Animal models of human intracranial aneurysms have been an integral part of the development of endovascular occlusion techniques for decades [1, 2, 3, 4, 5, 6]. Other authors have stressed that, ideally, these animal models would simulate the morphology and hemodynamics of human intracranial aneurysms. Specifically, animal models should mimic the high shear force of bifurcation or terminal aneurysms [2, 7]. Most previously developed animal models relied on a surgical creation, typically by arteriotomy and vein graft placement [2, 6, 8]. For example, recent reports detail an effective method for surgically created bifurcation aneurysms in the rabbit [9, 10].

Current research efforts in interventional neuroradiology have begun to extend beyond physical modifications of devices to embrace modern biologic techniques, including the addition of peptides and growth factors to embolic coils in hopes of improving intraaneurysmal scar formation [1, 11, 12, 13, 14]. Testing of these biologic modifications requires the development of animal aneurysm models that not only encompass the morphologic and hemodynamic characteristics of aneurysms, but also mimic the biologic and molecular milieu of human intracranial aneurysms. The use of surgically created vein-patch aneurysms becomes questionable in the age of biologic modification because the presence of disrupted arterial walls and suture lines yields unknown effects on the biology of the model.

We previously reported a technique using the rabbit for endovascular aneurysm creation in which occlusion of the left common carotid artery combined with intraluminal elastase incubation results in bifurcation aneurysms along the aortic arch [15]. In this report, we describe a new technique that is much simpler than our previously reported method, with significantly diminished procedural morbidity and mortality. Furthermore, the resultant morphology is that of an aneurysm arising from the apex of a curved artery, similar to an ophthalmic artery aneurysm in humans. The apex of a curving vessel is subjected to high shear stresses, similar to a bifurcation [16, 17, 18]. Our model relies on endovascular creation without surgical techniques. We propose this model as a preferred type for testing the new generation of biologically modified embolic devices.

Saccular aneurysms were noted in eight (89%) of nine rabbits. Aneurysm sizes ranged in width from 3 to 6.5 mm (mean, 4.5 mm; SD, 1.2 mm) and in height from 5 to 10 mm (mean, 7.5 mm; SD, 1.6 mm). The parent artery, the brachiocephalic artery, ranged in size from 3 to 5 mm (mean, 3.6 mm; SD 0.7, mm). The aneurysms arose from the proximal brachiocephalic artery at the apex of a 90° turn.

The aneurysms were evaluated angiographically and histologically at 2 (n = 3), 10 (n = 3), and 24 (n = 3) weeks. At 2 weeks, the morphology of the right common carotid artery stump was that of a saccular aneurysm arising from the apex of the curving brachiocephalic artery (Fig. 3A). Organized thrombus covered by a monolayer of cells contiguous with the endothelium of the aneurysmal cavity was seen in the dome of the aneurysm (Fig. 3B). The Verhoeff-Van Gieson stain, which stains elastin black, showed obliteration of the elastic lamina within the wall of the aneurysmal cavity (Fig. 3C). The elastic lamina was well preserved in the walls of the adjacent brachiocephalic and subclavian arteries.

At 10 weeks after aneurysm creation, the morphology was similar (Fig. 4A). Again organized thrombus in the dome of the aneurysm was shown (Fig. 4B). The predominate cells in the organized thrombus showed characteristics of both smooth muscle cells and fibroblasts. The monolayer of cells across the dome was contiguous with the endothelial layer of the aneurysmal neck and body. In one of the 10-week specimens, laminated (organizing) thrombus was present at the dome within the aneurysmal cavity (Fig. 4B).

At 6 months after aneurysm creation, compared with the earlier times, no significant change occurred in aneurysm appearance either radiographically or histologically, (Figs. 5A and 5B). No change was found in the organized thrombus in the dome of the aneurysm. This thrombus was separated from the aneurysmal cavity by a monolayer of cells contiguous with the endothelium (Fig. 5C).

In addition to the organized thrombus in the dome, four (50%) of the eight aneurysms contained a small amount of remodeling thrombus in the lumen of the aneurysm just beneath the dome. The remodeling thrombus, although small, was apparent angiographically as a small filling defect in the dome of the aneurysm (Fig. 4A). This small organizing thrombus was seen in two (67%) of three 2-week aneurysms, one (50%) of two 10-week aneurysms, and one (33%) of three 6-month aneurysms. In no case did the organizing thrombus fill more than 15% of the lumen.

In the single case in which aneurysmal dilatation was absent, a small aberrant artery arose from the proximal common carotid artery. At the time of aneurysm creation, the aberrant branch vessel was not noticed. However, slow washout of contrast material from the right common carotid artery was seen after the injection of contrast material to verify balloon occlusion of the origin of the right common carotid artery.

All animals tolerated the procedure well, and no periprocedure mortality occurred. The total time to construct an aneurysm was approximately 1 hr.

In this report we have characterized the morphology, patency rate, and histology of saccular aneurysms we created in New Zealand white rabbits. In all cases the aneurysm was located at the apex of the curve of the brachiocephalic artery, separate from the origin of the left common carotid artery. The patency rate was excellent, with no evidence of spontaneous thrombosis up to 6 months after the procedure. The elastic lamina was markedly attenuated, and the dome of the aneurysm was covered by a monolayer of cells contiguous with the endothelium of the aneurysmal neck at 2 weeks. Mean size was 4.5 × 7.5 mm, which we consider to be appropriate dimensions for testing of the current platinum coil devices.

