Effect of Amiodarone on Apixaban Exposure in Patients after Cardiac Surgery—A Population Pharmacokinetic Study

Abstract

Aim

To investigate the effect of amiodarone on apixaban pharmacokinetics in cardiac surgery patients with postoperative atrial fibrillation.

Methods

Apixaban concentrations of postoperative cardiac surgery patients with or without amiodarone therapy were quantified using liquid chromatography–tandem mass spectrometry (LC–MS/MS) in clinical routine. A population pharmacokinetic model was built using nonlinear mixed effects modeling in NONMEM^® 7.5 using first-order conditional estimation with interaction. The impact of amiodarone and creatinine clearance (CrCL) on apixaban exposure under various dosing regimens was analyzed using Simulx^® (Lixoft).

Results

A total of 33 patients with 76 apixaban concentrations were included. A one-compartment model best described the pharmacokinetics of apixaban with a clearance (CL/F) of 3.05 L/h, apparent volume of distribution (V_d/F) of 23.7 L, and an absorption rate constant (k_a) of 0.652/h. Interindividual variability (IIV) was observed in CL/F but not in V_d/F and k_a. The covariates amiodarone and CrCL were independently associated with apixaban CL/F. Under concomitant amiodarone therapy, simulations predicted an increase of 44–49% in apixaban area under the concentration–time curve (AUC), and AUC nearly doubled at CrCL 35 mL/min. A dose of 2.5 mg apixaban twice daily (b.i.d.) was identified as a potential dosing option in the CrCL range of 15–50 mL/min under amiodarone comedication.

Conclusions

Concomitant amiodarone therapy reduced apixaban CL/F and increased the risk of high exposure in patients with impaired renal function. A dose of 2.5 mg apixaban b.i.d. for a CrCL range of 30–50 mL/min under concomitant amiodarone therapy was identified as a new dosing option.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40262-025-01534-z.

Key Points

Amiodarone and creatinine clearance both have an effect on apixaban clearance.

In simulation scenarios of concomitant amiodarone therapy and impaired renal function, apixaban exposure nearly doubled at creatinine clearance of 35 mL/min.

The labeled dose of apixaban 2.5 mg b.i.d. was identified as a potential dosing option under amiodarone comedication in the creatinine clearance range of 15–50 mL/min.

Introduction

Postoperative atrial fibrillation is a relevant and common complication after cardiac surgery and is associated with increased morbidity and mortality [1, 2]. Rapid cardioversion is crucial to prevent hemodynamic instability and subsequent negative events such as renal failure [3–5]. For this purpose, amiodarone is the first-line pharmacotherapeutic option, although patient management might be complicated by possible drug–drug interactions [6]. Amiodarone and its metabolite N-desethylamiodarone (DEA) are moderate inhibitors of cytochrome P450 (CYP) isozymes as well as the drug transporters P-glycoprotein (P-gp) and breast cancer resistance protein (BCRP) [7–9]. This may be particularly relevant because direct factor Xa inhibitors such as apixaban are commonly used for stroke prophylaxis in cardiac surgery patients in whom high CHA₂DS₂-VA scores (formerly CHA₂DS₂-VASc score) are prevalent [10, 11]. Apixaban is a substrate of CYP3 A4, P-gp, and BCRP and is at risk of increased exposure if administered concomitantly with respective inhibitors or in patients with impaired transporter activity [9, 12, 13]. Indeed, several studies reported elevated apixaban trough concentrations (C_min) and bleeding events in patients treated with amiodarone and apixaban [14–16]. However, the extent of amiodarone inhibition on apixaban clearance (CL/F) and apixaban exposure has not been quantified so far, and it is unclear how this interaction should be managed in clinical practice. Especially postoperative cardiac surgery patients could be susceptible to increased apixaban exposure as additional risk factors for increased apixaban exposure (e.g., advanced age or impaired renal function) are frequently present in this patient population [4, 17, 18].

Therefore, a prospective population pharmacokinetic (PopPK) analysis using nonlinear mixed-effects modeling was conducted in a cohort of cardiac surgery patients receiving either apixaban alone or apixaban and amiodarone to analyze the impact of amiodarone on apixaban pharmacokinetics (PKs) and explore possible dosage reduction strategies.

