Objective: Grade I and II arteriovenous malformations (AVMs) have been considered safe to resect. However, unoperated low-grade AVMs have not been considered in previously reported series. The aim of this study was to examine all cases, both operated and unoperated, to identify any characteristics of low-grade AVMs that comprise a subgroup that might pose a relatively higher risk.
Methods: A prospectively enrolled AVM database included 237 patients in Spetzler-Martin Grade I or II. These patients were analyzed on the basis of demographic characteristics, angiographic and magnetic resonance imaging features, clinical presentation, method of treatment, and outcome.
Results: Surgery was performed in 220 patients in Spetzler-Martin Grade I or II. Seventeen patients did not undergo treatment because of poor neurological condition (six patients), patient refusal (nine patients), and perceived surgical difficulty (AVM size approaching 3 cm adjacent to Broca's area) (two patients). The overall surgical morbidity rate was 0.9%, and the mortality rate was 0.5%. Adverse outcomes occurred in 1 (0.6%) of 180 patients with AVMs located away from eloquent cortex and in 2 (5%) of 40 patients with AVMs adjacent to eloquent cortex. None of 28 surgical patients with deep venous drainage had an adverse outcome. All 219 patients who survived surgery underwent postoperative angiography that confirmed cure. No postoperative hemorrhage has occurred in 1143 patient-years of follow-up (mean follow-up, 5.3 yr).
Conclusion: When considering adverse outcome in the surgical series of Grade I and II AVMs alone, no statistical difference between non-eloquently located AVMs (0.6%) and eloquently located AVMs (5% adverse outcome) can be detected. However, consideration of all Grade I and II AVMs, both surgical and nonsurgical, may prove that a difference in outcome exists between these two groups masked by case selection. Generalization of the chances of adverse outcomes to all Grade I and II AVMs (both operated and unoperated) suggests that the risk of performing surgery on noneloquent brain in our series was 0.6% and that in eloquent brain could have been as high as 9.5%, had all such patients undergone surgery.