Discussion
This real-world nationwide cohort study, including 8686 propensity score-matched patients showed no significant difference regarding registered major bleeding events in patients with AF treated with amiodarone in combination with apixaban versus warfarin.
Amiodarone is an AAD used in patients with AF and is metabolised in the liver.3 7 It is well known that amiodarone interacts with the pharmacokinetics of OACs, including apixaban and warfarin.8 15 Thus, it is important to study the benefits and risks associated with different OAC combinations in patients treated with concomitant amiodarone. Our finding of a similar risk of major bleeding associated with apixaban versus warfarin in patients treated with amiodarone aligns with a few prior small observational studies of AAD (amiodarone, sotalol, flecainide, dronedarone, propafenone, dofetilide) and contemporary treatment with apixaban and warfarin.13 14 The findings also align with reports from systematic reviews which included data from both observational studies and subgroup analysis of direct oral anticoagulant (DOAC) trials.16 However, in a subgroup analysis of the ARISTOTLE trial, specifically comparing apixaban versus warfarin in patients treated with concomitant amiodarone, apixaban was associated with lower rates of major bleeding (HR: 0.61; 95% CI: 0.39 to 0.96).10 The differences in the findings from our study and the subgroup analysis from the ARISTOTLE trial might be explained by the difference in study design, the diverse definitions of major bleeding implemented by different studies and the larger sample size and the inclusion of real-life non-selected patients in the current study.
In this study, there was a significant numerically lower rate of intracranial bleeding with apixaban versus warfarin in patients treated with amiodarone. Similar findings have been reported in observational studies of dronedarone with concomitant use of apixaban and warfarin.17 Also, the ARISTOTLE trial reported that the risk for intercranial bleeding in patients on apixaban versus warfarin was statically lower with or without amiodarone.10 In contrast to the current study, earlier studies have reported higher rates of gastrointestinal bleeding with DOACs other than apixaban compared with warfarin.16 18 19 One possible explanation for this disparity could be the lower sensitivity associated with ICD codes for gastrointestinal bleeding (82.6%) compared with other bleeding outcomes.14
Another finding in our study was a numerically higher risk of all-cause mortality with apixaban compared with warfarin in patients treated with amiodarone. However, this has not been observed in prior subgroup analysis from DOAC trials or in systematic reviews in which apixaban compared with warfarin in amiodarone-treated patients was associated with an equal or lower risk of all-cause mortality.10 16 18 Explanations for this observation could be residual confounding and that patients treated with warfarin in the current study were, despite propensity score matching, more likely to receive concomitant treatment with cardioprotective drugs such as ACE inhibitors and statins, which could partially explain the slightly lower numerical rate of all-cause mortality in warfarin-treated patients.20 21
Warfarin has multiple interactions with food and other medication, mainly due to the complete hepatic metabolisation through the CYP enzymes (primarily CYP1A2, CYP2C9 and CYP3A4).22 Known interactions between other medications are fewer with apixaban than warfarin, but still present. Apixaban is partly metabolised by the CYP3A4 enzyme but is also eliminated through the P-gp efflux transporters, enabling renal clearance combined with gastrointestinal and hepatobiliary drug excretion.6 Amiodarone is a moderate inhibitor of the CYP2C9 and CYP3A4 enzymes and the P-gp transporter with the ability to increase the plasma concentration of both apixaban and warfarin.6 7 22 In the ARISTOTLE trial, apixaban was superior to warfarin in preventing stroke or systemic embolism and caused less bleeding.2 A more extensive interaction between apixaban and amiodarone may partially explain the findings in the present study, in which apixaban versus warfarin when combined with amiodarone was associated with a similar risk of major bleeding. Still, the findings in the present study suggests that apixaban is a safe OAC in patients with AF treated with amiodarone.
The recommendations from the US FDA and the European EMA include dose reduction recommendations for warfarin in patients who require concomitant amiodarone.3 7 While our study confirms the safety of apixaban in patients treated with amiodarone, future studies could explore plasma levels of apixaban in patients treated with amiodarone to fill some knowledge gaps. In clinical practice, the duration and dosage of amiodarone should be kept as short or low as possible to decrease the risk of amiodarone-associated side effects and possible interactions with apixaban.
The main strength of the current study is the large sample size representing real-world nationwide patients with no loss to follow-up. The Swedish national registries also provide unique opportunities to collect data and to combine registries, minimising missing data. To estimate medical therapy over time, the National Prescribed Drug Register was used. However, adherence to prescribed and collected medication cannot be ascertained. Missing data on patient weight prevented the determination of whether the prescribed dosage of apixaban aligned with recommended dose reduction criteria.3 Additionally, we lacked information about the true warfarin exposure in terms of INR and TTR. However, in the general warfarin population in Sweden, TTR is high, at around 77.1%.23 Due to the observational design of the study no evidence of causation could be established, and the definition of major bleeding was based on ICD-10 codes not defined by the International Society of Thrombosis and Haemostasis (ISTH). Although the definition of major bleeding used in this study has been validated, no clinical event adjudication was made.14 Another limitation of the study, potentially influencing the lack of significance for some outcomes, was the short follow-up. Due to the extracardiac toxicity of amiodarone, the treatment duration is often kept short, resulting in most patients only receiving short amiodarone prescriptions. Further limitations include confounding factors (eg, type of AF, use of NSAID) and a potential selection bias despite propensity score matching. Lastly, the primary outcome was a composite endpoint where patients were censored at the time of an event, leaving room for competing risk.