FormalPara Key points

In our cohort, levetiracetam clinical use was not associated with a higher percentage of patients having apixaban or rivaroxaban peak plasma concentrations (Cmax) below their therapeutic range, while the use of enzyme-inducing antiseizure medications was associated with direct oral anticoagulant Cmax below the therapeutic range.

Levetiracetam clinical use was not associated with a significant difference in apixaban-Cmax compared to patients receiving no antiseizure medications.

In multivariate linear regression, co-treatment with enzyme-inducing antiseizure medications was associated with a significant reduction in apixaban-Cmax, while a reduced dose, renal failure, female sex, and the clinical use of enzyme-inhibiting drugs were associated with higher apixaban-Cmax.

1 Introduction

The emergence of numerous antiseizure medications (ASMs) in the last 25 years has had a high impact on the management of epilepsy. Levetiracetam (LEV), introduced in 2000, offers distinct benefits that makes it a preferred choice over older ASMs. For example, at doses of 250–4000 mg/day, LEV has not been associated with an increased risk of major congenital malformation, in contrast to the risks associated with valproic acid (at doses > 650 mg/day) or carbamazepine (at doses > 700 mg/day) [1]. In addition, LEV seems to provide equivalent or superior seizure control, with less involvement in enzyme induction/inhibition-based drug–drug interactions compared with enzyme-inducing (EI)-ASMs [2]. Studies have shown that LEV is not a potent inducer or inhibitor of cytochrome P450 enzyme (CYP) or P-glycoprotein (P-gp), suggesting that it is unlikely to alter the pharmacokinetics of drugs that are CYPs or P-gp substrates [3].

Direct oral anticoagulants (DOACs) have emerged as a significant advancement in anticoagulation therapy, offering several benefits over vitamin K antagonists. DOACs, including apixaban, rivaroxaban, dabigatran, and edoxaban, have been suggested to have non-inferior effectiveness and superior safety in preventing thromboembolic events compared with warfarin [4, 5]. Direct oral anticoagulants also exhibit fewer food and drug interactions, making them more convenient for patients. However, one concern with DOACs is their potential for drug interactions with inducers of drug-metabolizing enzymes. Enzyme-inducing drugs (e.g., carbamazepine, phenobarbital, and phenytoin) can increase the metabolism of DOACs, specifically apixaban and rivaroxaban, leading to subtherapeutic anticoagulant concentrations and reduced efficacy [6,7,8,9,10]. Current recommendations advise clinicians to avoid combining EI-ASMs and DOACs. In cases where such a combination is unavoidable, considering alternative strategies such as plasma concentration-guided dosing may be required to maintain optimal therapeutic DOAC concentrations [11].

Indeed, EI-ASMs have been correlated with low DOAC plasma concentrations [7] as well as with an increased frequency of thrombotic events in patients treated with DOACs [6, 8, 9]. As LEV is an ASM not known to induce metabolizing enzymes, it has been suggested as a safer option for patients taking DOACs. However, there are no studies to support this practice and the European Heart Rhythm Association recommends caution in patients co-treated with LEV and DOACs [12].

In preclinical studies, LEV has shown a weak induction effect on P-gp [13, 14]; however, a clinical study in healthy volunteers showed that LEV did not affect the pharmacokinetics of the P-gp substrate digoxin [15]. Recent findings from pharmacovigilance studies further highlight concerns about the combination, revealing an elevated risk of reported ischemic stroke in individuals concurrently prescribed LEV and DOACs [8, 16].

Thus, we sought to explore the effect of LEV on apixaban and rivaroxaban plasma concentrations. Our study analyzed apixaban and rivaroxaban concentrations in patients co-treated with LEV in comparison to two control groups: (a) patients treated with EI-ASMs, known to reduce DOAC concentrations (‘positive’ control group) and (b) patients not treated with either LEV or EI-ASMs (‘negative’ control group).

2 Methods

2.1 Study Design

A retrospective cohort study was designed to include patients hospitalized in internal medicine, cardiology, and neurology wards in our medical center if they had an active apixaban or rivaroxaban prescription during hospitalization between August 2015 and December 2023 as reviewed by the institutional DOAC monitoring program. The program, implemented by clinical pharmacists who thoroughly assess each DOAC prescription to ensure its appropriateness and adherence to established guidelines and protocols, aims to minimize the risk of medication errors, optimize therapy, and enhance patient safety in the management of anticoagulation therapy [17]. This study was approved by the Institutional Review Board of the Hadassah University Hospital.

