Introduction

Neonatal hypoglycaemia is the most common metabolic disorder of the newborn infant.1 It is associated with brain damage in infancy2 and neurodevelopmental problems in childhood.3 Neuroimaging studies of infants soon after they have experienced neonatal hypoglycaemia have generally reported injury to posterior regions of the brain,2,4 although more widespread changes have also been observed.5 A meta-analysis of follow-up studies conducted to investigate neurodevelopmental outcomes following neonatal hypoglycaemia has suggested that neonatal hypoglycaemia is associated with visual-motor impairment and executive dysfunction in early childhood and with neurodevelopmental impairment, low literacy and numeracy in mid-childhood.3 Because increasingly complex demands are placed on the child as they transition through later developmental stages, more adverse effects of neonatal hypoglycaemia on neurodevelopmental outcomes may become apparent at older ages. However, the relationship between adverse neurodevelopment and brain imaging in childhood after exposure to neonatal hypoglycaemia has yet to be explored.

In the recently completed Children with Hypoglycaemia and their Later Development (CHYLD) Mid-childhood Outcome Study, neonatal hypoglycaemia was not associated with educational attainment or other neurodevelopmental outcomes at 9–10 years,6 despite an association with executive function and visual-motor integration at 4.5 years.7 However, the cohort as a whole had poorer than average educational attainment. Furthermore, in mid-childhood neonatal hypoglycaemia was associated with a smaller caudate volume in a magnetic resonance imaging (MRI) sub-study.8 This anatomical association is somewhat surprising in the context of no associations with neurodevelopmental outcomes at the same age. Furthermore, the association persisted even after adjusting for potential confounding and in various sensitivity analyses, suggesting a robust association.

The caudate is a subcortical C-shaped structure that receives afferent projections from nearly all regions of the cortex and is one of the main input nuclei, along with the putamen, to the basal ganglia.9 It is important for both cognitive and emotional processing.9,10 Smaller caudate volumes have been associated with lower intelligence quotient in children born preterm11 and poorer cognitive control and verbal learning/recall in youth exposed to alcohol in utero.12 These domains are critical for learning, and further investigation of the neurodevelopmental correlates of caudate volume in the CHYLD cohort is warranted. Our aim was to investigate the relationship between caudate volume and executive and emotional-behavioural function and educational achievement at 9–10 years in the CHYLD cohort and whether any relations were influenced by exposure to neonatal hypoglycaemia.

Methods

Participants

Participants were drawn from the CHYLD Study. The CHYLD Study is a prospective cohort study of children who were born at or after 32 weeks’ gestation with one or more risk factors for neonatal hypoglycaemia: maternal diabetes, preterm birth (<37 weeks’ gestation), small (<10th centile or <2500 g) or large (>90th percentile or >4500 g) birthweight, or other illness. Recruitment took place between December 2006 and November 2010 at Waikato Hospital, Hamilton, New Zealand. Babies were screened and treated aiming to maintain blood glucose concentrations ≥2.6 mmol/L (47 mg/dL) and two-thirds had masked continuous glucose monitoring (CGM) (CGMS Gold; Medtronic MiniMed). Details of the cohort are available elsewhere.7,13,14,15

All eligible participants (had not previously withdrawn or died) were invited to undergo neurodevelopmental assessment at 9–10 years.6 A subset of children born ≥36 weeks’ gestation was randomly selected to be invited to consider MRI scanning, based on their experience of neonatal hypoglycaemia in four exposure groups: none, mild (1–2 hypoglycaemic events 2.0–2.6 mmol/L), severe (any hypoglycaemia events <2 mmol/L or ≥3 hypoglycaemic events) or clinically undetected hypoglycaemia (interstitial episodes only). Hypoglycaemic events were defined as the sum of nonconcurrent blood and interstitial (CGM-based) episodes more than 20 min apart.6 A hypoglycaemic episode was defined as at least one consecutive blood glucose concentration <2.6 mmol/L and an interstitial episode was defined as at least 10 min of interstitial glucose concentrations <2.6 mmol/L.

