Introduction
Annually, approximately 10% of patients receiving chronic anticoagulation therapy undergo diagnostic or therapeutic procedures that are associated with bleeding risks and require therapy interruption [
1]. In particular, patients scheduled for major surgery have a high risk of bleeding. With the growing use of direct oral anticoagulants (DOACs), physicians must optimise periprocedural-DOAC management to balance the risk of bleeding with that of thromboembolic events. While clinical guidelines recommend that surgeries with high bleeding risks (e.g., major surgeries) should utilise temporary DOAC interruption, there are many less invasive procedures that carry a relatively low bleeding risk and do not necessitate interruption [
1,
2]. However, real-world data on the safety and periprocedural management of DOAC therapy in the setting of major surgeries with a high risk of bleeding are limited.
Previous studies have reviewed and assessed the pharmacological properties [
3] and periprocedural management of DOACs, including rivaroxaban, dabigatran, and apixaban [
4]. In the prospective, noninterventional Dresden registry, patients who underwent major procedures were significantly more likely to experience bleeding and major cardiovascular (CV) events as well as CV death when compared with patients who underwent minimal and minor procedures [
5]. Additionally, in the prospective, nonrandomised PAUSE trial, rates of major bleeding were higher in patients undergoing a high-bleeding-risk procedure treated with apixaban (2.96%) or rivaroxaban (2.95%) compared with dabigatran-treated patients (0.88%) [
6]. Notably, a subgroup of dabigatran-treated patients with creatinine clearance (CrCL) < 50 mL/min undergoing a high-bleeding-risk procedure had slightly longer preprocedural DOAC interruption compared with patients treated with apixaban or rivaroxaban undergoing a high-bleeding-risk procedure [
6].
Edoxaban is a DOAC approved for the prevention of stroke and systemic embolic events (SEEs) in patients with nonvalvular atrial fibrillation (AF) and for the prevention and treatment of venous thromboembolism (VTE) [
7‐
10]. Real-world data regarding periprocedural-edoxaban management are limited, especially in patients undergoing major surgeries.
The EMIT-AF/VTE programme (Edoxaban Management in Diagnostic and Therapeutic Procedures) was designed to investigate bleeding and thromboembolic events prospectively in patients with AF or VTE treated with edoxaban and undergoing procedures of varying risk levels [
11,
12]. Primary analysis of the EMIT-AF/VTE data showed low rates of periprocedural major bleeding, clinically relevant nonmajor bleeding (CRNMB), acute thromboembolic events, and acute coronary syndrome in edoxaban-treated patients who underwent a wide range of diagnostic and therapeutic procedures [
13]. The objective of this subanalysis is to compare the periprocedural management of edoxaban and clinical outcomes in patients who underwent major vs. nonmajor surgeries.
Observations and outcomes
Observations, including edoxaban interruption and clinical event data, were recorded from 5 days before the procedure until 30 days after the procedure. To enhance data capture, patients received a memory aid booklet at study enrolment, which was reviewed at the end of the study. Edoxaban therapy was considered as uninterrupted if treatment was administered on each day of the observation period, including at any time on the day of the procedure. Any interruption of edoxaban treatment was recorded as the number of days without administration of edoxaban (preprocedural days [5 days before procedure and at procedure day] and/or postprocedural days [within 30 days after procedure]).
The primary safety outcome was the incidence of major bleeding, as defined by the International Society of Thrombosis and Haemostasis (ISTH) [
14,
15]. Secondary outcomes included evaluation of periprocedural-edoxaban interruption, incidence of acute coronary syndrome (ACS), CRNMB, minor bleeding, all bleeding, all-cause death, CV death, and acute thromboembolic events (stroke, transient ischaemic attack, SEE). All major bleeding, CRNMB, ACS, and acute thromboembolic events were reviewed and unanimously adjudicated by the Steering Committee. Within each group, periprocedural-edoxaban interruption and clinical events were also analysed by renal function category (CrCL ≤ 50 vs. > 50 mL/min). The following parameters were documented at baseline: concomitant medications; HAS-BLED (Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile international normalised ratio, Elderly, Drugs/alcohol concomitantly) score; CHA
2DS
2-VASc (Congestive heart failure, Hypertension, Age ≥ 75 [doubled], Diabetes, Stroke [doubled], Vascular disease, Age 65–74 years, and Sex category [female]) score; renal function; details of edoxaban treatment; diagnostic/therapeutic procedures; and medical history.
