Article Text

Download PDFPDF

Catch-up growth and metabolic outcomes in adolescents born preterm
  1. Nicholas D Embleton1,2,
  2. Murthy Korada1,2,
  3. Claire L Wood1,
  4. Mark S Pearce2,
  5. Ravi Swamy1,
  6. Timothy D Cheetham3
  1. 1Newcastle Neonatal Service, Royal Victoria Infirmary, Newcastle Hospitals NHS Foundation Trust, Newcastle upon Tyne, UK
  2. 2Institute of Health & Society, Newcastle University, Newcastle upon Tyne, UK
  3. 3Department of Paediatric Endocrinology, Royal Victoria Infirmary, Institute of Genetic Medicine, Newcastle University, Newcastle upon Tyne, UK
  1. Correspondence to Dr Nicholas Embleton, Consultant Neonatal Paediatrician, Royal Victoria Infirmary, Newcastle Hospitals NHS Foundation Trust, Newcastle upon Tyne NE1 4LP, UK; Nicholas.embleton{at}ncl.ac.uk

Abstract

Background Accelerated infant weight gain in individuals born full term is linked to cardiovascular risk in adulthood, but data in those born preterm are inconsistent.

Objective To investigate the association between weight gain in infancy and childhood with later markers of the metabolic syndrome in adolescents who were born preterm.

Study design Longitudinal cohort study.

Setting Children born preterm with regular assessments of infant growth had auxology, body composition (dual X-ray absorptiometry), blood pressure, insulin sensitivity and lipid profile determined in adolescence.

Results We reviewed 153 children (mean gestation 30.8 weeks, median birth weight 1365 g) of whom 102 consented to venepuncture at a median age of 11.5 years. Adolescent height and weight standard deviation scores (SDS) were similar to population averages (0.01±0.92 and 0.3±1.2, respectively) and did not differ between infants when grouped according to degree of catch-up in weight gain in the immediate postdischarge period to 12 weeks of age. There were no significant associations between infant weight gain (change in weight SDS adjusted for length) and later metabolic outcome. However, there were strong associations between more rapid childhood weight gain (after 1 year of age) and subsequent body composition (higher fat mass %, fat mass index and waist circumference) and metabolic markers (higher fasting insulin, blood pressure and lower insulin sensitivity).

Conclusions The association of rapid weight gain on health is time critical in those born preterm; in early infancy, this does not impact on metabolic status in adolescence, in contrast to rapid weight gain in childhood, which should be discouraged. However, given the critical importance of brain growth in the neonatal period and infancy, further research is needed before strategies that discourage infant weight gain or catch-up can be recommended for infants born preterm.

  • Endocrinology
  • Growth
  • Neonatology
  • Diabetes

Statistics from Altmetric.com

What is already known on this topic?

  • Preterm birth is associated with an increased risk of later adverse metabolic outcomes.

  • Early growth failure in preterm infants results in ‘falling off’ growth centiles, with many demonstrating later catch-up growth.

  • There are associations between rapid catch-up and metabolic risk in later life, although data in preterm-born individuals are inconsistent.

What this study adds?

  • Patterns of early growth predischarge and postdischarge are not strongly associated with adverse metabolic outcomes in early adolescence.

  • Weight gain in childhood is strongly associated with adverse metabolic outcome in adolescence.

Introduction

There is increasing evidence that the pattern of growth during early childhood is associated with markers of the metabolic syndrome in later life.1 The risk appears greatest for low birthweight infants and those with in utero growth restriction. Growth acceleration and higher nutrient intakes in the immediate postnatal period are also associated with later metabolic risks in infants born preterm2 ,3 and underline the critical role of early-life nutrition. While there are data to show that preterm infants are at increased risk of adverse metabolic outcomes throughout childhood and early adulthood,4 ,5 the extent to which higher nutrient intakes increase the risk is uncertain and need to be balanced against data that show consistent associations between lower nutrient intakes (and/or growth rates) and worse cognitive outcomes. The strongest evidence for the cognitive benefits of enhanced nutrient intakes in preterm infants comes from randomised controlled studies (RCTs) in the first few weeks of life that show a persisting cognitive advantage in adolescence that is matched by changes in brain structure.6 However, an increase in adolescent insulin resistance and blood pressure associated with faster growth in the first two weeks of life was also observed.3