Methods

Setting

The study was performed in the cardiac surgery department of a large tertiary care university hospital. This cardiac surgery department covers a broad range of surgeries, from pacemaker implantation to heart transplantation.

Study Design

All patients aged ≥ 18 years after cardiac surgery treated with apixaban alone (control cohort) or in combination with amiodarone (interaction cohort) were eligible and were included in the study after providing informed consent. Patients with mechanical valve prothesis were excluded from the study. Patients were treated according to the drug label with apixaban 5 mg or 2.5 mg twice daily (b.i.d.) for stroke prophylaxis. The study was conducted in accordance with the declaration of Helsinki and its subsequent amendments, and all patients provided informed consent. A positive vote from the responsible Ethics Committee of the Medical Faculty of Heidelberg University was obtained before the start of the study (ethical approval no. S-523/2021).

Data Collection

Apixaban concentrations were measured in clinical routine at the time of routine blood sampling in residual blood and, therefore, were not necessarily trough concentrations, providing different time points after drug administration for modeling. Actual time points of drug administration and blood sampling were documented on standardized data sheets by clinical pharmacists of the study team on a daily basis, thus ensuring valid and comprehensive documentation.

Apixaban plasma concentrations were quantified as previously described by liquid chromatography coupled to tandem mass spectrometry (LC–MS/MS) [19]. The assay fulfilled all recommendations of the ICH M10 guideline for bioanalytical method validation. The lower limit of quantification was 0.5 ng/mL [19].

Population Pharmacokinetic Modeling

The PopPK model of apixaban was built using nonlinear mixed-effects modeling in NONMEM^® (version 7.5; ICON plc, Dublin, Ireland), which was executed through Pearl-speaks-NONMEM (version 5.0.0; Uppsala University, Uppsala, Sweden) [20]. First-order conditional estimation with interaction (FOCE-I) was used to estimate the model parameters. All data processing was performed in R (version 4.3.1, R Core Team, Vienna, Austria). The R packages ggplot2 (version 3.4.3), ggpubr (version 0.6.0), vpc (version 1.2.1), pwr (version 1.3), and GraphPad Prism (version 9.5, GraphPad Software, Boston, Massachusetts, USA) were used for plotting.

One-compartment and two-compartment models were considered for the structural PK model. Interindividual variability (IIV) of the PK parameters (log-normal distribution) and interoccasion variability (IOV), as additional intraindividual variability of the PK parameters across observed dosing occasions, were evaluated on all structural PK parameters. Combined, additive, and proportional error models were investigated to describe the unexplained residual variability of the observed concentration–time data around the model prediction.

Covariate testing was conceptually driven by the main hypothesis of the clinical study. Hence, the tested covariates included body weight (on both apixaban CL/F and apparent volume of distribution (V_d/F)), creatinine clearance (CrCL) according to Cockcroft and Gault’s equation on CL/F, as well as potentially relevant comedication (CYP inhibitors, CYP inducers, amiodarone, metamizole, rosuvastatin, and atorvastatin) on CL/F [14, 21]. CYP inhibitors and inducers were defined according to Flockhart’s table [22]. Continuous covariates were tested using linear and power models centered around the median covariate value of the population or centered to standard values. Apixaban dose was tested as a categorical covariate on bioavailability (F). To test amiodarone as a covariate, an established PK model for amiodarone [23] was encoded, and typical PK profiles were simulated within the same NONMEM control stream. An inhibition parameter for apixaban CL/F with concomitant administration of amiodarone was estimated. An attempt was made to estimate a threshold concentration for the (unobserved) amiodarone PK to capture the onset/offset of the interaction. Differences of post hoc estimates of CL/F were tested using a two-sided t-test.

The covariates were included using a stepwise forward approach followed by a more conservative backward elimination procedure. Inclusion of covariates was guided by a statistically significant improvement of the model fit using the likelihood ratio test (forward Δ objective function value [OFV] > 3.84, i.e., P < 0.05; backward ΔOFV > 6.63, P < 0.01) and reduction of variability components (lower variability indicates better model fit).