Classification of surgeries
Major surgeries were classified by a combination of criteria used in the Dresden registry and PAUSE study: relevant tissue trauma and high bleeding risk; utilisation of general or neuraxial anaesthesia; major intracranial, neuraxial, thoracic, cardiac, vascular, abdominopelvic, or orthopaedic surgery; or other major cancer or reconstructive surgery [
5,
16]. All major surgeries were considered high risk based on European Heart Rhythm Association (EHRA) bleeding risk levels, and nonmajor surgeries were assigned risk levels per EHRA periprocedural bleeding risk criteria.
Statistical analysis
Baseline data are presented as frequencies and/or as summary statistics. P-values for baseline characteristics were calculated using Fisher’s exact test. Pre- and postprocedural edoxaban interruption and clinical outcomes were compared between major vs. nonmajor surgery groups and between renal function subgroups (CrCL ≤ 50 mL/min vs. CrCL > 50 mL/min, calculated using the Cockcroft-Gault equation); data are presented as summary statistics (n, mean, standard deviation [SD]) for numerical parameters and absolute and relative frequencies for duration of interruption between major vs. nonmajor surgeries. Clinical event rates are presented as number of events per 100 surgeries. Clinical outcomes analyses were descriptive and exploratory; no statistical comparisons were made between subgroups. Edoxaban interruption duration data included all patients, both with and without interruption, to avoid selection bias. P-values for duration of edoxaban interruption were calculated using the Wilcoxon test.
Discussion
This subanalysis of the Global EMIT-AF/VTE programme assessed the periprocedural management of edoxaban and the occurrence of bleeding and thromboembolic events in edoxaban-treated patients who underwent major or nonmajor surgeries. To the authors’ knowledge, this analysis is the first to report treatment interruption and clinical events in patients treated with edoxaban undergoing major or nonmajor surgeries. Baseline CHA2DS2-VASc score, HAS-BLED score, and CrCL were similar between patients who underwent major and nonmajor surgeries. While major surgeries had a longer period of edoxaban interruption compared to nonmajor surgeries, low rates of all bleeding, major bleeding, CRNMB, and thromboembolic events were observed in both groups. These results suggest that longer periprocedural edoxaban interruption for patients undergoing major surgeries may help mitigate the bleeding and thromboembolic risk in this group.
The periprocedural management of DOAC treatment focuses on reducing the risk of bleeding without increasing the risk of thromboembolic events. Conversely, prolonged interruption of DOAC therapy may increase the risk of thromboembolism, most importantly ischaemic stroke, whereas an interruption that is too brief may increase the risk of bleeding. In our study, edoxaban therapy was not interrupted for 54 (19.6%) of the major surgeries, which may have been due in part to clinician interpretation of minor hemorrhagic risk and in part to the fact that 3 of these were unplanned (emergency/urgent) surgeries. Major surgeries carry a higher risk of bleeding, and most recommendations suggest longer interruption times compared with low- or minor-bleeding-risk surgeries [
17]. In line with these recommendations, the current study of routine clinical practice found major surgeries had longer edoxaban interruption times when compared with nonmajor surgeries. Notably, there were only 24.5% of major surgeries without any preprocedural interruption; 23.0% of major surgeries had one day or less of postprocedural interruption (interruption on day 0 and/or day 1). This agrees with findings from a subanalysis of the prospective Dresden registry, which reported data on 863 surgical or interventional procedures in DOAC-treated patients receiving predominantly rivaroxaban [
5]. Of the procedures reported, 87 (10.1%) were major surgical procedures, 641 (74.3%) were minor procedures, and 135 (15.6%) were minimal procedures [
5]. Despite having a smaller percentage of major procedures compared with our study, the Dresden study was similar to our analysis in that DOAC use was not interrupted in 22% of patients undergoing surgeries, and the majority of procedures were performed with DOAC interruption [
5].
