PDF
Citation
Print
INTRODUCTION
Stroke prevention is one of the pillars of management of patients with atrial fibrillation (AF), and guidelines advocate for the utilization of on-label dose direct oral anticoagulants (DOACs).1)2)3)4) The Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation-Thrombolysis in Myocardial Infarction 48 (ENGAGE AF-TIMI 48) trial showed high-dose edoxaban regimen (HDER) has similar efficacy to warfarin in preventing ischemic stroke in AF patients, while low-dose edoxaban regimen (LDER) increased ischaemic stroke risk but reduced major bleeding risk by 45% compared to warfarin.5) In light of these findings, HDER has been established as the standard dosage regimen for stroke prevention in patients with AF.
However, lifelong anticoagulation therapy poses bleeding risks. A sub-analysis of ENGAGE-AF trial showed similar net clinical outcomes of HDER and LDER as a result of lower bleeding risk with higher residual stroke risk in LDER, suggesting LDER could be useful for patients with high bleeding risk.6) The Edoxaban Low-Dose for Elder Care Atrial Fibrillation Patients (ELDERCARE-AF) trial demonstrated that very low-dose edoxaban (15 mg daily) reduces stroke risk without significant bleeding, offering potential for untreated, high-risk AF patients due to concerns over bleeding risk.7)8) Despite South Korea’s approval of 15 mg daily edoxaban, its clinical efficacy in real-world practice remains unassessed.
In this study, our aim was to assess the described the practice patterns, effectiveness, and safety of very low-dose edoxaban (15 mg once daily).
METHODS
An extended description of study methods is presented in Supplementary Data 1.
Ethical statement
The protocols were approved by the Institutional Review Boards (IRBs) of each canter and were conducted according to the principles of the Declaration of Helsinki (2013). Because the data were obtained retrospectively, the informed consent requirement was waived. The data were anonymized; thus, the study was exempt from the Institutional Review Board (IRB) review of Seoul National University Hospital (SNUH) (IRB No. H-2211-001-1372) and Asan Medical center (AMC) (IRB No. 2023-0058).
Study population
This was a retrospective observational study. Among the patients prescribed with DOAC for stroke prevention due to AF, those prescribed with edoxaban 15 mg between February 2016 and February 2023 were identified from the 2 tertiary hospitals (SNUH and AMC). The patient enrolment process is displayed in Figure 1 as a flowchart.
Covariates
To analyse the clinical characteristics of patients prescribed with edoxaban 15 mg once daily, the baseline characteristics of the patients were collected. The age, sex, comorbidities, presence of oral anticoagulation and the specific type and dosage of oral anticoagulant (OAC). The use of non-steroidal anti-inflammatory drugs (NSAIDs), dronedarone, and antiplatelet agents were recorded for concomitant medications.
The CHA2DS2-VASc score and modified HAS-BLED score were calculated using baseline comorbidities and medical history.9)10)11) The general health examination information consisted of weight. Laboratory results included haemoglobin and estimated glomerular filtration rate (eGFR).
Additionally, we assessed the eligibility of our population to the ELDERCARE-AF trial. Patients who met the criteria for very-low dose edoxaban prescription, as outlined in the ELDERCARE-AF trial, were additionally classified as the ELDERCARE-AF like population. Those who did not qualify for the ELDERCARE-AF like population were further evaluated for the appropriate dosage (30 mg or 60 mg) in accordance with the ENGAGE AF-TIMI 48 trial.
Study outcomes
The occurrence of thromboembolic events and bleeding events during follow-up period were collected. The thromboembolic events included stroke, transient ischemic attack (TIA), and systemic embolism (SE). The bleeding events included major bleeding, clinically relevant non-major bleeding, and minor bleeding. During the follow-up, the persistency of edoxaban 15 mg taken once daily was evaluated, as far as it could be assessed from the medical records.
Statistical analysis
For the description of continuous variables, the values are presented by means ± standard deviation or medians (interquartile range [IQR]), as appropriate. The categorical variables are presented as numbers (percentages). The annual risk of thromboembolic and bleeding events was derived from the observed event rate. To ascertain relative risk reductions, this observed annual risk was compared with the predicted average risk scores computed for the study population based on the CHA2DS2-VASc and HAS-BLED scores, serving as benchmarks for efficacy and safety evaluations, respectively.9)10)11)12) We calculated the expected risk by first assessing the risk for each individual patient and then averaging the values across the entire cohort.
Statistical significance was set at p<0.05. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA).
