Introduction

The management of brain arteriovenous malformations (AVMs) is in a state of flux. Recent studies, such as ARUBA, supporting medical management have changed the landscape of neurosurgical practice. Although there have been many arguments for and against the conclusions of the study, its legacy is a predisposition toward observation for the majority of practitioners. In this setting, it is important to understand which lesions are most amenable to treatment and which modalities are best suited.

High-grade AVMs are by definition larger, deeper, and more eloquent than other lesions, which makes treatment inherently more fraught. Patient selection is therefore paramount in deciding which patients we stand to help versus those at risk of harm. Although the factors comprising the Spetzler–Martin AVM grading scheme are those classically taught when considering AVM risks, there are a number of additional elements that should be considered in determining the optimal management strategy. We herein describe management options as well as the patient- and surgeon-level considerations that will guide optimal patient selection, treatment strategies, and execution.

Major controversies in decision making addressed in this chapter include:

1. 
   

   Is treatment indicated for grade IV and V AVMs?

   
   
2. 
   

   What combination of open, endovascular, and radiosurgical treatments should be employed?

   
   
3. 
   

   What is the optimal timing of each?

Whether to Treat

The decision to treat or not is driven by two primary factors: (1) the comparative risk of treatment versus observation and (2) patient preference.

Risks associated with treatment and observation have been the subject of numerous studies over the years. Ondra and colleagues published a landmark population-based study of AVM natural history and reported a hemorrhage rate on the order of 4% per annum. Of note, this figure excluded some 40% of patients that were selected for treatment. Subsequent studies that included all AVMs without regard for treatment status have found hemorrhage rates to be as high as 18% per year.

In reality, AVMs are heterogeneous and natural history risks vary significantly with anatomical variants. Stapf and colleagues studied a group of 622 AVMS with an overall hemorrhage rate of 6%. They reported several factors that positively correlated with hemorrhage: increasing age, initial hemorrhagic presentation, deep brain location, and exclusive deep venous drainage. For those without any of these anatomical risk factors, they observed an annual hemorrhage rate of 0.9%, whereas for those with all three anatomical risk factors the hemorrhage rate was as high as 34.4% ( 1 , 2 in algorithm ). The morbidity and mortality associated with these hemorrhages sets the benchmark against which any potential treatment strategy must be compared.

Treatment risks have been documented for all treatment modalities, but it must be noted that grade IV and V lesions are relatively underrepresented. For instance, ARUBA included only 23 grade IV lesions and no grade V lesions. Of these, only 8 were assigned to interventional management, whereas 14 were randomized to medical management. Regardless, the general consensus among studies is that high-grade AVMs present the highest treatment risks. Among Spetzler′s original 100 patients, upon which his grading scheme was based, there were 15 grade IV and 16 grade V lesions. Although there were no mortalities reported, minor morbidity occurred in 20 and 19%, respectively, for grade IV and V AVMs. Major morbidity accompanied 7 and 12% of grade IV and V cases, respectively ( 1 , 2 in algorithm ).

Although grading schemes allow practitioners to categorize and discuss lesions, they sometimes fail to account for nuances and complexities that may be important when considering how risks of observation and various treatments compare. Beyond size, drainage, and eloquence, a number of factors are now well established as contributing to risk of spontaneous hemorrhage, including AVM size, exclusive deep venous drainage, high intranidal blood flow, and AVM-related aneurysms ( 3, 4 in algorithm ). Furthermore, additional AVM characteristics have been correlated with treatment-specific outcomes and now form the basis for a multitude of modality-specific grading schemes, which will be discussed later. Thus, a more nuanced understanding of the individual lesion and patient circumstance can help guide the discussion between surgeon and patient as to the best treatment strategy for a particular lesion ( 5, 6, 7 in algorithm ).

Patient preference is another often underappreciated factor that should impact the decision-making process. The patient is the one who must ultimately live or die by the results of any treatment plan. Personal levels of risk tolerance should be thoroughly explored with the patient to understand how various treatment modalities may be received ( 5, 6, 7 in algorithm ). Furthermore, many express aversion to living with the uncertainty of future devastating hemorrhage, which in turn impacts how a patient values a given treatment option and what that patient determines acceptable level of risk. Thus, discussions regarding even very challenging lesions should be thorough, and any informed consent process should be well documented and understood by both surgeon and patient.

