Anticoagulation use and the risk of stroke and major bleeding in patients on hemodialysis: From the VIVALDI, a population‐based prospective cohort study

Evidence supporting the use of anticoagulation for the prevention of stroke and thromboembolism in patients with kidney failure on hemodialysis (HD) and atrial fibrillation (AF) is limited. We prospectively assessed the incidences of stroke and major bleeding, as well as anticoagulation strategies in patients on HD with AF.


| INTRODUC TI ON
Patients with kidney failure on maintenance hemodialysis (HD) are at high risk of thromboembolism, including ischemic stroke and systemic embolism, as well as of bleeding, 1,2 especially among those with atrial fibrillation (AF). 3 Presence of several comorbid conditions and advanced age in the population of HD patients 4 further increase the risks of both thromboembolism 5 and bleeding. 6 Clinicians face the challenge of balancing the competing risks of thromboembolism and bleeding complications when considering anticoagulation and antiplatelet strategies. Most patients on HD satisfy criteria for anticoagulation spelled out in guidelines 7,8 ; however, most patients are also at high bleeding risk per accepted scoring algorithms. 9 Data on population-specific risk factors and risk assessment models for AF patients with end-stage renal disease (ESRD) on HD are scarce.
In light of the fact that until recently there were no dedicated randomized trials for stroke prevention in HD patients, and HD patients had been excluded from all prior stroke prevention in AF trials, it is not surprising that statements in clinical practice guidelines are rather reserved when it comes to anticoagulation in patients on HD with AF. 7,8 In nonrandomized studies, there has been no agreement whether anticoagulation with vitamin K antagonists (VKAs) is of benefit compared with no treatment. [10][11][12][13] We aimed to prospectively assess the incidence and risk factors of thromboembolic events, major bleeding, and cardiovascular death in patients undergoing maintenance HD in a large European metropolitan area. We further aimed to investigate these outcomes in HD patients with AF on treatment with anticoagulation agents compared with no treatment and to assess whether the benefits of anticoagulation outweighed the risks in patients with AF and ESRD on HD.

| Study design and procedure
The Vienna Investigation of Atrial Fibrillation and Thromboembolism in Hemodialysis patients (VIVALDI) is a prospective populationbased cohort study, which was initiated to investigate the incidence of AF, thromboembolism, and bleeding in patients with kidney failure on maintenance HD. The VIVALDI study was approved by the local ethics committees and was conducted in accordance with the declaration of Helsinki and its later amendments.

Conclusions:
Although the nonrandomized nature of the study is prone to bias, anticoagulation with VKAs was not associated with decreased thromboembolic risk, but rather with increased risk of major bleeding and may be net harmful to patients with AF on HD.

K E Y W O R D S
anticoagulants, atrial fibrillation, chronic kidney failure, ischemic stroke, renal replacement therapy Essentials • There is limited evidence for the use of anticoagulants in patients on hemodialysis.
• An urban cohort of hemodialysis patients (N = 625) with up to 45 months of prospective follow-up.
• The event-rate for thromboembolic outcomes in AF patients (N = 238) was 48 per 1000 patient-years.
• Patients with AF did not benefit from anticoagulation but had increased risk of bleeding.
interviews with each patient, reviewed medical records at the participating dialysis centers, and verified findings with medical documentation and in consensus with the treating nephrologists.
Patients were prospectively followed for a maximum of 1350 days (45 months), and the occurrence of outcomes was assessed with one interim and one final personal interview, review of medical documentation at the dialysis centers, verification of events by the treating nephrologists, and verification with the Austrian death registry.
One male patient (0.16%) with AF was lost to follow-up.

| Outcomes
The prespecified primary thromboembolic outcome of the study was a composite of ischemic stroke, transient ischemic attack (TIA), and systemic embolism. Thromboembolic outcome was based on objective evidence in imaging records and after ruling out a readily identifiable cause such as a tumor, seizure, or HD-related transient neurological symptoms. The primary bleeding outcome was major bleeding as defined by the International Society on Thrombosis and Haemostasis. 14 Secondary outcomes were (1) the composite net clinical benefit or harm outcome of ischemic stroke, major bleeding, and cardiovascular (CV) death; (2) the composite major adverse cardiovascular outcome (3P-MACE) comprising CV death, myocardial infarction, and ischemic stroke; (3) cardiovascular death; and (4) all-cause mortality.
All outcomes were adjudicated by independent experts in neurology, cardiology, vascular disease, and hematology upon chart review and on basis of imaging evidence or autopsy findings. Autopsies were performed in only a few cases at the discretion of the treating physicians and not as a part of the study outcome verification.

