Combined MRI Ischemic Burden and Early EEG Background Improve Outcome Prediction After Pediatric Cardiac Arrest
Ashley Bach1, Matthew Kirschen1, France Fung1, Nicholas Abend1, Antara Mondal1, Steve Ampah1, Jimmy Huh1, Shih-shan Lang1, Ian Yuan1, Kathryn Graham1, Jeffrey Berman1, Arastoo Vossough1, Alexis Topjian1
1Children's Hospital of Philadelphia
Objective:
To assess the association of EEG background with MRI ischemic burden after pediatric cardiac arrest (CA) and determine whether addition of MRI ischemic burden to EEG and clinical variables improves outcome prediction.
Background:

The association of early EEG background with MRI ischemic burden and their combined association with pediatric CA outcome is unclear.

Design/Methods:

Retrospective single-center cohort study of pediatric CA from 2005-2019. Initial EEG background within 24 hours of CA was classified as normal, slow/disorganized, discontinuous/burst-suppression, or attenuated/featureless. MRI within 7 days of CA was assessed for ischemic burden, defined as brain tissue percentage with apparent diffusion coefficient (ADC)<650x10-6mm2/s. Outcomes were unfavorable neurologic status (Pediatric Cerebral Performance Category change ≥1 from baseline resulting in hospital discharge score ≥3) and death. Kruskal-Wallis test evaluated association of EEG with MRI. Logistic regression and likelihood ratio test assessed outcome prediction. Area under the receiver operating curve (AUROC) evaluated predictive accuracy.

Results:

We evaluated 90 children with median age 1.6(IQR:0.6-5.8) years. EEG background was normal in 16(18%), slow/disorganized in 42(47%), discontinuous/burst-suppression in 12(13%), and attenuated/featureless in 20(22%). Median percent of ischemic brain tissue was 5%(IQR:1-18%). Unfavorable neurologic status occurred in 58(64%), of whom 28(31%) died. Worse EEG background was associated with greater brain ischemia (χ2=27.9;p<0.001). A model of EEG background, number of epinephrine doses, post-arrest lactate and witnessed status yielded an AUROC of 0.90 for unfavorable neurologic status and 0.92 for death. Addition of MRI to this model significantly increased the AUROC from 0.90 to 0.92 for unfavorable neurologic status (χ2=4.8;p=0.03) and from 0.92 to 0.97 for death (χ2=19.1;p<0.001). 

Conclusions:
Early EEG background was associated with increased MRI ischemic burden within 7 days of cardiac arrest. Addition of MRI ADC to clinical factors and EEG background improved outcome prediction with high predictive accuracy. MRI and EEG, though associated, independently contributed to a multimodal model of outcome.
10.1212/WNL.0000000000201866