Late Endograft Failure
Improvements in preclinical testing will help reduce the risk of failure.
To view the figures that correspond to this article, please refer to the print version of our November/December issue, page 51.
Although Juan Carlos Parodi, MD, attempted to test endovascular treatment of abdominal aortic aneurysm (AAA) experimentally in the laboratory prior to clinically applying it, his lack of an animal model analogous to the human condition limited what he could learn from the simulation. Commercial device manufacturers began developing stent grafts soon after Dr. Parodi demonstrated clinical feasibility of the procedure. They faced important limitations, however, in their ability to predict stent graft function in humans by in vitro and laboratory in vivo modeling. These limitations included the lack of observed data on the physical forces within the aortoiliac vessels, the absence of naturally occurring animal aneurysm disease in a suitably sized vessel, and economic constraints that curtailed the development of complex simulations for preclinical testing. The investigational trials of newly developed endovascular stent graft (ESG) types therefore commenced with design efforts largely based on theory rather than observed data. Preclinical testing techniques may not have been relevant to actual clinical conditions, and researchers observed no biological response to the new treatment method. Considering this background as we enter the second decade of endovascular aneurysm repair (EVAR), it is not surprising that a range of device failure problems occur with variable frequency, and have affected all endografts to date. This article summarizes ESG device failures across many types, and discusses implications for future development.
AAA ENDOGRAFTING: THE BEGINNING
The first report of clinical endograft success pertained to an aortoaortic (“straight”) graft that used a single proximal anchoring stent. Follow-up observation showed blood flow reflux into the aorta above the distal extent of the endograft. Although this design appeared to adequately treat some patients, researchers adopted a standard modification for future endografting to include a second stent placed at the distal end of the fabric tube. This clinically learned lesson continues to influence ESG design improvement today. As Dr. Parodi applied it, endograft development was simple and direct. The design process has become much more involved now due to factors such as large company structures, teams of engineers, multicenter trial data, and the requirements and control of regulatory agencies that add complexity, cost, and time to the endeavor.
Dr. Parodi’s original report also hit upon a major factor influencing late results of ESG. “The size of the excluded AAA is considered to have decreased in three patients,” he wrote.1 This observation was surprising then and is still poorly understood today. Some excluded AAAs will shrink so much that they can virtually disappear, while others remain unchanged despite any evidence of endoleak. Other AAAs become angulated, shortened, or enlarged in sections. There is currently no way to predict this morphological change, which can exert forces to alter the configuration of endografts and contribute to device failure. The highly variable change in the morphology of the excluded aortoiliac arterial segment has proven to exert powerful effects upon the stent graft itself, however.
REASONS FOR DEVICE FAILURE
Patient selection is an important element in the complex equation of successful AAA endografting. The original anatomical features of the individual patient often influence the late results of ESG. The durability of ESG sealing and attachment may be compromised by accepting candidates with marginal anatomical features. For example, consider the upper attachment zone (also known as the proximal neck). Physicians who select patients whose proximal neck is so large that the endograft has minimal contact with the arterial wall, or insert endografts into a proximal neck that contains significant thrombus, may achieve only a temporarily adequate result. In contrast, a very short, angled proximal neck is more likely to produce type I endoleaks. The somewhat confusing literature describing proximal neck enlargement following ESG with self-expanding stents may include diverse patient genotypes, some of which could continue to enlarge in the course of developing a true juxtarenal aneurysm after the time of treatment.