2.2 Study Participants

Consecutive patients with an active prescription for DOAC (rivaroxaban or apixaban) with a corresponding DOAC peak plasma concentration (Cmax) from August 2015 to December 2023 were included. Some of these patients contributed data to a previous publication [7]. No patients in the study received dabigatran concomitantly with LEV; therefore, we excluded dabigatran from the study. No patient in our cohort had edoxaban concentrations as edoxaban was not marketed in Israel until 2023.

Electronic medical records of patients with an active prescription for apixaban or rivaroxaban were queried for DOAC plasma concentrations, anti-Factor Xa (anti-Xa) levels, medications administered, age, sex, weight, medical history, and laboratory data. Patients identified with an active prescription for apixaban or rivaroxaban were included in the study if an anti-Xa level corresponding to the DOAC plasma concentration was reported in the electronic medical record at a time correlated with the DOAC Cmax, namely 2–4 hours after administration of apixaban and rivaroxaban, as per the medical center’s protocol for the DOAC concentration measurement. Patients were excluded if the DOAC plasma concentration monitoring was performed because of acute bleeding, overdose, or non-adherence. Patients with acute thromboembolism were excluded as well. In patients taking LEV or ASM, we verified a minimum of 2 weeks of ASM administration prior to an apixaban or rivaroxaban plasma concentration measurement to ensure maximal the induction of CYP enzymes had been reached [11].

2.3 Determination of DOAC Plasma Concentrations

Apixaban and rivaroxaban concentrations were determined using a calibrated anti-Xa-based assay, following established methods that have been previously published [18, 19].

2.4 Data Analysis

In patients with multiple apixaban and rivaroxaban concentration measurements, only the first test was included. Direct oral anticoagulant concentrations were classified as below, within, or above the therapeutic range, defined as between the 5th and 95th percentile range for apixaban and rivaroxaban as reported by analyses of phase III studies [20, 21]. Therapeutic ranges for each drug, dose, and indication are defined in previous reports [20, 21].

According to the DrugBank database, strong CYP3A4/P-gp EI-ASMs identified in our study cohort included carbamazepine, phenytoin, phenobarbital, primidone, and oxcarbazepine [22]. According to the US Food and Drug Administration database, moderate CYP3A4/P-gp inhibitors included dronedarone, verapamil, diltiazem, fluconazole, and ciprofloxacin [23] as well as amiodarone [24]. Strong CYP3A4/P-gp inhibitors that were simultaneously used included ritonavir as a component of Paxlovid® [23].

Drug administration of EI-ASMs, or LEV, was determined based on the documentation of administration in the electronic medical record at the time of the first DOAC concentration test. These lists were subsequently reviewed by a clinical pharmacist (RG) and a clinical pharmacologist (MM). The proportions of patients with apixaban and rivaroxaban concentrations below the therapeutic range were compared in patients receiving EI-ASMs, patients receiving LEV, and those not receiving any ASM.

2.5 Statistical Analysis

Cohort characteristics were described using univariate statistics. Categorical variables were presented as numbers and percentages, and continuous variables represented by means or medians along with measures of variability. The differences between the three groups (patients receiving EI-ASMs, patients receiving LEV, and patients not receiving any ASM) was assessed using the Chi-squared test (χ2) for categorical variables, and analysis of variance with the post-hoc Tukey test or Mann–Whitney U-test and Kruskal–Wallis test for continuous variables, as appropriate. To compare patients’ characteristics between groups and regarding DOAC Cmax (below, within, or above the range), binary logistic regressions were employed using the Enter method.

As apixaban was the most commonly administered DOAC in our cohort (92%), an analysis of apixaban Cmax as a continuous variable was also performed. Apixaban plasma concentrations were log-transformed to ensure normality of data for parametric tests. Non-parametric tests were performed to validate results. A multivariate linear regression model of apixaban plasma concentrations was developed to evaluate variables influencing apixaban Cmax.

