A Percutaneous Model and Treatment With Stent Grafts in Sheep
Background Treatment of aortocaval fistulae (ACFs) by open surgery is known to be a high-risk procedure. In this experimental study, we developed a percutaneous model of ACF to test a new nonsurgical method of treatment of ACF using endovascular stent grafts.
Methods and Results An ACF was created percutaneously in eight sheep. Via a combined venous and arterial femoral approach, angioplasty balloons were inserted to occlude the infrarenal aorta and inferior vena cava (IVC). The IVC was punctured through the lateral aortic wall with a transjugular liver biopsy needle. The fistulous tract was then dilated with an 8-mm angioplasty balloon, and the animal was heparinized. Two weeks later, a 10-mm Cragg-Endopro-Stent was inserted into the aorta at the level of the fistula via a percutaneous femoral approach. ACFs were successfully created in all animals, with rapid dye shunt through the fistula and a 20% increase in cardiac pulsations. Follow-up angiograms at 2 weeks showed a patent ACF, and stent implantation excluded the fistula in every case. Angiographic and pathological examinations up to 6 months demonstrated normal aortic patency and persistent exclusion of the fistulae.
Conclusions ACFs were effectively treated by endovascular grafting of the aorta. The animal model and the stenting procedures were both performed percutaneously.
Aortocaval fistula (ACF) is a rare clinical condition that can be either spontaneous (80% of the cases), related to abdominal aortic aneurysm, or posttraumatic, usually due to lumbar disk surgery (20% of cases).1 The clinical presentation is varied and may be atypical.2
Surgical treatment is difficult because of venous hypertension and perivascular inflammatory reaction, and aortocaval clamping may be perilous.1 Despite recent surgical refinements such as transvenous balloon tamponade and autotransfusion,3 a new therapeutic approach has to be proposed because of the persistent high operative mortality.4
Percutaneous stent-graft implantation could avoid most of the drawbacks of open surgery of ACF, since recent reports have emphasized that it can effectively treat abdominal aortic aneurysms and arteriovenous fistulae in peripheral arteries.5 6
However, a model of ACF in a large animal species must be developed to test this technique before it is applied in humans. The present experimental study is a primary description of a new percutaneous model of ACF that was developed to assess its percutaneous treatment by endovascular stent graft.
All experiments were performed according to European Community rules of animal care.7 In eight sheep weighing 50 to 60 kg, an ACF was created percutaneously with a transjugular liver biopsy needle (Nycomed AS) via a combined venous and arterial percutaneous femoral approach.
After fasting for 24 hours, the animals were premedicated by intravenous injection of pentobarbital (20 mg/kg) and sodium thiopental (20 mg/kg), placed in a supine position, and intubated and anesthetized with 1% halothane. Oxygen arterial saturation and heart rate were monitored during the procedure.
Both groins were disinfected, and the right femoral artery and vein were cannulated before insertion of an 8F or 9F introducer sheath, respectively. The left femoral artery was subsequently cannulated, and the 9F introducer sheath of the transjugular liver biopsy needle was inserted.
An angiogram was first performed during simultaneous injection of contrast medium (Ultravist 370, Schering) into the aorta and inferior vena cava (IVC) to assess the size and the relative position of the vessels.
A 10-mm angioplasty balloon was then fully inflated in the infrarenal aorta to interrupt blood flow through the distal segment of the abdominal aorta. Similarly, a 20-mm angioplasty balloon was fully inflated in the IVC in front of the aortic balloon to fully distend the vena cava and to guide transaortic puncture of the vena cava.
The transjugular liver biopsy needle was inserted through the left 9F arterial sheath and advanced into the infrarenal aorta, 30 to 50 mm above the aortic bifurcation. During contrast injection into the vena cava, the tip of the needle, protected by its sheath, was gently pushed against the right lateral wall of the aorta until the left lateral wall of the opacified vena cava appeared deformed and displaced. From this position, the tip of the needle was further advanced to puncture the vena cava through the right lateral aortic wall. When venous blood was aspirated, a 0.035-in hydrophilic J guide wire (Terumo) was inserted into the vena cava while the venous balloon was deflated.
The fistulous tract was then dilated by advancing the sheath over its dilator over the guide wire from the aorta into the vena cava. An 8-mm angioplasty balloon was then inserted over the wire through the sheath and fully inflated in the aortic and vena caval wall.
