Pregnancy Insulin, Glucose, and BMI Contribute to Birth Outcomes in Nondiabetic Mothers

The retrospective cohort (the Cambridge Wellbeing Study) comprised mothers identified from routine clinical data in an electronic database of all deliveries at the Rosie Maternity Hospital, Cambridge, U.K., between 1999 and 2000. The data were merged to a biochemistry database of maternal venous blood glucose levels 60 min after a 50-g oral glucose load at 27–29 weeks of gestation as a routine screening test for gestational diabetes mellitus. Women with 60-min glucose levels >7.8 mmol/l were excluded from this study, as these women would have been considered at high risk of gestational diabetes mellitus (5). We identified data for 3,158 Caucasian women with normal glucose tolerance, who denied smoking or taking any medication at the time of glucose testing and who had a subsequent full-term (≥37 weeks of gestation) singleton delivery. Birth weight was measured at delivery by the midwife.

The prospective cohort (the Cambridge Baby Growth Study) was recruited from mothers attending ultrasound clinics during early pregnancy at the Rosie Maternity Hospital between 2001 and 2006. At 28 weeks of gestation, a 75-g oral glucose load (oral glucose tolerance test [OGTT]) was given. Venous blood glucose and insulin levels were measured at fasting and at 60 min, and capillary blood glucose was measured at 30, 90, and 120 min. Of a total of 708 women who underwent an OGTT, we excluded 34 (4.8%) who had known diabetes, a further 4 women who had a 120-min capillary glucose level >8.6 mmol/l, and 2 women who had fasting venous glucose >6.1 mmol/l. Complete data on glucose and insulin levels and offspring birth weight were available for 668 nondiabetic pregnant women with normal fasting glucose and glucose tolerance. Birth weight was measured by the midwife. Offspring skinfold thicknesses (biceps, triceps, subscapular, and suprailiac) were measured at birth (<7 days old). Measurements were performed in duplicate by trained research nurses using a Harpenden Skinfold Caliper (Holtain, Crosswell, Crymych, Pembrokeshire, U.K.).

Both studies were approved by the local ethics committee, Addenbrooke's Hospital, Cambridge, U.K. For the retrospective Cambridge Wellbeing Study, we only analyzed routinely collected and anonymized data, and therefore consent was not required. For the prospective Cambridge Baby Growth Study, written informed consent was obtained from all mothers by a research nurse.

In both cohorts, mother's whole blood glucose levels were taken from routine measurements in the Clinical Biochemistry Department, Addenbrooke's Hospital, U.K., using a standard glucose oxidase–based assay. In the Cambridge Baby Growth Study, fasting and stimulated insulin levels were measured in the Department of Pediatrics. Insulin was measured by enzyme-linked immunosorbent assay using a commercial kit (DSL, London, U.K.). Sensitivity was 0.26 mU/l. Intra-assay coefficients of variation (CVs) were 4.4 and 5.1% at 10.3 and 35.8 mU/l, and equivalent inter-assay CVs were 8.7 and 2.9%; this assay has no cross-reactivity with proinsulin at levels up to 1,000 pmol/l.

Insulin sensitivity was estimated using homeostasis model assessment (HOMA) (6). To estimate insulin secretion independent of insulin sensitivity, we calculated the disposition index as the insulinogenic index [(insulin 60 − insulin 0)/(glucose 60 − glucose 0)] (7) divided by the insulin resistance index (8). Offspring adiposity at ages 0, 3, 12, and 24 months was estimated as the average of the standardized scores for the four skinfold thickness measurements adjusted for age and sex.

For the retrospective Cambridge Wellbeing Study, glucose levels were categorized into five groups (≤4, 4.1–5.0, 5.1–6.0, 6.1–7.0, and 7.1–7.8 mmol/l). Differences in birth weight and birth outcomes between these five groups were analyzed by ANOVA and χ² tests for trend. Regression analyses were also performed using the continuous variables glucose and glucose² to test for nonlinearity.

