The effect of anastrozole on the single-dose pharmacokinetics and anticoagulant activity of warfarin in healthy volunteers

Abstract

Aims

The aims of this study were to determine the effects of the nonsteroidal, selective aromatase inhibitor, anastrozole, at steady-state concentrations, on the pharmacokinetics and pharmacodynamics of warfarin, and to assess whether or not anastrozole alone has any anticoagulant activity.

Methods

This was a randomized, double-blind, placebo-controlled, two-way crossover trial conducted at a single centre. The study comprised two treatment periods of 11 days, separated by a 3 week washout. Healthy male volunteers (n = 16, median age 28.5 years) were randomized to receive either anastrozole (7 mg loading dose on day 1, followed by 1 mg daily on days 2–11) in the first treatment period and placebo in the second treatment period, or vice versa. In addition to their randomized treatment, all volunteers received a single dose of 25 mg warfarin on day 3 of each treatment period. Blood samples for pharmacokinetic and pharmacodynamic assessment were taken at frequent intervals during each treatment period. The safety of volunteers was monitored throughout the study.

Results

Administration of anastrozole resulted in no clinically significant changes in the pharmacokinetics of either R- or S-warfarin compared with placebo for AUC (ng ml⁻¹ h) (glsmean, R-warfarin; anastrozole 93619.9, placebo 91127.91, 95%CI 0.988–1.068; S-warfarin; anastrozole 57129.21, placebo 55676.34, 95%CI 0.979–1.076), CL/F (ml min⁻¹) (glsmean, R-warfarin; anastrozole 2.23, placebo 2.29, 95%CI 0.937–1.012; S-warfarin; anastrozole 3.65, placebo 3.74, 95%CI 0.929–1.021) and t_1/2 (h) (lsmean, R-warfarin; anastrozole 55.40, placebo 55.15, 95%CI−2.083–2.592; S-warfarin; anastrozole 39.38, placebo 40.98, 95%CI−6.189–2.996). In addition, anastrozole had no clinically significant effect on the pharmacodynamic effects of warfarin, as assessed 240 h after warfarin dosing by measurement of prothrombin time (s) (glsmean, anastrozole 11.56, placebo 11.31, 95%CI 0.987–1.059), thrombin time (s) (glsmean, anastrozole 19.06, placebo 18.75, 95%CI 0.980–1.054) activated partial thromboplastin time (s) (glsmean, anastrozole 29.94, placebo 29.74, 95%CI 0.968–1.047) and factor VII (%) (glsmean, anastrozole 97.81, placebo 107.26, 95%CI 0.821–1.012). Anastrozole alone had no effect on these indicators of the clotting process.

Conclusions

Overall, there was no evidence to suggest that anastrozole has any clinically relevant effects on the pharmacokinetics of warfarin. Anastrozole had no effect on clotting mechanisms or on the pharmacodynamic activity of warfarin, as assessed by prothrombin time, thrombin time, activated partial thromboplastin time, and factor VII.

Introduction

Anastrozole is a potent, selective, nonsteroidal, oral, new-generation aromatase inhibitor of the triazole class [1]. It acts by the inhibition of the enzyme, aromatase, that catalyses the conversion of the androgens, androstenedione and testosterone, to oestrogens (oestrone and oestradiol), the major route of oestrogen synthesis in postmenopausal women. Anastrozole is available worldwide and is frequently administered for long periods of time to an elderly population of postmenopausal breast cancer patients, who may be taking other drugs concomitantly, such as warfarin. Warfarin has been shown to interact adversely with both tamoxifen [2–3] and the earlier nonselective aromatase inhibitor, aminoglutethimide [4], by different mechanisms. Further, a small number of reports of increased prothrombin times was received for patients seemingly stabilized on the oral anticoagulant, warfarin, and then administered anastrozole.