Conservative Management

One of the hardest management decisions a surgeon can make is the decision not to treat. Although this goes against the surgeon′s instinct to act, in many cases it is, in fact, the best of available alternatives. To be clear, for grade IV and V AVMs, observation should perhaps be the default course of action, and interventions should be undertaken if there are compelling reasons that shift the risk–benefit equation in favor of treatment ( 5, 11 in algorithm ). Resection may be associated with substantial morbidity, and radiosurgery and embolization are associated with low obliteration rates. Spetzler and colleagues performed a systematic pooled analysis of AVM outcome reports from the literature. Their first insight was that there was no significant difference in outcome between grade IV and V AVMs. They further recommended conservative management for high-grade AVMs unless a patient suffers hemorrhages or progressive neurological deficits ( 6–11 in algorithm ). Other mitigating circumstances that may favor treatment include the presence of flow-related aneurysms that are amenable to treatment by microsurgical or endovascular obliteration and the presence of steal-related deficits that might be ameliorated through endovascular embolization ( 3, 8, 10 in algorithm ).

Anatomical Considerations and Classification

Judicious patient selection is essential to avoid surgical complications and poor neurological outcomes with brain AVM. A number of classification schemes have been developed over the years incorporating a variety of anatomic, demographic, and radiographic features, each with its own emphasis, accuracy, advantages, and disadvantages. The most commonly used is the Spetzler–Martin grading scheme, a surgical risk tool that is based on the size, eloquence, and venous drainage of the lesion (▶ Fig. 55.1 ). The five Spetzler–Martin classifications have been further refined by Spetzler and Ponce to condense grades I and II into a low-risk class A, grade III into moderate-risk class B, and high-grade lesions assigned grades IV and V into high-risk class C. These classification schemes have value because they allow clinicians to simply convey the nature of a lesion and to transform complex management decisions into algorithms. However, particularly when considering high-grade lesions, the consequences of both observation and treatment are higher, and thus require a more nuanced appreciation of the lesion. Understanding an AVM′s subtleties allows the surgeon to know which should be left alone, which should be treated, and how.

To this end, a number of clinicians have analyzed their experiences to identify factors that determine the risks of surgery in order to assist them in this selection process. Lawton et al envisioned a grading system that would supplement rather than replace the already entrenched Spetzler–Martin grading system. This supplementary grading system adds three additional variables to the Spetzler–Martin scale. Points are assigned for patient age, hemorrhagic presentation, and AVM diffuseness, in a manner that is analogous to the Spetzler–Martin scoring system. Pediatric patients (age < 20 years) were assigned 1 point; adults (age 20–40 years) were assigned 2 points; and older patients (> 40 years) were assigned 3 points. Patients presenting with unruptured AVMs were assigned 1 point and ruptured AVMs 0 points. Diffuse AVMs were assigned 1 point and compact AVMs 0 point. These points were added together for a supplementary AVM grade that ranged from 1 to 5. Simplicity is critical to the utility of a grading scale, and the supplementary grading scale was designed with this in mind. The supplementary system was developed through analysis of 300 consecutive surgical patients. The supplementary grading scale had high predictive accuracy (area under the receiver operating characteristic [ROC] curve, 0.73 vs. 0.65 for the Spetzler–Martin grading system) and stratified surgical risk more evenly.

The supplementary grading system refines patient selection for AVM surgery. Clinical decisions begin with an analysis of nidus size, venous drainage, location, diffuseness, age, and presentation, generating Spetzler–Martin and supplementary grades. The Spetzler–Martin grade provides an initial risk estimate. The supplementary grade can be considered separately, with supplementary grades ≤ 3 having an acceptably low risk of AVM resection. It can also be added to the Spetzler–Martin grade, with combined grades ≤ 6 having an acceptably low surgical morbidity. An analysis of supplementary factors can impact a management decision by confirming the risk predicted by the Spetzler–Martin grade. For example, an AVM with a low Spetzler–Martin grade (grades I–III) may be favorable for microsurgical resection, and a low supplementary grade (I–III) may strengthen the recommendation for surgery. Lawton et al found that 62% of patients had low-grade AVMs according to both grading systems, and 85% of these patients were improved or unchanged after surgery. Conversely, an AVM with a high Spetzler–Martin grade (IV–V) may be unfavorable for microsurgical resection, and a high supplementary grade (IV–V) may strengthen the recommendation for nonoperative management. In these cases of matched Spetzler–Martin and supplementary grades, the supplementary grading system has a confirmatory role and may not alter management decisions.

