ABSTRACT
Aim
To describe cardiometabolic health profiles at age 5–6 years, their correlation with age at adiposity rebound (AR) and associations with maternal hyperglycaemia at 24–28 weeks' gestation (gestational diabetes, fasting (FPG), 1‐hour postload plasma glucose), in mothers without pre‐existing diabetes.
Methods
BMI, %fat mass, blood pressure (BP), FPG, HOMA‐IR and lipids were assessed in children from the EDEN study, a French bicentric birth cohort. Sex‐specific cardiometabolic health profiles were derived using principal component analysis and examined against age at AR with Pearson's correlation. Associations with maternal hyperglycaemia were studied using multiple linear regressions adjusted for parental factors.
Results
Among 674 children, four profiles were identified per sex: ‘higher adiposity, BP, insulin resistance (IR)’; ‘higher BP and lower adiposity’; ‘higher IR and lower adiposity’; ‘higher triglycerides and LDL‐c and lower HDL‐c’. The profile of ‘higher adiposity, BP, IR’ was correlated with earlier age at AR in both sexes. Higher maternal FPG was positively associated with the profile of ‘higher adiposity, BP, IR’ in boys and ‘higher triglycerides and LDL‐c and lower HDL‐c’ in girls.
Conclusion
‘Higher adiposity, BP, IR’ profile at age 5–6 years was associated with earlier age of adiposity rebound. Marginal associations were observed with maternal hyperglycaemia in pregnancy.
Keywords: adiposity rebound, cardiometabolic health, children, gestational diabetes mellitus, gestational glycaemia
Summary.
Previous studies examined maternal glycaemic disorders in relation to unique cardiometabolic outcomes in offspring, overlooking more integrated health profile.
Four cardiometabolic health profiles were identified in girls and boys at age 5–6 years, but maternal hyperglycaemia appeared to have a marginal influence on them.
The examination of cardiometabolic health profiles allows for a better depiction of the cumulative risk associated with the interrelated nature of these factors, which share common physiopathology and aetiology.
Abbreviations
- AR
adiposity rebound
- DASH
Dietary Approach to Stop Hypertension
- DBP
diastolic blood pressure
- EDEN
Etude des Déterminants pré et postnatals du développement et de la santé de l'Enfant
- FPG
fasting plasma glucose
- GDM
gestational diabetes mellitus
- HDL‐c
high‐density lipoproteins
- HOMA‐IR
homeostasis model assessment—insulin resistance
- IOTF
International Obesity Task Force
- LDL‐c
low‐density lipoproteins
- MetS
Metabolic syndrome
- PPG50g‐1hr
1‐hour postprandial plasma glucose after a 50 g glucose load
- SBP
systolic blood pressure
1. Background
Maternal gestational glycaemic disorders are the consequence of an inability to up‐regulate insulin production in response to increasing insulin resistance, leading to glucose overload for the foetus and inducing adverse perinatal outcomes [1]. Maternal hyperglycaemia, even below the diagnostic threshold for gestational diabetes mellitus (GDM), has been associated with adverse foetal development and a more detrimental cardiometabolic health trajectory in the offspring [2, 3]. However, investigations into offspring cardiometabolic risk factors have mostly been conducted separately in relation to maternal glycaemic disorders during pregnancy, neglecting potential intercorrelations. In adults, the constellation of interrelated cardiometabolic risk factors is termed metabolic syndrome (MetS) [4]. Adult MetS is defined by the presence of at least three of five cardiometabolic risk factors (excessive central adiposity, hyperglycaemia, elevated blood pressure, low high‐density lipoprotein cholesterol (HDL‐c), high triglycerides) and is associated with an increased risk of cardiovascular diseases and diabetes [5]. Defining MetS in children is challenging, particularly because cardiometabolic health parameters vary with growth and maturation. Yet, levels of individual cardiometabolic health parameters could track from childhood into adulthood [6]. The most holistic approach to evaluating children's cardiometabolic health is to use a combination (sum or average) of cardiometabolic health parameters with or without considering lifestyle behaviours, to define a cardiometabolic risk score [7, 8]. The limitation of this single score is that it fails to capture distinct heterogenous profiles of increased risk for cardiometabolic diseases. Previous evidence corroborates the existence of multiple cardiometabolic health profiles in children, not exclusively driven by variation in adiposity, suggesting more than one pathophysiology [9, 10, 11, 12]. Until now, children's cardiometabolic health profiles have been largely overlooked, especially in mid‐childhood, with only one study starting from preschool age [9]. In this study, our first objective was to describe cardiometabolic health profiles in 5–6‐year‐old children of European ancestry from a general population sample. We examined cross‐sectionally as well their correlates with the age at adiposity rebound (AR) known as a predictive marker of obesity and other adverse cardiometabolic outcomes in later childhood, adolescence and adulthood [13]. We hypothesised that the cardiometabolic health profiles are sex‐dependent, as boys and girls may have different sensitivity to an unfavourable developmental programming and, consequently, different susceptibilities to diseases [14]. Finally, we investigated associations between maternal gestational glycaemia in mothers without preexisting diabetes and GDM with children's cardiometabolic health profiles. As we acknowledge the potential influence of GDM treatment on our associations, we conducted analyses both with and without GDM mothers.
2. Materials and Methods
2.1. Cohort
Data from the ongoing French birth cohort study, EDEN (« Etude des Déterminants pré et postnatals du développement et de la santé de l'Enfant »), were analysed. The EDEN study comprised 2002 pregnant women recruited between 2003 and 2006 at 24–28 weeks of gestation from two university maternity clinics (Nancy and Poitiers). Exclusion criteria included multiple pregnancies, a known history of diabetes, inability to speak or read French, or any plans to move out of the region within the next 3 years. Additional details about EDEN have been published elsewhere [15]. Written consent was obtained from the mother for herself at inclusion and for her newborn child after delivery. GDM status was obtained from the obstetrical records for 95% (1897/2002) of mothers who continued to be followed after birth. Of these 1897 mothers, 1784 underwent the O'Sullivan test at 24–28 weeks of gestation and remained in the study after birth. Mothers from the Nancy centre also had fasting plasma glucose (FPG) recorded (n = 951). Among the mothers with either retrieved GDM status, measured 1‐hour postload plasma glucose (PPG50‐1hr), or measured FPG, 674 (main study sample), 633 and 275 respectively, had a child with at least one fasting blood parameter measured at the 5‐year follow‐up. The study flow chart is presented in Figure 1.
FIGURE 1.
Flowchart of study population selection. GDM, gestational diabetes mellitus; FPG, fasting plasma glucose; HDL‐c, high‐density lipoprotein cholesterol; LDL‐c, low‐density lipoprotein cholesterol; PPG50‐1hr, 1‐hour postload plasma glucose after a 50 g load of glucose.
The EDEN mother–child cohort study was approved by the ethics committee of Kremlin Bicêtre and by the Data Protection Authority, ‘Commission Nationale de l'Informatique et des Libertés’.
