Newborn size, and childhood growth, and cardiovascular disease risk factors at the age of 6 years; The Pune Maternal Nutrition Study

Introduction

South Asia has the highest rate of rise of type 2 diabetes in the world¹ and cardiovascular disease (CVD) is predicted to become the leading cause of death by 2015². Insulin resistance is an important risk factor for type 2 diabetes and CVD in South Asian populations^3-4. South Asians develop the insulin resistance syndrome (IRS), and related morbidity and mortality at a younger age and lower body mass index (BMI) than white Caucasians^5-6. This has been attributed to a characteristic South Asian body phenotype, comprising a low muscle mass, high percentage body fat, and tendency to central adiposity^7-8 (‘thin-fat’ phenotype⁹).

There is mounting evidence that small size at birth is associated with an increased risk of CVD, type 2 diabetes and the IRS^10-12. In the majority of these studies, the only measurement at birth is weight. Recently we described the detailed anthropometry of South Asian Indian newborns, and showed that they are markedly light in weight, have a reduced non-fat soft tissue mass, but are relatively adipose compared with white Caucasian newborns (the ‘thin-fat’ newborn phenotype)^9,13,14. We do not yet know how this relates to later CVD risk.

The risk of CVD, type 2 diabetes and the IRS is also related to growth in infancy and childhood. Studies in adults have shown that higher weight or BMI at the age of one year is associated with a decreased risk of disease, while rapid weight or BMI gain in childhood is associated with an increased risk^15-20 Studies in children have shown that features of the IRS are elevated in those who had grown heavier, taller or fatter relative to their peers^21-26. Childhood may be an opportune time to intervene to prevent excessive weight or BMI gain and thus prevent the development of insulin resistance and later disease. However, it is not clear which components of childhood body composition (adipose tissue, lean mass or skeleton) are related to later risk, nor do we know the critical time during childhood when interventions to limit weight gain may be appropriate²⁷. Most published studies were retrospective, with large dropout rates, making their interpretation difficult.

The Pune Maternal Nutrition Study is a prospective, population-based study of maternal nutrition, pregnancy outcome and CVD risk in the children, being carried out in a rural community in Western India. For the children, detailed anthropometric measurements were made at birth, and serially every six months during infancy and childhood. CVD risk factors were measured at the age of 6 years to test the following hypothesis: Low birthweight and the ‘thin-fat’ phenotype at birth, smaller size in infancy, and more rapid childhood growth, especially fat gain, will predict higher levels of components of the insulin resistance syndrome. Associations between maternal diet during pregnancy and cardiovascular outcomes in the children will be reported separately.

Patients and Methods

The Pune Maternal Nutrition Study methodology has been described previously⁹ (Figure 1). We prospectively studied non-pregnant married women from 6 villages near Pune, to record their menstrual dates and anthropometry. Women who became pregnant were studied twice during gestation, to measure their nutrition and fetal growth. There were 770 deliveries, including 8 stillbirths. The babies were measured within 72 hours of birth by one of 5 trained fieldworkers. Birthweight was measured to the nearest 25 g using a Salter spring balance, crown-heel length to the nearest 0.1 cm using a portable Pedobaby Babymeter (ETS J.M.B., Brussels, Belgium), triceps and subscapular skinfold thicknesses were measured to the nearest 0.1 mm, on the left side of the body, using Harpenden skinfold calipers (CMS Instruments, London, UK), occipito-frontal head circumference and mid-upper-arm circumferences (MUAC) were measured to the nearest 0.1 cm using fiberglass tapes (CMS Instruments). Of 762 liveborn babies, 702 were measured at birth. The duration of exclusive breastfeeding was recorded at one-year follow-up.

Statistical methods

Variables with skewed distributions (skinfolds, insulin and triglyceride concentrations and insulin resistance) were log transformed or inverted to satisfy assumptions of normality. All available data were used at each age. Measurements at birth were adjusted to a gestational age of 40 weeks using multiple linear regression. We use the term ‘size’ to describe anthropometric measurements. Size in infancy refers to measurements at 6 months and 1 year, size in childhood to measurements at 2,3,4 and 5 years of age, and current size to measurements at 6 years. Body composition refers to fat and lean mass measured using dual X-ray absorptiometry. Body composition measurements were adjusted to the age of 6 years using multiple linear regression. Associations between earlier size and fat and lean mass are reported using standardised beta coefficients (indicating the standard deviation (SD) change in the outcome per SD change in the predictor variable) to enable comparison of the effects of earlier size on these two body composition measurements. Associations between earlier size and CVD risk factors at 6 years were analysed by multiple linear regression, using three models: 1) including early size alone, 2) adjusting for current weight, and 3) including the interaction between early size and current weight.³⁰ Differences between the sexes in associations between earlier size and outcomes at 6 years were examined by including interaction terms (sex * the early size variable) in the regression models.

