ABSTRACT
OBJECTIVE: The objective of our study was to investigate the diagnostic accuracy of MRI and helical CT for endoleak detection. SUBJECTS AND METHODS: Fifty-two patients underwent endovascular aneurysm repair with nitinol stent-grafts. Follow-up data sets included contrast-enhanced biphasic CT and MRI within 48 hr after the intervention; at 3, 6, and 12 months; and yearly thereafter. The endoleak size was categorized as < or = 3%, > 3% < or = 10%, > 10% < or = 30%, or > 30% of the maximum cross-sectional aneurysm area. A consensus interpretation of CT and MRI was defined as the standard of reference. RESULTS: Of 252 data sets, 141 showed evidence for endoleaks. The incidence of types I, II, and III endoleaks and complex endoleaks was 3.2%, 40.1%, 8.7%, and 4.0%, respectively. The sensitivity for endoleak detection was 92.9%, 44.0%, 34.8%, and 38.3% for MRI, biphasic CT, uniphasic arterial CT, and uniphasic late CT, respectively. The corresponding negative predictive values were 91.7%, 58.4%, 54.7%, and 56.1%, respectively. The overall accuracy of endoleak detection and correct sizing was 95.2%, 58.3%, 55.6%, and 57.1% for MRI, biphasic CT, uniphasic arterial CT, and uniphasic late CT, respectively. CONCLUSION: MRI is significantly superior to biphasic CT for endoleak detection and rating of endoleak size, followed by uniphasic late and uniphasic arterial CT scans. MRI shows a significant number of endoleaks in cases with negative CT findings and may help illuminate the phenomenon of endotension. Endoleak rates reported after endovascular aneurysm repair substantially depend on the imaging techniques used.
