Principles of Neuroendovascular Management of AVMs: Goals, Timing, Techniques, and Outcomes

Introduction

Intracranial arteriovenous malformations (iAVMs) are among the most challenging and complex pathologies of the brain. After the poor results that were encountered with early attempts at the resection of iAVMs, Harvey Cushing was quoted as saying, “It would be nothing less than foolhardy to attack one of the deep-seated racemose lesions …. The surgical history of most of the reported cases shows not only the futility of an operative attack upon one of these angiomas but the extreme risk of serious cortical damage which it entails.” He was similarly pessimistic with respect to the more superficial variety of iAVM, saying, “…even with this latter and surgically speaking more favorable type, there is little encouragement to be had on the side of radical treatment.” The treatment challenges posed by iAVMs that have long been recognized by neurosurgeons and neuroscientists have motivated the search for less invasive, safe, and effective therapeutic options.

In 1960, Luessenhop and Spence described the first successful embolization of an iAVM; they used artificial embolic material and administered it directly via a surgically exposed left common carotid artery. Since then, several advancements have been made with regard to embolizing agents, access catheters, and balloon catheters. If the angioarchitecture of the iAVM is favorable, cure is possible with current endovascular techniques alone. Endovascular techniques are practiced in four different settings, each with an associated goal: (1) adjunctive—preoperative embolization to facilitate microsurgical resection or radiosurgery; (2) curative—embolization attempted for cure; (3) targeted therapy—to treat the source of bleeding; and (4) palliative—embolization to reduce arteriovenous shunting. In this chapter, we discuss goals, timing, techniques, and outcomes of neuroendovascular treatment of iAVMs.

Angiographic Evaluation of iAVMs

Digital subtraction angiography remains the gold standard for the evaluation of iAVM angioarchitecture. The goal of the angiogram is to identify the following features: feeding arteries; the location of the nidus and draining veins; the morphology, presence, and location of associated intracranial aneurysms; venous varices; and stenotic segments on arteries and veins. The identification of en passage feeding arteries is critical. This necessitates selective microcatheterization of the arterial pedicles. Arteriovenous shunting is confirmed by the visualization of an early draining vein during the arterial phase. A high-speed angiographic run with a higher number of frames per second may facilitate confirmation of the shunt. A 3D angiogram with 3D reconstruction is extremely valuable in understanding the architecture of the iAVM. It is important to note that a thrombosed AVM may not be detected on a cerebral angiogram. The iAVM nidus may also be missed on all imaging modalities if there is significant compression from an adjacent hematoma. Therefore it is prudent to repeat the angiogram once the hematoma and mass effect have resolved, usually in 2–4 weeks.

Other imaging modalities used to evaluate iAVMs include CT, CT angiography, MRI and MR angiography. These imaging modalities are limited in their sensitivity and ability to provide detailed imaging of iAVM angioarchitecture; however, each adds valuable information to aid the management approach. CT angiography provides better vascular detail of the iAVM, whereas MRI and MR angiography provide greater visualization of surrounding structures adjacent to the nidus. MRI is helpful in identifying thrombosed vessels as hyperintense signals and showing an associated hemorrhage at various stages of evolution. T2-weighted and gradient echo imaging sequences are the most sensitive to blood breakdown products. MRI can be important for preoperative planning for radiosurgery and microsurgery, as it helps to delineate the relationship of the iAVM with surrounding parenchymal structures.

Role of Classification in Patient Selection for Neuroendovascular Treatment

Spetzler-Martin grading remains the most widely accepted, reproducible, and utilized grading system for iAVMs. However, the Spetzler-Martin grade was developed to stratify the morbidity of microsurgical resection of iAVMs. Refinements of the original grading system have since been validated for radiosurgery, but Spetzler-Martin grading has not been validated to stratify the risk of neuroendovascular treatment of iAVMs. Developing a grading system to guide neuroendovascular decision-making has been a challenge because of variations in the location and size of the nidus and the number and size of the feeding arteries, location and number of draining veins, presence of associated aneurysms, rupture status of the iAVM, and source of bleeding, all of which influence neuroendovascular management. The considerable heterogeneity of iAVM angioarchitecture has prevented the development of a universally applicable classification system.

