Adriëtte Oostvogels

profile in childhood, blood pressure course (irrespective of hypertensive disorders) might yield different results: this needs to be studied to exclude blood pressure course during pregnancy as a potential underlying mechanism in the association between maternal pBMI and offspring’s cardiometabolic profile. In this thesis, accelerated postnatal growth modified the association between maternal pBMI and childhood cardiometabolic profile. Based on animal studies, this seemed to be due to dysfunctioning adipocytes and related secretion of pro- inflammatory cytokines. 77,78,115 Until now, few studies have examined the long-term outcome of this so-called ‘low-grade inflammation’. 115,116 Only one study showed positive associations between second-trimester C-reactive protein concentrations and body composition at age 7-10 years. 116 More research is needed to identify whether low-grade inflammation affects only body composition, or also other components of childhood cardiometabolic profile. Although postnatal growth did not mediate the association between pBMI and offspring’s cardiometabolic profile, it seems worthwhile to assess the role of growth throughout childhood. 50,51,117-119 To model the adiposity rebound better, follow-up data of the children in the ABCD-study (now available for age 11-12 years) should be used. Moreover, now that self-reported information of the children is available, differences in appetite, satiety and food preferences between children with mothers of normal weight and with overweight, can be assessed. Furthermore, sex differences should be studied in more detail as this will provide important sex data on disease aetiology. Moreover, although ethnic and socioeconomic inequaltities are reported in both maternal and cardiometabolic profile, further research is needed on ethnic and socioeconomic differences in the associations between maternal metabolic parameters and childhood outcomes, as this thesis demonstrated socioeconomic inequalities in growth to overweight. Finally, the analyses performed in this thesis were performed in a longitudinal cohort. This means that, although the determinant pBMI precedes the effect (i.e. cardiometabolic profile in childhood) and, therefore, one of the conditions of causality is fulfilled, no conclusions can be drawn with regard to causality. Future research should aim to assess causal pathways between maternal overweight and childhood cardiometabolic profile. For this, possible designs could include cross- cohort comparisons, Mendelian randomisation studies, or randomised controlled trials. First, a cross-cohort comparison that shows consistency between cohort studies would add to the credibility of the findings, but also cannot prove causality. 120 Second, a Mendelian randomisation study can be used to make causal inference: 232 Chapter 9