RESULTS
A total of 674 patients (mean age 78.3±9.1 years; 49.3% women) were finally included. During the study period, the number of very low-dose edoxaban new prescriptions increased over time with landmark events (Figure 2). Most of the edoxaban 15 mg once daily prescription was from the cardiology department (64.1%) followed by the neurology department (13.9%). Baseline characteristics are presented in Table 1. The 52.4% were persistent AF and mean CHA2DS2-VASc score of 3.9±1.6 and a mean modified HAS-BLED score of 2.0±1.1. Of total, 49.7% of the patients were aged over 80 years old, while 50.4% were below 60 kg (8.9% weighed below 45 kg), 40.2% had chronic kidney disease, and 5.9% had an eGFR of 15 to 30 mL/min/1.73 m2. Regarding comorbidities, 28.8% had a heart failure, 3.7% underwent transcatheter aortic valve replacement, 3.6% had bioprosthetic aortic valve replacement or mitral valve replacement, 3.7% had a recent percutaneous coronary intervention, and 19.7% had prior history of major bleeding. In addition, 15.3% were diagnosed with active cancer and more than half of the patients were accompanied by anaemia.
Figure 2
The temporal trend of edoxaban 15 mg prescription. The prescription of very low-dose edoxaban increased over time in Korea, with a substantial rise observed after landmark events.
AF = atrial fibrillation; ELDERCARE-AF = Edoxaban Low-Dose for Elder Care Atrial Fibrillation Patients; KFDA = Korean Food and Drug Administration; KHRS = Korean Heart Rhythm Society; RCT = randomized clinical trial.
Table 1
Baseline characteristics
Values are presented as number (%) or mean ± standard deviation.
AF = atrial fibrillation; AVR = aortic valve replacement; CABG = coronary artery bypass graft surgery; CAD = coronary artery disease; CKD = chronic kidney disease; CKD-EPI = chronic kidney disease epidemiology collaboration; ESRD = end stage renal disease; eGFR = estimated glomerular filtration rate; HD = haemodialysis; MVR = mitral valve replacement; PCI = percutaneous coronary intervention; TAVI = transcatheter aortic valve replacement.
*Prothrombin time international normalized ratio not included, and alcoholics considered as ≥8 alcohol drinks/weeks.
Among the subjects over the age of 80, 11.6% weighed below 45 kg, 8.1% had an eGFR between 15 and 30 mL/min/1.73 m2, 9.6% were using NSAIDs concomitantly, 22.7% were using antiplatelet drugs concomitantly, and 20.9% had a history of major bleeding.
Prior antithrombotic therapy before very low-dose edoxaban
The prior antithrombotic medication before edoxaban 15 mg are described in Table 2. Prior to the prescription of edoxaban 15 mg, 44.8% of the patients were not prescribed OACs. Among those who were prescribed, 12% received a standard dose DOAC prescription and 42.4% received a low-dose DOAC. Of the low-dose DOAC prescriptions, edoxaban 30 mg once daily was the most common (26.2%).
Table 2
Prior antithrombotic medication prescription history and concomitant medications including NSAID, p-glycoprotein inhibitor, and antiplatelet
Values are presented as number (%).
DOAC = direct oral anticoagulant; NSAID = non-steroidal anti-inflammatory drugs.
Among patients who have experienced prior bleeding events, low-dose DOAC was the most common medication (56.4% in major bleeding and 50.6% in minor bleeding) followed by no antithrombotic treatment (21.1% in major bleeding and 19.2% in minor bleeding) and antiplatelet agents (20.3% in major bleeding and 15.1% in minor bleeding) (Table 3). Among the 133 patients with major bleeding and the 33 with minor bleeding, 57.1% and 67% respectively were on anticoagulation therapy at the time of their bleeding events.
Table 3
The prior antithrombotic medication of patients with prior bleeding history
Concomitant medications
Concomitant medications are summarised in Table 2. NSAIDs were used by 12.3% of the population and antiplatelet agents were used by 26.7% of the population. The most common concomitant medication among the antiplatelet agents was aspirin single therapy (22.5%), followed by clopidogrel only (3.4%). Dronedarone was used in 1.8% of the population.
