Insulin Sensitivity and β-Cell Function During Early and Late Pregnancy in Women With and Without Gestational Diabetes Mellitus

Gestational diabetes mellitus (GDM), defined as increased fasting plasma glucose level (≥92 mg/dL) or increased plasma glucose level at 1 h (≥180 mg/dL) or 2 h (≥153 mg/dL) during a 75-g oral glucose tolerance test (OGTT) (1), occurs in ∼20% of pregnancies (2). GDM is typically diagnosed during mid- to late pregnancy, and most women with GDM do not have diabetes after delivery, although subclinical metabolic dysfunction can persist postpartum (2,3). GDM increases the risk of adverse obstetric outcomes and future type 2 diabetes and cardiovascular disease in both mother and child (2,4). Excess maternal adiposity is a major risk factor for GDM (2,3,5), but the specific metabolic alterations that cause GDM are not clear because of conflicting results from different studies. It has been proposed that GDM is caused by 1) greater insulin resistance (IR) than normally occurs during pregnancy (6–8) or 2) preexisting β-cell dysfunction that is identified during metabolic stress of pregnancy because of the failure of dysfunctional β-cells to adequately increase glucose-stimulated insulin secretion needed to maintain normal glycemia (3,6–8). However, these purported mechanisms are based on data from cross-sectional studies that compared β-cell function in women with and without GDM or studies that 1) compared β-cell function during late pregnancy and postpartum, 2) evaluated β-cell function by measuring plasma insulin concentration, rather than insulin secretion, and 3) included women who had GDM in a previous pregnancy, which could have influenced the results because GDM increases the risk of persistent metabolic dysfunction (2). We are not aware of any studies that evaluated the longitudinal gestational changes in both insulin sensitivity (IS) and β-cell function associated with GDM in women without a history of GDM and/or a family history of type 2 diabetes; this information is needed to reliably assess the physiological factors involved in GDM pathogenesis.

The purpose of the present study was to evaluate plasma glucose concentrations during fasting conditions and after glucose ingestion and the factors involved in regulating plasma glucose (IS with respect to glucose metabolism and β-cell function [β-cell responsiveness to glucose], assessed by using a frequently sampled OGTT) during early (15 weeks’ gestation) and late (35 weeks’ gestation) pregnancy in women with overweight or obesity who developed or did not develop GDM. We also evaluated plasma total free fatty acid (FFA) concentration during fasting conditions and after glucose ingestion and adipose tissue IS because of the important interactions in IS with respect to both glucose and fatty acid metabolism (9). We hypothesized that 1) IS with respect to glucose metabolism (Matsuda-IS index), adipose tissue IS (adipose tissue-IR [Adipo-IR] index), and β-cell responsiveness to glucose would be worse at 15 weeks’ gestation in women who develop GDM (GDM group) compared with those who do not develop GDM (No-GDM group); and 2) the pregnancy-induced decrease in IS would be greater and the pregnancy-induced increase in β-cell function would be blunted in the GDM compared with the No-GDM group.

A total of 195 women participated in this study, which represents a secondary analysis of a randomized clinical trial (10) that was approved by the Washington University Institutional Review Board and registered in ClinicalTrials.gov (registration no. NCT01768793). The parent trial included 267 participants and used an intention-to-treat analysis. In this secondary analysis, we only included the 195 participants who completed the frequently sampled OGTT at both 15 and 35 weeks’ gestation. Eligibility criteria included 1) self-identified as Black; 2) Medicaid recipient or living in a zip code associated with a median household income below the poverty level; 3) 18–45 years old; 4) BMI 25.0–45.0 kg/m² measured at the 15-week study visit; 5) singleton viable gestation at 15 0/7 weeks; and 6) glycosylated hemoglobin <6.5%. Exclusion criteria included diabetes or history of diabetes or GDM and use of medications that could affect study outcomes. A complete list of inclusion and exclusion criteria was reported previously (10). We specifically chose Black women with overweight or obesity who were Medicaid recipients or lived in a high-poverty zip code in the parent study, because of the need to develop interventions in this population, which is at high risk for adverse pregnancy-related outcomes and is often underrepresented in research studies.

The OGTT at the 35-week study visit was used to identify GDM and assign participants to the GDM and No-GDM groups. Among the 195 women in this study, 30 (15%) had GDM at 35 weeks, defined as fasting plasma glucose ≥92 mg/dL or plasma glucose ≥180 mg/dL at 1 h or ≥153 mg/dL at 2 h during a 75-g OGTT (1). Seven of 30 participants with GDM at 35 weeks and 12 of 165 participants without GDM at 35 weeks were diagnosed with GDM by their obstetricians at ∼24–28 weeks’ gestation (Supplementary Fig. 1). One participant in the GDM group and one in the No-GDM group were treated with insulin after this diagnosis was made and these women were excluded from the analyses; none of the other 29 women in the GDM group or the other 11 women in the No-GDM group received medications for GDM. The final study cohort included 193 participants (n = 29 in the GDM group and 164 in the No-GDM group). Participants diagnosed with GDM by their obstetrician before 35 weeks’ gestation received lifestyle (diet, physical activity) counseling. Excluding the 11 participants in the No-GDM group diagnosed with GDM by their obstetrician from the analysis did not affect any of the study conclusions. Eleven of the 29 participants in the GDM group and 18 of the 164 participants in the No-GDM group met criteria for GDM at 15 weeks’ gestation (Supplementary Fig. 1). Three of these 18 participants were diagnosed with GDM by their obstetrician before 35 weeks’ gestation (Supplementary Fig. 1). The reason why these 18 women no longer had GDM at 35 weeks is unclear, but this could be due to the variability in OGTT results because most of these women barely met criteria for GDM at 15 weeks’ gestation. The participants’ obstetricians were blinded to the 15-week OGTT results, so GDM therapy was not initiated on the basis of these results.