For the prospective Cambridge Baby Growth Study, correlations between continuous variables were tested by linear regression, adjusted for potential confounding factors (maternal age, height, smoking in pregnancy, parity, offspring sex, and gestational age). Results of these analyses are displayed as standardized regression coefficients (β). Insulin levels were positively skewed, and logarithms were calculated to achieve normal distributions.

In the retrospective Cambridge Wellbeing study, associations were identified between various birth outcomes and maternal glucose levels within the normal nondiabetic range (glucose level ≤7.8 mmol/l 60 min after a 50-g glucose load) (Table 1). Offspring birth weights increased by (mean ± SEM) 46 ± 8 g for each 1 mmol/l rise in maternal 60-min stimulated glucose level (P < 0.0001, n = 3,158), and a continuous trend was apparent across the range of normal glucose levels (Table 1). Each 1 mmol/l increase in maternal postchallenge glucose levels was also associated with a rise in the absolute risk of assisted vaginal deliveries (2.1 ± 0.8%, P = 0.006), emergency cesarean delivery (3.4 ± 0.8%, P < 0.0001), and elective cesarean delivery (3.1 ± 0.7%, P < 0.0001) (Table 1). No significant deviation from linearity was observed for any outcome (data not shown).

Maternal characteristics of the prospective Cambridge Baby Growth Study are displayed in Table 2. Among women with normal glycemia (fasting glucose <6.1 mmol/l and 120-min post–75-g OGTT glucose level ≤8.6 mmol/l and reportedly nondiabetic), distinct effects of maternal glucose levels and prepregnancy BMI were seen on infant birth size (Fig. 1). Mother's fasting glucose level was more strongly related to offspring skinfolds (β = 0.207, P < 0.0001) and weight at birth (β = 0.180, P < 0.0001) than to birth length (β = 0.097, P = 0.02). Conversely, mother's prepregnancy BMI was independently related to offspring birth length (β = 0.104, P = 0.009) and birth weight (β = 0.090, P = 0.04) but not to skinfolds at birth (β = 0.048, P = 0.2) (Fig. 1).

The risk of fetal macrosomia (birth weight >90th centile for gestational age and sex) was independently related to both the mother's fasting glucose (odds ratio 2.61 per +1 mmol/l [95% CI 1.15–5.93]) and prepregnancy BMI (1.10 per +1 kg/m² [1.04–1.18]).

Offspring skinfolds at birth were more strongly related to fasting glucose (β = 0.222, P < 0.0001) than to 30-min glucose (β = 0.086), 60-min glucose (β = 0.094), 90-min glucose (β = 0.074), or 120-min glucose (β = 0.114, all P < 0.05). Beyond the effect of fasting glucose, stimulated glucose levels did not further contribute to the variance in offspring skinfolds at birth (data not shown).

However, independent of the mother's glucose levels, the mother's insulin sensitivity (β = −0.129, P = 0.01) and insulin secretion (disposition index, β = −0.137, P = 0.02) were both inversely related to offspring skinfolds at birth (Table 3).

In the prospective Cambridge Baby Growth Study, prepregnancy BMI correlated positively with offspring adiposity at ages 12 and 24 months (both P < 0.05). However, the correlation between the mother's glycemia and offspring adiposity, which had been apparent at birth, disappeared from age 3 months onward (Fig. 2). Accordingly, higher maternal fasting glucose levels were associated with a decline in offspring adiposity between birth to age 12 months (β = −0.218, P < 0.0001).

In two contemporary U.K. cohorts, we observed that increasing maternal glucose levels within the normal nondiabetic range were consistently related to larger offspring birth size and increased risk of interventional deliveries. Maternal glucose levels specifically enhanced offspring adiposity at birth, and additional influences of maternal insulin sensitivity and insulin secretion were apparent. Postnatally, the effect of maternal glycemia on offspring adiposity rapidly disappeared during infancy, and over the same period a positive effect of maternal BMI became apparent.

