Both atrial fibrillation (AF) and chronic kidney disease (CKD) are highly prevalent, especially with increasing age and associated comorbidities, such as hypertension, diabetes, heart failure, and vascular disease. The relationship between both AF and CKD seems to be bidirectional: CKD predisposes to AF while onset of AF seems to lead to progression of CKD. Importantly, the concurrence of AF and CKD leads to an increased risk of thromboembolic complications including stroke, systemic thromboembolism, and myocardial infarction.1 Paradoxically however, CKD is itself a risk factor for bleeding.2

Stroke prevention is the cornerstone of AF management and requires use of oral anticoagulants (OAC), which are either vitamin K antagonists (VKA) (eg, warfarin), or the non-VKA oral anticoagulants (NOACs).3 While NOACs have been shown to be effective in mild-to-moderate renal dysfunction, there are a paucity of data regarding NOACs in severe and end-stage renal dysfunction (ESRD).

This review first discusses the evidence for NOACs in CKD. Second, we summarize the current knowledge regarding the efficacy and safety of NOACs to prevent AF-related stroke and systemic embolism in severe and end-stage renal disease.

Chronic kidney disease is classified into stages 1 to 5 by the Kidney Disease Improving Global Outcomes based glomerular filtration rate (GFR) or albuminuria present for > 3 months. The GFR may be estimated through use of the Modification of Diet in Renal Disease (MDRD) or Chronic Kidney Disease-Epidemiology Collaboration group (CKD-EPI) equations4 (Table 1). Severe renal dysfunction implies GFR less than 30mL/min/1.73m2.5 There is variability in the definition of ESRD in clinical trials; however a suggested framework to diagnose ESRD is symptomatic uremia necessitating chronic (> 30 days) renal replacement therapy.6 Renal replacement therapy may be delivered via extracorporeal (hemodialysis) or paracorporeal (peritoneal dialysis) modalities.7

The prevalence of AF in ESRD ranges from 7% to 27% in different studies, which is 10- to 20-fold higher than in the general population.8 For example, in the Chronic Renal Insufficiency Cohort (CRIC) prospective study of 3267 patients with mild-to-moderate CKD (mean GFR 43.6 ± 13.4mL/min/1.73 m2) AF was present in 18% participants, suggesting that the processes underlying AF onset may occur early in the course of CKD.9

Atrial fibrillation may also contribute to CKD progression. A subgroup analysis of the CRIC study found a higher rate of progression to ESRD in CKD patients with incident AF (11.8/100 person-years) compared with CKD patients that did not develop AF (3.4/100 person-years) over a mean follow-up of 5.9 years.10 Thus, the relationship between AF and renal dysfunction is bidirectional. Of note, the mortality associated with incident AF has been shown to be higher in patients with CKD (survival 63.4%-68.3% at 12 months) than without CKD (79.3% survival at 12 months).11

For this review, a search of published studies was performed using bibliographic databases (PubMed, Medline, Scopus, Cochrane database) and scanning reference lists from published articles. Search terms used were: “NOAC”, “chronic kidney disease”, “atrial fibrillation”, “dialysis”, “apixaban”, “rivaroxaban”, “edoxaban”, and “dabigatran”. Emphasis was placed on randomized controlled trials (RCTs) and original data.

Atrial fibrillation increases the risk of stroke in AF, but this risk is not homogeneous, and it depends on various risk factors. The more common and validated stroke risk factors have been used to formulate stroke risk stratification scores, such as the CHA2DS2VASc score.12 Of note, stroke risk is not static and regular reassessment is needed, given the increasing age and incident risk factors over time.13

Recognizing that CKD is additive to stroke risk, some studies have proposed the addition of CKD or renal surrogates (eg, proteinuria) to risk scores, such as the ATRIA score14 or the R2CHADS2 score.15 However, other studies have not shown an additive value for stroke prediction by considering CKD.16–18 This is perhaps unsurprising since CKD is highly associated with the individual components of the CHA2DS2VASc score. The CHA2DS2VASc score has been studied in dialysis patients, where it has been validated for stroke prediction.13

