Longitudinal changes in hippocampal morphology before and after temporal lobe epilepsy surgery

Abstract

Temporal lobe epilepsy (TLE) is associated with progressive neocortical thinning that may be prevented by resective surgery, but little knowledge exists on dynamic changes in the hippocampus. Here, we assessed progressive morphological changes of the hippocampus before and after TLE surgery. In this longitudinal case-control neuroimaging study, we included patients with unilateral drug-resistant TLE before (main cohort n = 23; replication cohort n = 81) or after (n = 54) anterior temporal lobe resection and healthy volunteers (n = 120) matched for age and sex. We compared hippocampal volumes and surface shape morphology on paired scans between groups using linear mixed effects models. We did not find accelerated atrophy of the ipsilateral hippocampus in chronic presurgical TLE (-2.1 µl/year) compared to normal aging (-3.6 µl/year, P = 0.78). In contrast, there was progressive hypertrophy of the contralateral hippocampus before epilepsy surgery (+1.4 µl/year) compared to normal aging (-4.1 µl/year), which was further accelerated after surgery (+4.0 µl/year, P = 0.01). We validated these findings in the replication cohort (n = 81). Factors contributing to contralateral hippocampal hypertrophy included left-sided epilepsy lateralization, ipsilateral hippocampal sclerosis, and visual memory encoding deficits before or after surgery. Conversely, a history of focal to bilateral tonic clonic seizures was associated with accelerated pre- and postsurgical atrophy of the contralateral hippocampus. Refractory TLE is associated with progressive hypertrophy of the contralateral hippocampus before and after resective surgery. These contralateral changes may represent compensatory plasticity that is related to pre- and postoperative disturbances in cognition, mainly affecting visual memory. In contrast to progressive cortical thinning, the ipsilateral hippocampus does not show additional atrophy in chronic TLE.

Keywords: MRI; epilepsy; hippocampus; seizures; surgery.

Graphical Abstract

Figure 1

Progressive morphological changes of the hippocampus in TLE before and after surgery. (A) Changes of ipsilateral (left) and contralateral (right) hippocampal volumes in healthy volunteers (grey), pre- (blue) and postoperative (orange) TLE patients. Each scan is represented by a dot and scans corresponding to the same subject are connected by a thin line. The three thick lines are linear regression lines of mixed effects models and their slopes represent the estimated rate of hippocampal volume change in each group. We used linear mixed effects models in all participants (presurgical TLE, n = 23; postsurgical TLE, n = 54; healthy volunteers, n = 120). Although mixed effects models were fitted with a variable intercept, for demonstration purposes we display the linear regression lines having the same intercept to improve legibility. (B) Subregional longitudinal changes in hippocampal morphology. We compared progressive changes in both hippocampi in presurgical TLE patients with healthy volunteers (left), the contralateral hippocampus in postsurgical TLE patients with healthy volunteers (middle), and the contralateral hippocampus in post- versus presurgical TLE patients (right). Blue clusters indicate significant progressive atrophy, red colours indicate significant progressive hypertrophy. We used linear mixed effects models in all participants (presurgical TLE, n = 23; postsurgical TLE, n = 54; healthy volunteers, n = 120). Significant P values were thresholded to P < 0.05 corrected for multiple comparisons using familywise error (FWE) correction by random field theory. The ipsi- and contralateral hippocampi are visualized from a superior and an inferior perspective. An approximation of major hippocampal subregional boundaries is overlaid on hippocampal surfaces. CA, cornu ammonis; DG, dentate gyrus; FWE, familywise error; HV, healthy volunteers; Sub, subiculum; TLE, temporal lobe epilepsy.

Figure 2

Progressive morphological changes of the hippocampus in TLE before versus after surgery in those with both paired pre- and postsurgical scans. The figure shows subregional longitudinal changes in hippocampal morphology. We compared progressive changes in the contralateral hippocampus in before surgery versus after surgery in TLE patients (n = 8) who had two scans before surgery and two scans after surgery. Blue clusters indicate significant progressive atrophy, yellow/red colours indicate significant progressive hypertrophy. Significant P values calculated using a linear mixed effects model in 8 subjects were thresholded to P < 0.05 corrected for multiple comparisons using familywise error (FWE) correction by random field theory. The ipsi- and contralateral hippocampi are visualized from a superior and an inferior perspective. An approximation of major hippocampal subregional boundaries is overlaid on hippocampal surfaces. CA, cornu ammonis; DG, dentate gyrus; FWE, familywise error; Sub, subiculum; TLE, temporal lobe epilepsy.

Figure 3

Progressive morphological changes of the hippocampus in the replication cohort with unilateral TLE before surgery. The figure shows changes of ipsilateral (A) and contralateral (B) hippocampal volumes in healthy volunteers (grey) and the replication cohort of unilateral preoperative TLE patients (blue). Each scan is represented by a dot and scans corresponding to the same subject are connected by a thin line. The three thick lines are linear regression lines of mixed effects models and their slopes represent the estimated rate of hippocampal volume change in each group. We used linear mixed effects models in all participants (replication TLE cohort, n = 81; healthy volunteers, n = 120). Although mixed effects models were fitted with a variable intercept, for demonstration purposes the linear regression lines are displayed having the same intercept to improve legibility. TLE, temporal lobe epilepsy.

Figure 4

Longitudinal changes of contralateral hippocampal morphology and their association with clinical variables. We correlated the subregional longitudinal changes in contralateral hippocampal morphology before and after epilepsy surgery with clinical variables. We assessed the association with younger age at epilepsy onset (A), longer epilepsy duration (B), left lateralisation of the epileptic focus (C), sclerosis of the ipsilateral hippocampus (D), smaller volume of the ipsilateral hippocampus (E), presurgical seizure frequency (F), presurgical focal to bilateral tonic clonic seizures (FBTCS, G), number of antiseizure medications at surgery (H), ongoing seizures after epilepsy surgery (I) and the size of the hippocampal remnant (J). Blue clusters indicate significantly less progressive hypertrophy, red colours indicate significantly more progressive hypertrophy. We used linear mixed effects models in 23 presurgical and 54 postsurgical TLE patients. Significant P values were thresholded to P < 0.05 corrected for multiple comparisons using familywise error (FWE) correction by random field theory. The contralateral hippocampi are visualized from a superior and an inferior perspective. An approximation of major hippocampal subregional boundaries is overlaid on hippocampal surfaces. CA, cornu ammonis; DG, dentate gyrus; FWE, familywise error; Sub, subiculum.

Figure 5

Longitudinal changes of contralateral hippocampal morphology and their association with neurocognitive testing results. We correlated the subregional longitudinal changes in contralateral hippocampal morphology with neurocognitive testing results. We assessed the association of presurgical hippocampal changes with worse preoperative verbal (A) or visual (B) learning, verbal IQ (C), and performance IQ (D). We also assessed the association of postsurgical hippocampal changes with postoperative decline in verbal (E) or visual (F) learning. Blue clusters indicate significantly less progressive hypertrophy, red colours indicate significantly more progressive hypertrophy. We used linear mixed effects models in 23 presurgical and 54 postsurgical TLE patients. Significant P values were thresholded to P < 0.05 corrected for multiple comparisons using familywise error (FWE) correction by random field theory. The contralateral hippocampi are visualized from a superior and an inferior perspective. An approximation of major hippocampal subregional boundaries is overlaid on hippocampal surfaces. CA, cornu ammonis; DG, dentate gyrus; FWE, familywise error; IQ, intelligence quotient; Sub, subiculum.