FormalPara Key Points

The risk of epilepsy is double in patients with atrial fibrillation (AF), possibly linked to silent stroke. Thus, a high proportion of patients with AF require the concomitant long-term administration of oral anticoagulation and antiepileptic therapy. Little is known about clinical interactions between antiepileptics (AEDs) and direct anticoagulants (DOACs).

In this prospective study, we found that patients with non-valvular AF treated with DOACs and antiepileptic drugs appear to have a relatively high rate of thromboembolic events.

Our findings suggest that the interaction between AEDs and DOACs might be of high clinical relevance. Indeed, caution is needed when prescribing AEDs to patients on concomitant treatment with DOACs.

1 Introduction

Direct oral anticoagulants (DOAC) are increasingly replacing vitamin K antagonists (VKA) for the prevention of stroke in atrial fibrillation (AF) and for the treatment of venous thromboembolism. In randomised clinical trials in these clinical settings, DOACs were shown to be non-inferior compared to VKAs concerning efficacy, with the advantage of a reduction in major bleeding [1, 2]. In addition, the use of DOACs in fixed doses without laboratory monitoring has led to an awaited improved practicality. Based on all these issues, current international guidelines recommend DOACs as anticoagulants of choice for patients with non-valvular AF [3].

DOACs have different pharmacokinetic profiles [4]. Direct factor Xa inhibitors such as apixaban, edoxaban and rivaroxaban are substrates of the cytochrome P450 (CYP) system, and especially the CYP3A4 isoform, while dabigatran is not substrate of cytochrome P450 system [4]. Their intestinal absorption and renal elimination are dependent on the permeability of glycoprotein (P-gp) efflux transporter protein system. Thus, concomitant treatment with drugs that influence the P-gp or CYP3A4 system might increase or decrease serum DOAC levels. Antiepileptic drugs (AEDs) seem to affect the P-gp and the CYP3A4 system [5, 6]. However, AEDs differ from each other in terms of mechanism and drug–drug interaction. While phenobarbital, phenytoin, carbamazepine, oxcarbazepine are established inducers of P-gp and the CYP3A4 systems, the pharmacokinetic properties of valproic acid and levetiracetam are still controversial [5,6,7,8].

Risk of epilepsy in patients with AF appears to be double compared to patients without AF [9]. Moreover, post-stroke epilepsy accounts for 30–40% of all cases of epilepsy in the elderly and it is a serious condition with increased mortality and reduced function outcome [10]. Thus, a high proportion of patients with AF require the long-term concomitant administration of oral anticoagulation and antiepileptic therapy.

To date, data on potential interactions between DOACs and AEDs are limited. The European Heart Rhythm recommendations do not support the use of commonly used AEDs (e.g. carbamazepine, levetiracetam, phenobarbital, phenytoin, topiramate and valproic acid) in patients concurrently taking DOACs [11]. These recommendations are based on expert opinions, DOAC summary of product characteristics and on animal or in vitro studies.

The aim of this study was to evaluate the incidence of thromboembolic and bleeding events in a cohort of patients with non-valvular AF concurrently treated with AEDs and DOACs.

2 Methods

2.1 Patients

Consecutive patients with non-valvular AF treated with DOACs were prospectively included in on-going prospective cohorts [12]. Data from patients treated with both DOACs and AEDs from the two cohorts were extrapolated and merged in a collaborative database and study. Patients were enrolled at the Internal, Vascular and Emergency Medicine-Stroke Unit in Perugia and at the Anticoagulation Clinic in Arezzo from 2014 to December 2018 [12]. These patients were prospectively observed until January 2020. Decisions concerning the choice of DOAC and AED for each individual patient were taken by the attending physician. In both cohorts, adherence to the criteria of the European Medicines Agency (EMA) for prescription of reduced doses of DOACs was suggested. Exclusion criteria were refusal of informed consent or treatment with anticoagulant agents other than DOACs. The study was approved by the Ethical Committee and/or Institutional Review Boards of the participating centres.

2.2 Data Collection

For each patient, the following data were collected: age, sex, comorbidities (hypertension, congestive heart failure, diabetes, previous stroke, vascular diseases, previous bleeding), previous vitamin-K antagonists use, type of AF (first diagnosed, paroxysmal and permanent AF), laboratory tests (e.g. creatinine clearance, eGFR according to Cockroft-Gault formula), type and dosage of DOAC, type of AEDs. Risks of stroke or bleeding were defined according to the CHA2DS2-VASc and HAS-BLED score [13, 14].

Patients’ observation started at the time of DOAC prescription and ended in case of death or permanent DOAC or AED discontinuation. Follow-up visits or telephone contacts were scheduled every 6 months or in case of signs/symptoms of study outcome events.

2.3 Study Outcome Events

The primary effectiveness outcome was the composite of ischaemic stroke, transient ischaemic attack (TIA) and systemic embolism (SE). The primary safety outcome was major bleeding (MB). Secondary outcomes of the study were (i) overall mortality and (ii) clinically relevant bleedings (CRB).

