Sex-Specific Associations of Gestational Glucose Tolerance With Childhood Body Composition

Project Viva is an ongoing prospective prebirth cohort study in which we recruited pregnant women at their initial prenatal visit from Harvard Vanguard Medical Associates, a multispecialty group practice in eastern Massachusetts, between April 1999 and July 2002. All mothers gave informed consent, and institutional review boards of participating institutions approved the study. All procedures were in accordance with the ethics standards established by the Declaration of Helsinki (16).

Of the 2,128 mothers with a live birth in Project Viva, excluded were those with missing or incomplete records on glucose tolerance testing (n = 47), those with a history of previous type 1 (n = 9) or type 2 (n = 7) diabetes and polycystic ovarian syndrome with glucose intolerance (n = 2), and 29 pregnancies with gestation <34 weeks. We followed the children with in-person visits just after delivery and at 6 months, 3 years, and 7 years of age and with annual mailed questionnaires at other ages. Of the 1,709 eligible for the school-age visit, 1,116 agreed to participate in an in-person visit. Anthropometry could be measured in 1,070, but not all participants accepted to undergo DXA. Our final sample included 958 children with data on maternal glucose tolerance during pregnancy, anthropometry measured at the school-age visit, and relevant covariates. A subsample of 760 of these 958 children had body composition measures with DXA.

Compared with the mothers who were not included in this analysis (n = 1,170), included mothers (n = 958) were more likely to be older (32.3 vs. 31.4 years at enrollment), white (69.7 vs. 62.5%), college graduates (70.0 vs. 60.1%), and nonsmokers (90.3 vs. 84.3%) and have household income >$70,000/year (74.6 vs. 65.3%). However, they did not differ substantially in terms of mean maternal prepregnancy BMI (24.9 vs. 25.2 kg/m²), gestational weight gain (15.4 vs. 15.6 kg) or gestational glucose tolerance status (normoglycemic: 83.1 vs. 82.2%). Father’s BMI was similar in the included and excluded groups (26.4 kg/m²). Children included in the analysis had slightly higher birth weight for gestational age z score than those who were not included (0.21 vs. 0.14).

Obstetric clinicians routinely screened all women for GDM at 26–28 weeks of gestation with a nonfasting oral glucose challenge test (GCT), in which venous blood was sampled 1 h after a 50-g oral glucose load. If the blood glucose exceeded 140 mg/dL, the clinician referred the woman for a fasting 3-h 100-g oral glucose tolerance test (OGTT). Abnormal OGTT results were a blood glucose >95 mg/dL at baseline, >180 mg/dL at 1 h, >155 mg/dL at 2 h, or >140 mg/dL at 3 h (17). We categorized women with two or more abnormal values on the OGTT as having GDM; those with one abnormal value on the OGTT as having an intermediate glucose intolerance (IGI) previously called impaired glucose tolerance (7); those with an abnormal GCT but a normal OGTT as having isolated hyperglycemia (IH); and the remaining women as having normal glucose tolerance. Those who were diagnosed with GDM were typically followed by a nutritionist, instructed to check their fasting blood glucose daily, and treated with diet and in some cases insulin (7). Mothers with IGI or IH did not have any further screening and were managed in the same way as women with normal GCT results.

During the in-person school-age visit, trained research assistants measured children’s weights (TBF-300A; Tanita, Arlington Heights, IL) and heights (calibrated stadiometer; Shorr Productions, Olney, MD). We calculated age- and sex-specific BMI percentiles and z scores using U.S. national reference data (18). The research staff measured subscapular (SS) and triceps (TR) skinfold thicknesses using Holtain calipers (Holtain, Crosswell, U.K.) and calculated the sum (SS+TR) and the ratio (SS/TR) of skinfolds. BMI z score and SS+TR represent overall adiposity, whereas SS/TR is a measure of central or truncal adiposity (19). To measure waist circumference, we used the method used in the National Health and Nutrition Examination Survey (NHANES) (20). For each measurement, the measuring tape was positioned parallel to the floor with the participant standing—abdomen relaxed, arms at the sides, and feet together—and facing the observer with the waist exposed (21). We measured waist circumference just above the right iliac crest at the mid-axillary line to the nearest 0.1 cm using a Hoechstmass measuring tape (Hoechstmass Balzer, Sulzbach, Germany). Research assistants followed standardized techniques and participated in biannual in-service training to ensure measurement validity (IJ Shorr; Shorr Productions) (22). Inter- and intrarater errors for skinfold measurements were within published reference ranges for all measurements (23).

