gms | German Medical Science

Objective: To assess the concordance of intraoperative language mapping by cortical stimulation (CSM) and presurgical functional magnetic resonance imaging (fMRI) despite the underlying fundamental methodical differences in patients with left hemispheric gliomas in the frontotemporal cortex.

Methods: A literature review on the various CSM-parameters applied in 16 different hospitals in the US, Europe and Australia was done. Our own data on 157 single language cortex stimulations, their correlation with the local BOLD-response in fMRI and a comparison to the available literature is shown (calculations done with CSM results as “gold standard”).

Results: In the different institutions, there was a wide variety of CSM-parameters. Stimulation amplitude ranged from 4 to 20 mA, single impulse-duration was 0,2 to 4,0 ms, stimulation frequency 40-60 Hz, and various wave-forms exist with a total duration of stimulation between 1 to 6 seconds. Our own data (50 Hz, amplitude 4-20 mA, impulse-duration 0,5 ms, 4-second-train of biphasic square wave pulses) showed a sensitivity of 35,5%, a specificity of 84,4% and a positive/negative predictive value of 33,3 / 85,0%, respectively, without a statistically significant dependency on the amplitude. A literature review on correlation analysis of CSM and fMRI showed that our local results are in good agreement with other series (specifcity from 61 to 98%, lower sensitivity, positive > negative predictive value – Rutten et al. 2002, Roux et al. 2003).

Conclusions: Our results show a good agreement with the published studies on CSM/fMRI-correlations. CSM reliably identifies regions essential for speech production, fMRI shows relative hyperperfusion of all regions of a network, essentially and supplementarily involved in language processing without focusing on their internal hierarchy, in comparison to a control condition. Due to this methodological difference, an excellent sensitivity or positive predictive value will hardly ever be realizable. Nevertheless, specificity and negative predictive value are quite high and show that both methods do safely identify regions free of language function. Since both methods may produce false positive or negative results, an individual analysis of fMRI-data in a robust paradigm with regard to the individual language abilities should be performed. Ideally, the same paradigm should be used for CSM and fMRI. Optimizing these complementary methods should be the main goal for the future, since identification of the functionally “silent” cortex is essential for planning the cortical transection.