Association Between Maternal Prepregnancy Body Mass Index and Plasma Folate Concentrations With Child Metabolic Health

Importance  Previous reports have linked maternal prepregnancy obesity with low folate concentrations and child overweight or obesity (OWO) in separate studies. To our knowledge, the role of maternal folate concentrations, alone or in combination with maternal OWO, in child metabolic health has not been examined in a prospective birth cohort.

Objective  To test the hypotheses that maternal folate concentrations can significantly affect child metabolic health and that sufficient maternal folate concentrations can mitigate prepregnancy obesity-induced child metabolic risk.

Design, Setting, and Participants  This prospective birth cohort study was conducted at the Boston Medical Center, Boston, Massachusetts. It included 1517 mother-child dyads recruited at birth from 1998 to 2012 and followed up prospectively up to 9 years from 2003 to 2014.

Main Outcomes and Measures  Child body mass index z score calculated according to US reference data, OWO defined as a body mass index in the 85th percentile or greater for age and sex, and metabolic biomarkers (leptin, insulin, and adiponectin).

Results  The mean (SD) age was 28.6 (6.5) years for mothers and 6.2 (2.4) years for the children. An L-shaped association between maternal folate concentrations and child OWO was observed: the risk for OWO was higher among those in the lowest quartile (Q1) as compared with those in Q2 through Q4, with an odds ratio of 1.45 (95% CI, 1.13-1.87). The highest risk for child OWO was found among children of obese mothers with low folate concentrations (odds ratio, 3.05; 95% CI, 1.91-4.86) compared with children of normal-weight mothers with folate concentrations in Q2 through Q4 after accounting for multiple covariables. Among children of obese mothers, their risk for OWO was associated with a 43% reduction (odds ratio, 0.57; 95% CI, 0.34-0.95) if their mothers had folate concentrations in Q2 through Q4 compared with Q1. Similar patterns were observed for child metabolic biomarkers.

Conclusions and Relevance  In this urban low-income prospective birth cohort, we demonstrated an L-shaped association between maternal plasma folate concentrations and child OWO and the benefit of sufficient folate concentrations, especially among obese mothers. The threshold concentration identified in this study exceeded the clinical definition of folate deficiency, which was primarily based on the hematological effect of folate. Our findings underscore the need to establish optimal rather than minimal folate concentrations for preventing adverse metabolic outcomes in the offspring.

In the United States, more than 50% of women of reproductive age are overweight or obese (OWO), with the prevalence at about 80% in non-Hispanic black women.¹ Maternal obesity has been linked to offspring obesity.^2,3 This transgenerational transmission may originate in utero⁴ and amplify the obesity epidemic in current and future generations.^5,6 However, questions remain regarding what early-life factors can enhance or mitigate the adverse effects of maternal obesity on child obesity and what underlying molecular mechanisms are involved.

Growing evidence suggests that maternal nutrition, through its effect on the fetal intrauterine environment, has a profound and life-long influence on child health.⁷ Among several specific nutrients that have been implicated, folate—an essential B vitamin involved in nucleic acid synthesis, DNA methylation, cellular growth⁸—is particularly important. The demand for folate increases during pregnancy due to fetal and placental growth and uterus enlargement.⁹ Research has shown the benefits of sufficient folate on reproductive, neurodevelopmental, and cardiovascular health^10-13 and, in particular, on preventing fetal neural tube defects^12,13 and counteracting adverse reproductive effects of environmental endocrine disruptors.¹⁰ However, in the United States, despite universal folic acid fortification of cereal grain products and the recommendation for women of reproductive age and especially pregnant women to take multivitamins,¹⁴ almost one-quarter of these women have insufficient folate to prevent neural tube defects.¹⁵ Maternal obesity has been linked with low blood folate concentrations^16,17 and, as such, obese mothers are at a higher risk for folate insufficiency compared with normal-weight mothers. From this perspective, studying the individual and combined effects of maternal obesity and folate concentrations offers another possibility to counteract maternal obesity–induced metabolic risk via the optimization of maternal folate status, a relatively safe and inexpensive strategy.

Folate can improve insulin sensitivity¹⁸ and act as a superoxide scavenger.¹⁹ In animal models, methyl supplements during pregnancy can prevent transgenerational amplification of obesity.²⁰ However, relevant data are lacking in humans. Two published observational studies based on retrospective recall of prenatal multivitamin intake did not find an association between prenatal vitamin intake and offspring adiposity^21,22; however, these studies did not measure maternal folate concentrations.