An arterial branch arose from the proximal right common carotid artery in the single animal in which the right common carotid artery stump failed to dilate. The arterial branch may be the tracheobronchial artery, an inconsistent branch that occasionally arises from the common carotid artery [19]. In the presence of this arterial branch, the elastase infusion is less effective because the elastase is carried away by the arterial branch. Hemodynamic factors such as the continued flow through the nascent aneurysm may also play a role in the lack of aneurysm formation. Fortunately, this anatomic variant is reported to be relatively uncommon [19]. We encountered it in one of nine animals.

The carotid occlusion was well tolerated by all subjects. The lack of significant morbidity associated with carotid occlusion relates to the patency of the circle of Willis and robust leptomeningeal collateral present in the rabbit.

Our aneurysm model is appropriate for the evaluation of endovascular devices because it yields saccular aneurysms that are similar to human intracranial aneurysms in many respects. First, the aneurysm occurs at a prominent curve along the brachiocephalic artery. This location subjects the aneurysmal neck to high shear stress [16, 17, 18] similar to that noted in ophthalmic aneurysms in humans. High shear stress is compounded by the proximity of the aneurysm to the aortic root. Second, the walls of our aneurysms simulate human aneurysms with the internal elastic lamina markedly diminished or absent [20]. The cellular composition of human intracranial aneurysms was the focus of previous studies [21, 22, 23, 24, 25]. These studies indicate that aneurysms lack an organized media, unlike the walls of our aneurysm model. Lending further support to our model as a reasonable simulator of human intracranial aneurysms, smooth muscle cells are found in abundance in the walls of human aneurysms. Third, the size of the aneurysmal sac is similar to that seen in humans, and the parent artery is similar in diameter to that seen in human intracranial arteries. Fourth, the endothelium of the aneurysm remains intact and free of suture lines, in contradistinction to surgically created vein-patch aneurysm models. Fifth, the rabbit represents the species with the greatest homology to the hematologic and coagulation systems of humans [26], and the blood pressure of the rabbit is similar to that of humans [27, 28]. Sixth, the size of the aneurysms remains stable for up to 6 months and thus is suitable for chronic experiments. Last, the wall of the aneurysm is arterial rather than venous similar to human aneurysms, another advantage over vein-patch aneurysm creation.

We have previously reported endovascular aneurysm creation in rabbits [15]. In our previous report, we used transfemoral access to place an occlusion balloon in the left common carotid artery, followed by endoluminal incubation of elastase in the arterial stump. Although this technique resulted in excellent long-term patency rates of the experimental aneurysms, the procedure for aneurysm creation was lengthy and associated with high morbidity and mortality (Cloft HJ, unpublished data). Furthermore, the resulting aneurysm was difficult to select for endovascular coil embolization procedures because of the morphology of the aneurysm neck in relation to the aortic arch (Cloft HJ, unpublished data). The technique reported in the current study is a significant improvement relative to the previous technique because essentially no procedural morbidity or mortality occurs, it can be completed in less than 1 hr, and it results in aneurysms that are easily catheterized in subsequent embolization procedures.

Elastase is gaining in popularity for use in the creation of aneurysm models [29]. Elastase-induced fusiform aortic aneurysms were originally reported in 1990 by Anidjar et al. [30]. Surgically created lateral aneurysms with intraluminal elastase infusion have recently been reported. These aneurysms have high rates of angiographic patency but disappointing patency rates based on histology [31]. The authors of this latter study speculated that the low patency rate was caused by the low flow in their lateral aneurysm model. The improved patency rate of our model over previous models may reflect the advantage of the high flow in the curving vessel aneurysm versus lateral aneurysm morphology.

Most previous reports of aneurysm models have relied on larger animal models, including swine and canine species [1, 3, 4, 6, 7, 8, 11, 32, 33]. The rabbit model used in our study offers advantages over these large-animal models, including the low cost for both purchasing and maintaining the animals, the similarity between rabbit and human coagulation systems and blood pressures [28], and the appropriate size of the aneurysm and parent vessel.

Even though our model offers many advantages over current models, several shortcomings exist. Because our model relies on distal occlusion of the common carotid artery, obligatory thrombus is initially present in the dome of the aneurysm. Fortunately, within 2 weeks complete reendothelialization across this clot occurs within the aneurysmal dome. Because the presence of thrombus in aneurysmal domes is commonly seen clinically, we do not see this as a significant limitation. Another potential criticism of our model is that the dome is supported by organized thrombus distally, rendering aneurysm regrowth after therapies such as coil placement unlikely. However, our model still allows assessment of coil compaction. Finally, a surgical cutdown is performed approximately 4 cm cephalad to the aneurysm itself. It is possible that some reaction to the surgery may confound experiments. However, no histologic evidence of inflammation occurred near the aneurysm. Several previous studies have reported marked inflammatory reactions induced by elastase [34, 35]. We noted no evidence of inflammatory reaction at any time. One potential explanation for this difference is that we incubated the elastase within an intact vessel. This procedure may limit exposure of elastase to surrounding tissues.

We thank Thomas D. Young and Sarah B. Hudson for assistance in preparing the histologic samples.

Address correspondence to T. A. Altes.

Received the 1999 American Roentgen Ray Society Executive Council Award at the annual meeting of the American Roentegen Ray Society, New Orleans, LA, May 1999.

T. A. Altes is supported by Radiological Society of North America Research and Education Fund Resident Research Grant.

D. F. Kallmes is supported by RSNA Research and Education Fund Scholar's Grant.