Plasma volume fraction in drug-resistant epilepsy: A DCE-MRI marker of microvascular pathology

Objectives Microvascular remodeling and blood–brain barrier dysfunction (BBBD) are increasingly recognized as contributors to epilepsy. However, commonly used vascular imaging markers are often state-dependent and lack spatial specificity. We aimed to (1) validate plasma volume fraction (vₚ) derived from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) as a sensitive marker of microvascular changes; (2) characterize the vₚ alterations in patients with drug-resistant epilepsy (DRE); and (3) assess the spatial relationship between vₚ abnormalities and BBBD. Methods We analyzed DCE-MRI images from 49 people with epilepsy (PWE) and 68 healthy controls across two sites. vₚ and BBB permeability were quantified using BBBdetect software . Voxel-wise vₚ was estimated using the extended Tofts model, and BBBD was quantified using slope-based permeability mapping. Both measures were summarized using modified z-scores, with suprathreshold abnormality defined as modified z-score > 2. We performed region-wise and lobe-wise analyses restricted to gray matter and trained supervised classifiers to distinguish PWE from controls using regional z-vₚ and z-BBBD features. Results Compared with controls, PWE showed increased voxel-wise and region-wise vₚ abnormality burden, with a non-uniform spatial pattern that includes prominent fronto-temporal elevations and frequent involvement of limbic regions. BBBD was common and spatially diffuse. Restricting analysis to regions with co-occurring suprathreshold vₚ and BBBD reduced spatial diffuseness relative to BBBD alone. Multivariate classification achieved encouraging test performance using vascular and barrier features (balanced accuracy = 0.81), with feature importance suggesting complementary contributions from vₚ and BBBD. Conclusion vₚ is a sensitive DCE-MRI-derived marker of microvascular abnormalities in DRE. Integrating vₚ with BBBD enhances the spatial specificity of abnormality patterns and shows encouraging concordance with the clinically suspected epileptogenic territories, warranting prospective validation against clinical reference standards.