Periprocedural anticoagulation in atrial fibrillation: Update on electrical cardioversion and ablation

Whereas for AF of more than 48 hours of duration no important changes have been reported, a stricter regimen has been introduced for anticoagulation therapy prior to ECV in the setting of short duration AF (Fig. 1; [3]). Previously, patients with AF < 48 hours were triaged for preprocedural heparin based on their stroke risk profile, but the large FinCV registry on 5116 cardioversions reported a clear increase in the risk of stroke when AF duration exceeded 12 hours. This also applied to patients with CHA₂DS₂-VASc [Congestive heart failure, Hypertension, Age ≥ 75 [doubled], Diabetes, prior Stroke [doubled]—Vascular disease, Age 65–74, Sex category] scores of 0–1 (Table 1; [11]). In the subset of 2,481 patients with neither preprocedural heparin nor subsequent anticoagulation therapy, the risk of thrombo-embolic events for CHA₂DS₂-VASc 0–1 was at least as high as for patients with CHA₂DS₂-VASc 2 [12]. Stroke rates were markedly lower when on antithrombotic therapy, with no events in CHA₂DS₂-VASc 0–1 in those on treatment versus 0.4% to those without [13]. Whereas in the 2010 European AF guideline preprocedural heparin was recommended based upon the estimated risk profile for stroke (Class I-B), the latest guidelines advocate preprocedural anticoagulation for all patients with AF < 48 hours, regardless of stroke risk factors (Class IIa-B) [3, 10]. In the context of stroke rates of 0.5–1% in the month after cardioversion, a properly powered randomised controlled trial is not realistic. This is why registry data currently represent the best available evidence on the benefit of standard anticoagulation prior to the ECV [13, 14]. Similar restrictions apply to the evidence used to change recommendations on the post-cardioversion antithrombotic regimen.

Another large registry, reporting outcome in 16,274 patients from Denmark, demonstrated that the risk of no anticoagulation therapy was similar for patients with a CHA₂DS₂-VASc score of 0–1 (odds ratio [OR] 2.21; 95% confidence interval [CI] 0.79–6.77) and those with a CHA₂DS₂-VASc score of ≥ 2 (OR 2.41; 95% CI 1.46–3.95) [14]. Whereas previously the recommendation of a 4-week anticoagulation strategy after ECV was restricted to patients with risk factors for stroke, these and other observations support the current advice to prescribe oral anticoagulation therapy after ECV in all patients [3, 10, 13‐16]. First, the risk of stroke increases after cardioversion, not only for patients with a high, but also for patients with a very low CHA₂DS₂-VASc score. For patients with a CHA₂DS₂-VASc score of 0–1, with a yearly stroke risk of up to 1.3%, the observed 30-day stroke rate post-cardioversion is about 0.5%. In the context of the estimated 1‑year stroke risk, this represents a marked increase [6, 10]. Even in patients with AF lasting less than 48 hours, the 30-day stroke risk has been reported to be considerably higher than the estimated 1‑month risk as would be expected based upon the annual stroke risk [3, 13]. In short, when a patient is referred for ECV, we predispose the patient to a higher risk of stroke, even when using the recommended postprocedural anticoagulation [6, 13]. About half of these strokes occur within 2–5 days [12, 17, 18]. Importantly, the pathophysiological mechanism of the thrombo-embolic events is multifactorial, and even more complex than during the natural course of AF [15]. Preprocedural duration of AF, CHA₂DS₂-VASc score and parameters that indicate atrial function recovery have all been associated with the risk of thrombotic complications [11, 13, 15, 19]. Importantly, the effect on electrical atrial remodelling not only applies to electrical, but also to pharmacological cardioversion [19]. These findings support the revisions in the present guideline, that endorse post-cardioversion anticoagulation in all patients with AF < 48 hours, irrespective of risk factors (Class I-B). As for anticoagulation beyond the 4‑week interval, no changes have been reported and the indication should be made based upon the CHA₂DS₂-VASc score (Class I-B).

