Endovascular Management of Arteriovenous Malformations for Cure

Published on 13/03/2015 by admin

Last modified 22/04/2025

Print this page

This article have been viewed 1195 times

CHAPTER 387 Endovascular Management of Arteriovenous Malformations for Cure

Robert M. Starke, Sean D. Lavine, E. Sander Connolly, Jr., Philip M. Meyers

Brain arteriovenous malformations (AVMs) are congenital vascular malformations and can be located in any part of the brain. The vascular nidus is made up of a complex tangle of abnormal veins and arteries devoid of an intervening capillary bed. Artery and vein are thus directly connected, which results in fistulous blood flow between them and is seen as “shunting” on a catheter arteriogram. Frequently, AVMs are asymptomatic and go undetected unless a clinical event such as hemorrhage or seizure takes place. Hemorrhage is the most common cause of functional/cognitive changes in patients with AVMs. Pathologic data suggest that approximately 12% of AVMs become symptomatic, whereas others are identified either incidentally on an imaging study or at autopsy.

The goal of treating AVMs is to eliminate the risk for hemorrhage and to preserve or improve functional neurological status. There are currently four major treatment options. The lesion can be monitored expectantly with the acceptance that the patient has a risk for the development of hemorrhage, neurological deficit, or seizure. Alternatively, one can attempt to eradicate the AVM. Microsurgery, radiosurgery, and embolization have all been used both alone and in various combinations to achieve eradication. In choosing between these various approaches, one must weigh the likelihood of total and permanent angiographic obliteration of the lesion against any delay in achieving the desired obliteration.

Early results with endovascular embolization of AVMs were disappointing. Selective access to most cerebral AVMs was not possible with the equipment available. Untargeted embolization was attempted with the use of Silastic beads or other objects.¹ High flow, it was hoped, would carry most of the embolic material to the AVM vessels.² Off-target occlusion occurred frequently, and procedural complication rates were significant. Since then, increased understanding of AVM morphology and significant refinements in endovascular technique have improved the risk-to-benefit ratio. Endovascular therapy has now become an important component in the therapeutic management of many brain AVMs. Currently, embolization is used in a variety of manners, including (1) as adjuvant therapy before definitive microsurgery or radiosurgery, (2) as palliative therapy for inoperable or otherwise incurable AVMs, and (3) as curative therapy (i.e., without surgical resection or radiotherapy). It is the curative use of embolization that we now consider.

However, before discussing curative embolization, it should be noted that no decision for treatment of any lesion as complex and heterogeneous as a brain AVM should be made in isolation. Rather, a multidisciplinary team made up of physicians with expertise in not only endovascular embolization but also microsurgical resection and radiosurgery should consider all available treatment options carefully. When decisions are made in this manner, definitive cure of brain AVMs by endovascular embolization as the sole treatment modality, however, is found to be desirable in only a select minority of cases. The reason is that the angioarchitecture of the AVM must permit solid casting of the AVM nidus with permanent embolic material in such a way that the draining vein is occluded only after the nidus is completely occluded. Failure to completely occlude the nidus first can lead to disastrous bleeding complications. This fact, combined with the frequently availability of less risky treatment alternatives, usually results in another treatment modality being chosen even when the anatomy is permissive.³^–⁷

Interventional Planning

Patient-Related Factors

Analysis of treatment options begins with patient-related factors such as symptoms, past medical history, psychological status, and family and social background. As with any other therapeutic decision making, the choice of intervention, with all the risks included, must be assessed in terms of the patient’s needs. A select number of patients may be unwilling to undergo surgery or, alternatively, are incapable of dealing with the anxiety of the delayed obliteration involved in radiosurgery. As with any type of surgical planning, the patient’s age, comorbid conditions, and functional status must be taken into account to assess operative risks. AVM patients tend to initially be seen at a younger age and with fewer medical conditions than patients requiring other vascular surgery, but they may have serious comorbidity as a result of the AVM itself. Particular attention must be paid to the patient’s clinical findings, and the highest priority is often the estimated risk for intracranial hemorrhage or rebleeding, as discussed in previous chapters. In patients with significant comorbid conditions precluding major general anesthesia and microsurgery and in patients with a significant risk for hemorrhage or recurrent hemorrhage, primary embolization may be the best treatment modality.

