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

Using multi-modal MRI, Caciagli et al. document abnormalities of hippocampal volume, shape and positioning in patients with juvenile myoclonic epilepsy and their unaffected siblings, which are associated with altered memory-associated functional activations. These findings represent novel genetic imaging phenotypes, and imply a neurodevelopmental mechanism with expression during the prenatal stage.


SUPPLEMENTARY MATERIAL 1. Sensitivity analyses for left hippocampal volumetry.
Inspection of Q-Q plots and normality tests revealed normal distribution of hippocampal volumetric data (Shapiro-Wilk test, all p>0.05 for the whole study sample as well for separate analyses in each study subgroup).
These results corroborate findings of the main analysis and show further convergence towards similar values for mean hippocampal volume in patients with JME and their siblings.

SUPPLEMENTARY MATERIAL 2. Subgroup discrimination via measures of hippocampal volumetry and positioning.
To complement validation of structural hippocampal anomalies in JME and siblings as endophenotypes, we assessed whether quantitative hippocampal measures would be accurate in achieving subgroup discrimination. Using receiver operating characteristic (ROC) curves, we assessed discrimination accuracy of (1) left hippocampal volume and (2) left hippocampal diameter ratio. The latter was chosen because of the significant group effects in the MANOVA for quantitative traits related to HIMAL and the statistically significant post-hoc comparisons for JME and siblings versus controls. We also investigated (3) the role of a combination of volume and positioning via entering the above measures in a principal component analysis. The first principal component obtained from left hippocampal volume and diameter ratio (eigenvalue: 1.38, accounting for 69% of the total variance) was considered as a composite left hippocampal marker. We also repeated all the above models controlling for age, gender and handedness. In the latter case, the first principal component By showing relatively high discrimination of (a) JME patients and (b) a combined JMEsibling group from controls, this supplementary analysis confirms co-segregation of left hippocampal morphological patterns in patients and their relatives, validating the endophenotypic potential of the quantitative markers identified in our study.

SUPPLEMENTARY FIGURE 1. Representative examples of HIMAL.
Abbreviations: CTR = healthy controls; HIMAL = hippocampal malrotation; JME = patient with juvenile myoclonic epilepsy; SIB = unaffected sibling of patient with JME. Examples of subjects with JME and bilateral HIMAL (A) or unilateral (left) HIMAL (B) are presented in the first row. In the second row, panel C, left-hand side, a scan of an unaffected JME sibling is shown, where HIMAL is associated with atypical morphometry of the inferior temporal sulci. Panel D provides an example of HIMAL in a healthy control. Abbreviations: CI = confidence interval; CTR = controls; JME = patients with juvenile myoclonic epilepsy; MNI = Montreal Neurological Institute; SIB = siblings of patients with juvenile myoclonic epilepsy.

SUPPLEMENTARY
Coordinates for mesiotemporal and extra-mesiotemporal activations are given in MNI space. When in bold, Pvalues for peak-level mesiotemporal activations are family-wise error rate (FWE) corrected for multiple comparisons using a small volume correction within a 12-mm diameter sphere, centred on the local activation maximum. P-values not in bold are uncorrected for multiple comparisons. Parameter estimates (i.e., betas) are reported along with their 95% confidence intervals (CI). For a given anatomical region, statistics are reported for up to three peak-level local activation maxima, ordered by statistical significance. For the JME group, there were no supra-threshold voxels. For completeness, we report coordinates of sub-threshold activation for the two locations showing significant group effects across all subjects (left hippocampus and left inferior frontal gyrus). Coordinates of mesiotemporal and extra-mesiotemporal group differences are given in MNI space. P values for peak-level mesiotemporal activations, all reported in bold font, are family-wise error rate (FWE) corrected for multiple comparisons (small-volume correction) using a 12-mm diameter sphere centred on the local activation maximum. For group differences regarding left mesiotemporal activation, Z-scores and P values in brackets refer to repeat models including left hippocampal volume as covariate.

Region
When in bold, P values for peak-level extra-mesiotemporal group differences are FWE-corrected for multiple comparisons across the whole brain (e.g. right middle frontal gyrus, JME-noHIMAL vs JME-HIMAL). When not in bold, P values for peak-level extra-mesiotemporal differences are uncorrected across the whole brain.
For a given anatomical region, statistics are reported for up to three peak-level local maxima, ordered by statistical significance. Coordinates for mesiotemporal and extra-mesiotemporal activations are given in MNI space. When in bold, Pvalues for peak-level mesiotemporal activations are family-wise error rate (FWE) corrected for multiple comparisons using a small volume correction within a 12-mm diameter sphere, centred on the local activation maximum. P-values not in bold are uncorrected for multiple comparisons. Parameter estimates (i.e., betas) are reported along with their 95% confidence intervals (CI). For a given anatomical region, statistics are reported for up to three peak-level local maxima, ordered by statistical significance. For the SIB group, there were no supra-threshold mesiotemporal voxels. Coordinates of sub-threshold mesiotemporal activation are reported for completeness.  Coordinates of mesiotemporal and extra-mesiotemporal group differences are given in MNI space. P values for peak-level mesiotemporal comparisons (reported in bold font) are family-wise error rate (FWE) corrected for multiple comparisons using a 12-mm diameter sphere, centred on the local activation maximum. Z-scores and P values in brackets refer to repeat models including left hippocampal volume as covariate. P values for peaklevel extra-mesiotemporal differences (all not in bold) are reported as uncorrected across the whole brain. are family-wise error rate (FWE) corrected for multiple comparisons using a 12-mm diameter sphere (small volume), centred on the local activation maximum. P-values not in bold are uncorrected for multiple comparisons. For a given anatomical region, statistics are reported for up to three peak-level local maxima, ordered by statistical significance.