Mietzsch et al1 report a secondary analysis of the multicenter, double-blind, randomized, placebo-controlled HEAL (High-dose Erythropoietin for Asphyxia and Encephalopathy) trial,2 describing an association between the blood glucose level within the first 12 hours after birth among infants undergoing cooling for moderate and severe neonatal hypoxic-ischemic encephalopathy (HIE). Infants were stratified based on their most extreme glucose value as having hypoglycemia (glucose level <50 mg/dL), normoglycemia (50-200 mg/dL), or hyperglycemia (>200 mg/dL). The study found that either hypoglycemia or hyperglycemia was associated with higher rates of death, and early hypoglycemia was also associated with worse neurodevelopmental impairment (NDI) for those who survived to 22 to 36 months.
These findings corroborate previous studies3–8 and have biological plausibility. During anaerobic metabolism (ie, global hypoxia-ischemia) glucose stores are depleted rapidly, leading to hypoglycemia, which can be worse in infants with liver injury.9,10 In contrast, hyperglycemia in patients with HIE receiving cooling therapy is linked to stress responses along with decreased glucose uptake and utilization.7,11,12 In both cases, hypoglycemia and hyperglycemia is likely a biomarker of the degree of the brain injury. This is supported by our early translational studies using a mouse model of neonatal hypoxic-ischemic brain injury, in which there was an association between extremely low or high early glucose levels and worse hippocampal, thalamic, and cortical injury, astrocytic activation, and altered neurobehavioral outcomes at later time points.13
The thresholds for hypo- and hyperglycemia used in the study by Mietzsch et al are operationally defined to allow appropriate statistical stratification based on frequency of NDI. As noted by Mietzsch et al, severity of illness influences glycemic control, as such patients with extreme glucose values will also have a higher pretest probability of an adverse outcome. Thus, an appropriate physiologic context for these thresholds is needed to support the reader in the interpretation and clinical application of the presented results.
Early studies defined neonatal hypoglycemia as blood glucose levels <47 mg/dL.14,15 More recently the American Academy of Pediatrics has proposed blood glucose thresholds in infants ≤4 hours from birth as low as 25 mg/dL (asymptomatic) or 40 mg/dL (symptomatic).16 The Pediatric Endocrine Society recommends a higher threshold of 70 mg/dL for neonates receiving glucose-containing intravenous fluids.17 Importantly, these definitions are only applicable for otherwise healthy neonates, assuming the intact bioavailability of alternative fuels for the brain, such as lactate and ketone bodies. In the setting of HIE, the bioavailability of these alternative fuels may be limited, and thus the conservative threshold of 50 mg/dL used in this study is most likely physiologically appropriate. For hyperglycemia, the threshold is usually based on glucose excretion in urine, which in healthy adults occurs at 198 mg/dL.18 Because of the immature glomeruli glucose resorption in neonates, glucose excretion occurs at much lower blood concentrations and more so in the setting of acute kidney injury, as can occur with HIE.19 Thus, neonatal hyperglycemia is often defined as a level exceeding 144 to 150 mg/dL.20,21 This study defines hyperglycemia as ≥200 mg/dL, which is above the typical physiologic definition but easy to operationalize for a secondary data analysis. Thus, it is possible that the poor outcomes described in this study can occur at lower thresholds of hyperglycemia. Another limitation is that the glucose measures from this study were taken from 2 different sources: point-of-care measures and laboratory-based serum measures. Point-of-care measures can be up to 15% lower than serum measures,22 leading to potentially inaccurate case allocation in the study.
The study adjusted for a wide range of potential confounders. However, there could be residual confounding. For example, mothers of hypoglycemic infants had lower overall education levels, larger family size, and more pregnancy complications, all of which could be risk factors for NDI.23,24 Furthermore, severity of illness during NICU admission is not studied in detail. These factors might be even more predictive than the glucose level.25–28 For example, infants in the “hypoglycemic” and “hyperglycemic” groups were delivered more often by emergent cesarean section, had greater need for resuscitation maneuvers in the delivery room, and more often developed evidence of multiorgan failure and seizures, all of which also increase the risk for NDI.
Despite these limitations, Mietzsch et al provide further evidence linking NDI and death with extreme glucose levels within the first 12 hours of life in infants undergoing cooling for moderate-to-severe HIE. Because these associations do not prove causality, future interventional randomized clinical trials in patients undergoing cooling for HIE are needed. At this stage, it is still unclear whether tightening the control of blood glucose levels would improve outcomes in this population.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2022-060965.
FUNDING: Funded by the National Institutes of Health (KO8NS096115, R.C.-V.; R01NS126549, RC-V, F.J.N.; 1R21NS123814, R.C.-V., F.J.N.; 1R01 HD110091, R.C.-V., K.A., V.J.B., F.J.N.), The Thomas Wilson Foundation (R.C.-V.), inHealth Dean Accelerator Award (K.A.).
CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no conflicts of interest to disclose.
Comments