Stroke was defined as a focal neurological defect, lasting at least 24 h while TIA as a transient episode of neurological dysfunction resulting from focal cerebral ischaemia not associated with permanent cerebral infarction. Bleeding was defined as a major event according to the International Society on Thrombosis and Haemostasis (ISTH) definition, if clinically overt and associated with a decrease in the haemoglobin level of 2.0 g per decilitre or more, transfusion of 2 or more units of red cells, if it was intracranial or retroperitoneal, if it occurred in another critical site, or contributed to death [15]. CRB were the composite of major bleeding and clinically relevant non-major bleeding. Clinically relevant non-major bleeding was defined according to the ISTH criteria as overt bleeding that did not meet the criteria for major bleeding but was associated with medical intervention, unscheduled contact with a physician, interruption or discontinuation of DOAC [15]. In case of death, the presumed cause was reported as adjudicated by the attending physician. The results of all the available tests used to confirm/exclude bleeding or thromboembolism as the cause of death were reported. Permanent discontinuation was defined as drug withdrawal for a period longer than 30 days.

2.4 Statistical Analysis

Categorical data were reported as frequencies and continuous data as mean ± SD or median (range). Study outcome events were presented as number of events per 100 person-years. We performed Kaplan–Meier survival analysis to estimate the time course of stroke/TIA/SE and of MB in the overall population and according to the type of AEDs: enzyme-inducing AEDs (phenytoin, phenobarbital, carbamazepine, oxcarbazepine) versus other AEDs.

Cox regression analysis was used to identify predictors of stroke/TIA/SE and of MB. The predictors included in the analysis were: age, sex, previous stroke or TIA, reduced dose of DOAC, previous use of VKA, CHA2DS2-VASc and HAS-BLED score. Results were reported as hazard ratio (HRs) with 95% confidence intervals (CIs). P-values ≤0.05 were considered statistically significant. Data were analysed with the SPSS/PC Win package 22.0.

3 Results

Of 2301 patients included in the two cohorts, 91 (3.9%) patients who were receiving DOACs in combination with antiepileptics were identified and included in this analysis. Mean age was 78 ± 9.5 years, 45 (49.5%) patients were female and 44 (48%) had a previous stroke or TIA. Main clinical features of study patients are reported in Table 1. Overall, 15 (16.5%), 25 (27.5%), 42 (46.2%) and 9 (9.9%) patients were treated with dabigatran, rivaroxaban, apixaban and edoxaban, respectively. Reduced doses of DOAC were prescribed in 34 (37.4%) patients. Before starting DOACs, 20 (22%) patients were on treatment with VKAs. Among study patients, 41 (45%) were on treatment with levetiracetam, 20 (22%) with valproic acid, 11 (12%) with phenobarbital, 10 (11%) with carbamazepine and 9 patients (10%) with other AEDs (Table 2). Accordingly, 24 (26%) patients were on treatment with enzyme-inducing AEDs and 67 (74%) patients with other AEDs.

Table 1 Main baseline clinical features
Full size table
Table 2 Comparison of the main clinical features of our study with post-marketing atrial fibrillation studies
Full size table

3.1 Study Outcome Events

Median follow-up was 17.5 ± 14.5 months.

During the study period, ischaemic stroke/TIA/SE occurred in 9 patients (5.7% patient-year). Of them, 8 patients had ischaemic stroke and 1 patient had a TIA (Fig. 1, Table 2). Ischaemic stroke was fatal in 3 patients (1.9% patient-year). The main characteristics of thromboembolic events are reported in Table 3. All patients who experienced a thromboembolic event were aged ≥ 75 years, 6 (67%) had a CHA2DS2-VASc score > 3 and a history of stroke. One patient who died from ischaemic stroke was on concomitant treatment with phenobarbital and levetiracetam. No differences were observed in terms of risk of ischaemic stroke/TIA/SE between patients treated with enzyme-inducing AEDs and patients treated with other AEDs (log-rank = 0.342).

Fig. 1
figure 1

Risk and incidence rates of ischaemic stroke/TIA/SE. pts-y patient-year, SE systemic embolism, TIA transient ischaemic attack

Full size image
Table 3 Main characteristics of the study patients with thromboembolic events
Full size table

MB occurred in 3 patients (1.9% patient-year), one of whom had fatal genito-urinary bleeding (0.6% patient-year) (Fig. 2, Table 2). In each patient who experienced a study outcome event, the DOAC dose was the one recommended according to EMA criteria.

Fig. 2
figure 2

Risk and incidence rates of major bleeding. pts-y patient-year

Full size image

Seven patients had CRB (4.5% patient-years). All-cause death occurred in 14 patients (8.7% patient-year). During the study period, 3 patients permanently discontinued anticoagulant treatment (1.9% patient-year). Reasons for discontinuation were starting of anti-hepatitis C therapy in one patient and progressive failure of renal function and TIA in the other.

3.2 Predictors of Primary Outcomes

At univariable analysis, increasing age was significantly associated with an increased risk of stroke/TIA/SE (HR 1.1, 95% CI 1.0–1.2, p = 0.04). The only predictor of MB was the HAS-BLED score (HR 4.5 for each point increase, 95% CI 1.0–20.5, p = 0.04). No association was observed between stroke/TIA/SE or MB and the other tested predictors.

