Associations of maternal BMI and gestational weight gain with neonatal adiposity in the Healthy Start study^12,3,45

INTRODUCTION

Pediatric obesity is a critical public health problem, with 18% of US elementary school-aged children classified as obese in 2009–2010 (1). Childhood obesity is associated with numerous cardiovascular and metabolic conditions including dyslipidemia, hypertension, and insulin resistance (2) and may lead to a reduced quality of life (3) and reduced life expectancy (4). Studies suggested that a child's trajectory toward becoming obese may be set by influences very early in life (5). In a study of children aged 5–14 y, who subsequently developed obesity, nearly one-half of them were already overweight at age 5 y, and more than one-third of them were large at birth (6). This early onset suggests an etiologic contribution from prenatal and early postnatal environments.

The intrauterine environment has been shown to influence offspring obesity risk in the context of maternal gestational diabetes and prepregnancy obesity (7, 8). Moreover, several studies showed excessive gestational weight gain (GWG) to be associated with increased risk of large-for-gestational-age birth (9, 10) as well as offspring overweight or adiposity in childhood (11, 12), adolescence (13), and adulthood (14). Childhood obesity likely reflects a combination of prenatal and postnatal influences. However, the adiposity of a neonate at birth cannot be influenced by postnatal exposures and, therefore, may provide a useful window into the effects of the intrauterine environment. Some studies have focused on neonatal body fat measured by using a variety of methods including skinfold thickness (15, 16), dual-energy X-ray absorptiometry (17), and air-displacement plethysmography (18). These studies reported positive associations of GWG with neonatal adiposity, but results varied regarding the role of maternal prepregnancy BMI and timing of GWG.

It has been proposed that early and midpregnancy GWG, which primarily represent increased maternal fat rather than the weight of the growing fetus (19), may be causally linked to offspring adiposity through the increased availability of maternal fuels such as glucose, amino acids, and free fatty acids (12, 20). However, few studies have examined associations between neonatal body composition and GWG occurring in specific periods of pregnancy.

Our aim was to estimate associations of maternal prepregnancy BMI and overall GWG as well as period-specific rates of GWG with neonatal adiposity measures in a large, ethnically diverse population. We hypothesized that maternal prepregnancy BMI and GWG would be positively associated with neonatal fat mass and the percentage of body fat, and early pregnancy, midpregnancy, and late pregnancy rates of GWG would be positively and independently associated with neonatal adiposity.

SUBJECTS AND METHODS

RESULTS

Of 1132 women enrolled in the Healthy Start study with a delivery date before 31 October 2013, 16 women experienced fetal demise, and 8 women terminated consent before delivery. Of the remaining women, 1000 individuals had a documented gestational age at birth ≥37 wk. Of these women, 827 individuals completed body-composition measurements ≤3 d of delivery. One participant was missing information on prepregnancy weight and was excluded, which left 826 participants in the main analysis. In period-specific analyses, a subset of 752 women with available weight measures in all 3 periods was used.

The 826 participants in the main analysis resembled the 1000 potentially eligible participants as well as the 752 participants in period-specific analyses with regard to maternal age, prepregnancy BMI, income, education, race-ethnicity, gravidity, smoking, mode of delivery, infant sex, and gestational age of the infant at birth (Supplemental Table 1). A slightly higher frequency of below-recommended GWG was observed in the 1000 potentially eligible participants (many of whom had no neonatal body-composition measures taken at birth) compared with that of the 826 and 752 women eligible for analyses.

Of the 826 women in the main analysis, the mean age at delivery was 28 y, and the range was 16–43 y (Table 1). Approximately 20% of participants were obese before pregnancy (BMI ≥ 30); the mean prepregnancy BMI was 25.8. Participants were predominantly non-Hispanic white (53%), Hispanic (24%), or non-Hispanic black (17%), with 6% of women from other racial and ethnic groups. As expected, the observed GWG was highest in women who were underweight before pregnancy and lowest in women who were obese before pregnancy. However, women who were overweight or obese before pregnancy were more likely to exceed IOM recommendations for GWG than were women who were normal weight or underweight before pregnancy (data not shown).

When the observed GWG was classified by 2009 IOM recommendations (26), 21% of participants gained insufficient weight overall, 27% of participants gained adequate weight, and 51% of participants gained excessive weight (Table 1). When we examined period-specific rates of GWG relative to IOM recommendations, 51% of participants had an excessive gain in early pregnancy, 69% of participants had an excessive gain in midpregnancy, and 65% of participants had an excessive gain in late pregnancy; however, only 23% of subjects had excessive GWG in all 3 periods of pregnancy (data not shown). Observed and predicted overall GWG measures had similar means and medians (Table 2) and were highly correlated (r = 0.92, P < 0.001). The mean observed weekly rate of GWG increased from early pregnancy (0.17 kg/wk) to midpregnancy (0.55 kg/wk) and remained relatively constant into late pregnancy (0.54 kg/wk) (Figure 1).

