Introduction

Direct oral anticoagulants (DOACs) have largely replaced vitamin K antagonists in the prevention of thromboembolic events associated with atrial fibrillation (AF), particularly in elderly patients, in whom AF prevalence reaches up to 20%.^1,2 However, DOACs have potential drug‑drug interactions of uncertain clinical relevance with several classes of drugs, including antiseizure medications (ASMs).³ Lower plasma DOAC concentrations and the subsequent reduced efficacy of anticoagulation have been reported in patients on ASMs, mainly those treated with cytochrome P‑450 3A4 (CYP3A4) and / or P‑glycoprotein inducers.⁴ Expert opinions of the European Heart Rhythm Association and the latest guidelines of the European Society of Cardiology (ESC) have shown that special caution should be exercised when using valproate, carbamazepine, or phenytoin along with DOACs.^5,6 The combination of valproate with agents that have unpredictable effects on CYP3A4 may result in variable outcomes.⁵ With respect to the new ASMs, such as brivaracetam, ethosuximide, lacosamide, pregabalin, and lamotrigine, there are no known interactions with DOACs.^5,6 Factors associated with the risk of low DOAC plasma concentrations in patients on ASMs are not well defined.^7,8 In individuals taking ASMs classified as CYP3A4 and / or P‑glycoprotein inducers, clinical failure of anticoagulant therapy with apixaban has been reported, as compared with patients not receiving ASMs,^8,9 probably due to increased expression of CYP3A4 and / or P‑glycoprotein resulting in lower plasma DOAC concentrations.⁹ In patients on ASMs other than CYP3A4 and / or P‑glycoprotein inducers, such as levetiracetam, anticoagulation failure was not observed.¹⁰ In a study conducted in 203 patients (186 on apixaban and 17 on rivaroxaban), some of whom were prescribed CYP3A4 and / or P‑glycoprotein inducers, subtherapeutic plasma DOAC concentrations were noted, except for the participants receiving levetiracetam.¹⁰ Lower rivaroxaban plasma concentrations have been documented in case reports describing patients on carbamazepine or oxcarbazepine.^11,12 Data on patients treated with anticonvulsants and dabigatran are scarce. Hager et al¹³ presented a single patient on phenytoin and concomitant dabigatran, in whom anticoagulation failed, resulting in atrial thrombus formation.

The aim of this case series was to show real‑life experience with co‑administration of DOACs and ASMs in AF patients with epilepsy.

Patients and methods

We retrospectively analyzed 27 consecutive AF patients treated with ASMs and DOACs in the outpatient neurology clinic at the St. John Paul II Hospital, Kraków, Poland, whose visits were scheduled between July 2024 and October 2024, and who, according to the attending neurologist, had indications for DOAC level assessment. All patients declared intake of the anticoagulant in the morning prior to blood draw. AF was diagnosed according to the 2024 ESC guidelines.⁶ Epilepsy was defined according to the standards proposed by the International League Against Epilepsy.¹⁴

ASMs were divided into 3 groups: 1) enzyme‑inducing ASMs (I‑ASMs), comprising CYP3A4 and / or P‑glycoprotein inducers, such as valproate, topiramate, and oxcarbazepine; 2) noninducing ASMs, including brivaracetam, lacosamide, and lamotrigine; and 3) the levetiracetam group, that is, ASMs with potential induction of P‑glycoprotein alone, with or without noninducing ASMs.⁴

The exclusion criteria were history of poor compliance, known cancer, advanced liver disease (class B or C Child–Pugh score), chronic kidney disease (stage G5), acute infections, recent surgery or trauma, and therapy with agents known to strongly affect DOAC levels (eg, rifampicin, verapamil). We also excluded patients with bipolar disease and depression.

Prior ischemic stroke diagnosis was based on symptoms and positive findings on computed tomography or magnetic resonance imaging, adhering to the World Health Organization criteria. Renal disease was defined as an estimated glomerular filtration rate (eGFR) below 60 ml/min/1.73 m².

This was a retrospective study; therefore, approval of the Bioethics Committee was not needed.

Results

We evaluated 27 patients with epilepsy (66.7% men; median [IQR] age, 73 [68–80^] years; 48% with paroxysmal AF) with a median (IQR) time since the AF diagnosis of 17.4 (8.1–22.7) months (Table 1). Most patients (88.9%) had focal epilepsy, whereas a history of ischemic stroke was the etiology of this disease in 48.1% of the cohort. More than half of the patients (59.3%) were using ASM in monotherapy, while the remainder received at least 2 different ASMs. As many as 51.9% of the patients were treated with apixaban, 22.2% with rivaroxaban, and 25.9% with dabigatran (Table 1). Reduced‑dose DOACs were used in 2 (14.3%), 3 (50%), and in 1 patient (14.3%) in each group, respectively. Other concomitant medications are listed in Supplementary material, Table S1.

