Commentary on “Demystifying the Benefits and Harms of Anticoagulation for Atrial Fibrillation in CKD”

Introduction In this evidence-based nephrology review, Garlo et al. (1) review the efficacy and safety of oral anticoagulants—including warfarin and newer direct oral anticoagulants (DOACs)—for the prevention of stroke and thromboembolism in patients with atrial fibrillation and CKD. Given the high incidence of both ischemic stroke and major hemorrhage among patients with CKD and atrial fibrillation, treatment decisions regarding oral anticoagulation can be challenging. The authors note that there is no evidence from randomized trials on the safety or efficacy of these medications in patients with CKD stages 4 and 5. However, findings from post hoc subgroup analyses of clinical trials in patients with CKD stage 3 and observational studies in patients with more advanced CKD indicate the following: A secondary analysis of the Stroke Prevention in Atrial Fibrillation III (SPAF III) clinical trial showed that dose-adjusted warfarin treatment is superior to low-dose fixed warfarin in preventing thrombembolism in patients with atrial fibrillation and CKD stage 3. There are conflicting results from different observational studies on the safety and effectiveness of warfarin among patients with CKD, including those with eGFR<30 ml/min per 1.73 m2 and those with ESKD. Differences in study populations and ascertainment of medication use and outcomes and differential approaches to confounder adjustment may account for this heterogeneity. Subgroup analyses of multiple randomized, controlled trials of DOACs in patients with CKD stage 3 have reported similar or greater efficacy for stroke prevention compared with warfarin or aspirin without excess risk of hemorrhage. A meta-analysis of these clinical trials reported a roughly 20% lower risk of stroke with DOACs in the subgroup with CKD stage 3. An observational study in patients on maintenance hemodialysis suggested a significant reduction in stroke and systemic embolism in patients receiving higher-dose apixiban compared with warfarin. However, there are conflicting results from different studies on the risk of bleeding with DOACs versus warfarin, with both higher and lower risks reported. Consistent with a recent Kidney Disease Improving Global Outcomes Controversies Conference report, the authors recommend DOACs over warfarin in patients with CKD stage 3 requiring anticoagulation for atrial fibrillation. They further recommend against DOACs in patients with CKD stages 4 and 5 due to insufficient evidence supporting their safety and superiority in this population. Two ongoing randomized trials comparing DOACs with vitamin K antagonists on risk of bleeding in patients with ESKD on hemodialysis should provide additional information regarding the relative safety of these treatments. Teaching Statement—Subgroup Differences in Randomized Trials Reports of randomized trials typically describe the average effect of a treatment among all participants in the trial. However, treatments may have different benefits and harms across specific subgroups of people. In this issue of the Clinical Journal of the American Society of Nephrology, Garlo et al. (1) review the risks and benefits of oral anticoagulation for atrial fibrillation in the setting of CKD. Part of the evidence base derives from subgroup analyses of randomized trials. For example, the SPAF III trial randomly assigned persons with nonvalvular atrial fibrillation and high clinical risk for stroke to either adjusted-dose warfarin (international normalized ratio =2–3) or fixed low-dose warfarin (1–3 mg/d). The effect of these treatments on the outcome of ischemic stroke or systemic embolism is shown below. - Adjusted-Dose Warfarin Low-Dose Warfarin Relative Risk of Ischemic Stroke or Embolism Comparing Adjusted-Dose with Low-Dose Warfarin (95% Confidence Interval) Participants Events Participants Events eGFR≥60 ml/min per 1.73 m2 242 5 259 16 0.33 (0.12 to 0.90) eGFR=30–60 ml/min per 1.73 m2 267 6 249 23 0.24 (0.10 to 0.58) These relative risks suggest a possibly greater benefit of adjusted-dose warfarin on ischemic stroke prevention among persons with CKD stage 3. However, differences in treatment effects seen in subgroup analyses require cautious interpretation, because such differences may or may not represent a true differential effect of the treatment on the outcomes under study. In general, subgroup differences are more likely to be valid when the following criteria are met. The magnitude of difference in the treatment effect across subgroups is large. The subgroups contain sufficient numbers of people and outcomes for comparison. The confidence intervals do not overlap with one another. The differential treatment effects are plausible on the basis of evidence from other studies. The differential treatment effects are replicated in other studies. In this example, the amount of difference in the relative risks across CKD subgroups is small: a 1.0−0.24=76% relative risk reduction in the CKD group versus a 1.0–0.33=67% reduction in the non-CKD group. The number of study participants in each group is reasonably large; however, the number of stroke outcomes is relatively small. The 95% confidence intervals for the relative risks in the CKD and non-CKD subgroups overlap with one another, suggesting that the observed difference may have arisen due to chance. Finally, there is no straightforward biologic rationale as to why adjusted-dose warfarin should be particularly beneficial for preventing stroke among persons with CKD stage 3. On the basis of these subgroup analyses and the absence of other supportive data, there is little reason to suspect that adjusted-dose warfarin has a differential effect on stroke reduction, specifically in CKD stage 3. Disclosures None.