Change in AAA Diameter
Early reports from May et al at the University of Sydney described what they saw in follow-up CT scans of patients 6 months or more following EVAR.2 They identified two groups: (1) 23 patients whose AAA diameter shrank a mean of 9 mm in 1 year and who showed no evidence of “contrast extravasation” (the Sydney group had not yet coined the term “endoleak”); and (2) four patients whose AAA sac increased significantly in diameter with evidence of endoleak. Additionally, and of continuing interest, there were three patients in this report whose AAA increased in size without evidence of contrast extravasation. It is unknown whether this observation was the result of an imaging that failed to show blood flow in the sac, or if the finding was due to pressure transmission through the thrombus. This report is important when considering late failures because it discounts the simplistic notion that clinical success can be evaluated by knowing only whether an endoleak is present. Many other investigators soon confirmed this line of thinking, and also indicated that AAA sac changes after exclusion were poorly understood.3-5
Interventionalists use diverse methods to measure morphologic changes in excluded vessels and endografts. Ordinary axial CT scans, duplex ultrasound, and arteriography, either singly or in combination, however, cannot accurately describe changes in angle, volume, and length of the relevant vascular structures. Two-dimensional imaging methods are inadequate to provide information describing three-dimensional objects.6 Conflicting reports about changes in aortic length following endografting illustrates this issue. One well-recognized European center concluded that, “… longitudinal shrinkage of the sac following [EVAR] led to buckling or kinking of the endograft within 1 year in 69% of patients. This appears to be an important source of delayed complications.”7 At the same time, another well-regarded European center possessing considerable aortic endograft experience concluded that, “… in this group of shrinking aneurysms after AAA repair, foreshortening of the excluded aortic segment appears not to be a clinically significant problem.”8 The apparent contradiction most likely stems from the problem of applying two-dimensional tools to three-dimensional objects, ie, the aorta and the endograft lying within it. Therefore, we must advance the present morphology assessment methods in order to improve the process of data acquisition and the assessment of graphic data (Figures 1 and 2).
ESG design variations have produced endografts that are stent-supported throughout their entire length, and others that are fixed by stents only at the proximal and distal attachment zones with unsupported fabric between them, much like conventional vascular prosthetic grafts. Although important differences exist in the design and location of the stents in the fully supported grafts, the stenting is intended to provide resistance to deformation and kinking. Endografts that are not fully stent-supported have greater flexibility that allows easier adaptability to tortuous and changing anatomy. Reports of the common need for adjunctive stenting to reduce external compression have revealed problems, however, with this approach.9 For example, one report of results with the Ancure device (Guidant Corporation, Indianapolis, IN) without stent support through the length of the endograft, noted that 46% of 88 devices inserted required adjunctive stenting for graft narrowing.10
Modular Component Separation
The separation of modular device components is a form of device failure that results in the rapid development of a large type III endoleak and possible AAA rupture. The problem of limb separation is specific to the modular endograft type in which components of the stent graft are assembled in situ (as opposed to other types that are one-piece in construction, although bifurcated in shape). Component separation has been observed with various manufacturers’ modular designs.11,12 Under nominal conditions after initial deployment of a modular stent graft, the frictional forces between its elements are sufficient to maintain a stable position. As the excluded aortoiliac anatomy changes shape in the process of shrinking, however, new and changing forces are applied to the endograft (Figures 3 and 4). ESG of different manufacturers vary in their column strength and resistance to bending forces. Stent grafts that are relatively flexible will bend in response to the changing aorta. Rigid grafts may accumulate tension until the yielding point gives way. In some cases, this point appears to be the junction between components of the endograft, as reported by Politz et al, and illustrated by their patient, whose endograft developed severe angulation 2 years postoperatively that was associated with limb separation, endoleak, and rupture.13 Zarins et al14 reported seven cases of late AAA rupture and commented on the contribution of operator error to the problem of limb separation. These investigators noted early postoperative radiographs showing that the contralateral limb had been inserted into the junction between the two components for an insufficient length, and was thus liable to separate more easily than if it had been positioned correctly.
Structural failure can occur independent of stress on the prosthesis that is induced by morphologic change in the vascular anatomy. Reports of three different types of endograft failure show how this occurs. An early one-piece design was introduced into clinical trials by Endovascular Technologies, Inc. (EVT) (Menlo Park, CA) and was available in both straight and bifurcated configurations. This endograft made unique use of penetrating hooks that were driven into the aortic wall at the proximal fixation point by balloon inflation and friction from a self-expanding stent. Soon after interventionalists began implanting this endograft in humans, the device exhibited problems with insecure proximal fixation associated with radiographic evidence of fractures of the anchoring hooks.