We recorded nontherapeutic peak DOAC concentrations in 8 patients (29.6%), and this group was similar to the remaining patients in terms of demographic and clinical characteristics. The only exception was body mass index (BMI), which was 8.6% higher in the group with subtherapeutic DOAC concentrations (Table 1). The percentage of patients with suboptimal plasma concentrations of the anticoagulant was the highest in the rivaroxaban group, as compared with the apixaban and dabigatran groups (P = 0.006), and was not related to the daily DOAC dose. Relative to apixaban or dabigatran, treatment with rivaroxaban was associated with a 5.9‑fold increased odds of plasma peak concentrations below the therapeutic range (OR, 5.9; 95% CI, 3.2–6.9; P<⁠0.001).

Subtherapeutic plasma peak apixaban or rivaroxaban concentrations were detected in patients taking valproate, oxcarbazepine, lamotrigine, and levetiracetam (Supplementary material, Table S1). We identified 1 individual with inappropriate DOAC dosing (patient No. 1; apixaban 2.5 mg twice daily instead of 5 mg twice daily), while the 7 remaining patients had too low DOAC concentrations despite the recommended anticoagulant dosing, including 3 patients taking valproate (Supplementary material, Table S1).

Analysis of the 7 patients with subtherapeutic peak DOAC levels (Supplementary material, Table S1) showed that patients No. 16 and 17, taking valproate, and patients No. 18 and 19, taking levetiracetam, had subtherapeutic rivaroxaban concentrations, while patient No. 2, also taking levetiracetam, had a low apixaban concentration. Patient No. 15, taking lamotrigine, had low plasma rivaroxaban concentration. Additionally, patient No. 21, taking 4 ASMs combined with dabigatran, had the plasma DOAC concentration below the lower limit of the recommended range. Four of the 12 patients on levetiracetam in monotherapy or combination with other ASMs had subtherapeutic DOAC levels, including 3 individuals treated with apixaban or rivaroxaban.

None of our patients experienced ischemic stroke or thromboembolism. We observed very high (81.5%) seizure freedom rates in the current study. In 5 patients (18.5%), including 4 with subtherapeutic plasma DOAC concentrations, the ASM treatment was not effective in terms of seizure control. The dose of the ASMs was increased for all of them; however, improvement was only observed in 2 cases (Supplementary material, Table S1).

Discussion

This report is the first to show that almost one‑third of Polish patients treated with ASMs and DOAC had subtherapeutic plasma peak DOAC concentrations. It highlights potential drug‑drug interactions in this group of patients, along with the increased risk of poor seizure control when DOAC levels are too low. Moreover, to our best knowledge, our cohort includes the largest proportion of epileptic patients with AF treated with ASMs and dabigatran reported in the literature. Dabigatran is currently infrequently used in Europe; however, in some countries, including Poland, it is still commonly prescribed. Our real‑life evidence also supported data on the interactions between I‑ASMs and DOAC concentrations.

It is well known that metabolism, absorption, and elimination of DOACs are mediated by CYP3A4 and P‑glycoprotein.³ All DOACs are subject to P‑glycoprotein efflux in the gut and kidney, and they have potential drug‑drug interactions, especially with I‑ASMs.³ The effect of levetiracetam on the DOAC concentration is weak or negligible, as demonstrated in a recent report by Goldstein et al¹⁰ and the ESC guidelines.⁶ However, findings from pharmacovigilance studies highlight concerns regarding the safety of this agent in patients on DOACs.^15,16 Our results support these concerns. Available reports yielded inconsistent findings regarding a clinical impact of such therapy in terms of stroke risk.^16-18 Taken together, the combination of levetiracetam and DOACs should be used with caution.

Among the patients using ASM in monotherapy, we found subtherapeutic rivaroxaban levels in 1 individual on lamotrigine. To our knowledge, so far there have been no similar reports on lowered rivaroxaban levels in patients on lamotrigine, and experts reported no such drug‑drug interactions.⁵ Therefore, other factors, such as irregular dosing, most likely contributed to the low DOAC concentration in this case. Notably, compliance is particularly important if the anticoagulant is taken once daily. This issue deserves further investigations.

We confirmed the results of previous studies^4,10 suggesting that the combination of valproate or oxcarbazepine with DOAC may be associated with lower plasma anticoagulant concentrations, which highlights potential clinical relevance of the drug‑drug interactions in everyday practice. Of note, we observed frequent concomitant use of valproate (similarly to carbamazepine in the case of dabigatran or rivaroxaban) and DOACs despite expert opinions and ESC guidelines discouraging such strategy as suboptimal given significant drug‑drug interactions.⁶

Importantly, we noted that the patients on ASMs exhibiting lower peak plasma DOAC concentrations had higher BMI than those with therapeutic DOAC concentrations. This finding supports a growing body of evidence showing that standard DOAC doses may be insufficient in patients with obesity, as reflected in a recent ESC consensus statement suggesting that this patient population may benefit from monitoring DOAC concentrations and individual dose adjustment, where appropriate.¹⁹ This recommendation is especially relevant in patients on ASMs, , who are at risk of suboptimal DOAC concentrations. Such measurements can help optimize both antiepileptic and anticoagulant treatment.