1 Introduction
Edoxaban is a novel, oral anticoagulant that inhibits factor Xa (FXa), which is located at the confluence of the intrinsic and extrinsic coagulation pathways, the primary site of amplification in the coagulation cascade [
1]. Edoxaban binds to both free FXa and FXa within the prothrombinase complex, therefore producing a dose-dependent decrease in thrombin generation [
2]. Edoxaban has recently been approved in Japan for prophylaxis against deep vein thrombosis in patients following hip and knee replacement surgery [
3]. In addition, ongoing phase 3 trials are assessing the efficacy and safety of edoxaban for the treatment and prevention of recurrences of venous thromboembolism (Hokusai-VTE [
4]) and prevention of stroke and systemic embolic events in patients with atrial fibrillation (ENGAGE-AF TIMI 48 [
5]). The Hokusai-VTE trial is evaluating edoxaban 60 mg once daily, while the ENGAGE-AF trial is evaluating edoxaban 60 and 30 mg once daily [
4,
5]. Edoxaban 60 and 30 mg once-daily doses were selected for ENGAGE-AF based upon a phase 2 dose-finding study in atrial fibrillation (AF) patients that showed that these 2 dosing regimens provided similar or less frequent bleeding than standard warfarin therapy [
6].
The pharmacokinetics (PK) of edoxaban have been extensively studied in healthy volunteers. Edoxaban is rapidly absorbed with peak concentrations observed at 1–2 h postdose and elimination is biphasic with a mean terminal elimination half-life (t
½) of 8.75–10.4 h [
7]. Edoxaban is primarily eliminated unchanged through multiple pathways, with approximately 50 % of systemically absorbed drug eliminated via renal excretion. The most abundant metabolites (M4 and M1) are formed through hydrolysis with minor contribution from cytochrome P450 (CYP) 3A [
8].
P-glycoprotein (P-gp) is an efflux transporter primarily expressed in the apical/luminal membrane of epithelia of the small intestine, hepatocytes, renal proximal tubules, and other sites. With broad substrate specificity and high transport capacity, P-gp can limit the systemic exposure of various xenobiotics by decreasing intestinal absorption and increasing renal excretion and biliary excretion [
9‐
11]. Strong P-gp inhibitors may increase systemic absorption and decrease elimination of P-gp substrates, resulting in increased exposure. The US Food and Drug Administration (FDA) now recommends that all investigational drugs should be evaluated for effect on potential P-gp activity [
12]. Results from transporter studies using Caco-2 cells and wild-type versus P-gp knockout mice indicate that edoxaban is a substrate for P-gp, but not for other commonly tested uptake transporters (eg, the organic anion transporter 1) [
13]. Modeling and simulation analyses, which include AF patients from a phase 2 dose-finding study, have demonstrated that concomitant edoxaban and strong P-gp inhibitors increase edoxaban exposure and the risk of bleeding [
14]. Therefore, it is important to assess the effect of P-gp inhibition on edoxaban PK by drugs that would be commonly co-prescribed in the AF population. The objectives of the 6 studies described here were to evaluate potential PK interactions between edoxaban and cardiovascular drugs that are known P-gp substrates (digoxin, atorvastatin, quinidine, and verapamil) and/or inhibitors (quinidine, digoxin, amiodarone, dronedarone, verapamil, and atorvastatin) and which may be prescribed to patients with AF [
11,
15‐
18].
2 Methods
2.1 Study Designs
The design of the quinidine, verapamil, atorvastatin, and dronedarone studies were based on standard drug-drug interaction designs using 2-period, 2-treatment crossover designs and taking into consideration the disposition of edoxaban and the interacting drug [
12]. Given amiodarone’s 58-day terminal elimination half-life (t
1/2) [
19], a single-sequence design was selected to prevent carryover. The study with digoxin used a dual-sequence, parallel design [
20]. Sample sizes were based on results from previous PK studies of edoxaban and variability of edoxaban PK parameters (eg, Ogata et al. [
7]). The dose of edoxaban was 60 mg, the highest dose being tested in phase 3 clinical studies. The highest recommended therapeutic doses were employed for all potentially interacting drugs. Studies were conducted at Celerion clinical research units in Neptune, NJ (quinidine, digoxin, amiodarone, verapamil, dronedarone), or Belfast, Northern Ireland, UK (atorvastatin). An institutional review board approved all protocols and each subject signed an informed consent prior to study entry. The studies were conducted in accordance with the international conference harmonisation guideline E6 for good clinical practice.
