Plasma Biomarkers of Brain Injury as Diagnostic Tools and Outcome Predictors After Extracorporeal Membrane Oxygenation*

Objective: To determine if elevations in plasma brain injury biomarkers are associated with outcome at hospital discharge in children who require extracorporeal membrane oxygenation. Design: Prospective observational study. Setting: Single tertiary-care academic center. Participants: Eighty children who underwent extracorporeal membrane oxygenation between June 2010 and December 2013. Interventions: None. Measurements and Main Results: We measured six brain injury biomarkers (glial fibrillary acidic protein, monocyte chemoattractant protein 1/chemokine (C-C motif) ligand 2, neuron-specific enolase, S100b, intercellular adhesion molecule-5, and brain-derived neurotrophic factor) daily during extracorporeal membrane oxygenation, using an electrochemiluminescent multiplex assay. We recorded clinical, neuroimaging, and extracorporeal membrane oxygenation course data. We analyzed the association of biomarker concentrations with favorable versus unfavorable outcome at hospital discharge. Favorable outcome was defined as Pediatric Cerebral Performance Category 1, 2, or no change from baseline. Patients had a median age of 3 days (interquartile range, 1 d–10 mo), and 56% were male. Thirty-three of 80 (41%) had unfavorable outcome, and 22 of 70 (31%) had abnormal neuroimaging findings during or after extracorporeal membrane oxygenation. Peak concentrations were significantly higher in patients with unfavorable outcome than in those with favorable outcome for glial fibrillary acidic protein (p = 0.002), monocyte chemoattractant protein 1/chemokine (C-C motif) ligand 2 (p = 0.030), neuron-specific enolase (p = 0.006), and S100b (p = 0.015) and in patients with versus without abnormal neuroimaging findings for glial fibrillary acidic protein (p = 0.001) and intercellular adhesion molecule-5 (p = 0.001). The area under the receiver operator characteristic curve for unfavorable outcome was 0.73 for a noncollinear biomarker combination. After removing collinear biomarkers, the adjusted odds ratios for unfavorable outcome were 2.89 (95% CI, 1.09–7.73) for neuron-specific enolase, using a cutoff of 62.0 ng/mL, and 2.15 (95% CI, 1.06–4.38) for glial fibrillary acidic protein, using a cutoff of 0.46 ng/mL. Conclusions: Elevated plasma brain injury biomarker concentrations during the extracorporeal membrane oxygenation course are associated with unfavorable outcome and/or the presence of neuroimaging abnormalities. Combinations of brain-specific proteins increase the sensitivity and specificity for outcome prediction.