Trained research assistants administered whole-body DXA scans with Hologic model Discovery A (Hologic, Bedford, MA) that they checked for quality control on visit days. We used Hologic software version 12.6 for scan analysis. A single trained research assistant checked all scans for positioning, movement, and artifacts and defined body regions for analysis. Intrarater reliability was high (r = 0.99). We calculated the DXA fat mass and fat-free mass indexes using the following formula: [total DXA fat mass (or fat free mass) in kg]/(height in meters)² (24). We also calculated the DXA trunk to peripheral fat mass ratio, a measure of central adiposity (25,26).

During pregnancy and in late childhood, using a combination of questionnaires and interviews, we obtained information about maternal age, race/ethnicity, education, parity, smoking during pregnancy, marital status, and household income. We collected information from prenatal medical records on serial pregnancy weights and blood pressure readings and infant birth weight and delivery date. Mothers reported their prepregnancy weight and height and the paternal weight and height. We calculated gestational weight gain as the difference between prepregnancy weight and the last clinically recorded weight before delivery. We derived gestational age from the last menstrual period or from the second trimester ultrasound if the two estimates differed by >10 days. Based on U.S. national natality data, we determined sex-specific birth weight for gestational age z scores (27).

At the school-age visit, mothers reported—separately for weekdays and weekend days—the number of hours per day that the children participated in light, moderate, and vigorous physical activities; watched television or other electronic screen media; and slept. Mothers also reported the frequency of children’s consumption of sugar-sweetened beverages, fast food, and fried foods.

We report age-, race/ethnicity, and sex-adjusted means for child body composition and behaviors reported at the school-age visit. Using multivariable linear regression, we built models based on bivariate associations and on our expectation of which variables would independently predict the outcomes as demonstrated by prior studies (8,9,28,29). Our basic model was adjusted for child age at examination (model 1) and subsequently adjusted for maternal prepregnancy BMI (model 2), gestational weight gain and paternal BMI (model 3), and maternal race/ethnicity, age, education, and parity and smoking during pregnancy, marital status, and household income (model 4: final model). When the outcome was a measurement of central adiposity, we additionally adjusted for child BMI because we were interested in fat distribution after controlling for overall body size. We further adjusted for birth weight for gestational age z score to assess whether it might be in the pathway linking gestational glucose tolerance with child adiposity (model 5).

Within the multivariable modeling context, we tested the interaction of gestational glucose tolerance and child sex on child body composition. Because we found evidence of an interaction of maternal gestational glucose tolerance with child sex on overall adiposity measured by DXA total fat mass, DXA fat mass index, and SS+TR (all P for interaction ≤0.04) and with BMI, albeit less so (P for interaction = 0.14), we present all the adiposity results separately according to child sex. We did not detect an interaction of gestational glucose tolerance and child sex on DXA fat-free mass (P for interaction = 0.85) or fat-free mass index (P for interaction = 0.48), so we present sex-adjusted results for fat-free mass with boys and girls combined (Supplementary Table 1). We performed all the analyses using SAS version 9.3 (Cary, NC).

Our sample included 43 (4.5%) mothers with GDM, 32 (3.3%) with IGI, 87 (9.1%) with IH, and 796 (83.1%) with normal glucose tolerance in mid-pregnancy (Table 1). Compared with normoglycemic mothers, those with an abnormal gestational glucose tolerance (GDM, IGI, or IH) were older and had higher prepregnancy BMI and lower pregnancy weight gain (Table 1). Mothers with GDM or IGI were more frequently black or Hispanic and more frequently smokers during pregnancy. Offspring of mothers with GDM, IGI, or IH all tended to have a higher birth weight and birth weight for gestational age z score and a lower gestational age compared with the offspring of normoglycemic mothers (Table 1).

