1356745 - Association of Atrial 4D Flow Hemodynamics with Stroke History, Race and Sex in AF Patients

Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia. AF has been shown to increase stroke risk due to thrombus formation in the left atrium (LA), with further increased stroke risk in the black AF population.¹ AF-associated reduction in LA blood flow velocities and increased blood stasis are key component of thrombus formation and may be associated with increased stroke risk. The purpose of this study was to investigate 4D flow MRI derived atrial flow differences between AF patients with and without history of cardioembolic stroke, while controlling for race (white and black populations) and sex.  

Methods:

56 AF patients (68 ± 8 years, 37 white/19 black, 36 males, 28 prior stroke history) were prospectively recruited to undergo whole heart 4D Flow MRI (fig 1). Following pre-processing, the LA was segmented (Mimics, Materalise) to mask the 4D flow data and calculate LA peak velocity and blood stasis (percent of the cardiac cycle a voxel velocity < 10 cm/s). Mean stasis over the LA volume was derived. Racial, sex, and stroke history differences in peak velocity, mean stasis, and LA volumes were assessed. These parameters and demographic data were also used for multivariable analysis, using forward/backward stepwise general logistic regression to identify independent predictors of stroke history.

Results:

Peak velocity, mean stasis, and LA volumes were not statistically different between white and black patients (fig 2). Mean LA stasis was nearly significantly different between AF patients with a history of stroke vs no stroke history (45 ± 14% vs 52 ± 15%, p = 0.05). Peak velocity, stasis, and LA volumes were significant predictors of stroke risk in the multivariable model, alongside age, with significant interaction terms between stasis with LA volume/peak velocity (fig 3). The McFadden pseudo R-squared value was 0.53, indicating excellent fit, and the multivariable model was able to highly discriminate between stroke history status (fig 3, right).

Conclusion:

Black and white AF patients showed similar 4D flow measures for peak velocity and atrial stasis. 4D flow MRI derived left atrial volumes and flow measures were significant predictors of history of stroke. Importantly, flow parameters and atrial morphology complemented each other when predicting stroke history, underscored by peak velocity and volume not being significant univariate predictors.  However, direct interpretation of coefficients is difficult due to the interaction effects. Inclusion of the interaction between sex and race may be due to all 8 black males having a prior stroke history, a limitation of the study. Further, the cardiac rhythm during 4D flow acquisition was not assessed. 4D flow parameters successfully helped to model stroke history, encouraging their use for further prospective stroke studies.