Rationale:
Epilepsy is a common co-morbidity of neurodegenerative disorders. Both epilepsy and dementia have been independently associated with higher white matter hyperintensities (WMH) on MRI. Nonetheless, the pattern of WMH distribution and burden remains understudied in dementia and co-morbid epilepsy. Our study aims to fill this knowledge gap.Methods:
This multicenter study draws on data from 42 Alzheimer’s Disease Centers across the US between 2005 and 2024. Participants were divided into three groups: (1) dementia with epilepsy (DemEpi); (2) Dementia without epilepsy (DemNoEpi); and (3) healthy controls (HC). All participants with DemEpi who had an MRI scan available were included. We used a fixed-ratio, optimal propensity score matching to match group 1 participants to groups 2 and 3. Matching was based on age, sex, and dementia severity for group 2, and on age and sex for group 3.
We selected high-resolution T1-weighted and FLAIR MRI scans for each subject to ensure reliable downstream analysis. Given the multicenter nature of the dataset, FLAIR images were first harmonized before the extraction of WMHs. WMH segmentation was then performed using the widely adopted Lesion Segmentation Tool (LST) with the Lesion Growth Algorithm (LGA).
In addition to whole-brain WMH analysis, we examined regional distributions across individual lobes. To do this, T1-weighted scans were co-registered to their corresponding FLAIR images using FreeSurfer’s mri_coreg tool. Brain segmentation was subsequently conducted with FreeSurfer, and WMH regions were mapped to specific anatomical labels based on these results. During analysis, we normalize the WMH volumes with respect to the total intracranial volume for each subject and log-transform the results.
Our analysis focused on WMH in the frontal, cingulate, occipital, temporal, parietal, insular, and precuneus lobes, evaluated separately for the left hemisphere, right hemisphere, and bilaterally combined. In addition, whole-brain WMH was also assessed.
Results:
Overall, 26 DemEpi, 128 DemNoEpi, and 129 HC met the inclusion criteria. We first applied the Kruskal-Wallis test to evaluate group differences in WMH among the three groups. This analysis revealed significant differences in all regions except the left insular lobe. To further explore these findings, we conducted post hoc pairwise comparisons using the Mann-Whitney U test.
When directly comparing the DemEpi and DemNoEpi groups, significantly higher WMH volumes were found in whole-brain, the combined cingulate lobes, and the right precuneus in the DemEpi group. Comparisons involving the HC group consistently showed significant differences across all regions, with higher WMH volume observed in both DemEpi and DemNoEpi compared to HC.
Conclusions:
Our results demonstrated that the presence of comorbid epilepsy in dementia is associated with higher overall whole brain, cingulate cortex, and right precuneus WMH volumes. These findings suggest that distinct WMH volume global and regional burden may be observed in comorbid epilepsy and may serve as a neuroimaging correlate of epilepsy in dementia.
Funding:
Dr. Zawar received support from Alzheimer’s Association and American Epilepsy Society for this work.