Jacobowitz et al, reporting the worldwide EVT experience, found three late ruptures among a total of 669 endografts, two of which showed hook fractures.15 Although hook fracture was not uniformly associated with adverse clinical consequences, the company suspended its clinical trial until they identified the cause and instituted manufacturing process changes to correct the problem. Matsumura and Moore reported 10 explants of early EVT grafts, eight of which demonstrated attachment system hook fracture.16 Most of these patients experienced migration with resultant endoleak that lead to planned conversion and then to open surgery; one patient experienced and survived AAA rupture. Brewster et al showed that a satisfactory course over a period of years with marked shrinking of the treated AAA is not sufficient to predict continued success.17 The researchers reported rupture 2.5 years following apparently successful EVT endografting. During the successful emergency surgery, they found a complete loss of distal fixation associated with hook fractures. The redesigned endograft was reintroduced into trials where it proved clinically successful and gained regulatory approval for commercial sale.18
The Stentor prosthesis (formerly MinTec, the Bahamas) is another example of an early endograft type that showed mechanical failure in clinical use. This graft employed a fabric tube that was constructed with a lengthwise suture line in its polyester fabric and supported by underlying nitinol stents. Although there were other late problems induced by morphologic changes, one complication that appeared to have a direct mechanical cause was rupture of the long suture line.19,20 Mechanical failure was cited for causing AAA rupture with associated aortoenteric fistula 17 months after the apparently successful functioning of an early Stentor endograft. The investigators concluded that, “[s]uture disruption between the internal support stents is a recognized complication in the first-generation Stentor device.”21 Schunn et al assessed the device’s long-term safety and efficacy in 190 patients, most of whom were treated with early-generation stent grafts.22 After observing changes in the endoprosthesis which they interpreted as suggestive of endograft disintegration in 30% of their patients treated with three different devices, the researchers concluded that, “… technical improvements in stent materials and design are necessary to guarantee long-term stability and safety of the device.”
Other Structural Problems
Even more recent designs of aortic endograft show evidence of mechanical failure. The environment inside the aorta is physically harsh, in terms of both mechanical forces and the corrosive effect of ionized blood.23 The mechanical bench-testing apparatus most manufacturers use to meet required testing for regulatory approval applies relevant stress to endografts under study, and allows physicians to rapidly accumulate data on the simulated effects of millions of cardiac cycles delivered in a compressed time interval. These tests are not comparable to the in vivo situation for many reasons, however, primarily because they lack the variable complexity of biological systems and chaotic interactions that continue to make empirical observations valuable, no matter how appealing the theoretical prediction of results might be.
Endograft designs still in development aim to achieve smaller-diameter endograft delivery systems. The benefits of placing stent grafts via smaller catheters include avoiding groin incisions and femoral dissection, and facilitating safer passage of potentially traumatic instruments through access and other vessels. However, reducing stent graft dimensions in order to minimize delivery systems increases the theoretical concern over the strength of the endograft and its ability to offer durable protection from a life-threatening condition. Early experiences with conventional aortic prostheses during the 1960s and 1970s demonstrated that some device failures were directly related to problems with material strength.24-26 This lesson is relevant in the investigation of aortic endografts today.27
FOLLOW-UP TO PREVENT FAILURE
At the end of this first decade of endograft experience, complete device failure resulting in post-EVAR rupture continues to be a problem in a small number of patients. Virtually every report of late results recommends mandatory, careful lifelong follow-up for all endograft recipients. Anecdotally, many centers with 4 or 5 years of endografting experience find that a significant number of patients either complain about their schedule for detailed follow-up and required CT imaging—or simply fail to meet it. Therefore, follow-up data outside of well-controlled clinical trials may not be reliable. Late data, however, is growing more and more important as reports of device failures accumulate, and some experienced endovascular centers have concluded that durability of an endograft cannot be evaluated with less than 3 years of follow-up.20
In 1995, Lumsden et al described two patients who experienced rupture following an early attempt at endograft exclusion.28 One rupture, which resulted in death, was attributed to poor patient selection because the distal anchoring zone was inadequate. The other patient had a persistent endoleak because the device failed to seal distally; this patient survived an emergency surgery 2 weeks following the original, incomplete procedure. This early failure report shows the interplay between patient selection and device characteristics, both of which can contribute to failure. Recent reports on significant cohort sizes give some indication of the prevalence of postendograft rupture, but the majority of these reports still encompass a short observation period. This issue was emphasized in the report of the US Vanguard trial describing outcomes for 268 patients receiving ESG followed from 12 to 36 months.29 AAA rupture occurred in three patients, and one patient was successfully managed by emergency surgery. The study also detailed a variety of other types of device failures and their management. The most important conclusion from this report, however, was that a 12-month follow-up observation period is insufficient to completely reveal safety issues (Figures 5 and 6).
