Objective: Memory complaints are common in patients after brain tumor, but is difficult to map memory functions during awake surgery, to preserve them. Thus we analyzed one of the largest data sets on clinical, surgical, and anatomical correlates of memory in patients with brain tumor to date, providing anatomical hotspots for short and long-term memory functions. Methods: A total of 260 patients with brain tumor (130 high-grade gliomas; 76 low-grade gliomas [LGG]; 54 meningiomas) were tested on 2 commonly used short-term memory (Digit Span Forward and Corsi Spatial Span) and 2 long-term memory tasks (Narrative Memory and Delayed Recall of Rey Figure). Patients were evaluated before and immediately after surgery and (for LGG) after 4 months and data analyzed by means of analysis of covariance and the voxel-based lesion-symptom mapping technique. Results: As expected, patients with high-grade gliomas were already impaired before surgery, whereas patients with meningioma were largely unimpaired. Patients with LGG were unimpaired before surgery, but showed significant performance drop immediately after, with good recovery within few months. Voxel-based lesion-symptom mapping analyses identified specific anatomical correlates for verbal memory tasks, whereas visuospatial tasks provided good sensitivity to cognitive damage but failed to show anatomical specificity. Anatomical hotspots identified were in line with both previous functional magnetic resonance imaging and clinical studies on other neurological populations. Conclusions: Verbal memory tasks revealed a set of specific anatomical hotspots that might be considered “eloquent” for verbal memory functions, unlike visuospatial tasks, suggesting that commonly used spatial memory tasks might not be optimal to localize the damage, despite an otherwise good sensitivity to cognitive damage.
