Abstracts

Rationale: Discordance between language lateralization occurs in ~14% of patients. The discordance rate is highest in people with bilateral or cross language lateralization. fMRI and Wada provide distinct information about language lateralization. fMRI indicates hemispheric involvement in language, while WADA indicates which hemisphere is critically required for language. Given the significant negative consequences that can result from inaccurate language lateralization (i.e., resecting dominant language areas), we aim to investigate which factors may contribute to discordance as a first step to understanding why the discrepancy occurs and which patients are at greatest risk of false language lateralization.

Methods: We completed a retrospective review of patients with discordant fMRI and Wada language lateralization. 31 patients were reviewed at BCH between 2002-2014. Of those, there were 9 cases of discordance between fMRI and Wada language lateralization. The majority of patients had a left sided seizure focus (7/9). Several patients also had an MRI lesion (7/9) or temporal seizure onset (7/9). Structural abnormalities included tumor (3/9), heterotopia (2/9), or large hemispheric disease (2/9). The 2 children who had MRI negative epilepsy were later found to have dysplasia via pathology. Pre- and post-operative IQ data was available for 6/9 patients.

Results: fMRI demonstrated bilateral language representation in 78% (7/9) of patients. Of these, 100% (7/7) had left language dominance on Wada testing. The two remaining patients (2/9) showed right language dominance on fMRI, while Wada demonstrated left or bilateral language dominance. Of those with post-operative IQ data, 50% (3/6) had stable verbal IQ (VIQ), 16% (1/6) had VIQ improvement, and 33% (2/6) had VIQ decline. Both of the children with decline had left Wada and left resection. One patient exhibited a large post-operative decline (27 VIQ points), while the other child experienced a small decline following a language sparing resection in the temporoparietal occipital junction (8 VIQ points). Half (3/6) of patients with discordant fMRI and Wada had impaired VIQ at baseline.

Conclusions: Discordance is more likely when fMRI demonstrates bilateral or right hemispheric language. False language lateralization of fMRI may also be more likely in patients with epilepsy related to acquired structural lesions (78% of our sample) that disrupt neurovascular coupling and affect the subsequent BOLD signal. Furthermore, many of our patients with discordant Wada and fMRI had low pre-surgical VIQ between 69-76. Increased language demands elicit greater bilateral activation in left language dominant individuals. For children with low IQ, the task would be more difficult, resulting in greater right hemisphere recruitment which may explain the discordance of fMRI language lateralization and the WADA procedure. Therefore, additional research is necessary to determine how task difficulty impacts the language activation map of children with language processing difficulties. This is crucial for accurate non-invasive language lateralization which ultimately plays an important role in surgical planning and outcomes.

Funding: Please list any funding that was received in support of this abstract.: N/A.