| |
Pharmacogenetic Characteristics of Indinavir, Zidovudine, and Lamivudine Therapy in HIV-Infected Adults: A Pilot Study
|
| |
| |
JAIDS Journal of Acquired Immune Deficiency Syndromes: Volume 42(4) 1 August 2006 pp 441-449
Anderson, Peter L. PharmD*; Lamba, Jatinder PhD ; Aquilante, Christina L. PharmD*; Schuetz, Erin PhD ; Fletcher, Courtney V. PharmD*
From the *University of Colorado at Denver and Health Sciences Center, Denver, CO; and St. Jude Children's Research Hospital, Memphis, TN.
Abstract
Objective: The aim of the study was to investigate relationships among indinavir, lamivudine-triphosphate, and zidovudine-triphosphate pharmacokinetics and pharmacodynamics with polymorphisms in CYP3A5, MDR1, MRP2, MRP4, BCRP, and UGT1A1 genes.
Study Design: Retrospective pilot investigation among 33 subjects who participated in a randomized pharmacological study of indinavir, lamivudine, and zidovudine. Subjects were defined as genetic variant carriers or not. Relationships were investigated with multivariable regression. Indinavir clearance was adjusted for African American race; triphosphates for sex; and HIV-response for study arm, drug exposure, and baseline HIV-RNA.
Results:
Genetically determined CYP3A5 expressors had 44% faster indinavir oral clearance versus nonexpressors (P = 0.002). MRP2-24C/T variant carriers had 24% faster indinavir oral clearance (P = 0.05).
Lamivudine-triphosphate concentrations were elevated 20% in MRP4 T4131G variant carriers (P = 0.004).
A trend for elevated zidovudine-triphosphates was observed in MRP4 G3724A variant carriers (P = 0.06).
The log10 changes in HIV-RNA from baseline to week 52 were -3.7 for MDR1 2677 TT, -3.2 for GT, and -2.2 for GG (P < 0.05).
Bilirubin increases were 2-fold higher in UGT1A1 [TA]7/[TA]7 genotypes.
No relationships were identified with BCRP.
Discussion: Novel relationships were identified among genetic variants in drug transporters and indinavir, lamivudine-triphosphate, and zidovudine-triphosphate concentrations. CYP3A5 expression was associated with faster indinavir oral clearance. These pilot data provide a scientific basis for more rational utilization of antiretroviral drugs.
Introduction
Up to 50% of patients experience adverse drug effects or virologic failure during combination antiretroviral therapy for HIV infection.1,2 Multiple factors contribute to how patients respond to therapy including viral drug susceptibility, patient adherence, and biological/genetic differences among patients in their pharmacokinetic and pharmacodynamic characteristics.3-5
In terms of pharmacokinetic variability, the coefficients of variation for protease inhibitor plasma area under the concentration time curves are generally 50% to 100% among adult HIV-infected patients after the same observed oral dose.6,7 The clinical pharmacology of nucleoside analog-reverse transcriptase inhibitors (NRTI) is governed by their intracellular triphosphate anabolites. Coefficients of variation for the intracellular concentrations of active NRTI-triphosphates in patient's peripheral blood mononuclear cells (PBMCs) are generally ≥50%.8,9 The expression and/or function of many enzymes contribute to the pharmacokinetic profile of antiretroviral drugs. These include the major cytochrome P450 and conjugation enzymes such as CYP3A and UDP glycosyltransferase (UGT), and active drug transporters such as P-glycoprotein (P-gp) and multidrug resistance-associated proteins (MRP). Additionally, many nuclear receptor systems such as pregnane X receptor regulate the expression of these important drug clearance enzymes.10
The pharmacodynamic profile of antiretroviral drugs is influenced by plasma drug concentrations (pharmacokinetics) and therefore the expression/function of the same enzymes as listed above. However, additional considerations are important. For example, P-gp expression on apical surfaces of the gut and proximal renal tubules would influence the blood concentrations and thereby the pharmacodynamics of substrate drugs.11 Moreover, P-gp expression on gonad-, placenta-, and brain-blood barriers and on lymphocytes may influence pharmacodynamics when the antiretroviral drug's effect-site resides in these protected tissues, even if plasma concentrations were held constant. Lastly, the expression/function of drug receptors and/or the substrates with which the drug competes would also be expected to influence the variability in clinical drug response. Genetic differences among people are one source of variability governing the expression/function of the enzymes responsible for antiretroviral drug pharmacokinetics and pharmacodynamics.
