WHAT IS ALREADY KNOWN ON THIS TOPICHOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICYIntroduction

The recent Glucose in Well Babies (GLOW) Study of healthy breastfed newborns demonstrated glucose and ketone profiles over the first 5 days that can be interpreted as showing two phases of neonatal glucose concentrations.1 2 The first phase immediately after birth is a period of low glucose concentrations associated with hypoketonaemia that is consistent with relative hyperinsulinism.3 This period, which we have termed ‘transitional hyperinsulinism’,4 reflects persistence of the lower glucose threshold for suppression of pancreatic insulin secretion that is needed to support the rapid rates of growth during fetal life.5 The second phase is a period of ketonaemia in conjunction with low glucose concentrations, which likely reflects a fasting response to the as-yet insufficient supply of calories from breast milk, as well as the resolution of transitional hyperinsulinism.

The significance of any particular low glucose concentration is obviously different between these two phases. During the transitional hyperinsulinism phase, ketone concentrations cannot be increased to supply fuel to support the metabolic needs of the brain when glucose concentrations are low. In contrast, during the second phase, elevated ketone concentrations may help compensate for the low glucose concentrations by providing an alternative fuel for brain metabolism.6 Low glucose concentrations represent a potential threat to the brain in the first phase, whereas during the second phase, it is primarily a sign of undernutrition.

In order to estimate the duration of the first, transitional hyperinsulinism phase, and the second, ketonaemic phase in individual infants, we examined the combination of plasma glucose and ketone concentrations over the first 5 days after birth in 66 healthy, mainly breastfed newborn babies using data from the GLOW Study. The results from these analyses also provided an opportunity to evaluate the utility of routine monitoring of ketones in the interpretation of plasma glucose concentrations in newborn infants.

Methods

The GLOW Study was conducted in Hamilton, New Zealand between November 2015 and August 2017 and has been previously reported.1 2 7 It was a prospective, longitudinal observational cohort study in healthy term newborns, to determine the pattern of glucose, ketones (beta-hydroxybutyrate (BOHB)) and lactate concentrations and feeding over the first 5 days. All newborns completed the study in their own homes and were primarily cared for by their parents. Newborns were fed according to maternal choice, and most were breast fed. The GLOW Study is registered with the Australian and New Zealand Clinical Trials Registry (ref: ACTRN12615000986572). Heel-stick blood samples were taken for analysis four times in the first 24 hours and then about every 12 hours thereafter.

For these analyses, 66 of the 67 healthy newborns who participated in GLOW were included. One infant who had persistent mildly elevated glucose concentrations of 6–7 mmol/L potentially suggestive of a possible glucokinase inactivating mutation was excluded. Glucose concentrations were measured by glucose oxidase method on either an Epoc blood analyser (Siemens Healthineers) or blood gas analyser, if the baby was still in the hospital (Radiometer ABL800 FLEX). BOHB concentrations were measured on a StatStrip point-of-care analyser (StatStrip Meter, Nova Biomedical) using a BOHB-dehydrogenase method. Samples were not routinely taken pre-feeding and some time points had missing data for either glucose or BOHB or both. Researchers and families were blinded to the results and remained so until after completion of all data collection. BOHB concentrations that were below the limits of detection (<0.1 mmol/L) were assigned a value of 0.05 mmol/L for analyses.

To identify the duration of transitional hyperinsulinism, we used the sum of the plasma glucose plus BOHB concentrations. This sum represents the total insulin-dependent metabolic fuels available for brain metabolism, based on the ATP equivalents for oxidation of glucose, BOHB and acetoacetate (AcAc) (31, 21.5 and 20 mol of ATP per mol, respectively).8 To adjust for the minor contribution of AcAc to total ketones, the ATP equivalent for BOHB was assumed to be the same as for glucose, making the total insulin-dependent fuel for the brain approximately equal to the simple sum of glucose and BOHB molecular concentrations, using units of mmol/L . We ignored minor contributions from lactate and pyruvate or amino acids which are not strongly insulin dependent and remain relatively constant beyond the first few hours after birth.2 Based on the Endocrine Society and Pediatric Endocrine Society-suggested normal range for plasma glucose concentration in older children and adults in the non-fasted state of 3.9–5.5 mmol/L (70–100 mg/dL),9 10 we chose a glucose plus BOHB concentration of 4 mmol/L as the boundary between states of hyperinsulinism and normal for insulin-dependent brain fuels (ie, the lower end of the normal range for glucose in the non-fasted state when BOHB concentrations are negligible). For analysing the duration of the second, ketonaemic phase of low glucose concentrations, ketonaemia was defined as BOHB concentrations >0.5 mmol/L.