The Health and Disability Ethics Committee (16/NTB/208/AM02) provided ethics approval for the study. Parents or primary caregivers provided written informed consent and the child provided assent.

Caudate volume

MRI scans were acquired at Midland MRI, Anglesea Imaging Centre, Hamilton or the Centre for Advanced MRI (CAMRI), Auckland, with a 3.0 Tesla Siemens Skyra using a 20-channel paediatric head coil. T1-weighted images were acquired with a magnetization-prepared rapid gradient-echo sequence [slice thickness = 0.85 mm; repetition time = 2000 ms; echo time = 3.51 ms; field of view = 210 mm; voxel size = 0.8 × 0.8 × 0.8 mm; flip angle = 9°].

Volumes of the caudate were extracted using the recon-all pipeline in FreeSurfer 6.0 image analysis suite (http://surfer.nmr.mgh.harvard.edu/), as described elsewhere.16,17,18

Neurodevelopmental measures

At age 9–10 years, children underwent a neurodevelopmental assessment that included measures of executive function, academic achievement, and social-emotional behaviour.6

Executive function was assessed using a tablet-based battery (Cambridge Neuropsychological Test Automated Battery). Measures of attention (Attention Switching Task [AST]), planning/memory (One-touch Stockings of Cambridge, Spatial Working Memory [SWM], Paired Associate Learning [PAL]) and inhibition (Stop Signal Task [SST]) were included. The AST measures a child’s ability to switch their attention between the direction of an arrow and the location of an arrow in the presence of distracting information that is given during testing. The outcome measures were median switching and congruent response latency. The One-Touch Stocking of Cambridge task involves solving problems to position coloured balls in the correct sequence on the screen in two, three and four moves. The outcome measures were the number of problems solved on first choice and median latency to the first choice. The SWM task involves finding a blue token in each of four, six or eight boxes that appear on the screen by selection and process of elimination. Between errors (number of times the subject revisits a box in which a token has previously been found) and a measure of strategy were the outcome measures. The PAL task involves correctly locating patterns that were previously shown within boxes on-screen. The outcome measures were total errors (adjusted) and the first attempt memory score (number of correct box choices made on the first attempt). The SST involves responding to an arrow stimulus, by selecting the correct side of the screen depending on the direction of the arrow. The participant must withhold (inhibit) the response if an audio tone is present. The main outcome measure was the stop signal reaction time (length of time between the go stimulus and the stop stimulus (audio tone) at which the subject is able to successfully inhibit their response in 50% of the trials); two categorical outcomes were also assessed (failed stop reaction time less than go reaction time, and mean failed stop reaction time increases).

Academic achievement was assessed using the e-asTTle school achievement tests of Reading Comprehension and Mathematics in English or Māori.19 Social-emotional behaviour was assessed using the Strengths and Difficulties Questionnaire (SDQ), both parent and teacher reported.20 The SDQ assesses emotional and behavioural difficulties in children aged 4–17 years. The 25 items are divided into five subscales. We included the total difficulties score (range 0–40) which comprises four subscales (emotional symptoms, conduct problems, hyperactivity/inattention, peer problems) and the fifth subscale, prosocial behaviour, which represents the strengths of a child. Borderline and abnormal thresholds for the total difficulties score were defined as a parent-reported score of ≥14 or a teacher-reported score of ≥12.

Statistical analysis

All analyses were conducted on SAS version 9.4.

Associations between caudate volume and neurodevelopmental outcomes were assessed using general linear models conducted on the overall cohort, in the subgroups of those who did and did not experience neonatal hypoglycaemia, and including neonatal hypoglycaemia as an interaction term. The models were conducted unadjusted and then adjusted for gestational age at birth, sex, birthweight z-score and age at MRI scan; the adjusted models are presented in the text.