In the present study, we set out to determine if variation in genes of pharmacological interest were associated with the pharmacokinetic and pharmacodynamic profile of indinavir (IDV), zidovudine (ZDV), and lamivudine (3TC). Our objective was to evaluate relationships between single nucleotide polymorphisms (SNPs) in these genes with the pharmacokinetic and pharmacodynamic profiles of these drugs in HIV-infected adult subjects who had participated in an intensive pharmacological study. The hypotheses were that SNPs in CYP3A5, MDR1, and MRP2 were associated with IDV oral clearance; SNPs in MRP4 and BCRP were associated with ZDV-triphosphate and 3TC-triphosphate concentrations; a UGT1A1 SNP was associated with IDV-induced bilirubin increases; and all SNPs besides that in UGT1A1 were associated with measures of antiviral response. Table 1 lists the genes and SNPs that were investigated and the rationale for inclusion in this study.12-28
DISCUSSION
Several significant relationships were identified between the selected genetic polymorphisms and the pharmacological characteristics of IDV, 3TC, and ZDV in this investigation among HIV-infected persons. Genetically determined CYP3A5 expression was associated with approximately 40% faster IDV CL/F after adjusting for African American race and other covariates. MRP2 C-24T variant carriers had 24% faster IDV CL/F after adjusting for African American race. With regard to ZDV-triphosphate and 3TC-triphosphate concentrations, MRP4 T4131G variant carriers had higher 3TC-triphosphate concentrations after adjusting for sex. MRP4 G3724A variant carriers had 58% higher ZDV-triphosphate concentrations after adjusting for sex, although there were only 3 variant carriers in this study. The magnitude of HIV-RNA decrease from baseline to the week 52 primary study endpoint was associated with MDR1 G2677T and baseline HIV-RNA. The HIV-RNA change was 1.5 log10 and 1.0 log10 greater in subjects with MDR1 2677 TT and GT genotypes compared with GG, respectively. The relationship was independent of study arm and drug exposures. The change in bilirubin levels during IDV-containing therapy was about 2-fold higher in subjects with UGT1A1 [TA]7/[TA]7 compared with [TA]6/[TA]7 or [TA]6/[TA]6. No relationships were identified between selected genetic variants in BCRP and the pharmacological parameters in this study.
This was a small, retrospective, hypothesis-generating pilot investigation, and as such, there are obvious limitations. First, the small number of variant carriers for some SNPs and the overall small sample size limited the statistical power and precluded a thorough evaluation of potential confounders. However, we did adjust pharmacogenetic relationships for known confounders (eg, sex for triphosphates, African American race for IDV CL/F, and study arm/drug exposures/baseline HIV-RNA for the change in HIV-RNA). Second, although statistical comparisons were reported only on hypothesized relationships (see second column in Table 1), there were multiple comparisons without statistical correction. We explored other unplanned relationships and found that 3TC-triphosphate concentrations were 14% higher in MDR1 G2677T variant carriers versus wild type after controlling for sex (P = 0.05) and that MRP4 variant at G3724A was associated with a 60% slower IDV CL/F than wild type after controlling for African American race (P = 0.04). A final limitation was that many additional genes and/or additional SNPs in the genes investigated in this study may be important in the pharmacology of these drugs, but were not quantified. It is necessary to point out that this study also had significant strengths in that the design was rigorous and controlled, and extensive pharmacological data were collected in each subject. The IDV oral clearances were averaged from up to 3 separate intensive pharmacokinetics studies in each subject and were adjusted for body weight. Similarly, an average of 9 triphosphate concentrations was determined in each subject. Thus, the findings in this pilot investigation warrant additional study.
Few investigations have rigorously evaluated CYP3A5 pharmacogenetics of antiretroviral drugs in patients, although most protease inhibitors are avid CYP3A substrates and pharmacokinetic variability is an important clinical issue. CYP3A5 expressor status was not associated with the plasma concentrations of nevirapine, efavirenz, and nelfinavir, all of which undergo significant biotransformation from other metabolic enzymes such as CYP2C19 and CYP2B6.33-35 In contrast, a recent study described about 2-fold faster saquinavir oral clearance in 6 genetically determined CYP3A5 expressors versus 14 nonexpressors.36 In the present study, IDV CL/F was more than 40% faster in 11 subjects with genetically determined CYP3A5 expression versus 22 CYP3A5 nonexpressors. Saquinavir, IDV, and all other protease inhibitors except nelfinavir are predominantly metabolized by CYP3A, so these later findings are biologically consistent.12,37-39 African American race was also associated with IDV CL/F in the present study, but when CYP3A5 expression and African American race were analyzed together, only CYP3A5 remained significant. This suggests that the African American race effect was mediated by racial differences in CYP3A5 expression. It is clinically important to identify sources of protease inhibitor pharmacokinetic variability. The current clinical approach to deal with this variability is to use ritonavir boosting. Although this strategy is associated with clinical success in a large number of patients, this one-size-fits-all approach to the problem ignores the underlying causes of pharmacokinetic variability. Such gaps in knowledge interfere with the development of any other rational strategies to improve the clinical use of protease inhibitors.