For each candidate model, the goodness-of-fit plots comparing the population predictions (PRED) and individual predictions (IPRED) to the observed data (dependent variable, DV), as well as residual plots such as the normalized prediction distribution error (NPDE) versus time after dose or IPRED were inspected. The predictive performance of the final model was evaluated using visual predictive checks (VPC, n = 500 simulations) using prediction-correction [24]. Parameter uncertainty was calculated using the log-likelihood profiling-based sampling-importance resampling technique (LLP-SIR) [25].

Sample size calculation was based on a two-sample t-test. For a power of 80%, a minimum sample size of 15 patients in each group was determined, assuming an effect size of θ = 1.5, a coefficient of variation of 0.37, and a given error level of α = 0.05 [26].

Exposure and Dosing Simulations

To analyze the effect of amiodarone on apixaban exposure, different dosing regimens, including scaling the dose to potential covariates, were simulated with the final model using Simulx (version2023R1; Lixoft, Antony, France). Therefore, apixaban profiles at the CrCL values 100 mL/min, 70 mL/min, 50 mL/min, 35 mL/min, 25 mL/min, and 15 mL/min were simulated for 1000 individuals with and without amiodarone. The threshold for apixaban dose reduction of 35 mL/min was derived from the factors indicating dose reduction in the drug label (i.e., age ≥ 80 years, serum creatinine ≥ 1.5 mg/dL, and weight ≤ 60 kg). The factors were inserted in the Cockcroft and Gault formula in various combinations of cutoff values (e.g., 79 years, 61 kg, serum creatinine 1.5 mg/dL) to estimate a cutoff value for the CrCL for the in-label dose of 5 mg apixaban.

The AUC at steady-state (AUC_ss) and C_min were calculated for the approved dose and for alternative dosing regimens considering amiodarone coadministration together with renal function. A C_min of > 230 ng/mL was used as a threshold for a potentially increased risk of bleeding as suggested in previous studies [13, 18, 27, 28]. For the risk of underdosing, a threshold of < 34 ng/mL was used reflecting C_min concentrations below the 5 th percentile of a labeled dose of 2.5 mg b.i.d. [13, 29].

Results

Overall, 33 patients were included in the study, contributing 76 apixaban concentrations. Of these, 15 patients (45%) had concomitant amiodarone and apixaban therapy after surgery. Comorbidities and demographics were equally distributed between the two groups (Table1). Coronary artery bypass graft (CABG) was the most frequent surgery, followed by aortic conduits due to aortic aneurysms. No CYP inhibitors or CYP inducers as defined by Flockhart’s table were prescribed to the patients [22].

Table 1.

Patient characteristics at inclusion

	Control cohort (n = 18)	Interaction cohort (n = 15)	Overall (n = 33)
Age (years) [min, max]	73.0 [32.0, 84.0]	70.0 [55.0, 87.0]	72.0 [32.0, 87.0]
Sex
Female	2 (11.1%)	2 (13.3%)	4 (12.1%)
Male	17 (94.4%)	12 (80.0%)	29 (87.9%)
Weight (kg) [min, max]	83.0 [61.3, 136.0]	83.0 [63.0, 103.0]	83.0 [61.3, 136.0]
Creatinine clearance (mL/min) ± SD	77.3 ± 38.8	74.8 ± 25.3	76.2 ± 33.3
Body mass index ± SD	28.1 ± 4.63	26.9 ± 4.27	27.6 ± 4.45
Previous stroke	1 (5.5%)	3 (20.0%)	4 (12.1%)
Diabetes mellitus type 2	7 (38.8%)	7 (46.6%)	14 (42.4%)
Past acute coronary syndrome	3 (16.6%)	6 (40.0%)	9 (27.7%)
Acute coronary syndrome < 1 month	0 (0%)	2 (13.3%)	2 (6.1%)
Hyperlipidemia	13 (72.2%)	11 (73.3%)	24 (72.7%)
Peripheral arterial disease	1 (5.5%)	0 (0%)	1 (3.0%)
Severe liver impairment	0 (0%)	0 (0%)	0 (0%)
Arterial hypertension	16 (84.2%)	12 (80.0%)	28 (84.8%)
Left ventricular ejection fraction (%) [min, max]	52.0 [20.0, 70.0]	54.0 [20.0, 60.0]	53.0 [20.0, 70.0]
Type of surgery
Aortic aneurysm	5 (27.7%)	5 (33.3%)	10 (30.3%)
CABG	9 (50.0%)	6 (40%)	15 (45.5%)
CABG + valve replacement	3 (16.6%)	2 (13.3%)	5 (15.2%)
Congenital heart defect	1 (5.5%)	0 (0%)	1 (3.0%)
Single valve replacement	1 (5.5%)	1 (6.6%)	2 (6.1%)
Comedication
Metamizole	18 (100%)	13 (86.6%)	31 (93.9%)
Rosuvastatin	2 (11.1%)	3 (20%)	5 (15.1%)
Atorvastatin	11 (61.1%)	8 (53.3%)	19 (57.6%)

