Classification Systems

Historical surgical grading systems

The first iAVM classification system was introduced by Alfred J. Luessenhop and Thomas A. Gennarelli in 1977. They rated supratentorial AVMs from grade I to IV, with an additional grade attributed for each arterial feeder with accepted nomenclature. To grade an AVM by this scheme, the primary contributing vascular territory is identified and points are added for each arterial feeder with standardized nomenclature from that territory. For example, an AVM in the frontal pole with three arterial feeders from the frontopolar artery and one from the orbitofrontal artery would be given a grade of II. Deviations from their algorithm include the following: lenticulostriate perforators were included as named arteries; choroid plexus AVMs were grade III, since they were supplied by one anterior and two posterior choroidal arteries; and corpus callosum AVMs were grade II if only fed by pericallosal arteries and grade III when also supplied (directly or indirectly) by the posterior cerebral artery (PCA). The classification’s upper limit was grade IV, since Luessenhop and Gennarelli deemed that an AVM of grade IV or higher was inoperable. The authors presented data from a series of 49 cases of supratentorial AVMs and reported the following postoperative morbidity and mortality rates: for grade I AVMs, 0%; for grade II, 5.9%; for grade III, 50%. All grade IV AVMs were managed conservatively (without surgery).

To ease classification and include cerebellar AVMs, Luessenhop and Rosa proposed an alternative schema that exclusively considered AVM size. Regardless of AVM location or the number of arterial feeders, grade I AVMs were lesions ≤ 2 cm; grade II were 2–4 cm; grade III, 4–6 cm; and grade IV, > 6 cm. The 1977 schema applied to supratentorial lesions, while the second version applied to all parenchymal AVMs rostral to the spinal cord. The authors reported morbidity and mortality rates of 0%, 6.7%, and 56% for grades I, II, and III, respectively. Similar to the prior publication, grade IV AVMs were deemed inoperable. While the classifications of Luessenhop and colleagues are not currently utilized, they established the foundation for subsequent iAVM classification systems.

In the October 1986 edition of the Journal of Neurosurgery , two distinct AVM classifications were published, one by Shi and Chen and the other by Spetzler and Martin. Shi and Chen graded iAVMs on a scale of I–IV based on scores for four components: AVM size, location and depth, arterial feeders, and draining veins. Each component was given a score ranging from 1 to 4. For AVM size, scores of 1, 2, 3, and 4, respectively, were given for sizes < 2.5 cm, 2.5–5 cm, 5–7.5 cm, and > 7.5 cm. For location and depth, AVMs with superficial and noncrucial location received a score of 1; superficial AVMs in functional areas, a score of 2; deep AVMs in the corpus collosum or basal ganglia, a score of 3; and brainstem or diencephalon AVMs, a score of 4. Arterial supply was scored as follows: 1 for a single superficial middle cerebral artery (MCA) or anterior cerebral artery (ACA) branch; 2 for multiple superficial MCA or ACA branches; 3 for branches of the posterior cerebral artery (PCA), deep MCA, deep ACA, or vertebral artery; and 4 for branches of all 3 major cerebral arteries or a vertebral or basilar artery. For venous drainage, AVMs with a single superficial draining vein with dural sinus drainage were scored 1; multiple superficial sinus draining veins, 2; drainage into a deep cerebral vein, vein of Galen, or straight sinus, 3; and deep drainage with associated venous ectasia, 4. To appropriately grade an iAVM with the Shi and Chen criteria, the lesion would be given a four-digit score. For example, an AVM with a 3-cm-diameter nidus, located on the cortex near Broca’s area with arterial supply from multiple branches from ACA and MCA territories and venous drainage into the sphenoparietal sinus would be given a score of 2223. When at least two of the four components share the highest grade, then the AVM is given that singular grade. If the highest score is only held by one component, then the AVM grade is intermediate. Therefore the final Shi and Chen classification of our example with the score of 2223 would be grade II–III. Since the Shi and Chen grading schema had intermediate grades, the total number of AVM classes was 7. In the series of 100 patients, the authors reported a combined morbidity and mortality rate of 0%, 0%, 0%, 16.7%, 20%, and 100% for grades I, I–II, II, II–III, III, and III–IV, respectively. (No rate was given for grade IV because no grade IV lesion was treated with complete resection in this series.) While both classifications proposed in 1986 were more sophisticated than prior grading systems, accounting for the architecture of draining veins, the schema proposed by Spetzler and Martin was more user-friendly. It eventually was widely popularized and was dubbed the Spetzler-Martin classification.