The majority of preterm infants are discharged home with a weight centile below their birth centile.7 In the months following hospital discharge (typically around 36–37 corrected weeks), some preterm infants demonstrate catch-up growth (upward crossing of weight and/or length centiles). Nutritional interventions in this postdischarge period have variably demonstrated short-term improvements in growth, body composition and bone density,8 but overall the data do not support any specific strategy and none have shown benefits on neurodevelopment.9 Observational studies have shown an association between infant weight gain and improvements in childhood cognition,10 but there remain concerns that faster growth in both the predischarge and early postdischarge period may increase later metabolic risks, with some authors cautioning against rapid early growth.11

The aim of this study was to determine whether growth patterns in the neonatal period, infancy and childhood are associated with markers of the metabolic syndrome in adolescents who were born preterm. We wanted to compare the metabolic impact of weight gain predischarge and in early infancy, when growth is strongly influenced by nutritional management, with the impact of childhood growth (after 1 year of age) when the determinants of weight gain are more closely linked to lifestyle factors. Specifically, we wanted to explore the pattern of catch-up growth (upward crossing of weight centiles) in the period between term and 12 weeks of age because this appears to be a sensitive time period for later adverse metabolic programming in individuals born at term.

Method

Study design

The study population consists of children born preterm (Newcastle Preterm Birth Growth Study) who have been followed since hospital discharge in 1993–1998 and is based on individuals born preterm at ≤34 weeks gestation originally recruited to one of two randomised controlled nutritional trials12–14 along with contemporaneously recruited control infants discharged breast feeding, and additional co-twins or triplets. The mean gestation of the original cohort was ∼30 weeks, and mean birth weight was ∼1.4 kg, and we excluded infants with major neonatal morbidities (severe neurological abnormalities or lung disease at hospital discharge). The cohort were regularly reviewed with standard auxological measures conducted until 2 years corrected age, prior to this current assessment in early adolescence.15

Clinical assessment

A single researcher (MK) reviewed the cohort at 9–12 years of age following signed parental and child consent. Standing height and weight (Seca Leicester Portable Height Measure and Seca model 708, Seca, UK) were measured at least twice and the average taken for analysis. Weight and height standard deviation scores (SDS), adjusted for sex and age, were calculated using the Lambda, Mu, Sigma calculator and British 1990 growth reference standards.16 ,17 Self-assessment of pubertal status used a validated Tanner scoring chart.18 Blood pressure (systolic/diastolic) was measured using a sphygmomanometer and an appropriately sized cuff at rest on two occasions 10 min apart and a final mean value calculated. Body composition was assessed using dual X-ray absorptiometry (DXA) (GE Lunar iDEXA and encore software V.11, GE Healthcare, UK) by one of two operators, with minimal interobserver variability, to produce an estimate of fat mass (FM) and lean mass (LM). We calculated FM index (FMI, total FM/height squared) as this is a more reliable marker of adiposity19 and similarly also calculated lean mass index (LMI).

Biochemical markers

Plasma glucose, lipid and insulin samples were taken in the morning between 09:00 and 10:00 hours following an overnight fast. Glucose and insulin were measured at baseline and then 30 min after 1.75 g/kg glucose (maximum 75 g) administered as part of an oral glucose tolerance test (OGT). Glucose samples were analysed immediately and samples for insulin stored at −80°C prior to batch analysis at Addenbrooke's Hospital, Cambridge. Plasma insulin was measured by an automatic immunoassay method (Auto DELFIA PerkinElmer Life Sciences, UK). The interassay coefficients of variation were between 1.9% and 3.1%.

