Abstract Body (Do not enter title and authors here): Introduction: There is a bidirectional relationship between atrial fibrillation (AF) and heart failure (HF) risk. Data regarding shared and unique pathways underlying each are limited.

Aims: Identify plasma proteins associated with both incident AF and HF and determine the extent to which associations with one outcome are conditional on interval occurrence of the other outcome.

Methods: Among participants in the community-based Atherosclerosis Risk in Communities study free of HF and AF at study visit 5 (2011-2013), we measured 4,955 plasma proteins (SomaScan aptamer-affinity assay). We performed parallel analyses of individual protein associations with incident HF and AF using multivariable Cox regression adjusted for demographic and clinical covariates at Bonferroni significance. We then repeated the analysis for each outcome in models censoring participants with interval occurrence of the other outcome (i.e. incident AF censoring for interval HF, and vice versa). We tested associations of AF-related proteins with echocardiographic and Zio patch measures at Visit 5.

Results: Among 4,118 participants included, mean age was 75±5 years, 60% were female, and 18% reported Black race. We identified 83 proteins associated with incident HF (n=355 incident events at 7 [IQR 6-8] year follow-up) and 24 with incident AF (n=517 incident events at 8 [7-9] years), of which 10 were associated with both. In models of incident HF censoring for interval AF, 63 (76%) proteins remained significantly associated with HF. In contrast, in the analysis of incident AF censoring for interval HF, only 4 (17%) proteins remained associated with AF, of which 3 were also associated with HF risk (Figure). These 4 AF-related proteins were associated with measures of LV diastolic function (LA volume, E/e’. average), LV mass index, LA function (LA reservoir, LA contraction), and supraventricular ectopy density (all p<0.05).

Conclusion: Nearly half of proteins reflecting AF risk also associate with HF, greater than 80% of AF-associated proteins are HF-dependent, and even those that are not correlate with measures of LV remodeling and dysfunction. These findings suggest that AF-related mechanisms among older adults are part of a wider HF syndrome.