After deflation of the aortic balloon and while the guide wire was left in place in the fistulous tract, an angiogram was performed to assess patency of the fistula. In the case of retroperitoneal bleeding, the aortic balloon was reinflated at low pressure at the site of the leak for 5 to 10 minutes to ensure hemostasis.
The animal was observed for 30 minutes after declamping; then another angiogram was obtained to confirm patency of the shunt and the absence of persistent leakage. If the shunt was occluded, the fistulous tract was redilated as described above.
At the end of the procedure, animals were given intravenous heparin (10 000 IU) and aspirin (500 mg) and were allowed to recover. The maximal duration of aortic clamping at any one time during these procedures was always less than 10 minutes.
Postoperative analgesia consisted of promethazine 2.5 mg · kg−1 · d−1 (Phénergan, Rhône-Mérieux) and dipyrone 100 mg · kg−1 · d−1 (Novalgine, Distrivet) for 5 days. Systematic prophylactic antibiotic therapy using 1 million IU/d of penicillin G (Bipeni, UVA) was also administered for 3 days after the operation.
The stent graft used in this experiment was the Cragg-Endopro-Stent 1 (Mintec), which consists of a 10-mm-diameter self-expandable stent composed of a nitinol wire bent on itself with a 6-0 polypropylene suture and covered with a 0.2-mm heparin-coated Dacron fabric.8 Nitinol is a nickel-titanium thermal-shape-memory alloy that gives the stent its self-expandable characteristics. It is preloaded in a 10F cartridge and is delivered by means of a pusher system in a long 10F sheath.
To evaluate short and long stents, we implanted a 30-mm stent in three animals (four stents) and a 60-mm stent in five animals (five stents).
Two weeks after creation of the shunt, a follow-up angiogram was performed to assess the patency of the ACF. When the fistula was still patent, the covered stent was inserted through a 10F introducer sheath via a percutaneous femoral approach and delivered into the aorta at the level of the fistula. Repeat angiograms were then performed every 3 minutes until demonstration of exclusion of the ACF. In the case of incomplete stent deployment, intra–stent-graft dilation was performed with a 12×40-mm angioplasty balloon (Boston Scientific). After stent implantation, the same postoperative treatment as that described above was administered to the animals, with the exception of aspirin and heparin.
Follow-up consisted of angiography before the animals were killed at 1 day to 6 months after stent implantation (one sheep at 1 day, one at 1 week, one at 3 weeks, one at 6 weeks, one at 4 months, and two at 6 months). The animals were examined by angiography to assess both aortic patency and exclusion of the ACF and then were killed with an overdose of 70 mg/kg pentobarbital (Doléthal, Vétoquinol). The aorta and IVC were harvested en bloc from the renal to the iliac vessels, washed with sterile saline, and fixed in 10% neutral buffered formalin solution (Sigma Chemical Co).
Gross Pathological and Histopathological Study
The samples were analyzed macroscopically before and after the stent graft was withdrawn. They were then embedded in paraffin and sectioned transversely and longitudinally for microscopic study of the outcome of the fistula and the modifications of the vessel wall covered by the stent graft. Sections were stained with a trichrome stain, and the architecture and inflammatory reaction within or around the vessel wall were studied.
Assessment of the Results
Early stent efficacy was evaluated in terms of adequate deployment, with a smooth appearance of the arterial wall and patency of the aortic lumen, exclusion of the fistula, and the appearance of collateral branches covered by the stent. Late stent efficacy was evaluated by the same criteria regarding tolerance and tightness of the stent graft. In addition, possible migration and deformation of the stent were carefully assessed on repeat angiograms in relation to bony landmarks.
An ACF was successfully created in every case, with a 20% increase in heart rate (110 to 130 bpm). The postoperative angiogram showed a rapid flow of contrast through the fistula from the aorta into the IVC (Fig 1⇓).
However, in five of eight animals, retroperitoneal bleeding was observed immediately after creation of the fistula. In the first case in our experiment, bleeding was successfully treated by immediate stent-graft implantation. Further experience in four other animals demonstrated that hemostasis could be achieved by 5- to 10-minute reinflation of the aortic clamping balloon at the level of the leakage. In three animals, this maneuver was unnecessary, since no leak was identified after declamping.
Repeat angiograms at 2 weeks showed a patent fistula in four cases, and the animals were stented at this time. In another three animals, however, the shunt was reoccluded (a pseudoaneurysm projected over this tract was observed in one of the reoccluded fistulae). In these three animals, a new shunt was created by the same method, and the animal was rescheduled for stenting 2 weeks later. All three ACFs remained patent at this time.