2.2. Outcomes
At the 5‐year child clinical examination, trained investigators measured children's weight, height, bioelectrical impedance and blood pressure following standard protocols. Weight was measured with light clothing using an electronic scale (SECA 888, Hamburg, Germany) and height was measured with a wall‐mounted stadiometer (SECA 208, Hamburg Germany). Using age, sex and weight and height measurements, body mass index z‐scores (Z‐BMI) were derived using International Obesity Task Force references [16].
Skinfold thicknesses were measured three times using the Holtain skinfold calliper (Chasmors Ltd., London, UK) and then averaged. Bioelectrical impedance was performed twice with a single‐frequency impedance analyser (BIA 101, Akern‐RJL, Italy) after a 5‐min rest. The percentage of body fat mass was derived from BIA, skinfold thicknesses, weight and height using the formula of Goran et al. [17]. Given that Goran et al.'s formula was developed with a study population comparable to ours—European ancestry children aged 4–6 years—it was considered more suitable for our study compared with existing fat mass equations.
Systolic and diastolic blood pressures (SBP, DBP) were measured using the COLIN 8800 oscillometer whilst the child was lying down after a 5‐min rest period, with three measurements spaced 2 min apart, and the last two measurements were averaged. During the same clinical examination, blood samples were taken after an overnight fast from the children. Blood glucose, serum insulin, total‐ and HDL‐c, triglycerides were measured, and low‐density lipoprotein cholesterol (LDL‐c) was calculated using the Friedewald‐formula for subjects with triglycerides levels < 4.0 mmol/L. Homeostasis model assessment—insulin resistance (HOMA‐IR) scores were calculated using the following formula HOMA‐IR = (fasting plasma glucose*fasting serum insulin)/22.5. Internal age‐ and height‐specific z‐scores of percent of fat mass were derived, and a quadratic term was added to account for the non‐linear relationship between adiposity and age during the AR. Internal age‐ and height‐specific SBP and DBP z‐scores were derived as well. Internal age‐specific z‐scores for blood measurements were calculated. All internal z‐scores were derived separately for each sex and further standardised on hospital centre to account for inter‐centre variability. Internal z‐scores corresponds to the standardised residuals obtained after regressing the outcomes on the covariates of interest, which include, age, centre, and, when applicable, height.
AR corresponds to the second rise in children's BMI curve occurring around 5–6 years of age. Individual BMI curves were modelled with mixed‐effects cubic models, using longitudinal data on the weight and height of children aged 18 months to 13 years, with the detailed modelling method published elsewhere [18].
2.3. Exposure
At the baseline clinical examination at 24–28 weeks of gestation, PPG50‐1hr was measured after a 50 g glucose load (O'Sullivan test) in mothers who had fasted overnight. In the Nancy center, FPG was also measured prior to the O'Sullivan test. Diagnosis of GDM was retrieved from obstetrical records. At the time of study, French hospitals diagnosed GDM using a universal screening at 24–28 weeks' gestation and from the start of pregnancy for women at risk (age ≥ 25 years, overweight or obese status, previous history of GDM or macrosomia or stillbirth, family history of diabetes, non‐European ancestry ethnicity), employing a two‐step process with the Carpenter and Coustan criteria [19].
2.4. Covariates
Maternal characteristics (age at delivery, educational attainment, parity, consumption or non‐consumption of alcohol and tobacco during the first trimester) and household income were collected during a face‐to‐face interview at baseline. Maternal pre‐pregnancy weight as well as paternal weight and height were reported, whilst maternal height was measured. Gestational age at delivery was determined based on the date of the last menstrual period or early standard ultrasound foetal measurement, according to clinician judgement.
Maternal diet before pregnancy was retrospectively assessed using a semi‐quantitative, validated Food Frequency Questionnaire (FFQ) completed at 24–28 weeks of gestation. The quality of maternal diet before pregnancy was evaluated by calculating an adherence score to the Dietary Approach to Stop Hypertension (DASH) diet, with a higher score indicating healthier dietary quality. Detailed methods are published elsewhere [20].
Maternal first‐trimester physical activity level was assessed at 24–28 weeks of gestation. Mothers reported the frequency and duration of physical activity of varying intensities from three domains: Occupational, Sports and Leisure‐time. These domains were summed into a total physical activity score, with a higher score indicating a more active individual.
2.5. Statistical Analysis
Participants sociodemographic characteristics and cardiometabolic health outcomes were described using means and frequencies. Distributions of the exposure and outcome variables were examined through histograms, and the linearity of the studied associations was assessed graphically. Variables with non‐Gaussian distribution, such as HOMA‐IR and triglycerides were log‐transformed.
Principal component analysis (PCA) was separately performed for each sex to identify cardiometabolic health profiles among children (N = 372 and 302, for boys and girls respectively). Variables included in the PCA were as follows: BMI, fat mass, SBP, DBP, FPG, HOMA‐IR, triglycerides, HDL‐c and LDL‐c z‐scores. The number of PCA components was determined based on a minimal eigenvalue of 1.0. Only variables with a profile loading greater than or equal to 0.30 were considered in the profile interpretation. However, labelling was based on the variables that contributed the most to the profile and shared features across sexes. Correlates between cardiometabolic health profiles adherence scores and age at AR were examined using Pearson's correlation. For each individual, a higher score on a profile corresponds to greater adherence to it. To relate the PCA identified profiles to the native variables, the cardiometabolic health outcomes of individuals in the highest quintiles of each profile adherence score were detailed in the Supporting Information (Table S1).
Associations between maternal glycaemia (GDM, PPG50g‐1hr, FPG (Nancy mothers only)) and children's cardiometabolic health profiles adherence scores were analysed using linear regressions, both unadjusted and adjusted for maternal characteristics (age, pre‐pregnancy BMI, parity, pre‐pregnancy DASH score, 1st‐trimester physical activity score, tobacco and alcohol consumption during pregnancy, educational attainment), paternal characteristics (educational attainment, BMI), study site (Poitiers, Nancy) and household income. The optimal set of adjustments required to examine the total effect was determined based on our narrative review of studies investigating the same association and our expertise in perinatal health, both integrated into a Directed Acyclic Graph (DAG) (Figure S1). Although paternal characteristics are not confounders per se, they were included as precision variables in our final adjustments. The study site was adjusted for to account for variations that may arise from differences between the participating centres. Our study sample has a homogeneous ethnic background, with 97% of participants of European ancestry. Ethnicity was therefore unlikely to have a major confounding influence on the associations studied and was not included in the adjustments.
For the adjusted associations, missing covariates were imputed using Markov Chain Monte Carlo multiple imputation under the missing‐at‐random assumption (n = 20). Detailed information on the imputation methods used and the relative efficiency of the imputation are shown in Table S2. Imputation of missing data on cardiometabolic outcomes in all children with at least one blood parameter measured were performed using the ‘missMDA’ package on RStudio [21]. This package relies on a regularised iterative PCA algorithm to impute missing data. Briefly, a first PCA is performed in which missing entries are imputed with an initial value (mean of the variable). The fitted PCA is then used to predict a new value for the missing one. These two steps are repeated until convergence. Evaluation of the robustness of the imputation performed using the regularised iterative PCA algorithm is presented in Figure S2. All analyses were performed using RStudio software (version 3.4.3; R Core Team (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R‐project.org/).