Architecture of the Arteriovenous Malformation

Patient characteristics must be used in conjunction with AVM architecture to formulate a treatment plan. Along with a thorough clinical examination, all AVM patients require detailed preoperative radiographic clarification of AVM anatomy and architecture and associated aneurysms. Magnetic resonance imaging is beneficial in elucidating the overall structure of the AVM and is the best modality to determine its anatomic location. The size of the nidus, proximity to eloquent parenchyma, and frequently deep venous drainage can also be determined to calculate the Spetzler-Martin grade. In addition, it can provide information about the best approach for endovascular management and the history of hemorrhage. Superselective digital subtraction angiography is then necessary to fully disclose the angioarchitecture of the AVM. A number of radiologic variables have been associated with an increased risk for hemorrhage, including small AVM size, elevated feeding artery pressure, periventricular or intraventricular location, basal ganglia location, deep venous drainage, impaired venous drainage, single draining vein, intranidal aneurysm, multiple aneurysms, and vertebrobasilar blood supply.⁸^–¹⁶ These parameters also influence the possible success of endovascular therapy because they determine the accessibility of the lesion with the microcatheter system, the number and type of feeding arteries, and the size of the AVM nidus and its hemodynamic properties.

Accessibility

Determination of accessibility of an AVM nidus starts with a patient’s past medical history and comorbid diseases. The extracranial or intracranial vasculature in patients with a history of congenital aortic or great vessel malformation or occlusive disease may preclude microcatheter access to the AVM. Further assessment during angiography is imperative because embolization in patients with AVM feeders that cannot be catheterized superselectively carry an increased risk for inadvertent occlusion of nontarget vessels and resultant stroke. If nidal embolization is not performed, embolization often remains incomplete. Angioarchitectural features that predict accessibility to an AVM with a microcatheter are the length, tortuosity, and diameter of the individual feeding vessel or vessels. Over time, hemodynamic alterations may lead to vasculopathic changes that inhibit microcatheter access, such as increased tortuosity, stenosis, or occlusion. Superselective angiography is useful to determine whether feeders terminate in the AVM nidus or continue distal to the AVM and supply normal brain parenchyma. Occlusion of these en passage vessels could lead to stroke if occluded in an effort to treat the AVM.

Feeding Vessel Morphology

The vascular supply to an AVM can be derived from pial and dural arteries. In some cases, this dual vascular supply can lead to confusion of brain AVMs for dural-based arteriovenous fistulas. Selective arteriography of intracranial and extracranial vessels can help distinguish between these dural- and pial-based shunts. Dural arteries contribute to AVMs in approximately 30% of cases.^17,¹⁸ Perforating arteries and choroidal feeders can supply deep-seated brain AVMs and lesions in the vicinity of the ventricular system. Care must be taken during endovascular occlusion of choroidal arteries, which frequently also supply important neural structures. Data on endovascular occlusion as the sole modality for the treatment of AVMs are limited. In one study of 307 patients undergoing 513 embolization sessions for curative, adjuvant, and palliative embolization of brain AVMs, accessibility was possible in only 415 embolization sessions (80%).¹⁹ Additionally, 10 of 62 complications (16%) occurred during microcatheter exploration before embolization and led to 1 death and 1 permanent neurological deficit. Similarly, in a study of patients undergoing embolization as the primary treatment modality for intraventricular or paraventricular lesions, 12 of 14 patients (86%) had lesions accessible to embolization.²⁰ In a series of 387 patients, Valavanis and Yasargil reported complete obliteration of 60% of sulcal AVMS but just 12.5% of gyral AVMs with endovascular therapy.²¹ The number of feeding arteries is a crucial determinant of the success of both delineation of the AVM and treatment. In one study in which the goal was primary curative embolization, the predefined criteria adopted for attempted curative embolization were based on radiologic interpretation of the technical feasibility and likelihood of complete embolization. AVMs with an increased likelihood of curative embolization had (1) a nidus that was accessible with the tip of the catheter, (2) three or fewer arterial feeders, (3) and a nidus that was not larger than 3 cm.²² In patients with angioarchitecture that makes vascular isolation of the AVM from surrounding brain particularly difficult, the risk-to-benefit ratio must be reassessed with regard to the expected therapeutic benefits of embolization. Although preoperative endovascular occlusion of AVMs is desirable in many cases, microsurgery or radiosurgery alone has a superior risk-to-benefit ratio.