The biochemical assessment of insulin sensitivity in young people is a controversial topic and is viewed by some as a trade-off between simplicity (the measurement of fasting parameters) and validity (the hyperinsulinemic clamp that is widely regarded as ‘gold standard’).20 We used the Homeostatic Model Assessment V.2 (HOMA2), which is derived from fasting values, as an assessment of insulin sensitivity.21 Fasting insulin concentrations correlate well with insulin sensitivity as measured by the hyperinsulinemic clamp in adults22 and fasting-derived surrogates have been found to be robust in some paediatric studies as well.23 On the other hand, fasting values have also been deemed to be a poor measure of insulin sensitivity.20 In addition to HOMA, we therefore performed a short OGT (described above) so that we could calculate the insulinogenic index and the oral disposition index (ODI).24 The insulinogenic index is the ratio of the difference in insulin concentration at 0 and 30 min to the difference in blood glucose at the same time points and represents a measure of insulin secretion. The ODI is a measure of β-cell function relative to insulin sensitivity and is derived from the insulinogenic index together with fasting insulin concentrations.24 Cholesterol, high-density lipoprotein (HDL) and triglycerides were measured using a colorimetric method (Roche Diagnostics GmbH, Mannheim, Germany) with interassay coefficients of variation of 1.6%, 0.9–1.4% and 1.6–2.0%, respectively.

Statistical analysis

We determined change in weight SDS in specific epochs: birth–hospital discharge, discharge–term, term–12 weeks corrected age (‘12 weeks’), 12 weeks–1 year and 1 year–current. Variables were logarithmically transformed when residuals deviated from homogeneity. Multiple linear regression models were used to estimate relationships between weight gain and metabolic outcomes after adjustment for potential confounding factors including gestation, birthweight SDS, requirement for mechanical ventilation (which we used as a proxy for illness severity), sex, current age at follow-up and pubertal status. We also adjusted for change in length during the growth period to determine the impact of ‘inappropriate’ weight gain, except for the predischarge period where a reliable measure of length at birth was infrequently recorded. The coefficient represents the size of the change in each outcome variable for every unit change in weight SDS during the specified time period.

The association of catch-up in weight postdischarge between term and 12 weeks on later outcomes was determined by categorising the pattern of catch-up in weight using the change in weight SDS: no catch-up (change in weight SDS<0), slow catch-up (change in weight SDS 0–0.67) and rapid catch-up (change in weight SDS>0.67). We chose a value of 0.67 because this represents the spacing between centiles on most growth charts and would be recognisable to clinicians. The effect of rapid weight gain postdischarge was examined by comparing the group with rapid catch-up to the group with no catch-up using multiple regression adjusted for likely confounders.

The study included children from two RCTs (the majority of whom were solely formula fed) and controls who were breast fed for variable periods of time. This meant that in addition to multiple other nutritional factors predischarge (eg, duration of parenteral nutrition), there were potentially seven different feed groups from the original trial assignations following hospital discharge, and we therefore did not attempt to examine relationships between dietary group and later outcomes. Mann–Whitney U or t tests indicated that the follow-up sample was broadly representative of the original cohort for gestation, birthweight SDS and birth weight. Levels of statistical significance were set at p≤0.05. Statistical analysis was performed using the statistical software package STATA V.11.0.

Results

There were 247 subjects recruited to the original trials, of whom 153 (67%) children were followed up at a median age of 11.4 years and 102 (52 girls) consented to venepuncture for metabolic profile. Categorising the entire cohort on the basis of catch-up growth in the period between term and 12 weeks resulted in group sizes of 36 (no catch-up), 38 (slow catch-up) and 24 (rapid catch-up), but due to 4 with missing values there were 98 children with complete data at term, 12 weeks and later childhood (table 1).

Table 1

Unadjusted clinical characteristics by subgroup of catch-up growth in weight from term to 12 weeks of age

Mean gestation (30±2 weeks), median birth weight (1365 g IQR 1160–1640 g) and age at discharge (36 weeks) were similar between boys and girls, and 62/102 infants (61%) had received mechanical ventilation in the neonatal period. There were no differences in baseline characteristics between the three postdischarge catch-up groups, except for those in the rapid catch-up group who were approximately 6 months younger than those with no catch-up when the assessment took place (11.3 vs 11.8 years, p=0.03).