Among 243 patients who received a variety of endograft types at the University of Sydney between 1992 and 1998, seven out of 17 patients had ruptured AAAs that required late conversion to open repair (six of those seven patients survived).30 The most recent follow-up data for the Ancure device (Guidant Corporation) (including patients with the redesigned attachment system only) reveal no late ruptures among 268 patients followed for 2 years.31 In 1996, a large European database was established to track self-reported late outcomes of AAA endografts provided by 88 centers. A recent report on late rupture risk among 2,464 patients with a mean follow-up of approximately 12 months showed 14 patients with confirmed rupture.32 These ruptures occurred between 3 and 24 months postoperatively, with 57% occurring later than 12 months; these finding reconfirm the conclusion that short-term data do not adequately inform us about endograft safety.
LESSER DEGREES OF FAILURE
Not all endograft failures are the complete calamity represented by late rupture. An increasing number of reports document the relatively high number of lesser degrees of failure that require merely secondary endovascular procedures or late conversion to open surgical repair. In a French trial, Becquemin et al33 emphasized the need for secondary intervention among 73 patients by reporting that the primary success rate at 12 months (defined as AAA exclusion without secondary intervention) was only 74%. The Vanguard clinical trial investigators have reported the need for secondary endovascular intervention with successful outcome for a variety of late problems that included migration, limb separation, and thrombosis.29 Overall, these indications for secondary treatment of milder forms of functional device failure arose within 24 months among 27 (10%) of 268 patients treated with endografts in the Vanguard trial. All incidents of limb occlusion and migration, as well as three endograft migration cases, were successfully treated with secondary endovascular procedures. Two additional endograft migration cases were successfully converted to open surgical repair.
Ohki et al at Montefiore Medical Center summarized a 9-year ESG experience and expressed a note of caution regarding an incidence of complications that increased with longer follow-up.34 Using seven different types of ESG to treat 239 patients, they encountered late AAA rupture (n=2), type I endoleak (n=7), ESG stenosis/thrombosis (n=7), and modular separation with type III endoleak (n=1). Secondary intervention was required in 23 (10%) of cases. Thus, they concluded that EVAR is often not a definitive procedure.
IMPROVING PRECLINICAL TESTING
It is likely that improvements in ESG design will stem from careful analyses of the lessons learned from recent device failures. In the early days of EVAR, endograft design was based largely on theoretical considerations; current devices have design features that are at least partially based on observed data. The data on endograft performance derive from human experience, however, a fact that severely limits the ability of investigators to measure performance in a controlled manner. We still lack a suitable animal model for studying endografts, and therefore depend on in vitro preclinical testing using artificial model simulation of the physical forces of the vascular environment to predict performance in vivo (Figure 7).
This process might be considered analogous to the common practice of testing a new aircraft design in the controlled conditions of a wind tunnel. Wind tunnel testing is now so sophisticated, however, that computer simulations using data from such testing can replace the need for physical modeling. In vitro endograft testing is nowhere near that level of accuracy. Rather than approximating the actual conditions in which an endograft must function, preclinical testing has a much more limited goal. Testing seeks to subject the endograft to physical forces similar to those within the aorta, and at a reasonably greater stress level, in order to provoke structural failure. This process, however, has a theoretical and incomplete basis. Proof of this incompleteness has come, unfortunately, in the form of the structural problems we have observed in the first several generations of endograft implantation, after clinical use was well underway.
Clinicians, government regulators, and the medical device industry have recognized that preclinical testing has not been successful in predicting the various endograft failure modes. This recognition prompted an FDA-sponsored workshop on endovascular graft preclinical testing in August 2001 that hosted experienced endovascular clinicians, medical device industry professionals, testing laboratory experts, and government regulators.35 These experts agreed that current testing inadequately models characteristics including anatomical changes in the aorta over time, vessel tortuosity, vessel wall characteristics, wear and fatigue dynamics on components suspended in the aneurysm sac, and pulsatility. Overall, these professionals felt that most tests, including finite element analysis and tests designed to model the in vivo condition, have been unable to account for the multitude of changing variables that affect the long-term results of endovascular grafts. It would be ideal to develop a single, sophisticated test apparatus that would be uniform for all endografts. Such a development seems unlikely for now, given the diversity of opinion about which parameters are most important to simulate, and how little data exist to validate their relevance. Although the process of improving preclinical testing has begun, it will take significant time, effort, and resources to accomplish anything similar to the functional utility of wind tunnel testing.