An in vitro study of the effect of anastrozole on human cytochrome P-450 enzymes showed that anastrozole had the potential to inhibit both the isoenzymes, CYP2C9 and CYP3A4, which are known to be responsible for the metabolism of S-warfarin and R-warfarin, respectively, but only at very high concentrations (30-fold higher than achieved clinically) [5]. These data suggest that anastrozole at therapeutic doses would not be expected to cause any clinically relevant interactions with other cytochrome P-450 metabolized drugs.

Therefore it was important to investigate directly the potential for drug interaction between anastrozole and warfarin at therapeutic doses. The purpose of this study was to detect any effects of steady-state concentrations of anastrozole on the pharmacokinetics and pharmacodynamics of either enantiomer of warfarin, and to assess whether or not anastrozole alone had any anticoagulant activity.

Methods

Subjects

This was a randomized, double-blind, placebo-controlled, two-way crossover trial in 16 healthy male volunteers (age range 22–49 years; median 28.5 years) conducted at a single centre. The study was approved by an independent Ethics Committee of LandesärzteKammer Baden-Württembürg and written, informed consent was obtained from each subject. Subjects were deemed to be healthy, based on medical history, a normal physical examination, and results of clinical laboratory and electrocardiographic tests (resting and 24 h continuous ambulatory monitoring). No subjects had taken any medication (including over-the-counter medicines) from 72 h before trial day 1. Volunteers were also excluded from the study if they had any significant abnormal results in clinical chemistry, haematology or urine analysis results or had received treatment during the past 3 months with any drug known to induce or inhibit the cytochrome P450 enzymes involved in the metabolism of warfarin.

Study protocol and drug analysis

The study was designed and monitored in accordance with the ethical principles required by the Declaration of Helsinki and performed within the European Guidelines on Good Clinical Practice.

The study comprised two treatment periods of 11 days separated by a 3 week washout period. Volunteer subjects were randomized to receive either anastrozole in treatment period 1 and placebo in treatment period 2 or vice versa. In each treatment period, volunteers received either anastrozole 7 mg (7 × 1 mg tablets) as a loading dose or matching placebo on day 1, followed by anastrozole 1 mg or matching placebo on days 2–11. The loading dose of 7 mg of anastrozole was selected on the basis of previous pharmacokinetic studies, to enable rapid achievement of a steady-state concentration of anastrozole, prior to the administration of warfarin. In addition to their randomized treatment, all volunteers received warfarin 25 mg (5 × 5 mg tablets) on day 3 of each treatment period, administered 2 h after anastrozole or matching placebo. No concomitant medication was allowed from 72 h before trial day 1 in the first treatment period until the end of the trial.

Blood samples (7 ml) for the assessment of warfarin pharmacokinetics were collected before dosing with warfarin, and at 13 preselected times up to 240 h after dosing with warfarin on day 3. Additional samples (10 ml) were collected at 1, 2 and 3 h after dosing and assayed for free warfarin. The plasma samples for analysis of warfarin were then frozen and stored at −20°C until analysis. R- and S-warfarin were extracted from an aliquot of human plasma (1 ml) using methylene chloride:hexane (1 : 5). Analysis was by normal phase high performance liquid chromatography with u.v. detection. Standard curves were generated by weighted (1/concentration squared) least-squares regression analysis of the peak height of R- or S-warfarin vs concentration using duplicate, 7-point calibration standards in the range of 25.0–1000.0 ng ml⁻¹ in human plasma. R- and S-warfarin plasma concentrations were calculated based on the respective peak heights, relative to the standard curves prepared on the same day. Extraction recovery of R-warfarin ranged from 90.4 to 102.0% and of S-warfarin from 91.0 to 94.5% over the range of 50.0–800.0 ng ml⁻¹. Assay performance was assessed using quality control samples at 50.0, 200.0 and 800.0 ng ml⁻¹ for both R- and S-warfarin. The mean accuracy of the quality control samples ranges from −4.1–3.1% for R-warfarin, and −5.0–0.8% for S-warfarin. The intra-assay precision (CV%) of the quality control samples ranged from 2.2 to 2.9% for R-warfarin and 1.6–2.8% for S-warfarin. The interassay precision of the quality control samples ranged from 5.8 to 8.0% for R-warfarin and 5.5–7.8% for S-warfarin.

