Minimal effect of MDR1 and CYP3A5 genetic polymorphisms on the pharmacokinetics of indinavir in HIV-infected patients

Abstract

What is already known about this subject

• Before this study, few data were available on the potential effect of genetic variants of P-glycoprotein or the CYP3A5 enzyme on the pharmacokinetic variability of protease inhibitors (PI).

• MDR1 C3435T polymorphism was often linked with the pharmacokinetic variability of nelfinavir. CYP3A5*3 polymorphism was linked with the pharmacokinetic variability of calcineurin inhibitors and was therefore strongly suspected of being one of the key factors in the pharmacokinetic variability of other CYP3A susbtrates.

What this study adds

• Our results showed that both MDR1 C3435T and CYP3A5*3 polymorphisms are involved in the pharmacokinetic variability of the absorption or elimination of indinavir, but probably jointly with other factors.

• The potent CYP3A inhibitory effect of ritonavir may hide the variability linked to genetic differences in the CYP3A5 gene, thereby reducing the overall pharmacokinetic variability of the boosted protease inhibitor.

• Genotyping MDR1 and/or CYP3A5 does not appear to be a clinically relevant factor in optimizing protease inhibitor boosted regimens.

Aims

The protease inhibitor indinavir is characterized by an important interindividual pharmacokinetic variability, which results from the actions of the metabolizing enzymes cytochrome P450 (CYP) 3A and the multidrug efflux pump P-glycoprotein (P-gp), encoded by MDR1. Using a population pharmacokinetic approach, we investigated the effect of several MDR1 and CYP3A5 polymorphisms on the pharmacokinetic parameters of indinavir in HIV-infected patients.

Methods

Twenty-eight patients receiving indinavir alone or together with ritonavir were included. Indinavir pharmacokinetics were studied over a 12 h interval. Genetic polymorphisms were assessed by real-time PCR assays and direct sequencing for MDR1 and by PCR-SSCP analysis for CYP3A5.

Results

The pharmacokinetics of indinavir were best described by a one-compartment model with first-order absorption. In the final model, the MDR1 C3435T genotype and ritonavir were identified as statistically significant covariates (P ≤ 0.001) for the absorption rate constant (95% confidence interval on the difference between CC and CT genotype 0.37, 5.53) and for clearance (95% confidence interval on the difference 5.8, 26.2), respectively. Patients with the CYP3A5*3/*3 genotype receiving indinavir alone had a 31% decrease in the indinavir clearance rate compared with patients carrying the CYP3A5*1/*3 genotype.

Conclusions

The MDR1 C3435T genotype affects the absorption constant of indinavir suggesting that P-gp may be implicated in its pharmacokinetic variability. Through its inhibition of CYP3A and P-gp, ritonavir could attenuate the pharmacokinetic variability linked to genetic differences, reducing significantly the interindividual variability of indinavir. However, genotyping MDR1 and/or CYP3A5 to optimize protease inhibitor boosted regimens does not seem clinically relevant.

Introduction

Long-term treatment response in HIV-infected patients varies considerably and many different factors have been implicated in this variability in drug response. HIV protease inhibitors present a wide pharmacokinetic variability leading to suboptimal or elevated plasma concentrations and this contributes either to viral failure, by the emergence of drug resistant viruses, or to increased toxicity [1]. This pharmacokinetic variability has been partly attributed to the extended hepatic metabolism of the protease inhibitors through cytochrome P450, especially the CYP3A4 and CYP3A5 isoforms from the CYP3A subfamily [2–5]. CYP3A5 expression appears to be highly variable and could play a greater role than CYP3A4 in the observed variability [6]. Moreover, HIV protease inhibitors are substrates for P-gp, encoded by the multidrug-resistance transporter gene MDR1[7, 8]. P-gp is a cellular efflux pump that is present in various tissues and cells including enterocytes and T-lymphocytes and plays an important role in drug disposition and response. P-gp has been involved in the pharmacokinetic variability of protease inhibitors by limiting oral absorption and diffusion through physiological barriers such as the blood–brain barrier [9–12]. Furthermore, recent reports have also suggested that P-gp may play a role in reducing the intracellular penetration of protease inhibitors [13, 14]. Cell lines over-expressing P-gp significantly reduced the accumulation of protease inhibitors in the lymphocytes, the major site of virus replication.