Size

In 1971, Doppman and coworkers introduced the term nidus, which refers to the area between the distal segment of the feeding arteries and the proximal segment of the draining veins where the arteriovenous shunt occurs.²³ According to Yasargil, the AVM nidus can be categorized into seven different types primarily based on size: (1) the nidus may be occult (not seen angiographically and not located at surgery); (2) the nidus may cryptic (not visible with angiography or at surgery but recognized histologically); (3) the nidus may be microscopic (visible as a site of arteriovenous shunting on angiography but not apparent on direct inspection of the brain); and (4) the nidus may small (1 to 2 cm), (5) moderate (2 to 4 cm), (6) large (4 to 6 cm), or (7) giant (>6 cm).²⁴ Improvements in high-resolution angiography^25,²⁶ have led to increased delineation of AVM angioarchitecture and most likely an increase in the number of AVMs in the microscopic category diagnosed.

The size of an AVM nidus influences the difficulty of surgical resection and radiosurgical obliteration and has been included in most classifications of AVMs.^6,²⁷ Size is also a critical determinant in planning primary curative embolization. Smaller brain AVMs are often more amenable to complete endovascular occlusion without increasing the rate of procedure-related complications. Studies have found that primary curative embolization was predominantly possible with brain AVMs less than 3 cm in diameter or smaller than 4 cm³.^{19–22,28–33} Future developments in embolic material and delivery catheters may change these assertions over time.

Larger AVMS frequently have additional dural and deep perforating arterial feeders. The therapeutic result of embolization may be limited by the accessibility of small perforating arteries (thalamoperforating, lenticulostriate, and choroidal arteries) and recanalization or vascular neogenesis in dural feeders. Larger AVMs with high rates of flow are prone to the development of stress-related vasculopathic changes, including aneurysm formation, tortuosity, stenosis, and occlusion, and result in insurmountable navigational challenges, even with current sophisticated microcatheter systems.

Hemodynamic Properties

In addition to the angioarchitecture of the nidus and feeders, the hemodynamic properties of an individual AVM nidus are crucial for successful endovascular embolization. The number, morphology, and velocity of the arteriovenous shunt determine whether embolic material can be deposited selectively and safely in the nidus. The nidus of a brain AVM can be compact or diffuse, although the diffuse angiographic appearance of a perinidal hypervascularity is often related to high-flow arterial angiopathy rather than diffuse segments of the AVM nidus. An AVM nidus can have multiple compartments and can be plexiform, fistulous, or mixed. In a plexiform nidus, selective angiography discloses a multitude of intranidal arteriovenous shunts. The diameter of these vessels is usually small, and the velocity of the shunt is generally low. In contrast, fistulous lesions lack the plexiform area of the nidus but typically involve a single-hole, high-flow arteriovenous fistula shunting directly into draining veins. This pattern of rapid arteriovenous shunting can be found in conjunction with a plexiform nidus in larger, mixed lesions and has a bearing on the method of treatment. The angioarchitecture of the AVM nidus and the velocity of the arteriovenous shunt are crucial determinants of whether lesions can be embolized totally and permanently. Haw and coauthors reported that a pure fistula or a nidus with a fistulous component was a significant predictor of complications after endovascular therapy.¹⁹ This stands in contrast to Valavanis and Christoforidis, who reported a dominant fistulous component in the nidus as being predictive of endovascular cure.³⁴ In mixed fistulous and plexiform lesions, an unrecognized fistulous component can result in embolization and occlusion of the draining vein with catastrophic hemorrhagic complications.¹⁹ When a single pathologic fistula is present, the lesion may be best suited for endovascular therapy with a high cure rate,³³ but these lesions result in a number of technical difficulties. According to Haw and colleagues, a wedged microcatheter position may not be possible, which decreases the accuracy of glue placement and may require increased glue concentrations and injection pressures. Rapid blood flow may increase the risk of placement of glue into the draining vein or reflux into normal arteries. Finally, these fistulas often require stiffer and larger diameter over-the-wire microcatheters, which increases the risk for arterial injury. An adequate distance between the microcatheter tip and the arteriovenous fistula is critical for safe embolization of these high-flow lesions.¹⁹

Technical Aspects

As outlined, the endovascular accessibility of the AVM greatly reflects the angioarchitecture of its feedings arteries. Significant improvements in the field of microcatheter technology now permit the superselective catheterization of most pial AVM feeders. An understanding of the different catheters, embolic agents, and endovascular techniques is critical to increase the opportunity for obliteration while decreasing complication rates.