Blood samples (7 ml) for the assessment of anastrozole pharmacokinetics were collected before dosing on trial days 1, 3, 7 and 11, and 2 and 10 h after dosing on day 3. Blood was collected into tubes containing lithium heparin anticoagulant. The samples were immediately inverted gently and centrifuged to separate out the plasma. The plasma samples were then frozen and stored at −20°C until analysis [6].

Pharmacokinetic data analysis

This was performed using noncompartmental methods. The pharmacokinetic parameters generated for this study were C_max, time to C_max (t_max), terminal rate constant (λ_z), t_1/2, CL/F, AUC(0,t) and AUC, although C_max and t_max were not subject to formal statistical analyses.

The constant λ_z was determined by log-linear regression on data from the terminal monoexponential phase of the concentration vs time profiles. AUC(0,t) was estimated according to the linear trapezoidal method. AUC, t_1/2, and CL/F were calculated using the following formulae:

where C_last is the last quantifiable concentration.

The computer programme TopFit 2.0 was used to generate these pharmacokinetic parameters [7].

Two subjects (subject 7, period 1 and subject 10, period 2) had concentrations of S-warfarin detected in the predose samples (30.5 and 28.6 ng ml⁻¹, respectively) but these were less than twice the limit of quantification (25.0 ng ml⁻¹). These values were not reset to zero prior to determination of the pharmacokinetic parameters.

The S-warfarin concentration values from the 166 and 240 h postdose samples for subject 9, period 1 could not be confirmed upon reanalysis and were therefore excluded from the pharmacokinetic analysis.

Pharmacodynamic assessments

The primary pharmacodynamic endpoints were the prothrombin time (PT) following the administration of warfarin, assessed using the PT AUC(8,94 h) and the PT at 166 and 240 h after dosing. The secondary endpoints were:

1. activated partial thromboplastin time (aPTT) and thrombin time (TT) following the administration of warfarin, assessed using AUC(8,94 h), and the results at 166 and 240 h after dosing

2. factor VII concentration following the administration of warfarin, assessed using AUC(8,22 h) and AUC(22,94 h), and the results at 166 and 240 h after dosing

3. PT, factor VII concentration, aPTT, and TT on day 3 before dosing with warfarin.

Blood samples (4 ml) for pharmacodynamic assessment were collected on days 1, 2, and 3 before dosing with warfarin, and at eight preselected times up to 240 h after dosing with warfarin on day 3. Blood samples were collected in tubes containing citrate anticoagulant. Plasma was separated immediately by centrifugation and kept frozen at −20°C up to the time of the analysis. All samples of one subject were analysed in one analytical run. The clotting assays were performed on a Behring Coagulation System from Dade Behring. Thromborel R, Thromborel S, Pathrombin SL and BC-Thrombin Reagent were used for the determination of PT, factor VII, aPTT and TT, respectively. Pharmacodynamic analyses were performed in the Department of Clinical Chemistry at the Albert-Ludwigs University of Freiburg, Germany.

Statistical analysis

All pharmacokinetic endpoints were analysed using an analysis of variance (anova) model, fitting for the effects of volunteers, periods and treatments (anastrozole or placebo).

AUC and CL/F data were log-transformed prior to analysis, as previous experience had shown them to be log-normally distributed. Plasma t_1/2 data were not log-transformed prior to analysis.

Assumptions of normality and consistency of variance were explored in all analyses and were found to be valid.

All pharmacodynamic endpoints were log-transformed prior to analysis and were analysed using an anova model, fitting for the effects of volunteers, periods and treatments (anastrozole or placebo). Using an ancova (analysis of covariance) model, the usefulness of day 1, predose values in increasing the precision of any treatment effects was investigated.

