The use of non‐vitamin K oral anticoagulants in dialysis patients—A systematic review

Abstract Non‐vitamin K oral anticoagulants (NOACs) are used for prevention of thromboembolic events, but their use in dialysis patients is debatable. This study investigated the available evidence for the use of NOACs in dialysis patients. Online databases were systematically searched for eligible studies including pharmacokinetic (PK) studies, cohort studies, and randomized control trials (RCTs) comparing NOAC with vitamin K antagonist (VKA) or no anticoagulant treatment. Newcastle Ottawa Scale and Cochrane Risk of bias tool were used for quality assessment. Twenty studies were identified (nine PK studies, two RCTs, and nine cohort studies). Most of the studies investigated apixaban or rivaroxaban. In dialysis patients, less accumulation was reported with apixaban and rivaroxaban compared to dabigatran and edoxaban. PK studies indicate that high dose apixaban or rivaroxaban should be avoided. The two RCTs (rivaroxaban/apixaban vs. VKA) were small and underpowered regarding stroke and bleeding outcomes. Most cohort studies found apixaban superior to VKA, whereas comparison of rivaroxaban with VKA yielded conflicting results. Cohort studies comparing apixaban high dose (5 mg) with low dose (2.5 mg) twice daily suggest a lower risk of stroke with high dose but also a higher risk of bleeding with high dose. Apixaban versus no anticoagulation was compared in one cohort study and did not lower the risk of stroke compared with non‐treated regardless of apixaban dosage. Widespread use of NOACs in dialysis patients is limited by adequately sized RCTs. Available evidence suggests a potential for use of apixaban and rivaroxaban in reduced dose.


| INTRODUCTION
Patients with chronic kidney disease (CKD) on dialysis treatment have a high prevalence of atrial fibrillation (AF). 1 Despite the high prevalence of AF, it is unclear whether dialysis patients should be anticoagulated. Non-vitamin K anticoagulants (NOACs) also known as direct oral anticoagulants (DOACs) are often used to prevent stroke in patients with AF 2 and as prophylaxis after thromboembolic events such as deep vein thrombosis and lung embolism. 3 Four different NOACs, dabigatran, 4 rivaroxaban, 5 apixaban, 6 and edoxaban 7 are currently available. Dabigatran is an oral direct thrombin inhibitor, whereas rivaroxaban, edoxaban, and apixaban are direct factor Xa inhibitors. [4][5][6][7] In non-renal patients, NOACs are now mainly used instead of a vitamin K antagonist (VKA) such as warfarin. 8 NOACs are easier to handle, since regular blood sampling for monitoring is not necessary unlike VKAs where the narrow therapeutic range is affected by several food and drug interactions. 9 All NOACs are in varying degrees renally excreted; however, pharmacokinetic (PK) properties are somewhat different in terms of renal clearance, the degree of protein binding, and the potential for removal by dialysis treatment as outlined in Figure 1. 3,8 Accordingly, there is a potential for drug accumulation if NOACs are used in patients with acute or chronic renal failure. 11 The effect and safety of NOACs compared with VKA treatment with warfarin was investigated in four large phase 3 trials in patients with AF: RE-LY, 12 ROCKET-AF, 13 ARISTOTLE, 14 and ENGAGE AF-TIMI 48. 15 However, none of these included end-stage renal disease (ESRD) patients on dialysis treatment. Currently, the European Heart Rhythm Association does not recommend use of NOACs in CKD patients with creatinine clearance (CrCl) ≤ 15 ml/min 16 and dialysis patients with AF are therefore in Europe predominantly treated with VKA. Yet, a large proportion of dialysis patients with AF receive no anticoagulation, due to the lack of available evidence and the known risks with VKA treatment. [16][17][18] Both NOAC and VKA treatment increase the bleeding risk significantly, which is a major hazard to the patients regardless of the drug used, and most cohort studies in dialysis patients have failed to demonstrate a benefit from VKA treatment in terms of stroke prevention. [19][20][21][22][23] Besides, VKA treatment, unlike NOACs, increases the risk of calcifylaxis, a life-threatening syndrome of vascular calcification, typically attributed to warfarin-induced deficiency of vitamin K-dependent calcification inhibitors. 24 Even without anticoagulant therapy, ESRD patients on dialysis have increased risk of bleeding usually attributed to uremia-induced platelet dysfunction and impaired interaction between platelets and the vessel wall. 25 Adding to this are factors such as frailty, comorbidity, high prevalence of antiplatelet drugs, [26][27][28] and the regular use of low-molecular-weight heparin (LMWH) given in order to minimize clotting in the dialysis filter during hemodialysis (HD) treatment. 29 Thus, when comparing bleeding risk in dialysis patients, HD patients have a 1.5-fold higher risk compared with peritoneal dialysis (PD) patients which could be attributed to LMWH. 30 HD patients also have a higher underlying risk of bleeding due to intermittent puncture of the vascular access with needles. 30 A previous bleeding history is another important risk factor, and dialysis patients belonging to this subgroup are reported to be at the highest risk of a new bleeding event. [30][31][32] Consequently, the added benefit of stroke prevention in dialysis patients with AF could be outbalanced by an increased risk of bleeding. 9 This, in conjunction with absence of large-scale randomized controlled trials (RCTs) in ESRD, makes use of NOACs in dialysis patients debatable. This debate has recently intensified after the Food and Drug Administration (FDA) approved use of apixaban in ESRD 33 causing increased utilization of apixaban among dialysis patients in the United States. 34 The aim of this systematic review was to search the existing literature in order to find out whether NOACs can be safely used in dialysis patients and if the PKs of NOACs are significantly different in dialysis patients. An additional aim was to locate intervention studies in dialysis patients comparing NOAC with VKA treatment including upcoming studies.