2.1.1 Quinidine
This was an open-label, randomized, 2-period, 2-treatment crossover study in 42 healthy volunteers. Subjects were randomized to receive 1 of 2 treatment regimens during the first period, followed by a washout of 7–10 days before the alternate treatment was administered during the second period. The treatments were a single, oral dose of quinidine 300 mg (Mutual Pharmaceutical Co.) on day 1, quinidine 300 mg three times daily on days 2–3, and a single dose of quinidine 300 mg on day 4, plus a single dose of edoxaban 60 mg on day 3; or edoxaban 60 mg once daily on days 1–4 and a single dose of quinidine 300 mg on day 3.
2.1.2 Digoxin
This was an open-label, dual-treatment sequence, parallel study in 48 healthy subjects assigned to digoxin (Lanoxin, GlaxoSmithKline) on days 1–7 (0.25 mg every 12 h on days 1–2, then 0.25 mg once daily on days 3–7) followed by coadministration of digoxin 0.25 mg once daily with edoxaban 60 mg once daily on days 8–14; or edoxaban 60 mg once daily on days 1–7 followed by coadministration of edoxaban 60 mg daily with digoxin on days 8–14 (0.25 mg every 12 h on days 8–9, then 0.25 mg once daily on days 10–14) [
20].
2.1.3 Amiodarone
This was an open-label, single-sequence study in 30 healthy volunteers. Subjects received 2 treatments in a fixed sequence: (1) a single, oral dose of edoxaban 60 mg on day 1, then (2) amiodarone 400 mg (Pacerone, Upsher-Smith Laboratories, Inc.) once daily for 4 days (days 4–7) plus a single oral dose of edoxaban 60 mg on day 7. There was a 3-day washout period between the 2 treatments.
2.1.4 Verapamil
This was an open-label, randomized, 2-period, 2-treatment crossover study in 34 healthy volunteers. Subjects were randomized to receive one of either 2 treatment regimens during the first period, followed by a washout of at least 7 days before the alternate treatment was administered during the second period. The treatments were verapamil 240 mg sustained-release (Calan SR, Pfizer Inc.) once daily on days 1–11 and edoxaban 60 mg on day 10; or edoxaban 60 mg once daily on days 1–4 and verapamil 240 mg sustained-release on day 3.
2.1.5 Atorvastatin
This was an open-label, randomized, 2-period, 2-treatment crossover study in 32 healthy volunteers. Subjects were randomized to receive one of either 2 treatment regimens during the first period, followed by a washout of at least 7 days before the alternate treatment was administered during the second period. The treatments were atorvastatin 80 mg (Lipitor, Pfizer) once daily on days 1–8 plus edoxaban 60 mg on day 7; or edoxaban 60 mg on day 1.
2.1.6 Dronedarone
This was an open-label, randomized, 2-period, 2-treatment crossover study in 34 healthy volunteers. Subjects were randomized to receive one of either 2 treatment regimens during the first period, followed by a washout of at least 7 days before the alternate treatment was administered during the second period. The treatments were edoxaban 60 mg on day 1 under fed conditions; or dronedarone 400 mg (Multaq, Sanofi) twice daily on days 1–7 under fed conditions plus edoxaban 60 mg on day 5 under fed conditions.
2.2 Blood Sampling and Analysis for Edoxaban
Blood samples were collected to determine plasma edoxaban concentrations at serial time points over 24 h (quinidine, digoxin, and verapamil) or 72 h (amiodarone, atorvastatin, and dronedarone) after administration of edoxaban. Plasma edoxaban concentrations were measured by a validated liquid chromatography tandem mass spectrometry (LC-MS/MS) method with calibration ranges of 0.764 to 382 ng/mL (Advion BioServices, Ithaca, NY) for all studies except for the digoxin study, which had a lower limit of quantitation of 1 ng/mL.
Blood samples were also collected to confirm exposure for the potentially interacting drugs. Validated LC-MS/MS methods were used to quantify plasma drug concentrations (quinidine and atorvastatin at PPD, Richmond, VA; amiodarone at Bioanalytical Systems, Inc, West Lafayette, IN; verapamil and dronedarone at Celerion, Lincoln, NE). Digoxin was analyzed by PPD using a validated radioimmunoassay procedure.
2.3 Clinical Laboratory Assessments
Samples for routine laboratory testing were collected and tested by a Clinical Laboratory Improvement Amendments-approved laboratory. Results of hematology, serum chemistry, coagulation, urinalysis, and fecal occult blood (FOB) testing were classified according to the clinical laboratory reference ranges.
2.4 Pharmacokinetic and Statistical Analyses
Pharmacokinetic parameters calculated from the individual plasma concentrations included area under the plasma concentration versus time curve (AUC) from time 0 to the last measurable concentration or to 24 h (AUClast or AUC0–24); AUC from the time of dosing extrapolated to infinity, calculated as (AUC0–inf); maximum observed plasma drug concentration (Cmax); time of maximum observed concentration (tmax); terminal t1/2, calculated as ln(2)/λz; and mean concentrations 24 h postdose (C24). All PK parameters were calculated using WinNonlin Professional version 5.2 (Pharsight Corp., Mountain View, CA) or version 4.0 (digoxin study only). Edoxaban PK parameters and plasma concentrations were summarized using descriptive statistics.