Assumptions of normality and consistency of variance were explored in the analysis and, where necessary, a nonparametric technique, in the form of a rank analysis, was used to validate the results of the main analysis. Ranks were fitted to an anova model, allowing for the effects of volunteer, period and treatment. Furthermore, where it was deemed necessary to validate the 95% confidence interval obtained from the parametric analysis, a Mann–Whitney test was performed from which nonparametric 95% confidence intervals were obtained. Rank analyses were performed for the AUC(8,94 h) and 166 h timepoint of PT, the 240 h timepoint of TT and the AUC(8,94 h) of aPTT. In addition, a Mann–Whitney analysis was used for the 166 h timepoint of PT.

Safety assessments

The safety of volunteers was monitored throughout the trial. Assessments were made by performing clinical, haematology, and urine analysis tests.

Adverse event assessments

The deterioration of any pre-existing condition or development of a new medical condition during or after exposure to trial treatment was considered to be an adverse event, irrespective of causality. Particular attention was paid to symptoms which were suggestive of haemorrhage.

Results

Demography

Sixteen Caucasian male volunteers entered this study. They had a median age of 28.5 years (range 22–49 years), a mean height of 181.0 ± 7.1 cm (range 170–195 cm) and a mean weight of 75.8 ± 7.8 kg (range 65–92 kg).

Pharmacokinetics

A summary of the results of the analysis of plasma AUC, CL/F and t_1/2 to assess the effect of concomitant administration of anastrozole on the pharmacokinetics of R-warfarin and S-warfarin is given in Table 1.

Table 1.

Statistical comparison of anastrozole and placebo treatment groups to assess the effect of concomitant administration of anastrozole on the pharmacokinetics of warfarin (R and S enantiomers).

	Anastrozole		Placebo
Parameter	n	glsmean	n	glsmean	Treatment effect	95% CI	P value
R-warfarin
AUC (ng ml−1 h)	16	93619.9	15	91127.91	1.027a	0.988,1.068	0.160
CL/F (ml min−1)	16	2.23	15	2.29	0.974a	0.937,1.012	0.163
t1/2(h)	16	55.40b	15	55.15 b	0.254c	−2.083,2.592	0.818
S-warfarin
AUC (ng ml−1h)	16	57129.21	15	55676.34	1.026a	0.979,1.076	0.259
CL/F (ml min−1)	16	3.65	15	3.74	0.974a	0.929,1.021	0.249
t1/2(h)	16	39.38 b	15	40.98 b	−1.596c	−6.189–2.996	0.466

^aglsmean of the anastrozole phase divided by the glsmean of the placebo phase.

^blsmean.

^clsmean of the anastrozole phase minus the lsmean of the placebo phase, CI: confidence interval, AUC: area under the curve, CL/F: apparent oral clearance, glsmean: geometric least squares mean, lsmean: least squares mean.

Figures 1 and 2 show the geometric mean total R- and S-warfarin plasma concentration–time profiles, respectively, for subjects receiving warfarin with concomitant placebo and anastrozole administration. These concentration–time profiles were essentially the same.

Figure 1.

Mean plasma concentrations of R-warfarin after a 25 mg single dose of warfarin followed by anastrozole 1 mg (▪) or placebo (○) daily.

Figure 2.

Mean plasma concentrations of S-warfarin after a 25 mg single dose of warfarin followed by anastrozole 1 mg (▪) or placebo (○) daily.

Table 1 shows the geometric lsmean CL/F of total S-warfarin (3.7 ml min⁻¹) to be greater than for R-warfarin (2.3 ml min⁻¹) with concomitant administration of placebo. Also, the geometric mean t_1/2 of S-warfarin (41.0 h) was shorter than for R-warfarin (55.2 h). As a result, the geometric mean AUC value for S-warfarin was smaller than for R-warfarin. No statistically significant differences were found in the AUC, CL/F and t_1/2 for either R- or S-warfarin comparing values with placebo and during concomitant administration of anastrozole. Furthermore, in the presence of anastrozole, the t_1/2 of S-warfarin did not increase or exceed that of R-warfarin. The geometric mean free R- and S-warfarin concentrations at 1, 2 and 3 h after warfarin dosing with concomitant placebo were also measured and found to be similar with both placebo and anastrozole administration.