MEETING THE CHALLENGE
The risk of late AAA rupture and the need for close follow-up and more frequent secondary intervention in patients treated with EVAR are costly burdens that induce patient anxiety and overextend resources. Although ESG offers intrinsic appeal as a less-invasive procedure, adverse events such as device failures continue to pose challenges. Achieving the full potential benefits of ESG for average-risk patients requires ongoing study by both manufacturers and clinical investigators.
Hugh G. Beebe, MD, is Adjunct Professor of Surgery at the Dartmouth-Hitchcock Medical Center and Director Emeritus at the Jobst Vascular Center in Toledo, Ohio. Dr. Beebe may be reached at (419) 471-2088; firstname.lastname@example.org.
1. Parodi JC, Palmaz JC, Barone HD. Transfemoral intraluminal graft implantation for abdominal aortic aneurysms. Ann Vasc Surg. 1991;5:491-499.
2. May J, White GH, Yu W, et al. A prospective study of changes in morphology and dimensions of abdominal aortic aneurysms following endoluminal repair: A preliminary report. J Endovasc Surg. 1995;2:343-347.
3. Malina M, Ivancev K, Chuter TA, et al. Changing aneurysmal morphology after endovascular grafting: Relation to leakage or persistent perfusion. J Endovasc Surg. 1997;4:23-30.
4. Broeders IA, Blankensteijn JD, Gvakharia A, et al. The efficacy of transfemoral endovascular aneurysm management: A study on size changes of the abdominal aorta during mid-term follow-up. Eur J Vasc Endovasc Surg. 1997;14:84-90.
5. Matsumura JS, Pearce WH, McCarthy WJ, et al, for the EndoVascular Technologies Investigators. Reduction in aortic aneurysm size: Early results after endovascular graft placement. J Vasc Surg. 1997;25:113-123.
6. Beebe HG. Imaging modalities for aortic endografting. J Endovasc Surg. 1997;4:111-123.
7. Harris P, Brennan J, Martin J, et al. Longitudinal aneurysm shrinkage following endovascular aortic aneurysm repair: A source of intermediate and late complications. J Endovasc Surg. 1999;6:11-16.
8. Wever JJ, Blankensteijn JD, Broeders IA, et al. Length measurements of the aorta after endovascular abdominal aortic aneurysm repair. Eur J Vasc Endovasc Surg. 1999;18:481-486.
9. Baum RA, Shetty SK, Carpenter JP, et al. Limb kinking in supported and unsupported abdominal aortic stent-grafts. J Vasc Interv Radiol. 2000;11:1165-1171.
10. Amesur NB, Zajko AB, Orons PD, et al. Endovascular treatment of iliac limb stenoses or occlusions in 31 patients treated with the ancure endograft. J Vasc Interv Radiol. 2000;11:421-428.
11. Maleux G, Rousseau H, Otal P, et al. Modular component separation and reperfusion of abdominal aortic aneurysm sac after endovascular repair of the abdominal aortic aneurysm: A case report. J Vasc Surg. 1998;28:349-352.
12. May J, White GH, Waugh R, et al. Life-table analysis of primary and assisted success following endoluminal repair of abdominal aortic aneurysms: The role of supplementary endovascular intervention in improving outcome. Eur J Vasc Endovasc Surg. 2000;19:648-655.
13. Politz JK, Newman VS, Stewart MT. Late abdominal aortic aneurysm rupture after AneuRx repair: A report of three cases. J Vasc Surg. 2000;31:599-606.
14. Zarins CK, White RA, Fogarty TJ. Aneurysm rupture after endovascular repair using the AneuRx stent graft. J Vasc Surg. 2000;31:960-970.
