Effect of multiple doses of montelukast on the pharmacokinetics of rosiglitazone, a CYP2C8 substrate, in humans

Abstract

Aims

To investigate the effect of multiple dosing with montelukast, a selective leukotriene-receptor antagonist, on the pharmacokinetics of rosiglitazone, a CYP2C8 substrate, in humans.

Methods

A two-period, randomized crossover study was conducted in 10 healthy subjects. After administration of oral doses of placebo or 10 mg montelukast daily for 6 days, 4 mg rosiglitazone was administered and plasma samples were obtained for 24 h and analyzed for rosiglitazone and N-desmethylrosiglitazone using high-performance liquid chromatography with fluorescence detection.

Results

During the montelukast phase, the total area under the time-concentration curve (AUC) and peak plasma concentration of rosiglitazone were 102% (90% CI 98, 107%) and 98% (90% CI 92, 103%) of the corresponding values during the placebo phase, respectively. Multiple dosing with montelukast did not affect the oral clearance of rosiglitazone significantly (90% CI 94, 105%; P = 0.50). The AUC ratio and plasma concentration ratios of N-desmethylrosiglitazone : rosiglitazone were not changed by multiple dosing with montelukast (90% CI 90, 103%; P = 0.14).

Conclusions

Multiple doses of montelukast do not inhibit CYP2C8-mediated rosiglitazone metabolism in vivo despite in vitro findings indicating that montelukast is a selective CYP2C8 inhibitor.

Introduction

Montelukast is a potent leukotriene receptor antagonist that blocks the action of cysteinyl leukotrienes [1]. Cysteinyl leukotrienes, lipid mediators released from inflammatory cells, produce airway oedema, mucus secretion, and eosinophil migration, which are reactions associated with the major findings of airway inflammation in asthma [2–4]. Therefore, montelukast has been used to reduce the signs and symptoms associated with asthma [5–7].

Several drug–drug interaction (DDI) studies [8–10] have shown that montelukast does not cause any significant DDIs with co-administered drugs, as a result of either pharmacodynamic or pharmacokinetic mechanisms. Consistent with these findings, a microsomal incubation study revealed that montelukast has a weak but noninhibitory effect on CYP1A2, CYP2A6, CYP2C19, CYP2D6, and CYP3A4-catalyzed reactions [11]. However, montelukast caused significant inhibition on CYP2C9 with a K_i value of 15 µmin vitro [11] but its inhibitory potency is weak compared with other CYP2C9 inhibitors [12–14]. In addition, montelukast does not affect the pharmacokinetics of S-warfarin, a substrate of CYP2C9 in healthy subjects [10]. Therefore, montelukast is believed to be a safe drug free from DDIs. Recently, Walsky et al. showed that montelukast was a selective and potent inhibitor of CYP2C8 with a K_i value of 0.0092–0.15 µm[15, 16]. Because its inhibitory potency is relatively strong, it may cause clinically significant DDIs with co-administered CYP2C8 substrates. In addition, in vitro results suggest that the area under the concentration-time curve (AUC) of CYP2C8 substrates could be increased at least two-fold. Therefore it is important to determine the clinical relevance of these in vitro findings.

Rosiglitazone is a novel insulin-sensitizing, oral antidiabetic agent of the thiazolidinedione class, used in the treatment of patients with type-2 diabetes mellitus, and is an effective and well-tolerated agent for lowering blood glucose concentrations [17]. Rosiglitazone is mostly metabolized to p-hydroxyrosiglitazone with subsequent conjugation and N-desmethylrosiglitazone and CYP2C8 is mainly involved in the N-demethylation and p-hydroxylation [18]. As rosiglitazone is a substrate for CYP2C8, it has the potential to interact with co-administered drugs that are inhibitors and inducers of this enzyme. Several drugs are known to interact with rosiglitazone leading to an increase or decrease in rosiglitazone concentrations in humans [19–22].