The influence of genetic variability in the MDR-1 gene on the clinical pharmacological characteristics of P-gp substrates, including antiretroviral drugs, in patients is contradictory. For example, in terms of the C3435T polymorphism, some studies have described statistically significantly lower nelfinavir concentrations in subjects with CC versus CT and/or TT, whereas other studies have described statistically significantly higher nelfinavir concentrations according to the same genotypes.35,40 In the present study, CC and GG genotypes at C3435T and G2677T, respectively, were associated with faster IDV CL/F (ie, lower IDV concentrations). However, when these genotypes were adjusted for CYP3A5 status, or African American race, the relationships were no longer significant. Several studies have also evaluated MDR1 genotypes according to responses to antiretroviral drug therapy. There has been better consistency across these studies in that subjects with CC and/or GG genotypes at C3435T and G2677T experienced poorer antiretroviral drug responses compared with the counterpart genotypes.33,35,40,41 Yet some studies have not detected relationships.42,43 The data in this study are consistent with these former studies in that subjects with MDR1 2677 GG experienced significantly poorer decreases in HIV-RNA from baseline to the primary study endpoint, independent of baseline HIV-RNA, study arm, and drug exposures. Further study of the impact of P-gp on the clinical pharmacological characteristics of protease inhibitors is needed.
There have been relatively few antiretroviral drug pharmacogenetic studies involving other transporters such as MRP2, MRP4, and BCRP. IDV is a substrate of MRP2, which lines the bile cannicular cells and the gut epithelia, with the potential consequence of limiting drug absorption and speeding drug elimination.19,20 Genetic variability in MRP2 has been identified, but the influence of common SNPs (frequencies >10%) on protein function is not clear.21 One study of 34 HIV-infected patients treated with other HIV protease inhibitors (saquinavir and lopinavir/ritonavir) described 3-fold higher saquinavir concentrations in patients with the MRP2 G1249A GG genotype compared with variant carriers (P = 0.009).22 In the present study, MRP2 G1249A variant carrier status was not related with the pharmacokinetics or pharmacodynamics of IDV, whereas variant carrier status at MRP2 C-24T was associated with faster IDV CL/F after adjusting for African American race.
MRP4 and BCRP were shown to efflux NRTI-monophosphates out of cells in vitro, which would presumably affect the downstream NRTI-triphosphate concentration and therefore potential NRTI activity. For example, intracellular ZDV-phosphate concentrations (mono-, di-, tri-) were dependent on the cellular expression of MRP4,23,24 and the pharmacological activity of ZDV and 3TC was dependent on the expression of BCRP.25 Two relatively common functional polymorphisms have been identified in the BCRP gene, one a G-to-A change at nucleotide 34 in exon 2 (Val to Met at codon 12) and a C-to-A change at nucleotide 421 in exon 5 (Glu to Lys at codon 141).26 However, we did not observe relationships between these polymorphisms and ZDV-triphosphate or 3TC-triphosphate concentrations.
No functional polymorphisms have yet been identified in the MRP4 gene, to our knowledge. In the present study, exonic splicing enhancer analyses were conducted to predict SNPs that had relatively high probability of altered mRNA splicing and potentially altered MRP4 protein expression because nonfunctional MRP4 protein has been described in the literature.44 A significant relationship was identified between carriers of the MRP4 T4131G variant and elevated 3TC-triphosphate concentrations, which suggests the T4131G variant may be associated with reduced expression/function (Fig. 2). MRP4 G3724A variant carriers had elevated ZDV-triphosphate concentrations, and slower IDV CL/F in unplanned analyses, which both would suggest reduced MRP4 expression/function with the variant. It should be emphasized that these are novel findings from a small pilot investigation. Future mechanistic studies are needed to verify the functional significance of these MRP2 and MRP4 SNPs.
