Contribution of Gestational Weight Gain on Maternal Glucose Metabolism in Women With GDM and Normal Glucose Tolerance

Fernanda L. Alvarado; Perrie O'Tierney-Ginn; Patrick Catalano


J Endo Soc. 2021;5(2) 

In This Article


Study Design

This is a secondary analysis from a prospective observational cohorts of women recruited before a planned pregnancy and followed through delivery. All of the subjects in the primary analysis were included in the present work. The primary objectives of the original studies were to characterize the longitudinal changes in maternal carbohydrate metabolism, body composition, and energy expenditure[17–22] in women with NGT and those developing GDM.

The original study procedures were conducted at Medical Center Hospital of Vermont, and at MetroHealth Medical Center, Cleveland, Ohio. Study visits in the clinical research units were conducted before a planned pregnancy during the follicular phase of the menstrual cycle, in early (12 to 14 weeks), and in late (33 to 36 weeks) pregnancy. For the purpose of this analysis, we only used the pre-pregnancy and late pregnancy data. The research protocol was approved by the 2 institutional review boards and written informed consent was obtained from each participant.

The population consisted of healthy women who were planning a pregnancy, not breastfeeding, and not using any contraception. They were nonsmokers with no known preexisting cardiometabolic disorders (hypertension, diabetes, thyroid disorders). Before conception, all subjects were given a 75-g oral glucose tolerance test (OGTT) as defined by the National Diabetes Data Group to confirm absence of preexisting diabetes. In the preconception period, all women were screened for GDM risk factors. Twelve of the women had a history of GDM or an abnormal glucose screening test in a prior pregnancy; and 25 had a family history of type 2 diabetes.[20] Women who developed GDM were diagnosed at the time of routine third-trimester screening.

Study Procedures

Two weeks prior to the preconception visit, women were instructed to follow a dietary regimen designed to standardize nutritional intake for each subject, to maintain weight before conception, and to allow appropriate weight gain during pregnancy. This regimen was identical to the diet employed in the treatment of gestational diabetes at our institution; which consisted of approximately 50% complex carbohydrates, 30% fat with an emphasis to avoid saturated fats, and 20% protein.[19]

The following tests were sequentially performed during a 3-day protocol: day 1, OGTT and islet cell antibodies; day 2, body composition and intravenous glucose tolerance test (IVGTT); and day 3, hyperinsulinemic-euglycemic clamp.[23] Detailed descriptions of the study procedures have been published.[17,18] Brief descriptions of the study procedures follow. At the time of the pre-pregnancy visit, subjects were matched for pre-pregnancy body mass index (BMI). We elected to estimate body composition because BMI in nonpregnant individuals has a wide variation when correlated with fat mass or percent body fat, and the relationships are less robust with advancing gestation.[24,25]

BMI was calculated using the subject's weight in kilograms divided by the square of her height in meters. Body composition was estimated by underwater weighing with adjustment for residual lung volume by helium dilution. The percent body fat was calculated according to Keys and Brozek to estimate fat-free mass.[18,19,26] Data are described as total kilograms of fat mass (FM), kilograms of fat-free mass (FFM), and percentage body fat (%BF) (kg of fat mass/kg of total body weight × 100).

Oral glucose tolerance test: pre-pregnancy subjects were given a 75-g OGTT using National Diabetes Data Group criteria for diabetes mellitus. During pregnancy women were given a 100-g OGTT. GDM was classified according to the Carpenter and Coustan criteria.[27] Venous plasma glucose concentrations were determined by the glucose oxidase method with a Yellow Springs glucose analyzer (Yellow Springs, OH).

