Classification of Brain Arteriovenous Malformations and Fistulas

12 Classification of Brain Arteriovenous Malformations and Fistulas

Derrick L. Umansky, Ricky Medel, and Peter S. Amenta

Abstract

The proper classification of brain arteriovenous malformations (AVMs) and dural arteriovenous fistulas (dAVFs) is critical to clinical decision making and assessment of treatment risk. The Spetzler–Martin grade is the most commonly cited and rigorously tested system when evaluating outcomes following resection. Multiple additions and supplements have been made over the past two decades. The evolution of endovascular and radiosurgical therapy has driven the need for additional classifications that capture the outcomes and risks associated with these modalities. dAVFs are a diverse group of lesions with complex angioarchitecture. The Borden and Cognard systems are currently the most referenced in the classification of dAVFs. This chapter presents a review of the systems currently used to classify intracranial AVMs and dAVFs. Microsurgical, endovascular, and radiosurgical classification schemes are discussed. Special attention is paid to the most referenced systems, particularly the Spetzler–Martin grading scale for AVMs and the Borden and Cognard systems for dAVFs.

Keywords: arteriovenous malformation, dural arteriovenous fistula, classification systems, Spetzler–Martin, Borden, Cognard, Barrow, Buffalo score, radiosurgical score

Key Points

The Spetzler–Martin grading system of arteriovenous malformations (AVMs) remains the standard in the assessment of risk of surgical resection.
Multiple grading systems for surgical, endovascular, and radiosurgical treatment of AVMs are currently in use.
Dural arteriovenous fistulas (dAVFs) are a heterogeneous group of lesions with complex and variable angioarchitecture.
The Cognard and Borden systems are the most commonly referenced classification schemes for dAVFs.

12.1 Introduction

Arteriovenous malformations (AVMs) and dural arteriovenous fistulas (dAVFs) represent challenging clinical entities that are associated with significant morbidity and mortality. Much of the difficulty in managing these malformations is derived from the great diversity that exists between individual lesions. Both AVMs and dAVFs possess complex angioarchitecture and may have relationships to critical functional anatomy. Clinical presentations range from incidental findings or minimally symptomatic to catastrophic life-threatening hemorrhage. Management decisions must weigh the risks and benefits of intervention against the natural history of the disease. Treatment planning must account for multiple variables, including patient age, symptomatology, history of intracranial hemorrhage, and the vascular anatomy of the lesion. Multiple classification systems have been proposed to capture the key elements of an individual lesion and assess the risk of intervention. These systems have the goal of creating a reliable, easily applicable, and accurate assessment that is of clinical utility. The development of these schemes is initially rooted in the microsurgical approach to vascular disease, given that intervention was only possible via this route prior to the modern era. Treatment of these lesions has undergone a transformation in recent years, and traditional microsurgical management is now frequently supplemented or even replaced by endovascular embolization and radiosurgery. As a result, modification and supplementation of the original systems has become necessary to capture the outcomes, challenges, and complications unique to these additional modes of therapy. This chapter presents a review of the systems currently used to classify intracranial AVMs and dAVFs. Microsurgical, endovascular, and radiosurgical classification schemes are discussed. Special attention is paid to the most referenced systems, particularly the Spetzler–Martin grading system for AVMs and the Borden and Cognard systems for dAVFs.

12.2 Materials and Methods

A wide range of materials were used in the composition of this chapter. For a historical perspective, many of the original papers pertaining to the classification of AVMs and dAVFs have been included. The reviews of the established grading systems are based on the landmark papers that defined their methodology and statistical significance. Multiple contemporary sources are cited in the explanation of the additions and supplements to these systems over the past two decades. The tables are adapted from the original papers to provide a resource for the trainee searching for a useful and rapid reference. Images are taken from the institutional experience in the neurosurgical department at the Tulane Medical Center.

