A Novel Method for Treatment of Abdominal Aortic Aneurysms Using Percutaneous Implantation of a Newly Designed Endovascular Device
Background Percutaneous implantation of a stent to bridge abdominal aortic aneurysm (AAA) may provide an alternative to surgical reconstruction in patients with this serious disorder. We developed a self-expandable, stainless steel, woven mesh endovascular device with a delivery catheter and studied its efficacy in a canine model of AAA.
Methods and Results Infrarenal AAAs were created surgically in eight adult dogs using autologous tissue. Two types of endovascular stents were used in this study: a plain or uncovered stent, about 14 mm in diameter in the unconstrained configuration, and a covered stent, coated by porous polyurethane, about 16 mm in diameter. All stents were successfully placed on the first attempt. Aortograms revealed a mean aneurysm diameter of 1.86±0.47 cm, an average of 70% larger than the reference aortic lumen before stent placement. After stent placement, aortograms showed that the aneurysmal cavity disappeared completely in three dogs treated with a covered stent and that the aortic blood flow into the cavity markedly reduced, with faint contrast filling the cavity in the remaining five dogs treated with an uncovered stent. The uncovered stent was intentionally placed across the major arterial branches in two dogs. No acute complications were encountered at the time of stent placement. Two dogs were killed shortly after the procedure for immediate evaluation of the device, which was found to be in place and patent. One dog in which a covered stent was placed was euthanized 2 weeks later because of paraplegia secondary to a spinal cord infarction noted 48 hours after stent placement. Postmortem study revealed thrombus occluding the stent lumen. The remaining five dogs tolerated the devices well and completed 4 weeks of follow-up. Premortem aortograms showed no residual aneurysmal cavity in four dogs and only a small cavity in one dog that had received an uncovered stent. All stents were fully patent with no thrombus and were either completely or partially surfaced by neointima. Importantly, the major arterial branches over which the uncovered stents were placed were widely patent without obstruction by neointima.
Conclusions This study demonstrates the feasibility of percutaneous implantation of this new device and its effectiveness in the treatment of surgically created AAA in our canine model. The covered stent was able to exclude AAA immediately upon deployment and is of potential value in the emergency treatment of leaking AAAs. The uncovered stent appears to safely bridge branch arteries as well as significantly reduce the angiographic size of the aneurysm and may be useful in the elective therapy of AAAs. These results are promising, and future clinical trials to investigate the safety and efficacy of this device in humans are warranted.
Abdominal aortic aneurysm (AAA) is a serious vascular disorder characterized by a permanent focal dilation of the abdominal aorta having a diameter of at least 50% greater than normal.1 This disease predominantly affects elderly men, and its incidence has substantially increased in individuals over age 55 years.2 3 4 5 6 Its true prevalence is unknown but is estimated to be between 3% and 6% in those over age 65 years in the general population.2 6 7 Approximately 100 000 AAAs are diagnosed each year,8 and it is currently the 13th leading cause of death in the United States.9 Rupture of the aneurysm carries a high mortality. Up to 80% of patients with ruptured AAA die before reaching the operating theater, and 50% of the remaining patients die during or shortly after emergency surgery.3 10 Elective surgery for AAA has a much lower mortality rate but still carries a 3% to 5% risk of death.2 3 Surgery is currently the only recognized effective treatment for AAA, and more than 40 000 aortic reconstructions have been performed annually in the United States in recent years.11 12 The demand for AAA repair is expected to increase as our population ages. A less invasive alternative to surgical AAA repair would be attractive, particularly if it could eliminate the risk of AAA rupture as well as the morbidity and mortality associated with surgical repair. It also might result in a significant cost saving in the medical care of these patients.
Interventional cardiovascular catheterization is a rapidly advancing field in which new techniques have been providing excellent results in the treatment of a number of cardiovascular diseases that formerly required surgical intervention. Endovascular stents are one of the advances in this field and have been effectively used to treat peripheral and coronary artery disease, pulmonary artery stenosis, and some types of congenital cardiac malformations.12 13
Recently, endovascular stents have been used experimentally in the treatment of aneurysms in animal models. Balko et al14 developed a polyurethane-coated expandable stent and tested it in AAA dog models. Later, Lawrence et al15 used a Dacron-wrapped Gianturco stent, Mirich et al16 tested a modified Gianturco stent covered with nylon, and Laborde et al17 tried a weft-knit Dacron tube with balloon-expandable stents in dogs for the same purpose. Although these devices could be percutaneously implanted to bridge and exclude an aneurysm, complications such as stent thrombosis and renal ischemia from placement across arterial branches occurred due to the design of the device.14 15 16 17
In an attempt to overcome these complications, we developed a self-expandable endovascular device (Schneider US Stent Division, Minneapolis, Minn) that can be implanted percutaneously and tested it by placing across surgically created AAA in dogs. The purpose of this study was to evaluate the feasibility of percutaneous placement of this device as well as its short- to intermediate-term safety and efficacy in treatment of AAA in this canine model.
