Neural network mapping of gelastic behavior in children with hypothalamus hamartoma

Abstract

Background: Hypothalamus hamartomas (HHs) are rare, congenital, tumor-like, and nonprogressive malformations resulting in drug-resistant epilepsy, mainly affecting children. Gelastic seizures (GS) are an early hallmark of epilepsy with HH. The aim of this study was to explore the disease progression and the underlying physiopathological mechanisms of pathological laughter in HH.

Methods: We obtained clinical information and metabolic images of 56 HH patients and utilized ictal semiology evaluation to stratify the specimens into GS-only, GS-plus, and no-GS subgroups and then applied contrasted trajectories inference (cTI) to calculate the pseudotime value and evaluate GS progression. Ordinal logistic regression was performed to identify neuroimaging-clinical predictors of GS, and then voxelwise lesion network-symptom mapping (LNSM) was applied to explore GS-associated brain regions.

Results: cTI inferred the specific metabolism trajectories of GS progression and revealed increased complexity from GS to other seizure types. This was further validated via actual disease duration (Pearson R = 0.532, P = 0.028). Male sex [odds ratio (OR) = 2.611, P = 0.013], low age at seizure onset (OR = 0.361, P = 0.005), high normalized HH metabolism (OR = - 1.971, P = 0.037) and severe seizure burden (OR = - 0.006, P = 0.032) were significant neuroimaging clinical predictors. LNSM revealed that the dysfunctional cortico-subcortico-cerebellar network of GS and the somatosensory cortex (S1) represented a negative correlation.

Conclusions: This study sheds light on the clinical characteristics and progression of GS in children with HH. We identified distinct subtypes of GS and demonstrated the involvement of specific brain regions at the cortical-subcortical-cerebellar level. These valuable results contribute to our understanding of the neural correlates of GS.

Keywords: Contrasted trajectories inference; Gelastic seizure; Hypothalamus hamartoma; Lesion network-symptom mapping; Neural basis.

Fig. 1

Flowchart of 149 consecutive participants who underwent multidisciplinary presurgical evaluation for epilepsy surgery. Eighty-three participants were excluded due to the exclusion criteria. Finally, 66 participants were enrolled in the study

Fig. 2

Relationship between disease progression and GS. a In the contrasted principal components (cPC) space, each subject is assigned to a PET metabolism trajectory. The subject’s position in the corresponding trajectory reflects the individual’s proximity to the healthy state (the background). An individual pseudotime index of PET metabolism is calculated and reflects the distance to two extremes (background or disease). GS-plus progresses from GS-only, but no-GS has an inverse direction. b Horizontal trend comparison of actual disease duration (gray bar) and pseudotime index (blue bar) across the three subgroups. Detailed statistical analysis is provided in Supplementary Table 2. c Correlation between actual disease duration and pseudotime index of each subgroup. Blue scatters indicate a significant correlation in the GS-only subgroup. GS gelastic seizures, PET positron emission tomography, *significant difference

Fig. 3

Predictive variables for GS. a Forest plot showing the effect of clinical variables on the prediction of GS. Diamonds and the horizontal line represent the individual variable effect (odds ratio) and 95% confidence interval, respectively. When the lines cross the solid vertical line, it indicates no effect and is marked with gray. b Bar plot and raincloud plots showing the intragroup comparison of clinical variables. GS gelastic seizures, HH hypothalamus hamartomas, onset age age of seizure onset, HHvol volume of hypothalamus hamartoma, nHHmeta normalized metabolism of HH, Endo endocrinologic dysfunction, NeruoCog neurocognitive sequelae, Intell intelligence decline, Memo memory decline, burden seizure burden

Fig. 4

Voxelwise lesion symptom mapping of HH associated with GS. a Overlap of individual HH segment maps superimposed on an averaged group-specific template, ranging from an HH volume in a single patient (blue) to overlap in all patients (red). b Group-level statistical parametric map displaying the metabolic connectivity difference between GS-only, GS-plus, and no-GS subgroups. The major significant clusters were the anterior cingulate cortex (ACC), medial prefrontal cortex (mPFC), somatosensory cortex (S1), caudate nucleus (CN), thalamus (Thal), and cerebellum. The values are expressed as ANOVA F values corrected for multiple comparisons using random field theory at P_FWE < 0.05 (cluster size > 500). All slices are presented in Supplementary Fig. 1. c Within-group comparison of the significant cluster. Detailed statistical analysis is provided in Supplementary Table 3. d In the GS-only subgroup, scatter plot analysis showed the metabolic correlation between other regions and the cerebellum. Only the thalamus was significantly positively correlated with the cerebellum (red dotted line), and S1 had a significantly negative correlation (blue dotted line). Detailed statistical analysis is provided in Supplementary Table 4. The matrix plot shows the intraregional correlation. GS gelastic seizures, HH hypothalamus hamartomas, ANOVA analysis of variance, CI confidence interval, *a significant difference