NEONATAL SOCIETY ABSTRACTS
Are newborn babies with substantial brain injury after hypoxia-ischaemia more likely to have multisystem involvement as reflected by routinely collected laboratory measures? A statistical approach
Presented at the Neonatal Society 2018 Spring Meeting (programme).
Rosser G1, Padhiar V, Openshaw-Lawrence N, Gupta N, Ekitzidou G, Ponnusamy V, Kappelou O, Evanson J, Shah DK
Barts and The London School of Medicine and Dentistry, Royal London Hospital, Southampton University Hospital, Homerton University Hospital, Ashford and St Peters Hospitals.
Background: In the event of perinatal hypoxia-ischemia (HI), it is believed that the supply to the brain, heart and adrenals is preserved at the expense of supply to other organ systems. Hypoxia-ischemic encephalopathy (HIE) and the associated brain injury in the newborn is an important cause of death and disability. Mild therapeutic hypothermia (TH) treatment is a recognised standard of care (NICE guidance 2010). Hypothesis: Babies with substantial brain injury on MRI after treatment with TH for HIE are more likely to have derangement of routinely collected laboratory measures, reflecting multisystem involvement, compared to those without brain injury.
Methods: With Research and Ethics approval (REC 13/LO/17380) and informed consent, blood samples were collected as part of routine care from 72 term newborn babies undergoing TH. A panel of 11 biomarkers was established reflecting haematopoietic, coagulation, renal, hepatic and inflammation systems. Cerebral MR images were classified into those predictive of favourable and unfavourable neurodevelopmental outcomes (1). Shrinkage logistic regression was applied to select biomarkers whose levels varied significantly between outcome groups. We assess the explanatory power of this model and test whether it remains applicable where blood products were required and in meconium aspiration and sepsis.
Results: Applying a t test to individual biomarker between the outcome groups revealed only one biomarker with a statistically significant distribution between the two groups; namely the platelet count. We construct a statistical model with modest sensitivity and good specificity based on peak and trough platelet counts. The model performs equally well in cases involving platelet transfusion and we observed no statistical support for treating these cases differently. In 14 cases of meconium aspiration, we identified significant differences in levels of haemoglobin, lymphocytes and C-reactive protein. Nevertheless, applying the platelet-based statistical model resulted in a good explanatory performance amongst this subset of cases.
Conclusion: Babies with brain injury predictive of unfavourable outcomes do not appear to have more multisystem involvement, as reflected by routinely collected laboratory values, compared to those without brain injury. However based on this small cohort, we have shown that platelet count is independently associated with brain injury outcome on MRI. A subset of babies who experienced meconium aspiration showed a significant difference in biomarker distribution, but is nonetheless supported by the platelet model. These findings need further study in a larger cohort.
Corresponding author: firstname.lastname@example.org
1. Rutherford et al. Lancet Neurology 2010
We are grateful to the Barts Charity, the families of the babies studied and the staff on the neonatal units.