| Definition of anticoagulation treatment
Anticoagulation treatment for stroke prevention in AF was based on chart and prescription review at the dialysis centers, and personal verification with the patient. Anticoagulation treatment was reevaluated at the interim data capture after 1 year and at the final data capture at the end of the study. Because anticoagulation treatment changed during the course of the observation time in some patients, anticoagulation treatment was analyzed in an intention-totreat approach. Intention-to-treat was considered if patients with AF at baseline had active anticoagulation treatment and maintained anticoagulation for at least 1 month, or if one of the primary or secondary outcomes occurred during the first month of observation while the patient was on anticoagulation. In patients with de novo AF, intention-to-treat was considered if anticoagulation was begun upon AF diagnosis and treatment was continued for at least 1 month.
Phenprocoumon was the VKA agent exclusively used in this cohort, and the target international normalized ratio for stroke prevention in AF was set at 2.0 to 3.0. Time in therapeutic international normalized ratio range (TTR) was calculated with the linear interpolation/Rosendaal method. 15 The low molecular weight heparin

| Statistical analysis
Patient characteristics at baseline are reported as counts and proportions or median values with the interquartile range (IQR), where appropriate. Differences between anticoagulation treatment groups were assessed with the chi-squared test for categorical variables and Mann-Whitney U test for continuous variables.
We recorded the incidence of primary and secondary outcomes in the total cohort and the AF group and calculated the incidence and 30-day case fatality rates.
To analyze risk factors for the primary outcomes in the full cohort and the AF group, we computed the univariable subdistribution hazard ratios (SHRs) and 95% CI, using competing risk regression according to

| RE SULTS
Baseline patient characteristics of the enrolled cohort (N = 625) are provided in Table 1 Figure 1 shows the cumulative probabilities of the composite outcome of thromboembolism, major bleeding, 3P-MACE, and net clinical benefit in AF patients with separate lines according to anticoagulation treatment.

| Secondary outcomes in the total cohort
Compared with non-AF patients without anticoagulation, AF patients treated with VKA also had increased risks for occurrence of primary and secondary outcomes (Table 5). Surprisingly, AF patients without anticoagulation did not have statistically increased risks for thromboembolic outcomes, major bleeding, CV death, and 3P-MACE compared with non-AF non-anticoagulated patients ( Table 5).

| DISCUSS ION
In this prospective population-based cohort study of HD patients encompassing the majority of patients undergoing HD in a large metropolitan area, we found an event rate of 29 per 1000 patientyears for the composite outcome of stroke, TIA, and systemic embolism, and 68 per 1000 patient-years for major bleeding. In contrast to most previous studies that used insurance claims to ascertain these events, all events were independently adjudicated by expert clinicians using imaging evidence. Patients with AF had higher event rates of all primary and secondary outcomes compared with the full cohort. AF patients receiving anticoagulation with VKA or LMWH did not benefit in terms of reduced thromboembolic risk or better survival compared with AF patients without anticoagulation, but their risk of bleeding was significantly and sizably increased, resulting in a two-fold increased risk for the net clinical harm outcome.
Despite being diligently adjudicated, the incidences identified were on the higher range of previous reports, where the incidence of ischemic stroke ranged from 10 to 38 per 1000 patient-years, [16][17][18][19] and 20 to 69 per 1000 patient-years for HD patients with AF. 16,18,[20][21][22][23][24][25] The inclusion of TIA into the primary thromboembolic outcome yielded only seven of the 40 events and was relatively low most probably because of nonadjudication because of other HD-related reasons mimicking TIA symptoms. Our results confirm that AF is not only exceedingly common in HD patients, 4,26 but a significant contributor to the risk of stroke and mortality in this population. 27 One clinical challenge when considering anticoagulation in patients with ESRD on HD lies in the evaluation of the risk of thromboembolism against the risk of bleeding in the absence of quality evidence from randomized trials in this specific population. However, the risk for other cardiovascular events such as myocardial infarction is also elevated because of high prevalence of cardiovascular risk factors including diabetes, hypertension, smoking, and obesity. Risk assessment, therefore, must involve multiple competing clinical outcomes and the treatment options that provide a net clinical benefit. TA B L E 5 Frequency of outcomes in AF patients and risk of outcomes associated with anticoagulation compared to AF patients with no anticoagulation and non-AF patients with no anticoagulation adjusted for CHA 2 DS 2 -VASc score and antiplatelet comedication

| CON CLUS ION
From this carefully conducted prospective study of HD patients in a large metropolitan area, we found that AF was a heavy burden on this population, associated with high risks of stroke, bleeding complications, and mortality. Our findings support the notion that HD patients with AF may not benefit from oral anticoagulation with VKA in terms of stroke prevention and may be exposed to undue harm from increased risk of major bleeding events compared with patients with AF not receiving anticoagulation. Thus, it is quite possible that anticoagulation treatment may result in net clinical harm. There is an unmet clinical need for effective and safe treatment options in this special patient population, and results of randomized controlled studies using VKAs, or other novel anticoagulants, are awaited with great interest.

ACK N OWLED G M ENTS
We acknowledge the members of the adjudication committee: Julia Riedl, Johannes Thaler, Christoph Kopp, Thomas Gremmel, and Fritz Leutmezer.

CO N FLI C T O F I NTE R E S T S
Dr. Winkelmayer has served as scientific advisor to Akebia,