One reasonable way to anticipate and conceptualize the cognitive correlates of specific epilepsy syndromes is through their primary underlying pathophysiology, that is, memory impairment characterizing temporal lobe epilepsy, executive impairment prominent in frontal lobe epilepsies, and attentional problems dominating in absence epilepsy [1
]. Memory problems can certainly be documented in temporal lobe epilepsy as would be expected given the primary neuropathology (hippocampal sclerosis) [41
], but when characterized in a comprehensive fashion, cognitive abnormalities appear more distributed and generalized than expected [12
]. While this “average cognitive profile” provides a somewhat different view of the neuropsychological consequences of temporal lobe epilepsy, clinical experience suggests that not all patients actually exhibit such a profile. That is, individual patients with mesial temporal lobe epilepsy may range from rather unscathed to severely affected mental status. Somewhat the same might be said regarding the structural abnormalities reported to be associated with temporal lobe epilepsy, abnormalities that have been found to extend beyond the epileptogenic hippocampus to affect a diversity of temporal and extratemporal lobe regions both ipsilateral as well as contralateral to the side of seizure onset [16
]. It has been our hypothesis that non-overlapping groups of patients may be identified who present with varying cognitive profiles and associated anatomic abnormality, findings that would identify important phenotypes for future research.
To address this issue, we adjusted (age, gender, education) cognitive test scores based on a healthy control population, derived cognitive domain scores, and used cluster analysis to determine whether meaningful profiles of cognitive status could be identified. Specific “cognitive phenotypes” were identified () with associated clinical seizure and demographic characteristics, increasing abnormalities in summary volumetric measures of total cerebral gray and white matter and hippocampi, and increasingly abnormal prospective (4 year) cognitive trajectories [25
]. Here we take advantage of sophisticated postprocessing techniques to comprehensively examine patterns of abnormality in cortical thickness and volumes of subcortical structures, callosal, and cerebellar regions across these cognitive phenotypes. The results are noteworthy for the regularity of evident neuroanatomic abnormality across the cognitive phenotypes, and perhaps surprising for the distribution and degree of abnormality.
First, the cognitive phenotypes were distinguished by abnormalities across very diverse cortical regions (left frontal lobe, left lateral temporal lobe, left and right parietal lobe, left and right occipital lobe), subcortical structures (bilateral hippocampus, thalamus, caudate, and gray matter of the cerebellum), and white matter (all regions of the corpus callosum). Second, the pattern of abnormality in all these abnormal regions was quite uniform in that those patients with the most impaired cognition (Cluster 3) exhibited the most significant cortical thinning (5 to 8%) and volumetric reductions across all subcortical (10 to 18%), cerebellar gray matter (14%) and white matter regions of interest (17 to 26%). These same anatomic measures were least abnormal in the most cognitively intact temporal lobe epilepsy group (<1 to 4% for cortical thinning, 4 to 10% for subcortical structures, 4 to 5% for cerebellar gray matter, 5 to 8% for regions of the corpus callosum). This symmetry between cognitive profile and brain structure speaks to the importance of underlying neuroanatomic abnormalities in the etiology of cognitive impairment compared to what might be considered to be more transient factors (e.g., epilepsy medications, epileptiform spike frequency) or clinical seizure characteristics (e.g., seizure frequency), although these variables certainly play a role. Third, the patterns of abnormality were often bilateral in nature affecting both cortical regions (), subcortical structures (thalamus, hippocampus, caudate), and cerebellar gray matter (), in conjunction with substantial white matter abnormality (). That there can be such widely distributed pathology in a localization related form of epilepsy is both surprising as well as a finding of concern. Fourth, cortical thinning was especially evident in posterior brain regions including parietal, occipital, posterior temporal and sensorimotor regions. We would have anticipated more anterior and frontal abnormalities, but these regions did not distinguish the clusters in a broad fashion, although there were unilateral frontal lobe differences. Abnormalities in these posterior regions have been reported in recent studies of cortical thickness in temporal lobe epilepsy [17
] and are found here to especially characterize the more cognitively impaired cohort (Cluster 3). Executive and motor functions are especially affected in Cluster 3 and it is possible that executive function is disrupted predominantly by abnormalities in subcortical structures known to be an intrinsic component of fronto-striatal systems that mediates these abilities [42
]. Finally, prior neuroimaging studies have presented modal profiles of structural abnormalities in temporal lobe epilepsy. As is the case for cognition, these modal profiles do not reflect the substantial degree of variability that may be present across patients. The interesting diversity of anatomic abnormality speaks to this underlying individual variability with clear implications for cognition.
Overall, these findings help characterize the neuroanatomic status of cognitive phenotypes that have associated patterns of prospective cognitive courses. As we have reported previously, the clinical and demographic features are not that strikingly different compared to the divergent nature of the cognitive and imaging results [25
]. Interestingly, despite the fact that these epilepsy patients were recruited from tertiary care centers that tend to care for more adversely affected patients, the largest cohort of patients (Cluster 1, 47% of subjects) exhibited comparatively intact cognition and the least affected brain structure, despite a prolonged course of epilepsy.
While we have presented considerable information regarding the identification of cognitive phenotypes; their clinical, demographic, and structural features as well as their prospective cognitive course, critical issues remain to be clarified regarding the origin, etiology, and epigenetics of these groups.
The limitations of this investigation should be recognized. The number of subjects is modest, we are not able to examine cognitive and structural patterns by lateralization of temporal lobe epilepsy, and as we are examining only patients with temporal lobe epilepsy the degree to which these phenotypes may be represented in other forms of epilepsy cannot be addressed. Future research should replicate these clusters and their associated features in a larger sample of patients.