Several alternatives to Spetzler-Martin grading have been proposed, yet no single system has gained widespread acceptance. Several investigators from our institution, including the senior author of this chapter, developed a grading system (called the Buffalo score) that accounts for the anatomical challenges unique to the endovascular treatment of iAVMs. The grading features include number of pedicles, diameter of the arterial pedicles, and location of the iAVM (eloquent vs noneloquent) ( Table 15.1 ). The grading system was derived from morbidity data from consecutive iAVM patients treated by endovascular means and was validated using multivariate regression analysis, although it has not yet been validated externally. Fig. 15.1 illustrates the differences between the application of the Buffalo and Spetzler-Martin grading systems. Another classification system was proposed by Feliciano et al. That grading system was based on a literature review and included factors such as number of pedicles, eloquence, and the presence of a fistulous component. None of these classification systems correlated with clinical outcomes in an independent cohort study by Gupta et al. Moreover, they are not applicable to transvenous iAVM embolization. Application of these classifications necessitates the evaluation of the fistulous component with selective microcatheterization of the iAVM. Although all these classifications may have clinical relevance, the selection of a treatment modality must be tailored to the individual case.

Curative Embolization

Endovascular options have largely been considered as adjunctive to microsurgery or radiosurgery; however, numerous small- and intermediate-size iAVMs can be completely obliterated with embolization alone. Cure may be feasible for small- to medium-size superficially located iAVMs with a compact nidus and pedicles accessible with a microcatheter that allows reflux of the embolic agent Onyx (Medtronic, Minneapolis, MN) for 2–3 cm and have recognizable proximal parts of the draining veins emerging from the nidus. Large iAVMs with feeders from multiple territories, deep location (such as the basal ganglia or brainstem), perforating artery feeders, or a diffuse nidus are considered unfavorable for endovascular iAVM cure via transarterial access.

Advances in devices and techniques for transvenous embolization have broadened the horizon of endovascular management of iAVMs. Some iAVMs that are not amenable to transarterial cure may be cured with transvenous embolization alone or in combination with a transarterial approach. Several studies have now reported the safety and feasibility of transvenous embolization. In fact, the cure rate associated with transvenous embolization is higher than that associated with a transarterial approach. However, compared to the transarterial approach, transvenous embolization is technically more challenging. This is because of difficulty with selective catheterization of the draining vein and delivery of embolisate into the nidus in a retrograde manner, against the high fistulous flow, without incurring nidal rupture and hemorrhage. Various strategies have been applied to reduce the high-flow shunting in order to facilitate the transvenous embolization procedure. These strategies include systemic hypotension and temporary balloon occlusion of arterial feeders.

We have described our experience with transvenous embolization using cardiac pacing to control the flow of blood through the iAVM. Transvenous iAVM embolization was successfully performed in 10 of 12 patients. (The procedure was abandoned in 2 patients because of technical difficulty.) Five of the 10 AVMs were infratentorial. All 10 had a single primary draining vein. Rapid ventricular pacing was used in 9 cases; intravenous adenosine injection was used in 1 case to achieve cardiac standstill. Complete AVM nidus obliteration was achieved in all 10 cases, with excellent neurologic outcome in 9 cases; it was accomplished with transvenous embolization alone in 2 of 10 cases and with staged transarterial followed by transvenous embolization in the others. Two of the successfully treated patients developed hemorrhagic complications intraprocedurally. One patient’s iAVM was managed conservatively and the other underwent surgical treatment, with AVM excision and hematoma evacuation; both made an excellent recovery and neither had any neurological deficit at 3 months after the embolization.

Transvenous iAVM embolization is suitable for small iAVMs (≤ 3 cm) with a single primary draining vein where transarterial embolization is not feasible due to lack of a definite arterial pedicle, presence of multiple small feeders, supply by tiny perforating arteries, or en passage feeding arteries. A combination of transarterial and transvenous approaches can be used in a concurrent or staged fashion. We prefer to perform staged embolization for larger (> 3 cm) iAVMs, to reduce the size of the nidus. Transvenous embolization may be considered if the iAVM is not obliterated after staged transarterial embolization and its features are favorable for a transvenous approach.