| Study selection
We screened a total number of 1650 articles ( Figure 2). NO PK studies in PD patients were found except for the ApiDP trial registered at the ClinicalTrials.gov database but without published data and could therefore not be included. An open-label RCT conducted in HD patients (RENAL-AF) was also found via ClinicalTrials.

| Risk of bias in the studies
Newcastle Ottawa Scale (NOS) was used to assess the risk of bias, and thereby, the quality of all eight cohort studies was included. 36 NOS ranks the studies with a star system based on three categories. The categories are (1) selection of study groups, (2) the comparability of the groups, and (3) the ascertainment of either outcome or exposure of interest. A high number of stars awarded to one study indicate that the study has high quality. The two RCTs included were quality assessed using the Cochrane Risk of Bias tool (ROSB). 37 The tool is based on seven categories: random sequence generation, allocation concealment, selective reporting, other sources of bias, blinding of participants and personnel, blinding of outcome assessment, and incomplete outcome data.
Each category is evaluated as having high, low, or unclear risk of bias.
F I G U R E 2 Selection of studies based on PRISMA-guidelines 3 | RESULTS

| Study characteristics
An overview of all selected studies such as study type (PKs study, retrospective cohort study, or RCT), number of patients, dialysis modality, and outcomes (bleeding and stroke) are provided in Tables 1-3. Upcoming studies are provided in Table 4.

| Quality assessment
Assessment of the quality of all cohort studies using this rank system resulted in nine high-quality studies with all studies receiving between seven and nine stars. Cochrane ROB tool resulted in one RCT with low risk of bias and one RCT with high risk of bias (See supporting information).

| Apixaban PK studies in dialysis patients
Four studies investigated apixaban PK in HD patients. [38][39][40][41] Wang et al. 38 showed that administration of single dose of 5 mg apixaban immediately after a HD session increased the area under the curve (AUC) with 36% in HD patients compared with non-renal patients.
Maximum concentration level predialysis and postdialysis only increased negligibly perhaps due to clearance of apixaban via the dialysate (Table 1). Mavrakanas et al. 39 also studied PK of apixaban in HD patients and showed that multiple doses (2.5 mg twice daily) caused a higher accumulated apixaban concentration in steady state compared with the single first day dose. A dose of 2.5 mg twice daily in HD patients was similar to 5 mg apixaban twice daily in healthy subjects. Furthermore, they found that 5 mg apixaban led to supratherapeutic levels in HD patients (Table 1). Van den Bosch et al. 40 compared PKs in HD patients exposed to 2.5 or 5 mg apixaban twice

| Rivaroxaban PK studies in dialysis patients
Two studies investigated PK of rivaroxaban in HD patients. 42

| Dabigatran PK studies in dialysis patients
Two studies investigated PK of dabigatran in HD patients. 44,45 Both studies showed that dabigatran was removed during HD. Khadzhynov et al. 44 showed that a minor redistribution of dabigatran occurred in the plasma after dialysis had ended. Stangier et al. 45 found twofold accumulation of dabigatran in HD subjects, and moreover, the drug was sixfold accumulated in subjects with severe renal impairment compared with healthy subjects (Table 1).