The main findings are reported for the total caudate volume. Additional analyses were conducted for the left and right hemisphere caudate volumes separately to explore the effect of laterality.

Results

In total, 248 participants were randomly selected to be invited to consider MRI, of whom 170 parents (69%) agreed to be approached and 111 (65% of those approached) consented to MRI. Following exclusion due to head motion, there were 101 participants with MRI and neurodevelopmental data available; 70 who had experienced neonatal hypoglycaemia (37 mild, 17 severe and 16 undetected) and 31 who had not. Most of the mothers of included children had post-secondary education (59%) and 40% had gestational diabetes mellitus (Table 1). There was a similar number of boys and girls, approximately one-third of the cohort lived in an area with high deprivation at birth and half of the cohort was of non-European ethnicity. There were no significant differences between children who had and had not experienced neonatal hypoglycaemia in any maternal or neonatal characteristics. Over 40% of children had low educational achievement, as assessed by the e-asTTle, and over 25% had parent- or teacher-reported emotional and behaviour difficulties (Table 2).

Table 1 Characteristics of the overall cohort and children who did and did not experience neonatal hypoglycaemia.
Full size table
Table 2 Associations between neurodevelopmental outcomes and caudate volume.
Full size table

In the overall cohort, caudate volume was not significantly associated with executive function measures (Table 2; for unadjusted models see Supplementary Table 1). However, there was evidence that the association between caudate volume and memory score on PAL (associative memory) may be influenced by neonatal hypoglycaemia as there was a positive association between caudate volume and memory score in children who had not experienced neonatal hypoglycaemia but not in those who had experienced hypoglycaemia (p for interaction 0.03; Fig. 1 and Table 2). No other interaction terms in the executive function tasks were significant.

Fig. 1: Interaction between neonatal hypoglycaemia and paired associate learning.
figure 1

Neonatal hypoglycaemia β –23, 95% CI –99 to 54, p = 0.56. No neonatal hypoglycaemia β 106, 95% CI 28 to 164, p = 0.01. Interaction p value = 0.03.

Full size image

Caudate volume was associated with emotional-behavioural regulation. The parent-reported Total Difficulties score and the proportion with borderline or abnormal score (≥14) were negatively associated with caudate volume, corresponding to a 60-mm3 decrease in caudate volume for every two-point increase in the Total Difficulties score. Parent-reported Prosocial score was positively associated with caudate volume (Table 2). There was no interaction between the caudate volume and neonatal hypoglycaemia for SDQ outcomes. However, in the subgroup analyses, among children who had not experienced neonatal hypoglycaemia SDQ, parent-reported total difficulties score and score ≥14 were negatively associated with caudate volume.

The analysis results were similar for left and right hemisphere caudate volume separately, although the associations between right caudate volume and SWM (between search errors) in the neonatal hypoglycaemia subgroup and with SDQ total difficulties score ≥14 in the no neonatal hypoglycaemia subgroup were not statistically significant (Supplementary Table 2).

Discussion

In a cohort of children born at risk of neonatal hypoglycaemia, caudate volume was associated with emotional-behavioural difficulties. Specifically, smaller caudate volume was associated with a greater number of parent-reported emotional and behavioural difficulties, and poorer prosocial behaviour. Furthermore, we found that although neonatal hypoglycaemia is associated with smaller caudate volume,8 it did not appear to influence the relationship between caudate volume and neurodevelopment, with the possible exception of associative memory.

The caudate is a key structure in cortical networks responsible for emotion regulation10,21 and integration of emotional information.22 We found that parent-reported social-emotional behaviour was related to smaller caudate volume, although teacher-reported behaviour was not. Discrepancies between these reports have been noted previously, possibly because teachers have a wider comparison to children of the same age and children may behave differently at home and at school.23,24 The negative relationship observed between total difficulties and caudate volume is consistent with previous literature linking altered structure of the caudate to attention deficit hyperactivity disorder (ADHD).25 It would be interesting to explore the relationship between ADHD and caudate volume in the current cohort. However, just five participants had parent-reported ADHD based on a single questionnaire item, which is too few for a reliable analysis.