Catheter Selection

Early attempts at catheterization and occlusion of blood vessels were made in aneurysm patients via an open surgical approach, which was associated with many of the same risks and morbidity as open neurosurgical procedures.¹ The first microcatheters that allowed access to AVM feeders were flow directed. Mounted calibrated leak balloons helped control flow and allowed the release of contrast material and cyanoacrylate embolic material, but these balloons were easy to overinflate and had a high rate of vessel perforation.³⁵

Subsequently, a torqueable guidewire system that allowed passage of a catheter was developed (Advanced Cardiovascular System, Santa Clara, CA). Modifications of this new generation of microcatheters, including softer distal segments, improved their versatility.^36,³⁷ A thick-walled proximal catheter segment controls torque and transmits longitudinal movements, the intermediate catheter segment is flexible but still transmits torque, and the distal segment is soft and thin walled. There are now many vendors of microcatheters designed for intracranial navigation. However, the greater tortuosity and fragility of cerebral arteries than extracranial systemic arteries requires specialized training for safe navigation.

Flow Control

When a microcatheter has been positioned in the feeding artery or nidus of an AVM, an individual compartment can be cast if surrounding blood flow can be controlled. As discussed, high-velocity arteriovenous shunting, unsuitable nidal architecture (i.e., fistulous single-hole compartments), or both can increase the risk for off-target embolization. If the angiogram shows congestion of the nidus (i.e., slow or stationary dye) after injection of embolic material, downstream (i.e., venous outflow) occlusion may have occurred. High-grade stenosis of draining veins is a risk factor for AVM rupture.²¹ Small amounts of glue passing through the shunt can occlude the stenotic portion of the draining vein. In multicompartmental AVMs with multiple feeders and one draining vein, casting of one compartment, including the venous outlet, can obstruct the draining system for the other compartments. Because there is no significant stromal support for pial arteries, particularly in and around the AVM nidus, hemorrhage is likely to occur. For this reason, venous occlusion must be avoided.

To overcome these difficulties, several techniques were historically proposed. First, some microcatheter designs permitted control of flow in the feeder during embolization (calibrated leak balloon catheter). Second, soft platinum wires (available in straight and coiled forms) were designed to be injected through the microcatheter in high-flow fistulas before glue embolization to slow the torrential blood flow.^38,³⁹ These liquid coils (Boston Scientific/Target Therapeutics, Fremont, CA) remain in the fistula site and prevent a large amount of glue from migrating through the fistula. Finally, embolization can be performed with the patient under induced systemic arterial hypotension. Although liquid coils are no longer manufactured and systemic hypotension is rarely if ever used, these techniques remain historically significant.

Embolic Agents

Cyanoacrylates

Cyanoacrylate polymers are the most widely used of the liquid agents. Zanetti and Sherman first introduced polymerizing cyanoacrylates as liquid embolic agents for endovascular embolization.⁴⁰ Because of their low viscosity, cyanoacrylates can be delivered through small, flexible, flow-directed catheters that can easily be manipulated to allow penetration deep into the AVM nidus. They can also be delivered quickly with infusion in generally less than 1 minute. Cyanoacrylates have high resistance to in vivo biodegradation.^41,⁴² An intensive inflammatory response helps maintain vessel occlusion and stimulates fibrosis. Although recanalization can occur, it is rare after complete embolization. The risk for recanalization is increased if there is proximal occlusion with minimal or no penetration of the nidus.³¹ In these cases in which the nidus remains active, collateral changes commonly develop and revascularize the nidus; such changes may be unfavorable for future endovascular occlusion and present additional challenges during surgical resection of the AVM.

