Preclinical extracranial aneurysm models for the study and treatment of brain aneurysms: A systematic review

Search strategy

The literature was reviewed to identify main animal models of experimental saccular aneurysm, their refinements, and technical modifications. We searched the database Medline/PubMed on 10 January 2020 using the key words “mouse,” “rat,” “rabbit,” “dog,” and “swine” in combination with “aneurysm” with the Boolean operator (AND). The search was restricted to “animals.” Two investigators (SM and FS) independently screened titles and abstracts for eligibility based on our predefined criteria, reviewed the full text of eligible studies, and confirmed articles for inclusion. Any disagreement about a particular study’s eligibility was resolved with consensus of the other authors. Additionally, we identified studies cited in previous reviews and added select articles by cross-reference checking until no other publications were found. Figure 1 outlines the applied search algorithm applied and reasons for exclusion in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. ¹²

OPEN IN VIEWER

Figure 1.

PRISMA flowchart of PubMed search strategy and selection process. Among more than 4000 titles and abstracts screened, 473 studies underwent detailed full-text analysis. From those, 68 models, including refinements and technical modifications, were identified in five different species.

Eligibility criteria and analyzed features

We considered in vivo extracranial aneurysm models in mice, rats, rabbits, dogs, and swine but excluded non-English publications and studies on intracranial aneurysms. In the rare case of a model in which an extracranial vessel was transposed or transplanted into the intracranial space, the model was, despite its anatomical location, assessed as an extracranial aneurysm model. For each included study, we recorded authors; year of publication; detailed technique of aneurysm creation; number of created aneurysms; time for creation; patency rate; histological findings; mortality and morbidity; and shape, size, and location of the aneurysms.

Animal models were defined with consideration to the large differences that occur in the natural course of aneurysms among species. Therefore, we considered the animal model to be unique when identical techniques of aneurysm induction were performed in either different species or a different anatomical location within the same species or with any variation in the aneurysm graft (arterial or venous pouch) or with modification by chemical or mechanical means. Besides differences in species, anatomical location, or graft characteristics, a model was defined as novel only if modifications of an existing technique were major, that is, resulting in significant changes in aneurysm size or shape, patency rate, time for creation, or morbidity and mortality. Any uncertainty about the novelty of a particular study was resolved by two authors based on the predefined criteria listed above.

Sidewall aneurysm model

The sidewall aneurysm model, described by German and Black, is the oldest microsurgical technique to produce experimental aneurysms (Figure 4, A1). ¹³ The two variations of this technique have used (1) an isolated vein graft sutured end-to-side onto the parent artery (Supplementary Figure A1A) and (2) an arteriovenous fistula created by end-to-side or side-to-side suturing of a vein onto the parent artery followed by immediate or delayed ligation of the fistula (Supplementary Figure A1B).^14,15 Modifications include weakening of the parent artery with the use of nitrogen mustard, elastase, or mechanical destruction and subsequent aneurysmal outpouching at the site of the damaged vessel segment. Although the size of arteriotomy can be standardized, the size of the venous pouch itself and ultimately the aneurysm volume varies greatly. Only use of an arterial graft can ensure the standardization of aneurysm size.

To date, the most standardized aneurysm model in terms of graft origin, aneurysm shape and dimensions, volume-to-orifice ratio, and parent vessel to aneurysm long axis angle is the rat sidewall aneurysm created by an arterial pouch (Figure 4, A2; Supplementary Figure A2). ¹⁶ The patency rate of sidewall aneurysms depends to a large extent on the animal used. Untreated sidewall swine aneurysms have a tendency for spontaneous thrombosis.^17–19 Unlike the tendency for spontaneous thrombosis in swine,^17–19 the sidewall venous pouch aneurysm models in the rat, rabbit, and dog achieved excellent rates of long-term patency, that is, 100% (84/84), ¹⁶ 95% (38/40), ²⁰ and 94% (81/86), ²¹ respectively, without need for an antithrombotic regimen.