Catch-up between term and 12 weeks corrected age

Height and weight SDS at follow-up were similar to population averages (0.01±0.92 and 0.3±1.2, respectively) and did not differ between catch-up groups. Unadjusted values are shown in table 1. There were no differences in auxological parameters, body composition, blood pressure, fasting glucose or insulin, or measures of insulin sensitivity. There were no differences in fasting cholesterol or triglycerides, but unadjusted HDL levels were slightly higher (p=0.03) in the rapid catch-up group. After adjustment for change in length, and likely confounders including current age, gestation, birthweight SDS score, sex, pubertal status and neonatal ventilation status (table 2), the effect on HDL disappeared.

Table 2

Adjusted differences in adolescent outcomes by subgroup of catch-up growth in weight from term to 12 weeks of age

However, after adjustment, those in the rapid catch-up group had lower LDL (low density lipoprotein) (p=0.01) compared to the no catch-up group, with values for the slow catch-up group being intermediate between the other groups. Fasting log-insulin was higher in the rapid catch-up compared with the no catch-up group (1.86 vs 1.35 mU/L, p=0.04) and there was a trend to a lower 30 min glucose concentration (6.5 vs 7.88 mmol/L, p=0.05).

To explore the effect of weight change prior to discharge, we conducted linear regression as shown in table 3. We also analysed the predischarge period by dividing the cohort into tertile groups of change in weight SDS from birth, but we did not identify any significant association with adolescent outcomes after adjustment for confounders (data not shown).

Table 3

Regression analyses of change in weight z-score in different epochs on outcome variables at follow-up

Gain in weight during infancy and childhood

To determine the association of patterns of weight gain during infancy and childhood on later metabolic outcomes, we performed a multiple regression analysis for each of the time periods described and adjusted for likely confounders and change in length during the period (table 3). We also performed the regression model with adjustment for the effect of growth in the previous time period (conditional growth) in order to account for any effect of catch-up/down growth in a subsequent interval after a preceding period of particularly slow or rapid growth. This had no significant effect on the magnitude or direction of effect of the outcome variables, except for a borderline negative effect of growth between term and 12 weeks on later waist circumference and LMI, which may be due to chance.

In infancy (time periods up to 1 year of age), there were no significant associations between weight and later outcomes. However, during the period between 1 year of age and current age (childhood growth), there were strong associations, even after adjustment for change in length SDS and likely confounders, with measures of body composition (higher FM%, FMI, LMI and waist circumference, all p<0.001), higher fasting insulin (p=0.002) and lower insulin sensitivity (p<0.001). Systolic and diastolic blood pressure was higher (p=0.006 and 0.005), HDL lower (p=0.001) and total cholesterol:HDL was higher (p<0.001).

Discussion

In children born preterm, we did not demonstrate an adverse association of weight gain pattern in infancy on metabolic outcomes in adolescence, although fasting insulin was slightly higher, and serum LDL levels lower in those with rapid catch-up growth between term and 12 weeks. There was no association of growth pattern during the predischarge period or any of the time periods up to 1 year of age and later metabolic outcome. In contrast, there were strong and highly significant associations between weight gain adjusted for length (and other confounders) and metabolic outcomes in later childhood during the period of childhood growth (between 1 year and current age).

These data come from a relatively large, prospectively followed cohort with detailed measures of growth throughout infancy. We used accurate measures of body composition (DXA) in later childhood instead of relying on calculated indices such as BMI, which cannot distinguish between LM and FM. We adjusted the data for likely confounders and examined the impact of weight gain after adjusting for changes in length. Our metabolic assessment was comprehensive and included measurements of fasting glucose, insulin and lipids, as well an assessment of insulin sensitivity, insulin secretion and β-cell function.

As with many longitudinal studies, we were only able to follow-up a proportion of the original cohort, and only two-thirds consented to invasive metabolic testing. However, those who took part did not differ from the original cohort on baseline characteristics (sex, gestational age, discharge age, birth weight). We did not recruit a contemporaneous control group, but the complexity of exposures experienced by preterm infants are very different at all stages of early life from those born at term, so determining the impact of nutritional exposures at precise time points will always be challenging. Although our original studies in this cohort examined the impact of different dietary regimens, we did not analyse for potential later effects because of small group sizes.

