NIMG-37. FUNCTIONAL MRI AND DTI TRACTOGRAPHY AS ADJUNCTS TO IMPROVE THE PREDICTIVE VALUE OF TMS FOR CORTICAL LANGUAGE SITES

Abstract

Abstract Surgeons must maximize the extent of resection while minimizing postoperative neurological morbidity in patients with eloquent gliomas. Transcranial magnetic stimulation (TMS) has emerged as a novel non-invasive technique to map the functional language cortex. However, multiple studies have documented the plethora of false positives that decrease the predictive value of TMS for intraoperative functional sites. Here we hypothesize that diffusion tensor imaging (DTI) tractography and functional magnetic resonance imaging (fMRI) can serve as adjuncts to identify true positive TMS preoperatively and increase the positive predictive value (PPV). We used a distortion correction algorithm to overlay the postoperative MRI with the preoperative surgical plan and correlated the resection versus preservation of certain structures with functional outcomes measured by pre to postoperative change in the Western Aphasia Battery (WAB) score in a cohort of 61 patients with language eloquent gliomas. We found that the resection of TMS points alone does not correlate with aphasic surgical deficits. We then seeded DTI tractography with TMS points and investigated the effect of progressively increasing fractional anisotropic (FA) thresholds. We found that the resection of TMS points with connecting white matter tracts identified at the 50%, 75%, and 85% FA thresholds strongly correlated with surgical deficits. (r=.38, p=.002; r=.61, p< .001, r=.54, p< .001) The PPV increased with increasing FA thresholds. The highest PPV was 50% at the 75% and 85% FA thresholds while the highest negative predictive value was 98% at the 75% FA threshold. We then combined fMRI with TMS and DTI tractography by only using fMRI-colocalized TMS points for tractography seeding. The PPV increased to 63% but the correlation coefficient with aphasic deficits stayed at .61. This study shows that adjunctive modalities can significantly increase the predictive value of TMS for preoperatively identifying true positive cortical sites of language function.