Quantitative MRI Measures and Cognitive Function in People With Drug-Resistant Juvenile Myoclonic Epilepsy

Abstract

Background and objectives: Neuroimaging studies have so far identified structural changes in individuals with juvenile myoclonic epilepsy (JME) when compared with controls. However, the underlying mechanisms of drug-resistant JME remain unknown. In this study, we aimed at characterizing the structural underpinnings of drug-resistant JME using MRI-derived cortical morphologic markers.

Methods: In this prospective cross-sectional 2-center study, T1-weighted MRI and neuropsychological measures of verbal memory and executive function were obtained in individuals with drug-resistant and drug-responsive JME recruited from epilepsy outpatient clinics and healthy controls. We performed vertexwise measurements of cortical thickness, surface area, and local gyrification index (LGI). Vertexwise group comparisons were corrected for multiple comparisons at a familywise error (FWE) of 0.05. The neuropsychological profile of disease subgroups was analyzed through principal component analysis.

Results: We studied 42 individuals with drug-resistant JME (mean age 29 ± 11 years, 50% female), 37 with drug-responsive JME (mean age 34 ± 10, years, 59% female), and 71 healthy controls (mean age 21 ± 9 years, 61% female). Surface area was increased in participants with drug-resistant JME in the left temporal lobe (Cohen d = 0.82 [-0.52 to -1.12], p_FWE < 0.05) when compared with the drug-responsive group. Although no cortical thickness changes were observed between disease subgroups, drug-resistant and drug-sensitive participants showed discrete cortical thinning against controls (Cohen d = -0.42 [-0.83 to -0.01], p_FWE < 0.05; Cohen d = -0.57 [-1.03 to -0.11], p_FWE < 0.05, respectively). LGI was increased in the left temporal and occipital lobes in drug-resistant participants (Cohen d = 0.60 [0.34-0.86], p_FWE < 0.05) when contrasting against drug-sensitive participants, but not controls. The composite executive function score was reduced in drug-resistant individuals compared with controls and drug-sensitive individuals (-1.74 [-2.58 to -0.90], p < 0.001 and -1.29 [-2.25 to -0.33], p < 0.01, respectively). Significant correlations were observed between executive function impairment and increased surface area in the precuneus and medial prefrontal regions (r = -0.79, p_FWE < 0.05) in participants with drug-resistant JME.

Discussion: We identified a developmental phenotype in individuals with drug-resistant JME characterized by changes in cortical surface area and folding complexity, the extent of which correlates with executive dysfunction. No association was observed between cortical thickness and disease severity. Our findings support a neurodevelopmental basis for drug resistance and cognitive impairment in JME.

Figure 1. Vertexwise Group Comparisons of Surface Features

Mass univariate analysis showing group comparisons of cortical thickness, surface area, and LGI (A) between individuals with JME and healthy controls (syndrome-related effects) and (B) between drug-resistant and drug-responsive individuals with JME (effects related to disease severity). Comparisons against the healthy control group are shown in (C) and (D). All surface features were corrected for age and sex while surface area was further corrected for total white matter volume. Clusters are color-coded according to the corresponding effect size estimates as reported by Cohen d (color bar). Clusters that survived multiple comparison correction using random field theory at p_FWE < 0.05 are outlined in black. FWE = familywise error; JME = juvenile myoclonic epilepsy; LGI = local gyrification index.

Figure 2. Performance in Neuropsychological Domains Across Groups

Raincloud plots show the distribution of composite cognitive construct scores representing (A) executive function and (B) verbal memory performance across healthy controls and individuals with drug-resistant and drug-responsive JME. In A and B, asterisks refer to p values for Bonferroni-corrected, age-adjusted, and sex-adjusted post hoc t tests calculated on the residuals from the analysis of covariance. Statistical details are provided in Table 3. **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 3. Correlation Between Neuropsychological Domains and Cortical Surface Features

Vertexwise partial Pearson correlation plots between the composite cognitive constructs for executive function (A), verbal memory (B), and the cortical surface features (cortical thickness, surface area, and LGI). Partial correlation analysis used age and sex as covariates for healthy controls and disease duration, sex, and total ASM load for the drug-sensitive and drug-resistant groups. Significant clusters that survived multiple comparison corrections using random field theory at p_FWE < 0.05 are outlined in black. Clusters are color-coded according to the partial Pearson correlation coefficient (r) (color bar). ASM = antiseizure medication; FWE = familywise error; JME = juvenile myoclonic epilepsy; LGI = local gyrification index.