Abnormal hippocampal structure and function in juvenile myoclonic epilepsy and unaffected siblings

Abstract

Juvenile myoclonic epilepsy is the most common genetic generalized epilepsy syndrome, characterized by a complex polygenetic aetiology. Structural and functional MRI studies demonstrated mesial or lateral frontal cortical derangements and impaired fronto-cortico-subcortical connectivity in patients and their unaffected siblings. The presence of hippocampal abnormalities and associated memory deficits is controversial, and functional MRI studies in juvenile myoclonic epilepsy have not tested hippocampal activation. In this observational study, we implemented multi-modal MRI and neuropsychological data to investigate hippocampal structure and function in 37 patients with juvenile myoclonic epilepsy, 16 unaffected siblings and 20 healthy controls, comparable for age, gender, handedness and hemispheric dominance as assessed with language laterality indices. Automated hippocampal volumetry was complemented by validated qualitative and quantitative morphological criteria to detect hippocampal malrotation, assumed to represent a neurodevelopmental marker. Neuropsychological measures of verbal and visuo-spatial learning and an event-related verbal and visual memory functional MRI paradigm addressed mesiotemporal function. We detected a reduction of mean left hippocampal volume in patients and their siblings compared with controls (P < 0.01). Unilateral or bilateral hippocampal malrotation was identified in 51% of patients and 50% of siblings, against 15% of controls (P < 0.05). For bilateral hippocampi, quantitative markers of verticalization had significantly larger values in patients and siblings compared with controls (P < 0.05). In the patient subgroup, there was no relationship between structural measures and age at disease onset or degree of seizure control. No overt impairment of verbal and visual memory was identified with neuropsychological tests. Functional mapping highlighted atypical patterns of hippocampal activation, pointing to abnormal recruitment during verbal encoding in patients and their siblings [P < 0.05, familywise error (FWE)-corrected]. Subgroup analyses indicated distinct profiles of hypoactivation along the hippocampal long axis in juvenile myoclonic epilepsy patients with and without malrotation; patients with malrotation also exhibited reduced frontal recruitment for verbal memory, and more pronounced left posterior hippocampal involvement for visual memory. Linear models across the entire study cohort indicated significant associations between morphological markers of hippocampal positioning and hippocampal activation for verbal items (all P < 0.05, FWE-corrected). We demonstrate abnormalities of hippocampal volume, shape and positioning in patients with juvenile myoclonic epilepsy and their siblings, which are associated with reorganization of function and imply an underlying neurodevelopmental mechanism with expression during the prenatal stage. Co-segregation of abnormal hippocampal morphology in patients and their siblings is suggestive of a genetic imaging phenotype, independent of disease activity, and can be construed as a novel endophenotype of juvenile myoclonic epilepsy.

Keywords: endophenotype; hippocampal malrotation; juvenile myoclonic epilepsy; magnetic resonance imaging; memory.

Figure 1

Qualitative and quantitative assessment of the hippocampus. (A) An example of a left hippocampus with HIMAL, as evidenced by a round hippocampal shape, loss of the lateral hippocampal convexity, the latter implying an enhanced radius of curvature (left-hand arrow), and verticalization of the DITS (right-hand arrow). The right hippocampus does not exhibit clear-cut features of HIMAL. (B) An example of a subject with normal hippocampi. The right-hand parts of A and B provide demonstrations of the three quantitative measures implemented in the study: diameter ratio, i.e. z divided by x; DITS height ratio, i.e. z divided by a, or value equal to zero in the absence of overlap between DITS and hippocampus, resulting in no measurable a; and parahippocampal angle, for a hippocampus with HIMAL (A) and for a normal hippocampus (B). See main text for details.

Figure 2

Activations related to subsequent verbal memory. Activations in relation to subsequently remembered verbal items (WR-WF) are shown across all subjects (A) and separately for each group [B, controls; C, siblings (SIB); D, patients with JME]. Statistical maps for mesiotemporal activations (orange-yellow scale) show voxels included in the 12-mm diameter spherical regions of interest used for multiple comparison correction, centred on local maxima, where peak-level significance at P < 0.05-FWE corresponds to z-scores >2.3). Extra-mesiotemporal activations are displayed at an uncorrected threshold (P < 0.01, cluster extent threshold of 10 voxels; ‘hot’ colour scale). ‘LH’ refers to a sagittal section of the left hemisphere. Colour bars reflect z-score scales for mesiotemporal (right) and extra-mesiotemporal activations (left). MNI coordinates and parameter estimates are provided in Supplementary Table 2.