3 Results

3.1 Patient Characteristics

Two hundred and three patients with 261 apixaban or rivaroxaban Cmax tests were analyzed. These included 242 apixaban tests (from 186 patients) and 21 rivaroxaban tests (from 17 patients). Five patients had concentration tests for both DOACs.

Of the 78 patients taking concurrent LEV and apixaban or rivaroxaban, 40 were excluded as the timing of blood monitoring did not correspond to their time to Cmax . Ten other patients were excluded from the LEV group for the following reasons: not fulfilling inclusion criteria (n = 2), non-documented concurrent DOAC-ASM administration (n = 2), LEV initiated after Cmax was obtained (n = 1), and concurrent LEV and EI-ASM treatment (n = 5), these patients were included in the EI-ASM group. The remaining 28 patients with concomitant DOAC and LEV were included in the analysis, in addition to 142 controls and 33 patients receiving EI-ASMs.

Of the analyzed patients (n = 203), 111 (55%) were female and 92 (45%) were male, with a mean (± standard error) age and weight of 78 ± 0.8 years, and 74 ± 1.8 kg, respectively, and mean plasma creatinine levels of 111 ± 5.2 µmol/L. There were no significant differences in baseline characteristics between the three groups, including mean age, weight, and creatinine level, as well as the proportion of reduced dosing per guidelines (Table 1). Baseline characteristics also did not differ in the analysis of patients treated with apixaban only.

Table 1 Patient characteristics
Full size table

Indication for DOAC treatment included atrial fibrillation in 177 patients (87%), venous thromboembolism in 22 patients (11%), and both atrial fibrillation and venous thromboembolism in one patient. In three patients, the indication was not recorded.

Fifty-four (27%) patients were treated with moderate combined CYP3A4/P-gp inhibitors (e.g., dronedarone, verapamil, diltiazem, amiodarone, fluconazole, or ciprofloxacin), and one with the strong CYP3A4/P-gp inhibitor, ritonavir (as a component of Paxlovid®).

Among the 28 patients treated with LEV and DOACs, the LEV daily dose range was 500–4000 mg/day, and in 25 of the 28 patients (89.3%) the range was 1000–3000 mg/day. There was no correlation between the LEV daily dose and DOAC Cmax (Spearman’s rho 0.37, p = 0.853).

3.2 Factors Associated with Apixaban and Rivaroxaban C max Below, Within, and Above the Therapeutic Range

A bivariate analysis of sociodemographic and clinical variables and apixaban and rivaroxaban concentration measurement is presented in Table 2. Female sex, plasma creatinine, and concomitant use of CYP3A4/P-gp inhibitors were statistically significantly associated with a higher proportion of patients having DOAC concentrations above the therapeutic range. The use of EI-ASM was significantly associated with a higher proportion of patients having DOAC plasma concentrations below the therapeutic concentration range, compared with patients co-treated with LEV and with the no-ASM control group.

Table 2 Bivariate relationship between patients’ characteristics (n = 203) and DOAC concentrations at first measurement
Full size table

3.3 Factors Associated with Apixaban and Rivaroxaban C max Below the Therapeutic Range

To determine the odds of apixaban and rivaroxaban concentrations being below the therapeutic range, we used the patients’ first recorded DOAC concentration (Table 3). The percentage of patients with an apixaban or rivaroxaban plasma concentration below the therapeutic range was 10.6% (15/142) among controls, 7.1% (2/28) among patients treated with LEV, and 36.4% (12/33) among patients treated with EI-ASMs (χ2, p = 0.001). Levetiracetam treatment was not associated with an increased odds of having a below-range apixaban or rivaroxaban plasma concentration (adjusted odds ratio = 0.70 [95% confidence interval (CI) 0.19–2.67], p = 0.61), while EI-ASM treatment was associated with an increased odds of a below-range apixaban and rivaroxaban concentration (adjusted odds ratio = 12.66 [95% CI 3.87–41.22], p < 0.001).

Table 3 Binary logistic regression of odds of below-range peak DOAC concentrations
Full size table

3.4 Factors Associated with Apixaban Cmax

As 92% of the patients studied were treated with apixaban and only a few patients received rivaroxaban, we calculated continuous variables in an analysis of patients taking apixaban. Of 191 patients with apixaban concentrations, 26 patients were taking LEV, 132 were not treated with any ASM, and 33 patients were taking EI-ASMs.