12.3 The Classification of Arteriovenous Malformations

12.3.1 AVM Classification and Microsurgical Resection

Spetzler–Martin Grading System

The primary goal of surgical resection of intracranial AVMs is the removal of the risk of intracranial hemorrhage and its associated morbidity and mortality. Complete removal may also benefit patients suffering from seizures, vascular steal phenomena, headaches, and other less severe but often debilitating sequelae. However, particularly in terms of intracranial hemorrhage, the risk of future rupture can only be predicted and, in fact, may never occur. To justify surgical intervention, one must be able to understand the potential risk of rupture, as well as the immediate morbidity and mortality of surgical resection.¹ In 1986, Spetzler and Martin proposed a grading system by which to predict the morbidity and mortality associated with open surgical resection of intracranial AVMs¹ (► Table 12.1). Taking into account the wide variability in angioarchitecture and anatomic location of these lesions, the authors composed a system that relied on three key features to simplify grading: AVM size, pattern of venous drainage, and relationship to eloquent tissue.

Table 12.1 Spetzler-Martin grading system

Graded feature
Size of AVM
Small (< 3 cm)	1
Medium (3–6 cm)	2
Large (> 6 cm)	3
Eloquence of adjacent brain
Noneloquent	0
Eloquent	1
Pattern of venous drainage
Superficial only	0
Deep	1

Note: In the Spetzler–Martin grading system, the grade = size + eloquence + venous drainage. AVMs with drainage into superficial veins are assigned 0 points, while those with drainage into deep veins are assigned 1 point. AVMs located in eloquent areas are assigned 1 point, while those in noneloquent areas are assigned 0 points. For instance, an AVM that is <2 cm in an eloquent area with deep venous drainage would have a grade of I + I +1 = III.

Source: Spetzler and Martin.¹

Nidus size, as measured with angiography, is considered small (< 3 cm), medium (3–6 cm), or large (>6 cm) and assigned a score of 1, 2, or 3, respectively. AVM size represents an important determinant of safety of resection for multiple reasons. Larger malformations are more likely to abut critical anatomic structures at the nidus margins. As a result, resection of larger lesions places a greater volume of functional tissue at risk. The size of the malformation also reflects the arterial supply and number of arterial feeding pedicles. Increasing AVM diameter places the nidus into an expanding number of vascular territories, thereby increasing the quantity of feeding pedicles and parent vessels supplying the AVM. The surgical approaches required to safely resect larger AVMs must account for blood supply from multiple territories and provide safe avenues by which to access those vessels.¹ Finally, as with most surgical procedures, larger lesions require a longer amount of time to remove. Patients with larger malformations are subjected to the risks of prolonged anesthesia, increased blood loss, and prolonged brain retraction.

Based on angiography, venous drainage is segregated into two distinct categories, superficial and deep drainage. Lesions possessing only superficial venous drainage are scored a zero. Superficial venous drainage is defined by venous outflow directed through the cortical venous system. AVMs exhibiting any element of deep venous drainage are assigned a 1. Venous drainage corresponds with accessibility of the lesion, as deep venous drainage places arterialized veins in a medial and potentially difficult location to access. These veins are often the subependymal veins of the ventricles and may be difficult to control, adding to blood loss, length of surgery, and volume of tissue resected.

Eloquence is assigned to structures and regions of the brain with known quantifiable function that, if injured, will result in a disabling neurologic deficit (► Fig. 12.1; Spetzler–Martin eloquent areas). AVMs confined to noneloquent regions are scored a zero, while AVMs involving at least one eloquent area are scored a 1. AVM nidus usually displaces surrounding tissue from its anatomic location; thus, eloquence is defined by the normal location of the affected anatomy. Obviously, dissection within or adjacent to eloquent tissue places the patient at a greater risk of postoperative impairment and increases the risk of resection. Computed tomography (CT) and magnetic resonance imaging (MRI) are useful in determining the anatomic relationships of the nidus.

