Three-dimensional ultrasound is a reliable alternative in endovascular aortic repair surveillance

Objective: Three-dimensional ultrasound (3D-US) has already demonstrated improved reproducibility with a high degree of agreement (intermodality variability), reproducibility (interoperator variability), and repeatability (intraoperator variability) compared with conventional two-dimensional ultrasound (2D-US) when estimating the maximum diameter of native abdominal aortic aneurysms (AAAs). The aim of the present study was, in a clinical, multicenter setting, to evaluate the accuracy of 3D-US with aneurysm model quantification software (3D-US abdominal aortic aneurysm [AAA] model) for endovascular aortic aneurysm repair (EVAR) sac diameter assessment vs that of computed tomography angiography (CTA) and 2D-US. Methods: A total of 182 patients who had undergone EVAR from April 2016 to December 2017 and were compliant with a standardized EVAR surveillance program were enrolled from five different vascular centers (Rigshospitalet, Copenhagen, Denmark; Catharina Ziekenhuis, Eindhoven, Netherlands; L'hospital de la Timone, Paris, France; Cleveland Clinic, Cleveland, Ohio; and The Christ Hospital, Cincinnati, Ohio) in four countries. All image acquisitions were performed at the local sites (ie, 2D-US, 3D-US, CTA). Only the 2D-US and CTA readings were performed both locally and centrally. All images were read centrally by the US and CTA core laboratory. Anonymized image data were read in a randomized and blinded manner. Results: The sample used to estimate the accuracy of the 3D-US AAA model and 2D-US included 164 patients and 177 patients, respectively. The Bland-Altman analysis revealed that the mean difference between CTA and 3D-US was −2.43 mm (95% confidence interval [CI], −5.20 to 0.14; P = .07) with a lower and upper limit of agreement of −8.9 mm (95% CI, −9.3 to −8.4) and 2.7 mm (95% CI, 2.3-3.2), respectively. For 2D-US and CTA, the mean difference was −3.62 mm (95% CI, −6.14 to −1.10; P = .002), with a lower and upper limit of agreement of −10.3 mm (95% CI, −10.8 to −9.8) and 2.5 mm (95% CI, 2-2.9), respectively. Conclusions: The 3D-US AAA model showed no significant difference compared with CTA for measuring the anteroposterior diameter, indicating less bias for 3D-US compared with 2D-US. Thus, 3D-US with AAA model software is a viable modality for anteroposterior diameter assessment for surveillance after EVAR.