Anastrozole was present in the plasma samples of all subjects who were administered anastrozole. Geometric mean concentrations were 36.4 ng ml⁻¹ prior to dosing with warfarin on day 3, and 22.5 and 20.9 ng ml⁻¹ prior to anastrozole dosing on days 7 and 11, respectively. These data indicate that the plasma concentrations of anastrozole throughout the warfarin dosing and sampling period were within the range seen in postmenopausal women with advanced breast cancer taking the clinically recommended dose [8].

Pharmacodynamics

The results of the analysis of PT, aPTT and TT, using AUC(8,94 h), and factor VII, using AUC(8,22 h) and AUC(22,94 h), together with the results at 166 h and 240 h after dosing for all four parameters, are summarized in Table 2.

Table 2.

Statistical comparison of anastrozole and placebo treatment groups to assess the effects of concomitant administration of anastrozole on the anticoagulant activity of warfarin.

	Anastrozole		Placebo
Parameter	n	glsmean	n	glsmean	Treatment effecta	95% CI	P value
PT (s)
Pre-dose (day 3)b	16	11.56	15	11.45	1.010	0.984,1.037	0.442
AUC(8,94 h)	16	17.45	15	18.36	0.951	0.860,1.052	0.299
166 hb	16	11.27	15	11.31	1.023	0.987,0.983	0.015
240 hb	16	11.56	15	11.31	1.023	0.987,1.059	0.193
aPTT (s)
Pre-dose (day 3)b	16	30.07	15	29.52	1.019	0.987,1.052	0.233
AUC(8,94 h)	16	34.96	15	36.07	0.969	0.919,1.023	0.237
166 hb	16	29.43	14c	31.72	0.928	0.889,0.968	0.002
240 hb	16	29.94	15	29.74	1.007	0.968,1.047	0.722
TT (s)
Pre-dose (day 3)b	16	18.64	15	18.23	1.023	0.991,1.055	0.148
AUC(8,94 h)	16	18.86	15	18.68	1.010	0.979,1.042	0.505
166 hb	16	18.72	14c	19.05	0.983	0.935,1.034	0.474
240 hb	16	19.06	15	18.75	1.017	0.980,1.054	0.350
Factor VII (%)
Pre-dose (day 3)b	16	109.59	15	103.34	1.060	0.925,1.215	0.366
AUC(8,22 h)	16	51.09	15	46.04	1.110	0.982,1.254	0.088
AUC(22,94 h)	16	30.94	15	28.67	1.079	1.908,1.284	0.355
166 hb	16	108.96	15	98.72	1.104	0.970,1.256	0.123
240 hb	16	97.81	15	107.26	0.912	0.821,1.012	0.079

^aglsmean of the anastrozole phase divided by the glsmean of the placebo phase.

^btimes relate to dosing of warfarin.

^cvalue not recorded for subject 0006. CI: confidence interval, PT: prothrombin time, aPTT: activated thromboplastin time, TT: thrombin time, AUC: area under the pharmacodynamic-time curve, glsmean: geometric least squares mean.

There was no carry-over of warfarin between treatment periods and no evidence of any anastrozole carry-over between treatment periods.

There were no statistically significant differences between treatment with anastrozole or placebo in the PT, aPTT, TT and factor VII concentrations prior to warfarin dosing.

PT AUC(8,94 h) and PT at 240 h postdose were essentially the same during both treatments. Analysis using an anova technique revealed no significant differences for either pharmacodynamic measurement (AUC(8,94 h), P = 0.299; 240 h, P = 0.193). Since the analysis of AUC(8,94 h) was affected by the presence of an outlying observation, the absence of a treatment effect on AUC(8,94 h) was confirmed by a nonparametric rank analysis (P = 0.560). The only statistically significant finding in this analysis was a reduction in PT (12.4%) at 166 h postdose (P = 0.015) with anastrozole compared with placebo.

