Dr. Bromley is a senior interventional fellow at the Dotter Interventional Institute, Oregon Health Sciences University, Portland, OR.

The placement of an endoluminal graft under radiological guidance is the latest advance in the treatment of aneurysmal disease of the abdominal aorta. Endografting represents a logical extension of endoan-eurysmorrhaphy, which has been the accepted surgical therapy for abdominal aortic aneurysms since the 1960s. ¹ In this latter procedure, the aneurysm is exposed surgically and opened longitudinally after proximal and distal hemostatic control has been obtained by surgical clamps (figure 1A). A synthetic fabric tube is then sutured to nondilated segments proximally and distally from within the aneurysm, and the sac is closed over the prosthesis (figure 1B and C). By avoiding resection of the aneurysm sac, aneurysm surgery was simplified significantly and the complication rate improved. ¹

The technique of endografting strives to further this end by means of specially designed "endografts," which can be introduced into the aneurysm lumen from a remote and minimally invasive access site (typically the common femoral arteries). These devices are positioned and deployed with radiological guidance. Proximal and distal fixation are obtained in nonaneurysmal segments by sutureless endoluminal fixation devices. The aneurysm sac is excluded from the circulation by a procedure performed from small groin incisions with minimal blood loss, no exposure of the aneurysm, and no need for aortic cross clamping.

Since the first radiologically guided, endoluminal repair of an abdominal aortic aneurysm in a human by Parodi and Palmaz ² in 1990, there has been an explosion of academic and commercial activity in this arena. Far more than arterial occlusive disease, endografting of aneurysmal disease has propelled the development of the new field of endovascular therapy. Continued advancement in this field is best served by cooperative efforts between those with the advanced training and expertise in the various radiological, endoluminal, and open surgical techniques necessary for procedural success and good patient outcome. This article will discuss the historical development of endografting and the clinical status of current technology and will also touch on endografting's future directions.

In the beginning there was wire

As a remarkable testament to human ingenuity, endoluminal approaches to the treatment of aortic aneurysms can be found as far back as the mid-19th century. The English surgeon Moore, in an effort to stimulate thrombosis, inserted 75 feet of fine iron wire into a thoracic aortic aneurysm in 1864. Unfortunately, the patient died from sepsis a few days later. ³ Wire placement was first used to treat an aneurysm in the United States in 1866 by Ranoshoff. ⁴ In 1879, the Italian physicians Buressi and Corradi ³ inserted wire into an aneurysm and attempted to induce coagulation with an electric current. In the first half of the 20th century Colt, ⁵ in England, and Blakemore, ⁶ in the United States, were proponents of the "wiring" method. These techniques, sometimes using up to 900 feet of wire, ³ produced mixed results and remained controversial. In cases of fusiform infrarenal aneurysms, the intent appeared to be to "strengthen" the aneurysm while maintaining a flowing channel through the middle. ^3,4 In fact, complete thrombosis was felt to be undesirable since it not infrequently led to fatal gangrene of the lower extremities. Nonetheless, reports appeared sporadically until the 1970s (figure 2). ⁴

It was not until Dubost reported on the successful resection and in-situ bypass of an abdominal aortic aneurysm in Paris, on March 19, 1951, that the era of surgical cure of abdominal aortic aneurysms was truly born. ^7,8 The initial flurry of enthusiasm for this new procedure, however, was later tempered by the identification of serious late complications resulting from degeneration of the cadaveric human aortic allografts (homografts) used to perform the bypass. ⁸ The 1950s became a period of intense investigation to identify a suitable synthetic material that could be handled easily and cut to length in the operating room, sutured to a blood vessel, and trusted to withstand the stress of the circulation in the patient for many years. Many different materials were investigated, including parachute silk, Vinyon-N, Orlon, Dacron, and Teflon, with the latter two ultimately proving to be quite suitable for arterial grafting. ^8,9 The availability of commercially produced synthetic grafts allowed surgical abdominal aortic aneurysm repair to become more widely applied but did nothing to reduce the severity of the surgery. The first step in this direction came in the form of an in-situ bypass without resection, attributed to Creech. ¹ His modification of the original endoaneurysmorrhaphy, described by Matas in 1902, allowed the bypass graft to be placed within the aneurysm lumen and ended the need for lengthy and difficult resection of the aneurysm sac from the retroperitoneum. This technique has been the standard since that time.