IDV (and atazanavir) causes the relatively benign adverse effect of raising unconjugated and total bilirubin concentrations via inhibition of UGT1A1. In this study, subjects who carried 2 [TA]7 alleles experienced about twice the bilirubin elevation as other subjects, which is consistent with previous studies in IDV-triphosphate and atazanavir-treated subjects.27,45 Although these data may seem inconsequential clinically because of the benign nature of bilirubin increases, the findings illustrate an important point: Genetic variability in drug targets influences drug response. Analogies can be drawn with variability in genes that encode drug targets for HIV replication (eg, reverse transcriptase or protease) where genetic mutations are well known to confer reduced susceptibility to drugs. Other analogies can be drawn with drug receptors that might underlie serious adverse effects, such as protease inhibitor-induced hyperlipidemia or mitochondrial toxicity.46 Therefore, future pharmacogenetic studies of antiretroviral drug therapy should include the analysis of drug targets, as well as drug metabolizing enzymes and drug transporters.
In conclusion, this study provides hypotheses and scientific direction for future investigations to elucidate the genetic determinants of antiretroviral drug pharmacokinetics and response. These investigations should incorporate a prospective design, sufficient sample sizes, and intensive pharmacological methods. Ultimately, knowledge from these studies will enable the most informed and rational use of antiretroviral medications.
RESULTS
Thirty-three subjects were evaluated. Baseline demographic, pharmacokinetic, and pharmacodynamic data are presented in Table 2A. Genotype data for all subjects are shown in Table 2B. All genotypes were in Hardy-Weinberg equilibrium, and the allele frequencies were consistent with those reported in the literature. African Americans had significantly different rates of "variant-carrier" status for several genotypes compared with non-African Americans (Table 2B). No subject carried the CYP3A5*7 allele.
Indinavir Pharmacokinetic Disposition
African American Race
African Americans had 56% faster median IDV CL/F compared with non-African Americans, 1.06 (0.72 to 1.20) versus 0.68 (0.53 to 0.91) L/h/kg; P = 0.02. Therefore, given that African Americans also had significantly different variant carrier statuses (Table 2B), the genetic analyses that follow were adjusted for African American race.
CYP3A5
Eleven subjects, including 7 of 7 African Americans, were CYP3A5 expressors. One subject had *1/*3 and *1/*6; however, no haplotype analysis was done. The subject was placed in the group with the most similar CL/F values (expressor), and in addition, the data were reanalyzed with the subject removed. The median IDV CL/F was 44% faster in CYP3A5 expressors versus nonexpressors; 0.92 (0.72 to 1.20) versus 0.64 (0.50 to 0.87) L/h/kg; P = 0.002; Figure 1. The IDV CL/F was 39% faster in expressors than nonexpressors after controlling for African American race (P = 0.04). Notably, African American race was not a significant variable in this analysis (P = 0.8; Fig. 1). Similarly, CYP3A5 expressor status was also independently associated with IDV CL/F after adjusting for MDR1 2677 GG versus TT or with MRP2 C-24T variant carrier status (described below). Finally, only CYP3A5 expressor status was associated with IDV CL/F in a backward-elimination multiple regression model with all these variables included (P = 0.001).
MDR1
The median IDV CL/F was 39% faster in subjects with 2677 GG versus TT, 0.78 (0.65 to 1.00) versus 0.56 (0.50 to 0.59) L/h/kg; P = 0.02, but the IDV CL/F in subjects with 2677 GT was not statistically different than either of these 2 groups, 0.81 (0.50 to 0.92) L/h/kg. Subjects with 3435 CC versus TT genotypes had marginally faster IDV CL/F; P = 0.053. However, when 2677 GG versus TT status and African American race were simultaneously analyzed, neither was independently associated with IDV CL/F (P = 0.1 and P = 0.09, respectively).
MRP2
Neither of the 2 MRP2 variant carrier statuses was associated with IDV CL/F in univariate models (P > 0.13). When each MRP2 variant carrier status was analyzed simultaneously with African American race, C-24T variant carriers had 24% faster IDV CL/F relative to C-24T wild type (P = 0.05).
