We would like to thank Dr. Massoud for his interest in our recently published arteriovenous malformation (AVM) model in sheep. As it was stated in our article, the purpose of the study was to evaluate the feasibility of creating an animal AVM model without the need for complex endovascular procedures. In his commentary of our article, Dr. Massoud made a strong defense of the swine model described by his group and criticized the sheep model as overly complicated and unrepresentative of a cerebral AVM. We will not even attempt to respond to all of his criticisms, because the response would need to be even longer than the original manuscript. We will simply try to delineate the rationale we followed in developing our experimental model and clarify some of his misunderstanding.
Despite some similarities to human AVMs, animal models are unable to reproduce the complex angioarchitectural features of human AVMs, such as arteriovenous shunts, multiple compartments in the nidus, tortuous feeders, intranidal vascular aneurysms, propensity to hemorrhage, variable hemodynamic resistance, abnormal venous drainage, “steal effect,” as well as the “perfusion pressure breakthrough” after treatment. Contrary to what Massoud likes to believe, this great variability in the morphologic and pathophysiologic features in AVMs makes it impossible to classify any one animal model as a “typical” AVM or to claim that it embodies most of the desired features. The fact is that, to date, none of the described animal models, including the swine, are truly representative of a human AVM.
Dr. Massoud emphasized in his original publication that endovascular occlusion of the right occipital, ascending pharyngeal, and the external carotid artery in the swine model was necessary to maximize blood shunting across the rete to the fistula. Their pilot study revealed that if these vessels were left unoccluded, they became major collateral pathways, diverting most of the blood flow away from the rete itself (1). Although Massoud argued in his letter that he eliminated the need for endovascular procedures in a subsequent study (2, 3), there was still the need to place a large 8F guiding catheter into the right carotid artery adjacent to the fistula in order to decrease the flow through this vessel and “thus allow for a greater contribution to the fistula from the cephalad aspect of the artery and thus increase the flow reversal down the drainage portion of the AVM model.” This was done in order to optimize the inadequate flow obtained through the rete in the swine.
Another “deficiency” of the sheep animal model is, according to Dr. Massoud, the existence of a single vessel connecting the two retia. We disagree with his interpretation of the images presented, because multiple vascular connections were clearly shown in the cast specimens as well as angiographically in our study (Fig 1A and B).
A, Photograph of the corrosion cast of the cerebrovascular anatomy on ventral view with 20° cephalad angulation shows two vascular connections in the posterior (curved arrow) and middle (straight arrow) portions as well as a dense vascular network connecting the anterior portion of the carotid retia. B, Close-up view of the retial structure on a left common carotid arteriogram after surgery confirms the vascular connections between both retia demonstrated on the corrosion cast
Dr. Massoud also pointed out in his letter that, in the sheep model, most of the flow diversion from the contralateral side to the fistula occurred at the level of the occipital arteries and not through the retia. We disagree with his interpretation and, in order to prove our point, we recreated the model in an additional animal and repeated the angiographic study after ligation of the occipital artery contralateral to the fistula. As expected, angiograms performed before and after ligation demonstrated, as expected, the same differences in density between the distal and proximal carotid artery (Fig 2A and B). Additionally, it was apparent in the postligation angiogram that the internal maxillary artery on the side of the fistula was densely opacified during the early phase, even before the carotid artery was opacified (Fig 2C), discounting Dr. Massoud's misconception that this segment had filled in a retrograde manner through the occipital artery. The relatively low density in the segment above the junction with the right occipital artery is easily explained, not as a result of diverted blood flow at the level of the occipital arteries, but instead as the final effect of the contrast column on the proximal and mid segments that are curved, as seen on the lateral projection, giving them a relatively higher density. This observation was demonstrated on the anteroposterior and lateral angiograms performed before and after surgery (4).
A, Postoperative left carotid arteriogram taken before occipital artery ligation shows the opacified left occipital artery (arrow). B, Left carotid arteriogram after occipital artery ligation. The early phase shows both retial structures and opacification of the internal maxillary artery (open arrow) on the anastomotic side. C, Late phase of carotid arteriogram shows opacification of the external carotid artery (long arrow) and jugular vein (short arrow) on the anastomotic side, confirming the direction of flow diversion through the retia toward the arteriovenous fistula
Dr. Massoud stated that the lack of validation of the morphologic and hemodynamic behavior of the sheep model made this model inadequate for scientific purposes. We would like to comment that, although hemodynamic assessments of the swine model, during AVM embolization by Massoud et al (2, 3), provided some interesting data (2, 3), the validity of these studies was questioned by several investigators (5, 6). Cockroff and Steinberg (5) expressed in their commentary that “the validity of the conclusions drawn depends on the model relevance to human AVMs.” They further expressed that “it remained to be determined whether these techniques will provide useful information on humans with complex AVMs.” Wakhloo et al (6), among others, also stated that the “use of blood flow velocity measurements in an AVM draining vein to quantify the extent of an embolization may have limited application.”
Rather than respond to all of Dr. Massoud's criticisms, which are mostly an ardent defense of “his model,” we think it is more important to keep in perspective that the importance of these models is their value for research and training in neurovascular interventions, because they do have arterial feeder(s), a nidal structure, draining vein(s), and high flow through the “AVM,” regardless of their morphologic characteristics.
References
- Copyright © American Society of Neuroradiology
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.