Survival of very preterm infants has improved over the last few decades. Motor outcome and rates of cerebral palsy have improved since 2000. However, mental outcomes remain unchanged.1^–3 At school age, preterm children are smaller, lighter and have smaller heads than their peers.4 5 Among those attending mainstream schools, cognitive and motor outcomes on formal testing compared less favourably with their peers, with higher occurrence of developmental difficulties in attention, literacy, numeracy and motor coordination.6 7 Head circumference appeared to be predictive of educational and motor outcomes.4 8^–10

Even small changes to the post-natal environment can have a considerable effect on outcome. Lucas et al found that preterm formula is superior to standard term formula in promoting growth and subsequent cognitive development.11 12 However, in very preterm infants, illness and feed intolerance in the early post-natal period lead to difficulties in providing adequate nutrition. Despite advances in the provision of parenteral and enteral nutrition, malnutrition in the first few weeks of life remains commonplace.13

Hyperalimentation is defined as a method of providing parenteral or enteral nutrition that contains macronutrients at amounts above current recommendation. In an earlier paper, the feasibility of providing additional protein and energy to very preterm infants in their first few weeks of life through hyperalimentation was evaluated.14 We described the effects of nutritional intake on head growth by measuring occipitofrontal circumference (OFC) at 36 weeks' postmenstrual age (PMA). The aims of this paper are to describe the relationships between developmental outcome and energy and protein intakes in the first 4 weeks of life, post-natal head growth and quantitative magnetic resonance imaging (MRI) findings at 40 weeks' PMA.

METHODS

A randomised controlled study of hyperalimentation was conducted in a tertiary neonatal intensive care unit, which serves a population of approximately 2.5 million in northwest England and north Wales. The intervention was described in detail in our earlier paper.14 In summary, 142 infants born before 29 weeks' gestation were randomly assigned to receive enhanced or standard parenteral and enteral nutrition from the first week of life to 34 weeks' PMA. Actual daily intake was recorded for each infant. Energy and protein deficits in the first 28 days of life were calculated by subtracting actual daily intakes from recommended intakes of 120 kcal/kg/day and 3 g/kg/day of energy and protein, respectively.15 16 OFC, lower leg length (LLL) and weight were measured at 36 weeks' PMA.17

Quantitative MRI was performed at 40 weeks' PMA using Philips Gyroscan T5-NT 0.5 Tesla scanner (Philips Medical Systems, Best, the Netherlands). The scans were performed in natural sleep with no sedation (feed and wrap). Total brain volume (TBV), cortical brain volume (CBV) and T2 relaxation times were measured. T2 relaxation times are time constants governing the decay of transverse magnetisation after the application of radio-frequency pulse.18 Tissue-water content and the interaction between water protons and tissue molecules determine the relaxation times. Rich water content generates a high T2 value. Change in the water-molecules ratio occurs on arrival of myelin precursors. Hence brain maturation is associated with shortening of T2 values.19 20 In this study, the T2 values of the fronto-parietal white matter were measured. Only inpatients or those who lived within an hour’s journey from Liverpool underwent MRI scanning. The infants had to be breathing in air or receiving only oxygen supplementation via low flow nasal cannulae, free of any active infections or methicillin-resistant Staphylococcus aureus (MRSA) colonisation and stable for transfer between hospitals. Scan sequences used were similar to those in a previous study.21 22 The second author (LA) who performed all the scan analyses was unaware of the infants’ group assignment. Due to events beyond our control, the magnetic resonance (MR) scanner was replaced with Phillips Achieva Nova 1.5 Tesla (Best, The Netherlands) 21 months after commencement of this phase of the study. Scan setting and sequences used remained the same.

The infants were invited for follow-up examinations at 3 and 9 months post-term. Infants were seen either in their own homes or in the outpatient department by parental choice. Bayley Scales of Infant Development II (BSID II) was performed by the first author who was not blinded to the infants’ feed assignment.23 Total raw scores in the mental (MDI) and motor (PDI) scales were converted to index scores. The index scores for each scale range from 50 to 150, with a mean of 100 (3.3 standard deviations on each side of the mean). Infants with severe delay who were not able to complete the test or with raw scores that lie outside the lower index range were given index scores of 49.

Data were collected according to defined criteria.14 “EBM only in first 4 weeks” is defined as infants who received expressed breast milk (EBM) exclusively in their enteral feeds for the first 28 days. “Mothers with higher education” is defined as education attainment beyond General Certificate of Secondary Education (GCSE) or equivalent (Year 11 in the British education system). Serial cranial ultrasound scans were performed using GE Ultrasound, Vivid 7 Pro (GE Medical Systems, Hovik, Norway), by suitably trained radiographers in the first 8 weeks of life, as part of their routine care.

