-based regimens have been a preferred choice for many patients compared to PI-based regimens because of lower pill burdens allowing for better compliance to antiretroviral therapy  and the better lipid profile associated with NNRTI usage [5, 6].
Currently, three NNRTIs including efavirenz (EFV), nevirapine (NVP), and delavirdine (DLV) are licensed for use in HIV-infected patients in the United States. EFV has been included in the majority of treatment guidelines as the preferred first-line regimen for HIV-infected adults  and children >3 year old  in the United States as well as outside of the United States . NVP has been used in HIV-infected infants, children and adults not only as a component of HAART [101-103], but also for the prevention of mother-to-child transmission (MTCT) [7-11].
Most patients experience an excellent response to NNRTI containing regimens, However a subset of individuals respond poorly, have slow virologic suppression and CD4+ lymphocyte increases, increased numbers of side effects or rapidly develop resistance. Although there are a number of reasons for these differential responses, recent pharmacogenetic studies suggest that at least some of these variable responses to NNRTI containing regimens are predictable based on variants in genes responsible for metabolizing and transporting antiretrovirals including PIs, NNRTIs and NRTIs . Compelling data, confirmed by several cohorts, demonstrate that the single nucleotide polymorphism (SNP) in CYP2B6 (CYP2B6-G516T or CYP2B6*6) significantly alters EFV pharmacokinetics in HIV-infected patients [13-21]. In addition, other studies have shown the importance of the CYP2B6 SNP on central nervous system (CNS) adverse effects related to EFV use [14, 15] and NVP pharmacokinetics [15, 22, 23].
This article reviews the metabolism of EFV and NVP, the impact of CYP2B6 genetic variants on hepatic CYP2B6 expression and function, and the effects of other genetic variants on pharmacokinetics, adverse events, and clinical responses to NNRTIs.
EFV is metabolized by the hepatic CYP. Among the CYP isoenzymes, CYP2B6 is the major enzyme converting EFV to its predominant metabolite, 8-hydroxyefavirenz , which is further hydroxylated to 8, 14-dihydroxyefavirenz by CYP2B6 (Figure 1) . EFV is also converted to 7-hydroxyefavirenz, which is a minor pathway, possibly by CYP2B6 and CYP2A6 [25, 26]. Their conjugated metabolites are excreted predominantly in the urine as glucuronides and sulfate conjugates [24, 25].
NVP is also metabolized by the hepatic CYP and converted to its metabolites. Among the metabolites, 2-hydroxynevirapine, 3-hydroxynevirapine, 8-hydroxynevirapine and 12-hydroxynevirapine are the major metabolites, mainly transformed by CYP3A4, CYP2B6, CYP3A4 and CYP2D6, respectively [27, 28]. The NVP metabolites are eliminated primarily in the urine in four glucuronidated conjugates of hydroxylated metabolites (Figure 2). The percentage of each metabolite found in urine is 23%, 32%, 2%, and 29% of total metabolites, respectively .
DLV is extensively biotransformed into several inactive metabolites, primarily by CYP3A4 with minor involvement of CYP2D6 . It is excreted into the urine and feces, primarily as dealkyl delavirdine and pyridinecleaved delavirdine . The hepatic CYP2B6 does not appear to be involved with DLV metabolism.
Over 100 SNPs have been identified within the CYP2B6 gene located on chromosome 19 . Among them, several important SNPs have been found to be associated with hepatic CYP2B6 expression and function (for summary see recent review ). In in vitro assays using human liver microsomes, several studies have demonstrated the impact of genetic variants on CYP2B6 protein expression and hydroxylation of CYP2B6 substrates. Lang et al. analyzed the expression of the CYP2B6 protein and its substrate metabolic transformation using S-mephenytoin N-demethylase activity in human liver samples. They observed significantly reduced CYP2B6 protein expression and decreased activity in human liver samples with the CYP2B6-C1459T (rs3211371) [CYP2B6*5 and CYP2B6*7] ; although these findings are yet to be confirmed. In contrast, Hesse et al. demonstrated that carrying the CYP2B6*6B allele was an important predictor of bupropion hydroxylation when the data were stratified with alcohol use; however, no effect of the genotype was found to be associated with CYP2B6 mRNA or protein expression . In another study, Xie et al. showed that the CYP2B6*6 variant enhanced cyclophosphamide hydroxylation . Lamba et al. evaluated CYP2B6 splicing variants in human liver samples and demonstrated that the CYP2B6-G516T, –C1459T, and intron 3 C15582T genotypes were predictors of hepatic CYP2B6 activity and varied according to sex and ethnicity . In an additional study by Desta et al., bupropion protein expression and activity, and EFV hydroxylation were significantly decreased in subjects with the CYP2B6*5, or CYP2B6*6 . Recent studies have shown several novel single nucleotide polymorphisms including CYP2B6*16 [18, 20], CYP2B6*27 and *28 , CYP2B6*26  which significantly altered EFV pharmacokinetics. In total, the CYP2B6-G516T genotype may serve as a tag SNP for several haplotypes and is an important predictor of hepatic CYP2B6 protein expression and hydroxylation of CYP2B6 substrates.
