Abstract Body (Do not enter title and authors here): Intro: Risk for dissection and rupture of aorta is determined, in part, by geometry. Diameter is the most commonly assessed risk factor, but other aspects of geometry also influence risk. In retrospective studies, greater aortic length, arch width, tortuosity, and aortic volume have been associated with adverse aortic events. One large-scale prospective analysis has linked 3D aortic geometry to future risk of arrhythmia and heart failure(Beeche et al., 2024). However, to our knowledge, 3D geometric measurements beyond length and diameter remain to be linked to adverse aortic events in a prospective cohort.

Aims: We sought to measure the geometric properties of the aorta and link them to future risk for aortic dissection.

Methods: We developed 3D semantic segmentation (voxel labeling) deep learning models and applied these to available magnetic resonance imaging from UK Biobank (application #41664), followed by performing centerline extraction. We defined a novel thoracic aortic measurement in which all centerline points were projected onto the same axial plane, with the distance between the two farthest points defined as the “projected aortic arch width.” Disease association was computed using Cox models.

Results: 3D aortic measurements were computed in 63,114 UK Biobank participants. For a composite of aortic interventions, aortic dissection, or death attributed to aortic disease (53 events), both ascending aortic diameter (hazard ratio [HR] 2.10 per standard deviation [SD], 95% confidence interval [CI] 1.92-2.29) and projected aortic arch width (HR 2.81 per SD, 95% CI 2.26-3.49) were associated when tested separately after adjustment for age and sex. When tested in a joint model, they remained independently associated: HR 1.93 (95% CI 1.73-2.14) for diameter and HR 2.07 (95% CI 1.62-2.65) for arch width.

Conclusions: A measurement extracted from 3D aortic geometry, the projected aortic arch width, is an independent risk factor for adverse aortic events.