SAMPLE SIZE AND STATISTICAL ANALYSIS

Sample size was calculated based on the primary outcome measure of this study, as described in our previous paper.14 Data analysis was performed by intention to treat using statistical software SPSS 12. Parametric and non-parametric data were analysed using student t test and Mann–Whitney U tests, respectively. Bivariate correlation and regression analysis were used where appropriate to determine factors associated with developmental outcome. Post hoc analyses of MRI and developmental outcomes according to sex were also performed.

RESULTS

One hundred and forty-two infants were recruited and randomised to hyperalimented or standard feeding regimen within a week of birth following written informed consent from parents (fig 1). One hundred and eleven infants survived to the time of discharge from the neonatal unit, 2 infants died at 3 months post-term (1 after the 3 months assessment). Six lived more than an hour’s drive from Liverpool, 19 were clinically unstable, ventilated or receiving significant respiratory support, 2 were colonised with MRSA, and parents of 18 infants declined MRI scan. Sixty-six infants underwent MRI scan. One baby was too restless while in the scanner, and the procedure was abandoned. The scans were performed at a mean age of 40.3 weeks' PMA (range 37.4–42.9 weeks' PMA) for infants in the intervention group (IG) and 40.7 weeks' PMA (range 39.6–42.9 weeks' PMA) for those in the control group (CG). Eighty-six infants attended follow-up assessments, 81 at 3 months and 71 at 9 months post-term. The median age of follow-up was 3.3 months (range 2.9–6.1) and 9.4 months (range 8.9–12.4), respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Schematic diagram showing outcomes of all eligible babies and number of babies who survived to 9 months post-term.

The baseline characteristics, nutrition and growth outcomes for the 86 infants assessed in the first year are shown in table 1. Among the 86 infants assessed, 15 infants in the IG compared to 11 in the CG had birth weights below 800 g, while 20 in the IG compared to 14 in the CG were born before 27 weeks' gestation. However, this was not statistically significant. Fifteen survivors in the IG compared to 11 in the CG had Clinical Risk Index for Babies (CRIB) scores of above 10. Fifteen infants in the IG and 14 in the CG did not fully receive their assigned feed. The reasons include transfer to another hospital, assigned feed unavailable (problems with stock) and parental choice (in one infant). There were no significant differences in baseline characteristics and growth outcomes at 36 weeks' PMA between survivors assessed in the first year and those not assessed. Thirty-two per cent of survivors seen in the first year compared to 4% of those not seen (p = 0.003) received EBM exclusively as enteral feed in the first 28 days.

MRI outcomes of the 65 infants who underwent MRI scan at 40 weeks' PMA are shown in table 2. Significant abnormalities seen on MRI include: periventricular leucomalacia (PVL) (1 in the IG and 3 in the CG), diffuse hyperintensity of the cerebral white matter on T1- or T2-weighted images (1 in the IG, 3 in the CG), ventricular dilatation (1 in the IG and 3 in the CG), porencephalic cyst (1 in the CG) and cortical ischaemia (1 in the CG). Six infants with normal cranial ultrasound scans in the first 8 weeks were found to have significant changes on MRI (one with bilateral periventricular cysts). T2 values, TBV and CBV were not measured in 2, 13 and 16 infants, respectively, due to poor scan quality. Seven infants in the IG and nine infants in the CG were scanned using the new 1.5 Tesla MRI scanner. Fity-five infants (24 in the IG and 31 in the CG) who underwent MRI scanning were seen again in the first year for developmental assessment.

There were no significant differences in the occurrence of gross abnormalities, TBV, CBV and T2 relaxation times between the two groups. Subgroup analysis by sex did not reveal significant differences. However when data from both groups were pooled, energy intake and energy deficit in the first 28 days of life correlated significantly with TBV (R = 0.35, R² = 0.12, p = 0.026; R = −0.35, R² = 0.12, p = 0.024, respectively). CBV, TBV and T2 values correlated significantly with standard deviation scores (SDS) of OFC at 36 weeks' PMA (R = 0.38, R = 0.6, and R = −0.38, respectively). TBV and CBV also correlated with weight (R = 0.46 and R = 0.32, respectively) and length SDS at 36 weeks' PMA (R = 0.56 and R = 0.41, respectively) but not OFC or weight at birth. There is a linear relationship between T2 relaxation time and TBV (R = −0.33) but not CBV.

Among the 86 infants assessed at 3 and 9 months post-term, 4 infants were known to have cystic PVL on either ultrasound or MRI scan (3 in the IG and 1 in the CG) and 1 had post-haemorrhagic hydrocephalus that required shunt insertion (CG). One infant showed signs of spastic tetraplegia (CG) and another dystonic cerebral palsy (IG). No significant differences were seen in MDI and PDI between the groups (table 3). Results were similar in subgroup analysis by sex. When data from both groups were pooled, energy and protein deficits at 28 days correlated significantly with MDI (R = −0.25, p = 0.03 and R = −0.32, p = 0.004, respectively) and PDI (R = −0.29, p = 0.01 and R = −0.3, p = 0.006, respectively) at 3 months post-term. The correlations were not significant at 9 months post-term.