Fig. 12.1 Spetzler–Martin areas of eloquence. Deep eloquent areas are depicted in the upper image and include the hypothalamus, thalamus, brainstem, and cerebellar peduncles. Cortical eloquent areas are depicted in the lower image and include the primary motor area, primary sensory area, language areas, and primary visual area. (Adapted from Spetzler and Martin.¹)

The grade of an individual AVM is calculated from the sum of the three scores, which can range from grade I to grade V. Grade I lesions are small, located in noneloquent regions, and display superficial drainage, and are therefore associated with limited morbidity and mortality with resection. Grade V lesions are large, involve eloquent anatomy, and display a component of deep venous drainage, rendering resection challenging, and are associated with significant operative risk. A total of 12 combinations of the grading criteria are possible. To test the validity of the system, the grading scale was applied to 100 AVMs resected by Spetzler and correlated with surgical complications. Complications were subdivided into minor deficits, major deficits, and mortality. There were no mortalities within the cohort. Grade I and II lesions were resected without major deficits. The percentage of major deficits progressively increased with each successive grade. Minor deficits were not observed in grade I AVM resections, and the incidence increased to 5 and 12% for grade II and III lesions, respectively. Grade IV and V lesions had the highest incidence of minor deficits. Examples of Spetzler–Martin grade II AVMs are seen in ► Fig. 12.2 and ► Fig. 12.3.

Fig. 12.2 Spetzler–Martin Grade II (S2V0E0) AVM. A 46-year-old man presented with a severe headache, nausea, and truncal ataxia. Sagittal (a) and coronal (b) MRI with gadolinium demonstrating a 3.2 x 2 x 2 cm AVM of the vermis and left tentorial surface of the cerebellum. (c) Preoperative left vertebral artery injection in the anteroposterior (AP) plane shows a hypertrophic left superior cerebellar artery feeding pedicle and superficial drainage to the tentorial sinuses. A supracerebellar-infratentorial approach was used to resect the AVM. (d) A postoperative left vertebral artery injection in the AP plane demonstrating complete resection of the AVM. The patient was discharged to rehab with persistent ataxia.

Fig. 12.3 Spetzler–Martin grade II (S1V0E1) AVM. A 42-year-old man presented to the clinic with a history of a remotely ruptured AVM located in the posterior left superior temporal gyrus. He reports a significant mixed aphasia resulted from the previous hemorrhage; however, he made significant improvement and is currently high functioning. Lateral projection of a left internal carotid injection demonstrates a compact 1.6 × 1.4 × 0.7 cm nidus fed by the angular branch of the middle cerebral artery. Superficial drainage via a single vein is directed to the superior sagittal sinus. The patient opted for radiosurgical intervention.

Further Classification of Spetzler–Martin Grade III AVMs

Due to its reliance on only three AVM characteristics, the Spetzler–Martin grading system is limited in the classification of intermediate grade lesions. This ambiguity is particularly true in regard to the scoring of grade III lesions, which are composed of four distinct subtypes.¹ The initial Spetzler–Martin paper demonstrated an incidence of 16% for minor and major neurologic deficits.¹ In a follow-up study, Hamilton and Spetzler reported a 2.8% permanent morbidity for grade III AVM resection.² Heros et al found the incidence of surgical complications to be 11.4% for resection of grade III lesions.³ De Oliveira et al initially proposed a reclassification of grade III AVMs into two additional subgroups, IIIA (“large”) and IIIB (“small and eloquent”), thereby reducing the number of Spetzler–Martin grade III subtypes from four to two.⁴ A total of 95.5% of patients with grade IIIA AVMs experienced good outcomes with embolization and resection, as opposed to only 70% of grade IIIB patients. A total of 27.8% of IIIB patients demonstrated a poor outcome and 2.1% died, which resulted in the authors recommending radiosurgery for this group.

