Introduction The treatment of abdominal aortic aneurysms (AAA) has changed significantly over the last several years since Parodi’s first description of an endoluminally-placed infrarenal aortic stent graft.1 The continued evolution and increasing prevalence of this technique has produced a dramatic change in the management of this patient population. Currently, almost 50% of all infrarenal AAA’s in the U.S. are treated with endovascular devices.2 There have been multiple endoluminal devices developed and approved for the treatment of AAAs, including the AnCure (Guidant Corporation, Menlo Park, Calif.), AneuRx (Medtronic, Santa Rosa, Calif.), Excluder (Gore Medical, San Francisco, Calif.), Zenith (Cook, Bloomington, Indiana), and Powerlink (Endologix, Inc., Irvine, Calif.) devices. The last four devices are currently available in the U.S. market. There are several well-documented advantages to endovascular aortic repair (EVAR);3–5 however, this technique is not without complications. Multiple short- and long-term complications, which can lead to aneurysm expansion and rupture, have now been recognized and include: endoleak, component separation, stent fracture, fabric erosion and graft migration. Patients require close long-term follow-up to ensure aneurysm stability or regression, and are typically examined with a CT scan at 1, 6, and 12 months and annually thereafter. Graft migration and proximal attachment graft failure are a result of the cumulative effects of pulsatile blood flow overcoming the stabilizing forces supplied by the aortic and iliac fixation sites (frictional and radial forces, columnar strength, suprarenal stent segment and stabilization barbs), and are associated with aneurysm enlargement and potential rupture. The etiologies of graft migration and proximal attachment failure are multifactorial in origin and have been demonstrated to occur following repair with all types of endovascular devices.6–12 The AneuRx graft, in the U.S. market since 1999, has received particular scrutiny in the literature regarding concern for proximal attachment failure and device migration. Investigators have defined device migration variably, making studies difficult to compare. Nevertheless, multiple studies have shown that AneuRx graft migration occurs relatively frequently. Conners et al defined graft migration as a change of 6 Cao et al have defined graft migration as a change of >/- 10 mm or more, and found migration in 15% of patients with a mean follow-up period of 28 months.7 More recently, Tonnessen et al analyzed device migration using a change of both 5 and 10 mm.8 Freedom from migration with the AneuRx device was 67.4% and 72% at 4 years, defining migration as 5 or 10 mm, respectively. We have recently reviewed our data and have shown similar results, with an overall device migration rate of 16.1% at 4 years, defining migration as a change in 10 mm or any change requiring therapy (SVS criteria).9 In addition, studies of the original AneuRx study cohort have shown a 7.8% incidence of migration at 4 years.13 Through these investigations, multiple risk factors have been identified, many of which can be avoided, decreasing the risk of AneuRx device migration to levels comparable to other approved devices. The Excluder graft has been in the U.S. market since 2002. Published data from the U.S. trial suggests that graft migration is an uncommon event, occurring in 1% of patients at 2-year follow-up.14 The Zenith graft, in the U.S. market since 2003, is also associated with low migration rates of 0 to 2.4%.8,15 There are limited data concerning patients treated with the Excluder and Zenith devices after approval to compare with the post-approval results of the AneuRx graft. The Powerlink device has been in the U.S. market since 2005, and the published results are limited, but the initial trial demonstrates a low early migration rate (2.2% at 2 years).16 This article will review risk factors, prevention and treatment for endovascular graft migration and proximal attachment failure. Prevention of Complications Proximal fixation and seal of an endoluminally-placed aortic stent graft is a critical component to successful aneurysm exclusion. Inadequate proximal fixation leads to an increased rate of device migration and endoleak, often requiring secondary repair. In recent reports, secondary repair was required in 32–85% of those with device migration, usually in the form of endoluminal repair.6–12 There are two major components important to the prevention of early and late complications of EVAR: appropriate patient and graft selection and accurate graft deployment. Patient and Graft Selection Appropriate patient selection is likely the most important factor in EVAR with any endovascular graft. The information for use (IFU) for the AneuRx stent graft was updated in October of 2002 from the original 1999 IFU. The original IFU