Electroencephalography and brain magnetic resonance imaging in asphyxia comparing cooled and non-cooled infants.

Abstract

The aim was to establish any differences in the predictive value of EEG and MRI for outcome in infants treated and not-treated with therapeutic hypothermia (HT) for perinatal asphyxia. We hypothesize that they are equally predictive and that combining both has the highest predictive value. We retrospectively compared data of infants with hypoxic-ischemic encephalopathy (HIE) who received HT (n=45) between September 2009 and December 2013 with those of infants with HIE born between January 2004 and August 2009, before HT was available (NT, n=37). All received conventional and/or amplitude-integrated EEG during the first days and early MRI (day 4-5). Associations of EEG, MRI and severe neurodevelopmental outcome (death or Bayley's -2SD below mean), were tested with a multivariable logistic regression analysis, corrected for HT. Forty-eight hours' EEG background pattern had a PPV of 92% and a NPV of 81% in HT, versus 100% and 58% in NT. MRI had a PPV of 71% and a NPV of 93% in HT, versus 82% and 75% in NT. The adjusted OR for adverse outcome was 0.013 (95% CI 0.002-0.154, p<0.001) for EEG background normalization within 48h and 32.19 (95% CI 4.84-214.25, p<0.001) for abnormal MRI. The predictive value of EEG and MRI is equal in cooled and non-cooled infants with HIE. Our data show a higher predictive value (death and severe outcome) for EEG compared to MRI. In HIE, persistent abnormal EEG background pattern until 48h, combined with abnormal early MRI is strongly predictive for poor neurodevelopment.