The genetic heterogeneity of CYP3A and MDR1 genes has recently become a major concern as a potential cause of variability in drug disposition [15, 16]. Several single nucleotide polymorphisms (SNPs) of MDR1, CYP3A4 and CYP3A5 genes have been described. Mutations C3435T in exon 26 and G2677T in exon 21 of MDR1 have been associated with differences in P-gp expression [17–19]. The 3435TT genotype has been associated with a lower level of P-gp expression in Caucasians. However, conflicting results regarding the genotype-phenotype correlation have been noted and it is still unclear whether alteration of P-gp activity or expression is correlated with pharmacokinetic variability [20–22]. Recently, it has been demonstrated that the C3435T silent polymorphism is in linkage disequilibrium with the SNP G2677T in exon 21, which could explain the discrepancies observed in P-gp expression in previous studies [18]. In this context, considering MDR1 haplotypes rather than isolated genotypes could be useful in identifying the functional significance of MDR1 SNPs on drug disposition [23–25].

The high variability of CYP3A5 expression has been partly attributed to the frequent SNP 6986A>G within intron 3 (CYP3A5*3 allele) that causes alternative splicing and protein truncation, leading to the loss of CYP3A5 enzyme activity [6]. Another rare mutation in exon 7, SNP 14690G>A (CYP3A5*6 allele), also results in a splicing defect [6]. Individuals with the CYP3A5*3 or *6 allele are therefore classified as nonexpressors. The correlation between this polymorphism and drug concentrations has been described with the calcineurin inhibitor, tacrolimus [26, 27]. Moreover, SNPs 27050A>G in intron 10 and 31611C>T in the 3′UTR occurred with the 6986A>G mutation in intron 3 [28, 29]. All of these linkages cause functionally defective alleles. In contrast, even though many CYP3A4 alleles have been identified, no correlation was found between these mutations and drug disposition, in particular with antiretroviral drugs [30].

We hypothesized that several genetic polymorphisms of MDR1 and CYP3A5 genes could explain the variability in the pharmacokinetics of protease inhibitors. Therefore, determination of MDR1 and/or CYP3A5 genotypes before starting a regimen could be useful to individualize HIV therapy through dose adjustment.

Indinavir is an HIV protease inhibitor that is widely used in highly active antiretroviral therapy. Indinavir was first prescribed at 800 mg three times daily and later ritonavir was added, because of its ability to increase plasma concentrations of indinavir and reduce plasma half-life, allowing twice daily dosing. Indinavir is characterized by a narrow therapeutic range and by concentration-dependent renal toxicity. Indinavir plasma concentrations require monitoring to optimize the dose and reduce the risk of toxicity [31]. Therefore, the management of patients receiving indinavir therapy may be improved not only by therapeutic drug monitoring but also by determining MDR1 and CYP3A5 genotypes to detect patients at a high risk of developing toxicity or virologic failure.

The aim of this study was to investigate the effect of several common MDR1 and CYP3A5 polymorphisms on indinavir pharmacokinetic parameters in HIV-infected patients. In order to evaluate the respective role of genetic polymorphisms and ritonavir-boosting on the pharmacokinetic variability of indinavir, we assessed the pharmacokinetics of indinavir in patients treated with indinavir alone or in combination with ritonavir.

Methods

Patients

In an open-label, prospective study, we included 28 HIV-1 infected Caucasian outpatients (five women and 23 men). Inclusion criteria were Caucasian ethnicity and stable regimen including indinavir for at least 1 month. All patients received two nucleoside analogue reverse transcriptase inhibitors in addition to either indinavir alone (n = 14) or indinavir with a fixed booster dose of ritonavir (100 mg twice daily) (n = 14). The indinavir dose was chosen by the physicians according to the patient's tolerance and therapeutic history.

