Brain arteriovenous malformation (AVM) is an important cause of hemorrhagic stroke. The etiology is largely unknown and the therapeutics are controversial. A review of AVM-associated animal models may be helpful in order to understand the up-to-date knowledge and promote further research about the disease. We searched PubMed till December 31, 2014, with the term “arteriovenous malformation,” limiting results to animals and English language. Publications that described creations of AVM animal models or investigated AVM-related mechanisms and treatments using these models were reviewed. More than 100 articles fulfilling our inclusion criteria were identified, and from them eight different types of the original models were summarized. The backgrounds and procedures of these models, their applications, and research findings were demonstrated. Animal models are useful in studying the pathogenesis of AVM formation, growth, and rupture, as well as in developing and testing new treatments. Creations of preferable models are expected.
Brain arteriovenous malformations (AVMs) are vascular anomalies where arteries and veins are directly connected through a complex, tangled web of abnormal vessels instead of a normal capillary network. There is usually high flow through the feeding arteries, nidus, and draining veins. AVMs represent a high risk for hemorrhagic stroke, leading to significant neurological morbidity and mortality in relatively young adults [
The management in the case of sudden bleeding is focused on restoration of vital function and prevention of recurrent hemorrhage, usually with some combination of surgical resection, embolization, and stereotactic radiotherapy. But all of these treatments pose a risk of serious complications, and the optimal treatment needs to be evaluated [
As considered to be embryonic origin and postnatal development, AVMs are highly dynamic rather than static [
Animal models are warranted to meet the needs mentioned above. Up to now, several experimental animal models have been developed in studying the AVM-related hemodynamics, pathogenesis, and treatments. Hence, a review was made about the background, the procedure, and the application of these models, and their advantages and disadvantages were briefly analyzed.
The aim of the review was to encourage creating more advantageous AVM models and promote further studies of the disorder.
We searched PubMed till December 31, 2014, using the term “arteriovenous malformation,” limiting results to animals and English language.
Two investigators read the titles and abstracts of the publications to find out the possibly relevant ones that described creations of AVM animal models or investigated AVM-related mechanisms and treatments using these models. The articles describing the creation of the dural arteriovenous fistula models or AVM lesions in other organs were excluded. Full texts of the selected articles were obtained, and those fulfilling our inclusion criteria were identified and finally summarized.
The emphases of the review were on the background, the procedure, and the application of each particular model. The chosen animals, the advantage, and the disadvantage of each model were also briefly discussed.
From the result of total 911 publications found according to the search term, we picked up more than 100 articles, by the inclusion criteria of either describing the creation of original or modified animal models or adopting these models to make experimental researches.
The animal models in the study of AVMs were diverse in accordance with research purpose, ranging from those based on the changes of the cerebrovascular circulation to those based on gene manipulation techniques. Eight different types of the original models were summarized and their highlights were shown in Tables
The highlights of the original models for AVMs.
Type | Author [Reference] | Year | Animal | Characteristics | Applications |
---|---|---|---|---|---|
Carotid jugular fistula (CJF) |
Spetzler et al. [ |
1978 | cat | |
|
Morgan et al. [ |
1989 | rat | |||
Bederson et al. [ |
1991 | rat | |||
Hai et al. [ |
2002 | rat | |||
Scott et al. [ |
1978 | monkey | |||
|
|||||
Intracranial arteriovenous fistula | Numazawa et al. [ |
2005 | dog | A venous graft shunting blood from a branch of the MCA to the SSS, the arterial territory as the blood stolen tissue surrounding AVMs | As above, more precisely in regional parenchyma, but not in the whole brain |
|
|||||
Rete mirabile (RM) as the AVM nidus | Chaloupka et al. [ |
1994 | pig | Inserting a needle to communicate the RM with the cavernous sinus | |
Massoud et al. [ |
1994 | pig | Establishing a CJF to retrogradely drain the blood from the RM | ||
Qian et al. [ |
1999 | sheep | |||
|
|||||
Venous plexus as the AVM nidus | Yassari et al. [ |
2004 | rat | Creating a CJF, arterialized venous vessels as an extracranial AVM lesion | To study molecular mechanism of AVM development and the effect of radiosurgery |
|
|||||
AVM-like lesions derived from implants | Pietilä et al. [ |
2000 | dog | A pedicled muscle graft implanted to the brain with an arteriovenous bypass | To emonstrate angiogenic mechanism of the AVM formation and development |
|
|||||
Xenograft arteriovenous fistula | Lawton et al. [ |
2004 | rat | Inserting an arterial graft from transgenic mice between the CCA and the EJV of nude rats | To evaluate the mechanism of radiotherapy and to develop novel therapies |
|
|||||
AVM tissue -implanted cornea model | Konya et al. [ |
2005 | rat | Transplanting human AVM tissues to the rat’s cornea | To evaluate the angiogenic property and its mechanism of human AVM specimens |
|
|||||
AVM lesions by gene manipulation | Details in Table |
AVM: arteriovenous malformation; MCA: middle cerebral artery; SSS: superior sagittal sinus; CCA: common carotid artery; EJV: external jugular vein.