As a sensitivity analysis, PCA was run on children with non‐missing cardiometabolic outcomes. A complete‐cases (with non‐missing covariates) analysis was conducted as well to challenge the robustness of our associations. Furthermore, given that all GDM mothers were treated after diagnosis, we accounted for the GDM treatment by removing GDM mothers from our models. Finally, analyses of the associations with GDM and PPG50g‐1hr were repeated in the sample with FPG to ensure that differences in findings were not due to variations among the included participants.
3. Results
Participants sociodemographic characteristics and cardiometabolic outcomes were described for the main study sample (n = 674) in Tables 1 and 2. Subsamples sociodemographic characteristics are detailed in the Supporting Information (Table S3). Overall, participants' characteristics were similar across study samples (Tables 1 and S1) except that mothers with measured FPG had relatively higher socio‐economic status and were less likely to consume alcohol during pregnancy. Mothers with overweight represented about a quarter of the included ones. Mothers delivered on average at 30 years of age (± 5), 5% had preterm delivery, 7% had GDM with more than half of them being treated with insulin, ~59% had at least a bachelor's degree, and 55% had a household income greater than €2300/month (Table 1). Compared with non‐GDM mothers, those with GDM were older (32 years of age (±5)), more likely to live with overweight or obesity, adhered to a healthier diet (DASH diet score), exhibited higher PPG50g‐1hr and FPG levels, and experienced lower weight gain across all trimesters of pregnancy (Table S3). In comparison, the non‐included mothers had a more disadvantageous socio‐economic background and lifestyle than the included mothers. At 5 years, 4% of the boys were living with overweight (IOTF definition) compared with 10% in girls (Table 2). Girls had more deleterious cardiometabolic health parameters than boys, except for HDL‐c and age at AR.
TABLE 1.
Characteristics of included and non‐included participants, France (Nancy & Poitiers), 2003–2006, EDEN study.
| n | Missing data (%) | Included mean (SD) no. (%) | n | Non‐included mean (SD) no. (%) | |
|---|---|---|---|---|---|
| Maternal characteristics | |||||
| Age at delivery, years | 674 | 0 | 30 (5) | 1233 | 29 (5) |
| Primiparous, % | 674 | 0 | 290 (43) | 1229 | 558 (45) |
| Educational attainment, % | 671 | < 1 | 1239 | ||
| No formal education/primary/secondary | 163 (24) | 386 (31) | |||
| Postsecondary | 114 (17) | 226 (18) | |||
| Undergraduate | 165 (25) | 249 (20) | |||
| Graduate/postgraduate | 229 (34) | 378 (31) | |||
| Monthly household income level, % | 670 | < 1 | 1243 | ||
| ≤ €1500 | 78 (12) | 249 (20) | |||
| €1501–€2300 | 222 (33) | 346 (28) | |||
| €2301–€3000 | 202 (30) | 299 (24) | |||
| ≥ €3001 | 168 (25) | 349 (28) | |||
| Pre‐pregnancy DASH score (range: 8–40) a | 672 | < 1 | 24.5 (4.1) | 1292 | 23.7 (4.3) |
| First‐trimester physical activity score (range: 3–15) | 660 | 2 | 5.7 (1.3) | 1218 | 5.5 (1.3) |
| Glycaemic parameters | 0 | ||||
| GDM, % | 674 | 44 (7) | 1230 | 79 (6) | |
| Treated by insulin, % | 44 | 25 (57) | 79 | 28 (35) | |
| PPG50g‐1hr at 24–28 weeks' gestation, mmol/L | 633 | 6 | 6.4 (1.4) | 1171 | 6.3 (1.4) |
| FPG at 24–28 weeks' gestation, mmol/L | 275 | 59 | 4.2 (0.4) | 689 | 4.2 (0.4) |
| Pre‐pregnancy BMI, % | 659 | 2 | 1225 | ||
| Underweight | 44 (7) | 117 (10) | |||
| Normal weight | 422 (64) | 805 (66) | |||
| Overweight | 132 (20) | 198 (16) | |||
| Obese | 61 (9) | 105 (9) | |||
| Tobacco consumer during pregnancy, % | 658 | 2 | 133 (20) | 1189 | 351 (30) |
| Alcohol consumer during pregnancy, % | 671 | < 1 | 332 (49) | 1247 | 486 (39) |
| Paternal characteristics | |||||
| Educational attainment | 622 | 8 | 1123 | ||
| No formal education/primary/secondary | 212 (34) | 424 (38) | |||
| Postsecondary | 124 (20) | 207 (18) | |||
| Undergraduate | 140 (23) | 210 (18) | |||
| Graduate/postgraduate | 146 (23) | 282 (25) | |||
| Overweight, % | 629 | 7 | 247 (39) | 1154 | 397 (34) |
| Obese, % | 629 | 7 | 59 (9) | 1154 | 93 (8) |
| Child characteristics at birth | |||||
| Boy, % | 674 | 0 | 372 (55) | 1229 | 628 (51) |
| Preterm birth, % | 674 | 0 | 37 (5) | 1231 | 73 (6) |
| Birth weight, g | 674 | 0 | 3281 (526) | 1225 | 3277 (505) |
Abbreviations: BMI, body mass index; DASH, dietary approaches to stop hypertension; FPG, fasting plasma glucose; GDM, gestational diabetes mellitus; PPG50‐1hr, 1‐hour postload plasma glucose after a 50 g load of glucose.
DASH score was previously derived [20].
TABLE 2.
Description of children's cardiometabolic parameters at 5–6 years by sex, France (Nancy & Poitiers), 2003–2006, EDEN study.
| n | Boys | n | Girls | |
|---|---|---|---|---|
| Age at clinical examination, years | 372 | 5.7 (0.1) | 302 | 5.6 (0.1) |
| Overweight, a n (%) | 372 | 16 (4) | 301 | 31 (10) |
| BMI IOTF z‐scores, SD | 372 | −0.13 (0.84) | 301 | 0.08 (0.90) |
| BMI, kg/m2 | 372 | 15.29 (1.2) | 301 | 15.43 (1.5) |
| Fat mass, % | 363 | 12 (3) | 292 | 17 (3) |
| Age at AR, years | 366 | 5.5 (1.4) | 288 | 5.4 (1.4) |
| SBP, mmHg | 372 | 100 (8) | 301 | 102 (9) |
| DBP, mmHg | 372 | 52 (9) | 301 | 54 (9) |
| FPG, mmol/L | 372 | 4.5 (0.4) | 302 | 4.4 (0.4) |
| HOMA‐IR | 342 | 2.2 (2.2) | 273 | 2.5 (1.7) |
| Triglycerides, mmol/L | 370 | 0.53 (0.19) | 301 | 0.58 (0.20) |
| HDL‐c, mmol/L | 371 | 1.51 (0.32) | 301 | 1.43 (0.31) |
| LDL‐c, mmol/L | 370 | 2.52 (0.60) | 301 | 2.73 (0.67) |
Note: Values are means (SD) or No (%).
