Abstracts

Rationale: Functional MRI is a reliable method for presurgical language mapping in children undergoing surgery for medically intractable epilepsy. To accurately identify regions of language function, fMRI paradigms must identify both frontal and temporal language areas. Children with epilepsy may have atypical language patterns, including crossed dominance, in which frontal and temporal lobe language areas are seen in opposite hemispheres. Therefore, it is not possible to determine hemisphere dominance for language unless both frontal and temporal areas are identified. The objective of this study is to identify fMRI language paradigms which may be used in combination to ensure adequate mapping of frontal and temporal lobe language areas in children with epilepsy. Methods: Functional MRI for presurgical language mapping was performed in 17 children, 6 male, age 8-17 years. 13 children were right handed. All children successfully completed two standardized visually presented paradigms; letter fluency (LF) and verb generation (VG) and one auditory semantic decision (AD) paradigm with reliable results. Studies were analyzed using AFNI and visually inspected and rated for lateralization and localization of frontal and temporal lobe language areas. Results: All three tasks were more likely to activate frontal than temporal language areas. LF activated only frontal lobe language areas in 5 children and both frontal and temporal areas in 12 children. VG activated only frontal lobe language areas in 7 children and frontal and temporal language areas in 10 children. AD activated both the frontal and temporal language areas in 15 cases. The best concordance was seen between LF and AD which agreed in frontal and temporal language activations in 12 /17 cases. The 3 language paradigms demonstrated the same result in only 5 of 17 studies (30%). In 8 of 17 (47 %), LF and VG produced the same patterns of activations. In 12 cases, the addition of a third task produced activations not seen using 2 paradigms alone. For example, in 3 cases in which LF and VG failed to demonstrate temporal lobe activations, AD produced temporal lobe activations. All tasks identified atypical language patterns; LF demonstrated atypical language patterns in 11 children, VG in 9 children and AD in 10 children. Conclusions: In 70% of this cohort of children with epilepsy, 3 fMRI language paradigms were necessary to completely demonstrate frontal and temporal lobe language areas. Functional MRI acquisition may be challenging in children, particularly those with epilepsy and comorbid cognitive and developmental issues. Despite this, all efforts should be made to acquire sufficient data to fully characterize language patterns for presurgical mapping in children with epilepsy.