CABG coronary artery bypass graft, SD standard deviation

Population Pharmacokinetic Modeling

A one-compartment model with first-order absorption and elimination described the observed apixaban concentration–time profiles well. A two-compartment model did not improve the model fit (ΔOFV −2.09). IIV was supported on CL/F, but not on V_d/F or the absorption rate constant k_a. IOV was not supported on any parameter. The VPC indicated a good predictive performance (Fig. 1). No trends were observed in the residual plots, indicating that the model structure was chosen adequately (Figs. 2 and 3).

Fig. 1.

Prediction-corrected visual predictive check of the population pharmacokinetic model of apixaban; 10 th to 90 th percentile (dashed lines) and median (solid line) and corresponding 95% confidence intervals of the model prediction. N = 500 simulations

Fig. 2.

Population predictions (PRED, left) or individual predictions (IPRED, right) versus observed apixaban concentration (DV)

Fig. 3.

Time after dose (left) or individual predictions (right) versus normalized prediction distribution error (NPDE). The dashed lines indicate the expected 2.5 th, 50 th, and 97.5 th percentile

Amiodarone comedication (ΔOFV: − 14.327, P = 0.00015) as well as CrCL (ΔOFV: − 6.758, P = 0.0093) were found as significant covariates of the population parameter CL/F, which remained in the model after the backward elimination procedure. The individual CL/F estimates in relation to these covariates are shown in Fig. 4. It was not possible to estimate an amiodarone threshold concentration for onset of the inhibition, and no difference between amiodarone loading or maintenance doses were observed. Hence, the threshold was set to a small value of 0.001 mg/L. The dose of apixaban was not a significant covariate of bioavailability (ΔOFV: −3.386, P = 0.065). There was no interaction between metamizole, rosuvastatin, or atorvastatin with CL/F.

Fig. 4.

Covariate relationships of apixaban CL/F versus amiodarone comedication (left) and CrCL (right). Solid line (right): estimated covariate relationship. P-value from comparison of post hoc estimates of CL/F using a two-sided t-test

The final model parameters are presented in Table 2. Apart from k_a and the proportional residual variability, the model parameters were estimated with adequate precision given the small-N dataset. Individual model predictions are reported in Supplementary Fig. S1.

Table 2.

Parameter estimates of the population pharmacokinetic model of apixaban

	Estimate	Lower CI 95%	Upper CI 95%
CL/F [L/h]	3.05	2.54	3.71
Vd/F [L]	23.7	16.11	31.7
ka [/h]	0.652	0.338	1.345
COV-CL/F-amiodarone [−]	0.679	0.556	0.820
COV-CL-CrCL [−]	0.279	0.073	0.506
IIV CL/F [% CV]	29.4	20.4	43.7
Prop. RUV [% CV]	31.4	25.8	39.3

Individual CL/F = 3.05 × (CrCL/100)^0.279 × 0.679 (for amiodarone comedication)