Using data on plasma folate concentrations from a well-established US prospective birth cohort, we sought to test the hypotheses that maternal plasma folate concentrations can significantly affect child metabolic health and that sufficient maternal folate concentrations can alleviate prepregnancy obesity-induced child metabolic risk as measured by child body mass index (calculated as weight in kilograms divided by height in meters squared), OWO, and metabolic biomarkers (insulin, leptin, and adiponectin).

Box Section Ref ID

This study included mother-infant pairs from the Boston Birth Cohort (BBC), which was initiated in 1998 with a rolling enrollment in Boston, Massachusetts, a predominantly urban, low-income minority population, as described previously.²³ At enrollment, each mother completed a questionnaire interview that assessed prepregnancy weight, height, race/ethnicity, education, smoking, parity, perceived stress during pregnancy, prenatal multivitamin intake, and a blood draw within 48 to 72 hours after delivery. Since 2003, all children who were enrolled in the BBC and planned to receive primary care at the Boston Medical Center were eligible for postnatal follow-up. A standardized questionnaire was used to assess postnatal demographic and environmental information.²⁴

As illustrated in the flowchart (eFigure 1 in the Supplement), of the 2891 mother-children pairs who were enrolled from 1998 to 2012 and followed up prospectively up to 9 years, this study included 1517 mother-children pairs who completed at least 1 postnatal well-child care visit beyond age 2 years at the Boston Medical Center and had complete data on maternal BMI and plasma folate concentrations. Maternal demographic characteristics and birth outcomes were comparable between the samples included in and excluded from this study (eTable 1 in the Supplement).

The study protocol was approved by the institutional review boards of Boston University Medical Center, Ann and Robert H. Lurie Children’s Hospital of Chicago, and Johns Hopkins Bloomberg School of Public Health. Written informed consent was obtained from all of the study mothers.

Maternal prepregnancy weight and height were primarily based on a standard maternal questionnaire interview within 2 to 3 days of delivery. At that time, the study mothers were not aware of their folate status and did not know their child’s future growth trajectory. Maternal BMI was categorized into 3 groups: normal weight (18.5-24.9), overweight (25-29.9), and obesity (≥30). Underweight mothers were removed from the analysis owing to small sample size.

Educational attainment was classified into high school and below vs college and above; maternal smoking during pregnancy was classified into 3 groups: never smoker, intermittent, or continuous smoker²⁵; and maternal race/ethnicity were classified as black, Hispanic, or other (which included white, Asian, Pacific Islander, and more than 1 race/ethnicity). Perceived stress during pregnancy was grouped into low vs high.²⁶ Maternal diabetes status was classified as nondiabetic or diabetic (either preexisting or gestational diabetes).²³ Infant breastfeeding history was collected in follow-up visits and 70.5% of the questionnaire interview was completed before the child reached 2 years of age. Breastfeeding was grouped into exclusively breast, exclusively formula, or mixed breast and formula.²⁴

Gestational age was determined by the first day of the last menstrual period and early prenatal ultrasonographic results²⁵ and categorized into term (≥37 weeks), which was further grouped into late term (≥39 weeks) and early term (37-38 weeks), and preterm (<37 weeks), which was further grouped into late preterm (34-36 weeks) and early preterm (<34 weeks).²³ Information regarding cesarean delivery and birth weight was abstracted from the electronic medical record.

Child weight and height were measured by medical staff during well-child care visits as documented in the electronic medical record. Body mass index z scores and percentiles were calculated using US national reference data.²⁷ Overweight or obesity was defined as a BMI in the 85th percentile or greater for age and sex.

Plasma folate and vitamin B₁₂ concentrations were measured by a commercial laboratory via chemiluminescent immunoassay using a MAGLUMI 2000 Analyzer (Snibe Co Ltd) with interassay coefficient of variation of less than 4%.¹¹

We assessed child plasma insulin (a marker of insulin resistance),²³ leptin (a marker of adiposity),²⁸ and adiponectin to leptin ratio (a marker of insulin sensitivity)²⁹ using established methods. Plasma insulin and leptin concentrations were determined using sandwich immunoassays based on flow metric xMAP technology on Luminex 200 machines (Luminex Corp).²³ Adiponectin was measured by enzyme-linked immunosorbent assay with an interassay coefficient of variation of less than 5.8%.