None of the patients had other concomitant medications affecting CYP3A and P-gp activities. Informed written consent was obtained from all of the patients and the study protocol was approved by the local Ethics Committee ‘CCPPRB-Marseille1’ (Timone Hospital, Marseille, France). Population characteristics are summarized in Table 1. All patients had an undetectable HIV-RNA on inclusion in the study. Patients were on an indinavir regimen for a median period of 26 months (3–76.5 months). All patients were contacted the day before the pharmacokinetic study to ensure that they took their antiretroviral treatment.

Table 1.

Characteristics of HIV-infected patients

Characteristics	Value (mean ± SD, range)
Number of patients	28
Sex (female/male)	5/23
Age (years)	41.0 ± 10.1 (27–77)
Body weight (kg)	63.5 ± 9.6 (43.8–81)
Height (cm)	171 ± 7.5 (150–182)
CD4 cells (/mm3)	552 ± 186 (144–919)
HIV RNA (copies ml−1)	<400*

^*all patients had undetectable HIV RNA.

Indinavir plasma concentration and population pharmacokinetic analysis

Blood samples were drawn at 0 (predose), 0.5, 0.75, 1, 1.5, 2, 3, and 12 h after a witnessed dose of indinavir. Indinavir plasma concentrations were assessed at steady-state by a validated high-performance liquid chromatography assay [32]. The limit of quantification was 20 ng ml⁻¹ and interassay variability ranged from 3.55% to 13.6%.

Data were analyzed with the nonlinear mixed-effect modelling software program NONMEM (version V, level 1.1, double precision) using the first-order estimation procedure [33]. Indinavir plasma concentrations were fitted to one- and two-compartment pharmacokinetic models with first-order absorption and elimination, by use of the NONMEM subroutine ADVAN2 and ADVAN4, respectively. Interindividual variability of the different pharmacokinetic parameters was estimated with a proportional error model. Residual variability was estimated with a mixed model, additive and proportional. The performance of the model was judged by both statistical and graphic methods [34]. The minimal value of the objective function as calculated by NONMEM was also used to assess the goodness-of-fit. An increase in goodness-of-fit is accompanied by a decrease in objective function, and this decrease is asymptotically distributed as a chi square distribution. Furthermore, standard errors were calculated by use of the COVARIANCE option of NONMEM. For graphic model diagnostics, the following graphs were compared: observed concentrations (DV) vs. predictions (PRED), weighted residuals (WRES) vs. time, weighted residuals vs. PRED, and individual predictions (IPRED) vs. DV.

A first analysis was performed to find the base model that best described the data. Once it has been defined, the influence of each covariate on the pharmacokinetic parameters was tested. These covariates were: ritonavir combination (with or without), MDR13435 genotype (SNP 3435C>T of exon 26), MDR1 2677 genotype (SNP 2677G>T of exon 21), CYP3A5 6986A>G genotype (SNP 6986 A > G in intron 3), CYP3A5 27050A>G genotype (SNP 27050A>G in intron 10) and CYP3A5 31611C>T genotype (SNP 31611C>T in the 3′UTR).

The diagnostic plots described above, the change in objective function, and the change in parameter variability were noted to select those which improved the model prediction. A decrease in the objective function value of at least 6.61 (chi-squared distribution with one degree of freedom for P < 0.01) relative to the base pharmacokinetic model was required for the addition of a single parameter in the model. The model with one covariate, which best described the data, was defined as the intermediate model. The covariates that significantly reduced the objective function at the previous step were then combined in a stepwise fashion with the intermediate model until no further improvement of the objective function occurred (full model). Model robustness was then evaluated by a backward elimination procedure where each covariate was removed in turn from the full model. An increase in objective function greater than 10.82 (P< 0.001) was required to keep the covariate in the final model.

Identification of MDR1 and CYP3A5 polymorphisms

Genomic DNA was extracted from whole blood using the QIAamp DNA Blood Minikit (Qiagen, Courtaboeuf, France). Table 2 shows the sequences of the primersused for identification of MDR1 and CYP3A5 polymorphisms.