The first cyanoacrylate used was isobutyl-2-cyanoacrylate, but it was discontinued after studies demonstrated that the agent possessed some carcinogenic potential in animals.³² Histoacryl-blue (N-butyl-2-cyanoacrylate [NBCA]) is now the embolization material of choice. NBCA is mixed with low-viscosity oily contrast media (e.g., Ethiodol ultrafluid), with or without additional tantalum or tungsten powder for radiologic visualization. NBCA polymerizes immediately on contact with free hydrogen ion, but the catheter is rinsed with dextrose to prevent premature initiation of such polymerization. The main disadvantage of NBCA relates to its adhesive properties and the risk of adherence of the catheter to the adjacent blood vessel wall, which can cause injury to the vessel or inhibit removal. Modifications of the cyanoacrylate formula endeavor to reduce its adhesiveness and increase its cohesiveness for AVM embolization.

EVOH Copolymer-DMSO Solvent

In the late 1980s, ethylene vinyl alcohol polymer dissolved in dimethyl sulfoxide (EVOH-DMSO) was first used for the embolization of AVMs.⁴³ This liquid precipitate is similar to NBCA and is currently available in the United States as Onyx (ev3, Plymouth, MN). This agent is viscous and precipitates in the biologic environment as the DMSO solvent diffuses into the lipid-rich environment. Although Onyx can be used to fill AVM vessels, it does not specifically adhere to the delivery catheter. This property permits more time for operator control over the volume and rate of delivery, yet ongoing fluoroscopic observation and periodic control angiography to assess the status of obliteration of the AVM are still necessary.

Preliminary studies found that DMSO induced vasospasm and angionecrosis.^44,⁴⁵ Further studies found that decreasing the volume of DMSO and its rate of introduction could reduce these effects.⁴⁶ In a study of 23 AVM patients, Jahan and coauthors reported one complication related to distal vasospasm that developed during injection and histopathologic evidence of angionecrosis in two AVM specimens resected 24 hours after embolization.⁴⁷ A recent study reported angionecrosis in 42.9% of patients, but subsequent studies have not commented on this effect.⁴⁸^–⁵² Although long-term data are lacking, Murayama and associates demonstrated no recanalization in swine after 6 months of follow-up,⁴⁶ and Jahan and colleagues reported no recanalization in a limited number of patients imaged up to 20 months after embolization.⁴⁷ In a series of 47 patients, Weber and coworkers reported an initial 84% nidal reduction, with recanalization seen in 11% on angiography at 3 months.⁴⁹ A study of patients undergoing microsurgical resection after Onyx embolization reported recanalization of embolized vessels in 14.3% on pathologic inspection,⁵³ but angiographic or pathologic recanalization rates have not been reported in other recent studies.^48,⁵⁰^–⁵² The use of Onyx is not without complications. Authors have reported cerebral artery perforation because relatively stiff-walled microcatheters must be used that can resist the high pressure needed to inject Onyx.⁴⁹ Catheters tend to become lodged in the Onyx cast as the polymer precipitates around the shaft, even though the precipitate is nonadhesive.⁴⁹ Embolization with Onyx requires increased fluoroscopy and procedure times in comparison to NBCA, and further investigation is needed to justify the increased radiation exposure and procedure time associated with Onyx.⁵⁴ Because Onyx remains relatively new at the time of this writing, its role in the treatment of cerebral AVMs remains to be defined.

Polyvinyl Alcohol

Polyvinyl alcohol (PVA) is the most commonly used particulate agent historically for the embolization of AVMs. PVA particles (50 to 1000 µm) can be used in various size ranges at the discretion of the operator. A microcatheter with a diameter large enough to accommodate the particle size selected is necessary to prevent occlusion of the microcatheter. The use of larger catheters decreases flexibility in the tortuous cerebral vasculature and makes superselective angiography of small or distal vessels more difficult.⁵⁵

Recanalization is also more likely to occur with particulates than with cyanoacrylates,⁵⁶^–⁵⁸ and therefore embolization with particulates is indicated only for preoperative embolization when surgical resection will follow closely. In a randomized clinical trial comparing embolization with NBCA versus PVA, the trial investigators found no difference in the degree of nidal reduction, number of vessels embolized, surgical resection times, transfusions, fluid replacement, or patient outcomes. There was a significant increase in the incidence of postsurgical hematomas in the PVA group.⁵⁵ This study led to approval of NBCA by the Food and Drug Administration for preoperative occlusion of brain AVMs.