The primary outcomes of interest included BMI z score (continuous variable) and OWO (binary) at the last well-child care visit, as well as metabolic biomarkers in early childhood. The primary predictors were plasma folate concentrations that were evaluated both as continuous variables and as categorical variables in quartiles (Qs). At first, we examined the association of maternal plasma folate concentrations with child BMI z score and OWO probability using smoothing plots (PROC LOESS). Given that there was no significant difference in the risk for OWO between Q2 and Q4, we grouped maternal folate concentrations into low (Q1) vs adequate (Q2-Q4) in the subsequent analyses as there is no standard, clinically meaningful cutpoint for low folate concentrations relevant to metabolic outcomes. Available clinical cutoffs are mainly for anemia and vary by definition.³⁰ Furthermore, we estimated the combined effects of maternal folate concentrations and prepregnancy BMI categories on child BMI and metabolic biomarkers z scores using linear regression models. As plasma insulin level, leptin level, and adiponectin to leptin ratio all had skewed distributions, these biomarkers were log transformed before calculating age- and sex-specific z scores. As a next step, we estimated the individual and combined effects of maternal folate concentrations and prepregnancy BMI categories on child OWO using logistic regression models. We tested the interaction of maternal prepregnancy BMI categories (as a categorical variable with 3 levels) and folate status (as a binary variable) on child BMI and metabolic biomarker z scores and odds of OWO by including cross-product terms in models with indicator terms for BMI categories and folate status. The effect modifications were tested with the likelihood ratio test using an a priori α value of .05. To further assess the robustness of the findings, we conducted stratified analyses by child age group (2-5 years vs 6-9 years), race (black only, the major race group in the BBC), and preterm vs term birth.

Covariables were selected based on the published literature and our previous studies in the BBC. In addition, we adjusted for maternal BMI categories (overweight/obese vs normal weight) to assess an individual effect of folate. To further address residual confounders, we performed propensity score–matched sensitivity analyses³¹ that compared maternal low folate concentration (<9 ng/mL; to convert to nmol/L, multiply by 2.266) with adequate folate (range, 9-81.9 ng/mL). We further adjusted for child plasma folate concentrations and examined combined effects of child folate concentrations with maternal folate on the risk for child OWO. Child age and sex were not included in the regression model because they were already accounted for when we defined the outcome variables. All statistical analyses were performed using SAS version 9.4 (SAS Institute).

This study was composed of 1517 mother-child pairs, including 1019 black pairs (67.2%) and 290 Hispanic pairs (19.1%). In all, 443 (29.2%) and 381 (25.1%) mothers were overweight and obese in prepregnancy, respectively, and a total of 590 (38.9%) children were OWO at ages 2 to 9 years. Overweight or obese children had higher birth weights, higher rates of formula feeding, and higher rates of maternal obesity and diabetes (Table 1).

The median (interquartile range [IQR]) for maternal plasma folate concentrations was 13.6 ng/mL (IQR, 9-19.6 ng/mL). Obese mothers had lower folate concentrations compared with normal-weight mothers: the geometric means of folate concentrations in normal-weight, overweight, and obese mothers were 13.5 ng/mL (95% CI, 12.9-14.2), 13.4 ng/mL (95% CI, 12.6-14.1), and 12.3 ng/mL (95% CI, 11.7-13.0), respectively (P for trend = .03). Rates of low (Q1) folate concentration among normal-weight, overweight, and obese mothers were 24.0%, 24.8%, and 27.0%, respectively.

Maternal plasma folate concentrations were inversely associated with child BMI z score and OWO, but the association was nonlinear (Wald χ² test P < .05). As shown in Figure 1A, there was a steep rise in BMI z score for children whose mothers’ plasma folate concentrations were below the 25th percentile (ie, 9 ng/mL). However, higher maternal folate concentrations beyond the median value did not confer additional benefits. Consistently, the increased risk for child OWO was mainly concentrated in the lowest folate quartile (Figure 1B): the multivariate odds ratio was 1.50 (95% CI, 1.10-2.04) compared with the higher quartiles (Table 2). Of note, very similar results were observed before vs after the adjustment of birth weight.

As shown in Table 2 and Figure 2, there was a significant combined effect of maternal folate concentrations and prepregnancy obesity on offspring BMI z score and risk for OWO. Children of obese mothers with low folate concentrations had a 0.89-unit increase (95% CI, 0.63-1.15) in BMI z score and a 3.05-fold increased risk (95% CI, 1.91-4.86) for OWO compared with those whose mothers with adequate folate concentration and normal weight. However, there was no significant interaction between maternal folate status and prepregnancy obesity (P > .05). Given maternal obesity, children of mothers with adequate folate concentrations were associated with a 43% reduction in the risk for OWO compared with children of mothers with low folate concentrations (odds ratio, 0.57; 95% CI, 0.34-0.95; P = .03).