Abbreviations: AR, adiposity rebound; BMI, body mass index; DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDL‐c, high‐density lipoprotein cholesterol; HOMA‐IR, homeostasis model assessment—insulin resistance; LDL‐c, low‐density lipoprotein cholesterol; SBP, systolic blood pressure; ST, skinfold thickness.
Children were categorised as overweight against IOTF definition [16].
Four cardiometabolic health profiles were identified in boys and girls, explaining 67% and 69% of the total variance, respectively. The loadings of each cardiometabolic parameter in each profile are detailed in Table 3. Overall, the features driving these profiles were similar in both sexes. The first profile, was characterised by ‘Higher adiposity, BP, insulin resistance (IR)’, explained 23% and 25% of the total variance in boys and girls, respectively. However, girls also exhibited greater levels of triglycerides. The second profile observed in boys and the third profile in girls displayed ‘Higher IR & lower adiposity’, accounting for 17% and 15% of the total variance, respectively. Additionally, boys showed higher levels of triglycerides and lower levels of HDL‐c. The third profile in boys and second profile in girls both displayed ‘Higher triglycerides, LDL‐c & lower HDL‐c’, explaining 14% and 17% of the total variance, respectively. Notably, girls showed lower BP, whilst boys exhibited lower FPG. Finally, the fourth profile in both sexes was characterised by ‘Higher BP & lower adiposity’, explaining 13% and 12% of the total variance, in boys and girls, respectively. Results in children with non‐missing cardiometabolic outcomes led to similar profiles (Table S4). The correlations between cardiometabolic health profiles adherence scores and age at AR are displayed in Table 3. Higher age at AR was correlated in both sexes with a lower adherence score to the profile ‘Higher adiposity, BP, IR’ and in girls only, with a higher profile score to the profile ‘Higher BP and lower adiposity’.
TABLE 3.
Cardiometabolic health parameters factor loadings in each identified profile (obtained from principal component analysis) (n = 674) a and correlates between cardiometabolic health profiles and age at AR (n = 653) b .
| Boys (n = 372) | Girls (n = 302) | |||||||
|---|---|---|---|---|---|---|---|---|
| “Higher adiposity, BP, IR” | “Higher IR and lower adiposity” | “Higher triglycerides, LDL‐c and lower HDL‐c” | “Higher BP and lower adiposity” | “Higher adiposity, BP, IR” | “Higher triglycerides, LDL‐c and lower HDL‐c” | “Higher IR and lower adiposity” | “Higher BP and lower adiposity” | |
| BMI | 0.76 | −0.36 | 0.10 | −0.30 | 0.73 | −0.02 | −0.18 | −0.54 |
| % Fat mass | 0.69 | −0.43 | 0.15 | −0.33 | 0.77 | 0.09 | −0.30 | −0.38 |
| SBP | 0.47 | −0.17 | 0.21 | 0.60 | 0.52 | −0.40 | −0.31 | 0.43 |
| DBP | 0.31 | −0.02 | 0.23 | 0.73 | 0.50 | −0.35 | −0.22 | 0.57 |
| HOMA‐IR | 0.60 | 0.64 | −0.22 | −0.03 | 0.54 | −0.09 | 0.66 | 0.06 |
| FPG | 0.54 | 0.62 | −0.36 | −0.06 | 0.37 | −0.13 | 0.71 | 0.03 |
| Triglycerides | −0.04 | 0.45 | 0.65 | 0.01 | 0.37 | 0.74 | 0.09 | 0.21 |
| HDL‐c | 0.09 | −0.41 | −0.57 | 0.24 | −0.15 | −0.64 | −0.13 | −0.27 |
| LDL‐c | 0.10 | −0.06 | 0.49 | −0.22 | 0.11 | 0.48 | −0.37 | 0.17 |
| % Explained variance | 23 | 17 | 14 | 13 | 25 | 17 | 15 | 12 |
| Boys (n = 365) | Girls (n = 288) | |||||||
|---|---|---|---|---|---|---|---|---|
| Correlation with age at AR | −0.47 c | 0.04 | 0.02 | −0.00 | −0.23 c | −0.08 | −0.02 | 0.22 c |
Abbreviations: AR, adiposity rebound; BMI, body mass index; BP, blood pressure; DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDL‐c, high‐density lipoprotein cholesterol; HOMA‐IR, homeostasis model assessment—insulin resistance; IR, insulin resistance; LDL‐c, low density lipoprotein cholesterol; SBP, systolic blood pressure.
All cardiometabolic health parameters introduced in the principal component analysis are z‐scores. Sex‐ and age‐specific BMI z‐scores were derived against IOTF references. All other cardiometabolic health parameters were standardised on sex, centre, and age. % fat mass was additionally standardised on height and age [2]. Blood pressure measurements were additionally standardised on height.
Correlation with age at AR was assessed using Pearson's correlation.
Two‐sided p < 0.05.
Results regarding the association between maternal glycaemia and child cardiometabolic health profiles adherence scores are presented in Table 4 for boys and Table 5 for girls. GDM and PPG50g‐1hr were not associated with any cardiometabolic health profiles. Higher FPG was associated with higher adherence scores to the ‘Higher adiposity, BP, IR’ profile in boys and to the ‘Higher triglycerides, LDL‐c & lower HDL‐c’ profile in girls. No further association was observed with maternal FPG. After the exclusion of GDM mothers, all association remained unchanged. Similarly, complete‐cases analysis and in FPG subsample yielded the same results.
TABLE 4.
Association between maternal glycaemia during pregnancy and children's cardiometabolic health profiles in boys, β [95% CI] a , France (Nancy & Poitiers), 2003–2006, EDEN study.