Alternative Embolysates

A number of other agents have been used for the embolization of AVMs, including Silastic pellets, blood clots, dura, silk sutures, and Gelfoam.⁵⁹^–⁶² Undiluted absolute ethyl alcohol (98% dehydrated alcohol injection—U.S. Pharmacopeia) has also been used for the embolization of AVMs. Of the 17 patient treated by embolization, 7 patients were cured by embolization alone, 3 patients after microsurgery, and 1 additional patient after radiosurgery. Despite these promising results, 2 patients with partially treated lesions died and 8 patients experienced complications related to therapy. Absolute ethanol also results in significant brain edema and can precipitate vasospasm. These significant side effects have not led to the widespread use of ethanol for embolization of brain AVMs.

Balloons and microcoils usually occlude vessels proximally rather than distally, but their role can be beneficial in the overall obliteration of AVMs. Balloons are no longer available in the U.S. market.

Risk Assessment

After a comprehensive evaluation has been performed, decisions can be made regarding the best management approach by comparing the natural history of the lesion with the intervention-related morbidity and mortality. When information about the natural history of AVMs is combined with knowledge of a patient’s characteristics and AVM morphology, a risk profile for future hemorrhage or rebleeding may be developed. A critical question is how the natural history for a particular patient of a given age with specific AVM architecture compares with the risk associated with treatment. The answer to this question helps determine the recommendation of whether to treat a particular AVM and, if so, with which treatment method or combination of methods.

Before endovascular therapy for an AVM, it is important to have a treatment plan, or strategy, and an understanding of the goals. This includes an understanding of the risks, benefits, and alternatives to embolization and the relationship of embolization to microsurgery and radiosurgery. The plan should include delineation of the targets for embolization, the best method of approach, optimal microcatheter selection, and appropriate embolic material.

Complications

In the literature there has been a wide range of reported morbidity (0% to 50%) and mortality (1% to 4%) after embolization of AVMs.^7,^{28–31,47,63–85} Rates are dependent on many factors: patient selection, embolic agents, means of delivering the occlusive agent, preexisting neurological morbidity, the goals of embolization (i.e., primary curative or adjuvant therapy), and the time of outcome assessment.* Typically, neurological deficits occur in 10% to 14% of patients, disabling deficits in 2% to 5% of patients, and death in approximately 1%.^19,^20,^22,^29,^55,^56,^65,^66,⁸⁷^–⁹³ Recent studies using EVOH-DMSO have reported similar results.^47,⁴⁹^–⁵³ When embolization is used as primary treatment, adjuvant therapy, or palliative therapy, the following factors are predictive of postembolization deficits: number of embolization procedures, number of arterial feeders embolized, eloquent location, location in the basal ganglia, deep venous drainage, size smaller that 3 cm, size larger than 6 cm, Spetzler-Martin grade III to V, venous penetration of the glue cast, and AVMs with a pure fistula or a nidus with a fistulous component.^19,⁸⁸^–⁹⁰ ^,94 We have found that that a significant number of patients experiencing deficits from embolization of brain AVMs will exhibit improvement over time.⁹⁴

Additionally, a patient’s age and general medical condition can alter the expected procedural risks. Although previous studies have not found these factors to be significantly predictive of postembolization deficits, this is most likely due to patient selection. The existence of atherosclerotic vessels increases the risk for embolic complications from dislodged atherosclerotic plaque. Additionally, risks from inappropriate endovascular technique and insufficient neurointerventional equipment are evident and not discussed.

The most common and serious complication of embolization is intraprocedural or delayed hemorrhage. Hemorrhagic complications associated with embolization occur in 1% to 2% of procedures (3% to 15% of patients).^73,^92,^94,⁹⁵ Hemorrhage during or after embolization may be due to perforation of the vessel during catheterization, rupture of the AVM, rupture of an intranidal aneurysm, or hemodynamic changes caused by alterations in feeder pressure, normal perfusion pressure breakthrough, or obstruction of venous outflow.^92,⁹⁶^–⁹⁹ We have found that intraembolization or postembolization hemorrhage is more likely with AVMs smaller than 3 cm in diameter or larger than 6 cm in diameter.⁹⁴ Small AVMs may be technically more difficult to embolize than medium-sized AVMs. The increased propensity of small, untreated AVMs to bleed has been demonstrated in many studies and may be due to increased pressure within small AVMs.¹⁴^–¹⁶ Deficits and hemorrhage are more likely to occur after the embolization of larger AVMs because it is more difficult to achieve complete embolization. Picard and coauthors reported that a volume of glue injected greater than 1 mL, venous embolization, and postembolization venous stagnation in or around the nidus were predictive of hemorrhage after embolization.⁹⁵