Stump aneurysm model

Fingerhut and Alksne ²² were among the first to ligate a major muscular branch of the femoral artery to create an arterial stump. Unlike all other bifurcation stumps, the tail artery stump in dogs (abdominal aortic trifurcation), described by Roach, ²³ represents a natural arterial bifurcation (Figure 4, B6; Supplementary Figure B6). Following these original descriptions, others have reported untreated arterial stump models at different anatomical locations and in various species. Untreated stumps are the simplest, fastest method to create an arterial pouch (even in relatively small species) and therefore are ideal models to screen potential embolization materials (Figure 4, B2; Supplementary Figure B2).^24–26

Influenced by previous experimental work of elastase-induced abdominal aortic aneurysms in rats, ²⁷ Cawley et al. ²⁸ infused elastase (via microcatheter from the femoral artery) into unilateral and bilateral surgically created stumps of the ECA in rabbits (Figure 4, B3; Supplementary Figure B3). Cloft et al. ²⁹ elaborated on a purely endovascular-based arterial aneurysm model. After blocking the origin of the left CCA 2 cm distal to the vessel origin using a detachable balloon, they infused the stump with elastase for 30 min (Figure 4, B1; Supplementary Figure B1). Altes et al. ³⁰ modified a version of this model that reduced creation time (<1 h) and procedural morbidity and mortality. They surgically exposed the right CCA, blocked its origin retrograde with a pliable balloon, and infused the distal stump with elastase. The group proposed several modifications to their right CCA model to customize resultant aneurysm morphology.^31–33 Because the right CCA model is more of a sidewall-type aneurysm (origins as bifurcation from the subclavian artery), Ding et al. ³⁴ adapted their previous modifications to the left CCA (originates as a trifurcation between the brachiocephalic artery and aortic arch) to create a more bifurcation type elastase-induced aneurysm in 2013.

Others proposed improvements for the elastase-induced CCA bifurcation stump model. Krings et al. ³⁵ advocated fluoroscopic guidance of elastase infusion to prevent leakage into aberrant tracheal arteries in effort to reduce mortality and morbidity (mainly attributed to hemorrhagic necrosis of the trachea). ³⁶ One year later, Hoh et al., ³⁷ also concerned by reports of failure of the original model due to aberrant branches from the right CCA, ³⁸ proposed temporary clip placement at the origin of the right CCA and infusion of the lumen under visual control (aberrant branches are ligated or coagulated). The technique reduced procedure time by 35 min, eliminated the need for fluoroscopy, and minimized the use of endovascular supplies. More recently, Kainth et al. ³⁹ proposed to ligate the left or right CCA 4–6 mm distal to the bifurcation, precisely brush the apex of the blind pouch using a 10/0 round-tip paint brush, and anchor the aneurysm to influence its orientation. This extravascular elastase application supposedly eliminates the need for angiography and endovascular supplies, preserves the femoral arteries for follow-up imaging or endovascular procedures, and allows for variation in the orientation of the aneurysm relative to the parent vessel.

The elastase-induced ECA bifurcation stump model also underwent modifications. Wang et al. ⁴⁰ proposed to ligate the ECA temporally using an aneurysm clip and infuse the ECA stump with elastase for 20 min. This surgical procedure is simple, fast, and requires no interventional support. Elastase-induced bifurcation stump models have also been introduced in the right ⁴¹ and left⁴² CCA in mice and in the abdominal aortic bifurcation in rats. ⁴³

The original elastase-induced right CCA bifurcation stump model in rabbits is one of the best characterized saccular aneurysm models. Serial angiography on days 1, 3, 5, 7, 14, and 21 demonstrated progressive enlargement, but thereafter the aneurysm size remained stable on one, two, three, and four months imaging. ⁴⁴ An excellent long-term patency of 92% was confirmed in a five-year follow-up study. ⁴⁵ Mortality rates associated with aneurysm creation and the embolization procedure were 8.4% (59/700) and 8.1% (43/529), respectively. ⁴⁶ Hemodynamic analysis revealed that most features of the elastase-induced right CCA model are qualitatively and quantitatively similar to many of human cerebral aneurysms. ⁴⁷ In a dose-escalation study, native bifurcation stumps failed to create saccular aneurysms. ⁴⁸ However, increased elastase concentration or incubation time or both demonstrated a trend for widening parent arteries but did not result in larger aneurysms cavities at three weeks follow-up. Rather, the aneurysms were mainly caused by the initial infusion of elastase and not by ongoing wall remodeling and degradation by endogenous proteases. ⁴⁹ Nevertheless, histological analysis has demonstrated that this model mimics wall types that have been identified in human intracranial aneurysms.^50,51