Table 2.

Sequences of primers used for determination of MDR1 and CYP3A5 polymorphisms

SNP	Primer sequences
MDR1 exon 26	Forward: 5′-GAGCCCATCCTGTTTGACTG-3′
3435C>T	Reverse: 5′-TCGATGAAGGCATGTATGTTG-3′
MDR1 exon 21	Forward: 5′-GGTTCCAGGCTTGCTGTAAT-3′
2677G>T/A	Reverse: 5′-AAAAGATTGCTTTGAGGAATGG-3′
CYP3A5 intron 3	Forward: 5′-CTTTAAAGAGCTCTTTTGTCTCTCA-3′
	Reverse: 5′-CCAGGAAGCCAGACTTTGAT-3′
CYP3A5 exon 7	Forward: 5′-ATAGTGGAAGGACGGTAAGAG-3′
	Reverse: 5′-GTGGATGAATTATACGATATGTG-3′
CYP3A5 intron 10	Forward: 5′-TGAGTTATTCTCTGGAGCTTC-3′
	Reverse: 5′-AGGCTCTGTCCAGTACTTTG-3′
CYP3A5 exon 13 (3′UTR)	Forward: 5′- ACTTTGCTTCCATCTTTTCTTC-3′
	Reverse: 5′- TATTGACTAAGTTGAAATCTCTG-3′

MDR1 C3435T polymorphism in exon 26 was analyzed by real-time PCR assays in a LightCycler^TM (Roche Diagnostic, Manheim, Germany). Fluorescent detection was performed with hybridization probes 5′-GCT GCC CTC ACA ATC TCT TCC (sensor) and 5′-LC Red640-TGA CAC CAC CCG GCT GTT GTC TCC A (anchor) (Tib Molbiol, Berlin, Germany). MDR1 G2677T/A polymorphism in exon 21 was obtained by direct sequencing of purified PCR products with the Dye Terminator Cycle Sequencing reaction kit on a Beckman Coulter CEQ 8000 sequencer (Beckman Coulter, France). For both exon 26 and 21, PCRs were performed in a total volume of 20 µl, using 200 ng of genomic DNA, 0.5 µm each primer (Table 2), 0.2 µm of deoxyribonucleoside triphosphate, 1X PCR buffer, 2 mm MgCL₂ and 0.16 U of Taq polymerase (Eurobio, France).

PCR followed by SSCP analysis was performed to genotype CYP3A5 as previously described with only slight modifications [35]. Details concerning the conditions for PCR and SSCP analysis are available on request from the authors. SSCP analysis was performed on a GenePhor Electrophoresis unit, using GeneGel Excel 12.5/24 or GeneGel Excel 15/24 kits (Amersham-Pharmacia Biotech, Uppsala, Sweden). The CYP3A5 polymorphisms analyzed included 6986A>G in intron 3, 14690G>A in exon 7, 27050A>G in intron 10 and 31611C>T in the 3′UTR.

Haplotype analysis

Haplotype analysis was restricted to the two MDR1 SNPs 3435C>T and 2677G>T on the basis of the linkage disequilibrium observed between both positions. The different haplotypes were defined as previously described [25]. Nine genotypes or haplotype pairs of MDR1 can be described with exclusion of the rare 2677A variant. Comparisons were performed between carriers and non carriers for each haplotype.

Results

Frequencies of MDR1 and CYP3A5 variants in HIV-infected patients

Table 3 shows the frequency of the different MDR1 and CYP3A5 polymorphisms. MDR1 SNPs frequency in Caucasian patients is consistent with previously published data [17, 36, 37]. The linkage disequilibrium reported for both 3435C>T and 2677G>T polymorphisms of MDR1 was also confirmed in our study population, with only five patients differing. The rare 2677A allele was not observed in our population.

Table 3.