| n | “Higher adiposity, BP, IR” | “Higher IR and lower adiposity” | “Higher triglycerides, LDL‐c and lower HDL‐c” | “Higher BP and lower adiposity” | |
|---|---|---|---|---|---|
| GDM | |||||
| Unadjusted | 372 | 0.13 (−0.52, 0.78) | 0.08 (−0.48, 0.64) | 0.04 (−0.48, 0.55) | 0.06 (−0.43, 0.56) |
| Adjusted | 372 | −0.18 (−0.83, 0.47) | 0.15 (−0.43, 0.72) | 0.08 (−0.47, 0.63) | 0.20 (−0.33, 0.73) |
| Adjusted, CC | 311 | 0.27 (0.48, 1.02) | 0.38 (−0.28, 1.04) | 0.18 (−0.44, 0.80) | 0.13 (−0.48, 0.75) |
| Adjusted, FPG subsample | 147 | −0.10 (−1.11, 0.92) | 0.23 (−0.63, 1.10) | 0.07 (−0.79, 0.93) | 0.05 (−0.72, 0.83) |
| PPG50‐1hr, mmol/L | |||||
| Unadjusted | 351 | 0.08 (−0.03, 0.18) | 0.04 (−0.05, 0.13) | −0.02 (−0.10, 0.07) | −0.07 (−0.16, 0.01) |
| Adjusted | 351 | 0.03 (−0.08, 0.13) | 0.04 (−0.06, 0.13) | −0.02 (−0.11, 0.07) | −0.06 (−0.14, 0.03) |
| Adjusted, CC | 296 | 0.04 (−0.08, 0.15) | 0.06 (−0.04, 0.17) | −0.03 (−0.13, 0.07) | −0.02 (−0.12, 0.07) |
| Adjusted, non‐GDM | 333 | 0.03 (−0.08, 0.15) | 0.03 (−0.08, 0.13) | −0.03 (−0.13, 0.07) | −0.07 (−0.16, 0.02) |
| Adjusted, FPG subsample | 147 | 0.02 (−0.16, 0.19) | 0.04 (−0.12, 0.20 | 0.08 (−0.07, 0.22) | −0.14 (−0.27, −0.01) |
| FPG, mmol/L | |||||
| Unadjusted | 147 | 0.83 (0.21, 1.45) b | 0.42 (−0.13, 0.96) | −0.23 (−0.76, 0.31) | 0.20 (−0.27, 0.66) |
| Adjusted | 147 | 0.75 (0.03, 1.48) b | 0.49 (−0.13, 1.11) | −0.36 (−0.98, 0.27) | 0.24 (−0.33, 0.82) |
| Adjusted, CC | 122 | 0.77 (−0.01, 1.55) | 0.64 (−0.07, 1.35) | −0.13 (−0.83, 0.57) | 0.13 (−0.52, 0.77) |
| Adjusted, non‐GDM | 138 | 0.84 (0.02, 1.66) b | 0.66 (−0.04, 1.36) | −0.39 (−1.11, 0.32) | 0.43 (−0.20, 1.06) |
Abbreviations: BP, blood pressure; CC, complete cases; FPG, fasting plasma glucose; GDM, gestational diabetes mellitus; IR, insulin resistance; PPG50g‐1hr, 1‐hour postload plasma glucose after a 50 g glucose load.
Models were imputed and adjusted for maternal characteristics (age, pre‐pregnancy BMI, parity, pre‐pregnancy DASH score, 1st trimester physical activity score, tobacco and alcohol consumption during pregnancy, educational attainment), paternal characteristics (educational attainment, BMI) and household income.
Two‐sided p < 0.05.
TABLE 5.
Association between maternal glycaemia during pregnancy and children's cardiometabolic health profiles in girls, β [95% CI] a , France (Nancy & Poitiers), 2003–2006, EDEN study.
| n | “Higher adiposity, BP, IR” | “Higher triglycerides, LDL‐c and lower HDL‐c” | “Higher IR and lower adiposity” | “Higher BP and lower adiposity” | |
|---|---|---|---|---|---|
| GDM | |||||
| Unadjusted | 302 | −0.11 (−0.74, 0.52) | 0.14 (−0.38, 0.65) | −0.46 (−0.94, 0.03) | 0.00 (−0.44, 0.44) |
| Adjusted | 302 | −0.55 (−1.18, 0.08) | 0.28 (−0.27, 0.82) | −0.50 (−1.01, 0.01) | 0.10 (−0.36, 0.55) |
| Adjusted, CC | 257 | −0.38 (−1.04, 0.29) | 0.05 (−0.52, 0.62) | −0.45 (−1.01, 0.11) | 0.03 (−0.46, 0.51) |
| Adjusted, FPG subsample | 128 | −0.52 (−1.59, 0.55) | 0.14 (−0.74, 1.01) | 0.00 (−0.84, 0.83) | −0.13 (−0.86, 0.61) |
| PPG50‐1hr, mmol/L | |||||
| Unadjusted | 282 | −0.04 (−0.17, 0.08) | 0.04 (−0.06, 0.14) | 0.00 (−0.10, 0.10) | 0.06 (−0.03, 0.14) |
| Adjusted | 282 | −0.07 (−0.20, 0.06) | 0.06 (−0.06, 0.17) | −0.02 (−0.13, 0.09) | 0.06 (−0.03, 0.16) |
| Adjusted, CC | 240 | −0.04 (−0.19, 0.10) | 0.05 (−0.07, 0.17) | 0.01 (−0.11, 0.13) | 0.06 (−0.04, 0.16) |
| Adjusted, non‐GDM | 264 | −0.03 (−0.18, 0.11) | 0.06 (−0.06, 0.19) | 0.01 (−0.11, 0.12) | 0.06 (−0.04, 0.17) |
| Adjusted, FPG subsample | 128 | −0.25 (−0.45, −0.04) b | −0.02 (−0.19, 0.16) | 0.02 (−0.16, 0.19) | 0.00 (−0.15, 0.14) |
| FPG, mmol/L | |||||
| Unadjusted | 128 | −0.25 (−0.86, 0.37) | 0.24 (−0.23, 0.71) | 0.26 (−0.19, 0.71) | −0.03 (−0.44, 0.39) |
| Adjusted | 128 | −0.46 (−1.13, 0.20) | 0.53 (0.00, 1.06) b | 0.23 (−0.29, 0.75) | 0.09 (−0.37, 0.55) |
| Adjusted, CC | 113 | −0.46 (−1.19, 0.26) | 0.49 (−0.05, 1.03) | 0.31 (−0.25, 0.86) | 0.02 (−0.48, 0.51) |
| Adjusted, non‐GDM | 119 | −0.51 (−1.36, 0.33) | 0.73 (0.06, 1.41) b | 0.16 (−0.50, 0.81) | 0.11 (−0.46, 0.67) |
Abbreviations: BP, blood pressure; CC, complete‐cases; FPG, fasting plasma glucose; GDM, Gestational diabetes mellitus; IR, insulin resistance; PPG50g‐1hr, 1‐hour postload plasma glucose after a 50 g glucose load.
Models were imputed and adjusted for maternal characteristics (age, pre‐pregnancy BMI, parity, pre‐pregnancy DASH score, 1st trimester physical activity score, tobacco and alcohol consumption during pregnancy, educational attainment), paternal characteristics (educational attainment, BMI) and household income.
Two‐sided p < 0.05.
4. Discussion
In the present study, four cardiometabolic health profiles were observed in preschool girls and boys. In both sexes, one profile exhibited greater overall cardiometabolic risk, characterised by an excess of adiposity, higher BP and greater insulin resistance. A greater adherence to this profile was correlated with an earlier age at AR, known to be a predictor of later risk of adverse cardiometabolic outcomes. Marginal associations were observed with maternal FPG during pregnancy and none with GDM or PPG50g‐1hr.
