Beyond the AJR: “Prediction of abdominal aortic aneurysm growth using geometric assessment of computerised tomography images acquired during the aneurysm surveillance period”

Summary of the Investigation Screening for abdominal aortic aneurysm (AAA) leading to prophylactic interventions and initiation of medical therapy has been shown to reduce all-cause mortality. Although the overall gain far exceeds the harm of overdiagnosis, the surveillance frequency and optimal timing for elective surgery in asymptomatic patients remain elusive targets. This goal requires accurate prediction of aneurysm growth and rupture. Current management guidelines are based on AAA diameter, but the pathophysiology of AAA cannot be fully explained by one parameter. Many patients undergo CT during their course of surveillance. CT can be used to develop geometric and fluid dynamic models with the goal to more accurately predict AAA growth rate. In a recent article by Chandrashekar et al. [1], the authors propose to use AAA diameter, undulation index (UI), and radius of curvature (RC) from CT-based volumetric models to predict AAA growth in a total of 192 patients who underwent elective repair of AAA. These geometric features were previously studied in the prediction of cerebral aneurysm rupture. In the current study, multinomial logistic and multiple linear regression models were trained, validated, and compared with predictions of AAA growth in a 2-year median preoperative interval (average baseline diameter in the training and test cohorts of 53.6 and 54.7 mm, respectively). AAA diameter, UI, and RC showed significant correlation with AAA growth as individual parameters. Three-variant models achieved areas under the ROC curves (AUCs) of 0.80 and 0.79 for the prediction of slow (< 2.5 mm/year) and fast (> 5 mm/year) growth, respectively. The prediction for growth rate was within 2 mm in 87% of cases.