4 Discussion

In this prospective cohort of patients with non-valvular AF treated with DOACs and AEDs and followed for nearly 1.5 years, we observed a relatively high rate of ischaemic stroke/SE (5.6% patient-year) and a rate of MB of 1.9% patient-year.

To date, little is known about the potential interaction between DOACs and AEDs since data are mainly available from animal or in vitro studies and there are only a few case reports and one retrospective study in humans [16,17,18,19,20,21,22,23]. To the best of our knowledge, this is the first study reporting on the incidence of ischaemic stroke/TIA/SE in a cohort of AF patients treated with DOACs and AEDs. Although a formal comparison cannot be performed, we found an overall rate of thromboembolic events of 5.7% patient-year, which was higher than rates observed in the main post-marketing cohort studies of patients with AF treated with DOACs [24,25,26]. These could partly be explained by the different patient profile. Indeed, our patients were older, had a higher thromboembolic risk score and a higher rate of history of stroke compared to patients included in the main AF post-marketing cohort studies.

It is well known that the potential interaction between DOACs and AEDs is based on the P-gp and CYP3A4 systems [4]. However, AEDs differ from each other in important pharmacokinetic characteristics and potential drug–drug interaction. In fact, phenobarbital, phenytoin, carbamazepine and oxcarbazepine are established inducers of both metabolic enzymes such as CYP3A4 and/or P-glycoprotein [7, 8]. In contrast, other AEDs such as valproic acid and levetiracetam have pharmacokinetic properties that are still questionable and studies documenting their effects on the concentrations of CYP3A and/or P-gp substrates are few and controversial [4]. Potential drug–drug interactions still remain unknown in other AEDs such as pregabalin, gabapentin, lamotrigine, lacosamide [4].

Carbamazepine is a strong inducer of CYP3A4 and P-gp transporter and can cause a marked reduction of DOAC in the plasma concentrations by 80% or more in humans [16,17,18]. In animal studies, induction of P-gp was found by levetiracetam, phenobarbital and phenytoin [19, 20]. However, the clinical data show that levetiracetam is not a P-gp inducer [27]. In contrast, in a recent case report, levetiracetam seems to reduce rivaroxaban plasma levels acting as P-gp inducer [28]. No effect was observed for levetiracetam and CYP3A4 activity in vitro [21]. In in vitro studies, valproic acid increased P-gp activities while controversial results have been reported for its potential interactions with CYP3A4 system [6, 22, 23].

Whether these potential interactions may have a clinically relevant significance is largely unknown. In a recent analysis of adverse event reports submitted to Food and Drug Administration (FDA) Adverse Event Reporting System, the risk of ischaemic and thromboembolic events in patients concomitantly treated with factor Xa-inhibitors, mainly apixaban and rivaroxaban, and enzyme-inducing AEDs, was higher than in those treated with other antiepileptic drugs [29]. In addition, in a retrospective study, DOAC concentrations were lower than expected in patients treated with carbamazepine, phenobarbital and phenytoin (50% of patients below the 5th percentile) [30]. In our study, we included a wide range of AEDs and found that patients concomitantly treated with DOACs and AEDs appear to have a relatively high rate of thromboembolic events. However, the study was not powered to detect potential differences between DOACs or AEDs. Indeed, whether this increased risk of ischaemic events is related to the enzyme inducing effect of some of these medications, or to the underlying clinical features such as history of stroke or pathological characteristics of these patients is unclear.

However, the number of patients treated with known enzyme-inducer AEDs was substantially lower than the number of patients treated with antiepileptics that are not considered enzyme inducers. In our study, the risk of ischaemic stroke/TIA/SE between these two groups of patients did not differ. This may in part be related to the limited number of outcome events occurring during the study period. However, we believe that a more cautious approach by physicians should be considered when prescribing carbamazepine, phenobarbital or phenytoin to patients on DOACs.

In our study, we found an incidence of major bleeding of 1.9% patient/year, lower than that reported in a recent retrospective study on Asian insurance claims data. A higher risk of major bleeding was observed in patients on DOACs and valproic acid (15.3% patient-year), phenytoin (13.5 patient-year) and levetiracetam (13.2% patient-year) compared to patients on DOACs alone [31]. However, data on bleeding-related potential covariate such as dosage of DOAC, liver and renal function as well as thromboembolic events were not reported in this study.

Finally, this study has some limitations. The limited number of patients included, the absence of a comparison group, the pooling of a wide range of anticoagulants and AEDs, each with distinct pharmacokinetic properties, and the lack of dosage of AEDs and that of DOACs or AEDs plasma concentration may limit the value of our results. However, the prospective data collection and the duration of the study follow-up are strengths of our study.

5 Conclusions

In conclusion, we observed that patients concomitantly treated with DOACs and AEDs appear to have a relatively high rate of thromboembolic events. Indeed, the interaction between AEDs and DOACs might be of high clinical relevance, leading to a potential increase in morbidity and mortality. Caution should be applied when considering anticoagulation treatment in AF patients treated with AEDs.