Zidovudine-Triphosphate and Lamivudine-Triphosphate Concentrations
Sex and African American Race
The median ZDV-triphosphate concentration was 230% higher in women compared with men (P = 0.003), and the median 3TC-triphosphate concentration was 160% higher in women compared with men (P = 0.002), as was reported previously.8 Three of 4 women were of African descent, which suggested possible confounding based on race. However, only sex was associated with ZDV-triphosphate and 3TC-triphosphate in a regression model with sex, study arm, and African American race analyzed together (P < 0.002). Therefore, it was concluded that sex was the most important potential confounder in the genotype analyses that follow below.
MRP4
The variant carrier status at MRP4 T4131G was associated with elevated median 3TC-triphosphate concentrations; 7396 (6515 to 8968) for wild type (TT) versus 8470 (7927 to 9386) for heterozygotes and 8882 (7315 to 9487) fmol/million cells for homozygous variants. In as much as concentrations in heterozygotes and homozygous variants were nearly identical, these 2 groups were combined. After adjusting for sex, MRP4 T4131G variant carriers had 20% higher 3TC-triphosphate concentrations than wild type (P = 0.004); Figure 2. The same trends were found with MRP4 T4131G versus ZDV-triphosphate, but the relationships were not statistically significant (P = 0.15). These relationships were consistent when triphosphate medians were recalculated with values between the assay detection limit and 0 used in place of BLQ.
The median ZDV-triphosphate concentration was 49% higher in MRP4 G3724A variant carriers versus wild type (GG); 64 (59 to 188) versus 43 (34 to 62) fmol/million cells (P = 0.03). One of the 3 variant carriers was female, and the relationship remained marginally significant when adjusted for sex (58% higher with variant; P = 0.06). The median 3TC-triphosphate concentrations were also higher in the G3724A variant carriers versus wild type (37%), but the relationship was not statistically significant (P = 0.25).
BCRP
None of the BCRP variants were associated with ZDV-triphosphate or 3TC-triphosphate concentrations in univariate analyses, or when the variants were analyzed simultaneously with sex.
Pharmacodynamics
MDR1
In an analysis of the magnitude of HIV-RNA reduction from baseline to the primary study endpoint (week 52, or study exit), subjects with 2677 TT and/or 2677 GT had significantly greater reductions in HIV-RNA compared with subjects who had 2677 GG; -3.7, -3.3, and -2.2 log10, respectively, as shown in Figure 3. The relationship was independent of baseline HIV-RNA, study arm, and concentrations of IDV, ZDV-, and 3TC-triphosphate. In a backward elimination multivariable model that included all of these variables, only G2677T variant carrier status and baseline HIV-RNA were retained in the final model (P < 0.006). Relationships were not found among any other study SNPs and HIV-RNA and CD4 responses.
Total Bilirubin Responses
Subjects with UGT1A1 [TA]7/[TA]7 had significantly greater elevations in total bilirubin concentrations during the study compared with subjects who had either [TA]6/[TA]7 or [TA]6/[TA]6; P < 0.005. The median increases were 2.5 mg/dL (2.0 to 3.2) for subjects with [TA]7/[TA]7 compared with 1.4 mg/dL (1.1 to 1.8) and 1.1 mg/dL (0.9 to 1.5) for subjects with [TA]6/[TA]7 and [TA]6/[TA]6, respectively. No other pharmacogenetic relationships were observed with bilirubin increases.
METHODS
The study was a retrospective, pilot, association study. The subjects had participated in a randomized trial of concentration controlled versus standard dosing of IDV, 3TC, and ZDV. Subjects were antiretroviral-naive HIV-infected adults with HIV-RNA in plasma above 5000 copies/mL and without active opportunistic infections. Subjects were randomized to standard dosing (IDV 800 mg every 8 hours plus twice daily 3TC 150 mg and ZDV 300 mg) or concentration controlled dosing, where doses were adjusted to attain plasma targets based on the recipient's pharmacokinetic parameters. The study details and primary results have been published elsewhere.29
All study participants underwent a full 8-hour pharmacokinetic profile at week 2, where observed doses of all study drugs were given, and serial blood samples were obtained at predose, 0.5, 1, 2, 3, 4, 5, 6, 7, and 8 hours postdose. Subjects fasted before and for 2 hours after dosing. Beginning at week 4 and continuing until week 24, a single blood sample was obtained at each monthly visit and the time postdose was recorded. Subjects could continue the study in the arm to which they were randomized for a total of 80 weeks. Intensive pharmacokinetic studies as described above were performed also at weeks 28 and 56, and single samples were obtained at 4-week intervals throughout. Intracellular 3TC-triphosphate and ZDV-triphosphate concentrations were determined in PBMCs that were harvested from 15 mL of blood taken 2 hours postdose at each intensive study (weeks 2, 28, and 56) and also at bimonthly visits (weeks 8, 16, 24, 36, 44, 52, 64, 72, and 80). A combined cartridge and competitive enzyme immunoassay was used to quantify ZDV-triphosphate, whereas a similar cartridge method with a liquid chromatography/mass spectrometry assay was used to quantify 3TC-triphosphate, as previously described.8 IDV plasma concentrations were quantified by high-performance liquid chromatography.30
A complete blood count with differential, a routine blood chemistry panel, HIV-RNA (Roche Amplicor Ultrasensitive Assay; Roche Diagnostic Systems, Branchburg, NJ), and CD4 lymphocyte counts were determined at regular intervals.
