Abstract
Objectives
Non-convulsive status epilepticus (NCSE) is misdiagnosed in >50% of cases in the emergency department. Computerized tomography perfusion (CTP) has been implemented in the hyper-acute setting to detect seizure-induced hyper-perfusion. However, the diagnostic value of CTP is limited by the lack of thresholds for hyper-perfusion and high inter-rater variability. This pilot case-control study aims at identifying the diagnostic value of reverse Tmax (rTmax) in differentiating NCSE vs acute ischemic stroke in the hyper-acute setting.
Methods
We enrolled patients with NCSE (Salzburg criteria-based diagnosis) and stroke cases 1:1 matched for clinical features and time of presentation. CTP standard maps (MTT-CBV-CBF) and rTmax maps were elaborated and rated by two experts in CTP blinded to the final diagnosis. Hyper-perfusion was adjudicated for standard CTP maps as an increase in CBF and a decrease in MTT, and for rTmax as the presence of a black area on 3, 2, and 1” threshold maps. Chronbach’s alpha was used for inter-rater agreement; receiver-operating curve (ROC) analysis was run to measure accuracy with AUC.
Results
Overall, 34 patients were included (17 NCSE, 17 stroke; onset-to-imaging 2 hours for both groups). People with NCSE were older and more frequently had a history of epilepsy. NCSE patients had hyper-perfusion on rTmax maps in 11/17 cases vs 0/17 in stroke. Intra and inter-rater reliability were higher for rTmax vs standard CTP maps (κ=1 vs κ=0.6). rTmax was 82% accurate in predicting NCSE vs stroke in the hyper-acute setting (0.67-0.97). Agreement between neuroimaging and EEG was limited at a hemispheric level for standard CTP maps, while rTMax had agreement with EEG largely reaching sub-lobar level.
Significance
rTmax mapping might represent a reliable tool to spot NCSE-induced hyper-perfusion with a threshold-based reproducible approach. Further studies are needed for validation and implementation in the differential diagnosis of focal neurological deficit in the hyper-acute setting.
JUL