Dr. Chuter is an Associate Professor in the Department of Surgery, Division of Vascular Surgery, University of California, San Francisco, CA.

Supported in part by grants from the Pacific Vascular Research Foundation.

Disclosures: Dr. Chuter has licensed patents to Cook, Inc. and Guidant Corp. Dr. Chuter has served as a paid consultant for Cook, Inc. (Bloomington, IN); Guidant Corp. (Indianapolis, IN); and W.L. Gore, Inc. (Flagstaff, AZ).

Stent-grafts have been used to treat a wide variety of arterial lesions, including traumatic disruption, dissection, fistula, and aneurysm. This article focuses on just one of them--abdominal aortic aneurysm (AAA)--yet there is still a wide variation in the technique of stent-graft insertion. Perhaps the most important determinant of operative technique is the type of device. Other variables include the patient's arterial anatomy, the mode of anesthesia, the type of imaging system, and the nature of the operating suite.

Device design

The success rates of stent-graft insertion have risen steadily over the past decade. ^1,2 Improvements in technique, patient selection, and preoperative planning have all played a role, but the main factor has been improved device performance.

Any delivery system can negotiate straight, wide iliac arteries, and any stent-graft can attach itself securely to long, healthy segments of infrarenal aorta. Unfortunately, elderly patients with large aneurysms seldom have such favorable arterial anatomy. Calcification, tortuosity, and close proximity of the aneurysm to vital arteries are common findings. Under these circumstances, the second- and third-generation devices available in Europe and Australia perform far better than the first-generation devices available in the United States. Adjunctive maneuvers used to be important as a means of extending the scope of early stent-grafts, but their role is diminishing as better delivery systems become available.

Anesthesia

There has been a trend away from general anesthesia toward regional or local anesthesia, as the operation has become shorter and less likely to end in open conversion.

General anesthetia may still be the preferred mode for a patient who cannot lie flat, cannot lie still, has had spinal surgery, or may require brachial access, especially when a long, complicated operation can be expected. Another minor advantage is the ability to suspend respiration during digital subtraction angiography.

Most endovascular aneurysm repairs are now performed under spinal or epidural anesthesia. One theoretical objection is the risk of epidural hematoma in patients who will be receiving large doses of heparin, and some anesthesiologists may refuse to proceed with the operation if they get a "bloody tap." An epidural catheter is usually removed in the recovery room, once coagulation has normalized, or the patient is kept overnight for postoperative analgesia. A minor disadvantage of this approach is the need for a urinary catheter.

Local anesthesia is feasible in many cases of bifurcated stent-graft insertion, ³ especially when using percutaneous technique. The effect on blood pressure is less than it is with general or regional anesthetic, which may be a particular advantage in the treatment of contained aneurysm rupture. However, local anesthetics have no effect on the ischemic leg pain that follows occlusion of the external iliac artery by a large delivery sheath, and they do not provide a large enough field of anesthesia for femorofemoral bypass in conjunction with aorto-uniiliac stent-graft insertion.

Arterial access

Femoral exposure

The standard approach to the femoral arteries is through a longitudinal incision, which allows easy exposure of the entire common femoral artery and its distal branches. While this may have advantages in operations for arterial occlusion, in endovascular operations it is best to stay away from the femoral bifurcation. The proximal femoral artery is usually softer, and is well away from collateral routes through the deep femoral artery to the superficial femoral artery and down the leg. An oblique incision at the level of the inguinal ligament avoids the groin crease and results in a lower incidence of wound infection and necrosis, especially in obese patients. ⁴

Femoral arteriotomy

First-generation delivery systems are large and blunt-ended. ^1,5,6 Such devices require a femoral arteriotomy. Once the artery has been opened, occlusive loops or clamps maintain hemostasis, leading to some degree of lower extremity ischemia. Instrumentation of the already opened artery also increases the risk of common femoral and external iliac arterial dissection. Most modern delivery systems ^2,7 have a smooth external profile that permits insertion into the punctured femoral artery by the Seldinger technique. A seal between the delivery system and the arterial wall at the puncture site maintains hemostasis. Flow through the femoral artery is not obstructed, unless the delivery system is unusually large or the artery is unusually small. Even when the primary delivery system is not amenable to direct insertion, the lower profile second- and third-generation devices ⁸ are small enough to be inserted through a sheath, in which case hemostastic sealing occurs between the sheath and the arterial wall and between the device and the sheath valve.