Lawton and colleagues demonstrated the influence of size on outcomes following the resection of grade III AVMs.⁵ Grade III lesions were subdivided into small (< 3 cm), medium/deep (36 cm), and medium/eloquent (3–6 cm) groups. Large AVMs (>6 cm) were not encountered, leading the authors to conclude that these lesions are nonexistent or exceedingly rare despite the fact that the Spetzler–Martin system allows for their grading. This finding is likely due to the fact that AVMs with a diameter greater than 6 cm almost always abut or involve eloquent brain or possess at least one deep draining vein, thereby rendering them a grade IV lesion. A total of 97.1% of patients with small grade III AVMs remained unchanged or improved neurologically. In total, 92.9% of patients with medium/deep grade III lesions experienced no change or improvement in their condition. Patients in the medium/eloquent group fared worse, with only 85.2% remaining unchanged or improved. Within this cohort, 14.8% of patients suffered a decline in condition or died following surgery, representing two-thirds of the neurological morbidity and death in the study. Surgical risk of small AVM resection was 2.9%, while it was 7.1 and 14.8% for medium/deep and medium/eloquent lesions, respectively. The incidence of morbidity and mortality of small grade III lesions is low and rivals that of Spetzler–Martin grade II lesions. This finding is in contrast to the reported higher incidence of complications in the de Oliveira series, where smaller AVMs in eloquent anatomy were subject to poorer outcomes.⁴ Conversely, in the Lawton series, resection of medium/eloquent AVMs is subject to considerably higher rates of morbidity and death that are comparable to grade IV AVMs.⁵ Medium/deep lesions were associated with a surgical risk similar to the entire class of grade III AVMs.¹^,⁵

The authors proposed a modification to the Spetzler–Martin grading system to better define the risk of resection of the four grade III subtypes⁵ (► Table 12.2). The four subtypes are relabeled with a “–” (lower surgical risk) or ” + ” (higher surgical risk). III– AVMs correspond to the small AVMs amenable to resection. III + AVMs represent the medium/eloquent group and are associated with the prohibitive risks seen with grade IV and V lesions. Grade III AVMs are medium/deep AVMs and occupy a zone of intermediate risk, which the authors concede remain a decision-making dilemma. These lesions must be addressed on an individual basis, and treatment must be based on a combination of patient wishes, surgeon experience, and the clinical picture as a whole.⁵ III* lesions are the > 6 cm AVMs that were not encountered in the study.

Table 12.2 Modification of the Spetzler–Martin grading scale

Lawton–Young AVM Grading System

Lawton and colleagues also proposed a grading scale to supplement the Spetzler–Martin score and add clarity to clinical decision making.⁶ The Lawton–Young supplementary grading system considers three variables significantly correlated with surgical risk, diffuseness, patient age, and hemorrhagic versus nonhemorrhagic presentation. The scoring of the three categories is summated to arrive at a cumulative score ranging from 1 to 5 (► Table 12.3). Diffuseness is a ssociated with elevated surgical risk due to the imprecise surgical boundaries that result when the nidus is not compact and easily defined. These vague margins increase the risk of nidal bleeding and transgression of eloquent structures. Age is considered to account for pediatric (< 20 years old) patients who tend to fare better following AVM resection due to limited systemic disease and a greater potential for neurologic recovery. Patients older than 40 years were assumed to have a greater incidence of systemic disease, thereby limiting their ability to tolerate surgery and recover from neurologic injury. Finally, the authors included hemorrhagic and nonhemorrhagic presentations, adding a point for nonhemorrhagic presentations. Hemorrhagic presentation was associated with a lower surgical risk due to hematoma-related fixed neurologic deficits that cannot be worsened by resection.⁷ Additionally, the presence of a hematoma may create or enhance dissection planes or obliterate a portion of the arterial supply.

Table 12.3 Lawton–Young supplementary grading system

The predictive accuracy of this supplementary system was tested against the Spetzler–Martin scale and a 10-point scale that is composed of a sum of the two scores (the supplemented Spetzler–Martin grade).⁶ The supplemented Spetzler–Martin grade was found to be the most predictive of surgical risk, while the original Spetzler–Martin score was the least predictive. The supplementary Lawton–Young score was significantly more accurate than the Spetzler–Martin score. The authors conclude that the supplementary score can be used to support the findings of the Spetzler–Martin score or provide an additional means by which to assess risk when there is a difference between the two scales.