At the school-age visit, mean (SD) age was 95 (10) months and mean BMI was 17.1 (2.8) kg/m² in boys and 17.2 (3) kg/m² in girls (P = 0.56). DXA fat mass and percentage fat mass were 6.6 (3.5) kg and 22.2% (5.7), respectively, in boys and 8.0 (3.7) kg and 26.8% (5.8) in girls (P < 0.0001). DXA trunk-to-peripheral fat mass ratio was lower in boys than in girls (0.57 [0.10] vs. 0.59 [0.10], P = 0.04) In boys and girls combined, mean weight or BMI, adjusted for age at examination, sex, and maternal race/ethnicity, did not greatly differ across the categories of gestational glucose tolerance (Table 1). However, offspring of mothers with IGI or GDM had a higher overall adiposity as measured by SS+TR and DXA (DXA percent fat mass: norm, 24.3%; IGI, 26.3%; and GDM, 26.4%). They also had a larger waist circumference (norm, 59.8 cm; IGI, 61.4 cm; and GDM, 61.9 cm) and more trunk fat (norm, 2.5 kg; IGI, 2.9 kg; and GDM, 3.2 kg). Other indicators of overall and central adiposity showed similar trends. Offspring of normoglycemic, IGI, and GDM mothers had similar fat-free mass. Child physical activity and intake of sugar-sweetened beverages and fried food did not materially differ according to maternal gestational glucose tolerance status (Table 1).

Table 2 presents the associations of gestational glucose tolerance with overall and central adiposity measured by DXA, according to child sex. Among boys, unadjusted analyses (model 1) showed that offspring of GDM mothers had a higher total fat mass (2.57 kg [95% CI 0.95–4.2]) and trunk-to-peripheral fat mass ratio (0.05 [0.003–0.09]) than offspring of normoglycemic mothers. Among girls, compared with offspring of normoglycemic mothers, fat mass was higher in female offspring of IGI mothers (2.57 kg [0.22–4.91]) but not among offspring of GDM mothers (−0.31 kg [−2.35 to 1.72]). In contrast to the boys, trunk-to-peripheral fat mass ratio in girls was not different for IGI or GDM (IGI 0.01 kg [−0.05 to 0.07] and GDM 0.00 kg [−0.05 to 0.05]; both vs. normoglycemic), although there was a suggestion of higher DXA trunk-to-peripheral fat mass ratio in offspring of IH mothers (0.03 [0.00–0.06]).

Additional adjustment for maternal prepregnancy BMI (Table 2 [model 2]) and other covariates (models 3 and 4) partially attenuated the associations of GDM (in boys) and IGI (in girls) with total fat mass (1.89 kg [95% CI 0.33–3.45] for GDM in boys and 2.23 [0.12–4.34] for IGI in girls) and, in boys, with trunk-to-peripheral fat mass ratio (0.04 [−0.01 to 0.09]) (Fig. 1).

Table 3 presents the adjusted estimates (model 4) of the association of gestational glucose tolerance with other measures of overall and central adiposity. Consistent with the results shown in Table 2 for the DXA trunk-to-peripheral fat mass ratio, SS/TR was higher in male offspring of GDM mothers (β = 0.34 kg/m² [95% CI 0.04–0.63]) (Table 3). However, we did not detect similar increases with waist circumference or DXA trunk fat mass adjusted for child BMI. In fact, it was the offspring of IGI, not GDM, mothers who showed a greater waist circumference (1.38 [0.20–2.56]).

Associations of gestational glucose tolerance with other measures of overall adiposity (SS+TR, DXA fat mass index, and BMI z score) were similar to those described for DXA total fat mass, although we did not detect an association of gestational glucose tolerance with BMI z score in boys (Table 3).