Persistency of edoxaban 15 mg prescription during follow-up
Among the 674 patients, 187 (27.7%) experienced change of prescription during follow-up (Supplementary Table 1). Dose increases of edoxaban to 30 mg or 60 mg were the most common cause (n=83, 12.3%) with clinical status improvement being the main reason. Overall, 37 patients discontinued OAC because of various bleeding events (9: gastro-intestinal bleeding, 4: genitourinary bleeding, 3: blood-tinged sputum, 3: epistaxis, 1: intracranial bleeding, 1: oral bleeding), thrombocytopenia (n=3), progression of anaemia (n=2), or worsening kidney function (n=1).
Observed rates of thromboembolic events and bleeding events
During a median follow-up duration of 8 (IQR 3–16) months, overall thromboembolic and bleeding events occurred in 16 (2.3%) and 88 (13.1%) patients, respectively (Supplementary Table 2). Specifically, 14 (2.1%) patients experienced a stroke, 1 (0.1%) had TIA, 1 (0.1%) had pulmonary thromboembolism, 26 (5.2%) patients experienced major bleeding, 32 (4.7%) clinically relevant non-major bleeding, and 30 (4.4%) had minor bleeding. The overall mortality rate was 5.2% (1.2% cardiac, 4.0% noncardiac death).
The expected risk based on CHA2DS2-VASc score and modified HAS-BLED score is described in Figure 3. In detail, the expected annual ischemic stroke risk based on CHA2DS2-VASc score was 5.2% and 7.3% for composite of ischemic stroke/TIA/SE events. The actual observed annual ischemic stroke rate was 2.0%, resulting in 61% relative risk reduction and annual observed rate of ischemic stroke/TIA/SE composite was 2.5% with relative risk reduction of 68% (p value <0.001 for both). Based on modified HAS-BLED score, the expected major bleeding risk was 2.2% compared to actual observed rate of 3.2%. The relative risk was 49% increased (p value=0.144).
Figure 3
Expected risk versus actual rate. Based on the CHA2DS2-VASc score, the RRR for ischemic stroke was 61%, and 68% for stroke/TIA/SE. According to the modified HAS-BLED score, there was a 49% increase in the relative risk of major bleeding.
RRI = relative risk increase; RRR = relative risk reduction; SE = systemic embolism; TIA = transient ischemic attack.
*Modified HAS-BLED score used.
Subgroup analysis on ELDERCARE-AF like population
Out of 335 patients aged ≥80 years, a total of 180 patients (53.7%) were eligible for the ELDERCARE-AF trial (forming the ELDERCARE-AF like population) (Supplementary Table 3). The most common additional criterion was ‘current use of an antiplatelet drug’ (n=93; 27.8%), followed by ‘history of bleeding in a critical area or organ or gastrointestinal bleeding’ (n=70; 20.9%), and low body weight (n=36; 10.7%). The mean CHA2DS2-VASc score of the ELDERCARE-AF like population was 4.8±1.5, and the mean modified HAS-BLED score was 2.6±1.1. Among the 155 patients (46.2%) older than 80 years who did not meet any criteria for the ELDERCARE-AF like population, 115 met the prescription criteria for 30 mg of edoxaban, and 40 met the criteria for 60 mg from the ENGAGE AF-TIMI 48 trial.
Among the 115 patients with edoxaban 30 mg criteria, 77 (23.0%) had low bodyweight (≤60 kg), 74 (22.1%) had low eGFR (15 to 60 mL/min/1.73 m2), and 9 (2.7%) had concomitant use of p-glycoprotein inhibitor (dronedarone or verapamil).
During a median follow-up duration of 7 (IQR 2–15) months, overall thromboembolic and bleeding events occurred in 8 (4.4%) and 7 (3.8%) patients, respectively. All thromboembolic events were strokes. The expected risks, based on the CHA2DS2-VASc score and the modified HAS-BLED score, are described in Supplementary Figure 1. The observed risk of ischemic stroke was 4.8%, which is a 28% relative risk reduction compared to the expected risk of 6.7% (p value=0.271). For the composite of ischemic stroke/TIA/SE, the observed risk was 4.8%, resulting in a 49% relative risk reduction from the expected risk of 9.4% (p value=0.007). In the case of major bleeding events, the observed annual risk was 4.3%, indicating a 79% relative risk increase from the expected risk of 2.4% based on the HAS-BLED score (p value=0.272).
