Abstract Body: Introduction: Atrial fibrillation (AF) is highly prevalent, and its health effects outside of stroke are under intense study. AF increases dementia risk, but underlying mechanisms are unknown. Stroke or cerebrovascular pathophysiology, possibly attenuated with anticoagulation, may be contributory.

Hypotheses: We assessed two hypotheses: (1) that higher levels of 9 circulating biomarkers (see Figure) would be associated with greater risk of incident cognitive impairment (ICI) in REGARDS participants with AF, and (2) that associations would be attenuated after adjusting for oral anticoagulant use.

Methods: REGARDS enrolled 30,239 persons >45 years old in 2003-07. AF was defined by self-report or ECG. ICI was defined during follow-up by robust cognitive norms based on the Montreal Cognitive Assessment and Six-Item Screener. Biomarkers were measured at baseline in participants with AF and no prior stroke. HRs of time to ICI by biomarkers were calculated by Cox proportional hazards models, with covariates shown in the Figure. Interaction testing by baseline oral anticoagulant use was also performed.

Results: Among 2,261 participants with baseline AF, mean follow-up was 10.6 years (mean age 66.5 years, 51% women, 28.7% Black), and there were 149 ICI cases (7%). Higher NT-proBNP, FVIII, and GDF15 were each associated with ICI (Figure), with the largest association for FVIII (HR_adj 2.41 per SD higher FVIII). Added adjustment for anticoagulant use did not attenuate associations, but the association of GDF15 was lower among those on anticoagulation (HR 1.0, 95% CI 0.4-2.6 compared to HR 1.8, 95% CI 1.1-3.0 without anticoagulation; p interaction 0.07).

Conclusion: Higher NT-proBNP, FVIII, and GDF15 were associated with ICI in this biracial cohort with AF. Risk associated with GDF15, an inflammation marker, might be reduced with anticoagulation, but further study is needed.