Eight mongrel dogs weighing 18.4 to 23.8 kg were used in this study. All animals were taken care of and treated strictly according to the principles outlined in “Position of the American Heart Association on Research Animal Use,” adopted November 11, 1984. The protocol was approved by the Animal Research Committee of Loma Linda University.
Construction of the Device
The device was a self-expandable, stainless steel, woven mesh endovascular prosthesis with a guiding catheter. It was constrained by a removable plastic sleeve; as the sleeve was withdrawn, the device returned to its original, unconstrained size, anchoring it against the vessel wall. Two types of stents were used in the study. One type was covered by porous polyurethane and constructed of 24 wire filaments, each 0.17 mm wide. It was constrained in an elongated configuration on a unistep 12F (1F=0.33 mm) delivery system. When the device was unconstrained, it had a diameter of 16 mm and a length of 40 to 60 mm (Fig 1A⇓). The other type of stent was constructed of 48 wire filaments, each 0.14 mm wide, without any covering over the wires. It was constrained on a unistep 10F delivery system. When this device was unconstrained, the diameter of the prosthesis was 14 mm and the length was 40 to 60 mm (Fig 1B⇓). Three covered stents and five uncovered stents were used in the trial.
Surgical Creation of AAA Model
Each dog was sedated with 2% sodium thiamylal; an endotracheal tube then was placed for mechanical ventilation with 1% to 1.5% halothane during the operation. After incising of the skin layer, a piece of rectus abdominis fascia about 3 to 3.5 cm long by 2 to 2.5 cm wide was taken. Upon entering the retroperitoneum, the aorta was exposed and cross-clamped below the renal arteries and above the aortic bifurcation. A 1-cm incision in the aorta was created, and a patch of the fascia previously taken was sutured over the hole created in the aorta. Two types of aneurysms were created: fusiform and saccular. Aortic cross-clamp time ranged from 35 to 45 minutes. Prophylactic cefazolin was given for 3 days.
Eight weeks was allowed for maturation of the aneurysms after surgery, in part to see if spontaneous obliteration or thrombosis of the aneurysm would occur. The dogs then were taken back to the catheterization laboratory, and under anesthesia an 8F sheath (USCI, CR Bard Inc) was advanced over a guide wire into the right femoral artery. A 6F angiographic pigtail catheter (Cordis Corp), over a 0.035-in (0.89 mm) Benson wire (Cook Co), was then advanced into the descending aorta 5 cm above the aneurysm, and an abdominal aortogram was performed by injection of 30 mL contrast dye at 20 mL/s to document continued patency of the surgically created AAA before stent placement. At the end of the aortogram, the pigtail catheter and the 8F sheath were removed from the femoral artery, leaving the Benson wire in the descending aorta. Under fluoroscopic control, a stent that had been premounted on a delivery catheter was passed over the Benson wire to the desired position and deployed by pulling back the covering plastic sleeve (Fig 2⇓). Immediately after stent placement, an aortogram was performed with the same amount of contrast material and injecting rate as the aortogram performed before stent placement to evaluate flow through the stent and the region of the aneurysm. In two dogs, uncovered stent was intentionally placed over the renal arteries in one dog and over the superior and inferior mesenteric arteries in another dog. All dogs were observed for recovery, and prophylactic cefazolin was given for 3 days. No anticoagulants or antiplatelet drugs were used before, during, or after the procedure.