15. Jacobowitz GR, Lee AM, Riles TS. Immediate and late explantation of endovascular aortic grafts: The endovascular technologies experience. J Vasc Surg. 1999;29:309-316.
16. Matsumura JS, Moore WS, for the Endovascular Technologies Investigators. Clinical consequences of periprosthetic leak after endovascular repair of abdominal aortic aneurysm. J Vasc Surg. 1998;27:606-613.
17. Brewster DC, Geller SC, Kaufman JA, et al. Initial experience with endovascular aneurysm repair: Comparison of early results with outcome of conventional open repair. J Vasc Surg. 1998;27:992-1003.
18. Moore WS, Kashyap VS, Vescera CL, et al. Abdominal aortic aneurysm: A 6-year comparison of endovascular versus transabdominal repair. Ann Surg. 1999;230:298-306.
19. Alimi YS, Chakfe N, Rivoal E, et al. Rupture of an abdominal aortic aneurysm after endovascular graft placement and aneurysm size reduction. J Vasc Surg. 1998;28:178-183.
20. Umscheid T, Stelter WJ. Time-related alterations in shape, position, and structure of self-expanding, modular aortic stent-grafts: A 4-year single-center follow-up. J Endovasc Surg. 1999;6:17-32.
21. Norgren L, Jernby B, Engellau L. Aortoenteric fistula caused by a ruptured stent-graft: A case report. J Endovasc Surg. 1998;5:269-272.
22. 666Schunn CD, Krauss M, Heilberger P, Ritter W, et al. Aortic aneurysm size and graft behavior after endovascular stent-grafting: Clinical experiences and observations over 3 years. J Endovasc Ther. 2000;7:167-176.
23. Guidoin R, Marois Y, Douville Y, et al. First-generation aortic endografts: Analysis of explanted Stentor devices from the EUROSTAR Registry. J Endovasc Ther. 2000;7:105-122.
24. Wesolow A. Vascular prostheses performance assessment of vascular prosthesis failure. Artif Organs. 1986;10:437-438.
25. Pourdeyhimi B, Wagner D. On the correlation between the failure of vascular grafts and their structural and material properties: A critical analysis. J Biomed Mater Res. 1986;20:375-409.
26. Mohr LL, Smith LL. Polytetrafluoroethylene graft aneurysms: A report of five aneurysms. Arch Surg. 1980;115:1467-1470.
27. Nunn DB. Structural failure of Dacron arterial grafts. Semin Vasc Surg. 1999;12:83-91.
28. Lumsden AB, Allen RC, Chaikof EL, et al. Delayed rupture of aortic aneurysms following endovascular stent grafting. Am J Surg. 1995;170:174-178.
29. Beebe HG, Cronenwett JL, Katzen BT, et al, for the Vanguard Endograft Trial Investigators. Results of an aortic endograft trial: Impact of device failure beyond 12 months. J Vasc Surg. 2001;33:S55-S63.
30. May J, White GH, Harris JP. Early and late conversion from endoluminal to open repair. Semin Vasc Surg. 1999;12:207-214.
31. Moore WS. Two-year follow-up of patients with EVT/Guidant (Ancure) bifurcated graft for endovascular AAA repair: Advantages and disadvantages of this device. Lecture presented at: The 27th Montefiore Vascular Symposium; November, 2000; New York, NY.
32. Harris PL, Vallabhaneni SR, Desgranges P, et al. Incidence and risk factors of late rupture, conversion, and death after endovascular repair of infrarenal aortic aneurysms: The EUROSTAR experience. J Vasc Surg. 2000;32:739-749.
33. Becquemin J, Bourriez A, D'Audiffret A, et al. Mid-term results of endovascular versus open repair for abdominal aortic aneurysm in patients anatomically suitable for endovascular repair. Eur J Vasc Endovasc Surg. 2000;19:656-661.
34. Ohki T, Veith FJ, Shaw P, et al. Increasing incidence of midterm and long-term complications after endovascular graft repair of abdominal aortic aneurysms: A note of caution based on a 9-year experience. Ann Surg. 2001;234(3):323-334.
35. Abel DB, Beebe HG, Dedashtian M, et al. Preclinical testing for aortic endovascular grafts: Results of an FDA workshop. J Vasc Surg. 2001;35:1022-1028.