Terminal aneurysm model

Terminal aneurysms are aimed at simulating human basilar tip and internal carotid artery (ICA) bifurcation aneurysms. Originally described by Strother et al. in 1992, they were constructed by forming an arch with end-to-end anastomosis of both CCAs (Figure 4, C1; Supplementary Figure C1). ⁵² One should be aware that the volume increases as the aneurysm matures during the first few weeks but thereafter remains stable (up to six months follow-up). ⁵³ With its distinctive hemodynamic stress, terminal aneurysms demonstrate 100% patency in rat (9/9) ⁵⁴ and rabbits (9/9) ⁵⁵ and 85% to 100% (11/13) ⁵⁶ to (16/16) ⁵³ in dogs. Even in swine where long-term patency is a concern, terminal aneurysms have been reported to have direct intraluminal blood flow with a 100% (3/3) patency rate at three-month follow-up. ¹⁹

Bifurcation aneurysm model, natural, and artificial variations

Complex aneurysm model

This group represents microsurgically created aneurysms that mimic rare forms of intracranial aneurysms such as giant sized, broad-based, no-neck, fusiform, blister-like, bisaccular, bi- or multilobular, or confluence artery aneurysms. Nishikawa et al. ⁷⁴ had already reported confluence aneurysms in dogs by inclusion of a vein graft into an intracranial lingual-basilar artery anastomosis. Jiang et al. ⁷⁵ designed a dog confluence aneurysm model that used both the CCA and a vein graft to closely mimic human vertebral confluence aneurysms (Figure 5, E2; Supplementary Figure E2). Complex shaped bilobular, bisaccular, and broad-neck microsurgical aneurysms were created using various combinations of vein grafts in the rabbit artificial bifurcation model (Figure 5, E4; Supplementary Figure E4A–E4C). ⁷⁶ Despite the complex angioarchitecture, the aneurysms proved excellent long-term patency. ⁷⁷

Varsos et al. ¹⁴ first attempted to create a giant aneurysm by inducing a fistula between the CCA and external jugular vein and ligating this fistula one week later to form a giant sidewall aneurysm pouch (Supplementary Figure E3B). Yapor et al. ⁷⁸ presented a modification with a one-stage approach (Supplementary Figure E3C). Almost two decades later, giant aneurysms (approximately 1 × 2.5 cm) were created in the canine terminal ⁷⁹ and rabbit artificial bifurcation ⁸⁰ models (Supplementary Figure E3A).

With the emergence of novel endovascular technologies that extended the indications for treatment to a broader range of intracranial aneurysm (IA) morphologies, new preclinical models were needed. More recently fusiform ⁸¹ and curved sidewall ⁸² aneurysm models were designed in rabbits and dogs. Darsaut et al. ⁸³ introduced a giant fusiform model that used a rhomboid venous patch sutured to the anterior aspect of the lingual artery that allowed testing of flow diverters (Figure 5, E5; Supplementary Figure E5A).

Based on this groundwork, Greim-Kuczewski et al. ⁸⁴ proposed to surgically include a venous patch as a sidewall aneurysm to mimic some morphological aspects of human intracranial dysplastic, fusiform, and blood blister-like aneurysms (Supplementary Figure E5B). To test flow diverter treatment of aneurysms with side branches, Darsaut et al. ⁸⁵ created complex terminal aneurysms that featured a side branch originating from the aneurysm sac (Figure 5, E8; Supplementary Figure E8). Similarly, Fahed et al. ⁸¹ created a complex fusiform rabbit aneurysm model that permitted more comprehensive testing for flow diverters (Figure 5, E6; Supplementary Figure E6). Avery et al. ⁸⁶ presented a refined fusiform aneurysm model by periarterial application of combined elastase and CaCl₂ in a rabbit CCA. Yan et al. ⁸⁷ and Nakayama et al. ⁸⁸ created canine CCA curved-artery models (Figure 5, E7). First, they printed artificial hollow tortuous tubes (based on human ICA stereolitography data) and then transposed the animal’s CCA into this hollow rod (Supplementary Figure E7A-C).

