Adriëtte Oostvogels

4 115 Maternal lipids and offspring’s lipids & glucose However, this study also has some limitations. First, although maternal lipid levels did not differ between the response and non-response groups, selection bias might have occurred as our final study population consisted of women that were older, higher educated, more often from Dutch origin and smoked less, but drank more alcohol during pregnancy compared to the non-response group (Supplementary table 1). This selective inclusion of a slightly ‘healthier’ study population may have led to an underestimation of the results, because of less variation in offspring’s lipids and glycaemic control. Second, insulin resistance was obtained by means of an indirect method, the HOMA2-model, and based on C-peptide. We had to impute 63.6% of the C-peptide levels since they were under the detection limit of 0.34. Sensitivity analyses with non- imputed C-peptide levels showed about the same results with also no evidence for an association when adjusted for all covariates (data not shown). Third, maternal lipid levels were determined based on one blood sample, taken in a non-fasting state randomly during the day; this could have resulted in a distorted maternal lipid profile. However, a large Danish study in adults showed that, in contrast to TG and TC, ApoA1 and ApoB did not alter after normal food intake. 29 In addition, high correlations (range: 0.90-0.99; p<0.001) were found between fasting and postprandial TG and TC. 30 Furthermore, in normal conditions the foetus is also exposed to non-fasting maternal lipids for most of the day. Finally, in this study, multiple tests were applied. Because we considered our analyses to be explorative, we did not apply the Bonferroni correction. Some statistical significant findings might therefore be the result of a type 1 error. However, our associations between maternal TC and ApoB and offspring’s’ TC and LDL were quite strong with remarkable sex differences (Table 2). Our results are a first indication that girls might be more affected by the maternal lipid profile. Further studies with specific attention for sex differences in the associations are necessary to confirm this hypothesis. Conclusion This explorative study shows an association between maternal lipid levels during early pregnancy and offspring’s lipid levels in childhood, with stronger associations in girls. No evidence was found for associations with offspring’s glycaemic control, or for a mediating role of offspring’s fat percentage. Additional research is required to determine to what extent genetic and environmental factors play a role. When our results are confirmed and maternal lipids have the potential to programme the