Inhibition of placental P-glycoprotein: impact on indinavir transfer to the foetus

Abstract

AIMS

To investigate the effect of P-gp inhibition on the maternal to foetal transfer of indinavir.

METHODS

Term human placentae (n = 12) were from non-HIV infected women. Maternal to foetal transfer of indinavir was examined in the absence and presence of P-gp inhibitors PSC833 (n = 7) or ritonavir (n = 5), in the perfused human placenta. Antipyrine and [³H]-vinblastine were included as markers of passive diffusion and P-gp transport, respectively. These markers and indinavir were added to maternal perfusate at 0 min; PSC833 or ritonavir was added at 25 min. Steady-state maternal to foetal transfer clearance was calculated during control and inhibitor phases. Indinavir and vinblastine clearances were normalized to antipyrine clearance (clearance index).

RESULTS

Indinavir clearance index increased between the control (0.25 ± 0.03) and PSC833 phases (0.37 ± 0.14) (95% CI of the difference −0.23, −0.002). Vinblastine clearance index increased from (0.25 ± 0.08) to (0.34 ± 0.06) in the control and PSC833 phases, respectively (95% CI of difference −0.14, −0.05). Indinavir clearance index was unchanged between control (0.34 ± 0.14) and ritonavir phases (0.39 ± 0.13) (95% CI of the difference −0.19, 0.08). Vinblastine clearance index increased from (0.24 ± 0.12) to (0.32 ± 0.12) in the control and ritonavir phases, respectively (95% CI of the difference −0.15, −0.009).

CONCLUSIONS

Maternal to foetal transfer clearance of indinavir and vinblastine increased following P-gp inhibition. The potential role for co-administration of P-gp inhibitors with PIs to reduce perinatal HIV transmission warrants further investigation.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• We have shown previously using the dually perfused isolated human placenta model that the maternal to foetal transfer of the antiviral protease inhibitor drug indinavir is substantially lower than the transfer in the opposite direction.

• This finding is not consistent with passive diffusion and indicates that a carrier-mediated mechanism is involved in retarding the movement in the maternal to foetal direction.

• The efflux transporter P-gp located in the apical membrane domain of the placental trophoblast cells has been implicated as the likely cause of the differential bi-directional transport.

WHAT THIS STUDY ADDS

• The present study also utilizes the human perfused human isolated placenta to investigate the possible inhibitory effects of the P-gp inhibitor PSC833 and the P-gp substrate/inhibitor ritonavir on the maternal to foetal transfer clearance of indinavir.

• The studies, which were conducted such that each placenta served as its own control, demonstrated a statistically significant increase in the maternal to foetal transfer of indinavir in the presence of PSC833 but not in the presence of ritonavir, a protease inhibitor that is often used in combination with other protease inhibitors in dual therapy.

• The lack of effect of ritonavir is most likely related to the relatively low inhibitory activity at the clinically relevant concentration used in this study.

Introduction

Mother to child transmission of HIV is the principal source of childhood HIV infections [1]. Administration of highly active antiretroviral therapy with protease inhibitors (PIs) during pregnancy, has decreased mother to child transmission of HIV to less than 2% [1]. However, as 17 million women are currently HIV infected [2], it is important to decrease further mother to child transmission of HIV. Anti-retroviral prophylaxis in the foetus is recognized to be important, and therefore preloading the foetus with potent antiretrovirals such as the PIs, may provide valuable protection against mother to child transmission of HIV [3, 4]. For this strategy to be successful, it is essential to understand the placental transfer of PIs and the factors that modulate the transfer.