After the study was completed, data were de-identified and institutional review board approval was obtained to retrospectively investigate selected genetic polymorphisms associated with the pharmacokinetic and pharmacodynamic profiles of the study drugs. Gene selection (ie, CYP3A5, MDR1, MRP2, MRP4, BCRP, and UGT1A1) was based on literature supporting a potential connection with IDV, 3TC-, or ZDV-triphosphate disposition in man (Table 1).12,16,20,23,25 SNP selection was based on a variant allele frequency of at least 10%, and potential for change in protein function and/or expression was based on literature or, in the case of MRP4, exonic splicing enhancer sequence analysis.13,17,18,21,26,28 These considerations are described in Table 1.
Complete Genotyping Conditions Are Available in the Appendix
Briefly, genomic DNA was isolated from stored PBMCs using a commercially available kit (QIAamp DNA Mini Kit, Qiagen Inc, Valencia, CA), according to the manufacturer's protocol. MRP2, BCRP, and UGT1A1 genotypes were determined by polymerase chain reaction (PCR; MyCycler, Bio-Rad Laboratories, Inc, Hercules, CA) followed by pyrosequencing analysis. Genotyping for CYP3A5*3 and *6 was performed as described in a previous publication.13 Genotyping for CYP3A5*7, MDR1 (G2677T and C3435T), and MRP4 (C1612T, G3463A, G3724A and T4131G) was performed by PCR amplification and direct automated sequencing. Sequences were assembled using the Phred-Phrap-Consed package (University of Washington, Seattle; http://droog.mbt.washington.edu/PolyPhred.html ) that automatically detects the presence of heterozygous single nucleotide substitutions by fluorescence-based sequencing of PCR products.31,32 Genotype data were subjected to internal and external quality assurance analyses. Approximately 15% of samples were regenotyped with the same assay, and more than 15% of the samples were genotyped by both direct sequencing and pyrosequencing. In all cases, there was 100% genotype concordance within and between the methods.
Data Analysis
Weight-adjusted IDV oral clearance [CL/F (liter per hour per kilogram)] was used in the analyses to account for differences in the subject's body size and to remove the potential effects of dose changes in the concentration-controlled arm. The average CL/F was calculated from all available pharmacokinetic studies for each subject. The median 3TC-triphosphate and ZDV-triphosphate concentration was calculated for each subject. Triphosphate concentrations below the limit of quantification (due to limited cell extract and/or low values) were removed from the data set and reanalyzed using a midpoint value between the assay detection limit and 0.8
Pharmacokinetic data were log-transformed (log base 2) before data analyses, unless otherwise noted. Subjects with at least one CYP3A5 *1 allele (wild type) were defined as CYP3A5 expressors, whereas others were CYP3A5 nonexpressors.14 Given the small subject numbers and the pilot nature of the study, genotypes were analyzed as wild type versus variant carriers, where variant carriers were heterozygotes and variant homozygotes. If 3 or more subjects were variant homozygotes, these subjects were also compared with wild-type homozygotes. The following covariates were coded 0 and 1 in the data set: CYP3A5 expressors versus nonexpressors; African American versus non-African American race; women versus men; and "variant-carrier" versus "wild type." Multivariable linear regression and Student t tests with equal variances were used to evaluate relationships. No corrections were made for multiple statistical comparisons; all comparisons were planned a priori (see potential association with study drug in Table 1). Fisher exact tests were used for differences in proportions. Alpha levels were set at 0.05. SPSS software, version 12, was used (Apache Software, Chicago, IL). Data are reported as median (interquartile range), unless noted otherwise.
|
|
| |
| |
|
|
|
