The reliability and clinical validity of functional magnetic resonance imaging in the assessment of language in pre-surgical patients with temporal lobe epilepsy.
Defining language lateralisation is important to minimise morbidity in patients treated surgically for temporal lobe epilepsy (TLE). Functional magnetic resonance imaging (fMRI) offers a promising, non-invasive, alternative strategy to the Wada test. Here, fMRI has been used to study healthy controls and patients with TLE in order (i) to define language-related activation patterns and their reproducibility; (ii) to compare lateralisation determined by fMRI with that from the Wada test; and (iii) to explore the usefulness of multiple fMRI language paradigms. 18 healthy controls (12 right-handed and 6 left-handed) and 24 pre-operative TLE patients (19 right-handed: 12 left-TLE, 7 right-TLE; 5 left-handed: 2 right-TLE, 3 left-TLE) were studied using fMRI. Four fMRI language paradigms used: phonetic and semantic fluency, and the naming of living and non-living things. The data for all 4 tasks were acquired during a single scanning session on two occasions. All patients also underwent Wada testing. In patients and controls, phonetic and semantic fluency tasks were robustly activating and strongly lateralising. Quantified language-related lateralisation from fMRI verbal fluency data was highly reproducible and concordant with the lateralisation of the Wada test. Both fluency tasks identified patients with atypical language lateralisation, including 4/12 right-handed patients with left-TLE and 4/5 left-handed TLE patients, regardless of the side of epileptic focus. In comparison, the two confrontational naming tasks were not strongly lateralising and did not reliably agree with Wada lateralisation in either 12 right-handed controls or 19 right-handed patients with TLE. However, there was a difference in the pattern of fMRI activation in right-handed pat ients with left-TLE. Left-TLE patients had a more right lateralised network of activation when naming living things relative to non-living things, suggesting that some patients may be at risk of a category specific naming decline for non-living things after left anterior temporal lobectomy. These results demonstrate that non-invasive fMRI measures of languagerelated lateralisation may provide a practical and reliable alternative to invasive testing for pre-surgical language lateralisation in patients with TLE. The high proportion of TLE patients showing atypical language lateralisation suggests considerable plasticity of language representation in the brains of patients with intractable TLE.