Considering the selective and potent inhibitory effect of montelukast on CYP2C8 in in vitro studies [15, 16], it is reasonable to suggest the possibility of a DDI between montelukast and rosiglitazone in vivo.

Therefore, the purpose of this study was to assess the effect of multiple doses of montelukast on CYP2C8 activity in humans using rosiglitazone N-demethylation as a CYP2C8 probe.

Methods

Subjects

Ten healthy Korean men (age 23–36 years, weight 67–80 kg) were included in the study after giving written informed consent. The subjects were confirmed as healthy by medical history, physical examination, and routine laboratory tests before they were enrolled in the study. All subjects were asked to abstain from alcohol and fruit juices during the study period.

Study design

The study protocol was approved by the IRB of Gil Medical Center, Incheon, Korea. A randomized, open-label, two-way crossover study was performed. The study phases were separated by a 3-week washout period, and the general study design was identical in both phases. The subjects were given either 10 mg montelukast (Singular®, MSD Korea, Seoul, Korea) or a matched placebo orally once daily for 6 days and the order of administration of two drugs was determined by the randomization schedule generated before the start of the study. After medication for five days with placebo or montelukast, a single oral dose of 4 mg rosiglitazone (Avandia®, GSK Korea, Seoul, Korea) was administered with water. The subjects were sitting when the drugs were administered, and they remained seated for the next 3 h after administration of rosiglitazone. A standard light meal was given after 4 h. The subjects were under direct medical supervision throughout the day of rosiglitazone treatment.

Sample preparations

Blood samples were drawn immediately before and at 0.5, 1, 1.5, 2, 2.5, 3, 4, 5, 6, 8, 10, 12, and 24 h after the administration of rosiglitazone. Blood samples, collected in heparinized glass tubes (Vacutainer®, Becton & Dickinson, Franklin Lakes, NJ), were centrifuged (209 g) for 15 min, and the separated plasma samples stored at −20 °C until assay.

Plasma rosiglitazone and its N-desmethylrosiglitazone concentrations were quantified using high-performance liquid chromatography with a slight modification of the previous method [23]. In brief, plasma samples (0.1 ml) with 20 µl of internal standard solution (50 µg ml⁻¹ of ketoconazole dissolved in methanol) were deproteinized using perchloric acid and directly injected onto the HPLC system. The mobile phase was composed of acetonitrile (45%) and 20 mm ammonium acetate (55%). The analysis was run at a flow rate of 0.15 ml min⁻¹ using a Capcellpak MG120 column (Shiseido, Japan) with a fluorescence detector operating at an excitation wavelength of 247 nm and an emission wavelength of 367 nm. The limit of quantification for rosiglitazone was 2 ng ml⁻¹ and the precision and accuracy for the quality controls were high, with a coefficient of variation of less than 8% within runs (intraday) and between runs (interdays). Because N-desmethylrosiglitazone is commercially unavailable, its concentration is given in arbitrary units relative to the ratio of the peak height of N-desmethylrosiglitazone to that of the internal standard in the chromatogram. The identity of the N-desmethylrosiglitazone peak (mass-to-charge ratio, 374) in the chromatogram was confirmed by mass spectrometry with the PE SCIEX API 3000 LC-MS-MS System (Sciex Division of MDS Inc, Toronto, Ontario, Canada). No additional peak was observed in the blank chromatograms from all subjects.

Pharmacokinetic calculations

The pharmacokinetic parameters of rosiglitazone and N-desmethylrosiglitazone were estimated by noncompartmental methods, using the software WinNonlin professional (version 4.1; Pharsight Co. Inc, Mountain View, CA). The maximum plasma concentration (C_max) and the time to reach C_max (t_max) were estimated directly from the observed plasma concentration–time data. The elimination rate constant (λ_z) was determined by linear regression analysis of the log-linear part of the plasma concentration–time curve. The total area under the plasma concentration–time curve (AUC) was calculated using the linear trapezoidal rule. The AUC from 0 to infinity (AUC(0, ∞) was calculated as AUC(0, ∞) = AUC_last+C_t/λ_z; where C_t was the last plasma concentration measured. The half-lives (t_1/2) of rosiglitazone and N-desmethylrosiglitazone were calculated from the following equation: t_1/2 = ln2/λ_z. The apparent oral clearance (CL/F) of rosiglitazone was calculated as CL/F = dose/AUC(0, ∞).