Insulin sensitivity was assessed using the hyperinsulinemic-euglycemic clamp. Briefly, peripheral insulin sensitivity was assessed after an 11-hour fast.[28] Infusion of [6,6-2H2] glucose for 2 hours was used to estimate basal endogenous glucose production. At the end of the 2-hour period, the clamp procedure was initiated with a constant infusion of 40 mU m−2min−1 of insulin, maintaining plasma insulin concentration of approximately 600 pmol m−2 mL−1. Plasma glucose was maintained at 90 mg/dL (5.0 mmol/L). The glucose and insulin concentrations were collected every 10 minutes during the last 40 minutes of the clamp; they were averaged and used to estimate the glucose disposal rate. Residual endogenous glucose production during the clamp was estimated by adding a calculated amount of [6,6-2H2]glucose to the 20% glucose infusion as described by Tserng and Kalhan.[23] The insulin (measured by radioimmunoassay) intra-assay coefficient of variation was 6% and the inter-assay coefficient of variation was 8%.[17]

The estimated insulin sensitivity was defined as the glucose infusion rate required to maintain plasma glucose at 90 mg/dL (5.0 mmol/L) during the clamp. To account for the lack of complete suppression of endogenous glucose production during the clamp in women developing GDM and the variability of plasma insulin concentrations during the clamp across time and between groups, we defined insulin sensitivity as the glucose infusion rate plus any residual endogenous glucose produced per kg of FFM during the clamp divided by the mean insulin concentration achieved during the hyperinsulinemic-euglycemic clamp.[23,29]

The first-phase insulin response was measured using IVGTT. Briefly, insulin response was measured in participants with less than 120% ideal body weight (n = 35) by infusing 0.5 g/kg glucose over a 3-minute bolus. Samples of glucose and insulin were obtained at baseline and at 1, 3, 5, 10, 15, 30, 45, and 60 minutes. For participants whose weight was greater than 120% ideal body weight (n = 11), the glucose bolus was 19 g/m2 body surface area. Using the trapezoidal rule, we estimated the first-phase response as the area under the curve (AUC) from 0 to 5 minutes.[23]

Disposition index was calculated to describe the degree of beta-cell compensation for decrements in insulin sensitivity. The disposition index is the product of the first-phase insulin secretory response (IVGTT insulin response) and the insulin sensitivity index (hyperinsulinemic-euglycemic clamp).

Statistical Analysis

Data describing maternal preconception demographics in GDM and NGT women are presented as mean ± SD. Difference between groups in absolute (Δ), from preconception to late pregnancy, and relative change (%Δ), which express the absolute change as a percentage based of the preconception period, were analyzed using paired Student t test. Frequency data were analyzed by Chi-square test. Maternal insulin sensitivity, insulin response, disposition index, and body composition (body weight, FM, %BF, FFM) were reported as the Δ and %Δ across those 2 time points.

All the absolute change (Δ) estimates were standardized and normalized using the Yeo-Johnson power transformation. The Yeo-Johnson power transformation was applied because it handles both positive and negative values.

All variables had complete data (n = 46), with the exception of insulin response (n = 33) and disposition index (n = 33). In order to treat missing data, linear regression was estimated using full information maximum likelihood (FIML). FIML is a validated statistical treatment for missing data and it accounts for the "missing at random" and "missing completely at random" assumptions by using all the available information and the same patterns as if there were no missing data. These results are similar to the multiple imputation procedure.[30,31] The R package lavaan was used to perform this analysis.[32]

Using simple linear regression, we analyzed whether Δ insulin sensitivity, insulin response, and disposition index were associated with maternal Δ in body weight, FM, %BF, and FFM, in both NGT and GDM women.

A multilinear regression model was assessed to explore whether there were other potential mediators of pregnancy that could be associated with the Δ of maternal insulin sensitivity, insulin response, and disposition index. We included maternal age, estimated gestational age at delivery, parity, maternal pre-pregnancy weight, and study group (NGT and GDM) as covariates. Residuals from models were checked for conformance to assumptions of normality and homeostasis. Models were assessed for multicollinearity by scoring the variance inflation factor.

For descriptive data, statistics were reported at the raw data level for ease of interpretation. Statistical analyses were performed with R studio, Boston, MA, version 3.6.2.[33] P values ≤ 0.05 were considered statistically significant.