Anatomical characteristics are associated with aneurysm sac regression after endovascular repair.

Abstract

Recent findings show that patients with abdominal aortic aneurysm (AAA) sac regression after endovascular repair (EVAR) have significantly better long-term outcomes than patients with a stable or expanding sac. Previous studies have not yet identified strong predictors of sac regression, but suggest that anatomical AAA parameters might play a role in the remodeling. This study aimed to conduct a comprehensive analysis of preoperative AAA anatomy to identify predictors of sac regression 1 year after EVAR. This retrospective two-center cohort study included patients with regressing or stable sacs one-year after elective infrarenal EVAR between January 2011 and December 2019. The lumen and intraluminal thrombus (ILT) of the AAA were automatically segmented. From this, the center lumen line and a range of anatomical characteristics were automatically derived, including volume, diameter and length measurements, (hostile) neck parameters, ILT distribution, radiodensities, calcification, and aortic curvature. Logistic regression was performed to identify the added predictive value of the anatomical characteristics. Enrolled were 289 patients (84% male, 72.0 ± 7.4 years old) of whom 46% experienced sac regression and 54% had a stable sac 1 year after EVAR. Significant differences between the regression and stable group were found for the relative neck ILT volume (1.2 ± 1.6 vs 1.7 ± 2.2; P = .029), location of maximum thrombus thickness (187° ± 109° vs 156° ± 105°; P = .018), variation in ILT circumference (25 ± 10% vs 23 ± 10%; P = .029), the roundness of the lumen (1.46 ± 0.24 vs 1.52 ± 0.25; P = .029), and variation in lumen radiodensity (77 ± 25 Hounsfield units vs 85 ± 24 Hounsfield units; P = .005). No significant differences were observed for the other volumes, lengths, and (hostile) neck parameters. Compared with a baseline model that included only clinical variables, the addition of the anatomical variables increased the model sensitivity from 42.1% to 58.3% and the overall accuracy from 59.9% to 66.3%. For the receiver operating characteristic curve, the area under the curve also significantly improved by 0.08 to 0.71 when the anatomical variables were added. Anatomical AAA characteristics relating to ILT play an important role in sac regression after EVAR. Furthermore, this study supports the existing evidence confirming that various mostly geometric neck parameters are not associated with sac regression, such as diameter, calcification, length, angulation, shape, and presence of a hostile neck. Although the anatomical analysis is of added value for predicting sac regression, further improvement is needed before clinical implementation.