Rivaroxaban vs Vitamin K Antagonist in Patients With Atrial Fibrillation and Advanced Chronic Kidney Disease

Background Treatment with vitamin K antagonists (VKAs) has been linked to worsening of kidney function in patients with atrial fibrillation (AF). Objectives XARENO (Factor XA-inhibition in RENal patients with non-valvular atrial fibrillation Observational registry; NCT02663076) is a prospective observational study comparing adverse kidney outcomes in patients with AF and advanced chronic kidney disease receiving rivaroxaban or VKA. Methods Patients with AF and an estimated glomerular filtration rate (eGFR) of 15 to 49 mL/min/1.73 m2 were included. Blinded adjudicated outcome analysis evaluated adverse kidney outcomes (a composite of eGFR decline to <15 mL/min/1.73 m2, need for chronic kidney replacement therapy, or development of acute kidney injury). A composite net clinical benefit outcome (stroke or systemic embolism, major bleeding, myocardial infarction, acute coronary syndrome, or cardiovascular death) was also analyzed. HRs with 95% CIs were calculated using propensity score overlap weighting Cox regression. Results There were 1,455 patients (764 rivaroxaban; 691 VKA; mean age 78 years; 44% females). The mean eGFR was 37.1 ± 9.0 in those receiving rivaroxaban and 36.4 ± 10.1 mL/min/1.73 m2 in those receiving VKA. After a median follow-up of 2.1 years, rivaroxaban was associated with less adverse kidney outcomes (HR: 0.62; 95% CI: 0.43-0.88) and all-cause death (HR: 0.76, 95% CI: 0.59-0.98). No significant differences were observed in net clinical benefit. Conclusions In patients with AF and advanced chronic kidney disease, those receiving rivaroxaban had less adverse kidney events and lower all-cause mortality compared to those receiving VKA, supporting the use of rivaroxaban in this high-risk group of patients.

T he prevalence of both atrial fibrilla- tion (AF) and chronic kidney disease (CKD) increases with advanced age and these patients are at increased risk for both thrombotic and bleeding events. 1,2vertheless, long-term oral anticoagulation (OAC) is utilized in most patients with AF and CKD to prevent thromboembolic stroke and systemic embolism.As CKD increases bleeding risk, patients with AF and advanced CKD stages $4 represent a high-risk population for which treatment decisions for or against OAC are especially challenging. 3,4C with vitamin K antagonists (VKAs) has been associated with accelerated calcification of coronary and extra-coronary arteries 5,6 as well as cardiac valves. 7,8A systematic review and meta-analysis revealed a significantly elevated OR (1.8, IQR: 1.43-2.24)for extra-coronary calcifications in patients treated with a VKA as compared to patients without anticoagulation or receiving other anticoagulants. 9The observed accelerated decline of kidney function in patients with AF and CKD treated with VKA 10 has also been attributed to these vascular side effects of VKA.However, there are additional concerns about the risk of anticoagulation nephropathy 11,12 during VKA treatment, although this condition may develop in response to any anticoagulant. 135][16][17][18][19] DOACs including rivaroxaban may not only lack a detrimental effect on arterial calcification but may even induce kidney sparing or preserving effects attributable to inhibition of protease-activated receptor-mediated proinflammatory effects in the vasculature 20 and in early vascular aging in CKD. 21ailable large registries assessing the effectiveness and safety of OAC in patients with AF included only a small proportion of patients with concomitant advanced CKD.Hence, in the large GARFIELD-AF (Global Anticoagulant Registry in the Field-Atrial Fibrillation) registry, physicians classified 10.9% of patients as having moderate-to-severe CKD, 22  France, Belgium, and Luxembourg). 23

Kreutz et al
Anticoagulation in Atrial Fibrillation and CKD treatment regimen when consented into XARENO (Supplemental Figure 1).Prespecified follow-up was at least 12 months followed by a planned extended data collection period for 1 up to 2 additional years.
OUTCOMES.The primary outcome of interest was the absolute change in eGFR in mL/min/1.73m 2 (as estimated by the Chronic Kidney Disease Epidemiology Collaboration equation) 24  To adjust for imbalances in patient characteristics between the rivaroxaban and VKA arms at baseline, we calculated propensity scores 25 based upon multivariable logistic regression using 42 distinct demographic, comorbidity, laboratory, and concurrent medication variables known to be risk factors for differential OAC exposure (Table 1, Supplemental Appendix).Estimated propensity scores were subsequently used to weight patients for analysis using overlap weighting (OLW). 26Propensity score OLW Kreutz et al Anticoagulation in Atrial Fibrillation and CKD    3).However, allcause mortality was found to be reduced with rivaroxaban compared to VKA (HR: 0.76; 95% CI: 0.59-0.98)(Table 3, Figure 2, Central Illustration).Taking rivaroxaban on the initial study visit was associated with a 24% lower relative hazard of discontinuing the anticoagulation therapy compared to VKA (HR: 0.76; 95% CI: 0.62-0.92)(Table 3).Our sensitivity analysis The flow of patients who participated in the registry is shown, including the number originally enrolled, and eligible participants included into the intention-to-treat population.The small group of 89 patients that were not treated with anticoagulation at the discretion of participating physicians were according to the study protocol 23    Annual event rates and adjusted HRs (95% CI) for the comparison between rivaroxaban and vitamin K antagonist (VKA) after propensity score overlap-weighted outcome analysis.CKD ¼ chronic kidney disease; XARENO ¼ Factor XA-inhibition in RENal patients with non-valvular atrial fibrillation Observational registry.
Kreutz et al Anticoagulation in Atrial Fibrillation and CKD showed that the hazard for the composite of any adverse kidney outcome with death as a competing risk was similar to the main finding (sub-HR: 0.60; 95% CI: 0.36-0.99).