The PK parameters for the interacting drugs were calculated when possible and summarized along with the plasma drug concentrations. All statistical summaries and analyses were performed using SAS version 8.2 (SAS Institute Inc., Cary, NC) or version 9.1.3 (dronedarone study only).
Comparisons of the PK of edoxaban administered with a potentially interacting drug versus edoxaban alone were performed using an analysis of variance (ANOVA) model with sequence, treatment, and period as fixed effects, and subject nested within sequence as a random effect. The ratios of the geometric least squares means (LSM) for the ln-transformed parameters Cmax, AUClast, and/or AUC0–inf were calculated for edoxaban alone versus edoxaban administered with the interacting drug. The 90 % confidence intervals (CI) of ratios were also calculated. If the 90 % CI was within the 80 to 125 % bioequivalence limit, it was concluded that there was no significant interaction between edoxaban and the coadministered drug.
2.5 Safety
Safety endpoints included vital signs, electrocardiograms (ECGs), adverse events (AEs), hematology, and serum chemistry.
4 Discussion
These studies explored the drug-drug interactions between edoxaban and 6 cardiovascular drugs that are likely to be commonly co-prescribed with edoxaban in the AF patient population and also have P-gp inhibition potential. These drugs display differing degrees of P-gp inhibition, with verapamil, quinidine, dronedarone, and amiodarone recognized as strong P-gp inhibitors [
12], while digoxin and atorvastatin are recognized P-gp substrates [
12,
16]. Coadministration of verapamil, quinidine, or dronedarone resulted in >50 % increases in total exposure measured as AUC and increases in the 24-h concentrations (a surrogate for trough concentrations) of edoxaban. Edoxaban coadministered with amiodarone increased total exposure by <50 % and 24-h concentrations decreased by approximately 26 %. Coadministration of either digoxin or atorvastatin had relatively minor effects on the PK of edoxaban.
The 90 % CI for the geometric LSM ratio values for AUC
0–inf and C
max of edoxaban were calculated per the standard test statistics described in regulatory guidelines. Salazar and colleagues used a pharmacometric model that predicted the effect of edoxaban exposure on bleeding to determine whether coadministration with these drugs would result in clinically meaningful increases in edoxaban exposure [
14]. Pharmacometric analysis of a phase 2 study in patients with AF determined that trough plasma concentrations were the most robust predictors of bleeding risk [
6]. Therefore, the observed edoxaban concentrations from the drug-drug interaction studies were modeled to predict trough concentrations at steady state. Increased exposure resulting in bleeding greater than the established limit (ie, greater than the observed warfarin bleeding) was considered clinically significant. On the basis of the results of the pharmacometric analysis [
14], verapamil, quinidine, and dronedarone were determined to have the potential for clinically meaningful effects on the disposition of edoxaban. Therefore, a dose reduction of 50 % has been recommended for coadministration of edoxaban with verapamil, quinidine, and dronedarone. The limited effects of atorvastatin and digoxin on edoxaban PK may be due to their weaker affinity for the P-gp transporter; therefore, these drugs, at therapeutic doses, are not considered strong P-gp inhibitors for edoxaban. Previous publications have cited verapamil, quinidine, amiodarone, and dronedarone as potent P-gp inhibitors based on the >25 % increase in AUC of digoxin, considered a probe substrate for P-gp [
12,
16,
24‐
26]. Although amiodarone has been shown to have strong P-gp inhibitory effects with other P-gp substrates, the increased edoxaban exposure was modest. Based on the pharmacometric analyses reported by Salazar et al. [
14], the increased edoxaban exposure associated with amiodarone did not result in significant bleeding based on these model predictions and, therefore, no dose adjustment was recommended.
The coadministration of these 6 cardiovascular drugs with edoxaban was considered safe and well tolerated within our study. There did not appear to be any significant increases in TEAEs recorded during any coadministration, and rates of TEAEs considered to be related to any study drug were low. However, the studies assessed relatively small numbers of healthy subjects and excluded those who had recently received prescribed or over-the-counter systemic or topical medications or herbal supplements. The total observed bleeding events in these studies were relatively minor in number and mild in severity; therefore, they offer limited insight into the bleeding potential across these dose regimens. With these caveats in mind, the incidence of bleeding-related AEs did not portend any increased bleeding when edoxaban was coadministered with one of these specific cardiovascular medications, nor was there any significant trend for increases in other AEs. However, patients with AF tend to be elderly and have cardiovascular comorbidities that may require treatment with one or more additional medications. The AE profile for edoxaban coadministered with these cardiovascular drugs will be further evaluated in patients with AF or venous thromboembolism in the large phase 3 ENGAGE AF-TIMI 48 [
5] and Hokusai-VTE [
4] studies, respectively.