suggested a proximal aortic neck of 10 mm in length, with diameters of 17–25 mm, allowing >/- 10% device oversizing. In addition, the distal attachments had to be 10–15 mm in diameter and 10 mm in length. Current recommendations suggest an aortic neck length of 15 mm with a maximum angulation of 45%. Oversizing of 10–20% is recommended. The iliac attachment should be >/- 25 mm in length. These new suggestions are more in line with the IFU of the other approved devices. Proximal fixation is of critical importance, as suggested by multiple series.6–12 There are a variety of neck characteristics that can potentially increase the risk of proximal graft failure. These include circumferential calcifications, extensive plaque/thrombosis in the neck and a reverse taper. None of these alone is a total contraindication for endovascular repair, but combinations of these factors and some of the others mentioned (angulation) will significantly increase the risk of proximal attachment failure and endovascular graft migration. In a recent review of our experience with the AneuRx graft, migration at 4 years occurred in 42% of patients that did not meet the current IFU criteria, versus a 4.5% rate of migration in patients that met the IFU criteria.9 In this study, a multivariant analysis was performed to examine potential factors that might be associated with endovascular graft migration. Neck angulation and endovascular graft distance from the lowest renal artery were the only factors that were significantly associated with graft migration. After careful patient selection, based on anatomical characteristics, the next critical decision is the size of the graft used. It is important to oversize the device 10–20% to have adequate proximal attachment and seal. Oversizing greater than 30–40% can lead to poor proximal seal due to kinking and uneven compression of the graft. This effect seems to be more pronounced with the Excluder graft since it has less radial force than other devices. Undersizing the graft, in particular attempting to treat borderline proximal necks (26–27 mm) with the AneuRx and Excluder devices (largest available devices are 28 mm and 28.5 mm, respectively), is likely to lead to poor apposition of the graft to the aortic wall and minimal radial force, leading to an unstable proximal fixation and seal. The selection of patients with suitable anatomy for endovascular treatment is paramount to having good short- and long-term results. Appropriate graft choice, based on the characteristics of the device, and appropriate sizing, based on the aneurysmal anatomy, are also critical factors. Accurate Graft Deployment In addition to appropriate patient and graft selection, accurate graft deployment is an essential component of successful EVAR. Several factors influence accurate deployment. These include maximizing the length of proximal and distal fixation, enhanced intra-operative imaging with intensifier angle correction, accurate device deployment, avoiding graft migration during secondary manipulations, as well as the use of aortic cuffs. The importance of maximizing the length of proximal fixation cannot be overstated. In a review of our experience, the mean distance from the orifice of the lower-most renal artery to the top of the AneuRx device was statistically significant between migrators and non-migrators at 6.5 mm and 2.8 mm, respectively.9 Similar results were found by Zarins et al,10 with a mean distance from lower-most renal artery to proximal graft of 11 mm and 8 mm in migrators and non-migrators, respectively. In addition, a migration rate of 19, 17, and 9% was demonstrated in patients with proximal fixation of less than 10, 10–15, or > 15 mm, respectively. The authors predict that each millimeter increase in length of proximal fixation decreases the likelihood of migration by 2.5%. Deploying the device as close to the renal arteries as possible increases the stability and seal of every endovascular device. The extent of distal fixation within the common iliac arteries has also been evaluated with respect to AneuRx graft migration.17 Heikkinen et al suggest that the stability of the AneuRx device increases and the migration rate decreases the closer the limbs are deployed to the hypogastric artery. Extensive overlap seems to increase the columnar support and stability of the AneuRx device, and it may also be the case with the other approved devices, since all these grafts are fully supported with a nitinol or stainless steel skeleton. Multiple intraoperative maneuvers can be performed to increase the accuracy of graft deployment just below the lowest renal artery. Partial deployment of the device with repeated injections to visualize the relationship of the renal arteries with the device being deployed is extremely helpful. This is an important part of the deployment technique of the AneuRx and Zenith devices as suggested