The AVM size (<3 cm, 3–6 cm, >6 cm), location (eloquent or non-eloquent), and venous drainage pattern (deep or superficial) were found to be important AVM characteristics when determining surgical risk and predicting postoperative neurological outcomes [28]. The Spetzler-Martin grading system (SMG) was developed using size, location, and venous drainage patterns to assign a numeric score (1–5) with the higher number predicting higher surgical risk with worse neurological outcomes (Table 48.1); an SMG of 6 has been used to imply an “unresectable” AVM. This system was validated and predicts that an SMG of “1” has a 0 % risk of deficit from surgical resection, while an SMG of “5” has a 31 % risk. The authors further divided surgical risk into “major” and “minor” neurological deficits (Table 48.2). Though the SMG has been very useful to many surgeons planning for AVM resection, it does have pitfalls. For example, surgical outcomes associated with the SMG will be different for individual surgeons depending upon their training, technical abilities, case volume, and experience. Additionally, interobserver variability for the SMG was 28 %, with venous drainage having the least interobserver variability and nidus size having the greatest [29].

A modification of the SMG was subsequently described where age, rupture status, and nidus diffuseness were considered important characteristics for assessing surgical outcome. This modification was published as the “supplementary grading system” (SGS) developed by reviewing 300 patients who underwent microsurgical AVM resection and determining their modified Rankin score, SMG, and SGS (Table 48.3) [30, 31]. The predictive accuracy of the SGS for postsurgical neurological deficits was higher (0.73) than the SMG (0.66). However, the SGS is intended to be an adjunct to the SMS and not a substitute. Therefore, the modification includes adding the SMG and SGS scores to obtain a combined score ranging from 0 to 10. Using this scoring system, a grade of 1–3 predicts low surgical risk, 4–6 moderate risk, and 7–10 high risk, for surgical AVM resection.

A three-tier classification system was recently described in order to guide AVM management based on the SMG [32]. AVMs were classified into “A” through “C” based on SMG with recommended management options according to the classification (Table 48.4). The recommendations based upon this classification system were Class “A” (SMG 1, 2), surgical resection; Class “B” (SMG 3), multimodality treatment; and Class “C” (SMG 4, 5), no treatment.

Various imaging modalities can be used to depict the anatomical features of an AVM, including computed tomography angiography (CTA), magnetic resonance angiography (MRA), and magnetic resonance venography (MRV) (Fig. 48.3). However, only catheter-based digital subtraction angiography (DSA) can demonstrate the anatomical and hemodynamic features of the AVM. Differentiating between arterial feeding pedicles and draining veins can be challenging on the static imaging such as CTA and MRA (Fig. 48.3). Additionally, clearly identifying the extent of the AVM nidus on MRA and CTA can also be challenging at times, particularly with diffuse lesions. Static images, however, unlike DSA, depict the brain tissue relative to the AVM and are therefore very important for determining the location of the AVM within the cerebral hemispheres. These static imaging modalities are particularly useful for determining AVM proximity to eloquent cortical and subcortical areas. Newer static imaging modalities such as functional MRI (fMRI) and diffusion tensor imaging (DTI) can be particularly useful for preoperative planning.

The emergence of FMRI in the 1990s and its continued development enable clinicians to identify eloquent cortical areas using real-time noninvasive mapping techniques. The patient is presented with a series of tasks during an MRI and the cortex can be mapped based on changes in blood flow and oxygen delivery, making this a functional and physiological test for detection of eloquent brain tissue. The proximity of eloquent areas relative to the AVM can then be viewed when planning the surgical approach and assessing surgical risk with the patient (Fig. 48.4). The combined use of noninvasive static imaging (MRA, fMRI, CTA) and minimally invasive dynamic imaging (DSA) allows the neurosurgeon to apply the SMG in a more precise manner when considering potential treatment options [33]. For example, an AVM clearly located in the eloquent cortex or deep white matter structures would likely result in an unacceptably high surgical risk, making embolization or SRS a safer option. fMRI is also helpful with AVM treatment planning because a small percentage of individuals exhibit cortical rearrangement, such that functionality is not where one would typically expect. These functional anomalies could have important implications for surgical planning when they are identified ahead of time and avoid unexpected poor outcomes from neurological deficits (Fig. 48.5).

DTI is a relatively new form of MRI technology that uses water diffusion in the brain tissue to map the white matter tracks (tractography) throughout the hemispheres (Fig. 48.6). The result is a color-coded map of the white matter tracts that can be viewed relative to the location of the AVM. White matter tracts originating from eloquent structures passing through the AVM nidus could be an important deterrent to surgical resection. For example, surgical resection of an AVM nidus adjacent to the white matter tracts connecting Broca and Wernicke areas may result in a conductive aphasia from disconnection of these fibers. Another example would be an AVM in Meyer’s loop resulting in visual field deficits after resection. In this way, DTI tractography can be an important tool for presurgical risk assessment and may guide the decision towards a minimally invasive therapy that is less likely to injure critical brain structures.