Functional Magnetic Resonance Imaging Based Language Center Localization Protocol Identifies Language Centers and Results in Clinical Outcomes Similar to Awake Intraoperative Testing

Introduction: The goal of surgery in intraxial brain tumors in proximity to eloquent language areas is maximal resection with preservation of language function. Awake intraoperative cortical stimulation is the standard of care for evaluating functional areas during resection. Functional magnetic resonance imaging (fMRI) is a tool that may help evaluate these areas without the risks of awake monitoring, but current techniques yield low (20-60%) specificity.

Methods: We describe a detailed single-center protocol for fMRI for tumor resection in proximity to eloquent language centers including: detailed pre-operative neuropsychology evaluation, image acquisition with diffusion tensor imaging and functional data post-processing, fusion to thin slice (1mm) anatomic imaging, integration into neuronavigation, and intraoperative use. We report outcomes of 110 patients undergoing surgery for intraxial brain tumor near functional language centers with language fMRI (50% with additional awake testing and 50% with fMRI data alone). In addition, we report a prospective series of patient comparing language localization by fMRI and awake testing.

Results: Broca’s area, Wernicke’s area, basal temporal language area, and arcuate fasiculus fibers are reliably identified and visualized on intraoperative neuronavigation software. We report heterogeneity of language center overlap in patients with multiple languages. We find no significant differences in extent of resection, post-operative language deficits, progression free survival, and overall survival between patients with our fMRI-based language localization protocol alone versus fMRI language identification with awake cortical language testing. We show > 90% sensitivity and specificity in our fMRI-based language center protocol for identifying Broca’s area and Wernicke’s area relative to awake intraoperative cortical stimulation with naming and comprehension tasks.

Conclusion : Together, these results suggest that our fMRI-based protocol robustly identifies language centers of the brain, including fiber tracts. We believe that in cases not suitable for awake intraoperative testing, this protocol may allow for improved extent of resection and preservation of eloquent language centers.