Statistical methods

The initial glucose measurement was taken as the lowest glucose concentration within the first 6 hours after birth. The age of resolution of transitional hyperinsulinism was estimated as the age at which the sum of glucose plus BOHB concentrations exceeded 4 mmol/L, using interpolation between the previous measurement and the first of the two sequential measurements where the sum exceeded this threshold. The age of resolution of ketonaemia was similarly determined using a threshold of 0.5 mmol/L, a concentration sufficiently high to be distinguishable from basal BOHB concentrations in the fed state (≤0.1 mmol/L). Means, SDs and centiles are reported where appropriate. Percentile curves were constructed using the skewness median coefficient of variation method11 and fitted with LMSchartmaker Light V.2.54 (Institute of Child Health 2011). Analyses and graphics were undertaken using Stata V.17, 2021 (Statacorp, College Station, Texas, USA).

Results

We included data from 66 healthy newborns from the GLOW Study (online supplemental table 1). There were 861 blood samples taken in total, with missing data in 14 (1.6%) samples. The mean (SD) initial glucose plus BOHB sum, obtained within 4 hours of birth, was 3.3 (0.7) mmol/L and was similar to the mean initial glucose concentrations reported in the GLOW Study (3.2 mmol/L (0.7)).1 The glucose plus BOHB sum rose over 36 hours to reach a mean of 4.3 (0.7) mmol/L in the period between 36 and 60 hours, compared with 3.4 (0.8), for glucose alone (figure 1). It then plateaued over days 4 and 5 to a mean of 4.9 (0.6) mmol/L, compared with 4.6 (0.6) for glucose alone. The fifth centile for the glucose plus BOHB sum on days 4 and 5 was 4 mmol/L.

Figure 1

Centiles of glucose plus BOHB plasma concentrations over the first 5 days of life. The grey area describes the 10th–90th centiles for plasma glucose concentrations only. BOHB, beta-hydroxybutyrate.

Most infants had initial values for the glucose plus BOHB sum well below the 4 mmol/L threshold, consistent with transitional hyperinsulinism, and this rose to reach concentrations above 4 mmol/L by 3 days of age, consistent with resolution of transitional hyperinsulinism (figure 2). The glucose plus BOHB sum then remained above the threshold of 4 mmol/L in most infants. However, the time to reach this threshold varied widely, from as early as 5 hours and as late as 96 hours (figure 3). The median duration was 40 hours, and 10 infants (15%) showed persistent hyperinsulinism beyond 60 hours. At 0–12 hours of age, 84% of the samples were in the hyperinsulinism region, compared with 20% at 48–72 hours, and only 6% at 96–120 hours of age (figure 4).

Figure 2

Six examples of plasma glucose, BOHB and their sum over the first 5 days of life. The orange dot indicates the time at which the glucose plus BOHB sum reached the threshold value of 4 mmol/L, indicating resolution of transitional hyperinsulinism (end of THI). BOHB, beta-hydroxybutyrate; THI, transitional hyperinsulinism.

Figure 3

Kaplan-Meier plot of the age of remission of transitional hyperinsulinism for 66 infants.

Figure 4

The relationships between BOHB and glucose plasma concentrations at three different postnatal ages. Double axes are presented to assist the interpretation of mg/dL units. The sloping lines indicate the threshold placement of 4 mmol/L for the ranges of BOHB and glucose—any point to the left of these lines indicates likely hyperinsulinism. BOHB, beta-hydroxybutyrate.

There was a weak negative relationship between the initial plasma glucose concentration measured after birth and the age at resolution of transitional hyperinsulinism (r2=0.21, p<0.001, figure 5).

Figure 5

The relationship between the first plasma glucose concentration and duration of transitional hyperinsulinism (HI). The solid line indicates the linear regression best fit; the grey zone indicates the 95% confidence zone. Regression equation is y=−13x+81, r2=0.21, p<0.001.

Following the initial period of transitional hyperinsulinism, 50 of the 66 (83%) infants had ketonaemia (BOHB concentrations >0.5 mmol/L). The median age at resolution of ketonaemia was 76 hours (range 41–>120 hours), but persisted beyond 120 hours in six infants (online supplemental figure 1). During this period of ketonaemia, 42 of 50 (84%) infants had plasma glucose concentrations below 4 mmol/L. The age at which ketonaemia resolved was inversely related to the percent weight gain over the first 5 days of age (r2=0.41, p<0.001; online supplemental figure 1). Eleven infants showed no ketonaemia at all, and five had ketonaemia prior to but not after completing resolution of relative hyperinsulinism.