The brachiofemoral guidewire

The brachiofemoral wire used to be considered a useful adjunct to stent-graft insertion. ⁹ Traction on both ends of the wire generated tension and provided the necessary stiffness for delivery system insertion through tortuous iliac arteries. This role has been surplanted by the combination of a stiff guidewire, such as Lunderquist (Cook, Inc., Bloomington, IN), and a trackable delivery system, such as the Zenith (Cook, Inc.), or the Excluder (W.L. Gore, Inc., Flagstaff, AZ). These days, the risks, including cerebral embolism and laceration of the subclavian orifice, are thought to outweigh the benefits.

Percutaneous technique

Stent-graft delivery systems are large enough that most surgeons prefer to expose the femoral artery at the start of the operation and repair it at the end. Others (mainly radiologists and cardiologists) have employed the Perclose system (Abbott Laboratories, Redwood City, CA) to eliminate the need for femoral exposure. ¹⁰ Because the sutures are placed prior to stent-graft insertion, this method has been dubbed the Preclose system of percutaneous AAA repair. In theory, the sutures repair the arterial laceration in much the same way as partial surgical repair. In practice, they owe much of their efficacy to closure of the femoral sheath. Success with this approach is most likely when the delivery system is small and the femoral artery is large and noncalcified.

Challenging anatomy

Iliac tortuosity

However tortuous and redundant the iliac arteries have become by the time a patient presents for aneurysm repair, they were once straight, and they can almost always be induced to become straight again. The keys to successful device insertion are a stiff guidewire and a trackable delivery system with a smooth external profile and long gradient of stiffness. Paradoxically, the most flexible systems are not always the easiest to insert, because it is rarely possible to eliminate all the stiff components from the device. The result is an abrupt change in flexibility where stiff and flexible segments meet, causing the device to buckle. For example, the relatively flexible Ancure system (Guidant Corp., Indianapolis, IN) tracks poorly, while the equally flexible Excluder system tracks well. At the other end of the stiffness spectrum, the AneuRx system (Medtronic, Minneapolis, MN) tracks poorly, while the Zenith system (Cook) tracks well.

An additional sheath sometimes helps by straightening the artery and providing a smooth channel through the iliac arteries. Other adjunctive maneuvers include mobilization of the external iliac artery, digital pressure on the apex of an iliac loop, brachiofemoral guidewire insertion, and access to the proximal common iliac artery, either directly or through a conduit. These had a more important role in the days of homemade or first-generation industry-made systems, but they are seldom needed when using narrower, more trackable delivery systems.

In problematic cases, delivery system removal may be more difficult and dangerous than insertion. Sheath withdrawal exposes the inner aspect of the stent-graft to the inner core of the delivery system while allowing the iliac arteries to return to their original, more tortuous position. Any bulges in the delivery system may then catch the stent-graft, leading to distal migration. This has been a problem with the AneuRx system, for example, in which the tip of the delivery system is a blunt metal knob. The problem is compounded by the lack of secure, barb-enhanced proximal fixation. Some authors advocate bracing the AneuRx stent-graft during delivery system removal using a sheath from the contralateral side. ¹¹

While iliac straightening is often a prerequisite for successful device insertion, the resulting change in position can complicate accurate deployment. Angiograms performed before device insertion may not provide an accurate guide to arterial position once a stiff delivery system is in place. Angiographic localization of the common and internal iliac arteries should be repeated immediately before deployment of the graft limb. Preoperative angiograms are useful as a guide to the most informative intraoperative view. In cases of iliac tortuosity, ipsilateral obliques are required to show the common iliac artery, while contralateral obliques show the internal iliac artery. The best compromise is often a simple anteroposterior (AP) projection, in which the position of the iliac bifurcation is indicated more by caliber change than by identification of the internal iliac orifice.

The orientation of the common iliac artery should be borne in mind when orienting a modular stent-graft. The proximal common iliac artery frequently directs a catheter toward the front of the aneurysm, in which case placing the contralateral limb of the primary stent-graft in a more anterior location may facilitate catheterization.