Several studies have identified a relationship between CYP2B6-G516T gene polymorphisms and plasma EFV pharmacokinetics in HIV-infected adults [13-20] and children  (Table 1). All available data have shown that the subjects with the CYP2B6-516-T/T genotype (homozygous variants) have consistently higher EFV concentrations compared to those with the CYP2B6-516-G/T (heterozygous variant) and G/G genotype (wild type). Additional research also identified differences in EFV pharmacokinetics in subjects with the CYP2B6-516-G/T and those with the CYP2B6-516-G/G genotypes [14, 16, 19, 20].
We evaluated the association between the CYP2B6-G516T genotype and EFV pharmacokinetics in 71 HIV-1 infected children receiving HAART . Children with the T/T genotype had significantly lower EFV oral clearance rates (3.0 L/h/m2) than those with the G/T genotype (5.7 L/h/m2, P = 0.02) and the G/G genotype (7.0 L/h/m2, P = 0.003) (Figure 3A). A multivariate analysis for EFV oral clearance including age, gender, race/ethnicity, and CYP2B6-G516T genotype also showed that age (P = 0.03) and the CYP2B6-G516T genotype (P = 0.005) were independently and statistically associated with EFV oral clearance.
Numerous studies have investigated the relationship between the CYP2B6-C1459T genotype and plasma EFV pharmacokinetics in HIV-infected adults [14, 17, 20, 37] and children . However, to date, no significant effect has been observed.
Compared to EFV, few studies have evaluated the relationship between the CYP2B6 genotypes and NVP pharmacokinetics [15, 22, 23] (Table 1). Rotger et al. assessed the CYP2B6-G516T in 59 HIV-infected subjects who received NVP as a component of HAART . Individuals with the CYP2B6-516-T/T had 1.7-fold higher mean plasma levels than those with the CYP2B6-516-G/G. Similarly, Penzak et al. evaluated 23 HIV-infected adults who received NVP and showed that the median NVP concentration in individuals with the CYP2B6-516-T/T (7.6 μg/mL) was higher than those in individuals with the G/G genotype (4.2 μg/mL) or with the G/T genotype (5.6 μg/mL) . Our recent data analyzing 126 HIV-1 infected children receiving NVP as a component of HAART demonstrated a significant association between the CYP2B6-G516T and NVP oral clearance . NVP oral clearance in children with the CYP2B6-516-T/T genotype (1.6 L/hr/m2) was significantly decreased compared to those with the -G/G (2.3 L/hr/m2, P = 0.001) and -G/T genotype (2.1 L/hr/m2, P = 0.003) (Figure 3B). A multivariate analysis for NVP oral clearance showed that the CYP2B6-G516T genotype (T/T genotype, P = 0.02) and concomitant PI (P = 0.05) were independently associated with NVP oral clearance. The CYP2B6-C1459T genotype was not associated with NVP oral clearance (P = 0.95).
Several major adverse effects are associated with EFV use including skin rash, abnormal liver enzymes, and notably CNS symptoms including insomnia, abnormal dreams, confusion, amnesia, and hallucination [101, 102]. These CNS adverse effects are associated with the higher levels of EFV in plasma [38-40], while others failed to demonsrate the association [41, 42]. Two studies have demonstrated the association between the CYP2B6 genotype and the incidence of CNS adverse effects [14, 15]. Haas et al. evaluated 156 patients who received EFV with different combinations of NRTI demonstrating that the CYP2B6-G516T genotype was associated with CNS symptoms at 1 week after initiation of treatment . Similarly, Rotger et al. evaluated 167 patients who received EFV as a component of HAART and demonstrated that the CYP2B6-516-T/T genotype was associated with a higher incidence of sleep disorders and fatigue .