In addition to using individual birth measurements to predict 6-year outcomes, a principal components analysis was carried out to derive composite measures describing the body proportions of the baby. From the original variables (newborn weight, length, head circumference, subscapular and triceps skinfolds and MUAC) we derived new variables or ‘components’ which are uncorrelated. In our data, the first principal component had high positive loadings for all birth measurements, indicating that this was a measure of overall body size (Table 1). The second component had strong positive loadings for skinfolds and negative loadings for length, head circumference and mid-upper-arm circumference. We used this component as an indicator of the ‘thin-fat’ newborn phenotype. Overall, 76% of the variation in size at birth was explained by the first 2 principal components. Both principal components were included simultaneously, with and without adjusting for 6-year weight, in multiple regression analyses to predict CVD risk factors at 6 years.

Growth refers to change in size over time. In order to examine independent associations of growth during different periods (birth-6 months, 6 months to one year, and every year up to the age of 6 years) with outcomes at 6 years, we used Royston’s conditional standard deviation (SD) scores method³¹. SD scores for all body measurements, and conditional SD scores, were calculated at exact ages of 6 months, and every completed year thereafter. Growth charts were constructed (not shown) separately for boys and girls, using a multilevel regression model (MLN software, University of London), from which age and sex-specific SD scores of achieved size at these time points, and SD scores of growth between successive time-points, conditional on previous size, were derived. These conditional SD scores are uncorrelated with each other and indicate whether growth during a particular time period is faster or slower than would be predicted from the average growth of the cohort and the child’s size at earlier time-points. The conditional SD scores can be used simultaneously in a regression model, along with size at birth, to determine associations of faster or slower growth at different time points with later outcomes. For the conditional SD scores analysis, we used data from children with complete measurements at birth, 6 months, and annually up to 6 years (n=566). Analyses were carried out using SPSS Version 11.0 and STATA 7.0. Effects of growth on different outcomes were compared by using the seemingly unrelated regression model (SUREG) command.

Results

Out of the 762 live births, 39 died before the age of 6 years (Figure 1); of 723 children available for follow up at 6 years, 698 children (97%) attended for measurements. Table 2 shows the children’s anthropometric measurements at birth, 1 and 6 years, and 6-year CVD risk factor and body composition measurements. Compared with international references (NCHS³² and WHO³³) the children were light, short and thin. Compared with a UK skinfold reference³⁴ (NCHS and WHO do not provide skinfold references) the children were relatively truncally adipose at 6 years, with subscapular skinfold SD scores higher than SD scores for weight, height and BMI. We limited our analysis to the following outcomes: insulin resistance (HOMA-R), systolic blood pressure, and plasma triglyceride, total and HDL cholesterol, glucose concentrations and fat and lean mass.

Discussion

We have measured body composition and risk factors for cardiovascular disease (CVD), in a large population-based cohort of rural Indian children who had serial anthropometry from birth to 6 years. As far as we know this is the first study to incorporate such detailed longitudinal measurements of body size with later body composition and CVD risk factors.

Larger overall size at birth (principal component 1) and larger newborn weight, length and MUAC predicted both fat and lean body mass at 6 years, but showed significantly stronger associations with lean mass. These findings are consistent with other studies in children and adults, and with the suggestion that fetal life is a critical period for the development of later lean body mass^35-37. Skinfold measurements at birth were unrelated to body composition at 6 years, and the ‘thin-fat’ newborn phenotype (principal component 2: a contrast between skinfolds and other body measurements), which we expected to be associated predominantly with adiposity at 6 years, was related equally to fat and lean mass.