Our data differ from a recent study where weight gain prior to term and catch-up weight after discharge were both associated with metabolic risk in young adults who were born with a similar degree of prematurity.11 There are several population characteristics that may explain differences between the studies including early nutrient exposures (eg, use of parenteral nutrition) and neonatal morbidity (eg, increased use of antenatal steroids, improved survival in more recent years). However, one major difference is in age at metabolic assessment. It is possible that young adolescence is too early to observe differences in measures of adiposity and insulin sensitivity if these are due to programming effects from infant weight gain, although other data clearly demonstrate effects of early nutrition in adolescence.3 Both the unadjusted and adjusted means for most outcomes related to infant growth were remarkably similar, suggesting a much larger sample size would be needed to determine differences using our methodology. Furthermore, given the rapid advances in neonatal care over the last three decades, outcomes in adulthood may not be generalisable to contemporaneous populations. Our study was conducted in an era where antenatal steroid use and surfactant were normal practice, but even so, many other neonatal practices (eg, the immediate commencement of parenteral nutrition) will have changed.

There are numerous challenges in interpreting observational data and only prospective RCTs with low attritional losses can determine causality. The only relevant RCT data in preterm infants with long-term follow-up available come from Singhal et al.3 Their data show that nutrient intakes in just the first 2 weeks of life, rather than later periods, have the strongest programming effects on later metabolism. In addition, the data of Singhal et al suggest that outcomes for the high nutrient intake groups were similar to a term control group. One possible interpretation is that it is lower nutrient intakes that are beneficial for metabolic outcomes rather than higher intakes being harmful per se compared with the background population.

While rapid catch-up in term infants, and especially those born growth restricted, is of no cognitive advantage but may result in metabolic harm, some data suggest that catch-up in preterm infants may also be harmful.11 However, this needs to be considered in the light of data showing clear cognitive advantages in infancy and later childhood of enhanced nutrient intakes in the predischarge period. At present, although data are limited, observational studies suggest there may also be a cognitive advantage associated with faster growth in infancy and early childhood.25 There are likely to be multiple mechanisms as well as potential genetic explanations26 for the association of higher infant growth rates with increased insulin resistance or improved cognitive outcome that require prospective controlled trials in order to determine causality.

It is important to note that while nutrient intakes predischarge can be carefully controlled, the majority of preterm infants in the immediate postdischarge period feed to satisfy caloric requirements.12 Manipulations in the ratio of calories to other nutrients such as protein would therefore be needed in order to modulate postdischarge growth patterns. This might prove challenging in infants who are breast feeding.

Until it is clear that enabling catch-up postdischarge is of no cognitive benefit, we feel it would not be sensible to recommend practices that restrict growth in the immediate postdischarge period. However, long-term benefits due to postdischarge nutrient enrichment (eg, specialised formula or use of breast milk fortifiers) are complex, remain to be determined and may be associated with either benefit or harm. Most importantly, if there are any long-term metabolic effects of infant growth, they appear to be smaller than the adverse metabolic impact of weight gain in childhood, which is amenable to alteration with lifestyle interventions and is of no cognitive or other health benefit.

Acknowledgments

The authors thank all the children and their parents, the nursing and medical staff on the Special Care Baby Unit, Royal Victoria Infirmary, Newcastle Hospitals NHS Foundation Trust and members of the research team who initiated the studies (Dr RJ Cooke) and conducted assessment in infancy (Dr KP McCormick, Dr I Griffin, Mrs S Eddy and Mrs M Henderson).

References

Footnotes

  • Contributors MK, TDC and NDE designed the study. MK and RS undertook data collection. CLW and MSP performed statistical analysis. NDE wrote the first draft of this manuscript. All authors have seen and approved the final draft.

  • Funding Nutricia UK provided funding for the initial controlled trials in infancy. Support for subsequent follow-up studies was provided by Novo Nordisk, Nutricia UK, and the Special Trustees Newcastle Healthcare Charity.

  • Competing interests None declared.

  • Ethics approval County Durham and Tees Valley Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Additional data are available by application to NDE.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Atoms
    R Mark Beattie