EFV is contraindicated for women of child bearing potential because it is known to be teratogenic, causing neural tube defects [43-45]. The impact of the CYP2B6-G516T genotype on the incidence of neural tube defect is currently unknown; however, it is possible that the CYP2B6-G516T genotype, which is associated with altering the plasma EFV levels, may impact on the incidence of neural tube defect in infants whose mothers are exposed to higher levels of EFV during the early pregnancy. In any event, women who are planning or at risk for becoming pregnant should be offered alternative antiretroviral regimens whenever possible.
Several major adverse effects are associated with NVP use including skin rash and abnormal liver enzymes. The frequency of Grade 3 or 4 increased liver enzymes in patients on NVP (4-18%) appears to be higher than those on EFV (1-8%) based on several clinical trials and cohort studies . Factors associated with hepatotoxicity include female gender, higher baseline and actual CD4+ T-cell counts [47, 48], and HLA-DRB1*0101 . Although NVP is not a substrate of P-glycoprotein , the ATP binding cassette, subfamily B, member 1 (ABCB1), which encodes P-glycoprotein, the ABCB1-C3435T (rs1045642) genotype has been reported to be associated with the incidence of hepatotoxicity in HIV-1 infected adults receiving NVP containing HAART regimens [51, 52]. Richie et al. evaluated the possible significant genotypes in 13 patients who developed NVP-associated hepatotoxicity with 49 matched controls . Using univariate analysis, the ABCB1-3435-T allele and the status of hepatitis B surface antigen were associated with a decreased likelihood of hepatotoxicity. Haas et al. also demonstrated the association between the ABCB1-C3435T genotype and NVP-associated hepatotoxicity in 53 patients compared to 108 controls . Using multivariate analysis, the ABCB1-C3435T genotype was significantly associated with reduced risk of NVP-associated hepatotoxicity. Interestingly, the CYP2B6-G516T genotype was not associated with risk of hepatotoxicity when assessed in multivariate analyses in both studies. The exact mechanism why ABCB1-C3435T was associated with a reduced risk of NVP-associated hepatotoxicity is currently unknown; however, it is possible that altered P-glycoprotein activity in the intestine associated with the ABCB1 variants  alters disposition of NVP and/or its metabolites that affects intracellular concentrations of NVP and toxicity in liver.
The actual influence of the CYP2B6-G516T genotypes on clinical responses has been evaluated in only a few studies [14, 17, 21, 54] (Table 1). Haas et al. evaluated the impact of the CYP2B6 genotype on virologic and immunologic responses over 24 weeks in 157 HIV-infected adults receiving EFV as a component of HAART and have shown no differences in immunologic (P = 0.15) or virologic responses (P = 0.74) among the patients with the CYP2B6-G516T genotypes after 24 weeks of therapy . A recent larger study in adults by Haas et al. also showed no correlation between the presence of CYP2B6-G516T polymorphisms and long-term virologic or immunologic response for up to three years . Another study by Motsinger et al. reported that virologic failure in patients receiving EFV was associated with a two-locus interaction between ABCB1-G2677T and CYP2B6-G516T . Our previous data in 72 children who received EFV as a component of HAART  did not demonstrate differences in immunologic or virologic outcomes when analyzed in association with CYP2B6-G516T gene polymorphisms despite significant differences in EFV oral clearance. However, our recent study in 126 children with advanced HIV infection who received NVP as a component of HAART demonstrated different findings . Children with the CYP2B6-516-T/T genotype had the greatest increase in CD4+ T-cell percentages (+9.0%) compared to those with the -G/G (+3.2%, P = 0.008) and -G/T genotype (+5.0%, P = 0.04) at week 12. This trend continued for CD4+ T cell percentages in children with the CYP2B6-516-T/T genotype (+10.5%) compared to those in children with the -G/G genotype (+4.7%, P = 0.01) and -G/T genotype (+8.2%, P = 0.06) at week 24. A multivariate analysis for change in CD4+ T-cell percentages showed that the CYP2B6-516-T/T genotype was the only covariate associated with a change in CD4+ T-cell percentages from baseline to week 12 (P = 0.03). Virologic response at weeks 12 and 24 was also evaluated among the three groups; however, no differences were observed (P = 0.86, P = 0.24, respectively). Although there are many factors involved in determining the response to antiretroviral therapy in HIV-infected patients, our study was the first report demonstrating that CYP2B6 genetic variants can significantly affect the clinical response to NNRTI-containing regimens. It is important to note the differences between our first and second studies. In our initial report, children were carefully monitored early in their treatment course with pharmacokinetic analysis. For children whose EFV plasma concentrations failed to achieve predetermined target levels, the dose of EFV was adjusted. Thus, the impact of specific CYP genotypes on virologic and immunologic clinical outcomes was limited because of the intensive pharmacokinetic monitoring of the children participating in this study. In contrast in our more recent study, HIV-1 infected children were treated with regimens containing an NNRTI without real-time monitoring of drug levels. Therefore, we believe that our current findings are more representative of children receiving chronic NNRTI containing antiretroviral therapy when routine pharmacokinetics is not performed.