Despite the focus on a stricter anticoagulation regimen, also in AF < 48 hours, little information is provided on the recommended dose of preprocedural heparin. In summary, most studies report for UFH a dose of 70–80 IU/kg, with a total bolus of about 5000 IU [20, 21]. The effect of UFH sets in quickly (within half an hour) and lasts for about 1–2 hours. As for enoxaparin, a subcutaneous dose of 1 mg/kg twice daily is most commonly reported; it has a slower onset of action (3–5 hours after subcutaneous injection) [20, 22, 23]. Please note that a vitamin K antagonist (VKA) prescribed as a post-cardioversion anticoagulation strategy in anticoagulation-naive patients requires bridging with enoxaparin or UFH until the international normalised ratio (INR) has reached the target range. As stated before, ECV by itself creates a pro-thrombotic state with events often seen within the first 2–5 days and thus early, optimal anticoagulation is a prerequisite [12, 17, 18].

Appreciating the low event rates in the setting of ECV, properly powered trials are a logistical challenge. Dabigatran was the first NOAC mentioned in the guidelines for the indication of ECV, based on a subgroup analysis from the RE-LY trial [24, 25]. Currently, subgroup analyses from all four large AF trials are available and several observational studies have been reported (Table 2; [24, 26‐33]). In addition, randomised trials for rivaroxaban, edoxaban and apixaban have been completed [7‐9]. Given their direct onset of the anticoagulant effect, NOACs are also promising drugs to substitute preprocedural heparin/LMWH. The only randomised trial that provides data on patients with AF < 48 hours is the EMANATE study, in which a subset of 342 patients received a loading dose of apixaban, administered at least two hours before cardioversion, the majority under guidance of transoesophageal echocardiography (TEE) [9]. Thus, most evidence of NOACs in the setting of cardioversion pertains to AF with a duration of > 48 hours (Table 3). To expedite early cardioversion, there was a minimal pretreatment duration with rivaroxaban of 1-5 days, and about 3 days with edoxaban (Table 4; [7, 8]). For apixaban, a 10 mg loading dose with subsequent cardioversion was possible (2 hours after intake), in other cases at least five regular doses of apixaban were required before cardioversion [9]. A meta-analysis on randomised data and information obtained from subgroup analyses demonstrates similar efficacy (0.4%) and safety (0.6%) between NOACs and VKA and thereby support the use of NOACs in the setting of ECV (Tables 5 and 6; [34]).

In summary, preprocedural anticoagulation for short duration AF is currently also recommended in patients without risk factors for stroke. Theoretically, the standard preprocedural administration of anticoagulation may carry a small, though somewhat higher risk of bleeding. With regard to post-cardioversion anticoagulation in all patients, this may have consequences as well, although the safety profile of NOACs is reassuring.

Importantly, for early, immediate cardioversions (for example, in case of haemodynamic instability) in anticoagulation-naive patients UFH intravenously provides an immediate anticoagulant effect. The only NOAC tested with early cardioversions is apixaban (loading dose 10 mg), but this still requires a 2-hour interval before to proceed. The other NOACs were used for at least one day before ECV was performed. As for elective cardioversions, a 3-week pretreatment period, with at least for 4 weeks of post-treatment has been studied with all NOACs.

In contrast to the situation with VKA, there is no objective tool to monitor the quality of anticoagulation of the preprocedural period with NOACs. Compliance is therefore of the utmost importance, and verification of compliance on the day of the ECV is crucial. Observations in patients treated with VKA have demonstrated that complications are much more frequent in those with a suboptimal intensity at the day of the procedure [18]. Although the randomised trials with NOACs on ECV do not provide very detailed definitions on preprocedural compliance, we advocate that none of the doses should be omitted in the three days before the procedure. The general rule that compliance is considered sufficient in case of an intake of 80% or more seems too arbitrary in the setting of cardioversion, as it does not cover the scenario that two missed doses in the three days before the procedure may still render a more than 80% intake [35].