Complex aneurysm models uniquely address a very narrow research questions about a specific aneurysm shape or flow condition. To date, these models are typically technically difficult, time-consuming to create (some even require two surgeons ⁷⁹ ), and are not well standardized or characterized.

Localization

The perianeurysmal space differs greatly between that of extracranial models and human brain aneurysms. With few exceptions, aneurysms models are not created in the subarachnoid space but in the soft tissues of the mediastinum,^29,30 neck,^28,93 leg,^22,94 retroperitoneal space^58,95 or within the abdominal cavity.^96,97 Nishikawa et al. created an intracranial sidewall aneurysm by suturing an autologous vein graft aneurysm onto the main branch of the middle cerebral artery (via transzygomatic approach), ⁶⁰ and sidewall and arterial confluence aneurysms were formed on the basilar and lingual arteries (via a transclival approach) in dogs. ⁷⁴ O’Reilly et al. ⁹⁸ relocated a rabbit vein pouch aneurysm via polyvinyl tubing from the right CCA into the subarachnoid space. However, technical difficulties, high mortality, and low patency rates prevented the further use of these techniques.

All other microsurgically created aneurysm models are surrounded by tissue that might affect intraluminal thrombosis and the healing response. The peritoneal cavity has the advantage of being less restrictive than the neck or other subcutaneous regions. Therefore, this region will potentially allow an aneurysm to grow if the wall is weakened. One specific advantage of aneurysm location on the renal artery is the possibility to control for potential ischemic embolic complications by postmortem assessment of renal infarction. ⁹⁵ Besides the location of aneurysm creation, one must also consider the diameter of parent arteries. Comparable to the diameter of great cerebral arteries where most human brain aneurysms are located is the size of the CCA in rabbits and the abdominal artery in rats.

Creation time

The simplest and fastest surgical techniques are the sidewall, bifurcation stump, and natural bifurcation models; these take 30–60 min for creation based on whether the aneurysm pouch needs additional modification. In comparison, terminal and artificial aneurysm models need up to 3 h for creation. The most demanding and time-consuming are complex aneurysm models, which require advanced microsurgical skills, an operating microscope, and occasionally two operators to create. ⁷⁹

Costs

Small experimental animals, such as mice and rats, are associated with lower costs than larger animals that require specialized experimental equipment, anesthesia, postoperative care, and housing. Additionally, surgery on small animals eliminates the need for a veterinary anesthesiologist and costs less for animal purchase, surgery, postoperative care, and housing. Surgery can usually be performed in the research laboratory, and general anesthesia is easily maintained by the operator with non-inhalational methods. Dogs and swine are the most expensive experimental animals. Operations are usually performed under costly general inhalational anesthesia by a veterinary anesthesiologist and additional technical assistants and must be performed in a designated animal operating theater.

Patency rate

Long-term patency without spontaneous thrombosis is a key characteristic and concern of any aneurysm model. Large series in rats, rabbits, and dogs demonstrated patency rates of 92.5%, ⁹⁶ 95%, ⁹⁹ 94%, ²¹ respectively, in 90° angle sidewall aneurysm models without any anticoagulation. However, when considering these excellent patency rates achieved by experienced researchers, it remains a matter of debate to what extent other factors are important measures in preventing thrombosis. These include extensive microsurgical training and associated technical factors (suture line, vein valves, badly placed sutures, or constricted aneurysm neck),^100,101 shape and size of arteriotomy, ⁹⁸ aneurysm volume-to-neck ratio, ¹⁰² angle to long axis of grafted venous pouches, ¹⁰³ number of sutures, tensionless anastomosis, and perioperative and postoperative management (compensation for fluid loss, pain management, antibiotics, vitamin complexes).^71,72 In rats and dogs, patency rates are better in artificial bifurcation models than in sidewall aneurysm models.^21,60,74 In dogs and swine, terminal aneurysm models exhibited better patency than sidewall aneurysm models.^19,56,66