One of the major concerns for patients who receive an NNRTI as part of their antiretroviral regimen is the development of resistant viruses, which requires only a single point mutation to confer high-level, cross-class resistance among NNRTIs . The impact of the CYP2B6 genotype on the development of NNRTI resistant virus is currently unknown; however, Ribaudo et al.  demonstrated that CYP2B6-G516T genotypes, which were associated with plasma EFV levels, predicted the duration of plasma EFV levels exceeding 95% inhibitory concentration. They concluded that the CYP2B6-516-T/T genotype or a prolonged half-life of EFV may predict increased risk for developing EFV resistance virus among patients who discontinued EFV-containing regimens.
We investigated the impact of the CYP2B6-G516T genotype and EFV oral clearance on HIV-1 infected children who developed NNRTI mutations (K103N and G190S) during the early stage of HAART (n = 10) and those who sustained undetectable plasma HIV-1 RNA during HAART (n = 31) . There was no difference in the distribution of the CYP2B6-G516T genotype or EFV oral clearance rate between two groups. It should be noted, however, that this was the same cohort of children who had dose adjustments based on EFV pharmacokinetics early in their treatment and the relatively small number of subjects available for study may have resulted in inadequate power to demonstrate a difference.
NVP has been used widely to prevent HIV-1 MTCT in developing countries [7-11]. However, a single dose of NVP is associated with the development of NVP resistant virus in mothers and their infants who become infected with HIV . The impact of the CYP2B6 genotype on the development of NVP resistant virus in mother and infant is unknown.
Although NVP or EFV is not a substrate of P-glycoprotein , associations have been reported between the ABCB1 genotype and EFV pharmacokinetics , and virologic outcome in HIV-infected adults receiving EFV containing HAART regimens . In addition, an inverse correlation between NVP intracellular concentrations and P-glycoprotein expression in PBMC has been reported , while another study demonstrated no association between EFV intracellular conc entrations and P-glycoprotein expression in PBMC . Recent data have suggested that the genetic polymorphisms for ABCB1-G2677T (rs2032582), which is highly associated with the ABCB1-C3435T genotype, influence the virologic outcome combined with the CYP2B6-G516T genotype in patients who receive EFV containing HAART regimens . In total, these data suggest that EFV, NVP, or their metabolites may be a substrate of P-glycoprotein.
We evaluated 14 pairs of cerebrospinal fluid (CSF) and plasma NVP levels from 11 children with known or suspected HIV encephalopathy, or presence of symptoms consistent with neurological decline attributable to HIV-related neurologic diseases . The median NVP CSF: plasma ratio was 0.62 in children with the ABCB1-3435-C/T or - T/T compared to 0.43 in children with the ABCB1-3435-C/C genotype (P = 0.01). No significant difference was observed when the ratios were compared with the CYP2B6-G516T genotype (P = 1.00). Although the number of CSF samples was limited, this finding provides additional support for NVP or its metabolites being substrates of P-glycoprotein .
Although there are no data available regarding the impact of age on hepatic CYP2B6 expression in children and adults, hepatic CYP activity is known to change with age and is often greater in young children (1-4 years old) compared to adults . This is also supported by studies showing that the clearance of CYP substrates including warfarin , antipyrine , and nelfinavir [63, 64] are increased in young children. Therefore, we examined if the CYP2B6-G516T genotype in young children would have a greater impact on EFV oral clearance than older children [21, 23]. EFV oral clearance was significantly greater for the younger children <5 years with the G/G genotype (9.7 L/h/m2) compared to those with ≥5 years with the G/G genotype (6.6 L/h/m2) (P = 0.03). However, no differences were observed in children with the G/T genotype (P = 0.71), or T/T genotype (P = 0.86). These data suggest that hepatic enzyme levels determined by age may need to be considered when evaluating the impact of genetic variants on drug pharmacokinetics of younger children. We performed the same analysis using the NVP cohort; however, no significant difference was observed when we analyzed the data by age . Because other CYP isoenzymes including CYP3A4 or CYP2D6 involve in NVP metabolism and the developmental maturation of each hepatic isoenzyme is different among the isozymes , the variations in these genes may affect the expression of hepatic CYP isoenzymes, which may lead to differences in NVP pharmacokinetics.