Total exposure of edoxaban with concomitant administration of quinidine, verapamil, or dronedarone increased <2-fold over the total exposure to edoxaban administered alone. While the pharmacometric analyses of the interactions between edoxaban and these 3 drugs suggest the potential for increased bleeding risk [
14], the magnitude of these P-gp inhibitory effects was less than that which can be observed with drug interactions involving other mechanisms such as drug metabolizing enzymes. For example, statins that are extensively metabolized by CYP3A have shown a 10-fold or more increase in blood levels when coadministered with strong CYP3A inhibitors [
12]. Therefore, the clinical relevance of the increases in exposure to edoxaban when coadministered with quinidine, verapamil, or dronedarone should be considered in the context of the therapeutic index for edoxaban, particularly for bleeding potential. The final determination of the relative effect of increased exposure on bleeding risk for edoxaban administered concomitantly with verapamil, quinidine, or dronedarone will be determined in the ENGAGE AF-TIMI 48 and Hokusai-VTE trials.
Cardiologists should be aware of emerging data characterizing the effect of the P-gp transport system on drug exposure and its potential for drug-drug interactions, especially for cardiovascular drugs with narrow therapeutic indices. While the potential for drug-drug interactions attributed to drug metabolizing enzymes are relatively well known, the results of these studies underscore the importance of drug-drug interactions involving P-gp. Potential drug-drug interactions involving transporters were the focus of a white paper published by the International Transporter Consortium in 2010 [
9] and also the recent release of the FDA draft guidance on drug-drug interactions [
12]. The FDA guidance states that all investigational drugs should be assessed using in vitro studies to determine the relative affinity of the compound for the P-gp transporter. Transporter assays measure the flux across various cell lines overexpressing P-gp (eg, Caco-2 and MDRK), and the relative flux (i.e., >2) can be used to define whether a drug is a P-gp substrate. The effect of P-gp on the disposition of various compounds has also been assessed in P-gp knockout mice relative to wild-type mice [
27]. The FDA guidance further states that P-gp substrates characterized by in vitro assays as having significant potential for drug-drug interactions should be investigated in clinical studies. Dose administration guidance involving transporters should continue to expand for newly approved drugs. The role of the P-gp transporter system in various cardiovascular drugs is the focus of a very recent article in the
Journal of the American College of Cardiology [
28] and highlights the importance of understanding P-gp interactions that may occur between commonly co-prescribed drugs.
Our results suggest that quinidine, verapamil, and dronedarone—cardiovascular drugs used to manage AF patients that are likely to be prescribed with edoxaban—present some potential risk of bleeding in these patients if edoxaban is not dose-adjusted. Other recently developed oral anticoagulants (dabigatran, rivaroxaban, and apixaban) are also P-gp substrates [
29‐
31]. Currently, in patients with normal renal function, no dose adjustments are recommended for patients who are also taking a P-gp inhibitor along with dabigatran or rivaroxaban [
29,
30]. A dose adjustment is recommended for patients with moderate renal impairment who are receiving dabigatran and ketoconazole or dronedarone. Prescribing information for both rivaroxaban and apixaban, which are both metabolized by CYP3A4 and are P-gp substrates, indicate significant interactions with ketoconazole, ritonavir, clarithromycin, and erythromycin [
30,
31]. Concomitant use of rivoraxaban with drugs that are both P-gp inhibitors and strong CYP3A4 inhibitors is to be avoided [
30]. Patients receiving apixaban should have their dose reduced when administered along with a dual P-gp and strong CYP3A4 inhibitor or avoid their concomitant use altogether [
31].
In conclusion, the results from this study indicate that the PK of edoxaban in healthy subjects is increased by coadministration of the potent P-gp inhibitors quinidine, dronedarone, and verapamil. Concomitant administration of each of these drugs with edoxaban significantly increased exposure to edoxaban as reflected by either C
max, AUC, or 24-h concentrations relative to bleeding potential predicted from modeling and simulation of exposure-response relationships in patients with AF [
14]. Atorvastatin, digoxin, and amiodarone did not significantly affect exposure to edoxaban. Coadministration of each tested drug with edoxaban appeared to be well tolerated in these short-term studies. The potential effects of coadministered P-gp inhibitors on the clinical efficacy and safety of edoxaban will be further evaluated in the large-scale phase 3 Hokusai-VTE and ENGAGE AF-TIMI 48 studies.
Acknowledgments
The authors would like to acknowledge editorial assistance provided by Evince Communications, Norwalk, CT, and by AlphaBioCom, LLC, King of Prussia, PA, which was funded by Daiichi Sankyo.