Statistical analysis

Results are expressed as means ± SD. The pharmacokinetic parameters of rosiglitazone were compared between the two phases with a paired t-test (two-tailed). The t_max data were compared with the Wilcoxon signed-rank test. All the data were analyzed with the statistical program Sigmastat® for Windows (version 3.1; SPSS Inc, Chicago, IL). Differences were considered statistically significant at P < 0.05.

Results

Mean plasma concentration-time profiles of rosiglitazone and N-desmethylrosiglitazone after administration of 4 mg rosiglitazone orally were similar between the placebo and montelukast phases (Figure 1).

Figure 1.

Mean plasma concentrations of rosiglitazone (A) and N-desmethylrosiglitazone (B) in 10 healthy subjects after a single oral dose of 4 mg rosiglitazone following 6 days pretreatment with placebo or 10 mg montelukast once daily. Results are shown as mean ± SD. Placebo phase (○), montelukast phase (•)

Multiple doses of montelukast did not show any statistically significant difference in pharmacokinetic variables for both rosiglitazone and N-desmethylrosiglitazone (Table 1). In the montelukast phase, AUC(0, ∞) for rosiglitazone and N-desmethylrosiglitazone was 102% (2004.4 ± 639.9 ng ml⁻¹ h vs. 2027.4 ± 602.5 ng ml⁻¹ h) and 98% (2002.3 ± 1188.0 arbitrary U ml⁻¹ h vs. 1971.1 ± 1163.3 arbitrary U ml⁻¹ h) of the value during the placebo phase, respectively. The oral clearance of rosiglitazone was 4.5 ± 2.0 l h⁻¹ and 4.4 ± 1.6 l h⁻¹ in placebo and montelukast phases, respectively (Table 1). The mean C_max values of rosiglitazone were also comparable without statistical significance between the placebo (271.6 ± 57.4 ng ml⁻¹) and montelukast (264.2 ± 55.3 ng ml⁻¹) phases (Table 1 and Figure 1). There was no significant difference between the AUC ratio of N-desmethylrosiglitazone : rosiglitazone for the placebo phase (0.54 ± 0.20) and the montelukast phase (0.52 ± 0.18) (P= 0.14) (Table 1).

Table 1.

Pharmacokinetic parameters of rosiglitazone and N-desmethylrosiglitazone after a single dose of 4 mg rosiglitazone in 10 healthy volunteers, following 6 days pretreatment with placebo or 10 mg montelukast once daily

	Placebo phase	Montelukast phase	Ratio (90% CI)	P value
Rosiglitazone
Cmax (ng ml−1)	271.6 ± 57.4	264.2 ± 55.3	0.98 (0.92, 1.03)	0.37
tmax (h)	1.2 ± 0.6	1.2 ± 0.7	1.24 (0.70, 1.78)	1.00
t1/2 (h)	5.7 ± 3.3	5.5 ± 2.2	1.03 (0.89, 1.16)	0.62
AUC(0,24) (ng ml−1 h)	1830.1 ± 454.0	1864.1 ± 452.2	1.02 (0.98, 1.07)	0.42
AUC(0, ∞) (ng ml−1 h)	2004.4 ± 639.9	2027.4 ± 602.5	1.02 (0.97, 1.07)	0.59
CL/F (l h−1)	4.5 ± 2.0	4.4 ± 1.6	1.00 (0.94, 1.05)	0.50
N-desmethylrosiglitazone
Cmax (arbitrary U ml−1)	53.7 ± 16.0	56.1 ± 16.0	1.05 (0.98, 1.13)	0.25
tmax (h)	8.1 ± 3.0	6.5 ± 1.7	0.88 (0.73, 1.02)	0.12
t1/2 (h)	27.3 ± 14.8	28.0 ± 18.8	1.00 (0.86, 1.15)	0.73
AUC(0,24) (arbitrary U ml−1 h)	959.7 ± 357.7	931.5 ± 309.8	0.99 (0.94, 1.03)	0.23
AUC(0, ∞) (arbitrary U ml−1 h)	2002.3 ± 1188.0	1971.1 ± 1163.3	0.98 (0.92, 1.04)	0.50
AUC ratio of N-desmethylrosiglitazone : rosiglitazone	0.54 ± 0.20	0.52 ± 0.18	0.97 (0.90, 1.03)	0.14

