Introduction
Increasing evidence suggests that greater maternal adiposity, circulating glucose and gestational diabetes (GDM) cause, via intrauterine mechanisms, greater offspring birthweight and ponderal index at birth [
1‐
3]. Even among non-diabetic mothers there is a linear association between fasting and post-challenge glucose concentrations during pregnancy and higher birthweight and greater birth adiposity [
4,
5]. Beyond birth, there is evidence that exposure to GDM in-utero results in greater offspring adiposity and overweight/obesity in infancy and early adulthood (to date there have been no reports for older ages) [
1,
6] but few studies have reported associations with fasting and post-load glucose following universal oral glucose tolerance testing, as opposed to studies in which the OGTT was confined to at-risk women. The latter is important, as studies in which universal risk factor screening followed by OGTT in only those with risk factors might result in bias due to preferential diagnosis testing in obese women [
1,
7,
8].
South Asian adults have a characteristic phenotype of proportionately greater adiposity, increased insulin resistance and higher rates of diabetes and cardiovascular disease compared with white European adults [
9‐
12]. Recent evidence suggests that these differences are present in children at age 9–10 years [
13] and at birth [
14‐
18]. We have previously shown (in the same cohort as this study) that despite having notably lower birthweight, infants of Pakistani-origin mothers had higher total fat mass (as indicated by cord-blood leptin levels) and similar skinfolds in comparison with infants of white British women [
18,
19]. Subsequently, we showed that greater circulating fasting and post-load glucose in Pakistani women at least partly explained the ethnic difference in birth fatness as measured by cord-blood leptin [
20]. There is inconsistent evidence regarding the role of maternal fasting and post-load gestational glucose in ethnic differences in offspring adiposity beyond birth or infancy [
8,
21] and, to our knowledge, no studies have explored whether maternal BMI and glycaemia traits differ in their associations with offspring measures of growth and adiposity between white European and South Asian mother–offspring pairs.
The aims of this study were as follows: (1) to determine the magnitude and direction of any differences in weight, height, BMI and triceps and subscapular skinfolds between white British and UK-born Pakistani-origin children at age 4/5 years and (2) to compare associations of maternal early pregnancy BMI, fasting and post-load glucose and GDM with offspring weight, height, BMI and triceps and subscapular skinfolds at age 4/5 years between white British and Pakistani mother–offspring pairs participating in the Born in Bradford (BiB) cohort study, all resident in the same UK city and receiving the same antenatal care.
Discussion
We have shown that, compared with white British women, Pakistani women have a lower early pregnancy BMI but higher gestational fasting and post-load glucose concentrations and higher prevalence of GDM. At age 4/5 years, their children are taller, have lower BMI and TSF thickness and similar SSF thickness. In both groups of women, greater maternal BMI was associated with higher offspring height, weight, BMI and TSF and SSF thickness, with some evidence of stronger associations in Pakistani compared with white British mother–offspring pairs. There was no evidence that maternal gestational fasting glucose, post-load glucose or a diagnosis of GDM were positively associated with offspring adiposity in either ethnic group in this study. Indeed, point estimates tended to be inverse (i.e. higher maternal glucose associated with lower BMI and skinfold thickness in both groups at age 4/5 years), though CIs were wide and included the null value.
In our previous work with this cohort, we found evidence that Pakistani children have greater relative general adiposity [
18]. Our results in this study are broadly similar in that Pakistani children remain lighter and have similar SSF thickness to white British children by age 4/5 years. However, we also see that by this age Pakistani children have lower TSF thickness, which suggests that they may have lower peripheral adiposity (indicated by TSF) but greater central adiposity (indicated by SSF). When we adjusted the ethnic differences in skinfolds for offspring BMI, the difference for SSF thickness increased and became positive, further suggesting greater relative central adiposity. However, this change may be the result of collider bias, given the inverse association of Pakistani ethnicity with BMI and positive association of SSF thickness with BMI, and may be spurious [
26].
Our findings are consistent with those of other studies comparing white European and South Asian children; these studies generally found similar or higher skinfold thickness, bioimpedance and fat mass but lower BMI in UK schoolchildren of South Asian origin when compared with those of European origin [
13,
28,
29]. We have previously reported that despite smaller size at birth, Pakistani children demonstrate some catch-up in infancy so that by age 2 years they remain lighter but are taller, [
25]. Our results reported here for the same children at age 4/5 years suggest that this pattern of ethnic difference in BMI and height continues through early-to-mid childhood. These findings are similar to results from the UK Millennium Cohort Study, which found that Pakistani children were born shorter but were taller at age 5 years than white British children [
30] but differ from results of a study of older children aged 9–11 years [
13] that compared South Asian children, rather than Pakistani children, with white Europeans, which might explain the different findings.