From today's perspective, it seems a logical next step in the progression of this therapy to introduce a graft into the lumen of the aneurysm from a remote access site without the need for exposing and opening the aneurysm. This moment arrived in September of 1990 when Parodi and associates ² performed, in Argentina, the first successful transfemoral endoluminal grafting of an abdominal aortic aneurysm in a human. The crude but innovative device used in this procedure was born of a melding of synthetic graft with catheter, stent, and imaging technology pioneered by interventional radiologists.

In 1964, Dotter and Judkins introduced the concept of catheter-mediated minimally invasive therapy for arterial disease when they reported a new technique for treating arterial stenoses with dilating catheters. ¹⁰ Searching for a means to improve on the results of his new "transluminal catheter dilatation" procedure, Dotter conducted experiments using tubular endovascular prostheses in the femoral arteries of dogs. Although all of the devices made of impervious plastic tubing resulted in thrombosis, he had encouraging results with coilspring prostheses made of stainless steel wire (figure 3), and introduced the concept of percutaneously placed endarterial tube grafts in 1969. ¹¹ Little activity occurred in this area for nearly 15 years when both Dotter and colleagues ¹² in Oregon, and Cragg et al ¹³ in Minnesota, published simultaneous reports on nitinol-based coilspring grafts, again in a dog model. Nitinol is a special alloy of nickel and titanium that is highly kink-resistant and possesses the characteristic of thermal shape memory. Exploiting these properties, these investigators were able to pass nitinol wire through the lumen of a catheter and have the wire shift to a coiled-tube state in the arterial lumen by exposure to body temperature or by flushing the delivery catheter with heated saline (60° C). The ability to pass the device as a straightened form (in the catheter lumen) and have it revert to a coiled tube of predetermined diameter (in the arterial lumen) meant that the diameter of the artery treated was no longer restricted by the size of the introducer catheter, as was the case in 1969.

The following year, Maass and coworkers ¹⁴ in Switzerland reported on a series of experiments they had begun in the early 1980s using a similar spiral-spring vascular endoprosthesis made of stainless steel. These devices became known as stents in a somewhat abstruse reference to Charles Stent, an English dentist who in the 19th century devised a splint to stabilize skin grafts. ¹⁵ Numerous publications followed throughout the 1980s as a variety of self-expanding and balloon-expanded metal endoluminal prostheses were developed and utilized in the treatment of arterial occlusive disease. In 1984, percutaneously introduced coiled wire experienced a brief reincarnation as a potential therapy for abdominal aortic aneurysms. That year, Cragg and associates ¹⁶ published further work on their nitinol coil grafts. In an article entitled, "Percutaneous Arterial Grafting," they reported the first percutaneous treatment of an experimental abdominal aortic aneurysm with a stent composed of a tightly wound tubular coil of nitinol (figure 4). ¹⁶ It was not long before wire and fabric were united and the stent-graft was born.

The stent-graft concept

Who was truly the first to conceptualize a stent-graft for the treatment of an aneurysm may never be known. The archives of the U.S. Patent Office contain a document entitled, "Method For Performing Aneurysm Repair" that describes a device created by M. Hasan Choudhury, which was filed in 1977 and patented in 1979 (figure 5). ¹⁷ Choudhury's device was to be composed of a single tube of surgical graft material, such as Dacron. It would be introduced into the body in a collapsed form on a catheter placed via a peripheral artery, such as the common femoral artery, and maneuvered into position to span the normal arterial segments proximal and distal to the aneurysm. This would be performed under fluoroscopic guidance. The catheter would contain channels to inject radiographic contrast. A guidewire was not mentioned. At the proximal and distal ends, "convoluted expansion rings" containing a "plurality of radially spaced anchoring pins" provided a means for attachment. Choudhury felt that "once in place, hemodynamic pressure [would] assure a continued fluid tight seal between the graft and the healthy vessel wall."