DISCUSSION
In this analysis, we examined the adoption of very low-dose edoxaban in real-world clinical practice, following the publication of efficacy and safety data and regional guideline updates, encompassing a substantial number of 674 patients. The main findings of our study are summarized in the Figure 4. Our findings indicated that (1) there was a discernible increase in the prescription of very low-dose edoxaban, (2) 26.7% of patients receiving the prescription met ELDERCARE-AF eligibility criteria, (3) regardless of ELDERCARE-AF eligibility, patients prescribed very low-dose edoxaban tended to be older, leaner, had reduced renal function, multiple comorbidities, a history of bleeding, or were on medications increasing their bleeding risk, and were often anaemic, (4) patients with a history of bleeding experienced further episodes, despite being prescribed low-dose DOACs, (5) the persistency of very low-dose edoxaban was 72.3% with the main cause of medication change being the improvement of their general condition, and (6) during a median follow-up of 8 months (IQR 3–16), we observed thromboembolic and bleeding event rates of 2.3% and 3.2%, respectively. The thromboembolic risk was lower than expected based on CHA2DS2-VASc scores, whereas the bleeding risk was consistent with expectations based on modified HAS-BLED scores.
Figure 4
In Korean practice, very low-dose edoxaban increased over time with low body weight, anaemia, chronic kidney disease, active cancer, concomitant antiplatelet agents, and prior major bleeding as the main cause. The use of very low-dose edoxaban reduced the expected risk of thromboembolic events.
AF = atrial fibrillation; AMC = Asan Medical Center; ELDERCARE-AF = Edoxaban Low-Dose for Elder Care Atrial Fibrillation Patients; KFDA = Korean Food and Drug Administration; KHRS = Korean Heart Rhythm Society; SE = systemic embolism; SNUH = Seoul National University Hospital; TIA = transient ischemic attack.
For adequate stroke prevention in patients with AF, it is recommended that DOACs be used at the labelled dose for each clinical trial.13)14) However, off-label underdosed DOAC prescriptions are common in real-world practice.15)16)17) While there is variability based on the type of DOAC, observational studies have indicated that off-label underdosing does not mitigate bleeding risk but rather elevates the likelihood of stroke compared to on-label dosing.18)19) An inherent limitation of observational studies lies in the preselection of patients for off-label underdosing, who are already predisposed to a high bleeding risk,16) thereby making it challenging to precisely demonstrate the efficacy and safety of off-label underdosing of DOAC.
The ENGAGE-AF TIMI 48 trial is the only source of insights into the efficacy and safety of LDER in a randomized clinical trial context, potentially representing off-label underdosing.5) In the ENGAGE-AF trial, outcomes revealed that the HDER exhibited non-statistically significant difference in the risk of ischemic stroke, while the LDER demonstrated a 43% increased risk of ischemic stroke compared to warfarin.5) Given the results of these studies, the HDER has been recognized as the benchmark dosage for mitigating stroke risk in patients diagnosed with AF. However, it should be notable that HDER has a similar risk of major bleeding to warfarin, while LDER reduces this risk by 45%.5)
Compared to a common comparator, warfarin, LDER was less effective but significantly safer, reducing major bleeding, fatal bleeding, intracranial haemorrhage, and gastrointestinal bleeding by 45%, 49%, 60%, and 30%, respectively, compared to warfarin, while HDER significantly reduced only intracranial haemorrhage among bleeding outcomes compared to warfarin.5) A post hoc analysis of these results comparing HDER and LDER directly, revealed that LDER was associated with a 31% higher risk of stroke/SE and a 36% lower risk of major bleeding than HDER.6) LDER also showed a 10% lower risk for the primary net endpoint, a composite of stroke/SE, major bleeding, and death.6) No significant difference was found between HDER and LDER for secondary and tertiary net endpoints.6) These findings provide valuable insights for clinicians considering LDER for patients with a high risk of bleeding. Similarly, a post hoc analysis of the Atrial Fibrillation and Ischemic Events with Rivaroxaban in Patients with Stable Coronary Artery Disease (AFIRE) trial reported that off-label underdose rivaroxaban use reduced the risk of major bleeding by 62% without increasing the risk of ischemic stroke and thromboembolic events.20)
The ELDERCARE-AF trial systematically examined the efficacy and safety profile of very low-dose edoxaban in comparison to a placebo group among elderly patients with AF who were at a high risk of bleeding.7) Compared to placebo, very low-dose edoxaban was associated with a lower risk of stroke/SE by 66% (p<0.001) and tended to be associated with a higher risk of major bleeding without statistical significance (hazard ratio, 1.87; 95% confidence interval, 0.90–3.89, p=0.09). The main findings of the ELDERCARE-AF trial were consistent in various subgroup by age strata (80–84 years, 85–89 years, and ≥90 years), body weight (>45 kg, ≤45 kg), and kidney function (creatinine clearance 15 to <30 mL/min, 30 to 50 mL/min, and >50 mL/min).21)22)23) Additional evaluation performed on ELDERCARE-AF like population in this study showed broadly similar results.
