Relationship between electroencephalography and magnetic resonance imaging findings after hypoxic-ischemic encephalopathy at term

Abstract

Hypoxic-ischemic encephalopathy (HIE) is a major cause of neonatal morbidity and mortality. Electroencephalography (EEG) and brain magnetic resonance imaging (MRI) are frequently performed in these infants, but the prognostic value of the combined use of EEG and MRI needs additional exploration. The purpose of this study was to investigate, in neonates with HIE, the role of early EEG and conventional MRI in the prediction of infants at risk for persistent encephalopathy at 18 months of age. Thirty-four term infants with HIE were enrolled in this prospective study. EEG was recorded within the first 72 hours after birth and a brain MRI scan was done between 1 and 4 weeks of age. Denver Developmental Screening Test II was performed at 6, 12, and 18 months of age. Three infants (9%) had mild HIE, 21 infants (62%) had moderate HIE, and 10 infants (29%) had severe HIE. The EEG background was normal, moderately, severely, and extremely discontinuous in eight (24%), three (9%), sixteen (47%), and seven (20%) neonates, respectively. EEG background activities correlated significantly with HIE severity (p = 0.0001). MRI findings significantly correlated with EEG background (p = 0.001). Normal MRI scans and minimal basal ganglia lesions were always associated with normal EEG background. Patients with severe basal ganglia and thalamic lesions in MRI (n = 2) had extreme discontinuous EEG background. For the prediction of poor outcomes, abnormal EEG background activity had a sensitivity (Sn) = 100%, a specificity (Sp) = 100%, positive predictive value (PPV) = 100%, and negative predictive value (NPV) = 100%, whereas values of abnormal MRI scans were Sn of 100%, Sp = 43%, PPV = 82%, and NPV=100%. EEG background activity is the best element to predict abnormal outcomes. Severe basal ganglia and thalamic injuries on MRI scans are associated with poor outcomes. Otherwise, MRI does not contribute to the prediction of outcomes at 18 months of age.