Juan Parodi, meanwhile, was a resident in vascular surgery at the Cleveland Clinic under Alfred Humpheries and Edwin Beven. In a retrospective article published in 1997, Parodi recalled fondly "the expression on Ed Beven's face when I told him that I foresaw a day when patients with aneurysms could be treated under local anaesthesia in the outpatient department." ¹⁸ Parodi apparently conducted some experiments in 1979 on a tubular polyester device with expandable stainless steel wire rings at the ends, using dogs, but abandoned the project because of disappointing initial results. ¹⁸

In 1986, Balko and associates ¹⁹ of New York were apparently first to describe in a scientific paper a device having the characteristics of a stent-graft for the treatment of an abdominal aortic aneurysm. These investigators created artificial aneurysms in the abdominal aortas of sheep using large Dacron patches and then several weeks later inserted their prostheses. These were composed of a constrainable wire metal framework (resembling Gianturco Z-stents) with a polyurethane coating. Cessation of sac pulsation and absence of bleeding when the aneurysm wall was lacerated purposefully determined successful acute exclusion of the aneurysm. Long-term follow-up was not obtained. The devices were inserted from femoral artery cutdowns, but the lack of fluoroscopic equipment necessitated they be positioned by direct palpation of the aneurysm neck during laparotomy. In their discussion, the authors mentioned the potential for collaboration between angiographers and vascular surgeons in the possible clinical application of such technology. Ultimately, radiologically guided aortic stent-graft placements were reported in experimental animals with very encouraging results on long-term follow-up. ^20-22 At about the same time, Palmaz was reporting the results of his newly designed balloon-expandable stent. Parodi met Palmaz in 1988 and immediately realized that this stent was the component he needed to achieve anchoring and sealing of an endoluminal graft. ¹⁸ Their collaboration resulted in a series of 62 canine experiments in Buenos Aires and culminated in the implantation of a device in a patient on September 7, 1990. ^2,18 In another parallel development, Harrison Lazarus, a vascular surgeon from Utah, had begun working on an endograft for abdominal aortic aneurysms during the mid-1980s. ²³ His early concept would ultimately evolve into a Food and Drug Administration (FDA)-approved device (discussed in the following sections).

Device design

Early devices were constructed from a piece of straight surgical tube graft, like Dacron, and modified by the addition of some form of endovascular attachment system. Lazarus experimented with and patented a device having an attachment system composed only of staples, at the proximal end, that would be driven into the vessel wall at the attachment site by a balloon device built into the delivery system. ²⁴ Stent technology was progressing rapidly during this period, however, and soon all devices incorporated some form of stent into the attachment sites. Lazarus's attachment device evolved into a self-expanding metal system similar in design to Gianturco Z-stents but with large anchors derived from the initial staple system. Parodi's group initially constructed devices with large balloon expandable stents for proximal attachment. The surgical approach was to modify the graft so that it could be attached from within the vessel. Interestingly, neither of these prototypes had a distal attachment mechanism. In fact, the first three patients in the clinical series that Parodi reported on in 1991 lacked any form of distal attachment device. ² Meanwhile Gianturco's group was building Z-stent devices long enough to span the infrarenal aorta and then covering the stents with graft material (figure 6). ^20-22 These stents came with or without wire barbs. Three separate design dichotomies were apparent from the outset and have persisted into current devices. One involves the method of obtaining device attachment within the vessel--radial force producing friction versus positive attachment with hooks or barbs. Another is the issue of structural support over the length of the graft material with stents. Finally there is the issue of utilizing stents that are self-expanding or balloon-expanded. The superiority of one approach over the other has not yet been resolved. Furthermore, it quickly became apparent during the initial clinical application of this technology that few patients could be treated successfully with a straight tubular graft.