As compared with adequate maternal folate concentrations, low folate concentrations were associated with an increase in the concentrations of insulin and leptin and a reduction in the adiponectin to leptin ratio in offspring. Children of obese mothers with low folate concentrations had a 0.39-unit increase (95% CI, 0.09-0.69) in plasma insulin z scores compared with the group with adequate maternal folate concentrations and normal maternal weight. A similar pattern was seen for leptin. In contrast, the children of obese mothers with low folate concentrations were associated with a 0.43-unit decrease (95% CI, 0.17-0.70) in their adiponectin to leptin ratio z scores. Given maternal obesity, the children of mothers with adequate folate concentrations had a 0.27-unit decrease (95% CI, −0.07 to 0.61) in plasma insulin, a 0.32-unit decrease (95% CI, 0.03-0.62) in leptin; and a 0.43-unit increase (95% CI, 0.14-0.72) in adiponectin to leptin ratio z scores compared with children of mothers with low folate concentrations (Table 3 and eFigure 2 in the Supplement). These associations were attenuated after further adjustment for child adiposity (eTable 2, eTable 3, and eTable 4 in the Supplement).

The associations described here did not differ appreciably across the following strata defined by child age (2-5 years vs 6-9 years; eTable 5 and eTable 6 in the Supplement), race (black only; eTable 7 in the Supplement), and gestation (preterm vs term birth; eTable 8 in the Supplement), or with additional controlling for other prenatal and perinatal factors including whether the index child was a planned pregnancy and whether the index child was delivered by cesarean birth (eTable 9 and eTable 10 in the Supplement). A propensity score–matched analysis showed that low maternal folate concentrations (<9 ng/mL), as compared with adequate concentrations (9-81.9 ng/mL), were associated with an increased risk for childhood OWO (odds ratio, 1.39; 95% CI, 1.04-1.86) (eTable 11 in the Supplement). When child metabolic biomarkers were limited within the first 2 years of life, the association between maternal folate concentrations and child metabolic biomarkers were not appreciably changed (eTable 12 in the Supplement).

Among 803 study children with postnatal folate concentrations, children had higher plasma folate concentrations than their mothers (geometric mean: 16.5 ng/mL; 95% CI, 16.1-17.0, vs 13.2 ng/mL; 95% CI, 12.8-13.6; P < .001). If the threshold for low folate concentration was defined as less than 9 ng/mL (maternal Q1 folate cutpoint), 72 children (9%) had low folate concentrations. The association between maternal folate and offspring BMI z score and the risk for OWO remained after further adjustment for child folate status (eTable 13 and eTable 14 in the Supplement).

To our knowledge, this is the first prospective birth cohort study to evaluate the individual and combined effects of maternal prepregnancy BMI and plasma folate concentrations on offspring metabolic outcomes including BMI, OWO status, and metabolic biomarkers. Our findings lend further support that maternal prenatal nutritional status may play an important role in child metabolic disorders.^2,3

Our data revealed a nonlinear L-shaped association: maternal plasma folate concentration in the lowest quartile was associated with increased risk for offspring OWO. Above this threshold, higher folate concentrations did not confer additional benefit, also suggesting a ceiling effect of folate. The threshold concentration identified in this study exceeded the clinical definition of folate deficiency (<4.4 ng/mL), which was primarily based on the hematological effect of folate.³² Our findings are in agreement with a recent study³³ that used a similar approach to estimate an optimal population red blood cell folate concentration for the prevention of neural tube defects. Taken together, these findings underscore the need to establish optimal rather than merely minimal folate concentrations for preventing adverse metabolic outcomes in offspring.

Our study suggests that adequate maternal folate concentrations could mitigate the adverse effects of maternal obesity on child metabolic risk. We demonstrated that, given maternal obesity, the risk for child OWO was associated with a 43% reduction if the mothers had adequate folate concentrations compared with low folate concentrations.

Our data on metabolic biomarkers lend further support to the biological plausibility of our findings. Previous mechanistic studies indicate that factors that create an unfavorable cardiometabolic intrauterine environment, such as increased insulin resistance, elevated glucose concentrations, and increased oxidative stress, in obese mothers may lead to OWO in offspring.^4,16,34,35 Consistently, we found that children with obese mothers who had lower folate concentrations had the most unfavorable metabolic profiles (increased insulin and leptin and decreased adiponectin to leptin ratio in childhood). These associations appear to be mediated by child adiposity. Our findings are in agreement with a previous antenatal micronutrient randomized trial showing that in-utero exposure to a folic acid supplement reduced the risk for metabolic syndrome in childhood.¹⁸ In contrast, an Indian study (of an observational cohort) reported a positive association between maternal folate concentration and offspring homeostatic model assessment–insulin resistance.³⁶ Notably, this study assumed a linear association between maternal folate concentration and offspring homeostatic model assessment–insulin resistance, and it did not analyze the combined effect of maternal folate and BMI on child metabolic risk. Our data suggest that inadequate maternal folate concentrations may have deleterious long-term metabolic effects beyond fetal overgrowth and child OWO, while adequate maternal folate concentrations can counterbalance the detrimental effects of maternal obesity on the offspring.

