Validating language localization by transcranial magnetic stimulation: A comparative study of adults and children

Transcranial magnetic stimulation (TMS) is increasingly used for language mapping in epilepsy surgery. However, due to small sample sizes and inconsistent definitions of overlap, when compared to direct cortical stimulation (DCS), TMS has shown variable sensitivity and specificity. Furthermore, performance of TMS in pediatric cohort is not well understood. This study aimed to standardize the TMS-DCS overlap criteria using a Euclidean distance method and then document the efficacy of TMS in pediatric and adult patients. Using this approach, we evaluated the statistical performance of TMS language localization in eight children (mean age 15) and seven adults (mean age 28) with refractory epilepsy all of whom underwent extra-operative DCS mapping. Overall, TMS showed 89.6% specificity, 81.5% negative predictive value, and 75% accuracy, with high false negatives and positives, resulting in 12.7% sensitivity and 22% positive predictive value (PPV). In subgroup analyses, children showed significantly higher sensitivity (16.3% vs. 5.3%; p = 0.026) and numerically higher PPV (28.9% vs. 8.7%) than adults. The results of this study add to prior evidence TMS may be most useful for predicting language-negative regions for surgery. In children, TMS can still provide useful adjunctive information where other language mapping modalities are less effective. The proposed validation method can be readily adopted across centers to optimize TMS parameters, aimed at improving its sensitivity and overall accuracy.