New devices, new challenges

By 1995, Parodi ²⁵ was reporting on a series of 50 patients in whom he had performed endograft procedures and was being referred to as the father of transfemoral abdominal aortic endografting. ²⁶ In Sydney, Australia, White et al ²⁷ were accumulating a similar quantity of patients. By February of 1993, the EVT device (Endovascular Therapeutics, Inc., Menlo Park, CA), derived from Lazarus's work, had already entered Phase I FDA clinical trials in the United States, and Moore ²⁶ from UCLA was reporting on the first 10 patients in 1994. Although Parodi was able to implant a significant number of straight tubular devices in Argentina, the experience in the United States indicated that <15% of patients with abdominal aortic aneurysms would be suitable for such grafts. ^26,28 Parodi's approach to patients without a suitable distal neck was to extend a tapered aortomonoiliac device into one iliac, embolize the contralateral side, and perform a femoral-to-femoral bypass. ²⁵ The Sydney team also utilized this technique. ²⁹ The impetus for a completely endovascular solution was strong, however, and as early as 1993, Chuter ³⁰ had constructed a bifurcated device and delivery system and used it successfully in dogs. He then made the first report of successful use of his device in a human in February 1994. ³¹ This was followed, in September, by a report of 5 patients successfully treated at the University Hospital in Nottingham, United Kingdom, using Chuter's bifurcated device. ³² The Sydney group published the preliminary results of their bifurcated device in 1994 as well, ²⁷ and EVT introduced a bifurcated device late that same year. ²³ Interestingly, the 5 patients in the Nottingham series represented only 17% of the 29 consecutive patients screened for possible treatment. ³²

As experience was gained, a problem list of challenges to be overcome in the development of new devices was developing. For all devices, the femoral and iliac arteries represented potential obstacles to the introduction of the bulky delivery systems either from atherosclerotic stenoses or from severe tortuosity. The latter could sometimes be overcome by stiff guidewires or by surgical traction on the arteries. Angioplasty of focal atherosclerotic lesions could be performed, but was of little help in diffuse disease or in patients with intrinsically small arteries (often the case in women). Parodi ²⁵ developed the approach of anastomosing a graft to the iliac arteries above the restrictive segment to provide a conduit for device delivery, but this approach is somewhat contrary to the minimally invasive theme of the endovascular paradigm. In 1994, the sheath used to introduce the EVT device was 28F and the Sydney device was 24F. The devices themselves needed to become smaller and more flexible.

According to the Sydney group, in general, patients requiring bifurcated grafts seemed to have larger and more complex aneurysms than those suitable for straight-tube grafts. ²⁷ Early in their experience, they deployed both limbs of a bifurcated device into the ipsilateral iliac artery on two separate occasions, necessitating conversion to open procedures in both patients. ²⁷ Although the details behind the "technical errors" are not discussed in their report, these mishaps highlight the need for accurate measurements of parameters never before considered in open surgery. Custom fitting of devices to individual patients required detailed preoperative analysis of computed tomography (CT) and angiographic data. Devices made too short would need to be rescued by open surgery or modular extension components, and devices made too long could result in confinement of the contralateral graft limb (as described above) or could cover and occlude internal iliac arteries.

Continued perfusion or late "reperfusion" of the aneurysm sac were also recognized quickly and led to the characterization of a new entity, the endoleak. This is defined as the persistence of blood flow outside the lumen of the endoluminal graft but within the aneurysm sac. ³³ A variety of endoleaks can develop and have been classified by a fairly uniformly accepted scheme (Table 1). However, continued expansion of aneurysmal sacs after endografting in the absence of identifiable endoleaks has led to the development of an additional concept, endotension. ^34,35 Whether this is a separate notion or an additional form of endoleak has led to some controversy over its definition and the endoleak classification system. ^34,35 Although most authorities seem to agree that Type I and Type III endoleaks should be addressed by further endovascular procedures or conversion to open surgery, the question of what to do with branch perfusion leaks (Type II endoleaks) remains controversial. ^34-36 The identification of hemodynamic pressures in these reperfusion circuits, comparable to those in the systemic circulation, has raised the level of concern, particularly among radiologists.

By early 1995, investigators began reporting the first instances of device deterioration. Hook fractures, and later frame breaks, were identified in follow-up of the EVT device resulting in a halt to enrollment in the FDA trial. ²³ The attachment system and hook configuration were redesigned, and the second generation of EVT grafts was subsequently introduced at the end of 1995. ²³ No new fractures have been reported since the design change, but other devices have developed similar metal fatigue problems, as well as disintegration of sutures and fabric erosions. ³⁷ Device migrations have developed particularly in short angulated necks as well as late disconnections of modular components. ^37,38 Two cases have been described in which patient positioning during subsequent (and unrelated) surgical procedures may have contributed to separation of modular components. ³⁹ Furthermore, longitudinal shrinkage of the aneurysm after successful exclusion may contribute to detrimental structural distortion of the grafts. ³⁸

