Cord blood metabolites and rapid postnatal growth as multiple mediators in the prenatal propensity to childhood overweight

Background The mechanisms underlying childhood overweight and obesity are poorly known. Here, we investigated the direct and indirect effects of different prenatal exposures on offspring rapid postnatal growth and overweight in childhood, mediated through cord blood metabolites. Additionally, rapid postnatal growth was considered a potential mediator on childhood overweight, alone and sequentially to each metabolite. Methods Within four European birth-cohorts (N = 375 mother-child dyads), information on seven prenatal exposures (maternal education, pre-pregnancy BMI, weight gain and tobacco smoke during pregnancy, age at delivery, parity, and child gestational age), selected as obesogenic according to a-priori knowledge, was collected. Cord blood levels of 31 metabolites, associated with rapid postnatal growth and/or childhood overweight in a previous study, were measured via liquid-chromatography-quadrupole-time-of-flight-mass-spectrometry. Rapid growth at 12 months and childhood overweight (including obesity) between four and eight years were defined with reference to WHO growth charts. Single mediation analysis was performed using the imputation approach and multiple mediation analysis using the extended-imputation approach. Results Single mediation suggested that the effect of maternal education, pregnancy weight gain, parity, and gestational age on rapid postnatal growth but not on childhood overweight was partly mediated by seven metabolites, including cholestenone, decenoylcarnitine(C10:1), phosphatidylcholine(C34:3), progesterone and three unidentified metabolites; and the effect of gestational age on childhood overweight was mainly mediated by rapid postnatal growth. Multiple mediation suggested that the effect of gestational age on childhood overweight was mainly mediated by rapid postnatal growth and that the mediating role of the metabolites was marginal. Conclusion Our findings provide evidence of the involvement of in utero metabolism in the propensity to rapid postnatal growth and of rapid postnatal growth in the propensity to childhood overweight. We did not find evidence supporting a mediating role of the studied metabolites alone between the studied prenatal exposures and the propensity to childhood overweight.


Figure S12
Results from sensitivity analyses excluding the preterm birth of (a) single mediation of gestational age on postnatal rapid growth through cord blood metabolites and (b) on childhood overweight through rapid growth, (c) on childhood overweight through cord blood metabolites, and (d) sequential mediation of gestational age on childhood overweight through cord blood metabolites and postnatal rapid growth.

Figure S13
Results from sensitivity analyses excluding children born by caesarean delivery of the effect of prenatal exposures on infancy rapid growth through the cord blood metabolites (N=276). 23

Cohort specific study population
ENVIRONAGE is an ongoing population-based prospective birth cohort study that aims at exploring the human ageing and its interaction with the environment 1 . From February 2010 onward, more than 2,000 mother-infant pairs have been recruited at delivery at the East-Limburg Hospital in Genk (Belgium). The study was approved by the ethics committee of the Hasselt University and the Hospital East-Limburg, Genk, Belgium and conducted in accordance with the Declaration of Helsinki. Anthropometric data were self-reported by the parents until the age of 2 years, and measured by trained-staff during a follow-up sub-study when the child was two years of age.
INMA is a population-based birth cohort including 3,768 mother-child pairs enrolled between 1997 and 2008 during the first prenatal visits at different health care centres Spain 2 . The study was approved by the ethics committee of the Hospital del Mar Medical Research Institute and conducted according to principles of the Helsinki Declaration. This study is based on 657 mother-child pairs recruited at two health care centres in Sabadell. Anthropometric data were collected via surveys or measured until at least adolescence.
Piccolipiù is a birth cohort of more than 3,000 mother-child pairs recruited in five Italian cities (Turin, Trieste, Viareggio, Firenze and Rome) 3 . Ethical approvals have been obtained from the Ethics committees of the Local Health Unit Roma E (management centre), of the Istituto Superiore di Sanità (National Institute of Public Health) and of each local centre. This study is based on 544 mother-child pairs recruited at the Sant'Anna Hospital in Turin. Follow-up anthropometric data collection surveys occurred at 6, 12 and 24 months after the delivery and then when the children turned 4 and 6 years with direct measurements at a clinical visit.
Rhea is a prospective mother-child cohort of 1,610 mother-child pairs recruited between 2007 and 2008 during the first trimester pregnancy visits at the public primary health care centres or hospitals in Heraklion, Greece 4, 5 . The Ethics Committee of the University Hospital at Heraklion approved the study protocols. Follow-up anthropometric measurements were collected via surveys or measured for the participants are available up to 11 years.
Informed consent for participation to each study was provided by parents or mothers.  Scores we assigned to each prenatal factor concurring to the obesogenic factors score calculation. The scores were based on a-priori knowledge, and for each prenatal exposure we reported at least one reference reporting an association with childhood overweight and obesity.     1.11-1.61)  1.61 (1.25-2.11) Rapid growth -4.39 (2.14-9.28) All the analyses are adjusted for sex of the newborns, child ethnicity, cohort membership and child age (for childhood overweight analyses only). The analyses of maternal pre-pregnancy BMI, maternal weight gain, smoking, age at delivery, gestational age and parity are additionally adjusted for maternal education, pre-pregnancy BMI, smoking, age at delivery, and parity.

Figure S1
Heatmaps showing Pearson correlation coefficients, scree plots, loading plots and plots showing first 10 contributors to the first and fourth principal components from principal components analyses of the 32 cord blood metabolic features under study in the population participating to (a) the single mediation analysis of the effect of prenatal exposures on postnatal rapid growth mediated by metabolites (N=375) and (b) to all the analyses on childhood overweight (N=249).

Figure S2
Results of the single mediation analysis of the effect of prenatal exposures on infancy rapid growth through the cord blood metabolites (N=375).

Figure S8
Results of sequential mediation of prenatal on childhood overweight defined using IOTF cut-offs through the cord blood metabolites and postnatal rapid growth (N=245).    S12 Results from sensitivity analyses excluding the preterm birth of (a) single mediation of gestational age on postnatal rapid growth through cord blood metabolites and (b) on childhood overweight through rapid growth, (c) on childhood overweight through cord blood metabolites, and (d) sequential mediation of gestational age on childhood overweight through cord blood metabolites and postnatal rapid growth. CI: confidence intervals, NIE: natural indirect effect; NIEM1: natural indirect effect via M1; NIEM2: natural indirect effect via M2; NDE: natural direct effect; OR: odds ratio; TE: total effect Excluding the preterm births, results were similar to main analysis: the effect of gestational age on rapid growth was mediated by cholestenone (NIE OR per each week of gestation= 0.89, 95% CI= 0.81-0.95, TE OR per each week of gestation= 0.54 per each week of gestation, 95% CI= 0.39-0.70) while mediation by U8 (m/z 269.1894) and PC (C34:2) attenuated and were not significant anymore; the effect of gestational age on childhood overweight was mediated by rapid growth (NIE OR per each week of gestation= 0.84, 95% CI= 0.71-0.94, TE OR per each week of gestation=0.83 per each week of gestation, 95% CI=0.59-1.16); no metabolite mediated the effect of any studied exposure on childhood overweight and the multiple mediation confirmed the effect of gestational age on childhood overweight was mostly mediated by rapid growth Figure S13 Results from sensitivity analyses excluding children born by caesarean delivery of the effect of prenatal exposures on infancy rapid growth through the cord blood metabolites (N=276).
Excluding children born by caesarean delivery, results were similar to main analysis: the effect of maternal low vs high education on rapid growth was mediated by metabolite U6 (m/z 289.2157, NIE OR of maternal low vs high education= 0.77, 95% CI= 0.55-1.00) and metabolite U8 (m/z 269.1894,