Maternal BMI before pregnancy and the weekly rate of continuous predicted GWG were both positively associated with neonatal body size and body composition (Table 3). For each 1-kg/m² increase in maternal prepregnancy BMI, neonatal fat mass increased by 5.21 g (95% CI: 3.54, 6.89 g), fat-free mass increased by 7.71 g (95% CI: 4.50, 10.91 g), and the percentage of body fat increased by 0.12% (95% CI: 0.08%, 0.16%). For each 0.1-kg/wk increase in the overall rate of GWG, neonatal fat mass increased by 23.95 g (95% CI: 17.37, 30.53 g), fat-free mass increased by 33.95 g (95% CI: 21.35, 46.56 g), and the percentage fat mass increased by 0.55% (95% CI: 0.37%, 0.72%). There was no evidence of a statistical interaction between prepregnancy BMI and continuous predicted GWG in their associations with any outcomes.

The differences in magnitude between associations with prepregnancy BMI and GWG may be explained by the different SDs of BMI (SD: 6.3) and GWG (SD: 0.15 kg/wk) in our study population. The changes in neonatal body-composition measures associated with 1-SD unit increases in prepregnancy BMI and GWG were of similar magnitude (data not shown).

Adjustments for covariates strengthened some associations but did not materially change the results. With the removal of each covariate individually from the fully adjusted model, we determined that the adjustment for the rate of GWG was primarily responsible for the higher adjusted association of prepregnancy BMI with neonatal fat-free mass. We attributed this finding to the fact that prepregnancy BMI was inversely associated with GWG (r = −0.27, P < 0.0001), and the results of the unadjusted model were biased because of confounding by GWG.

Women with insufficient overall GWG according to IOM guidelines had offspring with lower fat mass and a lower percentage of body fat but no significant difference in offspring fat-free mass compared with for women with adequate GWG in adjusted models (Table 4). Women with excessive GWG had offspring with greater fat mass and also greater fat-free mass than did women with adequate GWG but no significant difference in the offspring percentage of body fat in adjusted models. Findings were similar when observed period-specific rates were compared with IOM guidelines for the rate of GWG in each trimester of pregnancy (data not shown). There was no evidence of effect modification by category of prepregnancy BMI of associations between the IOM category of GWG adequacy and any neonatal adiposity outcomes.

All 3 continuous period-specific observed rates of GWG were associated with neonatal fat mass, fat-free mass, and the percentage of body fat in models that contained only one period-specific rate as a predictor (Table 5). When models were further adjusted for rates of weight gain in the other 2 periods, associations between each of the period-specific rates of GWG and outcomes of neonatal fat mass and the percentage of body fat remained significant, whereas the association of early pregnancy GWG and fat-free mass became nonsignificant. There were no significant differences between early pregnancy, midpregnancy, and late pregnancy rates of GWG in the strength of their associations with fat mass (P = 0.63), fat-free mass (P = 0.45), or the percentage of body fat (P = 0.64) as assessed by using an omnibus test in the mutually adjusted model. We assessed the degree of variance inflation in the fully adjusted model and showed no evidence of multicollinearity.

In a sensitivity analysis, we treated responses of do not know for household income as missing data, and we performed a multiple imputation to fill in these missing covariate values. A reanalysis with the imputed data did not change the interpretation of any results (not shown).

DISCUSSION

In a large, multiethnic prebirth cohort study, we showed positive and independent associations of continuous maternal prepregnancy BMI and GWG with neonatal adiposity measures. Period-specific rates of GWG in early pregnancy, midpregnancy, and late pregnancy were independently associated with neonatal fat mass and the percentage of body fat, and midpregnancy and late pregnancy rates of GWG were associated with neonatal fat-free mass. Excessive overall GWG by IOM recommendations was associated with fat mass and fat-free mass but not adiposity (percentage of body fat) in adjusted models.

Our results are consistent with the findings of previous studies that reported that maternal prepregnancy BMI was positively associated with neonatal fat mass (28, 29) and add additional support to the hypothesis that maternal GWG is independently associated with neonatal adiposity at birth. Four previous studies specifically examined the association between GWG and offspring body composition at birth, which was assessed by using a variety of methods as follows: skinfold measurements (15, 16), dual-energy X-ray absorptiometry (17), and air-displacement plethysmography (18). All 4 studies classified GWG as excessive, adequate, or insufficient by IOM guidelines. Each of these studies had ∼50% of participants with excessive GWG, which was similar to the proportion in our population. However, some of the previous studies excluded underweight women (15, 18), smokers (15), or women with abnormal glucose tolerance during pregnancy (16). Only one of the previous studies was conducted in an ethnically diverse population (18), and only one study examined GWG in early or late pregnancy (15). Our sample was ethnically diverse and was larger than in any of the 4 previous studies, which had sample sizes between 172 and 564 (15–18). Contrary to the results of some previous studies (16, 18) but consistent with the findings of another study (17), we observed no evidence of a statistical interaction between prepregnancy BMI and GWG in their associations with neonatal adiposity.

