Altered coupling of cerebral blood perfusion and neuronal activity in children with MRI-negative drug refractory epilepsy.

Background: Drug refractory focal epilepsy (DRFE) often gives rise to structural and functional damage in the brain accompanied by neuronal activity and cerebral vascular hemodynamics changes, which may result in neurovascular decoupling. However, neuroimaging evidence on neurovascular decoupling remains scarce. This study aimed to assess the manifestation of neurovascular coupling (NVC) in childhood DRFE using resting-state functional magnetic resonance imaging (rs-fMRI) and arterial spin labeling imaging (ASL). Methods: The rs-fMRI and ASL imaging data were obtained from 24 children with DRFE and 35 healthy controls (HC). Based on the collected data, degree centrality (DC) and cerebral blood flow (CBF) were calculated respectively. Across voxel CBF-DC correlations were calculated to evaluate the NVC within whole gray matter (GM), and NVC of brain region was assessed by the CBF/DC ratio, the whole-brain GM across voxel CBF-DC correlations and CBF/DC ratio of the 2 groups were compared. We further analyzed the relationships between the changed CBF, DC and CBF/DC ratio values and cognitive performance, clinical variables. Finally, we explored the value of machine learning methods for classifying DRFE and HC. Results: Compared with the HC group, the DRFE children showed higher across voxel CBF-DC correlations. Increased CBF/DC ratio located in the orbital part of the superior frontal gyrus, superior parietal lobule, and lower in the left inferior temporal gyrus and precuneus in the DRFE group. The brain regions of abnormal CBF, DC, and CBF/DC ratio were predominantly in regions in the default mode network, and the executive control network, and the abnormally CBF, DC values in some brain regions were significantly correlated to cognitive function. The classification model using CBF/DC ratio as features achieved the 72.8% accuracy, 0.764 area under the curve (AUC), 68.5% sensitivity and 87.5% specificity. The classification accuracy was higher than the model with other feature type (CBF: 67.8% accuracy; DC: 66.1% accuracy). Conclusions: The study reveals the cerebral blood perfusion, neuronal activity, global and regional NVC alteration in children with magnetic resonance imaging (MRI)-negative DRFE non-invasively, associated with lower cognitive performance. These findings indicate that NVC-based study can better integrate information of neuronal activity and cerebral hemodynamics, offering a new insight into the neuropathological mechanisms of DRFE and providing potential imaging biomarkers, and neurovascular decoupling may help clinical classification for childhood DRFE.