Anticoagulation

Although most groups use local irrigation of parent arteries and storage of grafts in heparinized saline, systemic application of systemic anticoagulant^58,80,104 or antiplatlet^39,75,76 drugs is also common. In a rare direct comparison of effect of systematic antiplatelet medication on the patency rate of venous sidewall CCA aneurysms in dogs, Kerber and Buschman observed greater success with aspirin than without; patency rates were 74% versus 45%, respectively, but the difference was not statistically significant. ¹⁰⁰ Excellent long-term patency rates in large series of sidewall, terminal, bifurcation stump, and arterial bifurcation models in rat, rabbit, and dog provide evidence against the use of systematic anticoagulation. One must be especially careful with the administration of aspirin because of its potential influence on aneurysm wall biology and recurrence after endovascular therapy.^105,106

Histopathology and aneurysm healing

Debate continues about whether the morphological and histological characteristics of human cerebral aneurysms are more accurately modeled by elastase-digested arterial stump models or by surgically created vein or arterial pouch aneurysms.^51,107 Nonetheless, there is consensus that models with a strong tendency for spontaneous thrombosis and excessive neointima formation are inappropriate for study of healing after embolization. Sidewall aneurysms (especially in swine) have demonstrated favorable healing reactions after coil embolization compared to artificial bifurcation models. ¹⁰⁸ However, bifurcation models may also demonstrate healing rates that far exceed those seen in human aneurysms after coil embolization. ¹⁰⁹ Recent research revealed that aneurysm healing and recurrence after endovascular treatment also depend on the aneurysm wall condition.^1,110–112 Therefore, models that depict various types of wall conditions will gain interest and become appropriate models for testing novel endovascular therapies.^50,110

Reproducibility

Standardized and reproducible aneurysm creation is vitally important to improve preclinical assessment of novel endovascular devices and enhance comparability of results between laboratories. To date, the most standardized aneurysm models in terms of graft origin, aneurysm shape and dimensions, volume-to-orifice ratio, and parent vessel to aneurysm long axis angle are created by an arterial pouch.^55,96 To a certain degree, neck size and aneurysm volume can also be controlled and adjusted in elastase-induced aneurysms^{31,33,113,114} and modified techniques have achieved more consistent/standardized aneurysm diameters. ³² In venous pouch aneurysm techniques (especially complex models), the angioarchitecture (size, shape, and flow condition) is less standardized. However, the heterogeneity of aneurysm angioarchitecture increases the external validity of the model. Therefore, both are indispensable: standard models can test and refine the basic characteristics of a novel endovascular device, whereas models with high aneurysm variability can provide more exacting tests.

Ethical considerations

Depending on the location, most sidewall and some bifurcation stump and natural bifurcation models allow for the creation of two or more aneurysms in the same animal. Such models not only reduce the overall number of animals needed but also allow testing of two treatment modalities side-to-side within a single animal. Except for a three- and four-aneurysm dog model, most artificial bifurcation and complex aneurysm models allow creation of only one aneurysm per animal.^66,67,76,79 Some aneurysm models depend on syngeneic grafts, that is, using two animals to produce one aneurysm.^16,61,96 In contrast, aneurysm models with simple and fast aneurysm creation facilitate research with a larger number of experiments, thus ensuring adequate statistical power with increased significance and ultimately fewer animals. With the exception of some historical models and thanks to modified techniques over the past decades, currently used aneurysm models harbor low rates of morbidity and mortality.

Limitation of the study

Despite systematic approach by following the PRISMA guidelines and two investigators who independently screened the literature, this review might not be exhaustive based on our inclusion and exclusion criteria. Especially in the case of recognition of a new model related to a major modification of an existing technique, we cannot exclude a certain arbitrariness in the selection process despite strict application of predefined criteria.

Taking the potential confounding effects of the chosen species and techniques into consideration, basic biological concepts of novel intracranial aneurysm therapies can be tested in a great variety of models available today. Our categorization of models into five main groups and discussion of advantages and disadvantages allows a quick and comprehensive insight into the field. Detailed submodel analysis facilitates prediction of specific aneurysm model characteristics and therefore supports researchers in study planning, the execution of experiments, and interpretation of the results.

In summary, this systematic review of PubMed database, covering the period between 1950 and 2020, identified five main models of extracranial saccular aneurysms in mice, rats, rabbits, dogs, and swine, and their subsequent refinements and technical modifications. This review may serve as a compendium for investigators to identify the most appropriate model from a wide range of different techniques that suits best their experimental goals, practical considerations, and laboratory environment.