DISCUSSION
KIDNEY OUTCOMES.In the current study, rivaroxaban use in AF patients with comorbid CKD was associated with a 38% reduction in the hazard of experiencing an adverse kidney outcome, including a 61% reduction in need for chronic KRT and a 49% reduction in progression of kidney function decline to an eGFR <15 mL/min/1.73m 2 vs a VKA.6][17] It is unclear whether the higher risk of adverse kidney outcomes with VKA compared to rivaroxaban observed in the current and prior real-world evidence studies [15][16][17] is due to a detrimental effect of VKAs on vascular injury and calcification, 5,9,27 kidney sparing or preservation effects of rivaroxaban possibly attributable to reduced protease-activated receptor-mediated inflammation 20,27 or some combination of both.randomized controlled trial. 28RENO did not aim and was not powered to show reductions in thrombotic or bleeding outcomes in patients with advanced CKD. 23 However, a recent patient-level network meta-analysis of the RCTs that compared DOACs with warfarin indicated that, compared with warfarin, standard-dose DOAC use was more effective to reduce the risk for stroke/systemic embolism across the spectrum of kidney function above CrCl of at least 25 mL/min. 29Another recent network meta-analysis in patients with AF and CKD (overall 19 studies) 30 including both subgroup   31 OAC resulted in a significant risk reduction (HR: 0.34; 95% CI: 0.19-0.61;P < 0.001) in the rate of stroke and systemic embolism as compared to placebo, while the risk for major bleeding was numerically but not statistically significantly higher (HR: 1.87; 95% CI: 0.90-3.89;P ¼ 0.09) in the OAC group and death rates were similar between groups. 31wever, the question whether low-dose treatment with the Factor Xa inhibitor edoxaban as compared to patients without OAC would impact on kidney outcomes was not addressed in this study. 31nual mortality was high in the XARENO study population which is in agreement with the mean age of about 78 years and the inclusion of a high-risk AF population in which the presence of advanced CKD contributed further to mortality. 32All-cause mortality was found to be reduced with rivaroxaban compared to VKA, but cardiovascular mortality was not.
Whether the reduction in all-cause mortality was a downstream effect of slowing the decline in kidney function or due to residual confounding specific to all-cause mortality or noncardiovascular death (w7 of 10 AF patients die of cardiovascular causes) 33 in XARENO is unclear.Finally, rivaroxaban was associated with better persistence to therapy than VKA; a finding that mirrors observations from prior realworld studies. 34UDY LIMITATIONS.This study has limitations worth discussion.The protocol prespecified that only patients with at least 3 months OAC pretreatment could be included, which limits generalizability of our findings to newly treated patients.However, this mandatory pretreatment phase was important to reduce the risk of selection bias at the time of enrollment, since OAC selection was already done as part of clinical routine and independent from the study procedures.Furthermore, thromboembolic and bleeding endpoints as well as treatment discontinuations for side effects are known to occur at a higher rate during the early phase of anticoagulant treatment. 35,36For newly treated patients, the need to establish a stable international normalized ratio at the beginning of VKA for chronic KRT and a decline to an eGFR <15 mL/min/1.73m 2 , when compared to use of VKA.
This occurred against a similar risk for net clinical benefit including stroke and thromboembolism and major bleeding events.XARENO thus provides important prospective observational evidence on the effectiveness and safety of rivaroxaban and VKA therapy when used in routine practice within the vulnerable group of patients with AF and advanced CKD.
Management of patients was at the discretion of the participating physicians.The study was registered with clinical trials.gov(NCT02663076).Inclusion criteria were a diagnosis of AF as diagnosed by the participating physicians, adult age ($18 years), and an eGFR between 15 and 49 mL/min/ 1.73 m 2 as estimated by the Chronic Kidney Disease Epidemiology Collaboration equation24 and an indication for anticoagulation.The XARENO protocol was approved by all responsible independent ethic committees and informed consent was obtained for all recruited patients.The study protocol has been reported (Supplemental Appendix).23To be included, patients had to be treated with rivaroxaban or VKA for at least 3 months prior to enrollment.Patients continued their ongoing anticoagulationA B B R E V I A T I O assigns weights to patients proportional to their probability of belonging to the opposing treatment cohort (ie, rivaroxaban patients were weighted by the probability of receiving VKA (or 1-the propensity score) and VKA patients were weighted by the probability of receiving rivaroxaban (the propensity score).OWL was chosen as the primary method for confounder adjustment because it allows for all eligible patients to be included in the analysis (unlike propensity score matching which typically results in sample size reduction in one or both cohorts), it assigns greater weight to patients in which treatment cannot be predicted and lesser weight to patients with extreme propensity scores preventing outliers from dominating the analysis and decreasing precision (a concern with inverse probability weighting) and because overlap weighting has the favorable property of resulting in the exact balance (absolute standardized differences [ASDs] ¼ 0%) for all variables included in the multivariable logistic regression model used to derive propensity scores.STATISTICAL ANALYSIS.All analyses used the intention-to-treat dataset.The difference in mean change in eGFR from baseline between groups was compared using a Student's t-test.Clinical outcomes were compared between groups using a propensity score OWL Cox proportional hazards regression model using a robust estimator.Patients were censored in the Cox models at the time of outcome occurrence, death, or end of study follow-up.Results are presented as HRs and corresponding 95% CIs.The proportionality assumption was assessed by evaluating Schoenfeld residuals and was not significant for any outcome.As a post hoc sensitivity analysis, we constructed a Fine and Gray competing risk regression model for the outcome of any adverse kidney outcome controlling for death.All database management and statistical analysis were performed using SAS version 9.4 (SAS Institute).No statistical adjustments for multiple hypothesis testing were performed.The modest sample size precluded meaningful subgroup analyses according to patient characteristics.A strength of the study was its pro-Overall 1,455 patients receiving either rivaroxaban (N ¼ 764) or a VKA (N ¼ 691) were included in the study and available for propensity score OLW outcome analysis (Figure 1).Patient characteristics including demographics, comorbidities, and comedications prior and following OLW are summarized in rivaroxaban and VKA arms (HR: 0.74; 95% CI: 0.40-1.34).Similarly, no significant differences were found for the composite net clinical benefit outcome (HR: 0.97; 95% CI: 0.72-1.31)or any of the individual component outcomes including no significant difference between groups for cardiovascular death (HR: 0.82; 95% CI: 0.54-1.25)(Table