After the completion of the COMPARE trial, uninterrupted use of VKA has become the standard, and bridging is not recommended anymore [40]. In this randomised study with over 1500 patients continuous use of VKA (INR 2.0–3.0) was associated with a significantly lower rate of thrombo-embolic events, and a 40–50% reduction in major bleeding as compared with a strategy with LMWH bridging [40]. Despite this improvement, the use of VKA has been associated with a disadvantage; the intensity of anticoagulation tends to fluctuate, and in about 20% of patients the INR is not in the target range prior to the procedure.

After the introduction of NOACs many observational studies have been conducted to describe experiences with these new agents in the setting of ablation for AF [44]. Especially the timing of holding and resumption of these new drugs was largely unknown. Given its direct onset of effect, timing of resumption may affect postprocedural bleeding. Detailed analyses of the observational studies have confirmed this, showing that the studies with the most marked rates of bleeding and tamponade were characterised by a very short NOAC-free interval. Appreciating that heparin is administered during the procedure, and the initially unknown effects of combined heparin and NOAC use, many of the first studies used a regimen of interruption of NOAC. The first observational evidence suggested numerically, though not statistically significantly higher rates of thrombo-embolic events for interrupted NOAC regimens, with similar or somewhat lower bleeding rates when compared with warfarin regimens [45]. Meta-analyses on all observational data comparing interrupted regimens of NOAC versus VKA showed no difference in thrombo-embolic events and bleeding [45]. With increasing familiarity with the new drugs, uninterrupted periprocedural NOAC use has become a more accepted strategy. For a fair comparison between NOAC and VKA with regard to bleeding complications, both should be used in a setting of uninterrupted anticoagulation. This is in contrast to a recent, randomised Asian study that showed lower bleeding rates in case of interrupted NOAC use versus continued use of VKA [46]. The available observational studies that compared continued periprocedural use of NOACs with uninterrupted VKA reported similar efficacy and safety [45]. Randomised controlled trials were eagerly awaited. Based on the reported rates of periprocedural stroke in the meta-analyses, it should be acknowledged that—a priori—these trials were not powered to address this endpoint. Therefore, the emphasis shifted towards ‘safety and feasibility’, although it should be appreciated that some of the trials have been criticised for the design and power to address major bleeding as well. As of yet, two trials are currently recruiting. First, the ELIMINATE (A prospective, randomised, open-label, blinded endpoint evaluation parallel group study comparing edoxaban versus VKA in subjects undergoing catheter ablation of non-valvular atrial fibrillation; NCT02942576) study in which uninterrupted anticoagulation with edoxaban versus VKA is the topic of interest. The trial will not be performed in the Netherlands and France due to alleged issues with the study design. However, calculations based upon the RE-CIRCUIT trial demonstrate that for the endpoint bleeding, power should not be an issue. The AXAFA trial has addressed the same study question for apixaban [9, 47]. The first randomised, though somewhat small study (n = 248) of NOACs versus VKA was VENTURE-AF, in which rivaroxaban 20 mg once daily resulted in similar bleeding rates as uninterrupted VKA [41]. The largest trial so far (n = 635) has been performed with dabigatran, where patients in the RE-CIRCUIT trial had significantly higher bleeding rates on VKA than on uninterrupted NOAC [42].

Just recently, an updated meta-analysis on 12 studies, with in total 4962 patients, suggested similar rates of thrombo-embolism (NOAC 0.08% vs. VKA 0.16%; OR 0.66, 95% CI 0.19–2.30) and demonstrated significantly lower bleeding rates (NOAC 0.9% vs. VKA 2.0%; OR 0.50, 95% CI 0.30–0.84, p < 0.01) when uninterrupted use of NOACs was compared with uninterrupted use of VKA, in observational and randomised trials taken together [44]. The first evidence on interrupted NOAC regimens looks promising with regard to safety, but additional information with regard to the embolic protection is warranted [48].