Abstracts

Rationale: Bottom-of-sulcus dysplasia (BOSD) is a clinically important but diagnostically challenging subtype of focal cortical dysplasia (FCD) type II. Visual and quantitative techniques are often used to search for subtle BOSD on neuroimaging in suspected patients. However, the frequency, significance and concordance of visible and measured imaging features on MRI and ¹⁸F-FDG-PET is unclear. _x000D_  _x000D_ We aimed to describe (1) the characteristic imaging features of BOSD on visual and quantitative assessment of MRI and PET and (2) the characteristic anatomical and functional network locations of BOSD.

Methods: A total of 81 children (48 male, median scan age=9.6 years) with focal epilepsy and MRI-positive BOSD (68% operated) who underwent 3T MRI and interictal ¹⁸F-FDG-PET were studied. Visual assessment of BOSDs on MRI was performed systematically by a neuroradiologist and neurologist, assessing cortical thickness, grey-white junction, cortical and subcortical signal, and the presence of a transmantle sign. Quantitative assessment of MRI and PET involved manual segmentation of BOSDs then measurement of cortical morphology, signal intensity and metabolism in Freesurfer software. Differences were calculated between (i) BOSD and a contralateral non-dysplastic mirror location within each patient; and (ii) BOSD and ipsilateral homotopic locations from a group of pseudo-controls. Finally, anatomical and functional network localization was determined using brain atlas templates.

Results: Characteristic visual features of BOSDs (Figure 1A) included cortical thickening (92%), grey-white junction blurring (100%), transmantle sign (65%), cortical T1 hyperintensity (57%), and subcortical FLAIR hyperintensity (37%). Quantitative imaging features (Figure 1B) that most distinguished BOSDs from non-dysplastic cortex included cortical hypometabolism, subcortical FLAIR hyperintensity, decreased T1 grey-white matter contrast, and cortical T1 hyperintensity. PET features distinguished BOSD more strongly than MRI features.   _x000D_  _x000D_ BOSDs were located in the frontal (62%), parietal (16%), insula (15%), temporal (6%) and occipital (1%) lobes. Fifty-nine percent were within or adjacent to eloquent cortex, most frequently motor cortex (58%). BOSD “hotspots” were in the middle frontal gyri and anterior insula cortex (Figure 2A). Fifty-two percent of BOSDs overlapped multiple functional networks (Figure 2B). The fronto-parietal control network was most frequently involved and the visual and limbic lobes and networks most infrequently.

Conclusions: Grey-white matter blurring and increased T1 cortical signal were frequently present on visual and quantitative analyses, but cortical thickening was more evident visually. On quantitative analysis, cortical hypometabolism was more discriminating than MRI features, suggesting that FCD detection tools may benefit from incorporating ¹⁸F-FDG-PET. Finally, BOSDs predominate in frontal, insula, and parietal cortical regions and frequently overlap multiple functional networks, the latter potentially contributing to diffuse electroclinical and cognitive manifestations of these tiny dysplastic lesions.

Funding: EML is supported by the Research Training Program and Clifford Family PhD Scholarships.