Several limitations should be acknowledged. We used maternal plasma folate concentrations taken at 1 to 3 days after delivery, which is at best a proxy of folate nutrition during the third trimester of pregnancy. While periconception and first-trimester folate concentrations are important for neural tube and brain development, fetal weight gain mostly occurs in the third trimester.³⁷ Our findings support that maternal folate concentration during the third trimester is important for child metabolic outcomes. Although we did not measure fetal folate concentrations, a previous study suggested a high degree of transplacental passage of maternal folate to the fetus.³⁸ Maternal prepregnancy BMI was primarily based on self-reported height and weight, thus it may be subject to reporting bias. Nevertheless, in a subset of the study population (n = 672), self-reported BMIs were compared with those taken from medical records and showed a high degree of agreement (r = 0.89; P < .001). In addition, exclusion of 1374 children for a variety of reasons may have resulted in selection bias, although the demographic characteristics were comparable with those of the included participants. Last, child plasma insulin, leptin, and adiponectin concentrations were measured in nonfasting samples. As we discussed in our previous report,²³ the timing of blood sampling occurred randomly (any time during the clinical hours), which may have introduced background noise and thus biased our results toward null.

We demonstrated an L-shaped association between maternal folate concentrations and child metabolic risk, suggesting both a threshold and a ceiling effect of folate. We found that low maternal folate concentrations during late pregnancy can increase child metabolic risk as measured by BMI z score, OWO, and metabolic biomarkers; conversely, sufficient maternal folate concentrations can mitigate the detrimental effects of maternal obesity on the offspring. Our findings underscore the need to establish and ensure optimal rather than minimal maternal folate concentrations for preventing offspring adverse metabolic outcomes, especially among obese mothers.

Corresponding Author: Xiaobin Wang, MD, MPH, ScD, Department of Population, Family, and Reproductive Health, Johns Hopkins University Bloomberg School of Public Health, Center on the Early Life Origins of Disease, 615 N Wolfe St, Baltimore, MD 21205 (xwang82@jhu.edu).

Accepted for Publication: March 24, 2016.

Published Online: June 13, 2016. doi:10.1001/jamapediatrics.2016.0845.

Author Contributions: Dr Wang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr Wang is the principal investigator of the Boston Birth Cohort.

Study concept and design: G. Wang, Hu, Mistry, Bartell, Zuckerman, Wang.

Acquisition, analysis, or interpretation of data: G. Wang, Hu, Mistry, Zhang, Ren, Huo, Paige, Hong, Caruso, Z. Ji, Chen, Y. Ji, Pearson, H. Ji, Zuckerman, Cheng, Wang.

Drafting of the manuscript: G. Wang, Paige, Z. Ji, Wang.

Critical revision of the manuscript for important intellectual content: G. Wang, Hu, Mistry, Zhang, Ren, Huo, Bartell, Hong, Caruso, Chen, Y. Ji, Pearson, H. Ji, Zuckerman, Cheng, Wang.

Statistical analysis: G. Wang, Hu, Zhang, Hong, Z. Ji, Chen, Y. Ji, H. Ji, Wang.

Obtained funding: Wang.

Administrative, technical, or material support: G. Wang, Zhang, Bartell, Caruso, Y. Ji, Pearson, Zuckerman, Wang.

Study supervision: Pearson, Zuckerman, Wang.

Conflict of Interest Disclosures: None reported.

Funding/Support: The Boston Birth Cohort is supported in part by the PERI grants 20-FY02-56 and 21-FY07-605 from the March of Dimes; grants R21ES011666, R21HD066471, R01HD041702, U01AI090727, R21AI079872, and R01HD086013 from the National Institutes of Health; and grant R40MC27443 from the Maternal and Child Health Bureau. Dr Zhang is supported by the intramural research program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We thank all of the study participants and the Boston Medical Center Labor and Delivery Nursing staff for their support and help with the study. We are also grateful for the dedication and hard work of the field team at the Department of Pediatrics, Boston University School of Medicine.