Discussion

This analysis of the GLOW Study data on glucose and BOHB concentrations during the first 5 days in healthy breastfed newborn infants shows that the initial phase of neonatal low glucose concentrations, likely due to transitional hyperinsulinism, appears to generally persist into the second day, with some infants continuing into the third day. The weak negative relationship between initial glucose concentration and duration of the transitional hyperinsulinaemic phase is consistent with a gradually increasing glucose threshold for suppression of pancreatic insulin secretion after birth, with infants who had higher initial glucose concentrations presumably further advanced in this process and therefore taking shorter time for its resolution. This variability may need to be taken into consideration in early nursery discharge protocols and in screening infants at risk of persistent forms of hyperinsulinaemic hypoglycaemia, such as perinatal stress hyperinsulinism and genetic hyperinsulinism.

Analysis of the duration of the second, ketonaemic phase of neonatal hypoglycaemia, based on a BOHB concentration of >0.5 mmol/L, suggested that this phase persisted beyond 75 hours of age in half of these infants, but did not occur at all in 11 infants. Persistence of ketonaemia may be a reflection of a fasting response and may be an indication that maintenance caloric intake had not yet been achieved, consistent with our finding of an inverse relationship between weight change and duration of ketonaemia. Presumably, the duration of hyperketonaemia may be shorter in infants who are formula, rather than breast fed, but further studies are needed to clarify this.

This analysis was intended to develop a method for combining plasma glucose and BOHB concentrations that would differentiate between low glucose concentrations due to hyperinsulinism from low glucose concentrations due to fasting. We focused on the important role that glucose and ketones play as major fuels supporting brain metabolism, as ketones can normally substitute for glucose as a brain fuel during times of fasting hypoglycaemia.6 The total fuel available for the brain at any time can be represented by the ATP equivalence of the concentrations of circulating substrates used by the brain, chiefly glucose, ketones, lactate and pyruvate. The latter two are not regulated by insulin and since their concentrations remain relatively constant under most conditions, have not been considered here. The ketones consist of BOHB and AcAc; the latter is not usually measured since it is unstable and amounts to only about one-quarter of total ketones. We accounted for the unmeasured AcAc contribution by inflating the BOHB concentration by 30% which made the molar ATP equivalents of total ketones and glucose approximately equal, ~31 mol ATP/mol. Thus, the concentrations of glucose and BOHB in mmol/L could simply be added together to yield the glucose plus BOHB sum as a measure of total insulin-dependent brain fuels.

We selected a value for the glucose plus BOHB sum of 4 mmol/L as the boundary between low glucose concentrations due to relative hyperinsulinism (hypoketonaemic hypoglycaemia) versus low concentrations due to fasting (ketonaemic hypoglycaemia). This was based in part on the normal range for glucose in older children and adults in the non-fasted (non-ketonaemic) state of 3.9–5.5 mmol/L (70–100 mg/dL).9 10 Our findings suggest that this threshold appears to have some clinical validity as it was consistent with the lower limit for the sum of glucose plus BOHB in newborns 96–120 hours of age after transitional hyperinsulinism had resolved in this study. Nevertheless, the occurrence of ketonaemia before the resolution of the transitional hyperinsulinism phase in five infants suggests that this threshold may not apply to all infants, and further studies in larger cohorts are required.

In this study, the 4 mmol/L threshold for the glucose plus BOHB sum was fairly easily detected, even though no attempt was made to control the time of blood sampling related to feeding. This suggests that the sum of glucose plus BOHB might potentially be applied to examine the duration of postnatal hyperinsulinism in other groups of newborn infants, such as those with perinatal-stress associated hyperinsulinism after birth asphyxia, intrauterine growth restriction or maternal diabetes. The sum of glucose plus BOHB measurements can be done at the bedside using point-of-care analysers, which have been shown to have excellent accuracy.12 This could potentially provide a simple alternative to doing a formal prolonged fasting test that measures the BOHB response at a time of hypoglycaemia followed by measuring the glycaemic response to glucagon.10 The sum of glucose plus BOHB might also be useful to screen for persistent forms of hyperinsulinism, including for high-risk infants with perinatal stress hyperinsulinism and for infants with congenital hyperinsulinism due to genetic defects in pancreatic beta-cell insulin regulation.

Strengths of this study are the relatively large sample size of healthy newborns who were mostly breast fed and large dataset with frequent measurements of glucose and BOHB using consistent methodology. However, these results should be validated in other populations including at-risk and formula-fed neonates.

In conclusion, measurement of plasma BOHB concentrations may be useful for helping interpret plasma glucose concentrations during the newborn period and for appropriate management of infants with neonatal hypoglycaemia. The first, transitional hyperinsulinism phase of low glucose concentrations may persist into the second and third day of life in healthy term newborns, and the subsequent, ketonaemic phase may last until the fourth or fifth day after birth.

Data availability statement

Data are available upon reasonable request.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants. The GLOW Study was approved by the Northern A Health and Disability Ethics Committee (ref: 15NTA). Participants gave informed consent to participate in the study before taking part.