Frequency of MDR1 and CYP3A5 polymorphisms in 28 HIV-infected patients

		Genotype % (n)
Gene	Polymorphism	wt/wt	wt/m	m/m
MDR1	3435C>T exon 26	28.5 (8)	53.5 (15)	18 (5)
	2677G>T exon 21	39 (11)	50 (14)	11 (3)
CYP3A5	6986A>G intron 3	0	36 (10)	64 (18)
	14690G>A exon 7	93 (26)	7 (2)	0
	27050A>G intron 10	79 (22)	21 (6)	0
	31611C>T exon 13 (3'UTR)	79 (22)	21 (6)	0

wt, wild-type; m, mutated.

For the CYP3A5 gene, the CYP3A5*3 allele (or 6986G) occurred with a frequency of 82% and was therefore consistent with other data in the Caucasian population [29]. The linkage described between CYP3A5 SNPs 27050A>G in intron 10, 31611C>T in the 3′UTR and 6986A>G in intron 3 was confirmed in our study. Among the nine patients heterozygous for either 27050A>G or 31611C>T mutations, two were also homozygous for 6986A>G, four were heterozygous for this SNP and three were heterozygous for the three SNPs. None of the patients was carrying the CYP3A5*6/*6 genotype (homozygous mutated for the SNP 14690G>A), confirming its very low frequency in the Caucasian population.

Population pharmacokinetic analysis

Patients received different indinavir doses in each group. In the group of patients without ritonavir, the indinavir dose was 800 mg three times daily (n = 10) except for four patients who received either 800 mg twice daily (n = 2), 600 mg three times daily (n = 1) or 400 mg three times daily (n = 1). These atypical doses of indinavir resulted from a previous dose adjustment due to intolerance. In the group of patients with ritonavir, the indinavir dose was 400 mg twice daily (n = 11), 400 plus 200 mg (n = 2) and 200 mg twice daily (n = 1).

Figure 1 shows the observed concentrations of indinavir (log-transformed) as a function of time for the two groups of patients without (A) and with ritonavir (B). The indinavir pharmacokinetic parameters estimated were oral clearance, volume of distribution and absorption rate constant (k_a). The one compartment model gave a better description of the data and thus was kept as the base model. The posthoc of the pharmacokinetic parameters obtained from the base model was used to investigate a link with the covariates.

Figure 1.

Observed indinavir plasma concentration vs. time profiles of HIV-infected patients (A) without RTV and (B) with RTV 100 mg twice daily. Indinavir concentrations were log-transformed. IDV = indinavir

A summary of covariate model development is presented in Table 4. Ritonavir was the covariate which best improved the model and was chosen to define the intermediate model. Of all the SNPs studied, only the CYP3A5*1/*6 genotype (SNP 14690G>A in exon 7) could not be tested in the population pharmacokinetic model because of the low occurrence of the mutation (2/28 patients were heterozygous). In the final model, ritonavir combination was included as a statistically significant covariate of clearance and MDR13435 genotype as a statistically significant covariate of k_a. The pharmacokinetic parameters of the final model are presented in Table 5. Indinavir clearance was significantly reduced by 38% in the presence of ritonavir (P< 0.001). The difference in clearance observed between both groups was 16.0 l h⁻¹ (95% confidence interval 5.8, 26.2). k_a was significantly higher in patients heterozygous for the SNP C3435T of MDR1 compared with patients with the wild-type genotype: 5.91 vs. 2.96 (P< 0.001), with a difference of 2.95 1 h⁻¹ (95% confidence interval 0.37, 5.53).

Table 4.

Summary of models evaluating the effect of covariates on the pharmacokinetic parameters of indinavir