Although gestational diabetes mellitus has long been associated with fetal macrosomia (9), our data support a growing number of studies that report a continuous effect of maternal glucose levels, even in the absence of maternal diabetes, on offspring birth weight and pregnancy outcomes (1,10–13). Most recently, the international multicenter HAPO Study reported that maternal fasting, 60-min, and 120-min glucose levels all showed similar linear effects on risks for macrosomia, cesarian delivery, neonatal hypoglycemia, and insulinemia at birth (1). Such findings indicate that the current criteria used to diagnose and manage gestational hyperglycemia need urgent reconsideration.

Our study shows for the first time that in addition to maternal glycemia, differences in maternal insulin sensitivity, insulin secretion, and BMI also have an impact on offspring size and adiposity. These findings suggest that maternal energy substrates other than glucose, such as free fatty acids or amino acids, may have an important independent influence on fetal growth. Although we used the HOMA model of fasting insulin sensitivity, smaller studies using the insulinemic-euglycemic clamp method showed that maternal insulin sensitivity during pregnancy contributes to both offspring body fat and fat-free mass (14).

The effect of maternal overweight or obesity on larger offspring size has been attributed mainly to higher pregnancy glucose levels. We now show that, independent of glycemia, prepregnancy BMI appears to influence offspring body length and weight, rather than adiposity. In contrast, the major effect of maternal glycemia on fetal growth was on adiposity (15). Fetal length may be determined at a very early stage during gestation, whereas increased adiposity may result largely from glucose transport to the fetus during the third trimester. In animal models, maternal nutrition during the periconceptual period may have important effects on subsequent fetoplacental growth (16). In humans, differences in fetal length may therefore reflect early gestational placentation and periconceptual maternal nutrition. Our findings therefore indicate that the overweight or obese mother might somehow enhance early placentation and early transport of nutrient to the fetus, possibly by affecting gene methylation (17).

Offspring of diabetic pregnancies appear to have increased long-term risks of obesity (18,19) and impaired glucose tolerance (18). It has been suggested that fetal exposure to glucose increases risks of overweight and type 2 diabetes in later life (13). However, such observational studies cannot completely separate the specific effects of gestational glucose levels from inherited genetic susceptibility to overweight and type 2 diabetes. Our data show that the effect of maternal glycemia on offspring adiposity is only transitory and disappears soon after birth. Therefore, any deleterious long-term effects of fetal glucose exposure on type 2 diabetes risk must be more subtle, possibly affecting fetal β-cell capacity or adipocyte distribution and function rather than overall adiposity. In contrast, we observed that the “transgenerational” effects of maternal BMI on offspring adiposity emerge early in infancy and could be explained by transmission of genetic susceptibility to obesity (20) or by parental influences on infant nutrition.

In summary, the linear nature of the associations between maternal glycemia and adverse birth outcomes leads to uncertainties in the optimal definition of gestational hyperglycemia. In addition to intensive interventions in high-risk individuals, according to epidemiological principles, such linear associations indicate that population-based strategies to reduce glycemia in all pregnant women would have a greater impact on reducing adverse pregnancy outcomes in similar Westernized settings (21). Furthermore, the aim of such strategies should also be to reduce maternal prepregnancy BMI and improve insulin sensitivity as these factors may influence fetal growth independent of maternal glycemia.

This project was supported by a research grant from Wellbeing, Royal College of Obstetricians and Gynaecologists, U.K. The Cambridge Baby Growth Study was supported by the European Union Framework 5, the World Cancer Research Fund, and the Medical Research Council. B.D. is supported by the Throne-Holst Foundation, Uppsala University, Her Royal Highness the Crown Princess Lovisa's Foundation for Children's Health Care, the Wera Ekstrom Foundation, and the Swedish Society of Medicine. We also acknowledge the support of the Cambridge Biomedical Research Center.

We thank all the families in the Cambridge Baby Growth Study, research nurses, staff at the Addenbrooke's Wellcome Trust Clinical Research Facility, and midwives at the Rosie Maternity Hospital. We thank Angie Watts for generous technical assistance.