| Edoxaban PK studies in dialysis patients
Only one study was found. Parasrampuria et al. 46 investigated edoxaban PK in an open-label phase 1, randomized crossover study in HD patients. The patients received a single dose of edoxaban 15 mg 2 h prior to HD or in between HD sessions (off-dialysis). HD caused a minor decrease in AUC but did not affect edoxaban levels significantly. HD was deemed ineffective in terms of edoxaban removal (Table 1). Different doses or prolonged treatment was not tested.

| Apixaban stroke and bleeding outcomes in dialysis patients
The only apixaban RCT, RENAL-AF, 47 showed that apixaban 2.5 or 5 mg twice daily resulted in similar rates of stroke and bleeding as warfarin-treated HD patients (Tables 2 and 3). Unfortunately, the study was stopped earlier than planned. Only 154 patients were enrolled (out of the 760 patients originally planned) causing it to be significantly underpowered and difficult to interpret. Four 34,48,49,56 retrospective cohort studies compared apixaban with warfarin in ESRD patients on dialysis. Only three 34,48,56 of them included PD patients. In terms of stroke prevention, Stanton et al. 48 and Reed et al. 49 compared apixaban with warfarin treatment and found similar stroke occurrence ( Table 2). The largest cohort study by Siontis et al. 34 also found similar stroke rates in matched cohorts of dialysis patients with AF treated with either apixaban or warfarin. A lower risk of stroke was found with 5 mg twice daily compared with either 2.5 mg twice daily apixaban or warfarin ( Table 2). Mavrakanas et al. 50 compared apixaban treatment (both 5 mg and 2.5 mg twice daily) with no anticoagulation treatment in dialysis patients with AF. Interestingly, apixaban use was not associated with a lower risk of stroke compared with no anticoagulant therapy. A trend toward fewer ischemic strokes was seen with apixaban, but it was offset by a trend toward more hemorrhagic strokes especially with 5 mg twice daily.
Thus, compared with patients who did not receive any anticoagulants, a significantly higher incidence of fatal or intracranial (major) bleeding was seen in the subgroup of patients treated with standard apixaban dose (5 mg twice daily) but not in patients who received the reduced apixaban dose (2.5 mg twice daily) (Tables 2 and 3). Despite lower risk of bleeding, low dose was also associated with an increased risk of ischemic stroke or myocardial infarction compared with no anticoagulation. Stanton et al. 48 and Sarratt et al. 56 found that apixaban was associated with a similar risk of major bleeding as warfarin treatment (Table 3). Siontis et al. 34 and Reed et al. 49 found lower risk of major bleeding when apixaban was compared with warfarin. Siontis et al. 34 also reported that both 5 and 2.5 mg twice daily were associated with significantly lower risk of major bleeding compared with warfarin (Table 3). Finally, a retrospective cohort study by Miao et al. 51 compared apixaban with rivaroxaban in HD patients and found no significant difference regarding the risk of major bleeding and stroke but did not compare the outcomes with warfarin-treated patients (Tables 2 and 3).

| Rivaroxaban stroke and bleeding outcomes in dialysis patients
The Valkyrie study 54,55 was the only RCT found that compared rivaroxaban with warfarin and assessed bleeding, stroke, and calcification outcomes in HD patients. Progression of vascular calcification was similar in rivaroxaban, rivaroxaban plus vitamin K supplement, and warfarin-treated HD patients. Although not powered for events, the risk of major bleeding was lower with rivaroxaban compared with warfarin-treated patients (  9 showed that the event rate of stroke and major bleeding was higher in rivaroxaban-treated compared with warfarintreated HD patients (Tables 2 and 3).