There was just one significant interaction between the caudate volume and neonatal hypoglycaemia in the relationship with neurodevelopmental outcomes, indicating that smaller caudate volume may be associated with fewer correct choices on the first attempt at a visual associative memory task only in children who had not experienced neonatal hypoglycaemia. This interaction may be driven by the opposite direction of effect observed in children who had experienced neonatal hypoglycaemia. However, this association was not significant and any further discussion about the meaning of a negative relationship between caudate volume and visual memory in the neonatal hypoglycaemia group would be entirely speculative. The caudate nucleus has strong connections to higher-order association areas such as the dorsolateral prefrontal cortex9 and has previously been linked with executive function in children.11,12 Although we found associations between SWM and social-emotional behaviour difficulties and caudate volume in subgroups of children who had and had not experienced hypoglycaemia, the lack of an interaction effect for these outcomes necessitates caution when interpreting these findings. We performed 42 tests for the subgroup analyses and could therefore reasonably expect eight false-positive results. It is therefore possible that the hypoglycaemia subgroup findings are due to Type I error.

We have previously reported that neonatal hypoglycaemia is not associated with altered educational or other neurodevelopmental outcomes in mid-childhood,6 and this study suggests that hypoglycaemia may not alter the relationship between caudate volume and neurodevelopmental outcomes. Together, these suggest that the observed association between neonatal hypoglycaemia and reduced caudate volume8 may not be causal, or that if it is, it has little observable clinical relevance. However, all children in this cohort were tested and treated with the intention of maintaining blood glucose concentrations at or above 2.6 mmol/L, which may have minimized any potential effects on later outcomes. Randomized trials are needed to determine whether neonatal hypoglycaemia actually causes altered brain volumes and neurodevelopmental outcomes and at what glucose thresholds.

Strengths of the study include that children were assessed on a wide range of neurodevelopmental outcomes that included academic performance, executive functions and socio-emotional behaviour. In addition, there was a high percentage of low educational achievement and emotion and behaviour difficulties, which may have increased the power to detect associations with caudate volume. Low educational attainment in this cohort has been discussed comprehensively in our previous report6—briefly, we postulate that the risk factors for neonatal hypoglycaemia and their social determinants may themselves. have a negative effect on neurodevelopment. Examining the left and right hemisphere caudate volumes separately also did not change the findings. This is one of the few studies to conduct brain MRI in the mid-childhood following neonatal hypoglycaemia. However, the sample size was modest, particularly in the group that had not experienced neonatal hypoglycaemia, and the confidence intervals around many of the beta coefficients were relatively wide, suggesting that these estimates of relationships between caudate volume and functional outcomes were imprecise. Furthermore, we focussed on caudate volume based on the most robust finding in our previous report of the MRI findings at 9–10 years related to hypoglycaemia in this cohort;8 however, children who had experienced neonatal hypoglycaemia also had smaller thalamus and subcortical grey matter volumes as a percentage of total brain volume compared to children who had not. As these findings were not significant for the absolute volumes (mm3), it is unlikely they are contributing to the association between caudate volume and emotional-behavioural problems. In addition, none of the included children underwent neonatal MRI; therefore, we have no information about any acute changes after neonatal hypoglycaemia. Two children had evidence of other changes on MRI scan (one a well-defined splenium lesion of the corpus callosum and the other posterior white matter injury, particularly in the parieto-occipital regions). However, as there were just two children with incidental findings, they are unlikely to have substantially influenced the results. Future studies with larger sample sizes would be beneficial to determine if differences in brain structure associated with neonatal hypoglycaemia have an effect on neurodevelopment.

In summary, smaller caudate volume is associated with increased emotional-behavioural difficulties in the mid-childhood. However, exposure to neonatal hypoglycaemia was not found to significantly influence this relationship.