Anticoagulation to prevent ischemic stroke and neurocognitive impairment in atrial fibrillation: the BRAIN-AF randomized clinical trial

Abstract

Individuals with atrial fibrillation (AF) are at increased risk of stroke, cognitive impairment and dementia. Observational studies suggest that anticoagulation may reduce the risk of cognitive decline in patients with AF and elevated thromboembolic risk, implicating subclinical cerebral emboli as a potential mechanistic link. Whether anticoagulation prevents cognitive deterioration in patients with AF at low risk of stroke remains uncertain. Here we conducted a multicenter, double-blind, placebo-controlled trial in which participants with AF and low thromboembolic risk (CHA₂DS₂-VASc scores of 0 or 1, excluding female sex) were randomized 1:1 to receive rivaroxaban 15 mg daily or placebo. The primary outcome was a composite of cognitive decline (≥2-point drop in Montreal Cognitive Assessment), stroke or transient ischemic attack with a motor deficit or aphasia. The trial was halted after meeting the predetermined futility criterion following a planned interim analysis, with 1,235 of the intended 1,424 participants (919 men; 316 women) enrolled. Over a median follow-up of 3.7 years, the primary outcome occurred in 256 (20.7%) participants, at an annual rate of 7.0% with rivaroxaban versus 6.4% with placebo, yielding a hazard ratio of 1.10 (95% confidence interval (0.86-1.40); P = 0.46). Conditional power analysis indicated a 1.2% probability of achieving a statistically significant treatment effect if the trial had been continued to its planned total of 410 events. Major bleeding occurred in two patients treated with rivaroxaban (0.09% per year) and five patients treated with placebo (0.21% per year). In conclusion, despite the high incidence of cognitive decline observed among patients with AF and low stroke risk, the BRAIN-AF trial, which tested a low dose of rivaroxaban to prevent stroke, transient ischemic attack and cognitive decline in patients with prior AF, was stopped early due to futility. ClinicalTrials.gov registration: NCT02387229 .

Fig. 1. Flowchart of participant distribution throughout the study.

‘Study completion’ refers to patients who either attended the final visit scheduled at trial cessation or had a final visit upon turning 65 years of age before the trial ended.

Fig. 2. Primary efficacy outcome.

a–c, The primary efficacy outcome (cognitive decline, stroke or TIA) according to randomization to rivaroxaban versus placebo is shown for the main analysis (a), censored analysis (b) and on-treatment analysis (c). The outcome is interval-censored and graphically displayed using non-parametric maximum likelihood estimates of survival curves for interval-censored data, resulting in stepwise curves. Numbers at risk are approximated by assuming that events occurred at the upper bound of the interval. Hazard ratios, 95% CIs and P values were obtained from proportional hazards regression models fitted to handle interval-censored data and stratified for study phase. The trial was terminated early for futility. By the time of study closure, 256 of 410 planned primary events had occurred.

Fig. 3. Subgroup analyses of the primary outcome.

Error bars represent point estimates of hazard ratio ± 95% CI obtained from proportional hazards models fitted to handle interval-censored data. Models included terms for treatment, subgroup and treatment × subgroup interaction and were stratified by study phase (except for the subgroup analysis on study phase). The size of each square (point estimate) is proportional to the number of events. P values refer to the test of the treatment × subgroup interaction terms. All tests were two-sided, and no adjustment for multiple comparisons were done.