Other complications of iliac tortuosity include kinking and limb thrombosis. Once the delivery systems and guidewires are out, the iliac arteries tend to return to their former tortuous state. Many fully stented devices have enough stiffness and enough support to resist kinking. Additional support, in the form of a Wallstent (Boston Scientific, Natick, MA), is required when the original prosthesis is largely unstented (Ancure) or when the degree of tortuosity overwhelms the supportive potential of the original stent skeleton. ¹² The best time to make this determination and add the necessary stents is at the original operation. Some form of completion assessment, either angiography or intravascular ultrasound (IVUS), must follow removal of all other sources of support, such as stiff guidewires or elements of the delivery system.

Iliac aneurysm

If the common iliac artery is too large or too irregular for secure, hemostatic im-plantation of the graft limb, the prosthesis has to extend to a more distal implantation site in the external iliac artery. The size threshold for common iliac implantation depends on the maximum size of the graft limb. This has been a major limitation in the application of the first-generation systems, which had a limited range of limb diameters. With the AneuRx system, the range can be extended by the addition of an aortic cuff, as a "bell-bottom" extender. ¹³ The Excluder system also suffers from this limitation. Other stent-grafts have wide iliac limbs as part of the primary system. One example is the Zenith system, with limb diameters up to 24 mm, and the Talent system (Medtronic), with limb diameters up to 22 mm.

A rich network of cross-pelvic collaterals generally allows one internal iliac artery to be isolated from flow without serious ill effects. Buttock claudication is common, but self-limited. These collaterals are also a potential source of flow back into the iliac and aortic aneurysms. If the common iliac orifice is nondilated, the stent-graft will obliterate the connection between the two aneurysms. In the absence of a route of egress, there will be no flow through the common iliac artery and little potential for continued perfusion, pressurization, or rupture. When the neck of the common iliac artery is wide, the trunk of the internal should be occluded using large coils. Smaller coils and more distal embolization should be avoided, because they block intrapelvic collaterals and produce more severe buttock claudication. ¹⁴

When both common iliac arteries are too large for stent-graft implantation, the reconstruction must be more complicated. We favor the use of a stent-graft with an iliac bifurcation and outflow to both internal and external branches. ¹⁵ Others have described techniques of external to internal iliac bypass, either by endovascular ¹⁶ or by surgical ¹⁷ means.

Unfavorable neck

Conventional endovascular technique dictates that the proximal margin of the stent-graft must end below the renal arteries. Successful repair depends on the presence of a nondilated segment of aorta (the neck) between the renal arteries and the aneurysm. As always, device characteristics determine the acceptable limits of anatomic distortion, but in general, endovascular repair is more likely to fail when the neck is short, wide, angulated, irregular, or thrombus-lined. ^18-21 In the short term, an unfavorable neck impairs sealing, leading to a type I endoleak, aneurysm pressurization, and rupture. In the long term, an unfavorable neck impairs attachment, or fixation, leading to migration. The end result, a type I endoleak, is the same.

Type I proximal endoleak is easier to prevent than to remedy. The issue is more one of patient selection, rather than intraoperative technique. Nevertheless, there are intraoperative maneuvers that may be used to optimize device performance when faced with unfavorable anatomic features.

The potential ill effects of an angled neck are often underappreciated during patient selection; hence, the attempted treatment of patients with severe angulation. Retrospective analysis of the results shows important device-related differences. Devices that depend on "column strength" for proximal stent position lack the flexibility to accommodate neck angulation. The AneuRx is a good example. Devices with active proximal attachment, such as the Ancure or the Zenith (Flex), perform better, but only if they are implanted with enough redundancy for the outer margin of the graft to hug the outer margin of the neck. It is no good deploying the device in a fully extended, or even tensioned, state. In the case of the Zenith, one must perform the steps of deployment in a different order, with proximal stent deployment preceding deployment of the distal half of the main body. This imparts the necessary redundancy without impairing the accuracy of proximal stent placement.

In cases of neck angulation, it is particularly important to place the proximal end of the stent-graft as close as possible to the renal arteries. The longer the effective implantation site, the more likely it is that the stent-graft will align itself with the long axis neck, rather than with the long axis of the aneurysm. Only when the two are co-axial will there be a seal. Of course, neck angulation itself can impede accurate stent deployment. In the presence of AP angulation, the neck is seen best in a craniocaudal view. In addition, neck angulation can make it difficult to predict the exact position of the stent-graft.