Recent articles illustrate the feasibility of translating HIV pharmacogenomics into clinical practice [36, 66]. Gatanaga et al. demonstrated that patients with the CYP2B6 *6/*6 and *6/*26, who had extremely high plasma EFV concentrations, had persistently suppressed plasma HIV-1 RNA while receiving reduced doses of EFV (200-400 mg/day, instead of 600 mg). They also showed improvement of CNS-related symptoms after the dose adjustment. In addition, Torno et al. presented a case of an HIV-infected patient who has maintained durable virologic suppression of HIV infection with a 400 mg daily dose of EFV and the pharmacogenetic study and therapeutic drug monitoring supported the dose reduction. They suggested that the potential applications of combined pharmacogenetic testing/therapeutic drug monitoring in guiding efavirenz-based therapy.
Several important genetic variants have been reported that alter NNRTI pharmacokinetics and the risk for adverse effects. The CYP2B6-G516T is associated with EFV and NVP pharmacokinetics, and the incidence EFV-related CNS adverse effects. In addition, the ABCB1-C3435T genotype has been associated with the incidence of NVP hepatotoxicity, although NNRTI are thought not to be a substrate of P-glycoprotein. At present, therefore, the relationship between the ABCB1 genotype and hepatic P-glycoprotein functional activity remain uncertain.
In total, currently available data suggest that CYP2B6 variants are important predictors of EFV and NVP pharmacokinetics. It is likely, therefore, as antiretroviral therapy evolves to the optimization of treatment regimens for the individual patient that determination of the CYP2B6 genetic variants will be important for optimizing NNRTI containing regimens.
Pharmacogenetics promises to provide important clues to understanding the variable responses of patients to antiretroviral treatment regimens. Our findings and those of others support an important effect of CYP2B6-G516T genotypes on EFV pharmacokinetics in HIV-1 infected children and adults. In addition, there are several genetic variants reported to alter the concentrations of plasma EFV and NVP that are associated with the risk for adverse effects. These findings will require validation in additional cohorts.
Several important areas remain to be investigated. First, the impact of CYP2B6 genotypes on MTCT and development of resistant virus in mothers and their infants after receiving a single dose of NVP is of particular interest. Second, the association of genetic variants with the development of NNRTI resistance with different antiretroviral backgrounds is an area in need of investigation. Third, because NVP is an inducer of CYP, the effect of CYP2B6 genotypes on long-term antiretroviral treatment remains to be examined. Fourth, the relationships between NNRTI and P-glycoprotein in lymphocytes, the blood brain barrier, and other anatomical sites warrants further investigation. Fifth, because age is an important factor to determine the metabolism of drugs in children, the contribution of age into PK data needs to be in consideration when analyzing the pharmacogenetic data in children. Finally, studies of genetic variations of nuclear receptors which regulate the transcription of CYP and other transporters responsible for metabolizing and transporting NNRTIs are of considerable importance.
These and other questions will need to be addressed in order for determination of genetic variation to become a useful tool in optimizing the treatment of patients.
Importance of NNRTIs in treatment of HIV-1 infection
Metabolism of NNRTIs
Genetic variants and hepatic CYP2B6
Genetic variants and NNRTI pharmacokinetics
Genetic variants and clinical responses
Genetic variants and adverse effects
The research presented was supported in part by the Pediatric AIDS Clinical Trials Group (PACTG), the International Maternal Perinatal Adolescent AIDS Clinical Trial (IMPAACT) Network (grant AI068632), and grants from the National Institute of Allergy and Infectious Diseases (grant 5K23AI-56931 to A.S. and grant AI-36214 to the Virology Core UCSD Center for AIDS Research).
Akihiko Saitoh, University of California, San Diego Department of Pediatrics Division of Infectious Diseases 9500 Gilman Drive, La Jolla, CA 92093-0672 Telephone: (858) 534-7258, Fax: (858) 534-7411 ; Email: ude.dscu@hotiasa..
Stephen A. Spector, University of California, San Diego Department of Pediatrics Division of Infectious Diseases Member, Center for Molecular Genetics Member, Center for AIDS Research 9500 Gilman Drive, La Jolla, CA 92093-0672 Telephone: (858) 534-7170, Fax: (858) 534-7411 ; Email: ude.dscu@rotcepsas.