gms | German Medical Science

Objective: Losing language function following tumour resection has considerable impact on quality of life, hence appropriate presurgical planning is needed. Navigated transcranial magnetic stimulation (nTMS) is a non-invasive method that may be effective in evaluating which cortical regions are important for different aspects of language ability. Language function requires interaction between perisylvian regions, connected by the arcuate fasciculus (AF). nTMS and diffusion imaging methods can be combined to assess the relationship between cortical stimulation and underlying white matter. The AF is subdivided into a direct long segment (LS; fronto-temporal) and two indirect segments, anterior (AS; fronto-parietal) and posterior (PS; parieto-temporal). These may contribute to different aspects of language function. We set out to identify if there were common speech errors specific to different segments of the arcuate fasciculus.

Methods: 22 patients with left hemisphere gliomas were recruited, who underwent nTMS language mapping. Trains of 5 stimuli at 5 Hz frequency were applied over the perisylvian cortex while the patient named images of different semantic categories, displayed as black and white drawings on a screen. Misnamings were categorized into motor speech errors and semantic errors. On average 80 cortical areas were targeted with nTMS three times in each subject in the left hemisphere. Diffusion imaging was perfomed and the three branches of the AF reconstructed using deterministic spherical deconvolution tractography. The nTMS stimulation locations were then overlaid with the images to identify overlap.

Results: All patients produced motor errors, and 17/22 produced semantic errors. Over 50% of total errors were at locations corresponding to the AF. Motor speech errors occurred at locations corresponding to the AS in 16/22 patients, the LS in 10/22 patients and the PS in 8/22 patients. Significantly more motor errors occurred at stimulation points at terminations of the AS than the PS (p <0.03). Semantic errors were less common. Of the total errors overlapping with the AF more occurred in sites corresponding to the posterior segment than the other segments (AS in 3/17 patients, LS in 2/17 and PS in 7/17).

Conclusion: Our results support the idea of the AF as the major language tract. Moreover, the different types of errors elicited by nTMS may be related to stimulation of different segments of the arcuate fasciculus. This supports current theories that propose the posterior segment may be important for object recognition to verbal recall (and hence, semantic processes) and the anterior segment for verbal recall to speech production, motor aspects of speech. The results of this preliminary work may be important for evaluating risk pre-operatively, and further, the appropriate focus for speech mapping during awake surgery.