% CV for IIV calculated by $\sqrt{e^2-1}$

CI confidence interval, CL/F apixaban clearance, COV covariate, CrCL creatinine clearance, IIV interindividual variability, k_a absorption rate constant, Prop proportion, RUV residual unexplained variability, V_d/F volume of distribution

Exposure Analysis

Simulations using the approved doses showed that median exposure increased by 35% with a CrCL of 35 mL/min for treatment with apixaban 5 mg b.i.d. without amiodarone (Supplementary Table S1). In simulations, concomitant amiodarone therapy considerably increased apixaban exposure over all CrCL scenarios compared with no amiodarone therapy. Consequently, the median AUC_ss nearly doubled (+95.1%) between apixaban 5 mg b.i.d. monotherapy at a CrCL100 mL/min and apixaban 5 mg b.i.d. under amiodarone at a CrCL of 35 mL/min.

Simulated profiles without amiodarone in the CrCL range of 35–100 mL/min had a relatively low frequency of C_min exceeding the presumed increased-bleeding-risk threshold concentration of 230 ng/mL (Fig. 5). However, under concomitant amiodarone therapy, the frequency of C_min above this threshold increased considerably. For apixaban 5 mg b.i.d. at CrCL 35 mL/min, 32.8% of simulated profiles were above the 230 ng/mL threshold, and nearly every fourth profile at CrCL 50 mL/min exceeded 230 ng/mL under concomitant amiodarone therapy. At a CrCL < 35 mL/min the labeled dose of apixaban 2.5 mg b.i.d. had a low prevalence of C_min above the threshold (Fig. 5). In contrast, a CrCL > 100 mL/min resulted in 8.1% of C_min below the threshold for potential underdosing of 34 ng/mL for the labeled dose of 5 mg apixaban without amiodarone.

Fig. 5.

Apixaban C_min for the labeled doses of 5 mg b.i.d. or 2.5 mg b.i.d. in 1000 simulated individuals with differing CrCL with (blue boxes) or without amiodarone (green boxes). Italic numbers above boxes show the proportion of profiles with C_min > 230 ng/mL of the respective scenario. Italic numbers below boxes show the proportion of profiles with C_min < 34 ng/mL. Dashed lines indicate an expected C_min range of 34–230 ng/mL (5 th, 95 th percentile) [13, 28]. b.i.d. twice daily, C_min trough concentrations, CrCL creatinine clearance

Different dose regimens under concomitant amiodarone therapy were explored for the CrCL ranges of 35–100 mL/min and 15–35 mL/min by scaling the dose to the renal function (Supplementary Figs. S2 and S3). As a result, a dose of 2.5 mg apixaban b.i.d. with amiodarone comedication in the CrCL range of 15–50 mL/min yielded a similar C_min distribution compared with 5 mg apixaban b.i.d. without amiodarone (Fig. 6). Considering the predicted concentration profiles, doses of 2.5 mg apixaban b.i.d. in the CrCL range of 15–50 mL/min and 5 mg b.i.d. for CrCL > 50 mL/min appear to be a potential dosing strategy under concomitant amiodarone therapy (Fig. 7).

Fig. 6.

Apixaban C_min distribution of the labeled dose (green boxes) and a potential dosing strategy for amiodarone comedication (blue boxes) at different CrCL ranges. Italic numbers above boxes show the proportion of profiles with C_min > 230 ng/mL of the respective scenario. Italic numbers below boxes show the proportion of profiles with C_min < 34 ng/mL. b.i.d. twice daily, C_min trough concentrations, CrCL creatinine clearance

Fig. 7.

Predicted apixaban concentrations over 48 h using the available strengths of 2.5 mg and 5 mg b.i.d. (simulated for 1000 individuals). Green shaded plots show the labeled dose. Blue shaded plots show the proposed dosing scheme with amiodarone comedication. Dashed lines represent the 5 th and 95 th percentile; the solid line represents the median. The dotted line indicates the bleeding threshold of C_min: 230 ng/mL apixaban [18, 27]. b.i.d. twice daily, CrCL creatinine clearance

Discussion

In this prospective PopPK analysis of apixaban in 33 cardiac surgery patients, amiodarone comedication significantly reduced apixaban CL/F by approximately 30%. In scenarios of impaired renal function (CrCL 35 mL/min) with 5 mg b.i.d. apixaban under amiodarone therapy, the median daily apixaban AUC_ss nearly doubled. Reducing the apixaban dose to 2.5 mg at CrCL< 50 mL/min under amiodarone comedication was identified as a potential dosing option.

