Abstract Body: Introduction During imaging follow-up of unruptured intracranial aneurysms (IA), changes in cerebrovascular (CV) anatomy are commonly found. While the focus is on the IA site, clinical images capture evidence of vascular remodeling and longitudinal change in the Circle of Willis. This study hypothesizes an association between IA and broader CV changes. This study developed a CV longitudinal atlas (CV shape vs. age) to determine a shared trajectory of CV morphological change in IA patients and identified clinical risk factors associated with IA which affected this trajectory.
Methods A total of 405 longitudinal standard clinical magnetic resonance angiography (MRA) image studies were included in the study. On the average there were 3.68 ± 1.34 MRA studies for each IA patient, with follow-up time ranging from 2.5 to 14 years. The Deformetrica was used to analyze the shape changes in the DICOM images. Specifically, a machine learning diffeomorphic analysis of arterial shape with age was performed. A hybrid geodesic regression/Bayesian atlas 4D longitudinal atlas model that inferred the average trajectory of arterial change from patients with overlapping ages of follow-up was created. Parameters mapping individual IA patients to the average trajectory were then analyzed based on patient characteristics.
Results The general longitudinal atlas trajectory consisted of longer cerebral arteries as patient age increased. The Kruskal-Wallis test associated several factors related to IA patient medical histories with this pattern of change. Specifically, diabetes mellitus (p=0.016) and no previous subarachnoid hemorrhage (p=0.023) were associated with a faster median rate of CV change (1.5-fold and 2.7-fold, respectively). Hypertension (p=0.0004), atherosclerosis (p=0.001), and ICA IA (p=0.013) were associated with CV morphology more advanced along the CV-age trajectory.
Conclusion We developed the first longitudinal atlas model of CV morphological change over time. We used this new technique to investigate the association of IA location and related risk factors with the trajectory of CV change. We identified characteristics that affected the speed and degree of CV change in individual patients. Risk factors like diabetes and hypertension impact a variety of vascular diseases, and our findings indicate IA patients with these factors may have broad CV change and may benefit from review of the Circle of Willis, in addition to the IA site during MRA follow-up.