Portal de Periódicos da CAPES

Morphologic evaluation of ruptured and symptomatic abdominal aortic aneurysm by three-dimensional modeling

2014; Elsevier BV; Volume: 59; Issue: 4 Linguagem: Inglês

10.1016/j.jvs.2013.10.097

1097-6809

An Tang, Claude Kauffmann, Sophie Tremblay-Paquet, Stéphane Elkouri, Oren K. Steinmetz, Florence Morin-Roy, Laurie Cloutier-Gill, Gilles Soulez,

Vascular Procedures and Complications

ObjectiveTo identify geometric indices of abdominal aortic aneurysms (AAAs) on computed tomography that are associated with higher risk of rupture.MethodsThis retrospective case-control, institutional review board-approved study involved 63 cases with ruptured or symptomatic AAA and 94 controls with asymptomatic AAA. Three-dimensional models were generated from computed tomography segmentation and used for the calculation of 27 geometric indices. On the basis of the results of univariate analysis and multivariable sequential logistic regression analyses with a forward stepwise model selection based on likelihood ratios, a traditional model based on gender and maximal diameter (Dmax) was compared with a model that also incorporated geometric indices while adjusting for gender and Dmax. Receiver operating characteristic (ROC) curves were calculated for these two models to evaluate their classification accuracy.ResultsUnivariate analysis revealed that gender (P = .024), Dmax (P = .001), and 14 other geometric indices were associated with AAA rupture at P < .05. In the multivariable analysis, adjusting for gender and Dmax, the AAA with a higher bulge location (P = .020) and lower mean averaged area (P = .005) were associated with AAA rupture. With these two geometric indices, the area under the ROC curve showed an improvement from 0.67 (95% confidence interval, 0.58-0.77) to 0.75 (95% confidence interval, 0.67-0.83; P < .001). Our predictive model showed comparable sensitivity (64% vs 60%) and specificity (79% vs 77%) with current treatment criteria based on gender and diameter at the point optimizing the Youden index (sensitivity + specificity − 1) on the ROC curve.ConclusionsTwo geometric indices derived from AAA three-dimensional modeling were independently associated with AAA rupture. The addition of these indices in a predictive model based on current treatment criteria modestly improved the accuracy to detect aneurysm rupture. To identify geometric indices of abdominal aortic aneurysms (AAAs) on computed tomography that are associated with higher risk of rupture. This retrospective case-control, institutional review board-approved study involved 63 cases with ruptured or symptomatic AAA and 94 controls with asymptomatic AAA. Three-dimensional models were generated from computed tomography segmentation and used for the calculation of 27 geometric indices. On the basis of the results of univariate analysis and multivariable sequential logistic regression analyses with a forward stepwise model selection based on likelihood ratios, a traditional model based on gender and maximal diameter (Dmax) was compared with a model that also incorporated geometric indices while adjusting for gender and Dmax. Receiver operating characteristic (ROC) curves were calculated for these two models to evaluate their classification accuracy. Univariate analysis revealed that gender (P = .024), Dmax (P = .001), and 14 other geometric indices were associated with AAA rupture at P < .05. In the multivariable analysis, adjusting for gender and Dmax, the AAA with a higher bulge location (P = .020) and lower mean averaged area (P = .005) were associated with AAA rupture. With these two geometric indices, the area under the ROC curve showed an improvement from 0.67 (95% confidence interval, 0.58-0.77) to 0.75 (95% confidence interval, 0.67-0.83; P < .001). Our predictive model showed comparable sensitivity (64% vs 60%) and specificity (79% vs 77%) with current treatment criteria based on gender and diameter at the point optimizing the Youden index (sensitivity + specificity − 1) on the ROC curve. Two geometric indices derived from AAA three-dimensional modeling were independently associated with AAA rupture. The addition of these indices in a predictive model based on current treatment criteria modestly improved the accuracy to detect aneurysm rupture.