In patients taking EI-ASMs, mean apixaban Cmax was 126 ng/mL (95% CI 101–153 ng/mL) lower than in patients treated with LEV (164 ng/mL [95% CI 134–194]) and lower than controls )179 ng/mL [95% CI 163–197]) [log-transformed data, analysis of variance, p = 0.005, verified using the Kruskal–Wallis test). In the post-hoc Tukey test, apixaban Cmax was not different among patients taking LEV and controls (p = 0.778), while patients taking EI-ASMs had significantly lower concentrations than controls (p = 0.003) and there was a trend towards lower concentrations in patients taking EI-ASMs as compared with patients taking LEV (p = 0.136) (see Fig. 1).

Fig. 1
figure 1

Apixaban peak plasma concentrations (mean, 95% confidence interval, analysis of variance, p = 0.005). conc. concentration, EI-ASM enzyme-inducing antiseizure medication, LEV levetiracetam

Full size image

3.5 Multivariate Analysis: Variability in Apixaban C max

We performed a multivariate linear regression to analyze variables influencing apixaban concentrations (Table 4). Figure 2 presents the results of the multivariate linear regression model.

Table 4 Multivariate linear regression model for log-transformed peak apixaban concentrations (r2 = 0.40)
Full size table
Fig. 2
figure 2

Multivariate linear regression model for log-transformed apixaban peak plasma concentrations (beta coefficient, 95% confidence interval). The line at 0 represents no influence on apixaban peak plasma concentration changes. Factors contributing to a decrease in plasma concentration feature on the left, while on the right, factors contributing to an increase in plasma concentration are presented. CYP cytochrome P450, P-gp P-glycoprotein, yrs years

Full size image

As expected, reduced apixaban dosage was associated with decreased apixaban Cmax. Female sex and plasma creatinine were associated with increased apixaban Cmax. In addition, LEV was not associated with the variability in apixaban Cmax. Drugs that inhibit CYP3A4/P-gp were significantly associated with an increase in apixaban concentrations, whereas EI-ASMs were associated with significantly lower apixaban Cmax.

4 Discussion

We have shown that the proportion of patients with apixaban and rivaroxaban Cmax below the therapeutic range was similar amongst patients taking LEV and negative control patients not treated with either LEV or EI-ASMs, and significantly lower compared with positive control patients co-treated with apixaban or rivaroxaban and EI-ASMs. Similarly, in the sub-analysis of patients treated with apixaban, we found that apixaban Cmax in patients taking LEV was similar to that of the no-ASM control patients, and was higher than the Cmax in patients co-treated with DOACs and EI-ASMs.

Female sex was associated with higher apixaban concentrations. This may be explained by the fact that while weight is a continuous variable, apixaban dose reduction is based on a weight cut-off value (60 kg) leading to more female patients being on the lower end of the weight spectrum but still receiving the higher dose as they do not qualify for a dose reduction.

We have previously shown that EI-ASM use was associated with subtherapeutic DOAC concentrations [7] and treatment failure [6]. The European Society of Cardiology recommends avoiding concurrent use of EI-ASMs with DOACs, and also warns against co-medication of LEV with DOACs [12]. As a result, clinicians may avoid LEV in post-stroke epilepsy in patients treated with DOACs [25], even though LEV is considered first-line therapy for post-stroke epilepsy [26] because of its efficacy [27]. The potential mechanisms associated with possible LEV-DOAC interactions are unclear. Although preclinical studies showed a possible P-gp induction by LEV [13, 14], pharmacokinetic studies have indicated that LEV does not have a substantial clinical impact on P-gp transport or CYP3A4 metabolism [3, 15]. A case report showed decreased rivaroxaban concentrations with LEV administration [28]. A small observational registry study including patients taking dabigatran, rivaroxaban, and apixaban suggested that LEV is not likely to be a potent inducer of DOAC transport and metabolism [29]. This study [29] found that DOAC trough concentrations in patients taking concurrent LEV were within therapeutic trough ranges and no thromboembolic events were recorded in this study [29] as well as in a recent study of Ip et al. [9]. Similarly a pilot study of DOAC drug interactions reported a low probability of low DOAC concentrations with concomitant LEV [30].

