Introduction Grade 3 AVMs are diverse due to variations in size (S), location in eloquent cortex (E) and presence of central venous drainage (V). These variations may have implications for management and outcome which were evaluated in this study.

Methods 100 patients with grade 3 AVMs were treated from 1984 to 2010. The AVMs were categorized by Spetzler-Martin characteristics as follows: Type 1, S1E1V1, Type 2 S2E1V0, Type 3 S2E0V1 and Type 4 S3E0V0. The occurrence of new neurological deficit, functional status (mRS) at discharge and follow-up, and radiological obliteration were correlated with demographic and morphologic characteristics.

Results 100 patients (49f, 51 m, age 5–68 years, mean 35.8 years) were evaluated. Size of AVMs were <3 cm (28), 3–6 cm (71) and >6 cm (1); 86 AVMs were in eloquent cortex, and 38 had central drainage. The AVMS were type 1 (28), type 2 (60), type 3 (11) and type 4 (1). Embolization was performed in 77 patients (175 procedures), surgery in 64 patients (74 surgeries), and radiosurgery in 49 patients (44 primary, 5 postoperative). Mortality following the management of these AVMs was 1%. 14 patients (14%) had new neurological deficits, with 5 (5%) disabling (mRS >2), and 9 (9%) non-disabling (mRS <2) events. Patients with type 1 AVM (small AVMs) had the best outcome, with 1/28 (3.6%) having a new neurological deficit as compared to larger AVMs (13/72, 16.7%, p<0.002). Older age (>40 years), size >3 cm, and non-hemorrhagic presentation predicted occurrence of new deficits (p<0.002). Gender, eloquence and venous drainage did not confer any benefit. 89 patients had adequate follow-up to be included in the obliteration analysis. AVM was obliterated in 78 patients (87.6%), 69 of them (78.6%) on angiogram and 9 on MRI/MRA. There was no difference between obliteration rates between different types of AVMs, Size, eloquence and drainage. Age, gender and clinical presentation also did not predict obliteration.

Conclusions Multi-modality management of Grade III AVMs results in a high rate of obliteration which does not appear to be influenced by size, venous drainage or eloquent location. However, the development of new neurologic deficits did correlate with size, while eloquence and venous drainage did not influence this. We propose sub-classifying the grade III AVMs according to their size (<3 cm and >3 cm) to account for treatment risk.