Figure 3

Group comparisons for subsequent verbal memory activations. The figure shows group comparisons for activation related to subsequently remembered verbal items (WR-WF) in patients with JME compared with controls (A), along with subgroup analyses for JME with HIMAL (B) and JME without HIMAL versus controls (C), and for the direct comparison of JME with and without HIMAL (D). Comparison of JME siblings against controls is shown in E. Statistical maps for mesiotemporal activations (cyan scale) show voxels included in the 12-mm diameter spherical regions of interest used for multiple comparison correction, where peak-level significance at P < 0.05-FWE corresponds to z-scores >2.3). Extra-mesiotemporal activations are displayed at an uncorrected threshold (P < 0.01, cluster extent threshold of 10 voxels; blue-green colour scale). ‘LH’ refers to a sagittal section of the left hemisphere. Colour bars reflect z-score scales for mesiotemporal (right) and extra-mesiotemporal activations (left). MNI coordinates are provided in Supplementary Table 3.

Figure 4

Activations related to subsequent visual memory. Mesiotemporal activations in relation to subsequently remembered visual items (FR – FF) are shown across all subjects (A) as well as separately for each group [B, controls; C, siblings (SIB); D, patients with JME]. Statistical maps for mesiotemporal activations (orange-yellow scale) show voxels included in the 12-mm diameter spherical regions of interest used for multiple comparison correction, where peak-level significance at P < 0.05-FWE corresponds to z-scores >2.3). Extra-mesiotemporal activations are displayed at an uncorrected threshold (P < 0.01, cluster extent threshold of 10 voxels; ‘hot’ colour scale). Colour bars reflect z-score scales for mesiotemporal (right) and extra-mesiotemporal (left) activations. ‘LH/RH’ refer to sagittal sections of the left/right hemisphere, respectively. MNI coordinates and parameter estimates are provided in Supplementary Table 4.

Figure 5

Subgroup analyses for subsequent visual memory. The figure shows subgroup analyses comparing mesiotemporal activation for subsequently remembered visual items (FR − FF) in JME with and without HIMAL against healthy controls (A and C, respectively), along with the direct contrast of JME with and without HIMAL (B). Statistical maps for mesiotemporal activations (orange-yellow scale) show voxels included in the 12-mm diameter spheres used for multiple comparison correction, where peak-level significance at P < 0.05-FWE corresponds to z-scores >2.3. Extra-mesiotemporal activations are displayed at an uncorrected threshold (P < 0.01, cluster extent threshold of 10 voxels; ‘hot’ colour scale). ‘LH/RH’ refers to a sagittal section of the left/right hemisphere, respectively. Colour bars reflect z-score scales for mesiotemporal (right) and extra-mesiotemporal (left) activations. MNI coordinates are provided in Supplementary Table 5.

Figure 6

Structure-function relations. The scatterplots highlight the association between the composite morphological indicator of left hippocampal positioning and activation of the left hippocampus (orange-yellow scale) and middle frontal gyrus (‘hot’ colour scale) for successfully encoded verbal items. (A) Analysis of all study subjects. (B) Analysis of a subgroup composed of patients with JME and their siblings. Local activation maxima, corresponding to the area of maximal correlation between structural and functional metrics, are shown on coronal sections. Maps for mesiotemporal effects show voxels included in the 12-mm diameter sphere centred on the local maximum, where peak-level significance at P < 0.05-FWE corresponds to a z-score >2.3. Extra-mesiotemporal activations are displayed at an uncorrected threshold (P < 0.01, cluster extent threshold of 10 voxel), corresponding to z scores >2.3. Colour-bars reflect z-score scales for mesiotemporal and extra-mesiotemporal activations. The P-values reported on the scatterplots for hippocampal effects are small-volume corrected for multiple comparisons (asterisk); the P-values for middle frontal effects are uncorrected for multiple comparisons. MNI coordinates are provided in Supplementary Table 6.