Purpose: To evaluate the predictability of endoleak.
Materials and methods: Thirteen women and 60 men (mean age, 69.8 years) underwent transfemoral insertion of endoluminal stent-grafts for treatment of aortic aneurysms. Follow-up included helical computed tomography (CT) at 3-month intervals. In the cases of endoleak, angiography also was performed to document the number of leak sites, their size and position, the feeding artery, the size of the aneurysm, the amount of thrombus, and the visualization of the lumbar arteries and inferior mesenteric artery. These data were correlated (Student t test) with the probability of endoleak.
Results: A total of seven (10%) endoleaks were identified at CT in 68 patients. The feeding vessels were lumbar arteries in three cases, the inferior mesenteric artery in three cases, and the median sacral artery in one case. Of all factors, only the number of lumbar arteries visualized preoperatively (P <.005) had a significant correlation with probability of endoleak. In 71% (five of seven patients) of the cases of lumbar endoleak, four lumbar arteries were patent, whereas among the 61 patients with successfully repaired aneurysm, only eight (13%) had four patent lumbar arteries. Endoleaks were never found in the primarily thrombosed sections of an aneurysm.
Conclusion: Prediction of endoleaks with absolute certainty remains elusive. The single correlating risk factor identified from the data was patency of four or more lumbar arteries visualized preoperatively at CT.