The highlights of AVM models by gene manipulation.
Type | Author [Reference] | Year | Animal | Characteristics | Applications |
---|---|---|---|---|---|
AVM lesions by gene manipulation | Bourdeau et al. [ |
1999 |
mouse | Generating |
To investigate the pathogenic mechanisms of AVMs in genetic factors |
Oh et al. [ |
2000 |
mouse | Generating |
||
Xu et al. [ |
2004 | mouse | Focal virus-mediated VEGF gene transferred in the brain of |
To investigate the pathogenic mechanisms of AVMs in genetic and environmental factors | |
Hao et al. [ |
2008 | mouse | Focal virus-mediated VEGF gene transferred in the brain of |
||
Sung et al. [ |
2009 | mouse | Conditional knockout of |
To investigate the pathogenic and hemorrhagic mechanisms of AVMs | |
Choi et al. [ |
2014 | mouse | Conditional knockout of |
||
Walker et al. [ |
2011 | mouse | Focal virus-mediated Cre and VEGF gene transferred in |
To investigate the pathogenic mechanisms of AVMs and to test the potential treatments | |
Choi et al. [ |
2012 | mouse | Focal virus-mediated Cre and VEGF gene transferred in |
||
Chen et al. [ |
2014 | mouse | Conditional knockout of |
To evaluated the role of endothelia in the pathogenesis of AVMs | |
Mahmoud et al. [ |
2010 | mouse | Conditional knockout of |
||
Milton et al. [ |
2012 | mouse | Mating |
To investigate the hemorrhagic mechanisms of AVMs and to test the potential treatments | |
Choi et al. [ |
2014 | mouse | Mating |
||
Murphy et al. [ |
2008 | mouse | Induced overexpression of constitutively active |
To investigate the pathogenic mechanisms of AVMs in genetic factors | |
Yao et al. [ |
2013 | mouse | Generating |
||
Nielsen et al. [ |
2014 | mouse | Deleting |
AVM: arteriovenous malformation; VEGF: vascular endothelial growth factor.
To explain a phenomenon that brain tissue surrounding the AVM lesion is subject to swelling and hemorrhage immediately following surgical excision of the lesion, Spetzler et al. firstly suggested that the chronic ischemic brain tissue near high flow AVMs might experience a loss of vascular autoregulatory capacity, the theory of normal perfusion pressure breakthrough (NPPB), by using carotid-jugular fistula (CJF) model in cats [
Animal models with carotid-jugular fistulae. (a) Spetzler’s model, (b) Morgan’s model, and (c) Hai’s model. CCA: common carotid artery; ICA: internal carotid artery; ECA: external carotid artery; EJV: external jugular vein; IJV: internal jugular vein.
Therefore, a modified CJF model in rats was introduced by Morgan and colleagues. They made an end-to-end anastomosis of both rostral ends of the CCA and the EJV (the internal jugular vein in rats is hypoplastic, and the cerebral venous blood drains mainly to the EJV) on the right side and ligated the caudal ends of both vessels and the ipsilateral external carotid artery (ECA), creating a functional arteriovenous fistula between the circle of Willis and the right lateral sinus (Figure
Besides, “occlusive hyperemia” was also suggested to be related to the brain edema and hemorrhage following the large AVM resection. High blood flow and mass effect of AVM lesions might cause obstruction of the venous outflow and stagnation of arterial inflow in their adjacent parenchyma, with subsequent worsening of the existing hypoperfusion and ischemia in these tissues. Bederson et al. evaluated this presumption in a rat CJF model by a proximal CCA to distal EJV anastomosis with contralateral EJV occlusion [
Rats were mostly chosen as the model animal probably because they are economic and accessible in spite of their anatomical differences related to humans. CJF models were also tried in monkeys; however, they were hard to handle, expensive to create, and also with intricate ethical concerns [
Carotid-jugular fistulae resulted in the hemodynamic changes in whole brain or predominantly the hemisphere in the fistula side, but not in the regional parenchyma. A dog model with local cerebral hypoperfusion was tried using an intracranial arteriovenous fistula [
Both extracranial and intracranial arteriovenous fistula models lacked a real AVM nidus, these models were focusing on the hemodynamic and pathophysiological changes of AVM adjacent parenchyma, but not the AVM lesion itself.