C_max, peak plasma concentration; t_max, time to peak plasma concentration; AUC(0,24), area under the plasma concentration–time curve, from 0 to 24 hours; AUC(0, ∞), area under the plasma concentration–time curve from time 0 to infinity; CL/F, oral clearance. Data are expressed as mean ± SD.

Discussion

The results of the present study indicate that montelukast produced no significant changes in the plasma concentration of rosiglitazone and thus caused no inhibitory effect on its pharmacokinetics in vivo. In addition, the pharmacokinetic characteristics of its metabolite, N-desmethylrosiglitazone, were also not affected by montelukast.

It has been reported that several drugs such as gemfibrozil (K_i = 69 µm) [24], ketoconazole (K_i = 2.5 µm) [25], and trimethoprim (K_i = 29–32 µm) [26] have an inhibitory effect on CYP2C8-catalyzed reactions in vitro. These inhibitors exhibited an inhibitory effect on CYP2C8-catalyzed reaction in vivo[19, 20, 22] but the magnitude of the interaction was weak (i.e, 1.4-, 2.3-, and 1.5-fold increase of rosiglitazone AUC in the presence of trimethoprim, gemfibrozil, and ketoconazole, respectively).

Recently, in vitro studies demonstrated that montelukast is a potent and selective inhibitor of CYP2C8 with a K_i value of 0.0092–0.15 µm[16]. Therefore, considering the stronger inhibitory potency of montelukast, co-administration of montelukast with CYP2C8 substrates may lead to an elevation of the level of co-administered CYP2C8 substrates by DDI in vivo. Furthermore, Walsky et al.[16] predicted the possibility of DDIs that could be caused by montelukast on CYP2C8 substrates and estimated more than a two-fold elevation of the AUC of CYP2C8 substrates with montelukast. Therefore, we could reasonably hypothesize that montelukast would cause a significant interaction with co-administered CYP2C8 substrates through CYP2C8 inhibition in vivo, although we did not observe any change in the pharmacokinetics of rosiglitazone by montelukast.

Theoretically, the potency of DDIs based on the inhibition of hepatic metabolism can be calculated from the K_i and the appropriate concentration of the inhibitor ([I]) of the metabolic enzyme in the liver, provided that the fraction of substrate metabolized by the inhibited CYP pathway is 1, using the following predictive model: AUCinhibited/AUCcontrol = 1 + [I]/K_i[27–30]. In the current study, we could not determine the concentrations of montelukast in plasma because it is commercially unavailable. Therefore we estimated the changes of rosiglitazone AUC based on the K_i value of montelukast (0.0092–0.15 µm) and the C_max of montelukast (1 µm) using the scaling model described above [16, 31]. From this we expected montelukast to increase the AUC of rosiglitazone more than eight-fold and possibly by up to 110-fold on the assumption that the drug completely inhibits the metabolism of rosiglitazone. Because the range of the calculated K_i values for montelukast is large, it causes a large variability in the AUC of rosiglitazone. Walski et al.[16] suggested that the carboxylic acid structure of montelukast has a binding site for substrate binding pocket on CYP2C8 described from recent X-ray crystallographic studies [32]. Therefore, structural diversity resulting from genetic polymorphism might be responsible for the wide range of K_i values for montelukast.

