Abstracts

Rationale: Left and right temporal lobe epilepsy (LTLE and RTLE) are associated with white and grey matter abnormalities. Measures of white and gray matter integrity have each been used to predict neurocognitive performance in these patients [1, 2]. Less known is which of these structural factors, separate or combined in an interactive model, produce the strongest relationship with neurocognitive performance. The purpose of this study was to investigate the effect of specific white matter tracts, hippocampal volumes, and their interaction, on expressive language and verbal learning performance in LTLE and RTLE patients.Methods: Structural MRI, diffusion tensor imaging, and fMRI were obtained from 21 LTLE, 23 RTLE, and 28 healthy controls. All subjects were left hemispheric dominant and no difference existed between groups in age or epilepsy duration. Atlas-guided tractography was used to extract four tracts bilaterally: parahippocampal fasciculus, uncinate fasciculus, fronto-temporal arcuate, and fronto-parietal fibers. After investigating the correlation between hippocampal volume (HV) and fractional anisotropy (FA) of the four tracts, hierarchical stepwise regression was conducted to determine the best predictors of the Boston Naming Test (BNT) and the California Verbal Learning Test (CVLT-TL, Total Learning) within the TLE groups. The interaction between HV and FA was included for tracts with significant predictive power. All results represent effects after accounting for epilepsy duration.Results: Smaller HV was found for longer epilepsy durations bilaterally in LTLE and ipsilaterally in RTLE. Controls showed no correlation between HV and FA in any of the tracts. In RTLE patients, right HV positively correlated with FA of the right uncinate, right fronto-parietal fibers, and left arcuate. In LTLE patients, left HV positively correlated with FA of all four tracts bilaterally. After regressing out seizure duration, BNT performance was predicted by FA of the parahippocampal tract ipsilateral to the disease in both TLE groups (LTLE R²=0.17; RTLE R²=0.39). CVLT-TL performance was predicted by the left arcuate in RTLE (R²=0.35), and by the interaction of right HV with left parahippocampal FA in LTLE (R²=0.45).Conclusions: An abnormal link between gray (HV) and white (FA) matter properties arises in TLE patients. The association between HV and FA is more extensive in LTLE patients, involving bilateral white matter tracts. A functionally adaptive interaction between white (left parahippocampal tract) and gray (right hippocampal volume) matter properties was found in LTLE, with better structural integrity on these measures predicting better verbal learning. In these left hemisphere dominant patients, this pattern suggests potential adaptive reorganization of verbal learning networks. Given the link between epilepsy duration and selective white and gray matter compromise, the interaction between the latter suggests early intervention is important to reduce cognitive deficits and protect the integrity of available cognitive reserves.