		Run model	Objective function value	DOFa
1	Base model		2963	–
Intermediate model
CL/F
2	with and without ritonavir*	θ1 × RTV + (1 − RTV) θ6	2940	22
3	MDR1 3435C>T	θ1 × 3435CC + θ6 × 3435CT + θ7 × 3435TT	2956	7
4	MDR1 2677G>T	θ1 × 2677GG + θ6 × 2677GT + θ7 × 2677TT	2962	1
5	CYP3A5 *1/*3 (6986A>G)	θ1 × 6986AG + θ6 × 6986GG	2962	1
6	CYP3A5 27050A>G	θ1 × 27050AA + θ6 × 27050AG	2963	0
7	CYP3A5 31611C>T	θ1 × 31611CC + θ6 × 31611CT	2962	1
ka
8	MDR1 3435C>T	θ3 × 3435CC + θ6 × 3435CT + θ7 × 3435TT	2945	17
9	MDR1 2677G>T	θ3 × 2677GG + θ6 × 2677GT + θ7 × 2677TT	2957	5
10	CYP3A5 *1/*3 (6986A>G)	θ3 × 6986AG + θ6 × 6986GG	2962	1
11	CYP3A5 27050A>G	θ3 × 27050AA + θ6 × 27050AG	2963	0
12	CYP3A5 31611C>T	θ3 × 31611CC + θ6 × 31611CT	2960	3
V/F
13	MDR1 3435C>T	θ2 × 3435CC + θ6 × 3435CT + θ7 × 3435TT	2962	1
14	MDR1 2677G>T	θ2 × 2677GG + θ6 × 2677GT + θ7 × 2677TT	2962	0
15	CYP3A5 *1/*3 (6986A>G)	θ2 × 6986AG + θ6 × 6986GG	2962	1
16	CYP3A5 27050A>G	θ2 × 27050AA + θ6 × 27050AG	2958	5
17	CYP3A5 31611C>T	θ2 × 31611CC + θ6 × 31611CT	2962	0
Final model
18	CL/F (RTV) and ka (MDR-1 3435)		2927	14**

^adecrease of objective function

^*best intermediate model

^**DOF from best intermediate model; CL/F clearance; V/F volume of distribution; k_a,rate of absorption constant.

Table 5.

Indinavir population pharmacokinetic parameters of the final model

	Estimation
Parameter	Final estimate	SE of the estimate	95% CI	95% CI on the difference
CL/F (l h−1)
With RTV	26.3	2.85	20.6, 32.0	5.8, 26.2
Without RTV	42.3	4.33	33.6, 51.0
V/F (l)	69.9	5.84	58.2, 81.6
ka (h−1)
MDR1 3435CC	2.96	0.94	1.08, 4.84	0.37, 5.53*
MDR1 3435CT	5.91	0.92	4.07, 7.75
MDR1 3435TT	3.22	1.12	0.98, 5.46
Intersubject variability (ω2) (% coefficient of variation)
ω2 (CL/F)	0.19 (43%)	0.06	0.07, 0.31
ω2 (V/F)	0.31 (55%)	0.11	0.09, 0.53
ω2 (ka)	6.7 (258%)	5.83	4.96, 18.36
Residual variability (σ2)
σ2 proportional	0.18 (42%)	0.02	0.14, 0.22
σ2 additive (ng ml−1)	198	35.1	127.8, 268.2

CL/F clearance; V/F volume of distribution; k_arate of absorption constant; 95% CI 95% confidence interval

^*95% CI on the difference between CC genotype and CT genotype

At the same time, we analyzed the effect of the haplotypes of MDR1, which combine the effect of the two SNPs and are thought to be more predictive of pharmacokinetic variability of drugs. We described five haplotype pairs among the 28 patients. Haplotypes 11 (C-G) and 22 (T-T) were present, at least in one allele, in 23 and 17 patients, respectively. None of the haplotypes significantly decreased the objective function or improved the model fit (data not shown).

Influence of MDR1 and CYP3A5 polymorphisms on groups with or without ritonavir

In patients receiving indinavir alone, we observed a 31% decrease in clearance in patients carrying the CYP3A5*3/*3 (6986GG) genotype compared with patients carrying the CYP3A5*1/*3 (6986AG) genotype: 35 l h⁻¹vs. 50.4 l h⁻¹. No difference in clearance between the different genotypes was reported, especially for CYP3A5 6986A>G SNP, in patients treated with indinavir plus ritonavir (19.6 l h⁻¹ for genotype *3/*3 vs. 21.9 l h⁻¹ for genotype *1/*3). Neither the k_a nor the volume of distribution was affected by the different CYP3A5 polymorphisms studied, whatever the treatment prescribed.