Although the PIs are routinely used in pregnancy [1], there are limited data on the mechanisms and extent of placental transfer. When PIs are administered during pregnancy, the concentrations in foetal plasma at the time of delivery are low in comparison with the corresponding concentrations in maternal plasma, suggesting a low extent of transfer from mother to foetus [5]. Previous studies in the isolated perfused human placenta revealed low maternal to foetal transfer of ritonavir [6, 7], amprenavir [8] and saquinavir [9, 10], and relatively low transfer of lopinavir [7]. Recently, our group investigated the bidirectional transfer of indinavir in the isolated perfused human placenta [11]. Indinavir transfer was three-fold lower in the maternal to foetal compared with foetal to maternal direction, indicating passive diffusion was not the sole mechanism of placental transfer [11]. Several transport systems, including P-glycoprotein (P-gp), have been identified in placental trophoblast cells [12]. P-gp is strategically localized to the apical (maternal facing) membrane, and active efflux by P-gp appears to restrict the maternal to foetal transfer of xenobiotics [12]. The PIs are substrates for P-gp [13], and the observed differential transfer of indinavir across the placenta [11], is consistent with P-gp limiting indinavir transfer to the foetus.

The ciclosporin analogue, PSC833, is a P-gp inhibitor [13]. Studies in mice revealed co-administration of PSC833 increased the transfer of PIs across the gastrointestinal [14], blood-brain [14, 15], and blood-testis [15] barriers. In a study in pregnant mice, saquinavir concentrations were seven-fold higher in mdr1a/1b(–/–) compared with mdr1a/1b(+/+) foetuses [3]. In addition, co-administration of PSC833 to pregnant dams increased saquinavir concentrations in mdr1a/1b(+/+) foetuses, similar to those observed in mdr1a/1b(–/–) foetuses [3]. The PI ritonavir has been shown to inhibit P-gp and increase the uptake of saquinavir in cell culture [16, 17]. Furthermore, Van Praag et al. reported that following addition of ritonavir to indinavir therapy in HIV-infected men, concentrations of indinavir increased approximately two-fold in cerebrospinal fluid and eight-fold in seminal plasma, findings that were not fully explained by increased plasma indinavir concentrations [18]. This is consistent with ritonavir inhibiting P-gp at the blood-brain and blood–testis barriers in humans.

The aim of the present study was to investigate the effect of PSC833 and ritonavir on the maternal to foetal transfer of indinavir in the human isolated perfused placenta; to substantiate the purported role of P-gp in the differential placental transfer of indinavir observed previously in the human placenta [11].

Methods

Materials

Indinavir sulphate was generously donated by Merck Research Laboratories (New Jersey, USA). PSC833 (valspodar) and ritonavir were provided by Novartis (Basel, Switzerland) and Abbott Laboratories (Illinois, USA), respectively. Analytical grade antipyrine, dextran 70, tetramethylammonium perchlorate, trifluroacetic acid and verapamil hydrochloride were purchased from Sigma-Aldrich (Castle Hill, Australia). [³H]-Vinblastine (specific activity, 1.85 GBq mmol⁻¹) was from Amersham Biosciences (Castle Hill, Australia), d-glucose from BDH Chemicals (Poole, England) and methanol from Biolab (Melbourne, Australia). All other chemicals were analytical grade or equivalent.

Patients and clinical study protocol

The study protocol was approved by the Ethics Committees of the Royal Women's Hospital and Monash University. Written informed consent was obtained from all participants prior to enrolment. Placentae were obtained from women who had uncomplicated, term pregnancies following normal vaginal deliveries or elective Caesarean sections. Women were excluded if they had HIV/AIDS or any chronic diseases, or were treated with indinavir or other chronic drug therapy. Prior to perfusion, placentae were carefully inspected and those with visible tears were discarded.

Placental perfusion technique

The dual in vitro perfusion of the isolated human placenta was adopted from the method of Penfold et al.[19], as described previously [11]. Human placentae were perfused using protein-free Krebs-Ringer (Henseleit) solution (pH 7.4) with the addition of 3% dextran and 0.1% glucose [19]. Maternal and foetal flow rates were approximately 14 and 7 ml min⁻¹, respectively. During the initial stabilization period (20 min), the maternal circuit was operated in an open, single-pass mode, while the foetal perfusate was recirculated. The viability of the placenta was assessed during this time by monitoring pressure in the foetal circuit and loss of perfusate from the foetal reservoir. The placental preparation was discarded if there was a change in pressure (>10 mmHg) or a loss of perfusate (>3ml h⁻¹). Temperature and pressure were monitored continuously during the perfusion.