12.3.2 AVM Classification and Endovascular Therapy

Spetzler–Martin Grading and Embolization

Advancements in endovascular intervention have greatly impacted current strategies for the management of AVMs. The advent of liquid embolic agents, particularly Onyx, flowdirected and detachable-tip microcatheters, and balloonassisted techniques have expanded the armamentarium available to treating physicians. Embolization is currently used prior to surgical resection to reduce flow through the nidus, obliterate difficult to access deep feeders, and reduce intraoperative blood loss. Endovascular intervention is also used in conjunction with radiosurgical management, for palliation of symptoms and curative obliteration.⁸^,⁹^,¹⁰^,¹¹^,¹²^,¹³

Multiple authors have applied the Spetzler–Martin scale or its components to the endovascular management of AVMs with variable results. Hartmann et al reported no significant relationship between the Spetzler–Martin grade or its individual components and the incidence of postembolization complications.¹⁴ Gobin et al found increasing embolization complication rates with increasing Spetzler–Martin grade in patients undergoing embolization and radiosurgery.¹⁰ Viñuela et al demonstrated the incidence of embolization-associated complications to increase with each subsequent Spetzler–Martin grade.¹⁵ Haw et al reported a significant association between the incidence of complications and the location of the nidus in eloquent tissue.¹⁶ Kim et al applied the Spetzler–Martin grading to a series of patients treated with embolization performed as a curative procedure or as part of a multidisciplinary treatment regimen.¹⁷ The relationship between the incidence of complications and grade was not significant, but did demonstrate a clear trend of increasing complications with increasing lesion grade.

The Buffalo Score

Endovascular embolization of AVMs subjects patients to many of the same risks as open surgical resection; however, the means by which these complications occur are drastically different. Ischemic complications arise from occlusion of normal arteries following embolism into an en passage vessel, reflux from a feeding pedicle, or thromboembolism from prolonged catheterization. Penetration of embolic material into the venous outflow prior to complete nidus obliteration results in outflow restriction and potentially subsequent rupture. Superselective catheterization of arterial feeders risks perforation and intracranial hemorrhage. To account for these unique complications, Dumont et al proposed a grading system, the Buffalo score, to assess the risk of endovascular intervention for a particular lesion.¹⁸

The grading system relies on three characteristics: the number of feeding pedicles, the diameter of the feeding pedicles, and the eloquence of the surrounding brain.¹⁸ Based on angiography, the number of feeding pedicles is classified as one to two, three to four, or five or more. Embolization of each pedicle is an individual intervention; thus, an increasing number of pedicles is equated with an increasing possibility of iatrogenic injury.¹⁷^,¹⁸ The diameters of the feeding pedicles are measured within 1 cm of the nidus on angiography and classified into two groups, large (mostly, diameter >1 mm) or small (mostly, diameter < 1 mm). Smaller pedicles are a more challenging target for superselection and are more prone to wire perforation and subsequent hemorrhage. Additionally, liquid embolic agents may have a greater tendency to reflux into normal parent vessels when injected into smaller diameter pedicles. The eloquence of tissue adjacent to the nidus represents the third risk factor.¹ Risk of neurologic deficit due to embolization is greater when embolizate is injected into eloquent regions. The cumulative grade for an individual lesion is calculated by combining the scores for each of the three variables (► Table 12.4).

Table 12.4 Buffalo score

Graded feature	Points assigned
Number of arterial pedicles
1 or 2	1
3 or 4	2
5 or more	3
Diameter of arterial pedicles
Most >1 mm	0
Most ≤ 1mm	1
Nidus location
Noneloquent	0
Eloquent	1

Note: When calculating the Buffalo score, the AVM grade is the sum of the points assigned for the number of pedicles, diameter of the arterial pedicles, and the nidus location.

Source: Dumont et al.¹⁸

Only gold members can continue reading. Log In or Register to continue