The relationships between developmental outcome in the first year and growth outcomes at 36 weeks' PMA, and MRI outcomes at 40 weeks' PMA are shown in table 4. Highly significant correlations exist between weight and both mental and motor outcomes in the first year, as well as T2 relaxation time and mental outcome. After adjusting for birth weight, gestation, sex, antenatal steroid use, mode of delivery, maternal age, maternal education and exclusive EBM use in enteral feeds in the first month, the relationships remain significant between length, weight, TBV and T2 relaxation time, and mental outcome; and between weight and motor outcome.

DISCUSSION

As far as we are aware, this is the first paper to describe the relationships between early nutrition, post-natal growth, quantitative MRI findings and developmental outcome in the first year among very preterm infants. The preterm brain is vulnerable to insults and injury especially in the early post-natal period. Illnesses such as sepsis and chronic lung disease (CLD) increase the body’s metabolic demands and quickly lead to energy deficit. Brain growth requires substrates including amino acids and fatty acids; it also utilises glucose as a form of energy. Energy and protein deficit are common among very preterm infants. Brain volume at term may be improved by reducing energy deficit in the early post-natal period. Mental outcome in the first year appears to be closely related to growth at 36 weeks' PMA as well as brain volume and maturation at 40 weeks' PMA.

The role of early nutrition on the growth and neurodevelopment is unclear. Other factors, including illness severity and CLD may have stronger influences.24^–26 Nevertheless, the potential of reducing energy and protein deficits through greater attention to the provision of early post-natal nutrition should not be overlooked.

In their study, Lucas et al randomised infants to receive preterm or term formula, and breast milk supplemented with term or preterm formula. At 18 months post-term, PDI in infants fed solely preterm formula were significantly better than those fed solely term formula. Later formal cognitive assessments at 7.5–8 years however showed that only boys fed preterm formula had an advantage in verbal and overall IQ.27 The infants in this study are of a higher gestation (mean around 30 to 31 weeks) and birth weight (mean around 1300 to 1400 g). The length of time the infants were given parenteral feeds would have been significantly shorter. The preterm formula they used has the same protein content (2 g per 100 ml) to our control group.

Among the infants assessed in the first year, there appeared to be slightly more infants of lower gestation and birth weight in the intervention group. Similar proportions were also seen among all the survivors. This may contribute to the lack of detectable difference in the developmental outcome between the two groups. In addition, nearly a fifth of the infants assessed in both groups did not complete their assigned feeding regimen, many because they were transferred back to the referring hospital. Standards of neonatal care may also vary among smaller units.

MRI at 40 weeks' PMA may be better at predicting cognitive and motor outcomes than cranial ultrasound scanning.28 29 There are currently no imaging correlates for the spectrum of neurocognitive impairment seen in preterm infants.30 In their study, Abernethy et al evaluated preterm children born before 32 weeks at 7 years of age without major disability and attended mainstream school.21 They found an association between longer T2 relaxation times in the cerebral white matter and minor motor impairment (Movement ABC below fifth centile). These results are in keeping with our finding that longer T2 relaxation times are associated with lower motor index scores.

The limitations of this paper include the lack of blinding and the relatively small number of infants followed up to 9 months post-term. Only OFC measurements at 36 weeks' PMA and MRI analysis were blinded.14 Complete blinding was not possible due to practical reasons. However, the use of a single observer should improve the consistency of the developmental assessments. MR scanning was only performed in a selected group of infants who were well enough clinically to undergo this investigation and who lived within the geographical region. Although the MR scanner was changed part way through the study, this only affected 16 infants, with similar numbers in each group. The small numbers followed up in the first year may not be representative and may have influenced the results. However, the follow-up rate was similar in both groups. The poor follow-up rate may reflect the general poor attendance, attitude towards healthcare and mobility among the population we serve. Of note is the higher proportion of infants who had exclusive EBM in the first month among those followed up compared to those not followed up.

Developmental outcome in the first year was not our primary outcome. The sample size was calculated based on OFC at 36 weeks' PMA. This study serves as a pilot to inform future research. To determine the effect of early nutrition on MDI, 120 infants in each group are required in order to show a 0.5 SD difference in outcome. A multi-centre study may be preferable in order to recruit such numbers.

In conclusion, improving early nutrition by reducing energy deficit in very preterm infants may improve brain growth and maturation. There is a good correlation between OFC measurement and total brain volume. Quantitative MRI including the measurement of T2 relaxation time has the potential to predict both mental and motor outcomes. There may be a critical period of brain growth that is influenced by nutritional deficit. Ongoing studies to evaluate the effect of early nutrition and growth on school-age outcomes are underway.