Study design

Following the stabilization period, both the maternal and foetal circuits were operated in an open, single-pass mode for the remainder of the perfusion. In each placenta, the maternal to foetal transfer clearance of indinavir was examined in the presence of antipyrine (marker of passive diffusion) [20], and [³H]-vinblastine (marker of P-gp mediated transport) [21]. A clinically relevant concentration of indinavir (7.60 mg l⁻¹) was added to maternal perfusate, corresponding to plasma unbound concentrations [22]; the concentration of antipyrine was 20.0 mg l⁻¹ and vinblastine 0.03 mg l⁻¹ (5.00 μCi l⁻¹). In all perfusions, the three analytes were added to the maternal reservoir (0 min); the foetal inflow perfusate was drug-free.

The maternal to foetal transfer of indinavir, vinblastine and antipyrine in the absence and presence of P-gp inhibitors, PSC833 (n = 7) or ritonavir (n = 5), was determined. A concentration of PSC833 specific for P-gp inhibition in vitro (1.21 mg l ⁻¹ = 1.00 μM) [23], and a clinically relevant concentration of ritonavir (0.22 mg l ⁻¹ = 0.40 μM) [24] were selected. The inhibitors were added to the maternal perfusate to mimic the situation in vivo in which the P-gp inhibitor would be administered to the mother.

In each placenta, the maternal to foetal transfer of indinavir, vinblastine and antipyrine was examined alone for 25 min (control phase). At the end of the control phase, the P-gp inhibitor (PSC833 or ritonavir) was introduced, and the maternal to foetal transfer of indinavir, vinblastine and antipyrine was examined for a further 25 min (inhibitor phase); hence, each placenta was used as its own control. Maternal and foetal perfusate samples (3 ml) were collected at 3 min intervals to determine the maternal to foetal transfer clearance at steady state during the control and inhibitor phases. Maternal and foetal perfusate samples were stored at −20°C until the time of analysis.

Drug analysis

Perfusate indinavir concentrations were determined using a validated HPLC assay [11]. Calibration standards (0.10–10.0 mg l⁻¹) and quality control (QC) samples (0.25, 1.00 and 10.0 mg l⁻¹) were used in each analytical run. The measured concentrations of the QC samples were within 15% of the nominal concentrations, and reproducibility was within 15% as assessed by coefficients of variation at 1.00 mg l⁻¹ and 10.0 mg l⁻¹. For the low QC, the corresponding values were within 20%. The P-gp inhibitors, PSC833 and ritonavir, did not interfere with the chromatography of indinavir or the internal standard, verapamil. Perfusate antipyrine concentrations were determined using a validated HPLC assay [25]. The measured concentrations of the QC samples (0.50, 5.00 and 20.0 mg l⁻¹) were within 10% of the nominal concentrations and reproducibility was also within 10% for all QC samples. The chromatography of antipyrine and internal standard, phenacetin, was not influenced by PSC833 or ritonavir. [³H]-Vinblastine concentrations in perfusate were analyzed using liquid scintillation counting (Packard Tricarb 1900CA scintillation counter).

Pharmacokinetic analysis

The maternal to foetal transfer clearance at steady state (CL_M→F) during the control (19, 22, 25 min) and inhibitor (44, 47, 50 min) phases for indinavir, antipyrine and vinblastine were calculated as described previously [11] (Equation 1).

(1)

where F_out is foetal outflow concentration, Q_f is foetal perfusate flow rate, and M_in is maternal inflow concentration of the respective analyte. The CL_M(r)F-values of indinavir or vinblastine during the control and inhibitor phases were normalized to the respective CL_M(r)F of antipyrine, and termed the clearance index (Equation 2).

(2)

Statistical analysis

Group data are presented as mean ± SD. Differences between the maternal to foetal transfer clearance of antipyrine in the first and second 25 min phases, and differences between the maternal to foetal clearance indices in the control and inhibitor phases for indinavir and vinblastine were assessed using the paired Student's t-test; P < 0.05 was regarded as statistically significant. Statistical analyses were performed using SPSS for Windows version 11.5 (SPSS Inc., Chicago, Illinois).