In comparison with cardiometabolic health profiles observed among older children and adolescents, we consistently found a profile characterised by an excess of adiposity and greater insulin resistance [9, 10, 11, 12], and another characterised by an unhealthier lipid profile [10, 11, 12]. Apart from these two profiles, the remaining identified profiles in the literature displayed significant variations. Potential reasons for this discrepancy may stem from variations across studies in the population age and ethnicity, available cardiometabolic health parameters, and methodological considerations such as stratification by sex. Unfortunately, very few studies investigating cardiometabolic health profiles have stratified by sex despite growing evidence suggesting that boys and girls have different susceptibilities to diseases, explained by sexual dimorphism in systems regulating energy homeostasis and sexually dimorphic responses to early life programming [14]. Consistent with Ramachandran et al., that stratified by sex and used a similar set of cardiometabolic health parameters, the contribution of lipid levels to the identified profiles differed by sex [12]. From our results, in boys and girls aged 5–6 years, a cardiometabolic health profile, combining higher adiposity, BP and IR was associated with an earlier age at AR, a well‐recognised marker of later cardiometabolic health [13].
Our findings indicate that stronger adherence to the profile characterised by ‘Higher adiposity, BP, insulin resistance (IR)’ in boys and to the profile ‘Higher triglycerides, LDL‐c and lower HDL‐c’ in girls was associated with higher maternal FPG during pregnancy. However, these scores were not related to either GDM or PPG50g‐1hr. According to the literature, offspring of mothers with GDM were at greater risk of an increased adiposity and glycaemic disorders, even after adjusting for maternal BMI, a major confounder [22]. In a population with a greater prevalence of GDM, we did not observe any association between GDM and the offspring's cardiometabolic health. This lack of association could be potentially explained by GDM treatment, allowing better glycaemic control, which might dilute the associations [23]. From our results, half of GDM mothers were treated with insulin and they had in average a healthier diet during pregnancy and hence, lower gestational weight gain across all trimesters. After excluding GDM mothers, the strength of the positive association between maternal FPG and the adherence score to the profile ‘Higher triglycerides, LDL‐c and lower HDL‐c’ in girls increased by 38%, whilst the magnitude of the positive association between maternal FPG and the adherence score to the profile ‘Higher adiposity, BP, IR’ in boys increased by 12%. This might suggest that GDM treatment has more or less influence on some cardiometabolic health profiles in children of GDM mothers and that there may be a residual risk nonetheless. Compared with GDM or PPG50g‐1hr, FPG had received less attention once studying the detrimental effect of maternal gestational hyperglycaemia on offspring health outcomes. This is explained by controversies about FPG fluctuations throughout pregnancy and its poor predictive value by itself in GDM diagnosis [24]. However, evidence shows that higher FPG during pregnancy could rather be a surrogate marker of an overall more adverse cardiometabolic health profile during pregnancy [25]. To explain the observed sex‐differential association between gestational glycaemia and children cardiometabolic health, evidence suggests that there is a sex‐specific programming of prenatal exposures on later health [26]. More precisely, boy foetuses could be more sensitive to a hyperglycaemic in utero environment [26, 27, 28]. Yet, sex differences are poorly investigated regarding the association between gestational hyperglycaemia and children cardiometabolic outcomes which restrains our ability to compare our results.
Our findings need to be interpreted cautiously. The study population is highly educated and composed mostly of individuals with homogenous ancestry. This may limit the generalisability of our findings to other populations. Compared with the most commonly used approach to investigate children's overall cardiometabolic health, that is, cardiometabolic risk scores (a sum of all levels of cardiometabolic health parameter), studying profiles has one major limitation. Cardiometabolic health profiles are, by construction, data‐driven and hence depend on the characteristics of the studied population (age, sex, ethnicity, lifestyle, etc.). We also acknowledge that the prevalence of overweight children at age 5 years in the EDEN cohort study is relatively low, especially in boys (4%) which could also hinder our likelihood of observing a substantial association with maternal hyperglycaemia. Yet, based on the American Heart Association's cardiovascular health metrics (ideal body mass index, physical activity, diet, blood pressure, cholesterol and glucose levels and exposure to passive smoking), only 34% of children aged 5–6 years had ideal cardiovascular health in the EDEN cohort study [29]. In this study, adiposity was assessed using BMI z‐scores and % fat mass, with the latter calculated from BIA outputs and skinfold thickness measurements using the Goran et al. formula (R 2 = 0.91 compared with DXA estimation) [17]. Although these may not be the most accurate measures, when taken together, they provide a reasonable proxy that does not significantly affect the validity of our findings.
Our study demonstrated several strengths. The examination of cardiometabolic risk in childhood often focuses on isolated health parameters, such as excessive adiposity, which leads to overlooking the cumulative risk associated with the interrelated nature of these factors. A widely used approach to assess this cumulative risk is to calculate a cardiometabolic risk score that combines the levels of all the cardiometabolic health parameters [7]. However, children who display different combinations of cardiometabolic health parameters can share the same score, hence the same level of risk, which is a strong assumption. The profile approach depicts more granular information on clusters of cardiometabolic health parameters that share common physiopathology and aetiology. PCA is a valuable tool for identifying profiles of interrelated risk factors within a population whilst retaining most of the variance present in the data. In comparison with other clustering methods, recent advancements in PCA allow for the handling of missing data, which helps to minimise selection bias [21]. Studying similarities and dissimilarities in children's cardiometabolic health profiles across different populations should, therefore, represent an opportunity to identify prevention strategies from early ages by distinguishing context‐specific variations from universal trends. Finally, unlike our study, previous studies examining the associations between maternal hyperglycaemia and offspring cardiometabolic outcomes adjusted for mediators (such as gestational age or birthweight) and did not adjust for other known confounders, such as parental socio‐economic status and maternal lifestyle [2, 30, 31]. This could lead to biased findings in the existing literature resulting from a greater level of residual confounding and a collider bias [32].
5. Conclusions
Multiple cardiometabolic health profiles were observed since preschool age, including one that suggests a potentially elevated risk for future cardiometabolic disease. Children's cardiometabolic outcomes primarily vary at subclinical levels, forming together a risk continuum that provides insights into future health. Analysing these factors in isolation offers limited clinical value, highlighting the need for a collective assessment. Our findings emphasise that strategies for preventing childhood cardiometabolic health issues should address a broader range of risk factors rather than relying mainly on adiposity screening. Marginal associations were observed with maternal hyperglycaemia. This study contributes to document the influence of perinatal health on cardiometabolic health programming. Yet, future replication of our study in other populations at different developmental stages and with different environment, ethnic and cultural backgrounds is warranted to investigate shared aetiology of children's cardiometabolic health profiles across populations.
Author Contributions
Wen Lun Yuan: conceptualization, investigation, writing – original draft, writing – review and editing, formal analysis, data curation. Aminata H. Cissé: methodology, data curation, resources, writing – review and editing. Muriel Tafflet: methodology, writing – review and editing, data curation, resources. Elissa El Khoury: investigation, formal analysis, data curation, writing – review and editing. Olfa Khalfallah: resources, writing – review and editing. Laetitia Davidovic: writing – review and editing, resources. Marie Aline Charles: funding acquisition, conceptualization, writing – review and editing, project administration. Barbara Heude: conceptualization, writing – review and editing, methodology, validation, project administration, supervision, resources.
