Highlights
- The BRAIN-AF trial evaluated whether 15 mg of rivaroxaban daily could prevent cognitive decline, stroke, or TIA in patients with atrial fibrillation (AF) and low thromboembolic risk (CHA2DS2-VASc score 0-1).
- The trial was stopped early for futility after an interim analysis showed a 1.2% probability of achieving a statistically significant treatment effect.
- Cognitive decline occurred at a high rate in both the rivaroxaban and placebo groups (7.0% vs. 6.4% per year), suggesting that factors beyond macro-thromboembolism may drive neurocognitive impairment in low-risk AF populations.
- Safety outcomes remained favorable, with low annual rates of major bleeding in both the treatment and control arms.
Background: The Interplay Between Atrial Fibrillation and Cognition
Atrial fibrillation (AF) is traditionally managed with a focus on preventing ischemic stroke and systemic embolism. However, a growing body of evidence suggests that the burden of AF extends beyond acute neurological events to include chronic, progressive cognitive impairment and dementia. Observational studies have consistently identified AF as an independent risk factor for cognitive decline, even in the absence of clinical stroke. The hypothesized mechanism often involves subclinical cerebral micro-emboli, which may lead to cumulative white matter damage and silent infarcts.
Current clinical guidelines strongly recommend oral anticoagulation (OAC) for patients with elevated CHA2DS2-VASc scores. However, the benefit of OAC in patients with a low thromboembolic risk (scores of 0 or 1, excluding female sex) remains a subject of intense debate. While these patients are at low risk for major strokes, they may still be vulnerable to the subtle, cumulative neurocognitive effects of AF. The BRAIN-AF (Blinded Randomized Trial of Anticoagulation to Prevent Ischemic Stroke and Neurocognitive Impairment in Atrial Fibrillation) trial was designed to address this clinical uncertainty by investigating whether low-dose rivaroxaban could mitigate these risks in a low-risk population.
Study Design and Methodology
The BRAIN-AF study was a multicenter, double-blind, placebo-controlled randomized clinical trial. The study enrolled 1,235 participants out of an intended 1,424 from various clinical sites. The inclusion criteria focused on individuals with documented AF and low thromboembolic risk, defined as a CHA2DS2-VASc score of 0 or 1 (excluding the point for female sex).
Participants were randomized in a 1:1 ratio to receive either rivaroxaban 15 mg once daily or a matching placebo. It is notable that the 15 mg dose is lower than the standard 20 mg dose typically used for stroke prevention in AF patients with normal renal function; this choice was intended to balance potential neuroprotective benefits against the risk of bleeding in a lower-risk cohort.
The primary outcome was a composite endpoint consisting of:
1. Cognitive decline, defined as a drop of 2 points or more on the Montreal Cognitive Assessment (MoCA).
2. Ischemic stroke.
3. Transient ischemic attack (TIA) associated with a motor deficit or aphasia.</n
Secondary outcomes included individual components of the primary composite, major bleeding events, and overall mortality. The trial was designed with a planned interim analysis to assess for efficacy or futility.
Key Findings: The Futility of Anticoagulation for Neuroprotection
The BRAIN-AF trial was halted prematurely following a planned interim analysis which met the predetermined criterion for futility. At the time of cessation, 1,235 participants (919 men and 316 women) had been enrolled, with a median follow-up period of 3.7 years.
Primary Outcome Analysis
The primary composite outcome occurred in 256 participants (20.7% of the total cohort). Specifically, the annual rate of the primary outcome was 7.0% in the rivaroxaban group compared to 6.4% in the placebo group. The resulting hazard ratio (HR) was 1.10, with a 95% confidence interval (CI) of 0.86 to 1.40 (P = 0.46). These data indicated no significant difference between the two groups, and a conditional power analysis estimated only a 1.2% chance of achieving statistical significance had the trial reached its full enrollment of 1,424 participants.
Cognitive Decline Incidence
Perhaps the most striking finding of the trial was the high incidence of cognitive decline across the entire study population, regardless of treatment. A drop of 2 or more points on the MoCA scale was the primary driver of the composite endpoint. This high rate of decline in a relatively young, low-risk population underscores the significant burden of neurocognitive impairment associated with AF, yet it simultaneously demonstrates that low-dose rivaroxaban is not an effective intervention for this specific issue.
Safety and Bleeding Risks
Regarding safety, the trial reported that major bleeding was rare. Only two patients in the rivaroxaban group (0.09% per year) and five patients in the placebo group (0.21% per year) experienced major bleeding events. This confirms that while the intervention was not effective for the primary endpoint, the low-dose rivaroxaban regimen was relatively safe in this low-risk demographic.
Expert Commentary and Mechanistic Insights
The results of the BRAIN-AF trial present a complex challenge to the prevailing hypothesis that subclinical micro-emboli are the primary cause of cognitive decline in AF. If micro-emboli were the dominant factor, one would expect a potent anticoagulant like rivaroxaban to show at least a trend toward neuroprotection. The lack of any such trend—and the high rate of decline in both groups—suggests that other mechanisms may be at play.
Potential alternative mechanisms for AF-related cognitive decline include:
1. Cerebral Hypoperfusion: The irregular heart rate in AF can lead to beat-to-beat variability in stroke volume, potentially causing chronic cerebral hypoperfusion and subsequent white matter disease.
2. Shared Risk Factors: AF and dementia share common precursors, such as hypertension, diabetes, and systemic inflammation. These factors may drive neurodegeneration independently of embolic events.
3. Micro-hemorrhages: It is also possible that even low-dose anticoagulation could increase the risk of cerebral micro-bleeds, which might offset any benefit gained from preventing micro-infarcts.
Critics and commentators have also questioned whether the 15 mg dose of rivaroxaban was sufficient to achieve the desired neuroprotective effect, or if the MoCA scale is sensitive enough to capture the specific types of cognitive changes associated with AF-related vascular injury. However, the early termination for futility strongly suggests that standard anticoagulation strategies are not the answer for cognitive preservation in low-stroke-risk patients.
Conclusion and Clinical Implications
The BRAIN-AF trial provides high-quality evidence that for patients with AF and low thromboembolic risk (CHA2DS2-VASc 0-1), initiating low-dose rivaroxaban does not prevent cognitive decline, stroke, or TIA. While the study confirms the high prevalence of cognitive deterioration in AF patients, it effectively closes the door on the use of OACs as a primary neuroprotective strategy in this specific subgroup.
For clinicians, these results reinforce the importance of adhering to current guidelines, which do not mandate anticoagulation for patients with a CHA2DS2-VASc score of 0 (men) or 1 (women). Future research should focus on identifying the non-embolic pathways of AF-related cognitive impairment, potentially investigating rate and rhythm control strategies, aggressive management of cardiovascular risk factors, or novel therapeutic targets that address cerebral perfusion and neuroinflammation.
Funding and ClinicalTrials.gov
The BRAIN-AF trial was supported by various research grants and institutional funds (detailed in the full manuscript). ClinicalTrials.gov registration: NCT02387229.
References
Rivard L, Khairy P, Talajic M, et al. Anticoagulation to prevent ischemic stroke and neurocognitive impairment in atrial fibrillation: the BRAIN-AF randomized clinical trial. Nat Med. 2026;32(1):297-305. doi:10.1038/s41591-025-04101-y.