Results

In preliminary perfusions to assess the viability of placentae over the perfusion duration in the absence of PSC833 or ritonavir, the mean clearance index of indinavir in the first 25 min (0.30 ± 0.11) was not different from that in the following 25 min (0.31 ± 0.13) (95% CI of the difference −0.15, 0.12; P = 0.49). The mean clearance index of vinblastine in the first 25 min (0.25 ± 0.13) vs. the following 25 min (0.26 ± 0.18) was also not different (95% CI of the difference −0.39, 0.37; P = 0.77). These results indicate the constancy of the maternal to foetal transfer clearances of indinavir and vinblastine, and the viability of the preparation across the 50 min perfusion period.

In the presence of the P-gp inhibitor, PSC833, the foetal outflow concentrations of indinavir and vinblastine increased significantly. The mean clearance index of indinavir increased from the control (0.25 ± 0.03) to PSC833 (0.37 ± 0.14) phases (95% CI of the difference −0.23, −0.002; P = 0.047) (Figure 1A). Similarly, vinblastine mean clearance index increased between the control (0.25 ± 0.08) and PSC833 phases (0.34 ± 0.06) (95% CI of the difference −0.14, −0.05; P < 0.003) (Figure 1B).

Figure 1.

Clearance indices of (A) indinavir and (B) vinblastine in the control (mean of values at 19, 22 and 25 min) and PSC833 (mean of values at 44, 47 and 50 min) phases. The lines join data from the same placenta (n = 7). Mean clearance index of indinavir in the control (0.25 ± 0.03) and PSC833 (0.37 ± 0.14) phases (95% CI of the difference −0.23, −0.002, P = 0.047). Mean clearance index of vinblastine in the control (0.25 ± 0.08) and PSC833 (0.34 ± 0.06) phases (95% CI of the difference −0.14, −0.05, P < 0.003)

The effect of ritonavir (P-gp inhibitor) on the foetal outflow concentrations of indinavir and vinblastine was also assessed in other placentae. The mean clearance index of indinavir was not significantly different between the control (0.34 ± 0.14) and ritonavir phases (0.39 ± 0.13) (95% CI of the difference −0.19, 0.08; P = 0.31) (Figure 2A). In contrast, the mean clearance index for vinblastine increased from the control (0.24 ± 0.12) to ritonavir phases (0.32 ± 0.12) (95% CI of the difference −0.15, −0.009; P = 0.04) (Figure 2B).

Figure 2.

Clearance indices of (A) indinavir and (B) vinblastine in the control (mean of values at 19, 22 and 25 min) and ritonavir (mean of values at 44, 47 and 50 min) phases. The lines join data from the same placenta (n = 5). Mean clearance index of indinavir in the control (0.34 ± 0.14) and ritonavir (0.39 ± 0.13) phases (95% CI of the difference −0.19, 0.08, P = 0.31). Mean clearance index of vinblastine in the control (0.24 ± 0.12) and ritonavir (0.32 ± 0.12) phases (95% CI of the difference −0.15, −0.009, P = 0.04)

Discussion

As strategies involving preloading the foetus with PIs have been proposed to reduce mother to child transmission of HIV [3, 4], it is essential to elucidate the mechanisms regulating the placental transfer of PIs. Previously our group demonstrated differential bidirectional transfer of indinavir (and vinblastine) in the human placenta [11], suggesting potential involvement of placental P-gp in retarding indinavir and vinblastine transfer to the foetus. In the present study, the placental transfer of indinavir and vinblastine was examined in the presence of two P-gp inhibitors, PSC833 and ritonavir [13, 16, 17]. Even though the change in transfer of indinavir and vinblastine was small in some placentae following addition of either P-gp inhibitor, the mean maternal to foetal clearance index of indinavir and vinblastine increased significantly in the presence of PSC833. Similarly, ritonavir significantly enhanced the mean maternal to foetal clearance index of vinblastine, but there was no significant change in the mean clearance index of indinavir. While the number of placentae studied was relatively small, these results indicate P-gp restricts the maternal to foetal transfer of indinavir and vinblastine, consistent with the placenta acting as a pharmacological barrier in humans.