Complications rates when embolization is used as a primary modality have been similar in patients achieving complete obliteration.¹⁹^–²² These AVMs are more likely to be accessible and small and have fewer feeding vessels. Although accessibility and decreased number of feeding vessels are associated with decreased complications, we have found small (<3 cm) versus medium (3-6 cm) size to be predictive of neurological deficits, intraprocedural hemorrhage, and infarction after embolization used as adjuvant therapy to microsurgery or radiosurgery.⁹⁴

Given the therapeutic options available for patients harboring a brain AVM (i.e., open surgery and radiosurgery), the potential procedure-related risks associated with embolization must be compared closely with the expected risk-to-benefit ratio. If total endovascular occlusion of the AVM can be achieved only with a high risk for procedural complications, combined or other treatment strategies must be considered (i.e., embolization in conjunction with surgery or radiosurgery).

Outcomes

In most published series, the number of totally and permanently embolized AVMs is small. In overall cohorts of patients treated by embolization, success rates of curative embolization are mainly in the range of 5% to 20% but vary greatly from 0% to 70%.*

In an analysis of 1246 cases from 1960 to 1995, the overall cure rate was 5%. There was no significant difference between cure rates of AVMs published before 1990 (n = 708, 4%) and those published after 1990 (n = 538, 5%). In a study by Fournier and associates, 7 (14%) of the 49 patients exhibited morphologic cure by embolization alone as shown on 2-year follow-up angiograms.³¹ In a series of 101 patients, Vinuela and coauthors reported a 9.7% cure rate with embolization alone. Cures were achieved in patients with small AVMs and few feeding pedicles.²⁹ In a series of patients in whom curative embolization was primarily used for high-grade AVMs, Deruty and colleagues achieved a cure rate of just 5%.⁶³ In a series of 125 patients undergoing embolization before radiosurgery, Gobin and collaborators reported a cure rate of 11.2%. Increasing feeding pedicles and AVM volume were also inversely related to complete obliteration.³⁰ In a selective referral group of 150 patients, Wikholm and coauthors reported a cure rate of 13.3%. Again, size was the critical determinant of occlusion; 71% of the AVMs smaller than 4 cm³ were obliterated versus 15% of AVMs 4 to 8 cm³.⁸⁴ Willinsky and coworkers reported the results of embolization of small brain AVMs. Although only a minority of these patients had deep-seated brain AVMs, the cure rate with embolization alone was 27%.³³

Curative embolization rates have been similar in patients treated with Onyx and NBCA and range between approximately 15% and 50%.⁴⁹^–⁵² ¹⁰² In a study of 93 patients treated with Onyx, Weber and associates reported that embolization was curative in 20% of patients. Onyx resulted in high occlusion rates (volume reduction >90%) when the AVM was in a supratentorial and cortical location, the nidus was compact and plexiform, and there was a small number of supplying (direct) feeders and one superficial draining vein. Onyx did not achieve a high occlusion rate (<70%) in AVMs with multiple compartmental draining veins, in AVMs with multiple supplying arteries (especially leptomeningeal, en passant, or perforating feeders), and in AVMs with a diffuse nidus.⁴⁹ In a study of 44 patients treated with Onyx, van Rooij and coauthors reported curative embolization rates of 16%.⁵¹ In a study by Mounayer and colleagues of 94 patients treated with Onyx over a span of 210 sessions, treatment was concluded in 53 patients at the time of publication. In this select cohort, curative embolization was achieved in 26 of 53 patients (49%).⁵⁰ In a study of 101 patients treated with Onyx by Katsaridis and associates, only 52 patients had received complete treatment and further embolizations were planned in 49 patients. In the select number of patients receiving complete treatment, embolization was curative in 28 of 52 patients (54%).⁵²

In locations that make microsurgery particularly difficult, embolization may be an alternative in carefully selected patients. AVMs of the basal ganglia and thalamus may have a higher annual hemorrhage rate ¹⁰³ and worse clinical course.¹⁰⁴ In a study by Hurst and colleagues of 14 AVMs located in the thalamus or basal ganglia, cure by embolization was achieved in 15% of cases.⁷² In a study by Paulsen and collaborators of 38 AVMs in the thalamus or basal ganglia, 1 AVM was cured by embolization alone and 14 of 37 lesions were cured by combined modalities.⁷⁹