However, in cases of mismatched Spetzler–Martin and supplementary grades, the supplementary grading system may alter management decisions and therefore has a more important role ( 7–9 in algorithm ). Lawton et al found that 28% of patients had low Spetzler–Martin grades and high supplementary grades, and 41% of these patients were neurologically worse after surgery, which is a higher morbidity than that of Spetzler–Martin grade IV AVMs. Insight provided by the supplementary grade might have discouraged the recommendation for surgery in some of these patients. Similarly, 7% of patients had high Spetzler–Martin grades and low supplementary grades, and 29% of these patients were neurologically worse after surgery. This proportion of worsening was lower than the 35% morbidity for the overall group of Spetzler–Martin grade IV and V AVMs, and equivalent to the 30% morbidity seen for grade III AVMs. Again, insight provided by the supplementary grade might have encouraged the recommendation for surgery in some of these patients. Spetzler–Martin grade III AVMs have surgical risks that depend on the subtype, with small grade III–lesions associated with lower risk and medium/eloquent grade III+ lesions associated with higher risk. In addition to considering the grade III subtype, considering the supplementary grade may influence surgical decisions for AVM patients at the borderline between high and low risks. The University of California San Francisco (UCSF) experience demonstrates that factors outside of the Spetzler–Martin grading system improve the prediction of neurological outcome after AVM resection, and that a simple supplementary grading system can be easily applied at the bedside to refine patient selection for AVM surgery.

In addition to surgical grading schemes, there have been radiosurgical and endovascular grading schemes developed to address the particular risks and selection features for other treatment modalities.

This is necessary because the factors that are important to AVM obliteration by stereotactic radiosurgery are different from the factors in the Spetzler–Martin grading system. Inoue et al identified size, morphology, and hemodynamics as factors most predictive of radiosurgical outcome. The authors classified AVM hemodynamics as Moya type (small-caliber feeding arteries, compact nidus, and veins that drain in the venous phase of the angiogram), shunt type (large-caliber feeding arteries, indistinct nidus, and veins that drain early in the arterial phase), or mixed. AVM morphology was classified as homogeneous or heterogeneous. Radiosurgical results in 30 patients demonstrated that small, homogenous, Moya-type AVMs had the best obliteration rates and that other AVMs with shunt and mixed hemodynamics would benefit from preradiosurgical embolization.

Pollock and Flickinger developed a more quantitative radiosurgical classification scheme called the radiosurgery-based AVM grading system. Multivariate analysis of data from 220 patients identified five variables associated with radiosurgical success, defined as complete nidus obliteration without new or worsening neurological deficit: AVM volume, age, location, previous embolization, and number of draining veins. The latter two factors added little to predictive accuracy and were dropped from the final equation for outcome: AVM score = 0.1 × (volume in cm³) + 0.02 × (patient age in years) + 0.3 × (location of lesion [frontal or temporal = 0; parietal, occipital, intraventricular, corpus callosum, or cerebellar = 1; and basal ganglia, thalamic, or brainstem = 2]). Patients with a composite score of less than 1 had excellent outcomes, whereas only 39% of patients with scores greater than 2 had excellent outcomes. The predictive equation was validated at a different institution using a separate patient cohort of 136 patients. To reduce the complexity of the scheme, the location variable was modified to be two tiered, with lesions in the hemispheres, corpus callosum, and cerebellum assigned 0 points, and those in the basal ganglia, thalamus, and brainstem assigned 1 point. This modified equation was validated on an additional 247 patients. Although this grading system was developed at a center using exclusively Gamma Knife radiosurgery, it has been validated at LINAC-based centers as well. The scale was based on outcome data after a single radiosurgery procedure. Radiosurgery has a long latency period and many radiosurgical patients require repeat radiosurgery or surgery to achieve complete obliteration of the lesion. Therefore, outcome data used for radiosurgical grading scales require diligent follow-up.

Endovascular embolization of AVMs has been used both as a surgical adjunct and as a potentially stand-alone curative therapy. Consequently, there have been several attempts to create an endovascular classification scheme. The Buffalo system takes into account three features: the number of arterial pedicles, the pedicle diameter, and eloquent location. A higher complication incidence would be expected for patients with a higher score. The Vienna group produced a classification based on AVM size, number of feeding arteries, and pial versus perforating feeding arteries. Low-grade AVMs that were small, had less than two feeding arteries, and were not supplied by perforators were suitable for endovascular therapy, whereas high-grade AVMs that were large (>4 cm), had more than four feeding arteries, and were supplied by perforators were not suitable for endovascular therapy. Improved embolic agents like Onyx and more navigable microcatheters have increased the rates of curative AVM embolization. As technology and techniques evolve and obliteration rates increase, there will undoubtedly be additional classification schemes that will help select patients for curative endovascular intervention.

As treatment for AVM continues to evolve toward multimodality care, there is a need for highly discriminative risk assessment models that take into account all treatment options. As these are relatively rare lesions, this will likely require collaborative efforts across treating centers to develop a fund of data that can power meaningful risk prediction and help clinicians better match lesions and treatment options.