Using an anova technique, the treatment effect on TT AUC(8,94 h) was not statistically significant (P = 0.505), nor was it statistically significant on TT determined at 166 h postdose (P = 0.474) and 240 h postdose (P = 0.350). Since the analysis of TT at 240 h postdose was affected by the presence of an outlying observation, the absence of a treatment effect was again confirmed by a nonparametric rank analysis (P = 0.562). The aPTT AUC(8,94 h) and the aPTT at 240 h postdose were not statistically different between the two treatments (P = 0.237 and P = 0.722, respectively), as determined by anova. The absence of a treatment effect AUC(8,94 h) was again confirmed by a nonparametric rank analysis (P = 0.884), since the analysis of aPTT AUC(8,94 h) was affected by the presence of an outlying observation. The only statistically significant finding in this analysis was a reduction in aPTT (11.1%) at 166 h postdose (P = 0.002) with anastrozole administration compared with placebo.

There was no statistically significant treatment effect, as determined using an anova technique, on factor VII AUC(8,22 h) (P = 0.088) or AUC(22,94 h) (P = 0.355) or in the percentage of factor VII determined at 166 h (P = 0.123) and 240 h (P = 0.079) postdose, in the presence of anastrozole or placebo.

Tolerability

There were no serious adverse events during the trial. The trial medication of anastrozole in the study was well tolerated by volunteers.

Discussion

Warfarin is a mixture of R- and S-enantiomers, with both enantiomers being active as anticoagulants; the S-enantiomer has a shorter half-life and greater potency than the R-enantiomer [9]. The R- and S-enantiomers of warfarin are metabolized by different mechanisms involving different cytochrome P-450 isoforms [10–11]. The risk of complications with warfarin usage may be increased with concomitant drug therapy, with some drugs potentiating, and others inhibiting, the effect of warfarin, leading to changes in its anticoagulant effect [12]. Drug interactions involving warfarin can be divided into pharmacokinetic effects, which alter the plasma concentration of the drug, and pharmacodynamic effects, which increase the risk of bleeding, without altering the plasma concentration of warfarin [9].

In this healthy volunteer study, the mean plasma concentrations of anastrozole achieved throughout the warfarin dosing and sampling period were within the range seen in postmenopausal women with advanced breast cancer taking the clinically recommended dose of the drug [8]. The pharmacodynamic results demonstrated that sufficient warfarin was administered to produce a clinically significant effect.

No statistically significant differences were found in the CL/F, t_1/2, AUC, C_max, t_max or protein binding of R- or S-warfarin between anastrozole and placebo. In both the absence and presence of anastrozole, total R-warfarin was cleared more slowly than S-warfarin, resulting in the expected higher plasma concentration of R- than of S-warfarin [9]. These results indicate that anastrozole does not affect any of the P450 enzymes involved in warfarin metabolism, as reflected by its pharmacokinetics.

Following warfarin administration, all of the pharmacodynamic parameters showed predictable results, regardless of the coadministered treatment. The statistically significant reductions in PT and aPTT were small, only evident at a single timepoint and in the opposite direction to that which would be expected on the basis of inhibition of warfarin metabolism. These reductions are not considered clinically significant. However, these data do not exclude the possibility that different results might be obtained in patients stabilized on therapeutic doses of warfarin.

Overall, there was no evidence to suggest that anastrozole has any clinically relevant effects on the pharmacokinetics, including protein binding or anticoagulant activity, of warfarin. These results confirm the prediction that, although high concentrations of anastrozole inhibit the cytochrome P450 enzymes, CYP3A4 and CYP2C9, in vitro, these effects are not translated into the clinical situation [10]. In this respect, anastrozole differs from other endocrine therapies, namely, tamoxifen and aminoglutethimide [2–4].

The results of this study clearly indicate the absence of an effect of anastrozole on a single therapeutic dose of warfarin. They support the conclusions of previous studies, suggesting that therapeutic doses of anastrozole have no clinically significant inhibitory effect on P450 enzymes in healthy volunteers. They predict that anastrozole will not be associated with clinically significant drug interactions by this mechanism.