Measurement of Aneurysm Size
Aneurysm size was determined angiographically using a computerized digital analytical system, AngioComm StatVIEW Digital Recorder (ImageComm Systems, Inc). The measurements were independently performed by one of the authors (C.E.R.) and a catheterization laboratory technician (F.L.H.). Since no measurement variation was larger than 10%, a third observer was not involved. The transverse diameter of the contrast column was measured at its widest point, from the outer wall of the aneurysm to the opposite wall of the aorta, or at its widest portion of the aneurysm, whichever measurement was larger. This was compared with a normal or reference aortic diameter defined as the average of the “normal” aortic diameter immediately above and below the aneurysm. Two measurements of the diameter were taken, and the mean value was presented.
The first two dogs (dogs 1 and 2) were killed shortly after stent placement to evaluate the position and function of the stent. Dog 3 was recatheterized and euthanized 2 weeks after stent placement because of decubitus ulcer formation as a result of paraplegia. The remaining five dogs underwent follow-up aortogram at the completion of the study 4 weeks after stent placement and were euthanized under deep sodium pentobarbital anesthesia by injection of potassium chloride solution, with subsequent removal of the stented segment of the aorta.
Gross and Histopathologic Study
Gross examination of each segment of stented aorta was performed by longitudinal opening of the aorta and stent, with inspection of the lumen. The aneurysmal cavity also was opened and its contents examined. Specimens were fixed in 10% buffered formalin for routine light microscopy. Portions of the aneurysm wall and contents were submitted for routine processing, paraffin embedding, and light microscopic study of hematoxylin-eosin–stained sections by standard methods. An attempt was made to study the interface of the stent with the aortic intima after removal of the embedded stent wires as carefully as possible from the neointimal tissue, with variable success. Portions of the stent with aortic wall were also fixed in 2% buffered glutaraldehyde for scanning electron microscopy in dogs 4 through 8.
Surgical creation of the AAA was successful in each dog. There were no complications encountered either during the procedure or during the 8-week period of aneurysm maturation, and continued patency of the aneurysms was documented before stent placement.
The stent was successfully deployed and placed across the aneurysm on the first attempt in all eight dogs. Aortogram before stent placement revealed a mean aneurysm diameter of 1.86±0.47 cm, which was 70±40% larger than the reference aortic lumen (1.09±0.08 cm) (Table⇓) (Fig 3A⇓ and Fig 4A⇓). Aortogram immediately after stent placement showed effective exclusion of flow from the aneurysm, with reduction in the size of the angiographically demonstrated lumen to a mean diameter of 1.16±0.35 cm, which was no different than the reference aortic lumen (1.09±0.16 cm) (Table⇓). The aneurysmal cavity disappeared in all three dogs receiving a covered stent (Fig 3B⇓), and the aortic blood flow into the aneurysmal cavity markedly diminished, with only faint contrast filling the cavity in all five dogs with uncovered stents (Fig 4⇓, B and C). No acute complications were observed during the placement. The average procedure time was 30 minutes, and average fluoroscopy exposure was 3 minutes.
Follow-up Angiographic Study
One dog that had received a covered stent (dog 3) developed paraplegia secondary to spinal cord infarction 48 hours after stent placement. An aortogram indicated that the stented section of the aorta was completely occluded, with only very limited aortic flow from collateral arteries distal to the stent, and this animal was euthanized 2 weeks after stent placement because of decubitus ulcer formation. The remaining five dogs (dogs 4 to 8) tolerated the device well for the entire 4-week follow-up period, with no complications, and underwent follow-up aortogram 4 weeks after stent placement (Table⇑). The diameter of the stented aortic lumen was essentially the same as immediately after stent implantation. The aortograms showed all stents to be patent, with vigorous flow through the stented section of the aorta (Fig 3C⇑ and Fig 4D⇑). The cavity of the aneurysm could not be visualized in four dogs (two with covered stents and two with uncovered stents). Slight cavitary filling with contrast was observed in dog 5, which had received an uncovered stent, but the angiographic size of the aneurysmal cavity appeared to have further reduction (Fig 4D⇑). Importantly, the major arterial branches bridged by the uncovered stent were maintained with unrestricted flow through these arteries, and no angiographic evidence of the residual aneurysm was revealed (Fig 5⇓).