The most common cause of a type I endoleak is a disparity between the orientation of the stent-graft and the neck. The inflation of a large balloon will always force the two into alignment. Sometimes they remain coaxial after balloon deflation; sometimes the stent-graft rotates out of position again as the balloon deflates, in which case additional fixation is needed in the form of a large Palmaz stent (P40-10, Cordis Corp., Miami Lakes, FL). These maneuvers should be accomplished at the time of stent-graft implantation, for two reasons. First, a proximal type I endoleak has been known to cause aneurysm rupture in the early postoperative period. Second, large angioplasty (or cardiac valvuloplasty) balloons require a 12F to 14F sheath. Percutaneous access through a recent operative site is unsafe, and reopening the groin incision increases the risk of wound infection.

When the stent is longer than the neck, as it often is, one has to choose between stenting the suprarenal aorta or stenting the portion of the stent-graft that lies within the aneurysm. Suprarenal stenting is probably safer, because the degree of aortic angulation at the renal arteries is usually less than half the degree of angulation between the neck and the aneurysm. Moreover, the suprarenal stent has no potential for injuring the stent-graft, whereas a Palmaz stent in the proximal end of the aneurysm has a potentially deadly free distal margin. Pulsatile movement of the graft would have the unyielding Palmaz stent as its fulcrum. As the Vanguard experience demonstrated, metal edges in contact with a moving graft lead inevitably to fabric erosion. ²²

The large Palmaz stent also has a role as a remedy for type I leakage between the proximal end of the stent-graft and the bulgy, irregular neck. The extent of posterior bulging is best appreciated on three-dimensional computed tomography or angiography in a lateral view.

Grossly inaccurate stent-graft deployment is the most common and most treatable cause of a type I endoleak. The treatment is to insert a second prosthesis (aortic cuff), at a higher level, to extend the overall conduit, which increases the functional length of the neck and enhances the seal. The accuracy of proximal stent placement, and the associated rates of cuff use vary widely according to the type of device. Even experienced users have high rates with devices such as the AneuRx ²³ and Talent. ⁷

Unfortunately, proximal extension does not ensure secure fixation, because extension cuffs lack any means of attachment, other than friction. The cuff can migrate distally with the primary stent-graft, or it can stay in place and separate from the primary stent-graft. Both lead to catastrophic failure.

Stent-graft rescue

Most of this paper concerns the conduct of the primary stent-graft implantation, the goal being to minimize the chance of primary failure. Unfortunately, successful stent-graft implantation does not always translate into durable protection from rupture. Many of the early devices have shown high rates of late failure. The worst offenders, such as the Vanguard, ²² are no longer marketed. However, early generation devices, such as the AneuRx, are still being placed in large numbers in the United States where, until recently, there were few alternatives. Migration rates of 19% to 32% ^24­26 have been reported for AneuRx devices at 3-year follow-up. Such cases usually require re-intervention to eliminate endoleak and prevent rupture.

In general, the most common causes of late failure are migration, component separation, and fabric erosion. The usual approach has been to reline the leaking segment, and only the leaking segment. This may be successful in the short-term only to fail again in the long-term. For example, a short aortic cuff may abolish the endoleak in cases of AneuRx migration, but migration often recurs, because the new cuff is no more securely attached to the aorta than the original stent-graft, the connection between stent-grafts relies on friction alone, and the overlap zone between stent-grafts is relatively short. In the long-term, it is safer to use a tapered uni-iliac stent-graft with active, barb-mediated proximal attachment to exclude the original stent-graft entirely. Femoro-femoral bypass adds significantly to the duration of the procedure, but not to the potential for late-failure, which is low.

Conclusion

Both the technique and the technology of endovascular aneursym repair continue to evolve. As devices improve, adjunctive maneuvers become less important and a wider range of patients become candidates for repair. However, there will always be patients whose anatomy lies beyond the limits of device performance. The secret of success is to know where those limits lie and select patients accordingly. Meanwhile, device companies will be working to identify and eliminate the causes of failure in the long term. A third source of progress is the development of entirely new systems with fenestrations and branches for aneurysms that encroach on the origins of vital arteries. Examples include aneurysms of the aortic arch, ²⁷ aneurysms of the thoracoabdominal aorta, ²⁸ and aneurysms of both common iliac arteries. ¹⁵ These techniques are currently applicable in only a few specially selected cases, but there may come a time when they are applied more widely.