The influence of maternal GWG on neonatal adiposity may differ according to the timing of weight gain during pregnancy. The only previous study that specifically addressed the timing of GWG showed that early excessive weight gain was associated with greater neonatal fat mass and the percentage of body fat than was late excessive weight gain (15). We examined the rate of GWG in 3 periods of pregnancy and observed positive associations between the rate of gain in each of the 3 periods and neonatal fat mass and the percentage of body fat, even after holding constant the rate of gain in other periods. We also showed independent associations between midpregnancy and late pregnancy rates of GWG and neonatal fat-free mass. We interpret the positive associations between early and midpregnancy GWG and neonatal adiposity as evidence of the influence of maternal fat accumulation and the intrauterine environment. The interpretation of the observed associations between late pregnancy GWG and fat mass or fat-free mass is complicated by the fact that fetal and placental weights constitute a larger fraction of weight gain in late pregnancy (19). However, we observed positive associations between the rate of GWG in all 3 periods and the neonatal percentage fat mass, which suggested that neonatal fat mass may increase more rapidly than fat-free mass in response to maternal GWG.

To make our results comparable with those of previous studies, we further examined the adequacy of observed weight gain according to IOM guidelines in association with neonatal body-composition measures. We showed positive associations between excessive GWG by IOM guidelines, relative to adequate GWG, and neonatal fat mass and fat-free mass. However, we did not observe a significant association between excessive GWG and the neonatal percentage of body fat. These results suggest that excessive GWG by IOM guidelines is more-strongly associated with total body mass than adiposity. Higher total body mass at birth has been associated with higher risk of subsequent child, adolescent, and adult obesity (30–33).

One strength of the analysis was the use of a predictive model to standardize the length of time over which overall GWG was calculated. Conventionally, GWG is calculated by subtracting prepregnancy weight from the last observed pregnancy weight. Because the estimate of the last observed pregnancy weight depends crucially on the length of the pregnancy, this approach may produce biased effect estimates in studies of outcomes that are also associated with gestational duration (34). With the use of a predictive model, we estimated GWG for a standard pregnancy length of 39 wk. Overall observed and predicted measures of GWG were highly correlated, which was likely related to the fact that we restricted our analysis to term pregnancies (34). With the use of an accurate predictive model and common time interval, we eliminated one source of bias in estimating associations between GWG and neonatal body composition, both of which may depend on the gestational age at birth. One limitation of our approach was that the 2-stage modeling procedure we used may have introduced some degree of measurement error in the exposure. We anticipated, on the basis of recent statistical literature (35), that this error would not have produced a bias in the estimate but may have inflated the variability of second-stage effect estimates.

There are plausible biological mechanisms by which maternal obesity and GWG may lead to increased neonatal adiposity. The fetal overnutrition hypothesis proposes that excess maternal circulating glucose crosses the placenta, resulting in increased fetal insulin secretion (36). Fetal hyperinsulinemia, in turn, may promote adipogenesis or adipose cell hypertrophy (37). Higher circulating free fatty acids and triglycerides associated with maternal obesity and excessive GWG may cause increased fatty acid transfer to the fetus, thereby promoting the development of larger or more-numerous fetal fat cells (20).

In addition to the mechanisms linking GWG to neonatal adiposity, other biological mechanisms were proposed by which fetal overnutrition may influence offspring's later risk of obesity. Long-term effects may be mediated through dysregulation of the hypothalamic-endocrine system that regulates appetite and satiety (38, 39) and the epigenetic alteration of genes related to leptin signaling (40). It is possible that all of these physiologic mechanisms operate together to increase offspring's risk of obesity in addition to or perhaps synergistically with a genetic predisposition.

The 2009 IOM report on weight gain during pregnancy emphasized the need for an investigation into the long-term effects of GWG on child health (26). Our study addressed that need by using data from a large, ethnically diverse cohort of pregnant women and their infants to examine the influence of maternal prepregnancy BMI and GWG on neonatal adiposity, which is linked to offspring's future risk of obesity (17, 41). We focused on neonatal adiposity to eliminate the influence of postnatal factors that may confound studies linking GWG to childhood obesity. We could not rule out residual confounding by shared genetic factors.

In conclusion, taken together with previously reported associations between GWG and childhood obesity (10–12, 14, 42–45), these findings suggest an obesogenic influence of GWG that may persist into childhood and beyond. Long-term follow-up of participants in Healthy Start and other large cohorts will be important to further our understanding of the developmental origins of obesity, thereby enabling future studies and programs aimed at primordial prevention of obesity and associated chronic diseases.

Supplementary Material