Chronic kidney disease also predisposes to an increased risk of bleeding. Although numerous bleeding risk factors have been described including various biomarkers, bleeding and stroke risk factors are often similar, thus stroke and bleeding rates track each other. In this sense, the HAS-BLED score, which incorporates the more common validated bleeding risk factors, has been proposed to assess bleeding risk.19 The appropriate use of the HAS-BLED score is to draw attention to the modifiable bleeding risk factors, and to ‘flag up’ the high bleeding risk patients for more frequent review and follow-up.20 Of note, an approach simply focused on modifiable bleeding risk factors alone is an inferior strategy to HAS-BLED for bleeding risk prediction.13,21,22

Thromboprophylaxis in AF requires oral anticoagulants, which can be well-managed VKA (eg, warfarin) or NOACs (eg, apixaban, rivaroxaban, edoxaban, and dabigatran).3 Although warfarin and NOACs act on the coagulation pathway they differ in their mechanism of action. Warfarin inhibits the vitamin K-dependent clotting factors II, VII, IX and X to influence the international normalized ratio (INR): a low INR increases clotting risk whereas a high INR increases bleeding risk. By contrast, the NOACs target individual clotting proteins: apixaban, rivaroxaban, and edoxaban directly inhibit factor X and dabigatran is a direct thrombin inhibitor23 (Figure 1).

Figure 1.

Mechanism of action of nonvitamin K oral anticoagulants.

(0.08MB).

A prospective study of 565 patients on warfarin demonstrated that individuals with severe renal dysfunction (GFR < 30mL/min per 1.73kg/m2) required lower warfarin doses and were less time in therapeutic range (TTR). Moreover, the incidence rate of major hemorrhage was greater in severe renal dysfunction (30.5 per 100 patient-years) compared with mild (6.2 per 100 patient-years) and moderate (8.3 per 100 patient-years) renal dysfunction.24 In addition, warfarin use is also complicated by multiple drug and food interactions and a reduction in vitamin K-dependent matrix G1a protein that results in increased vascular calcification.25 Warfarin-related nephropathy resulting from glomerular hemorrhage and tubular obstruction by red cell casts is also seen more frequently in patients with CKD.26

Nonvitamin K antagonist oral anticoagulants have changed the landscape for stroke prevention in AF, although regional differences are evident.27 In contrast to warfarin, NOACs have fewer drug and diet interactions, have a rapid onset of action and shorter half-life, and do not require regular laboratory monitoring.28 The short half-life means that adherence and compliance is crucial, and patients need education and counselling.29

When glomerular filtration is impaired, clearance of NOACs is decreased, thus their plasma half-life is prolonged. This may result in increased total drug exposure or area under the curve (AUC), which increases the risk of bleeding complications.30 Dabigatran has significant renal clearance (80% renal excreted), with lower renal excretion seen for edoxaban (50%), rivaroxaban (33%) and apixaban (27%).31 There are limited data for NOACs in severe and ESRD (Cockcroft-Gault creatinine clearance [CrCl] < 25-30mL/min) as these patients were excluded from the phase 3 randomized trials32 (Table 2).

The phase III trials supporting the use of NOACs to prevent thromboembolism in AF were performed with specific NOAC dosing and patient exclusion criteria. Patients with CrCl < 30mL/min (dabigatran, rivaroxaban, and edoxaban) or CrCl < 25mL/min (apixaban) were excluded from the pivotal clinical trials.33–36

