O7.10 * CORRELATION OF PRE-OPERATIVE LANGUAGE FMRI WITH INTRA-OPERATIVE DIRECT CORTICAL STIMULATION FINDINGS IN GLIOMA SURGERY IDENTIFIES POTENTIAL FALSE POSITIVE FMRI ACTIVATIONS WITH CONVENTIONAL ANALYSIS

Abstract

BACKGROUND: Functional MRI (fMRI) has an increasingly established role in surgical planning for glioma patients. However, the presence of neuro-vascular decoupling in the vicinity of the tumor, along with other tumor specific phenomena, raises at least theoretical concerns of potential inaccuracies. Several studies have attempted to correlate fMRI findings with cortical mapping performed using direct cortical stimulation (DCS). Results with respect to language mapping in particular have not been consistent and have focused upon investigating the sensitivity of fMRI to detect eloquent tissue. The possibility of false positive activations that could lead to regions of tumor being incorrectly labeled as inoperable has not been fully explored, in part due to the challenge of accurately recording DCS findings. METHODS: Ten patients with suspected primary or recurrent low-grade gliomas underwent pre-operative fMRI including 3-4 language paradigms on a Phillips 3T Achieva scanner. Comprehensive neuropsychological assessment was also completed prior to awake craniotomy. fMRI data processed using Phillips proprietary IViewBOLD was made available using the BrainLab VectorVision neuro-navigation system for planning of the craniotomy. The surgeon was blinded to fMRI outcomes for the duration of cortical mapping to minimize bias. DCS was completed using a bipolar stimulator tracked continuously using the BrainLab system networked via the VVLink interface to a research system running 3D Slicer. DCS outcomes were recorded by the surgeon using a panel of footswitches to trigger a custom logging module. fMRI data was re-processed using Statistical Parametric Mapping (SPM) 8 to generate t-statistic maps with a simple threshold of 3 applied analogous to the standard clinical implementation. DCS outcomes within 10mm of a cluster of ≥10 voxels above the threshold were analyzed. RESULTS: A mean of 125.4 [range 64-178] unique DCS outcomes were recorded per subject with 60 [32-100] being at the current level at which a positive outcome was noted. A total of 15 fMRI activation clusters above the threshold level could be evaluated across the 10 patients [0-3 per patient]. 7 activations (46.7%) were confirmed with positive DCS findings (i.e. speech disturbance induced by DCS). However, 8 activations (53.3%) from 4 patients had more than one negative DCS finding and no positive DCS outcomes within a 10mm radius. 10 individual positive DCS outcomes [0-4 per patient] with no corresponding fMRI activation were also noted. CONCLUSION: The presence of fMRI activations that were not confirmed on DCS, the current gold standard for cortical mapping, raises the possibility of false positive fMRI activations. This could unnecessarily restrict tumor resections with a potential negative impact upon outcome and merits further investigation in particular as to the potential benefit of alternate fMRI analysis approaches.