Along a similar line, studies have demonstrated continued dilation of the infrarenal neck in some patients following endograft placement, raising the concern that Type I endoleaks could develop, possibly years after a graft has been implanted. ⁴⁰ Evidence has also appeared that little tissue incorporation occurs at the attachment site with these devices in humans, despite evidence to the contrary in experimental animals. ⁴¹ Device migrations and dilating necks have fostered the use of positive fixation devices and suprarenal fixation elements in newer generation devices. Their value, however, remains theoretical until long-term follow-up data on their use becomes available. So it appears that, in the foreseeable future, patients implanted with abdominal aortic endografts will be subjected to a lifetime of close surveillance.

Another unique set of challenges that has arisen out of this evolving field includes the questions of who should perform these procedures and where they should be performed. Since the size of the delivery system required to implant an aortic endograft necessitates surgical exposure at the access site, surgical skills have been obligatory from the onset and remain so. The potential for completely percutaneous systems, however, is on the horizon (see below) and will no doubt significantly alter the playing field in the future. Although some vascular surgeons choose to ignore the unique skills and experience of their colleagues in interventional radiology and set out to rediscover basic catheter and wire techniques on their own, ⁴² this cannot be of any benefit to the patient or the field of endovascular therapy. ^43,44 An interventional radiologist would not be surprised to hear that surgeons implanting the large and stiff early EVT device in a tortuous aneurysm without an over-the-wire technique deployed a device into the thrombus necessitating open repair. ⁴⁵ Collaborative teams of experts in vascular surgical and endoluminal techniques, such as the archetypal Miami Vascular Institute, are to be applauded. This cooperative approach has also been fostered at the Dotter Interventional Institute where we evaluate endograft candidates and perform endograft procedures in conjunction with the vascular surgeons (figure 7).

Current devices

Nearly 50 years ago, pioneers of aortic synthetic graft surgery employed a seamstress to construct a suitable fabric tube for a patient after the abdomen had been opened and measurements obtained in the operating room. ⁹ The graft was then autoclaved and sewn into place. Similarly, the first stent-grafts were handmade for each patient out of available materials, although modern cross-sectional imaging techniques obviated the need to open the patient's abdomen. As was the case with synthetic grafts, it was not long before dedicated endografts for the repair of abdominal aortic aneurysms were being manufactured commercially. In the United States, the marketing of these devices is subject to regulations enforced by the FDA. Endografts are included in the regulations for Class III devices and are subject to a premarket approval system that requires clinical testing to establish safety and efficacy prior to marketing. Two bifurcated devices have completed this process and are marketed in the United States for the treatment of abdominal aortic aneurysms. These are the Ancure endografting system (previously the EVT graft; Guidant Cardiac and Medical Division, Menlo Park, CA) and the Aneurx stent-graft (Medtronic Inc., Minneapolis, MN). Both of these devices received FDA approval in September 1999. There are at least nine other devices currently in various stages of premarket development. Table 2 summarizes many of the features of these devices.

Despite that fact that both contain self-expanding stent components, the Ancure (figure 8) and Aneurx (figure 9) devices are products of very different design philosophies. The unibody design of Ancure mimics that of a standard surgical graft, while the Aneurx design is modular. The modular design permits greater versatility in matching a patient's anatomy, as the device is constructed inside the patient, but the junctions between individual parts provide potential places for early and late graft failures (see the discussion of endoleaks above). The stent skeleton provides support against compression over the entire length of the Aneurx device but reduces its mechanical flexibility. Conversely, although the Ancure device boasts greater flexibility, particularly in angulated necks, the unsupported iliac limbs have been subject to a greater number of thromboses. This has been improved through the liberal use of Wallstents (Boston Scientific, Boston, MA) by many interventionalists. ² Finally, the self-expanding stents at the ends of the Ancure device provide a means to initiate deployment of the anchoring pins, but otherwise do little to hold the device in place. Once the graft is released from the delivery catheter and the pins have contacted the wall, they are driven into place by the operator with a balloon. The pins provide the primary means of securing the graft. In contrast, the Aneurx device relies entirely on the force of friction between the self-expanding stents and the vessel wall.