Discussion

We have investigated the impact of several polymorphisms of the MDR1 and CYP3A5 genes on the pharmacokinetic parameters of indinavir in HIV-infected patients. Identification of a correlation between drug pharmacokinetics and genetic differences could provide a useful tool to individualize HIV therapy through dose adjustment, thus reducing toxicity or virologic failure. The MDR1 SNP C3435T was the only genetic factor associated with a statistically significant difference in the pharmacokinetics of indinavir in HIV-infected patients. Despite the small size of the study population, we showed that patients with the MDR1 3435CT genotype had a significantly higher k_a (two-fold increase) compared with patients carrying the 3435TT genotype; patients with the 3435CC genotype showed a similar profile to those with the 3435TT genotype. As expected, ritonavir was identified as a significant covariate of indinavir clearance and this effect was taken into account in the final model [38]. This means that the effect of MDR1 C3435T polymorphism on the k_a of indinavir occurred whatever the treatment arm, with or without ritonavir.

P-gp, widely expressed in the small intestine and liver, plays a significant role in drug disposition. In humans, the role of P-gp in drug absorption is more difficult to prove. Two studies reported that co-administration of ciclosporin, a P-gp inhibitor, enhances the oral bioavailability of the anticancer drugs paclitaxel and docetaxel [39, 40]. Among protease inhibitors, ritonavir has been described as an inducer and inhibitor of P-gp [41, 42]. Even though P-gp inhibition is less well described than for the CYP3A enzyme, an inhibitory effect of ritonavir could have consequences in terms of treatment efficacy and toxicity. The variability observed in P-gp expression could therefore also alter the extent of absorption and elimination of protease inhibitors. SNP C3435T was first associated with variability in P-gp expression. Patients carrying the 3435TT genotype displayed a lower P-gp expression leading to increased digoxin plasma concentrations [17, 37, 43]. However, these results have been contradicted by other recent studies showing either no difference or a lower concentration for the 3435TT genotype [18, 44]. Fellay et al.[30] reported that patients with the 3435CC genotype presented with higher nelfinavir plasma concentrations than patients with the 3435CT and 3435TT genotypes, although the latter group had a lower level of P-gp expression. A linkage disequilibrium between SNPs C3435T and G2677T has been suggested as a possible cause for these conflicting results [18]. G2677T polymorphism may contribute to the observed variability in P-gp expression. Other recent studies have highlighted the importance of considering the haplotype of MDR1 rather than the isolated genotype to predict drug variability [25]. Unfortunately, neither the G2677T polymorphism nor the MDR1 haplotypes predicted the variability observed in the pharmacokinetic parameters of indinavir in this study. This has also been demonstrated in an in vitro genotype-phenotype study and in vivo in healthy volunteers and in patients receiving tacrolimus [21, 45–47].

Surprisingly, in our study, it was the heterozygous variant of MDR1 C3435T which showed the highest k_a value, and not the mutated 3435TT genotype, which is supposed to have the lowest level of P-gp expression. However, in our study, patients with the 3435TT genotype had a significantly lower k_a and similar median MDR1 mRNA concentration (1.45 vs. 1.56, data not shown) than patients with the 3435CT genotype. The limited number of patients carrying the 3435TT genotype (n = 5) may explain the lack of a genotype-phenotype correlation. A recent study reported a similar profile with nelfinavir in HIV-infected children [48]. Carriers of the 3435CT genotype had higher 8 h postdose concentrations and lower clearance rates than patients with the CC or TT genotypes, who had similar pharmacokinetic values and a similarly limited number of patients.

In summary, the involvement of P-gp in drug absorption is clear, but the usefulness of MDR1 polymorphisms to predict drug variability appears to be less certain. Caution should be taken when interpreting the data because of the likely impact on the transporter function regulation both of the drugs themselves and of the progression of the disease. These factors could contribute to the dissimilar results between patients and healthy volunteers observed in studies. The implication of other transport processes such the multidrug resistance related-protein 1 (MRP-1) or the organic anion transporting polypeptide (OATP) transporter family could mask or modify a potential P-gp effect. Last but not least, the extensive hepatic metabolism of indinavir through the CYP3A isoenzyme enhances drug variability and could partly hide the effect of P-gp.