A one-compartment model with first-order absorption best described the apixaban PK. The estimated CL/F, Vd/F, and k_a values align well with previous reports that utilized one-compartment and two-compartment models [30–32]. No effect of apixaban dose on bioavailability was found in this study, contrasting previous findings [30]. Of the different tested covariates, CrCL and amiodarone had an effect on CL/F. Previously observed covariates such as age or body weight were not confirmed in our patients, and no influence of other comedication was observed [30–32]. Furthermore, recent acute coronary syndrome (ACS) had been reported to reduce CL/F but had a low prevalence in this study and was not a significant covariate [30]. However, owing to generally high ACS prevalence in the cardiac surgery population, this might be a potential risk factor of reduced clearance in clinical practice.

The effect of CrCL on apixaban CL/F is well established, and reduced P-gp and BCRP activity have been described to reduce apixaban CL/F and increase exposure by up to 30% [30, 32–34]. The combination of impaired renal function and decreased BCRP activity has been reported to increase apixaban exposure up to 2.6-fold, which is line with the findings of this study [33].

However, the clinical interpretation of these findings is still difficult, as associations between increases in apixaban AUC and the incidence of bleeding are inconclusive so far. While in orthopedic patients an association was observed and US Food and Drug Administration (FDA) approval data reports up to a 70% higher risk for major bleeding in 1 year with a twofold increase in AUC, no association has been seen in data of patients with venous thromboembolism [31, 35–37]. Different studies reported a threshold of C_min > 230 ng/mL for increased risk of bleeding, which was exceeded in more than 30% of simulated profiles with CrCL 35 mL/min and amiodarone in this study [18, 34]. In combination with the reports of clinically relevant bleeding in patients treated with amiodarone and apixaban or other P-gp inhibitors, these associations appear to be of relevance for clinical practice [14–16, 38]. In addition, cardiac surgery patients have a particularly high prevalence of characteristics, such as postoperative renal failure, impaired hepatic function after cardiopulmonary bypass, or a high age, that are associated with a higher risk of bleeding under apixaban treatment [3, 39–41].

In contrast to edoxaban, no dose adjustment for concomitant administration of P-gp inhibitors has been included in the drug label of apixaban [13, 42]. Given the available strengths of 2.5 mg and 5 mg, an off-label dose of 2.5 mg apixaban could be evaluated for concomitant amiodarone therapy in a CrCL range of 15–50 mL/min. Especially in patients with unstable renal function or risk of overestimation of renal function (e.g. patients with sarcopenia), clinicians should consider the risk of over exposure with concomitant amiodarone therapy, in particular after cardiac surgery.

A high variability of apixaban exposure and the risk of underdosing and low exposure with subsequent thromboembolic events must also be considered [29]. Therapeutic drug monitoring could be utilized in such cases, as recommended by guidelines for special patient populations and clinical scenarios [28, 43]. However, currently there are no prospective randomized trials evaluating benefits of such an approach with regard to clinical endpoints such as bleeding incidence or thromboembolic events [28]. The 2021 guideline of the European Heart Rhythm Association on the use of non-vitamin K antagonists defined potential indications for DOAC plasma concentration measurement, e.g., obesity, comedication with risk of interactions, advanced renal impairment, or severe liver impairment [28].

However, target concentrations have not yet been defined on the basis of controlled trials, and subsequent dosage adjustment are discouraged by the guideline and should be considered only in selected cases. Nevertheless, as apixaban exposures outside the expected range have been reported to be associated with an increased risk of bleeding and thromboembolic events, it may be worthwhile to investigate whether targeting median exposure ranges by model-informed dosing improves safety and efficacy of apixaban therapy in patients treated with amiodarone [18, 29, 44]. With regard to this study, patients in the CrCL range of 30–50 mL/min with amiodarone comedication could be screened for overexposure by trough measurements and be evaluated for potential model-informed dose reduction of 2.5 mg in a future study scenario.