In previous work, using cord-blood leptin as a marker of total infant fat mass, we found that greater fatness at birth in Pakistani infants compared with white British infants was at least in part mediated by greater gestational glucose in Pakistani women [
20]. In the analyses reported here, we found positive associations of maternal BMI with offspring height, weight, BMI and SSF and TSF thickness but we did not identify any robust associations between gestational glucose levels and offspring adiposity at age 4/5 years. The results of previous studies of the association of maternal gestational glycaemia with offspring size/adiposity seem to vary depending on the age at which offspring outcome is assessed. Mendelian randomisation analyses suggest a causal effect of greater maternal gestational glucose on birthweight and ponderal index [
2] and matched-within-sibship analyses suggest a causal positive effect by the time offspring are aged 18 years [
6]. However, two previous studies of largely white European offspring assessed in infancy found that the positive association of maternal gestational glucose with offspring birth adiposity disappeared within a few months after birth and remained null until at least 24 months of age whereas maternal BMI remained positively associated with offspring adiposity [
31,
32]. Similarly, a study of Singapore mother–offspring pairs found that maternal fasting glucose was not associated with adiposity to age 36 months (whereas pre-pregnancy BMI was) [
27]. By contrast, a study carried out in the USA found 1 h post (50 g) oral glucose challenge concentrations were positively associated with offspring BMI at age 36 months [
33]. Two separate studies of mother and offspring pairs participating in the HAPO (Hyperglycemia and Adverse Pregnancy Outcome) study have reported inconsistent associations between hyperglycaemia and offspring adiposity: participants resident in China showed a positive association with offspring adiposity at age 7 years, especially in girls [
21], whereas participants from Northern Ireland (UK) showed no association with offspring adiposity at ages 5–7 years. [
8] Taken together, these studies suggest a tendency for the positive effect of maternal gestational hyperglycaemia on birthweight and ponderal index to attenuate in infancy and early childhood at least up to age 4/5 years. Our study, which is considerably larger than previous studies, provides support for this age-related change, though it is possible that the lack of positive associations of GDM and fasting and post-load glucose with offspring adiposity in our study is because those diagnosed with GDM were treated and this treatment attenuates any true positive associations. We do not have information on treatment in the BiB study and so to explore this we repeated analyses after removing any women diagnosed with (and hence treated for) GDM and found results for associations of maternal BMI, fasting and post-load glucose with offspring adiposity remained unchanged. While it is possible that some women who had fasting/post-load glucose values close to the thresholds used for GDM diagnosis were encouraged to modify their diet, they would not have been treated with metformin or insulin and the possible inclusion of such women is unlikely to have attenuated a true positive association to such an extent that point estimates are weakly negative.
While we and others have found positive associations of pre-pregnancy or early pregnancy maternal BMI with offspring subsequent adiposity, including in infancy and early childhood, Mendelian randomisation analyses, negative control studies (comparing maternal and paternal BMI associations with offspring adiposity) and within-sibship analyses, conducted in largely white European populations, suggest that associations of maternal early or pre-pregnancy BMI with subsequent offspring adiposity are unlikely to be due to causal intrauterine effects [
6,
34,
35]. Thus, the associations observed here cannot be considered causal. Previous studies have suggested interactions between maternal BMI and gestational glucose measurements in their associations with offspring adiposity [
27] and also possible differences between associations in girls and boys [
21]. In our larger study, we found no evidence for such interactions.
The key strengths of our study are its large size and the availability of universal OGTT data for all participants. We have measurements of BMI and peripheral and central skinfolds and adjusted for a wide range of potential confounding factors. Up to 23% of offspring had missing outcome data; missing data was more common in white British children than in Pakistani children, reflecting how Bradford’s Pakistani community are more likely to remain in the city and have children in inner-city schools. However, the distributions of outcome measurements were similar in those without missing data and results were similar whether we used the multivariable imputed datasets or the complete case data. Gestational weight gain has also been considered as a possible developmental overnutrition risk factor [
1]; however, we do not have data on this in our cohort and so could not explore this here. Our results are from two homogeneous groups and do not necessarily generalise to all South Asians and white Europeans.
In conclusion, at age 4/5 years Pakistani children are taller than white British children but they are lighter and consequently have lower BMI. Assessment of skinfolds suggest that they may have lower levels of peripheral adiposity (lower TSF thickness) but that they may be relatively more centrally adipose (similar SSF thickness). In both ethnic groups, children of mothers with higher BMI were larger and more adipose, with these positive associations tending to be stronger in Pakistani children than in white British children. By contrast, none of the maternal gestational glycaemia measurements were associated with offspring size and adiposity in either ethnic group. As maternal glycaemia is not related to offspring adiposity in either group at 4/5 years it cannot explain ethnic differences in size and adiposity that we, and others, have observed at this age.