No migration of the stent was noted in the dogs. The two dogs that were killed shortly after stent placement (dogs 1 and 2) showed a fully deployed, patent stent across the site of the aneurysm, with no clot. Dog 3, which had angiographic evidence of thrombosis as described above, was found to have thrombus filling the entire lumen of the stent as well as the cavity of the aneurysm. In the remaining five dogs, all stents were patent, with no significant clot in the lumen. The stents were either totally or partially covered with neointima, and the saccular aneurysms were essentially completely filled with clot (Fig 6A⇓), while the fusiform aneurysms showed adherence of the stent to the aortic wall with contraction of the aortic (aneurysm) diameter but without significant thrombosis. In dog 5, which had angiographic evidence of a small residual aneurysm, the stent wires were covered by neointima around the edges of the aneurysm and the cavity was mostly filled with clot. In two dogs in which the uncovered stents had been intentionally placed across the renal arteries and superior and inferior mesenteric arteries, the patency of those arteries was maintained without obstruction by neointima (Fig 6B⇓).
Light and Scanning Electron Microscopy
The covered stents showed fibroblastic and histiocytic permeation of the coating fabric by light microscopy, with patchy chronic inflammation and variable endothelialization of the surface (Fig 7A⇓). A fibroblastic reaction was focally present in the intima beneath the stent. Some areas showed fibroplasia of the intimal surface as well, with a microlayer of adherent thrombus. One dog that received a covered stent (dog 4) showed focal penetration of a stent wire into the aortic media, with associated chronic inflammation. This was uncommon, however, and most of the stent wires appeared to rest on the aortic intima, with or without a thin layer of intervening intimal fibrous reaction. The dog with thrombosis of a covered stent (dog 3) showed focal necrosis of the inner aortic wall, apparently secondary to the thrombosis. Histological study of tissue reaction associated with the uncovered stent was considerably more difficult, as the neointimal tissue was very thin and easily disrupted by removing the stent wires before sectioning. However, we were able to demonstrate a mild histiocytic and giant cell reaction to the wires, with a thin overlying layer of neointima composed largely of myofibroblastic-type cells, and little reaction in the underlying aorta. The clot in the aneurysms typically showed peripheral organization (Fig 7B⇓).
Scanning electron microscopy showed the neointimal surface to be covered by cells with the appearance of fibroblasts, with variably sized islands of endothelial cells partially covering the fibroblasts (Fig 8⇓, A and B). The endothelial cells in some areas appeared to have migrated onto the neointimal surface through porelike spaces between the wires.
Aortic abdominal aneurysmectomy has been the only effective therapy to prevent rupture of AAA and to save life when rupture occurs. However, it is a major surgical procedure, with substantial attendant complications, particularly in the elderly or patients with underlying cardiovascular, kidney, or pulmonary disease. Furthermore, it has been estimated that the direct cost of AAA repair is $2.3 billion per year in the United States alone.18 Thus, AAA imposes a considerable economic burden on already strained healthcare resources.
Further refinement and simplification of surgical techniques in AAA repair seem unlikely at present.19 However, nonsurgical ablation by means of implantation of a specially designed device to bridge AAA has been tested in animal experimental studies and appears promising. This alternative is expected to result in less morbidity and mortality than surgical repair and has the potential to significantly reduce the cost of treatment of AAA.
Several intraluminal devices have been tested previously in the treatment of AAA in animal models. Balko et al14 used a polyurethane graft with a Nitinol and/or stainless steel frame to bridge the artificial aneurysm with a 15F catheter. The aneurysms were successfully excluded, and the patency and position of the device were confirmed shortly after the device implantation. However, no follow-up study was performed. Important disadvantages associated with this device were the potential for premature expansion inside the catheter due to its material (Nitinol), which required constant cold saline irrigation, and the large size of the catheter required for placement, limiting its application.
Lawrence et al15 developed an expandable Gianturco stent covered with Dacron and tested it in normal aorta of dogs. However, because of the nonexpandable nature of its Dacron covering as well as its tendency to wrinkle, there was concern regarding its long-term promotion of thrombosis and fibrogenesis, with possible eventual stenosis or thrombosis. In addition, side branch arteries were occluded by the Dacron covering.
To overcome the disadvantages of the device, Mirich et al16 percutaneously placed a modified Gianturco stent in six dogs with surgically created AAA. The modified stent was covered by porous nylon material to allow stretching and to maintain patency of the side arterial branches where covered by the stent. It had a fully expanded size of 11 to 12 mm with a 12F delivery catheter. One of the dogs had incomplete stent expansion due to a defect of the device, with subsequent cephalad stent migration and death as a result of occlusion of the renal arteries. In the remaining dogs, the aneurysm was successfully sealed off. There was no evidence of device migration or luminal narrowing. Microscopic examination revealed neointimal coating of the fabric strands across side branch arteries and evidence of renal ischemia in one dog, indicating that these vessels might eventually be occluded.