For patients participating in the Dresden registry, bleeding and cardiovascular event rates were low, similar to this subanalysis [
5]. Notably, the present study stratified patients undergoing major and nonmajor surgeries by pre- and postprocedural interruption, whereas the Dresden study analyzed periprocedural-DOAC interruption in major, minor, and minimal procedures [
5]. The Dresden study also did not stratify patients by time of interruption relative to the procedure, nor did it specify whether DOAC use on the day of a procedure was considered uninterrupted [
5]. Overall, bleeding and cardiovascular events were < 6% for all procedures; patients who underwent minimal and minor procedures vs. those who underwent major procedures had significantly higher rates of any bleeding (2.2% and 4.5% vs. 16.1%), major bleeding (0% and 0.5% vs. 8.0%), CRNMB (1.5% and 3.1% vs. 8.0%), major CV events (0% and 0.8% vs. 4.6%), or CV death (0% and 0.2% vs. 2.3%) at day 30 ± 5 after the procedure [
5].
Additionally, results from this study are comparable with those from the PAUSE study. However, while the PAUSE study used a predefined interruption protocol for surgeries with different bleeding risks, the EMIT-AF/VTE study left the periprocedural-edoxaban management to the attending clinician without any influence of a study protocol [
11,
16]. Therefore, the results of this analysis with low bleeding and ischaemic events in major surgeries suggest that clinicians made the right decision to confine the risk of bleeding in high-risk major surgeries while not increasing the risk of preoperative ischaemic events. Furthermore, compared with the PAUSE trial, this study used a stronger definition for major surgeries that combined the criteria utilised in both the PAUSE and Dresden studies; this improved definition may reduce the risk of selection bias within our study [
5,
11,
16].
In patients with AF, renal dysfunction is a risk factor for both thromboembolic and bleeding events [
18,
19]. Current guidelines recommend a reduced dose of DOACs in patients with renal impairment (CrCL ≤ 50 mL/min) [
20]. In the current study, treatment resumption was protracted in patients with CrCL ≤ 50 mL/min vs. CrCL > 50 mL/min; approximately 70% of patients with CrCL ≤ 50 mL/min resumed edoxaban by day 30 vs. 90% of patients with CrCL > 50 mL/min. With regards to clinical event rates, in a subanalysis of the ENGAGE AF-TIMI 48 trial, patients on a high-dose edoxaban regimen with moderately reduced renal function (CrCL 30–50 mL/min) had numerically lower rates of major bleeding when compared with patients with CrCL > 50 mL/min [
21]. In the current study, the rate of all bleeding events for patients with CrCL ≤ 50 mL/min vs. those with CrCL > 50 mL/min undergoing major surgeries was numerically lower, while the rate of all bleeding events for patients with CrCL ≤ 50 mL/min vs. those with CrCL > 50 mL/min undergoing nonmajor surgeries was numerically higher. This may be due, in part, to a longer periprocedural interruption time in patients with CrCL ≤ 50 mL/min vs. those with CrCL > 50 mL/min undergoing major surgeries, whereas in the nonmajor surgery group, there was no difference in interruption duration between renal function subgroups. These results support the safety of the clinician-driven, edoxaban-management regimen in vulnerable populations, such as patients with renal impairment. However, bleeding event rates (number of events per 100 surgeries) were low overall, regardless of renal function or surgery group (< 6 for all outcomes).
Limitations of this subanalysis include the lack of a DOAC-comparator arm and the lack of formal statistical comparisons between groups for the periprocedural management of edoxaban and clinical outcomes. Additionally, edoxaban management was not standardised, as it was at the discretion of the investigator; however, this enabled patient-individualised treatment. EMIT-AF/VTE is a global programme with data from 326 centres comprising a large number of patients undergoing a wide range of major or nonmajor surgeries in routine clinical practice, including a high percentage of patients (20.8%) with CrCL ≤ 50 mL/min, representing a strength of this analysis. As a large observational study, these data complement randomised controlled trial data, reflecting edoxaban management in current clinical practice without the guidance of a predefined study protocol.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.