Although we did not consider some factors such as hepatic drug uptake, the rate of elimination, or intrahepatic montelukast concentrations in this equation [33], the predicted findings show a substantial discrepancy from the values of previously expected data (i.e. 2.1-fold elevation of AUC compared with baseline) [16]. In addition our findings suggest that expectation of DDIs in vivo from in vitro findings has limitations that are not fully understood. We could not clearly explain the discrepancy between predicted and actual results. However it is reasonable to suggest that the concentration of montelukast in plasma might be too low to inhibit CYP2C8-catalyzed rosiglitazone metabolism, so the potential for a DDI between montelukast and rosiglitazone might be weak. However, we administered montelukast at a therapeutic dose (10 mg day⁻¹) daily for 6 days, and considering the pharmacokinetics of montelukast [31], the time should have been sufficient to reach a steady state. Therefore, the concentrations used to predict the inhibition (1 µm) should have been adequate. On the other hand, the inhibitor concentration (montelukast) used in the equation above is just the concentration around the metabolic enzyme CYP2C8 intracellularly. Because of no available data in the literature about the concentrations of montelukast in tissue, we could not predict the exact concentration of montelukast available to the CYP2C8 enzyme in hepatocytes. However, considering that the volume of the distribution of montelukast is very low (8–11 l) [31], suggesting the extent of tissue distribution is very small and thus most of the drug is located in the extracellular space, the concentrations available to CYP2C8 in the liver may be far lower and thus montelukast may not reach liver cells at therapeutic concentrations, resulting in less DDI through CYP2C8. In addition, the target site for montelukast is cysteinyl leukotriene receptors which are located on the outer cell membrane [34].

In vitro studies have shown that CYP2C8 is the principal enzyme responsible for the elimination of rosiglitazone with CYP2C9 contributing to a lesser extent. Therefore, considering the large abundance of CYP2C9 isoform compared with CYP2C8 [27], uninhibited rosiglitazone metabolism through CYP2C9 by montelukast might be compensating for inhibited metabolism through CYP2C8. However, considering the far lower intrinsic clearance of rosiglitazone by CYP2C9 compared with that by CYP2C8 and the inhibitory effect of montelukast on CYP2C9 activity [16], this assumption is unlikely.

A recent in vitro study has shown that rosiglitazone is an inhibitor of OATP-C (organic anion transporter peptide-C) [35], which might suggest that rosiglitazone is a plausible substrate of the transporter. However, to date, there has been no observation that montelukast is a substrate and inhibitor of this transporter in vivo.

Interestingly, quercetin, another selective and potent inhibitor of CYP2C8 used in in vitro studies, did not influence the pharmacokinetics of rosiglitazone in vivo[36]. Montelukast and quercetin have far lower inhibitory constant, suggesting more strong inhibitory potency, than other inhibitors such as gemfibrozil [24] and trimethoprim [26]. However, trimethoprim and gemfibrozil exhibited inhibitory effect on CYP2C8-catalyzed reaction in vivo[13, 20, 26] whereas montelukast and quercetin did not [36]. From these findings, it might have a limitation to simply extrapolate the in vitro findings to the prediction of DDI possibility in vivo. Even though we are unclear of the exact mechanism as to why the estimated results of a DDI in vivo do not match the actual findings, it is obvious that in order to predict more accurately the in vitro–in vivo correlation of DDIs, other factors that represent the real concentration available to metabolic enzymes, such as tissue/blood partition coefficient and plasma protein bindings should be considered. For example, it has been reported that trimethoprim has a high liver/plasma partition coefficient [28] whereas that of quercetin is relatively lower [37].

In conclusion, the results of the present study indicate that despite the belief that montelukast is a selective and potent inhibitor of CYP2C8 in vitro, it does not inhibit CYP2C8-mediated rosiglitazone metabolism in vivo. Therefore, multiple doses of montelukast have a low probability of interacting with drugs that are metabolized by CYP2C8, including rosiglitazone.