The European Society of Cardiology recommendations for moderate CKD (GFR, 30-59mL/min) are based on secondary analyses of phase 3 NOAC trials.37 By contrast, the Food and Drug Administration (FDA) in the United States have approved reduced dose dabigatran 75mg twice daily, apixaban 5mg twice daily (apixaban 2.5mg twice daily if ≥ 80 years old ≤ 60kg) and rivaroxaban 15mg twice daily in patients with CrCl 15 to 29mL/min mainly based on pharmacological modelling data for CrCl 15 to 29mL/min (Table 3).38-42

In a meta-analysis of 13 878 AF patients with moderate CKD, the NOACs were compared through assessing surface under the cumulative ranking (SUCRA) curves. Dabigatran 150mg twice daily was the most efficacious (SUCRA 0.96) followed by apixaban, rivaroxaban, and edoxaban (SUCRA 0.67; 0.53; 0.51 respectively). Apixaban (SUCRA 0.84) and edoxaban (SUCRA 0.61) had the most favorable safety profile.54

Atrial fibrillation in dialysis patients poses a complex clinical dilemma. The highest incidence of AF is seen in hemodialysis patients (15.1%) compared with nonend-stage CKD (9.6%) and in normal renal function (2.6%).2 End-stage renal dysfunction increases bleeding risk secondary to platelet dysfunction (disturbance of platelet α-granules, fibrinogen fragments binding to the glycoprotein (GP) IIb/IIIa receptor on platelets, uremic toxins and abnormal calcium mobilization all contribute to reduced aggregation and adhesion of platelets).34

There are no RCT investigating the use of NOACs in severe and end-stage renal disease. Even ESRD patients were excluded from the RCTs of warfarin for stroke prevention in AF. Nonetheless, data on warfarin use comes from observational cohorts, where good-quality anticoagulation control is crucial to minimize the risks of stroke and bleeding in these high-risk patients.55,56 By contrast, observational data from North America, which generally do not report TTR, suggest that ischemic stroke is not reduced but serious bleeding significantly increased in patients with end-stage renal failure (ESRF).57,58

For example, Harel et al. performed a meta-analysis of 14 observational studies examining the use of warfarin in the prevention of ischemic stroke in ESRD. There was no significant difference in ischemic stroke or intracranial hemorrhage in patients treated with warfarin. The key limitations of this meta-analysis were that the definition of stroke and bleeding varied between studies (for instance transient ischemic attack was accounted for as a primary outcome in certain studies but not recorded in others). The quality of warfarin anticoagulation (ie, TTR) was also not accounted for in the included studies.59,60

Alternatively, a prospective study of warfarin treatment in ESRD (median TTR 54%) showed that higher TTR reduced the risk of bleeding,55 implying that warfarin use in ESRD must be accompanied by adequate INR surveillance for maximum benefit. In a Danish registry study over a 15-year study period, patients with AF were subdivided into those with nonend-stage CKD (11 128 patients) or on renal replacement therapy (1728). There was a lower risk of death on warfarin therapy in both nonend-stage CKD and renal replacement therapy (hazard ratio [HR], 0.64; 95% confidence interval [95%CI], 0.60 -0.69 and HR, 0.85; 95%CI, 0.72-0.99 respectively). In nonend-stage CKD, a lower composite risk of fatal stroke and bleeding (HR, 0.71; 95%CI, 0.57-0.88) was observed.61

The clearance of drugs during hemodialysis is enhanced for small molecules (< 1500 daltons) and intravascular unbound drugs (low volume of distribution) as they pass easily through the dialysis membrane.30 Given these considerations, dabigatran is known to be the only dialyzable NOAC. Surprisingly, registries of ESRF patients report the use of NOACs in these patients. Indeed, data from the US Renal Data System show that NOAC use over warfarin increased from 0.16% to 29.16% (P-for-trend < .001) between October 2010 and December 2015.62