It is important to note that in the series just described, patients were often treated with embolization as an adjuvant to radiosurgery or microsurgery. The patients typically had significant reductions in size that facilitated microsurgery or radiosurgery. The rate of complete obliteration may be different in patients specifically selected for curative embolization. In a study by Yu and colleagues, 27 patients underwent embolization. Ten patients were specifically treated with the goal of curative embolization because they preferred the reduced invasiveness of embolization to surgical excision. These patients had AVMs in which the nidus was not larger than 3 cm, the number of feeders did not exceed three, and the nidus was accessible with the tip of the catheter. The success rate of endovascular embolization in patients treated with curative intent was 60% (6 of 10), and the overall cure rate was 22% (6 of 27). Follow-up angiography performed at 17 to 32 months showed complete obliteration of the AVM nidus in the 6 patients after initial embolization, and these patients remained asymptomatic 5 to 7 years after treatment.²²

In a study of 306 patients undergoing 513 embolizations, embolization was curative in 29 patients (9%). In the 415 sessions in which the AVM was accessible to deposition of embolic material, the rate of overall curative embolization was 7%. In 64 patients in whom the primary goal was curative embolization, the AVM was accessible in 55 patients and 17 AVMs were cured by embolization, for curative embolization rates of 26% in the total group and 31% in those with accessible lesions.¹⁹

Valavanis and Yasargil reported a 40% cure rate in a series of 387 embolized patients. In a subgroup of 182 AVMs that were embolized with curative intent, they reported a 75% cure rate.²¹ Direct, dominant feeding arteries, a monocompartmental nidus, and a dominant fistulous component of the nidus, without perinidal angiogenesis, were the chief variables predictive of endovascular obliteration. These authors did not find size or number of feeding pedicles to be an important determinant of the potential for endovascular cure.

In well-selected patients, curative embolization may be beneficial for AVMs located in regions with a poor response to microsurgery or radiosurgery. In a study of 14 patients with AVMs in intraventricular or paraventricular locations, curative embolization was achieved in 6 of 12 lesions that could be approached. In carefully selected patients, curative embolization may be appropriate because surgical treatment of AVMs in these locations is more difficult ¹⁰⁵ and decreased rates of obliteration¹⁰⁶ and poorer outcomes¹⁰⁷ have been observed with radiosurgery.

In a select number of patients with a low risk profile, primary embolization may be an acceptable form of treatment. The patient in Figure 387-1A to C had a small AVM that was easily accessible with only one arterial feeding vessel from the left pericallosal artery. This high-flow lesion had varicoid venous drainage, which placed the patient at increased risk for hemorrhage. This lesion was successfully cured after embolization with NBCA under monitored anesthetic care and provocative neurological testing. The patient returned at 3 months and recently at 6 months (Fig. 387-1D to F) for angiography, which confirmed complete obliteration of the lesion.

FIGURE 387-1 Cerebral arteriovenous malformation characterized by a single branch fistula in the left parietal lobe before and after complete endovascular occlusion. A-C, Cerebral arteriography of the left internal carotid artery in the frontal and lateral projections shows rapid arteriovenous shunting in the left parietal lobe (arrows). Complete occlusion was accomplished with N-butyl-2-cyanoacrylate (Cordis Neurovascular, Miami, FL). D-F, Surveillance arteriography after 6 months shows durable occlusion of the vascular malformation without recanalization (open circles).

Recanalization

In any AVM treated by curative embolization, complete angiographic embolization of an AVM does not guarantee that future hemorrhage will not occur because the nidus of the AVM remains and delayed recanalization through collateral channels is a possibility.¹⁰⁸ There is some evidence that patients with incompletely embolized AVMs are at increased risk for future morbidity and mortality from hemorrhage.^84,^109,¹¹⁰ A study by Heindenreich and colleagues demonstrated that an incomplete reduction of greater than 60% resulted in a 19-fold increased risk for hemorrhage.¹¹¹ Complete endovascular cure is considered possible with the use of cyanoacrylate, but the degree of permanent obliteration is probably high yet remains unknown.¹⁰⁸