Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Improving the Validity, Feasibility, and Reliability of Presurgical Language Mapping in Refractory Epilepsy
Poster Session B, Friday, October 25, 10:00 - 11:30 am, Great Hall 3 and 4
Aoife Reardon1-3, Sonia Brownsett1-3, Lisa Gillinder1,4, David Copland1-3, Katie McMahon5; 1University of Queensland, 2Surgical Treatment and Rehabilitation Service (STARS), 3La Trobe University, 4Princess Alexandra Hospital, 5Queensland University of Technology
Background: Neurosurgical interventions are often indicated to achieve seizure freedom in chronic refractory epilepsy. However, with resection comes the risk of aphasia, an acquired language deficit, which can significantly impact upon post-surgical quality of life. Whilst clinicians endeavour to minimise risk through pre-surgical language mapping protocols, evidence suggests the prevalence of post-surgical aphasia may be underestimated. Limitations of currently implemented fMRI language mapping paradigms include omission of known language regions (most notably, the anterior temporal lobe), limited consideration of task monitoring, inadequate tailoring of task difficulty to patient ability, poorly matched baselines, and limited consideration of longitudinal reliability. Aim: To investigate the validity, feasibility, and intrasubject reliability of a combination of current recommended, evidence based, and novel paradigms in a control group, using a test-retest design. Method: Seven language mapping tasks (autobiographical writing, noun semantic decision, verb semantic decision, adaptive semantic, adaptive rhyme, sentence completion, and silent word generation) were completed by healthy controls (n=20) at two timepoints. Tasks were evaluated in terms of 1) feasibility to complete within the MRI scanning environment; 2) validity with reference to a priori regions of interest; and 3) test-retest reliability, by calculating a similarity metric between activation maps obtained at each timepoint. Results: Through monitoring motion artefact, self-report, and number of correct trials (where applicable), all tasks were deemed feasible to complete within the MRI scanning environment. Activation maps for group level univariate contrasts revealed verb semantic decision, adaptive semantic, and autobiographical writing tasks most reliably activated known language regions of interest, with current recommended clinical paradigms (sentence completion and silent word generation) failing to activate all areas of surgical interest, most notably the anterior temporal lobe. Mean intrasubject test-retest reliability for all tasks fell within the fair range (0.4–0.6), with the exception of the adaptive semantic task, which was in the good range (0.6-0.75). Conclusion: Current recommended language mapping paradigms are grounded in historic neuroanatomical models of language, and fail to reliably map all areas of surgical interest, including the most frequently resected region in patients with refractory focal epilepsy, the anterior temporal lobe. Clinical paradigms should incorporate a combination of tasks (written autobiographical, adaptive semantic, adaptive rhyme, verb semantic decision) to ensure that the functional language network is comprehensively mapped. This will ensure patients and clinicians are able to make informed treatment choices when evaluating the risk-benefit trade-off of ongoing seizures against the possibility of post-surgical aphasia.
Topic Areas: Disorders: Acquired,