To assess potential mediation, we additionally adjusted the final model (Table 2 [model 4]) for birth weight for gestational age z score (model 5) and found no attenuation of the association of GDM with fat mass in boys (1.85 vs. 1.89 kg) or the association of IGI with fat mass in girls (2.24 vs. 2.23 kg) (Table 3).

For DXA fat-free mass, the other categories of gestational glucose tolerance did not significantly differ from offspring of normoglycemic mothers (model 1, IH 0.09 kg [95% CI −0.22 to 0.40], IGI 0.86 kg [−0.50 to 2.23], and GDM 1.0 kg [−0.31 to 2.31]). Results were similarly null for fat-free mass index (Supplementary Table 1).

In this cohort study of pregnant women and their children, male offspring of mothers who had GDM, but not IGI, exhibited higher overall adiposity at the school-age visit than offspring of normoglycemic mothers. However, girls of IGI, but not GDM mothers, had higher adiposity.

Although chance is one plausible explanation for this sex difference given the relatively small number of children in each stratum of exposure and outcome, another potential explanation relies on the observation that male and female fetuses seem to have different strategies in utero. Eriksson et al. (30) hypothesized that “boys live dangerously in the womb.” Indeed, male fetuses grow faster in all dimensions than female fetuses (31) as early as the preimplantation stage (32), but they seem to invest less in placental growth. Male placentas are smaller than those of female fetuses at any given birth weight (30). Male placentas are also more efficient; at any placental weight, male fetuses tend to be heavier. This can result in less reserve capacity in the male fetus in the presence of a stressful event (33). In case of shortage of key nutrients, male fetuses attempt to compensate, more or less successfully, by expanding their placental surface in late gestation (30). In utero, boys seem more responsive to the mother’s current diet and metabolism than girls, who appear more influenced by their mother’s lifetime nutrition and metabolism (30,34). As a consequence, male fetuses may be more vulnerable to variations in nutrient intake or transfer during pregnancy. In contrast, female fetuses seem to take a more cautious route that favors survival. They grow more slowly in weight and length and they have a larger placenta for a given birth weight than boys (30). Female neonates also have higher insulin concentrations at birth and higher adiposity for a given birth weight, indicating that they allocate more resources toward fat mass than do male fetuses (35).

These characteristics may result in differential sensitivity to hyperglycemia in pregnancy and in differential long-term programming in male and female fetuses. Ricart et al. (36) showed at birth in >9,000 newborns that GDM was a predictor of macrosomia only in males. In 84 newborns born to GDM mothers, maternal fasting blood glucose was the major predictor of adiposity measured by air displacement plethysmography in male newborns but had little effect on adiposity in females. Conversely, maternal BMI was the primary predictor in female newborns but not in males (37). A recent report from a large randomized controlled trial of treatment for mild GDM by the U.S. Maternal-Fetal Medicine Network reported a greater reduction of birth weight and fat mass in male than in female neonates (38). Studies also reported sex-specific associations with measures of adiposity in childhood. A study of >600 French children born to GDM mothers paired with children born to non-GDM mothers (i.e., normoglycemic, IH, and IGI combined) showed that only the male offspring of GDM mothers had a higher BMI at 5–7 years after adjustment for prepregnancy BMI, gestational weight gain, caloric intake, physical activity, screen time, and a range of sociodemographic variables (14). In contrast to our findings, Krishnaveni et al. (15) showed that female offspring of Indian mothers with GDM had higher adiposity at age 5 and 9.5 years than offspring of mothers without GDM. They did not find an increased adiposity in male offspring of GDM mothers, although they noted some signs of metabolic dysfunction at age 9.5 years. These dissimilar findings could be explained by genetic, cultural, or environmental differences in the two populations. In particular, body composition differs in Indian and white populations (39).