The carotid rete mirabile (RM) of the swine is a special vascular structure with a tangle of microarteries and arterioles situated at the termination of each ascending pharyngeal artery (APA) as it perforates the cranial base. The two sides of the RM, which are connected with each other across the midline, are also supplied by other small collateral arteries and effuse to form internal carotid arteries ipsilaterally (Figure
Anatomic basis and features of the swine AVM model. (a) Schematic representation of the normal left carotid arterial anatomy of the swine. The carotid rete mirabile is situated at the termination of the APA. ICA: internal carotid artery; ECA: external carotid artery; CCA: common carotid artery; IMA: internal maxillary artery; MMA: middle meningeal artery supplying the ramus anastomoticus; RA: ramus anastomoticus; AA: arteria anastomotica; APA: ascending pharyngeal artery; OA: occipital artery; BA: basilar artery; CW: circle of Willis; EJV: external jugular vein. (b) Schematic representation of the AVM model after creation of a right carotid-jugular fistula. Arrows indicate direction of flow, that is, from the left CCA to both retia mirabilia via the three feeding arteries (the left APA, RA, and AA), and retrograde down the right APA toward the right carotid-jugular fistula. Note balloon occlusion of the right ECA.
To address this shortfall, Chaloupka et al. produced a high flow arteriovenous shunt in the swine RM by inserting a needle through the orbit to create communications between the rete and the surrounding cavernous sinus [
Massoud et al. developed a distinguished swine AVM model with induced high blood flow across both retia, by surgical formation of a side-to-side arteriovenous fistula between the CCA and the EJV with the ligation of the CCA proximal to the fistula on the right side [
Based on Massoud’s model, modified swine AVM models were introduced. They posed a higher pressure gradient closer to values found in human AVMs, thereby reducing the rate of spontaneous thrombosis in the rete [
Besides, in the pig, the natural structure of carotid RM is also seen in the other artiodactyl animals such as the sheep, goat, ox, and cat, but not in the dog, rabbit, and rat. Whether the swine RM models can be duplicated in the other animals was unknown, except for a feasibility study in the sheep [
Anatomic basis and features of the sheep AVM model. Arrows indicate direction of flow, that is, from the left side of the carotid artery through both retia mirabilia, retrograde to the right carotid artery and jugular vein following surgical creation of an anastomosis. CCA: common carotid artery; ECA: external carotid artery; IMA: internal maxillary artery; RA: ramus anastomoticus; AA: arteria anastomotica; EJV: external jugular vein.
In 2004, Yassari et al. described a rat model with the sham AVM nidus simply by ligating the left EJV at the confluence of the subclavian vein and making an end-to-side anastomosis of the EJV to the CCA [
The arteriovenous fistula of the rat arteriovenous malformation model. 1: fistula; 2: arterialized jugular vein; 3: nidus; CCA: common carotid artery; EJV: external jugular vein.
Further analysis in this model demonstrated that the nidus vessels underwent morphological changes from normal veins to those similar to immature vessels in human AVMs, including heterogeneously thickened walls, splitting of the elastic lamina, and thickened endothelial layers [
The activation of vascular cells in the nidus made it a unique model for studying the occlusive effect of radiosurgery on AVM vessels, because little was known about the molecular mechanisms of radiation mediated vascular obliteration. One study using the model showed that the expression of endothelial adhesion molecules in the nidus cells changed after radiosurgery [
Both the AVM lesions with simulated niduses using the RM and the venous plexus did not actually locate in the cerebral parenchyma. Pietilä et al. developed a novel model with an induced AVM lesion in the dog brain [
There were some highlight features of this model resembling the appearance of AVMs in human, including thickening and fibrosis of the draining venous wall, new formation of vessels, and vascular proliferation, surrounding brain tissues with signs of ischemia and hemorrhage. Although an exquisite surgical technology was required for producing the animal model, it might help discovering the pathological mechanisms involved in AVM development.
Currently, radiosurgery was a kind of less invasive treatment for AVMs. It took a therapeutic effect by obliterating the AVM nidus, with a low obliterating rate and a latency period up to 2 years. Further understanding of the mechanism of radiosurgery might be helpful to develop advanced pharmacological therapies to improve the occlusive effects based on conventional radiosurgery.
For this purpose, the xenograft arteriovenous fistula model was created, as a segment of main arteries from transgenic mice was interposed between the caudal end of the CCA and the rostral end of the EJV in immune-deficient nude rats [
In this model, the arteriovenous fistula with radiation pretreatment reproduced distinct radiation arteriopathy as observed in resected human AVM specimens pretreated with radiosurgery. If radiation pretreatment would result in a specific molecular change in the fistula graft, or if the fistula graft from different transgenic mice would have a different response to radiation, this model probably yielded clues to the vascular targeting therapy and the gene therapy. One study had detected that some robust but modified radiation responses occurred in Endoglin and eNOS knockout transgenic arteriovenous fistulae [
The model was technically feasible and the overall angiographic patency rate was about 50%. However, there was a time limitation of 4 months for allowing transplanted tissues to retain their phenotypes due to the rejection reaction.