Nine stent grafts, 10 mm in diameter, were implanted in eight animals. Four 30-mm stent grafts were used in three animals, and five 60-mm stent grafts were used in the other five animals. In the animal that had both an ACF and a pseudoaneurysm, two 30-mm stent grafts were used, one to cover the pseudoaneurysm and one to exclude the fistula.
Stent delivery was hampered by technical problems related to the device in four cases. In two cases, the difficulty was due to initial design of the loading cartridge and resulted in one acute aortic thrombosis, and in two cases, delivery was imprecise because of sudden slippage of short (30-mm) stent grafts into the aorta while the sheath was being retrieved. Despite these difficulties, adequate occlusion of the fistula was obtained, and the aorta remained patent after stent-graft delivery in all but one case.
Sequential angiography after stent-graft delivery showed that the fistula was occluded within 10 minutes. The IVC was no longer opacified after 3 minutes, but occlusion of the fistulous tract itself was completed over the ensuing minutes. In three cases, intra–stent-graft dilation with a 12-mm balloon was necessary to exclude the fistula by achievement of full stent deployment. Lumbar arteries covered by the stent graft were immediately occluded in all cases and were still not opacified on the 10-minute angiogram.
On follow-up angiograms, the aorta remained patent and the fistula and/or pseudoaneurysm remained occluded; in addition, no migration of the stent graft was observed.
Occlusion of the lumbar arteries persisted in all cases (up to 6 months). However, in one case, a pair of lumbar arteries initially occluded by the stent graft were reopacified by adjacent lumbar arteries on the 6-month angiogram.
In another case, the right renal artery covered by the stent graft was still patent on the 3-week angiogram, probably opacified by persistence of an open space between the aortic wall and the fabric of the stent graft (Fig 2⇓). In this case, we assumed that reopacification of the renal artery was due to recoil of the stent graft, despite intra–stent-graft dilation performed on the day of stent-graft implantation.
When the animals were killed, the vena cava appeared to be adherent to the aorta (Fig 3⇓), with no evidence of organized clot or recent bleeding. In addition, a mild inflammatory reaction with enlarged lymph nodes was observed in the animals killed before 3 weeks.
After the aorta was opened longitudinally, the luminal surface of the stent was covered with a thin fibrinous translucent layer, with no evidence of thrombus formation on its surface. The stent graft was easy to remove, and its abluminal surface appeared to be free of thrombus. In addition, the impression of the stent struts was clearly visible on the aortic wall. At the level of the shunt, the scar of the fistula presented as convergent folds directed toward the site of the arterial wound. Examination of the stent graft showed no evidence of deterioration of the nitinol wires, polyethylene sutures, or fabric either in the dilated or in the nondilated stent grafts.
Longitudinal sections of the ACF revealed an organized clot within the fistulous tract (Fig 4⇓), consisting of fibrous connective tissue containing a large amount of collagen and many inflammatory cells. A 3-mm hole was visible at the level of the arterial and venous wounds. A mild neointimal proliferation without evidence of inflammatory reaction was noted within the aortic wall exposed to the fabric material of the stent graft. The stent struts induced thinning of the arterial wall, predominantly affecting the inner third of the media associated with fractures of elastic laminae and mild cellular proliferation. Examination of the perivascular hypertrophic lymph nodes showed a nonspecific inflammatory appearance.
Surgical treatment of ACFs is currently performed by transaortic closure of the fistula and restoration of continuity with a prosthetic graft, as initially described by Cooley in 1955.9 However, this operation remains dangerous: excessive bleeding from engorged retroperitoneal veins, paradoxical embolization of intraluminal thrombus, and frequent association with congestive heart failure in elderly patients results in high operative mortality, ranging from 22% to 51% of patients.1
Endovascular methods are rapidly being developed to treat a variety of vascular disorders, because they allow less traumatic interventions. The use of stent grafts was recently reported in the treatment of arterial trauma and arteriovenous fistula in peripheral vessels5 10 and could be transposed to larger vessels. However, this approach must be evaluated experimentally to assess both the procedure and the device, as performed for other aortic diseases.11 12
In our experiment, we demonstrated that it is possible to cure large, well-organized ACFs by obliterating the aortic orifice, similarly to Cooley’s operation but percutaneously. We have also shown that it is possible to treat various forms of abdominal aortic injuries and that stent grafting of the aorta can be used to cure acute traumatic lesions and pseudoaneurysms of the aorta.
Although this type of repair can be considered to be minimally invasive, the immediate and delayed tolerance of this therapeutic approach needs to be demonstrated.
