gms | German Medical Science

Objective: Repetitive navigated transcranial magnetic stimulation (rnTMS) is increasingly used for preoperative cortical language mapping due to its advantage of non-invasiveness and usability in preoperative testing and planning. To further elucidate the pattern of language errors induced by rnTMS on a group level, subjects’ individual space data were transferred to a normalized space template (Montreal Neurological Institute, MNI).

Methods: We collected language rnTMS data of patients with tumor or vascular disease in language related regions between 2011 and 2018. The coordinates of positive and negative stimulation responses were registered to MNI space. Subsequently, we computed the distributions of errors and non-errors (er, n errors /n stimulations=error rate) derived from our nrTMS based stimulations. To define the anatomical location and distribution of er, the cortex was segmented using automated anatomical labeling (AAL).

Results: We included 171 patients in this study with tumors or vascular lesions in the left or right perisylvian area. On each hemisphere an average of 196.3±66.9 rnTMS trains was applied, inducing a mean er of 5.5%. We observed no significant difference concerning er distribution (p=0.145) between left and right hemispheres in patients with left hemisphere tumors (n=134). However, we found a highly significant correlation (p<0.001) between er distribution and pathology localization (anterior vs. posterior), with more errors posteriorly in anterior tumors and vice versa. Additionally, we observed a significant difference (p=0.044) in error distribution between tumor patients (n=149) and vascular disease patients (n=22), indicating higher er (5.7%) in tumor patients and lower er (3.7%) in vascular disease patients.

Conclusion: Analysis of rnTMS language mapping data in normalized space applying automated anatomical labeling allows for standardized analysis of large data sets. In contrast to earlier studies, no difference in er was observed between left and right hemisphere. Tumor location highly significantly impacted the pattern of induced language erros intrahemispherically. Moreover, a significant difference in error distribution depending on the type of disease (tumor vs. vascular) was observed.