Incidence and linguistic quality of speech errors: a comparison of preoperative transcranial magnetic stimulation and intraoperative direct cortex stimulation

Abstract

Objective: Given the interindividual variance of functional language anatomy, risk prediction based merely on anatomical data is insufficient in language area-related brain tumor surgery, suggesting the need for direct cortical and subcortical mapping during awake surgery. Reliable, noninvasive preoperative methods of language localization hold the potential for reducing the necessity for awake procedures and may improve patient counseling and surgical planning. Repetitive navigated transcranial magnetic stimulation (rnTMS) is an evolving tool for localizing language-eloquent areas. The aim of this study was to investigate the reliability of rnTMS in locating cortical language sites.

Methods: Twenty-five patients with brain tumors in speech-related areas were prospectively evaluated with preoperative rnTMS (5 Hz, train of five, average 105% resting motor threshold) and navigated direct cortical stimulation (DCS; bipolar, 50 Hz, 6-8 mA, 200-μsec pulse width) during awake surgeries employing a picture-naming task. Positive and negative stimulation spots within the craniotomy were documented in the same MRI data set. TMS and DCS language-positive areas were compared with regard to their spatial overlap, their allocation in a cortical parcellation system, and their linguistic qualities.

Results: There were over twofold more positive language spots within the exposed area on rnTMS than on DCS. The comparison of positive rnTMS and DCS (ground truth) overlaps revealed low sensitivity (35%) and low positive predictive value (16%) but high specificity (90%) and high negative predictive value (96%). Within the overlaps, there was no correlation in error quality. On DCS, 73% of language-positive spots were located in the pars opercularis and pars triangularis of the frontal operculum and 24% within the supramarginal gyrus and dorsal portion of the superior temporal gyrus, while on rnTMS language positivity was distributed more evenly over a large number of gyri.

Conclusions: The current protocol for rnTMS for language mapping identified language-negative sites with good dependability but was unable to reliably detect language-positive spots. Further refinements of the technique will be needed to establish rnTMS language mapping as a useful clinical tool.

Keywords: direct cortical stimulation; functional mapping; glioma surgery; language; navigated transcranial magnetic stimulation; oncology; surgical technique.