PSC833 significantly increased the maternal to foetal clearance index of indinavir and vinblastine. Ritonavir enhanced the maternal to foetal clearance index of vinblastine; however, there was no significant change in the mean clearance index of indinavir. These results indicate P-gp restricts the maternal to foetal transfer of indinavir and vinblastine, consistent with the placenta acting as a pharmacological barrier in humans.

In the current study, the mean maternal to foetal clearance index of indinavir increased 1.5-fold after addition of PSC833. These results confirm our earlier suggestion that the maternal to foetal transfer of indinavir is modulated by P-gp [11]. Similarly, there is evidence that the placental transfer of another PI, saquinavir, is also modulated by placental P-gp [10]. Using the isolated perfused human placenta model, Molsa et al.[10] observed a three-fold higher maternal to foetal transfer of saquinavir in the presence of PSC833. There are several possible reasons for the difference in magnitude of the increase or difference in placental transfer of indinavir (1.5-fold) and saquinavir (three-fold) in the presence of PSC833 in the two studies. Firstly, unlike the present study design, Molsa et al.[10] did not use each placenta as its own control to assess possible inhibition of P-gp by PSC833, therefore the results were possibly influenced by interplacenta variability. Secondly, this experimental design in the study by Molsa et al.[10] probably contributed to the finding that the bidirectional transfer of antipyrine (marker of passive diffusion) was not equal as expected [11]; the maternal to foetal transfer of antipyrine was six-fold lower than the foetal to maternal transfer [10]. Furthermore, the measurement of placental drug transfer used by Molsa et al.[10] was different from that in the current study. Moreover, the concentration of PSC833 chosen by Molsa et al.[10] was higher, which may have resulted in more extensive P-gp inhibition. Finally, there is evidence that the two PIs vary in their affinity for P-gp; saquinavir has a higher affinity for P-gp compared with indinavir [26].

In the present study, vinblastine was used in tracer quantities as a marker of P-gp mediated transport [21]. The maternal to foetal transfer of vinblastine increased significantly following inhibition of P-gp with PSC833 or ritonavir. This is consistent with studies in cultured human trophoblast cells which demonstrate vinblastine cell uptake to be significantly enhanced in the presence of various P-gp inhibitors [21, 27]. Ritonavir (1.40 μM) has been shown to inhibit vinblastine transport in cell culture [13]. The current study demonstrated that a clinically relevant concentration of ritonavir (0.40 μM) [24], inhibited vinblastine transport by P-gp, thereby increasing its transfer across the placenta in the maternal to foetal direction.

Ritonavir (K_m = 0.060 μM) has a higher affinity for P-gp compared with indinavir (K_m = 0.47 μM), as illustrated previously using High Five insect cells expressing human MDR1 gene [13]. In addition, ritonavir (IC₅₀ = 3.80 μM) is a more potent P-gp inhibitor than indinavir (IC₅₀ = 44.0 μM) as shown in Caco-2 cells [15]. Nevertheless, ritonavir did not increase the maternal to foetal transfer of indinavir in the present study. This is similar to results in mice in which the placental transfer of saquinavir was not changed in the presence of ritonavir [28]. It is important to note that a clinically relevant concentration of ritonavir was chosen in the current study (0.40 μM) [24], as dual PI therapy with ritonavir is used in pregnancy [1], and hence the results would be relevant for clinical practice. This concentration (0.40 μM) however, was less than the IC₅₀ value (3.80 μM) mentioned above [15], and thus there was less likelihood of extensive P-gp inhibition. These results suggest that dual PI therapy with ritonavir is unlikely to achieve a clinically significant increase in indinavir transfer to the foetus.

