Does eloquence subtype influence outcome following arteriovenous malformation surgery?

Although numerous arteriovenous malformation (AVM) grading scales consider eloquence in risk assessment, none differentiate the types of eloquence. The purpose of this study was to determine if eloquence subtype affects clinical outcome. This is a retrospective review of a prospectively collected clinical database of brain AVMs treated with microsurgery in the period from 1997 to 2017. The only inclusion criterion for this study was the presence of eloquence as defined by the Spetzler-Martin grading scale. Eloquence was preoperatively categorized by radiologists. Poor outcome was defined as a modified Rankin Scale (mRS) score 3-6, and worsening clinical status was defined as an increase in the mRS score at follow-up. Logistic regression analyses were performed. Two hundred forty-one patients (49.4% female; average age 33.9 years) with eloquent brain AVMs were included in this review. Of the AVMs (average size 2.7 cm), 54.4% presented with hemorrhage, 46.2% had deep venous drainage, and 17.0% were diffuse. The most common eloquence type was sensorimotor (46.1%), followed by visual (27.0%) and language (22.0%). Treatments included microsurgery alone (32.8%), microsurgery plus embolization (51.9%), microsurgery plus radiosurgery (7.9%), and all three modalities (7.5%). Motor mapping was used in 9% of sensorimotor AVM cases, and awake speech mapping was used in 13.2% of AVMs with language eloquence. Complications occurred in 24 patients (10%). At the last follow-up (average 24 months), 71.4% of the patients were unchanged or improved and 16.6% had a poor outcome. There was no statistically significant difference in the baseline patient and AVM characteristics among the different subtypes of eloquence. In a multivariate analysis, in comparison to visual eloquence, both sensorimotor (OR 7.4, p = 0.004) and language (OR 6.5, p = 0.015) eloquence were associated with poor outcomes. Additionally, older age (OR 1.31, p = 0.016) and larger AVM size (OR 1.37, p = 0.034) were associated with poor outcomes. Unlike visual eloquence, sensorimotor and language eloquence were associated with worse clinical outcomes after the resection of eloquent AVMs. This nuance in AVM eloquence demands consideration before deciding on microsurgical intervention, especially when numerical grading systems produce a score near the borderline between operative and nonoperative management.