In the present study, we analyzed several CYP3A5 polymorphisms, two of which (6986A>G and 14690G>A) led to nonexpression of the CYP3A5 protein. Although none of the CYP3A5 polymorphisms was significantly associated with the pharmacokinetics of indinavir, a 31% decrease in clearance was observed in patients receiving indinavir alone, who were carriers of the CYP3A5*3/*3 (6986GG, ‘non expressors’) genotype. Interestingly, this decrease was not observed in patients receiving indinavir together with ritonavir. This observation is the first reported with protease inhibitors. In two previous studies in HIV-infected adults and children, no significant association was found between CYP3A5*3 polymorphism and the pharmacokinetics of nelfinavir [30, 48]. In contrast, in renal transplant recipents, carriers of the CYP3A5*3/*3 genotype had higher dose-adjusted concentrations both for tacrolimus and ciclosporin [26, 27]. Unfortunately, our results did not reach significance, probably due to the limited number of patients in the group without ritonavir (n = 14, 10 CYP3A5*3/*3 and four CYP3A5*1/*3). The significance of the discrepancies observed between studies on the role of P-gp and/or CYP enzymes in drug variability underlines the need to pay close attention to study design, sample size, drug association and pharmacokinetic evaluation. In our study, pharmacokinetic measurements were assessed at steady state using seven samples per patient. Many studies have assessed pharmacogenetic correlations on a single drug concentration value. Although our study was limited because of the small number of patients, it was the first report evaluating the impact of MDR1 and CYP3A5 polymorphisms using a population pharmacokinetic approach.

The impact of ritonavir on indinavir clearance was much higher than genetic factors. Among the protease inhibitors, indinavir possesses good oral bioavailability but has an extensive hepatic metabolism. The role of ritonavir as a booster is well known and it is widely used in HAART to improve virologic response by avoiding suboptimal concentrations, leading to the appearance of resistance [38]. As expected, the decrease in oral clearance by ritonavir was observed in our study between the two groups of patients. However, the most important observation concerns the lower clearance reported only in patients receiving indinavir alone and carrying the CYP3A5*3/*3 genotype. Indeed, this suggests that without ritonavir, indinavir plasma concentrations may differ between patients according to their genetic status, leading to differences in either drug response or toxicity. In contrast, when ritonavir is added, interpatient variability in indinavir plasma concentrations is lower and genetic differences do not appear as a significant factor in this variability. This is particularly interesting insofar as indinavir is no longer administered without ritonavir. It emphasizes the benefits of the ritonavir boost.

In conclusion, the effect of the MDR1 3435CT genotype on the k_a of indinavir suggests that P-gp could play a role in the absorption of indinavir and probably of other protease inhibitors. While genetic factors such as MDR1 C3435T or CYP3A5*3 polymorphisms contribute slightly to the variability observed in the pharmacokinetics of protease inhibitors, other known or unknown environmental factors probably have a more significant role in this drug variability. Moreover, the high CYP3A5-inhibition potency of ritonavir could hide the variability caused by genetic differences and thus reduce the overall pharmacokinetic variability of the protease inhibitor added [49]. Therefore, genotyping MDR1 and/or CYP3A5 does not appear to be a clinically relevant factor in optimizing protease inhibitor-boosted regimens. On the other hand, using therapeutic drug monitoring to evaluate the overall pharmacokinetic variability of indinavir in patients seems more appropriate, allowing rapid dose adjustments according to the patient's tolerance and virologic response. This also improves the management of physiologically unpredictable drug–drug interactions, which occur throughout the patient's life and may seriously affect plasma concentrations of the protease inhibitors.

This work was supported in part by grants from Merck Sharp Dohme Laboratories and from the Direction de la Recherche Clinique de l'AP-HM (AORC n°38/2001).