No perioperative complications were observed in our experiment, and all animals recovered immediately after insertion of the stent graft. In addition, since this method does not require aortic clamping, no operative mortality was observed during percutaneous treatment despite a hyperdynamic circulation, as suggested by the increased heart rate.
In most cases, exclusion of the fistula was obtained only after a 5- to 10-minute delay, possibly due to insufficient impermeability of the fabric to blood and progressive occlusion of the small holes of the fabric, as usually observed with most polyester grafts, or to inappropriate size or mechanical characteristics of the device. With regard to stent mechanical properties, we have observed that full opening of the stent graft requires intra–stent-graft balloon dilation. However, delayed recoil of the extremities of this stent graft is still possible, as observed in the animal in which, despite initial exclusion after intrastent dilation, the right renal artery was reopacified on a control angiogram at 3 weeks. This drawback of the device is of clinical interest, since some cases of persistent filling of the aneurysmal sac after endovascular stent grafting of abdominal aortic aneurysms have been similarly attributed to a passage of blood between the stent graft and the aortic wall. On the basis of our various experiments, we believe that the diameter of the self-expandable or balloon-expandable device must be slightly greater than that of the aorta.
We also found that the length of the stent graft influenced accuracy of stent delivery, because only short stent grafts tended to suddenly jump out of the sheath during delivery. Insufficient contact of a short stent graft against the aortic wall when its distal extremity was delivered from the sheath could probably not prevent further longitudinal displacement.
Midterm tolerance of this stent graft was also well documented in our study, without any inflammatory changes in the arterial wall or narrowing of the aortic lumen at 3 to 6 months. The inflammatory periaortic lymph nodes could be secondary to retroperitoneal bleeding caused by creation of the fistula rather than to the material of the stent graft itself (fabric or nitinol alloy), as previously reported in humans.13
Finally, although this study was not designed to assess pathophysiological aspects, the variations of heart rate related to creation of the fistula and their reversal after closure of the fistula suggest that this model could also be used to study high-output congestive heart failure in large animals. The classic model of cardiac overload reported to date was described in rats, but this species is too small to test an experimental therapeutic approach to shunt closure.14 15
In conclusion, this percutaneous animal model is useful in the study of complex interventional maneuvers and assessment of percutaneous aortic endografting procedures. It could also be used to test new vascular stent grafts, assessing both their biological tolerance and mechanical properties.
Above all, this study demonstrated the feasibility and the efficacy of treatment of experimental ACFs by endovascular placement of stent grafts into the aorta. We believe that the use of stent grafts in this clinical condition could represent a significant therapeutic advance, particularly because it allows closure of the fistula without requiring open surgical access and aortic clamping. On the basis of the results of this experimental study, an organized clinical trial in humans could be started to assess the efficiency of this approach in patients suffering from ACF.
This work was supported by a full grant (931401) from INSERM (Institut National de la Santé et la Recherche Médicale). The procedures were performed at the Research Laboratory CR2i of INRA (Institut National de la Recherche Agronomique) and APHP (Assistance Publique–Hôpitaux de Paris), and the samples were examined at the INSERM Unit 367. The authors wish to thank J.P. Albert, C. Bourgeois, and G. Pigneaud of INRA and N. Blanck and A. Laurent of APHP for technical assistance at the CR2i (Centre de Recherche en Imagerie Interventionnelle, APHP-INRA, Bat 402, Domaine de Vilvert, 78352 Jouy-en-Josas, France). The authors also wish to particularly thank J.B. Michel (INSERM Unit 367) for sample examination and A. Saul for manuscript revision.
- Received July 6, 1995.
- Revision received December 7, 1995.
- Accepted December 19, 1995.
- Copyright © 1996 by American Heart Association
Aorto-caval fistula. Lancet. 1991;338:415-416. Editorial.
Directive CEE 86/609. J Officiel Communautés Européennes. December 18, 1986;L358:1-28.
Cooley DA. Discussion of Javid H, Dye WS, Grove JW, Julian OC. Resection of ruptured aneurysms of the abdominal aorta. Ann Surg. 1955;142:623.
Sapoval MR, Gaux JC, Long AL, Azencot M. Hoffman O, Schönfeld R, Raynaud A. Transient periprosthetic thickening after covered stent implantation in the iliac artery. Am J Roentgenol. 1995;164:721-723.
Huang M, Hester RL, Guyton AC. Hemodynamic changes in rats after opening an AVF. Am J Physiol. 1992;262:H846-H851.