Birth measurements were largely unrelated to CVD risk factors at 6 years. Lower birthweight was associated with higher systolic blood pressure only after adjusting for 6-year weight. This contrasts with findings in a cohort of 8-year old urban Pune children, among whom, after adjustment for current weight, lower birthweight was associated with higher blood pressure, HOMA-R and cholesterol concentrations²⁴. An explanation could be relatively less post-natal growth in the rural children. SD scores for weight, height and BMI for the urban children were -1.21, -0.60, -1.60 (boys) and -0.91, -0.41, -1.23 (girls) according to the NCHS reference³². We previously speculated that the ‘thin-fat’ newborn phenotype would be associated with an adverse CVD risk profile at 6 years⁹; however, it was unrelated to CVD risk factors. Our data do not suggest important associations of small size at birth per se with an adverse CVD risk factor profile in prepubertal children in this population.

As expected, weight growth was associated with fat and lean mass, height growth with lean mass, and skinfold growth with fat mass at 6 years. Somewhat unexpectedly, height growth was also associated with fat mass, which could just reflect larger size of the body. Faster MUAC growth predicted lean and fat mass approximately equally as expected from the components of MUAC (bone+muscle+fat). Previous studies have shown associations of weight, weight gain and length gain in infancy with lean mass in later life^37-40. Unlike the previous studies (systematically reviewed)^41,42 we are able to predict fat and lean mass more specifically.

There is little information on associations between growth in early life and cardiovascular risk factors in childhood. In our study, slower MUAC growth between birth and 6 months was associated with higher insulin resistance at 6 years. This may be a chance finding, or indicate an association between growth of both fat and lean mass in the early post-natal period and the later development of insulin resistance. Larger body size or faster growth from birth to 6 months did not predict an adverse CVD risk factor profile at 6 years. There is currently controversy over the long-term effects of infant weight gain. Studies in adults suggest that lower weight or BMI at one year predicts an increased risk of CVD and type 2 diabetes^{11,15,17- 19}. However, among children born preterm who took part in randomised controlled trials of different infant feeds, those with more rapid weight gain in early infancy had increased CVD risk factors in childhood⁴³. Our data do not suggest strong effects one way or the other of changes in body size during the first 6 post-natal months on CVD risk factors.

After 6 months, higher weight and height and faster growth in these measurements, were associated with higher 6-year blood pressure and insulin resistance. The mechanisms linking rapid growth to CVD risk factors are unknown. Lever proposed that the endocrine ‘drivers’ of growth also stimulate vascular smooth muscle hypertrophy^44,45. Another possible mechanism is through changes in body composition, for example increased fat mass relative to lean mass. However, faster weight and height gain predicted lean mass more strongly than fat mass at 6 years. Larger earlier size or faster growth could be proxies for larger current size. Our findings would then imply that simply having larger fat and lean mass has adverse effects on CVD risk factors. The association between fat mass and insulin resistance probably reflects an excess release of fatty acids and adipocytokines. The link between larger lean mass and higher risk factors is difficult to explain. It could reflect functional characteristics or intra myocellular fat deposition⁴⁶. Blood pressure is physiologically higher in taller people, and height is positively correlated with lean body mass. However, our data showed a positive association between 6-year lean mass and systolic blood pressure, independently of height, and other studies have shown positive associations between muscularity and blood pressure^47,48. Brenner has proposed that the number of renal nephrons, which is fixed during early development, may become inadequate later in life in the face of higher body weight, especially lean body mass, leading to hypertension⁴⁹.

In conclusion, our data show that larger size at birth and more rapid growth in early childhood is associated with higher lean body mass and increased CVD risk factors at 6 years. It is striking that these associations are found in a rural Indian population, among children who are growing slowly, and many of whom may be categorised as ‘under-nourished’ according to WHO definitions⁵⁰. The findings may not be generalisable to populations in western countries where children are born heavier and remain larger throughout childhood, but would be applicable to a large number in developing countries. The effects we observed were small; an increase of one third of an SD in weight in any year after infancy was associated with, on average, an increase of 6% in 6-year insulin resistance and 0.6 mmHg blood pressure. However, even such small changes, when observed in childhood and across the normal range in a population can translate into important effects in public health terms⁵¹. Associations between 6-year CVD risk factors and growth during infancy were smaller compared to growth later in childhood. It has been recommended that weight gain should be promoted during infancy in under-nourished populations. This is based on the evidence that this favours immediate survival⁵², and may prevent stunting and improve neuro-cognitive development^53-55. Our findings do not conflict with this advice. However, children who grow rapidly at later ages in childhood may be at increased risk of type 2 DM and CVD.