in their IFU. In addition, for very accurate deployment, the relationship of the aortic neck and the renal arteries has to be optimally visualized. Most infrarenal aortic necks have some degree of anterior angulation as the aneurysm enlarges and the aorta elongates (Figure 1). For optimal visualization, the image intensifier needs to be at 90º to the infrarenal neck just below the lowest renal artery. An estimate of this angulation can be obtained from the preoperative evaluations and may assist in planning, but the anatomy will change intraoperatively as stiff wires and bulky devices are introduced through the patient’s arterial anatomy. The controlled deployment of the AneuRx and Zenith devices allow for precise device deployment. The proximal markers are readily visualized and serve as a guide to adjust the cranio-caudal angulation of the image intensifier until it is perfectly perpendicular to the device. After repeat imaging of the renal arteries, the device can be safely and accurately positioned just below the lowest renal artery to have the maximal proximal attachment and seal (Figure 2). The Excluder device has a less controlled deployment technique that some physicians have modified to approximate the deployment technique of the other two devices. With increased neck angulation, these maneuvers are even more critical to avoid short- and long-term complications. Once the main body has been deployed, migration during further manipulation must be avoided. The current delivery system of the AneuRx device allows removal of the system with significant ease, and the risk of distal migration of the graft at that time is minimal. In the rare circumstance that the delivery system is difficult to remove, a variety of “buttressing techniques” can be used, including deployment of the contralateral limb with iliac attachment to increase graft stabilization and columnar strength. During any manipulations within the endograft, care should be taken to avoid any downward forces on the graft, like those that can occur during introducer retrieval, balloon molding, and sometimes necessary manipulations over the graft bifurcation used for contralateral access. Lastly, if possible, one should avoid the use of aortic cuffs. This is more prevalent when an angled proximal neck is present. The use of multiple cuffs and components in the proximal neck often reflect the degree of difficulty of the anatomy and in addition is associated with an increased risk of type I and III endoleaks (Figure 3). They are associated with poor stability within each other, with relatively limited overlap in the best of circumstances, most commonly with the AneuRx and Excluder grafts, due to the relative short bodies of the bifurcated grafts and short aortic cuffs. Treatment Options Once a patient has been identified as having a graft which has proximal attachment failure with or without migration, one must decide what treatment is necessary. If limited migration has occurred, the patient can be observed if a type I or type III endoleak is not identified, the aneurysm is not increasing in size and there is good proximal attachment remaining. If patients require treatment for migration with or without an endoleak, most of them can be treated endoluminally, with open surgical conversion rarely necessary. Endoluminal treatment options include aortic cuffs (transrenal or infrarenal), bifurcated devices and aorto-uni-iliac grafts. Provided the patient has a suitable proximal neck to extend the fixation of the device, proximal aortic cuffs have been very successful initially. However, we have found that a significant number of secondary failures occur with cuff placement, either with or without suprarenal fixation in patients with unfavorable proximal necks. This is most commonly secondary to component separation with continued device migration (Figure 3).18 Other options to consider in these patients include placement of a new bifurcated graft, provided the distance from the lowest renal artery to the bifurcation of the old graft is > 70 mm. Another option is the use of an aorto-uni-iliac system that provides stability proximal to the device below the renal arteries and distal to the prior reconstitution in the iliac arteries or at least within one of the stable iliac limbs (Figure 4). A new bifurcated graft or aorto-uni-iliac devices are both attractive options, significantly decreasing the risk of component separation seen with proximal aortic cuffs in general. Conclusion Aortic neck attachment failure and proximal migration of endovascular grafts is a challenging problem, infrequently requiring open conversion. However, with proper patient selection, precise initial device placement, and appropriate use of available devices for secondary endovascular repair in patients with device migration, open surgical conversion should continue to be a rare necessity (Figure 5).