There is only one other unibody design among the devices listed in Table 2. The PowerLink System (Endologix Inc., Irvine, CA) is a single-piece bifurcated graft created from polytetrafluoroethylene (PTFE) with an endoskeleton of self-expanding stainless steel wire (figure 10). The main body stent is composed of a single wire, which is purported to eliminate any potential for stent separation. The company states that the long body, reaching to the aortic bifurcation, will reduce the chances of late migration. Cuffs are available to extend the device proximally and distally to accommodate challenging anatomy. Experience with the device is limited in the United States, but Phase II FDA Trials have been ongoing since July 2000. The device is marketed in Europe and South America.

All of the other devices in Table 2 are modular designs and are fully supported by stents or stent-like elements. The Excluder (W.L. Gore and Associates, Flagstaff, AZ), Talent (Medtronic World Medical, Sunrise, FL), and Zenith (Cook Inc., Bloomington, IN) have all completed Phase II FDA trials in the United States. The aortic segment of the Talent device is available in diameters up to 36 mm. This is the largest diameter of any of the devices. The Zenith device derives a significant amount of its design from the extensive experience developed in Perth, Australia. It is notable for the high suprarenal attachment combined with positive fixation elements present on all versions of the devices (figure 11). The Zenith, Talent, and Ancure devices are also available in aortomonoiliac configurations. The Lifepath AAA graft system (Baxter Edwards Lifesciences LLC, Irvine, CA) is the only device on the list to rely on balloon-expandable stent elements (which the company refers to as "wireforms") that are present in the upper segment of the bifurcate body and throughout the other modular components. Only the limbs of the bifurcate body have self-expanding stents. The wireforms are also woven into the graft fabric to remove the need for sutures. Wireform fractures identified during long-term follow-up led to a halt of Phase II trials with this device in April 2000. These fatigue issues have been addressed, and the company anticipates restarting clinical trials before 2002. The Excluder also avoids sutures and suture holes by attaching the nitinol to the PTFE with polyethylene tape. ⁴⁶ Of the devices that have completed Phase II trials, the Excluder currently has the lowest profile at 18F.

What does the future hold?

In 1999, the Health Services Research and Development Service of the Veterans Health Administration's Office of Research and Development released a 22-page report assessing the effectiveness of endovascularly placed grafts for the repair of abdominal aortic aneurysms. ⁴⁷ This group reviewed the data available up to May 1998 to address issues such as the establishment of VA referral centers for minimally invasive vascular surgery procedures. Specifically, they wished to address the lower morbidity, mortality, and/or healthcare costs offered by endovascular treatments. Their report concluded the "currently available literature represents studies that are methodologically inadequate to answer the questions addressed in this report." The only proven benefits of endoluminal grafting for abdominal aortic aneurysms, to date, are less blood loss and a shorter hospital stay. ⁴⁷ The risk of aneurysm rupture has not yet been eliminated since late ruptures have occurred during the follow-up of both of the FDA-approved devices. ^23,48 While efforts will undoubtedly continue to make technical advances like fully percutaneous abdominal aortic endografting a reality (Figure 12) and to extend the range of patients who can be treated, carefully controlled trials are needed to prove this technology alongside the gold-standard open procedure. Attention must also be directed at better defining the need for "sac management" since it is becoming clear that, in many cases, the endograft itself is insufficient to exclude the aneurysmal sac from the systemic circulation (see the discussion of endoleaks in the previous section).

Conclusion

Techniques for the treatment of abdominal aortic aneurysms have evolved dramatically from the first primitive attempts to coil the lumen with wire in the 19th century. The mid-20th century saw the historic first open resection and in-situ grafting of an abdominal aortic aneurysm. Now, at the beginning of the 21st century, transfemoral endoluminal grafting has become a reality. Although the technology for endoluminal repair has developed through the contributions of both vascular surgeons and interventional radiologists, these groups have often worked in isolation. Since the shared expertise of these groups is greater than the sum of their separate parts, the evolution of the field of endovascular therapy and the future treatment of patients will be best served by cooperation. The first decade of the 21st century provides an exciting opportunity for these two specialties to prove the benefits of this promising technique. *