LAUSR

Objective Using neuronavigated repetitive transcranial magnetic stimulation (rTMS) to disrupt language function during an online-task (e.g., picture naming) is a novel mapping technique which has recently been introduced for presurgical diagnostics in brain tumor patients. However, the method has some limitations due to its rather low specificity. Beyond its dependency on the task, the distinct error categories may have a major effect on the mapping results. We, here, studied the reliability and the spatial localization of rTMS-induced naming errors related to motor-speech function in contrast to semantic language errors. Methods 13 right-handed, healthy volunteers were investigated in 3 consecutive sessions (short-term: 2–5 days, long-term: 21–40 days) by 10 Hz-rTMS. First, the primary motor representation of the face (M1) was mapped by single-pulse TMS. Then, the minimal stimulation intensity which was required to disrupt rhythmic tongue movements by at least three out of five consecutive stimulations over the M1 hotspot was assessed by two independent observers. The “motor inhibition threshold” (MIT) determined the stimulation intensity for the following rTMS language mapping. Online-rTMS was then applied during a picture-naming task using black-and-white drawings of everyday objects which were presented simultaneously (triggered, delay = 0 ms) with the rTMS. The rTMS bursts lasted for 1.5 s (15 pulses). The continuous stimulated area of the left hemisphere had the following anatomical boundaries: middle temporal sulcus (caudal), angular gyrus/ middle parietal sulcus (mediodorsal), hand knob and middle frontal sulcus (medial), triangular part of the inferior frontal gyrus (rostral). Errors were rated by two independent raters by post hoc video analysis. The error frequency (number of errors per 100 TMS trains) as well as the spatial representation and the reliability of distinct motor-speech-related errors, i.e., speech arrest and dysarthria, were compared to language errors associated with semantic processing, i.e., anomia and semantic paraphasia. Test-retest-reliability was assessed by the average intraclass correlation coefficient for a fixed rater (ICC). Results Overall, speech/ language errors were rare events. Amongst the four analyzed error categories, dysarthria represented the most frequent error ( 4.0 ± 2.5 % ) whereas only very few semantic paraphasias were induced ( 0.5 ± 0.4 % ). Motor-speech errors were found significantly more often than semantic language errors ( p 0.01 ). In comparison to semantic errors, the test-retest-reliability of the error rates was lower for motor-speech errors (ICC = 0.45 vs. 0.64). By contrast, the spatial reliability was higher for motor speech errors and showed a strong clustering over M1 and the premotor cortex, especially for the category “arrest”. Conclusion RTMS language mapping is a promising novel technique but the results are still hard to interpret and highly variable. The great majority of errors seems to be at least partially due to motor network disruption rather than being induced by interference with language processing. Further studies, combining different neuroimaging techniques and online-EEG, are mandatory to gain further insight into the underlying mechanisms and the brain-state-dependency of rTMS during language tasks.