This study has several limitations. Firstly, this was a real-world study using sampling in clinical routine with a certain risk of imprecision in sampling and documentation. Although CL/F and Vd/F were estimated with good precision, limited data for the absorption phase impacted precision of the estimated parameter k_a. In addition, the use of residual blood instead of timed sampling might carry an additional risk of bias. Further, the small sample size and few samples in the absorption phase may have prevented the estimation of a significant effect of apixaban dose on bioavailability. Hence, the absence of such an effect cannot be excluded from the present study. To account for this, the study had a prospective design, and a comprehensive and timely documentation process was implemented to ensure high documentation quality. To assess parameter uncertainty in small-N studies, the bootstrap technique may not be reliable as noted by Comets [45] and Broeker [25]. Hence, LLP-SIR was used to assess parameter uncertainty as a robust method in small-N datasets [25]. However, still, covariate identification and estimation might have been affected by the study size [46]. Nonetheless, effects of renal function on apixaban PK have been consistently reported in literature, and the effect size of amiodarone appears to be comparable to studies with low P-gp and BCRP activity [30, 32–34]. In addition, the PK of the active amiodarone metabolite DEA, a potent P-gp and CYP inhibitor, was not specifically modeled [47, 48]. However, DEA concentrations increase over time, potentially influencing apixaban PK in long-term therapy. [49, 50]. Hence, the results represented herein may not fully represent the long-term therapy setting.

In addition, no bleeding and thromboembolic events were detected during the study period in any patient. Evaluation of clinical outcomes was limited because patients often were discharged timely after apixaban initiation. Thus, no long-term conclusions between exposure and bleeding outcomes can be drawn from this study owing to the short follow-up. Furthermore, cardiac surgery patients are a distinct patient population, potentially limiting transferability of the results, but our model estimates align with published models and suggest potential relevance also to other populations [30–32].

Conclusions

Potentially relevant effects of concomitant amiodarone therapy and creatinine clearance on apixaban clearance were observed in this study of cardiac surgery patients treated for postoperative atrial fibrillation. Consequently, patients treated with amiodarone who have impaired renal function appear to be at a higher risk of increased apixaban exposure. Exploratory analyses suggest that adjusting the apixaban dose at CrCL < 50 mL/min in such patients may help to compensate for the reduced apixaban CL/F; nevertheless, monitoring of apixaban concentrations appears necessary to avoid potential underexposure, given the high interpatient variability in exposure.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Competing Interests

The authors declare that they have no competing interests.

Contributions

Conceptualization—Benedict Morath, Kathrin I. Foerster, David Czock, Marcin Zaradzki, Matthias Karck, Walter E. Haefeli, and Sebastian G. Wicha; methodology—Benedict Morath, Ute Chiriac, Jürgen Burhenne, and Sebastian G. Wicha; formal analysis and investigation—Benedict Morath, Kathrin I. Foerster, Ute Chiriac, Jürgen Burhenne, Torsten Hoppe-Tichy, David Schrey, and Sebastian G. Wicha; Writing—original draft preparation—Benedict Morath, Ute Chiriac, and Sebastian G. Wicha; writing—review and editing—Benedict Morath, Kathrin I. Foerster, Ute Chiriac, Marcin Zaradzki, Torsten Hoppe-Tichy, David Schrey, Jürgen Burhenne, David Czock, Matthias Karck, Walter E. Haefeli, and Sebastian G. Wicha; resources—Torsten Hoppe-Tichy, Matthias Karck, and Walter E. Haefeli; supervision—Walter E. Haefeli and Sebastian G. Wicha.

Financial Interests

The authors declare that they have no financial interests.

Funding Statement

Open access funding enabled and organized by Projekt DEAL.

Data Availability

Data are available upon reasonable request.

Ethics

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. A positive vote of the Ethics Committee of the Medical Faculty of Heidelberg University was obtained before the start of the study (S-523/2021).

Consent to Participate

Informed consent was obtained from all individual participants included in the study.