FIGURE 1
FIGURE 1 Flow Diagram With Patient Disposition in the XARENO Registry not included in the propensity score overlap weighting outcome analysis.XARENO ¼ Factor XA-inhibition in RENal patients with non-valvular atrial fibrillation Observational registry.
THROMBOEMBOLIC AND BLEEDING OUTCOMES.In respect to thrombotic and major bleeding outcomes, no significant difference in the composite net clinical benefit outcome or any of the individual component outcomes was observed when comparing rivaroxaban and VKA in agreement with the ROCKET-AF (Rivaroxaban Once-daily, oral, direct factor Xa inhibition compared with vitamin K antagonism for prevention of stroke and Embolism Trial in Atrial Fibrillation)

FIGURE 2
FIGURE 2 Main Outcomes

22 CONCLUSIONS
therapy (when subtherapeutic and supratherapeutic effects are common) would have clearly biased the study in favor of rivaroxaban.Therefore, the mandatory pretreatment phase was an important measure to reduce confounding within our study.Moreover, although data on measurements of serum creatinine or estimation of GFR were prospectively collected every 3 months during follow-up, due to the noninterventional design of the study, the data set was heterogeneous and measurements occurred inconsistently over time and could have been additionally influenced by restrictions during the COVID-19 pandemic.37Thus, the power to assess outcomes such as absolute decline of eGFR over time or other established outcomes based on serum creatinine measurements such as doubling in serum creatinine or >30% decreases of eGFR was limited.On the other hand, while these are reasonable kidney outcomes to evaluate, they are only surrogate in nature.Outcomes such as the need for chronic KRT arguably carry greater clinical relevance.Regardless of the optimization of the methodology and the number of covariates used in propensity score analysis, residual confounding cannot be fully excluded.38The number of patients analyzed in XARENO seems modest, however, the registry included largely patients with at least CKD stage 3b (94%) and the sample size is still comparable with the number of patients with advanced CKD that were included in other large registries. 3,Among patients with AF and CKD, use of rivaroxaban was associated with a reduction in patients' risk of adverse kidney outcomes including the need J A C C : A D V A N C E S , V O L . 3 , N O . 2 Anticoagulation in Atrial Fibrillation and CKD

TABLE 3
Propensity Score Overlap-Weighted OutcomesValues are %/y unless otherwise indicated.Annual event rates and adjusted HRs with 95% CIs for the comparison between rivaroxaban and vitamin K antagonist (VKA) after propensity score overlap weighting outcome analysis.a Individual components of any adverse kidney outcome.b Individual components of net clinical benefit.