Our results are concordant with these findings, showing that LEV use was not associated with decreased apixaban Cmax while EI-ASM use was associated with decreased apixaban concentrations. Previous studies were inconsistent regarding the risk of thromboembolic events in patients treated with concurrent LEV and DOAC [8, 9, 11, 16, 31]. Gronich et al. showed a more than two-fold higher risk of stroke or systemic embolism when LEV was co-administered with DOACs. A nationwide cohort study showed an increased risk of stroke in patients taking LEV [31]. A prospective cohort study also found a relatively high rate of stroke in patients taking DOACs together with ASMs compared with previously published rates of stroke in the AF population takings DOACs. The studied ASMs included LEV, valproic acid, and EI-ASMs [32].

A recent study of the FDA Adverse Event Reporting System (FAERS) database identified a significant reporting odds ratio of 3.57 for ischemic stroke in patients taking the DOAC-LEV combination [16]. Conversely, according to a larger study by Ip et al., patients taking LEV with DOACs were not at an increased risk of stroke, venous thromboembolism, or death [9]. In addition, concurrent LEV use has been associated with major bleeding in patients taking DOACs [33].

In a recent analysis, we have examined the extent of changes in various pharmacokinetic parameters of DOACs in patients co-treated with EI-ASM [11]. The results of the current study combined with our previous analysis suggest that any possible increased thromboembolic risk in patients concurrently treated with LEV and DOAC is not likely to be explained by the induction of DOAC’s metabolizing enzymes or by other pharmacokinetic mechanisms.

The mechanism for the increased risk of stroke that was observed with LEV in some [8, 16, 31] but not all [9] previous studies is not clear. Levetiracetam use by itself has been suggested to be associated with an increased risk of stroke [34]. Given the diverse pharmacokinetic variability among the various DOACs, the anticipated impact of LEV on the pharmacokinetics of individual DOACs would be expected to differ [11]. Nonetheless, the ROR of LEV with each DOAC was found to be similar [16], thereby reinforcing the notion that the heightened risk of stroke is not likely to be a result of a pharmacokinetic interaction. Other proposed mechanisms for the increased risk of stroke among patients treated with LEV include an increased inherent thromboembolic risk in patients with epilepsy [35] or in patients using ASMs [34, 36], or a recurrent stroke risk in patients with post-stroke epilepsy [16, 37].

The findings from this study suggest that the effect of LEV on the metabolism of apixaban and rivaroxaban is unlikely to be clinically significant. Thus, the previously reported LEV effect on the treatment failure of DOACs is not due to the induction of these DOACs’ metabolism or P-gp-mediated transport.

Our study has several limitations. The retrospective observational design of the study does not allow a cause–effect relationship to be established between the use of LEV or EI-ASM use and the variability in DOAC concentrations. The second consideration is the relatively small sample size of the study. This limits the use of analytical statistics, especially for uncommon outcomes. Results of the binary logistic regression should be interpreted with reservation, and larger studies are needed to confirm these findings. Nevertheless, currently, this is the largest study performed in a clinical setting to examine the effect of LEV on apixaban or rivaroxaban plasma concentrations. In addition, the results of this analysis are consistent with the multivariate linear regression of apixaban concentrations performed in this study. In order to reduce the variability associated with clinical conditions, we did not include patients in whom monitoring of DOAC plasma concentrations was performed because of acute bleeding, acute thromboembolism, overdose, or non-adherence.

Validated therapeutic concentration ranges have not been established for DOACs. However, clinical guidelines have suggested using DOAC concentrations to guide decisions in the following clinical situations: (a) surgical procedures in patients taking DOACs; (b) patients with extreme bodyweights; (c) patients with severe kidney failure; and (d) patients with clinically relevant drug interactions [12].

5 Conclusions

As substrates of CYP3A4 and P-gp, apixaban and rivaroxaban are prone to clinically significant drug interactions with EI-ASMs and other medications that affect their metabolizing enzymes. Guidelines have suggested caution for the use of these DOACs with LEV because of concern for the risk of thromboembolism and stroke. Our study suggests that pharmacokinetic mechanisms are unlikely to explain a previously reported increased risk for ischemic stroke in patients co-medicated with LEV and apixaban and rivaroxaban.