Ethics Statement
The EDEN mother–child cohort study was approved by the ethics committee of Kremlin Bicêtre (CPP reference 02‐70, December 2002; CCTIRS reference 02‐189, July 2002) and by the Data Protection Authority, ‘Commission Nationale de l'Informatique et des Libertés’ (CNIL reference 902267, December 2002).
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Figure S1. Graphical representation of the hypothetical causal relationships between the variables included in the fully adjusted model, with the following colour‐coded groups: red circle for confounders, white circle for precision variables, empty blue circle for mediators, green for exposure and non‐empty blue circle for the outcome. Arrows represent the hypothetical direction of influence between variables.
Figure S2. Illustration of the robustness of the missMDA imputation used.
Table S1. Cardiometabolic health outcomes of individuals in the highest quintiles of each identified pattern.
Table S2. Imputation method used for cardiometabolic health parameters and covariates.
Table S3. Characteristics of participants with evaluated PPG50g‐1hr, FPG, with GDM and non‐GDM.
Table S4. Description of cardiometabolic health patterns (children with non‐missing cardiometabolic outcomes) (n = 594).
Acknowledgements
We would like to thank Gilles Paradis for his valuable input in reviewing this work. We are indebted to the participating families, the midwife research assistants (Lorraine Douhaud, Sophie Bedel, Brigitte Lortholary, Sophie Gabriel, Muriel Rogeon and Monique Malinbaum) for data collection, and Patricia Lavoine, Josiane Sahuquillo and Ginette Debotte for checking, coding and data entry. The EDEN mother–child cohort study group includes I. Annesi‐Maesano, J.Y. Bernard, J. Botton, M.A. Charles, P. Dargent‐Molina, B. de Lauzon‐Guillain, P. Ducimetière, M. de Agostini, B. Foliguet, A. Forhan, X. Fritel, A. Germa, V. Goua, R. Hankard, B. Heude, M. Kaminski, B. Larroque, N. Lelong, J. Lepeule, G. Magnin, L. Marchand, C. Nabet, F. Pierre, R. Slama, M.J. SaurelCubizolles, M. Schweitzer and O. Thiebaugeorges.
Funding: This work was supported by LONGITOOLS, a project funded by the European Union's Horizon 2020 research and innovation programme [grant number 874739]. WLY received postdoctoral fellowship from Fondation de France. The EDEN cohort study is supported by Foundation for Medical Research (FRM), National Agency for Research (ANR), National Institute for Research in Public Health (IRESP: TGIR cohorte santé 2008 program), French Ministry of Health (DGS), French Ministry of Research, Inserm Bone and Joint Diseases National Research (PRO‐A) and Human Nutrition National Research Programmes, Paris–Sud University, Nestlé, French National Institute for Population Health Surveillance (InVS), French National Institute for Health Education (INPES), the European Union FP7 programs (FP7/2007–2013, HELIX, ESCAPE, ENRIECO, Medall projects), Diabetes National Research Program (through a collaboration with the French Association of Diabetic Patients (AFD)), French Agency for Environmental Health Safety (now ANSES), Mutuelle Générale de l'Education Nationale (MGEN), French National Agency for Food Security and the French‐speaking association for the study of diabetes and metabolism (ALFEDIAM).
Data Availability Statement
The data underlying the findings cannot be made freely available for ethical and legal restrictions imposed, because this study includes a substantial number of variables that together, could be used to re‐identify the participants based on a few key characteristics and then be used to have access to other personal data. Therefore, the French ethics authority strictly forbids making these data freely available. However, they can be obtained upon request from the EDEN principal investigator. Readers may contact barbara.heude@inserm.fr to request the data. The analytic code will be made available upon request pending application and approval.
References
- 1. Metzger B. E., Buchanan T. A., Coustan D. R., et al., “Summary and Recommendations of the Fifth International Workshop‐Conference on Gestational Diabetes Mellitus,” Diabetes Care 30, no. Suppl 2 (2007): S251–S260. [DOI] [PubMed] [Google Scholar]
- 2. Tam W. H., Ma R. C. W., Ozaki R., et al., “In Utero Exposure to Maternal Hyperglycemia Increases Childhood Cardiometabolic Risk in Offspring,” Diabetes Care 40, no. 5 (2017): 679–686. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Ylostalo T., Saha M. T., Nummi T., Harjunmaa U., Salo M. K., and Vuorela N., “Maternal Weight, Smoking, and Diabetes Provided Early Predictors of Longitudinal Body Mass Index Growth Patterns in Childhood,” Acta Paediatrica 113, no. 5 (2024): 1076–1086. [DOI] [PubMed] [Google Scholar]
- 4. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) Final Report (2002). [PubMed]
- 5. Malik S., Wong N. D., Franklin S. S., et al., “Impact of the Metabolic Syndrome on Mortality From Coronary Heart Disease, Cardiovascular Disease, and All Causes in United States Adults,” Circulation 110, no. 10 (2004): 1245–1250. [DOI] [PubMed] [Google Scholar]
- 6. Camhi S. M. and Katzmarzyk P. T., “Tracking of Cardiometabolic Risk Factor Clustering From Childhood to Adulthood,” International Journal of Pediatric Obesity 5, no. 2 (2010): 122–129. [DOI] [PubMed] [Google Scholar]
- 7. Kamel M., Smith B. T., Wahi G., Carsley S., Birken C. S., and Anderson L. N., “Continuous Cardiometabolic Risk Score Definitions in Early Childhood: A Scoping Review,” Obesity Reviews 19, no. 12 (2018): 1688–1699. [DOI] [PubMed] [Google Scholar]
- 8. Lloyd‐Jones D. M., Allen N. B., Anderson C. A. M., et al., “Life's Essential 8: Updating and Enhancing the American Heart Association's Construct of Cardiovascular Health: A Presidential Advisory From the American Heart Association,” Circulation 146, no. 5 (2022): e18–e43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Chen W., Srinivasan S. R., Elkasabany A., and Berenson G. S., “Cardiovascular Risk Factors Clustering Features of Insulin Resistance Syndrome (Syndrome X) in a Biracial (Black‐White) Population of Children, Adolescents, and Young Adults: The Bogalusa Heart Study,” American Journal of Epidemiology 150, no. 7 (1999): 667–674. [DOI] [PubMed] [Google Scholar]
- 10. Lambert M., Paradis G., O'Loughlin J., Delvin E. E., Hanley J. A., and Levy E., “Insulin Resistance Syndrome in a Representative Sample of Children and Adolescents From Quebec, Canada,” International Journal of Obesity and Related Metabolic Disorders 28, no. 7 (2004): 833–841. [DOI] [PubMed] [Google Scholar]
- 11. Stoner L., Weatherall M., Skidmore P., et al., “Cardiometabolic Risk Variables in Preadolescent Children: A Factor Analysis,” Journal of the American Heart Association 6, no. 10 (2017): 1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Ramachandran A., Snehalatha C., Yamuna A., Murugesan N., and Narayan K. M., “Insulin Resistance and Clustering of Cardiometabolic Risk Factors in Urban Teenagers in Southern India,” Diabetes Care 30, no. 7 (2007): 1828–1833. [DOI] [PubMed] [Google Scholar]