One study has directly compared bleeding rates for patients with AF and ESRD on hemodialysis and initiated on rivaroxaban, dabigatran, or warfarin. The highest rate of major bleeding was for ESRD patients treated with dabigatran (83.1 events per 100 patient-years) and rivaroxaban (68.4 events per 100 patient-years) compared with warfarin (35.9 events per 100 patient-years). Furthermore, mortality in patients taking dabigatran (19.2 deaths per 100 patient-years) was higher than in those on rivaroxaban (16.2 deaths per 100 patient-years) and warfarin (10.2 deaths per 100 patient-years) raising concern about the use of dabigatran in ESRD.63

Ongoing studies may further elucidate the use of NOACs on ESRD patients. The RENAL-AF NCT02942407 trial (Renal Haemodialysis Patients Allocated Apixaban Versus Warfarin in AF) is randomizing 762 patients to receive 2.5 to 5mg twice-daily apixaban or warfarin to measure outcomes of major bleeding episodes, stroke, and mortality over a 15 month period (completion May 2019). The AXADIA trial is a prospective, multicenter study in Germany that has enrolled 222 patients to examine the safety of apixaban vs the VKA phenprocoumon in AF patients on hemodialysis (completion April 2019).38 Future studies should further consider differences in the efficacy and safety of patients maintained on hemodialysis vs peritoneal dialysis as the excretion of NOACs may vary between these renal replacement therapies.

In the clinical environment, we must endeavor to use CrCl in the estimation of renal function to decide an appropriate anticoagulation strategy as it reflects the clinical trials most closely. Indeed, previous studies have shown a disparity in GFR estimation when the Cockcroft-Gault, CKD-EPI and MDRD equations are used.52

Patients prescribed NOACs should also have their renal function monitored to ensure that they are consistently prescribed the correct dose. For instance, a study in primary care has shown that annual renal function monitoring identifies patients who are over- or undercoagulated.74 Furthermore, compliance with renal function monitoring has been shown to increase the likelihood of adequate NOAC dosing at 1-year follow-up.75

We must additionally consider adjusting NOAC doses in acute-on-chronic kidney injury. In an observational study of 162 patients with concomitant AF and heart failure, the measured fluctuation in kidney function merited dosage adjustment (44%, 35%, 29% of patients on dabigatran, rivaroxaban and apixaban, respectively, required dose adjustment).76

The decision to anticoagulate patients with concomitant AF and CKD depends on CKD stage, given the fine balance between the prevention of thromboembolism and excess bleeding especially in ESRF (Figure 2). In moderate CKD, the available data indicate NOACs are at least noninferior to warfarin in the prevention of stroke and systemic embolism with similar to superior safety profile. In severe renal failure (GFR 15-20mL), the use of NOACs is not routinely recommended as the pivotal trials excluded these patients. The EHRA practical guide generally favors the use of warfarin in this group of patients.48 Although the FDA has approved use of reduced dose NOACS in severe renal impairment, this is based on pharmacokinetic studies rather than prospective clinical trials.

Figure 2.

Atrial fibrillation management algorithm. When warfarin is used we should aim for time in therapeutic range ≥ 70%. AF, atrial fibrillation; NOAC, new oral anticoagulant.

(0.21MB).

An individualized approach is required to evaluate the risk vs benefit of anticoagulation in patients with AF and ESRD as there is evidence for and against its use. There are no RCT exploring the use of NOACSs in ESRD. Of note, the largest available head-to-head trial of outcomes in patients taking dabigatran, rivaroxaban or warfarin raised concern regarding the use of NOACs (in particular dabigatran) in hemodialysis due to increased bleeding risk.77 Thus, at present warfarin has more supportive evidence for its use in AF patients with ESRD to prevent stroke and systemic thromboembolism.

G.Y.H. Lip is a consultant for Bayer/Janssen, BMS/Pfizer, Medtronic, Boehringer Ingelheim, Novartis, Verseon and Daiichi-Sankyo and a speaker for Bayer, BMS/Pfizer, Medtronic, Boehringer Ingelheim, and Daiichi-Sankyo; he declares not directly receiving any personal fees derived from these activities.

.