More recently, Laborde et al17 developed a weft-knit Dacron tube with balloon-expandable stents that were placed across artificial AAAs in eight dogs. Initial success in excluding AAA was achieved in all the dogs. However, two had early occlusion caused by torsion of the device as a result of defective folding inside the delivery sheath. At follow-up, the remaining dogs had patent Dacron grafts, but four showed evidence of kinking due to shrinkage of the artificial aneurysm. Histopathologic study demonstrated endothelialization that was complete on the stents and partial in the artificial aneurysms.
We used two types of new endovascular devices in the treatment of AAA in a dog model. Except for one dog with early stent occlusion, we did not encounter any complications in the placement or use of this device. The two types of this device have somewhat different features and therefore may have potentially different clinical applications. The covered stent would appear likely to be most useful in the emergency treatment of ruptured or acutely expanding AAAs, since in our study it appeared to exclude aneurysms immediately upon placement and therefore might be able to stop bleeding from a leaking or acutely expanding aneurysm. However, because of its fabric coating, it could not be allowed to bridge the entrance of any major arterial branch because it would also interrupt blood flow through that artery. Fortunately, most human AAAs occur below the renal arteries, so it would not be necessary to bridge those arteries with a covered stent. Exclusion of spinal arteries did not appear to be a problem in our dogs; signs of spinal cord ischemia were not observed except in the one dog whose stent thrombosed shortly after placement. However, only three dogs were treated with the covered stent, and more animal experimental studies are certainly required after further refinement and improvement in device design and technique.
The uncovered stent has not been reported to be used in experimental animals, but its features suggest it to be potentially useful in the elective therapy of stable AAAs. This device was able to significantly reduce the angiographic size of most aneurysms, providing a framework for neointimal growth while also maintaining the patency of branch arteries. In the two animals in which these stents were placed across major arterial branches, those branches were widely patent, with unrestricted blood flow at 4 weeks after stent placement. This unique aspect of the uncovered stent holds promise for the long-term treatment of AAA, although longer follow-up is needed. The effect of the uncovered stents on aneurysm size was in some cases less dramatic than that of the covered stent; however, it was able to either obliterate or substantially reduce the size of the aneurysm in all cases by the end of the 4-week follow-up period. This is an important achievement because a strong correlation has been demonstrated between aneurysm size and the likelihood of rupture.2 3 6 10
It appears likely that the mechanism of closure of the saccular aneurysms by the uncovered stent is by reduction of blood flow in the aneurysm cavity secondary to the shear forces introduced by the wires that cross the aneurysm opening. This is suggested by only faint contrast filling the aneurysmal cavity immediately after placement of the stent. The mechanism of shrinkage of the fusiform aneurysms is less clear. However, the almost immediate reduction in the angiographically demonstrated diameter of the aorta in the region of the aneurysm in two of the three dogs with fusiform aneurysms suggests that the shear forces introduced by the stent wires may redirect blood flow back toward the lumen and away from the dilated aortic walls, allowing for healing and contraction of the dilated aorta. The one case in which the diameter was slightly greater immediately after stent placement appears to be due to placement of a relatively large stent in a fairly small dog, which resulted in some initial stretching of the aorta. This case, however, went on to show normalization of the aortic diameter with 4 weeks of follow-up. The obliteration of the aneurysms in all cases appears to be due to the presence of the stent and not to spontaneous thrombosis, as angiography demonstrated immediate changes in the effective size of the aorta after stent placement. Furthermore, the aortogram before stent placement showed that the size of the aneurysms appeared to be larger rather than reduction from the original created size in the initial 8 weeks after aneurysm creation, a period designed to serve as a control for any possible spontaneous healing, regression, or thrombosis of the aneurysms.
Placement of the device was feasible and straightforward and appears to be safe. We did not encounter any difficulty in positioning or deploying the stent. However, several considerations must be kept in mind. One must be certain that the stent is in the correct position before withdrawing the covering sheath for deployment since it is not retrievable once deployed. Care also must be taken not to bridge any major arteries with a covered stent, otherwise kidney or other visceral damage will almost certainly ensue.