- 13. Zhou J., Zhang F., Qin X., et al., “Age at Adiposity Rebound and the Relevance for Obesity: A Systematic Review and Meta‐Analysis,” International Journal of Obesity 46, no. 8 (2022): 1413–1424. [DOI] [PubMed] [Google Scholar]
- 14. Dearden L., Bouret S. G., and Ozanne S. E., “Sex and Gender Differences in Developmental Programming of Metabolism,” Molecular Metabolism 15 (2018): 8–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Heude B., Forhan A., Slama R., et al., “Cohort Profile: The EDEN Mother‐Child Cohort on the Prenatal and Early Postnatal Determinants of Child Health and Development,” International Journal of Epidemiology 45, no. 2 (2016): 353–363. [DOI] [PubMed] [Google Scholar]
- 16. Cole T. J. and Lobstein T., “Extended International (IOTF) Body Mass Index Cut‐Offs for Thinness, Overweight and Obesity,” Pediatric Obesity 7, no. 4 (2012): 284–294. [DOI] [PubMed] [Google Scholar]
- 17. Goran M. I., Kaskoun M. C., Carpenter W. H., Poehlman E. T., Ravussin E., and Fontvieille A. M., “Estimating Body Composition of Young Children by Using Bioelectrical Resistance,” Journal of Applied Physiology (1985) 75, no. 4 (1993): 1776–1780. [DOI] [PubMed] [Google Scholar]
- 18. Cisse A. H., Lioret S., de Lauzon‐Guillain B., et al., “Association Between Perinatal Factors, Genetic Susceptibility to Obesity and Age at Adiposity Rebound in Children of the EDEN Mother‐Child Cohort,” International Journal of Obesity 45, no. 8 (2021): 1802–1810. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Carpenter M. W. and Coustan D. R., “Criteria for Screening Tests for Gestational Diabetes,” American Journal of Obstetrics and Gynecology 144, no. 7 (1982): 768–773. [DOI] [PubMed] [Google Scholar]
- 20. Aubert A. M., Forhan A., de Lauzon‐Guillain B., et al., “Deriving the Dietary Approaches to Stop Hypertension (DASH) Score in Women From Seven Pregnancy Cohorts From the European ALPHABET Consortium,” Nutrients 11, no. 11 (2019): 2706, 10.3390/nu11112706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Josse J. and Husson F., “missMDA: A Package for Handling Missing Values in Multivariate Data Analysis,” Journal of Statistical Software 70, no. 1 (2016): 1–31. [Google Scholar]
- 22. Patel S., Fraser A., Davey Smith G., et al., “Associations of Gestational Diabetes, Existing Diabetes, and Glycosuria With Offspring Obesity and Cardiometabolic Outcomes,” Diabetes Care 35, no. 1 (2012): 63–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Hillier T. A., Pedula K. L., Schmidt M. M., Mullen J. A., Charles M. A., and Pettitt D. J., “Childhood Obesity and Metabolic Imprinting: The Ongoing Effects of Maternal Hyperglycemia,” Diabetes Care 30, no. 9 (2007): 2287–2292. [DOI] [PubMed] [Google Scholar]
- 24. Corrado F., D'Anna R., Cannata M. L., Interdonato M. L., Pintaudi B., and Di Benedetto A., “Correspondence Between First‐Trimester Fasting Glycaemia, and Oral Glucose Tolerance Test in Gestational Diabetes Diagnosis,” Diabetes & Metabolism 38, no. 5 (2012): 458–461, 10.1016/j.diabet.2012.03.006. [DOI] [PubMed] [Google Scholar]
- 25. Harreiter J., Simmons D., Desoye G., et al., “IADPSG and WHO 2013 Gestational Diabetes Mellitus Criteria Identify Obese Women With Marked Insulin Resistance in Early Pregnancy,” Diabetes Care 39, no. 7 (2016): e90–e92. [DOI] [PubMed] [Google Scholar]
- 26. Oken E., Morton‐Eggleston E., Rifas‐Shiman S. L., et al., “Sex‐Specific Associations of Maternal Gestational Glycemia With Hormones in Umbilical Cord Blood at Delivery,” American Journal of Perinatology 33, no. 13 (2016): 1273–1281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Lu J., Zhang S., Li W., et al., “Maternal Gestational Diabetes Is Associated With Offspring's Hypertension,” American Journal of Hypertension 32, no. 4 (2019): 335–342. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Le Moullec N., Fianu A., Maillard O., et al., “Sexual Dimorphism in the Association Between Gestational Diabetes Mellitus and Overweight in Offspring at 5–7 Years: The OBEGEST Cohort Study,” PLoS One 13, no. 4 (2018): e0195531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Climie R. E., Tafflet M., van Sloten T., et al., “Cardiovascular Health at Age 5 Years: Distribution, Determinants, and Association With Neurodevelopment,” Frontiers in Pediatrics 10 (2022): 827525, 10.3389/fped.2022.827525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. W. L. Lowe, Jr. , Lowe L. P., Kuang A., et al., “Maternal Glucose Levels During Pregnancy and Childhood Adiposity in the Hyperglycemia and Adverse Pregnancy Outcome Follow‐Up Study,” Diabetologia 62, no. 4 (2019): 598–610, 10.1007/s00125-018-4809-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Scholtens D. M., Kuang A., Lowe L. P., et al., “Hyperglycemia and Adverse Pregnancy Outcome Follow‐Up Study (HAPO FUS): Maternal Glycemia and Childhood Glucose Metabolism,” Diabetes Care 42, no. 3 (2019): 381–392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Hernandez‐Diaz S., Schisterman E. F., and Hernan M. A., “The Birth Weight “Paradox” Uncovered?,” American Journal of Epidemiology 164, no. 11 (2006): 1115–1120. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Graphical representation of the hypothetical causal relationships between the variables included in the fully adjusted model, with the following colour‐coded groups: red circle for confounders, white circle for precision variables, empty blue circle for mediators, green for exposure and non‐empty blue circle for the outcome. Arrows represent the hypothetical direction of influence between variables.
Figure S2. Illustration of the robustness of the missMDA imputation used.
Table S1. Cardiometabolic health outcomes of individuals in the highest quintiles of each identified pattern.
Table S2. Imputation method used for cardiometabolic health parameters and covariates.
Table S3. Characteristics of participants with evaluated PPG50g‐1hr, FPG, with GDM and non‐GDM.
Table S4. Description of cardiometabolic health patterns (children with non‐missing cardiometabolic outcomes) (n = 594).
Data Availability Statement
The data underlying the findings cannot be made freely available for ethical and legal restrictions imposed, because this study includes a substantial number of variables that together, could be used to re‐identify the participants based on a few key characteristics and then be used to have access to other personal data. Therefore, the French ethics authority strictly forbids making these data freely available. However, they can be obtained upon request from the EDEN principal investigator. Readers may contact barbara.heude@inserm.fr to request the data. The analytic code will be made available upon request pending application and approval.