Abstracts

Rationale:
Using a memory encoding functional MRI (fMRI) paradigm, we previously showed that left lateralised frontal and temporal activations were predictive of greater memory decline after temporal lobe surgery in people with temporal lobe epilepsy (TLE). Memory fMRI was the strongest independent predictor compared to standard outcome measures such as pre-surgical clinical neuropsychometry scores, however some centres may have limited access to fMRI. The relationship between hemispheric and behavioural lateralisation of memory in TLE is unclear. The aims were therefore to (1) examine the relationship between memory fMRI hemispheric and behavioural lateralisation, and (2) create predictive models of hemispheric lateralisation based on behavioural measures.
Method:
Ten patients with left TLE (LTLE), 8 patients with right TLE (RTLE), and 21 healthy controls were administered a memory encoding and retrieval fMRI paradigm of faces and words. A behavioural memory test (“The Pair Test”) was also administered to provide measures of verbal and visual memory. Lateralisation of memory activation was examined with lateralisation index (LI) values, for both memory processes (encoding and retrieval) and both materials (verbal: words, and visual: faces). We conducted: (1) Pearson correlations between LI values and memory scores to establish the relationship between behavioural and hemispheric lateralisation, (2) multiple linear regressions to predict LI values based on memory scores. 
Results:
Better verbal memory was associated with lateralised activation towards the pathological medial temporal lobe (MTL) during word encoding (LTLE: R=0.66, p=0.07; RTLE: R=-0.76 p=0.03), with right lateralisation during word retrieval for LTLE (R=-0.844, p=0.004), and with left lateralisation for RTLE during word retrieval (R=0.76, p=0.03). Better visual memory was associated with lateralisation to the MTL contralateral to the pathology during face encoding (LTLE: R=-0.52, p=0.1; RTLE: R=-0.54, p=0.2), with bilateral activation during face retrieval for LTLE, and with right lateralisation during face retrieval for RTLE (R=0.7, p=0.08). Multiple linear regressions showed that for LTLE, individual verbal memory LIs could be predicted based on verbal memory scores. The models 95% of the variance (R2=0.95, p=0.024). For RTLE group, individual visual memory LIs could be predicted based on visual memory scores. The model explained 93% of variance (R2=0.93 p=0.03).
Conclusion:
The relationship between behavioural and fMRI hemispheric lateralisation varies for encoding and retrieval memory processes. These results support the distinction between encoding and retrieval as distinct processes and emphasise the importance of examining the functional organisation of both memory processes. For the prediction of individual functional hemispheric lateralisation using behavioural data, verbal but not visual memory LIs could be predicted for LTLE, and visual but not verbal memory LIs could be predicted for RTLE. This dissociation is in accordance with material-specific memory impairments. This predictive model of memory functional LI may be of value in the pre-operative assessment and counselling of people with TLE in centres without fMRI facilities.
Funding:
:National Institute for Health Research University College London Hospitals Biomedical Research Center (grant number 229811)