| Dabigatran stroke and bleeding outcomes in dialysis patients
Only one retrospective cohort study by Chan et al. 9  Abbreviations: BID, (bis in die) = twice a day; CKD, chronic kidney disease; ESRD, end-stage renal disease; HD, hemodialysis; NOAC, non-vitamin K oral anticoagulant, PD, peritoneal dialysis; RCT, randomized control trial, VKA, vitamin K antagonist. a ISTH criteria: composite of any acute, clinically overt bleeding plus at least one of the following: a decrease in hemoglobin of 2 g/dl or more, bleeding that required the transfusion of 2 units or more of blood products (packed red blood cells or fresh frozen plasma), bleeding in at least one critical site (intracranial, intraspinal, intraocular, pericardial, intra-articular, intramuscular with compartment syndrome, or retroperitoneal), or fatal bleeding.
b Authors did not report HR.
c Adjusted odds ratio.
Gov identifier patients had a significantly higher bleeding risk including fatal events compared with warfarin-treated patients. The stroke rate was similar; however, according to the authors, there were too few strokes in the cohort to detect meaningful differences in stroke and arterial embolism between these two groups. (Tables 2 and 3). Table 4 shows upcoming RCTs investigating NOACs and VKA in HD and PD patients, and one finished unpublished PK study in PD patients. All upcoming NOAC-trials use apixaban.

| DISCUSSION
This study reviewed the existing literature on the use of NOACs in dialysis patients. Of the 20 studies included, nine were PK studies.
Only two studies were RCTs, and the majority (nine) were retrospective cohort studies. Six studies were recruiting/ongoing (five RCTs) or without results (one PK study). PK studies in dialysis patients showed that multiple doses of apixaban (2.5 mg twice daily) gave higher exposure at steady state compared with single dose. The RENAL-AF PK substudy 41 showed that 5 mg apixaban twice daily or 2.5 mg apixaban twice daily if patients were ≥80 years or weighed ≤60 kg was similar to apixaban levels in non-ESRD patients. 14,41 However, the results were only extracted from an abstract with limited description of data collection and methodology. More valid data on apixaban PK can be found in the PK study by Mavrakanas et al. 39 despite a lower number of patients. This study showed that treatment for 8 days resulted in significant accumulation with both 2.5 and 5 mg apixaban twice daily.
Especially 5 mg twice daily led to supratherapeutic levels, whereas the reduced dose of 2.5 mg twice daily resulted in drug exposure that was comparable with that of the standard dose in patients with preserved renal function. Interindividual variability could be underestimated due to the low number of patients, but unlike most other PK studies, multiple doses were given over 8 days which is arguably better compared with single dose studies when assessing drug accumulation. Thus, the best apixaban PK study suggests that a dose of 5 mg twice daily should be avoided in dialysis patients regardless of age and weight.
Considering the PK of rivaroxaban, the use of 10 mg in ESRD dialysis patients resulted in plasma concentration similar to 20 mg rivaroxaban in healthy subjects from the ROCKET-AF study. No accumulation was seen after multiple doses of 10 mg daily. 42 Dabigatran was removed by dialysis; however, minor redistribution was seen after HD was ended, which indicates accumulation of dabigatran which therefore should be avoided in ESRD patients. 44 Edoxaban is not removed by dialysis but whether accumulation occurs over time with prolonged treatment has not been tested. 46

| Limitations
The biggest limitation of this review is the lack of adequately sized

| CONCLUSION
This systematic review showed that of all four NOACs apixaban and rivaroxaban in low dose appear to be the best candidates for safe use in dialysis patients. Firstly, the PK properties of NOACs vary and apixaban and rivaroxaban were associated with less drug accumulation compared with edoxaban and dabigatran. Secondly, apixaban and rivaroxaban were evaluated in the only two RCTs conducted so far with similar stroke rates as warfarin treated and without significantly more bleeding events. Yet, caution is advised since both studies were underpowered regarding stroke and bleeding outcomes. Cohort studies comparing apixaban high dose (5 mg) with low dose (2.5 mg) suggest a lower risk of stroke with high dose but also a higher risk of bleeding with high dose apixaban.
Apixaban versus no anticoagulation was compared in one cohort study and did not lower the risk of stroke compared with nontreated regardless of apixaban dosage. Cohort studies comparing rivaroxaban with VKA yielded conflicting results. Currently, widespread use of NOACs in dialysis patients is limited by adequately sized RCTs, but several apixaban trials are underway. Clinicians are advised to individualize treatment and carefully weigh the risk of stroke versus bleeding especially in patients with a prior history of bleeding before prescribing NOACs to dialysis patients.