The surgically resected human AVM lesions were valuable specimens for the histopathological study. When the specimens were transplanted into the corneal micropocket of the rats, they kept alive and growing. The angiogenic activity of the implanted tissues could be repeatedly measured according to a standard of neovascularization assessed by microvessel counts and VEGF expression [
Based on the model, the implanted AVM tissues showed the highest angiogenesis compared to other cerebrovascular disorders, cavernous malformation, and venous angioma, indicating that the AVM niduses were more likely to be active and progressive. The implanted AVM tissues previously treated with embolization exhibited the highest angiogenic activity, followed by untreated and gamma knife treated AVM tissues; this might explain why AVM recurrence after intravascular embolization was more common. Moreover, this rat cornea model containing human AVM tissues could be used for evaluating molecular mechanisms of the neovascularization process over time [
Hereditary hemorrhagic telangiectasia (HHT) is an autosomal dominant vascular disorder characterized by recurrent nosebleeds, mucocutaneous telangiectases, and AVM formations in the brain and other visceral organs [
Knockdown of
The conditional knockout technique with Cre/LoxP recombination system made it possible to delete target genes at the planned time or in the expected cells, because the Cre enzyme expression could be precisely controlled. Conditional deletion of both
Antenatal deletion of both
Other genes involved in angiogenesis would also be manipulated to create AVM models. Taking essential roles in vascular development and remolding, Notch signaling pathway was upregulated in human AVMs and might be an important molecular regulator of AVM pathogenesis [
The lack of matrix Gla protein (Mgp) also caused AVMs in mice. Cerebral enlarged vessels and direct connections between arteries and veins were detected in the
Cerebrovascular abnormalities, AVM formations, and hemorrhage occurred spontaneously in some cases where relevant genes were directly or conditionally deleted at the antenatal or postnatal stages, although in most cases, the model mice either displayed minimal vascular lesions or obvious vascular lesions out of the brain. The spontaneous cerebral AVM lesions partially simulated the natural clinical course of the disease, but the lesions lacked uniformity and reproducibility in size and location. Focal angiogenic stimulation based on gene deficiency helped to create adult onset models of induced AVM lesions in the brain. These models containing comparable AVM lesions might be more suitable for mechanism and therapeutic studies. In spite of posing disadvantages such as complicated procedures, high expanding, and being time consuming, the models by gene manipulation were unique for investigating the AVM pathogenesis and testing new therapies.
As shown in Tables
An ideal AVM model, which completely shared the same anatomic, physiologic, biological, and clinical features as human AVM disease, was lacking. Even the transgenic mice model carried out with spontaneous but systematic vascular malformation lesions could not fully represent the sporadic cases mostly seen in clinic. In spite of limitations, these various models provided assistance to answer particular questions in the study of AVMs.
The origin of AVM is still a mystery. It was generally believed that the vascular disorder was initiated during embryonic development. However, evidences from animal models demonstrated that postnatal formations of AVMs were possible, due to the two causal factors of angiogenic stimulation and gene deficiency. With genome-wide association study, investigators attempted to identify mutant genes associated with AVM susceptibility in sporadic AVM patients. The possible involved genes included Alk1, Eng, interleukin-6 (IL-6), and interleukin-1
The mechanisms that underlie AVM growth and progression remain poorly understood. Abnormally high blood flow and shear forces in nidal vessels activated molecular pathways in smooth muscle cells and ECs. Hypoperfusion and hypoxia in the nidal and surrounding tissues stimulated angiogenesis and inflammatory reactions. Both of them lead to vascular proliferation and remodeling [
Intracranial hemorrhage is the most severe and most common clinical presentation of AVM patients. Risk factors associated with AVM rupture include certain genetic mutations, intranidal aneurysms, exclusive or restricted venous drainage, deep or infratentorial location, and history of previous hemorrhage [
Among the conventional treatments, microsurgical resection is currently recommended for Spetzler-Martin Grades I and II AVMs. For high-grade AVMs, combined treatments are often used lacking a standard procedure. Given that the majority of high-grade lesions cannot be treated without relatively high morbidity and mortality, new biological therapies and gene therapies are under development aiming toward vascular remodeling. A study showed that losartan, an angiotensin II receptor antagonist, attenuated abnormal blood vessel morphology in the
We hope this review would provide the basic of currently available AVM models. The diverse techniques and methods displayed here might shed light on the creation of preferable AVM models in the future, overall promoting further studies of the disease.
The authors declare that there is no conflict of interests regarding the publication of this paper.
The paper was supported by grants of the National Natural Science Foundation of China (no. 81000489) and Shanghai Municipal Science and Technology Commission Foundation (no. 13140903300).