Previously our group demonstrated the foetal to maternal transfer of indinavir and vinblastine to be two- to three-fold higher than maternal to foetal transfer [11]. In the current study, the mean maternal to foetal transfer of indinavir and vinblastine increased 1.5-fold following addition of the P-gp inhibitor PSC833. There is evidence to suggest that in the presence of P-gp inhibitors, foetal to maternal drug transfer is likely to be reduced [21]. Therefore the transfer of indinavir and vinblastine in each direction across the placenta following P-gp inhibition may be more similar than suggested by comparing data from the bidirectional studies [11] with the present inhibition studies. It is possible that higher concentrations of PSC833 or ritonavir may have achieved more extensive inhibition of placental P-gp, but this was not investigated in the current study as the intention was to use clinically relevant concentrations of the P-gp inhibitors.

In addition to P-gp, other placental transport systems may be involved in modulating the transfer of indinavir and vinblastine. The differential bidirectional transfer of these drugs demonstrated previously [11] suggests involvement of efflux transporters localized to the apical membrane of placental trophoblast cells, such as P-gp, MRP2 and BCRP [12]. Indinavir and vinblastine are substrates for MRP2 [29, 30], but not BCRP [31, 32]. PSC833 and ritonavir may not have inhibited MRP2 in the placenta at the concentrations used in the current study [29, 33], and thus it is possible that MRP2 may be involved in the active efflux of indinavir and vinblastine. However, it is important to note that the level of expression of MRP2 in the placenta is very low [34], hence this transporter is unlikely to contribute significantly to limiting the transfer of these drugs to the foetus. Therefore, the available evidence suggests that P-gp is the predominant transporter mediating the maternal to foetal transfer of indinavir and vinblastine.

In the current study, interplacenta variability was observed in the effect of PSC833 and ritonavir on the transfer clearance of indinavir and vinblastine. The relatively small number of placentae studied precludes an assessment of the full extent of the actual variability. The variability may be due to polymorphisms in the MDR1 gene encoding for P-gp. Tanabe et al. reported the T-129C allele resulted in significantly lower placental P-gp expression [35]. The G2677(A,T) allele was also associated with lower P-gp expression in the placenta, although the results were not statistically significant [35]. Similarly, Hitzl et al. observed polymorphisms in the MDR1 gene at positions 3435 and 2677 were associated with significantly lower placental P-gp expression [36]. Furthermore, P-gp expression was markedly lower when mother and foetus were homozygous for the TT allele at 3435 [36]. Molsa et al. investigated the influence of MDR1 polymorphisms on the placental transfer of saquinavir [10]. The authors concluded the genotype did not affect the PSC833-induced difference in the maternal to foetal transfer of saquinavir, but acknowledged that the number of placentae with various genotypes was low [10]. Recently, increased expression of MDR1 mRNAs and P-gp in placentae from HIV-1 infected women, compared to noninfected women, has been reported [37]; the relative roles of the infection and maternal administration of antiviral drugs in the former group of women to the increase in placental P-gp was not clear. The implications of heterogeneity in P-gp expression, and induction of P-gp in infected women under treatment, for placental drug transfer require further investigation.

In conclusion, this study confirmed P-gp modulated the maternal to foetal transfer of indinavir in the human placenta. In the presence of the P-gp inhibitor, PSC833, the mean maternal to foetal clearance index of indinavir increased significantly. In comparison, there was no significant change in the mean clearance index of indinavir following addition of a clinically relevant concentration of ritonavir. In the presence of each of the P-gp inhibitors, the change in indinavir transfer was variable; this may have been due to genetic polymorphisms in the MDR1 gene resulting in variable placental P-gp expression, and warrants further investigation. Thus, the results are consistent with the notion that P-gp mediated efflux creates the placenta as an important pharmacological barrier in humans, thereby creating the foetus as a pharmacological sanctuary. Co-administration of P-gp inhibitors may be useful in strategies involving preloading the foetus with PIs to decrease mother to child transmission of HIV. Inhibition of placental P-gp, however, is a complex challenge since maternal as well as foetal factors must be taken into consideration, and thus further research into this area is required.