In this study, the majority of observed bleeding events originated from the gastrointestinal or genitourinary systems. Out of 26 major bleeding events, none resulted in fatality and most of them were predominantly manageable. In the ELDERCARE-AF like population, 6 out of 8 major bleeding events were traced back to the gastrointestinal tract, one subdural haemorrhage was due to head trauma, and one was due to anaemia. Our data suggests that these bleeding events are generally reversible and manageable. Given the irreversible consequences of stroke, we suggest that patients with frailty should also be prescribed with oral anticoagulation, even if administered at very low doses, rather than leaving such high-risk patients untreated.
While results reported from the ENGAGE-AF TIMI 48 trial and ELDERCARE-AF trial have increased our understanding of the clinical outcomes of very low-dose edoxaban, there is still limited reported on the penetration of very low-dose edoxaban prescribing in clinical practice and the actual patterns of very low-dose edoxaban prescribing by clinician. Within our cohort, the consideration of very low-dose edoxaban prescriptions extended beyond patients who met all ELDERCARE-AF eligibility criteria, encompassing individuals who were generally frail and exhibited high risk for bleeding. Clinicians might show a preference for prescribing off-label underdosing due to concerns about bleeding while still aiming to prevent stroke. It will be intriguing to observe how this prescribing pattern influences patient outcomes. A recent study reported a comparison of clinical outcomes between very low-dose DOACs (edoxaban 15 mg once daily, dabigatran 110 mg or 150 mg once daily, apixaban 2.5 mg once daily, or rivaroxaban 10 mg once daily) and regular-dose DOACs (edoxaban 60/30 mg once daily or other labelling-dosage DOACs) in patients aged 80 years or older with a CHADS2 score of 2 or greater, utilizing a Taiwan nationwide population-based cohort.24) Despite not exclusively focusing on very low-dose edoxaban, this observational study revealed that very low-dose DOACs demonstrated comparable risks of ischemic stroke/SE and major bleeding compared to regular-dose DOACs. However, very low-dose DOACs were associated with higher risk of major adverse limb events, venous thrombosis, and all-cause death compared to regular-dose DOACs.
Nevertheless, assessing the effectiveness and safety of off-label underdosing in observational studies proves challenging due to the complexity of analysing an underdose while accounting for all confounding factors influencing the physician’s decision to opt for underdosing.
Furthermore, oral anticoagulation is but one component of the current holistic approach to risk stratification, including attention to risk factors and comorbidities.25) Multimorbidity, polypharmacy and frailty are common in elderly patients especially in association with AF, with major implications for outcomes, including bleeding.26)27)28) Such a holistic approach to AF management has been associated with better outcomes,29) leading to its recommendation in guidelines.3)
This study has several limitations. First, data collection was limited to Asia, specifically South Korea, with a small sample size and a short to mid-term follow-up period, conducted in a retrospective manner. Therefore, caution is needed when generalizing our findings to non-Asian populations. Second, only patients prescribed with edoxaban 15 mg once daily were evaluated. This limited the study’s ability to assess the proportion of patients who were prescribed with edoxaban 15 mg in the overall anticoagulation treatment population. It also prevented a outcomes comparison between the group of patients who received the dose they were originally prescribed and those prescribed with edoxaban 15 mg. Third, similarly, a single-arm study, it was not possible to compare outcomes with other edoxaban doses or with a no OAC group. Fourth, in more than half of the cases, the cause of the stroke could not be determined due to the patients’ poor medical conditions. Lastly, although our study, similar to the ELDERCARE-AF study,7) did not include data on proton pump inhibitors, the use of proton pump inhibitors or H2 blockers may have influenced bleeding event rates by reducing the risk of gastrointestinal bleeding.
Despite these limitations, the significance of this study lies in its attempt to examine patient characteristics and short-term follow-up outcomes as very low-dose edoxaban begins to penetrate clinical practice. In reality, there is still a lack of abundant clinical data on edoxaban 15 mg. More large-scale, long-term follow-up data will be needed in the future.
In Korean practice, very low dose edoxaban increased over time in patients with low body weight, anaemia, chronic kidney disease, active cancer, concomitant antiplatelet agents, and prior major bleeding, who were previously left untreated. Such use reduced the expected risk of thromboembolic events in these high-risk patients.
XML Download