This study shows that non-LGA and LGA OGDM display different BMI SDS trajectories during childhood, both groups reaching slightly higher values than those of the 2009 (Fifth) Dutch Growth Study. Although LGA OGDM had the higher scores, both subgroups of OGDM had a mean BMI SDS of less than +1 SD when they reached early adolescence, indicating that their risk for becoming overweight in the short term is likely to be limited. The height SDS for non-LGA and LGA OGDM was comparable with that of the 2009 background population. In both the Dutch Growth Study and our OGDM group, BMI SDS compared with the 1980 nationwide study was increased, suggesting a gradual shift in offspring BMI reference values and an increased incidence of overweight/obesity in both the reference group and the OGDM since the 1980s .
It is becoming increasingly clear that a hyperglycaemic intrauterine environment is responsible for an increased risk of diseases in childhood and later adulthood. The relatively overnourished offspring of gestational diabetic pregnancies are prone to later development of obesity and diabetes. This is due, not only to genetic susceptibility , but also to exposure to the abnormal intrauterine environment with potential for epigenetic changes in the fetal phenotype . These intrauterine phenomena contribute to childhood obesity and possibly to obesity and diabetes in adulthood and an increased risk for cardiovascular disease [36, 37]. Thus, it is evident that preventive strategies are needed. This matter is intuitively most relevant for offspring likely to be most at risk: those born LGA. Therefore, we analysed the growth trajectories of non-LGA and LGA OGDM separately.
Several cross-sectional studies on OGDM have found evidence for overweight/obesity in this population [7, 9,10,11,12,13]. However, although the available longitudinal BMI data in OGDM indicate a higher childhood BMI [17, 19, 20], with sex and maternal prepregnancy BMI as significant contributing factors [18, 21], results cannot be compared due to differences in methodology. In the only other longitudinal study in which LGA and non-LGA infants were studied separately, at age 4–7 years LGA OGDM were found to have a higher BMI than either non-LGA OGDM or LGA offspring from women without diabetes . In a cross-sectional prospective cohort of 6- to 11-year old infants, no difference in BMI was found between LGA and non-LGA OGDM and the offspring of mothers without diabetes in pregnancy .
We found that maternal BMI did not influence the BMI SDS trajectory of OGDM. This is in contrast to most studies in which maternal BMI was taken into account [8, 9, 17, 22, 39, 40] and might be due to the relatively low incidence of maternal obesity in our study group (mean BMI, 25.8 kg/m2; 30.9% overweight [BMI 25–29.9 kg/m2] and 22% obese [BMI ≥30 kg/m2] [9, 22, 40]. This is relevant given the higher BMI in most GDM studies from a range of countries [9, 20]. Apparently, this profile of women with GDM is common in our region and may represent a clinically different entity (with a different outcome for their offspring) when compared with other studies. Previous studies have shown that in 16–17-year-old adolescents, the relationship between maternal gestational diabetes and offspring BMI was attenuated after correction for maternal prepregnancy BMI  and weight gain in pregnancy . Furthermore, offspring of mothers with gestational diabetes and a normal BMI did not show an increase in obesity, in contrast to offspring of obese mothers with gestational diabetes [9, 40]. Maternal obesity is, therefore, a strong predictor of childhood obesity  and a higher childhood BMI is related to maternal BMI  in pregnancies complicated by GDM. In our cohort of mothers with GDM, the relatively low maternal BMI may explain the limited effect on childhood BMI.
In the literature we did not find any studies comparing the growth trajectories of OGDM with those of ODM1 and ODM2 separately. Such a comparison is essential when studying (epi)genetic pathways of obesity. Figure 3 clearly shows that ODM2 have the highest risk of childhood obesity. The most striking difference between the GDM, type 1 diabetes and type 2 diabetes groups was the maternal BMI, which was higher in women with type 2 diabetes (31 kg/m2), compared with women with type 1 diabetes (24 kg/m2) and those with GDM (26 kg/m2). This highlights the importance of maternal BMI with respect to infant growth.
Only a few studies are available that examine a heterogenic group consisting of ODM1, ODM2 and OGDM [15, 41,42,43]; ODM1 and OGDM have been compared as separate groups in only two studies. Overweight rates in OGDM and ODM1 were twice those of the background population , with a higher risk for obesity at age 18–20 years in OGDM and only a weak association for ODM1 .
This study was performed at a single centre, where pregnant women with GDM were diagnosed and treated according to the same protocol.
A longitudinal control group was not available for comparison. We could, however, compare our data with nationwide data from the Dutch population [26,27,28]. The reported BMI and height SDS were calculated based on these population values and the latter were plotted as graphs for comparison . Unfortunately, it was not possible to calculate statistical significance between the OGDM subgroups and the nationwide population because the complete set of raw (cross-sectional) data from the Dutch Growth Study was not available for analysis.
In this study, we only received informed consent from 23% of all eligible mothers. The response rate was lower than that reported in the literature (ranging from 46% to 66% [10, 18, 20, 39]; although response rates were not reported in most studies), illustrating that this cohort is hard to trace and recruit. This issue is present explicitly or implicitly in many studies. A prospective cohort study would be the best way to include and follow-up the offspring of pregnant women with GDM.
In the Netherlands, GDM is most prevalent in the subgroups of the population that also have a high incidence of type 2 diabetes, including natives from Turkey, Morocco, Suriname and the Caribbean islands. These groups were under-represented in the current study, possibly because some of these minority populations are under-represented in our region or because the parents and offspring in these groups are sometimes difficult to trace.
Given the ethnic heterogeneity of our study population, a subgroup analysis was not performed. BMI SDS of OGDM in our cohort might have been higher if more Mediterranean-Dutch children had participated . Although the mean growth trajectories, as depicted in the figures, appeared to be different, statistical significance was not always reached for non-LGA OGDM vs LGA OGDM, which may partly be due to the small numbers of Mediterranean-Dutch children involved.
Tanner stage and onset of puberty were not recorded in the current study and, thus, the impact of this natural process on the current group is unknown. By using SDS values from the nationwide study, in which children at different stages of puberty were included, the effect of puberty on height and BMI SDS trajectory is evened out.
In conclusion, this is the first study to analyse longitudinal growth trajectories in OGDM up to the age of 14 years with separate height and BMI SDS for LGA and non-LGA offspring. Growth trajectories were only slightly above those of the Dutch reference population (e.g. Dutch Growth Study), indicating that the risk of OGDM becoming overweight in adolescence seems limited, at least in a population with a low incidence of maternal obesity. Compared with offspring of women with pregestational diabetes, we showed that ODM2 are at highest risk of becoming overweight in early adolescence, followed by LGA ODM1. Non-LGA ODM1 had a growth trajectory comparable with that of the reference population. This study gives a broad view on the intrauterine origins of adult disease in offspring of women with diabetes. Parents and healthcare workers should carefully consider the risk of obesity in offspring of women with diabetes, especially infants born LGA, and should initiate preventive measures where possible. These may include diet and lifestyle coaching for children and parents.