Complications of cerebral arteriovenous malformation embolization: Multivariate analysis of predictive factors

OBJECTIVE: Embolization is an important therapeutic modality in the multidisciplinary management of arteriovenous malformations (AVM); however, prior series have reported a wide variability in overall complication rates caused by embolization (10-50% neurological deficit, 1-4% mortality). In this study, we reviewed our experience with AVM embolization and analyzed factors that might predict complications and clinical outcomes after AVM embolization. METHODS: We analyzed our combined neurovascular unit's results with AVM embolization from 1993 to 2004 for the following outcomes measures: 1) clinically significant complications, 2) technical complications without clinical sequelae, 3) discharge Glasgow Outcome Scale score, and 4) death. To determine embolization efficacy, we analyzed perioperative blood transfusion and rate of AVM obliteration. Univariate and multivariate analyses were performed for patient age, sex, history of rupture, history of seizure, associated aneurysms, AVM size, deep venous drainage, eloquent location, Spetzler-Martin grade, number of embolization stages, number of pedicles embolized, and primary treatment modality. RESULTS: Over an 11 year period, 295 embolization procedures (761 pedicles embolized) were performed in 168 patients with embolization as the primary treatment modality (n = 16) or as an adjunct to surgery (n = 124) or radiosurgery (n = 28). There were a total of 27 complications in this series, of which 11 were clinically significant (6.5% of patients, 3.7% per procedure), and 16 were technical complications (9.5% of patients, 5.4% per procedure). Excellent or good outcomes (Glasgow Outcome Scale >= 4) were observed in 152 (90.5%) patients. Unfavorable outcomes (Glasgow Outcome Scale 1-3) as a direct result of embolization were both 3.0% at discharge and at follow-up, with a 1.2% embolization-related mortality. In the 124 surgical patients, 96.8% had complete AVM obliteration after initial resection, and 31% received perioperative transfusion (mean 1.4 units packed red blood cells per surgical patient). Predictors of unfavorable outcome caused by embolization by univariate analysis were deep venous drainage (P < 0.05), Spetzler-Martin Grade III to V (P < 0.05), and periprocedural hemorrhage (P < 0.0001) and by multivariate analysis were Spetzler-Martin III to V (odds ratio 10.6, P = 0.03) and periprocedural hemorrhage (odds ratio 17, P = 0.004). CONCLUSION: In a single-center, retrospective, nonrandomized study, 90.5% of patients had excellent or good outcomes after AVM embolization, with a complication rate lower than previously reported. Spetzler-Martin grade III to V and periprocedural hemorrhage were the most important predictive factors in determining outcome after embolization.