Our results also suggest that the offspring of IGI and IH women in this study are noticeably different from the offspring of normoglycemic mothers. Unfortunately, many earlier studies did not separately examine offspring of IGI mothers, but this group has recently received enhanced attention after results of the Hyperglycemia and Adverse Pregnancy Outcomes (HAPO) study that showed a continuous association of maternal glycemia during pregnancy with offspring adiposity at birth (2). In a sample of >9,000 mother-child pairs, Hillier et al. (40) showed that offspring of IGI mothers, but not IH, had a higher risk of childhood overweight than offspring of normoglycemic mothers, but they did not mention any sex-specific associations. The female-only associations that we show in offspring of IGI mothers have never been reported before. While we did not have data on degree of control or adherence to treatment in women with GDM in this study, a possible explanation for the observed difference is that exposure of the fetuses to excess glycemia may differ in intensity and timing among those whose mothers have IGI versus GDM. In addition, sensitivity may also differ in males and females, which may result in differential long-term programming. However, our results should be tempered by the fact that if we could have used the recent criteria recommended by the International Association of the Diabetes and Pregnancy Study Groups, a large proportion of the women diagnosed with IGI in our study would have been diagnosed with GDM (41).

We did not find an effect of gestational glucose tolerance on fat-free mass in boys or girls.

Similarly, Catalano et al. (42) did not show a significant difference in lean mass measured by DXA in 9-year olds between the offspring of GDM (mean [SD] 25.0 [6.7] kg) and of normoglycemic mothers (24.5 [4.9] kg, P = 0.72). Chandler-Laney et al. (8) showed a positive, albeit modest, correlation of maternal glucose measured during pregnancy with children’s lean mass at 5–10 years of age after adjustment for height and sex (r = 0.37, P = 0.07). It will be important to reevaluate these associations during and after puberty, since muscle mass acquisition takes place largely at puberty in a sex-specific manner.

Previous studies have suggested that the association of exposure to the hyperglycemic environment with later child obesity is only in part explained by its influence on birth weight (3,7,8,29). Our finding that adjustment for birth size did not markedly attenuate associations with increased childhood adiposity confirms these reports.

Strengths of this study include prospectively collected longitudinal data beginning in early pregnancy; detailed assessment of demographic, socioeconomic, and biological family and child characteristics; and research standard measurements of child body composition. We assessed overall and central adiposity using DXA as well as other anthropometric measures.

One limitation of this study is the lack of information on glycemic control in the GDM group. The sample size also prevented us from assessing whether women being categorized as IGI because of an abnormal fasting glucose value or an abnormal 1-h, 2-h, or 3-h postglucose value modifies the association with child adiposity. Indeed, it has been shown that the metabolic implications of impaired glucose tolerance in pregnancy vary in relation to the timing of the abnormal glucose value from the diagnostic OGTT (43). Another limitation is attrition. It is possible that loss to follow-up could have introduced bias, but many baseline variables were similar in the study sample and in those excluded. Generalizability may be limited given the relatively higher socioeconomic status of the study participants.

In conclusion, maternal hyperglycemia in pregnancy was associated with excess adiposity but not greater lean body mass in offspring. In girls, IGI but not GDM was associated with greater offspring adiposity, whereas in boys we found the opposite. A different sensitivity of male and female fetuses in utero may result in differential programming and metabolic consequences in the long run.

This research was supported by grants from the National Institutes of Health (K24 HD 34568, K24 HD069408, R01 HL 64925, R01 HL 68041, R01HL 75504, and P30 DK092924).

The fellowship of N.R. was cofunded by the Société Francophone du Diabète and Lilly. No other potential conflicts of interest relevant to this article were reported.

N.R. developed the study aim, obtained funding, analyzed data, wrote the manuscript, and contributed to interpreting the results and to revision of the article. M.W.G. developed the study aim, obtained funding, and contributed to interpreting the results and to revision of the article. S.L.R.-S. provided statistical insight and contributed to interpreting the results and to revision of the article. E.E. contributed to interpreting the results and to revision of the article. E.O. developed the study aim, obtained funding, and contributed to interpreting the results and to revision of the article. M.W.G. and E.O. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Parts of this study were presented in abstract form at the Pediatric Academic Societies Annual Meeting, Boston, Massachusetts, 28 April–1 May 2012, and at The Obesity Society 30th Annual Scientific Meeting, San Antonio, Texas, 20–24 September 2012.

The authors thank the staff and participants of Project Viva.