Thrombosis and Endothelialization
Minimal thrombosis and rapid endothelialization are critical issues for the success of stent placement.20 All stents in our study showed at least partial neointimal endothelialization at the end of 4 weeks and generally minimal thrombus on the aortic luminal aspect of the stent. Palmaz et al21 22 23 24 have demonstrated the absence of thrombosis over a long time period after aortic graft placement. This would suggest that anticoagulation is not necessary after aortic stent placement because of the vigorous blood flow. However, Schatz20 pointed out that metallic stents, regardless of design and configuration, have an inherent thrombogenic effect without adequate anticoagulation. In our study, one stent was completely occluded by thrombus, which might be prevented by administering anticoagulants, and similar stent thrombosis has been reported by others.12 16 17 Thus, anticoagulation may be important for stent placement even in the aorta. Further studies are warranted to clarify this important issue.
Although infection and colonization of the stented area did not occur in our study, this potential complication must be given serious consideration. Prophylactic administration of antibiotics, as used here, is recommended for a short period after stent placement. Once a stent is completely covered with neointima, the likelihood of infection as well as thrombosis may be minimized, so that long-term antibiotic therapy probably is not indicated.
The surgically created aneurysms in the animal model are not necessarily comparable in configuration to the most common AAAs in humans, particularly with respect to the opening of the aneurysmal cavity, which was proportionally smaller than in human AAAs. It is uncertain whether neointimal surfacing of an uncovered stent would be as rapid and complete in these larger aneurysms as it was in our study. Further studies are indicated in this regard.
The present study was based on short-term and immediate follow-up after stent placement in a relatively small sample size. Although the data are encouraging, longer follow-up and a larger sample size are needed to establish both the long-term efficacy and safety of this procedure.
Our study serves to document the feasibility of percutaneous placement of an endovascular stent in the treatment of AAA in an animal model. Short-term follow-up suggests this to be a relatively safe and effective method for nonsurgical treatment of AAA in the dog and deserving of serious consideration for use in humans when further refinements in technique are accomplished. If proven safe and effective for humans as well, it has the potential for substantially reducing the morbidity and mortality associated with AAA.
This study was supported in part by a grant from the Schneider US Stent Division, Pfizer Hospital Products Group, Minneapolis, Minn. The authors wish to thank Erlinda L. Kirkman, DVM, for experimental animal assistance, Frank L. Haynes for catheterization laboratory assistance, and Dennis O’Malley and Debra Torbitt for technical assistance in the scanning electron microscopy of the study material.
Reprint requests to Dr Carlos E. Ruiz, Division of Cardiology, Department of Medicine, Loma Linda University Medical Center, Room 2426, 11234 Anderson St, PO Box 2000, Loma Linda, CA 92354-0200.
Presented in part at the 67th Annual Scientific Sessions of the American Heart Association, Dallas, Tex, November 1994.
- Received July 11, 1994.
- Revision received October 31, 1994.
- Accepted November 26, 1994.
- Copyright © 1995 by American Heart Association
Melton LJ III, Bickerstaff LK, Hollier LH, van Peenen HJ, Lie JT, Pairolero PC, Cherry KJ, O’Fallon WM. Changing incidence of abdominal aortic aneurysms: a population-based study. Am J Epidemiol. 1984;120:379-386.
Fowkes FGR, Macintyre CCA, Ruckley CV. Increasing incidence of aortic aneurysms in England and Wales. BMJ. 1989;298;33-35.
National Center for Health Statistics. Detailed Diagnosis and Procedures. Rockville, Md: 1984.
Vital Statistics of the US. CA 37:6, 1987.
National Center for Health Statistics, Graves EJ. Detailed Diagnoses and Procedures, National Hospital Discharge Surgery, 1988: Vital and Health Statistics, Series 13, No. 107. Washington, DC: US Government Printing Office; 1991. DHHS publication No. (PHS) 91-1768.
Mullins CE, O’Laughlin MP, Vick GW III, Mayer DC, Myers TJ, Kearney DL, Schatz RA, Palmaz JC. Implantation of balloon-expandable intravascular grafts by catheterization in pulmonary arteries and systemic vein. Circulation. 1988;77:188-199.
Lazarus HM. Endovascular grafting for the treatment of abdominal aortic aneurysms. Endovasc Surg. 1992;72:959-968.
Schatz RA. A view of vascular stents. Circulation. 1989;79:445-462.