The parent study randomly assigned participants to receive standard Parents-as-Teachers (PAT) education or PAT education plus lifestyle counseling (PAT+). The standard PAT curriculum provides information on child development and parent–child interactions during 10 home visits. The PAT+ curriculum added diet and physical activity counseling to the standard PAT curriculum to help prevent excessive gestational weight gain. The incidence of GDM did not differ between the PAT and PAT+ groups; 16 of the 95 participants (17%) assigned to the PAT group and 13 of the 98 participants (13%) assigned to the PAT+ group in the present study developed GDM. Therefore, the percentages of participants assigned to PAT and PAT+ in the GDM group were not different from the percentages in the No-GDM group (55% in the GDM group and 49% in the No-GDM group were assigned to standard PAT; 45% in the GDM group and 51% in the No-GDM group were assigned PAT+).

Participants were admitted to the Clinical Translational Research Unit at 15 and 35 weeks’ gestation, after they fasted (10–12 h) overnight. Blood samples were obtained immediately before and at 10, 20, 30, 60, 90, and 120 min after ingesting a 75-g glucose drink. Plasma glucose, insulin, C-peptide, and FFA concentrations were determined as previously described (11,12). IS with respect to glucose metabolism was assessed using the Matsuda-IS index, which correlates with IS assessed by the hyperinsulinemic-euglycemic clamp procedure in pregnant women (13,14). The Adipo-IR index was calculated as the product of fasting plasma FFA and insulin concentrations (12). Insulin secretion rate (ISR) was calculated by fitting the plasma C-peptide concentration values during the OGTT to a two-compartment model with population-based C-peptide kinetic parameters and volume of distribution values (15). At 35 weeks’ gestation, the C-peptide volume of distribution was assumed to be 1.4 times the value at 15 weeks because plasma volume expands by ∼40% between 15 and 35 weeks’ gestation (16). Plasma insulin clearance rate (ICR) at time zero was calculated by dividing the basal ISR by the basal plasma insulin concentration; the ICR after glucose ingestion was calculated as the integral of the insulin disposal rate to plasma insulin concentration ratio between two time points (17). The insulin disposal rate was calculated from the ISR and the change in the insulin pool size (assuming a distribution volume of 117 mL/kg in women with overweight or obesity [18]) at 15 weeks, and 1.4 times that value at 35 weeks because of fluid expansion (16). β-Cell function, defined as β-cell responsiveness to glucose, was assessed during the first 30 min of the OGTT when plasma glucose was increasing to peak values.

The statistical significance of differences in metabolic variables at 15 weeks’ gestation and gestational weight gain between the GDM and No-GDM groups was evaluated by using the Student t test for independent samples. The statistical significance of differences in metabolic outcomes at 35 weeks’ gestation between the GDM and the No-GDM groups was evaluated by ANCOVA, with the 35-week value as the dependent variable, group as the independent variable, and the 15-week value of the metabolic outcome as the covariate. Repeating the statistical analysis with parent study treatment (i.e., PAT or PAT+) as an additional factor did not affect any of the conclusions. Skewed data sets were log-transformed before analyses to achieve a normal distribution. Group differences in β-cell function were evaluated by using a generalized linear mixed model with 1) ISR as the dependent variable; 2) group, time during the OGTT, and testing visit (15 weeks and 35 weeks) as fixed factors; and 3) plasma glucose concentration and/or the Matsuda-IS index as covariates. Receiver operating characteristic (ROC) analysis evaluated the prediction accuracy of the metabolic variables that were different between the GDM and No-GDM groups at 15 weeks’ gestation. Youden’s J statistic was used to determine optimum cutoff values for risk prediction. The positive predictive value for these cutoff values was calculated as the number of true-positive cases divided by the sum of the number of true-positive and false-positive cases assuming the prevalence of GDM observed in our study. A P value of ≤0.05 was considered statistically significant. Statistical analyses were performed by using SPSS Statistics, version 26.

The primary study outcome was IS assessed with the Matsuda-IS index. On the basis of the data from a study that found IS was ∼30% lower in women with GDM than those without GDM (19), we estimated that our study would be able to detect a 30% difference with a two-tailed test, an α-value of 0.05, and a power of 0.91.

Participants’ age, prevalence of overweight and obesity, and BMI at 15 weeks’ gestation were not different between the GDM and the No-GDM groups (Table 1). Gestational weight gain tended to be greater in the GDM than in the No-GDM group (Table 1).

At 15 weeks, 191 of 193 participants had normal fasting plasma glucose concentration (<100 mg/dL [20]), whereas two (n = 1 each in the No-GDM and GDM groups) had impaired fasting plasma glucose concentration (≥100 mg/dL [20]). However, the fasting plasma glucose level was higher in the GDM than in the No-GDM group (P < 0.05; Table 1). Glucose tolerance, defined as plasma glucose at 2 h during the OGTT (20), was not different between the GDM and the No-GDM groups (Table 1, Fig. 1A) and most participants (19 of 29 [66%] in the GDM group and 134 of 164 [82%] in the No-GDM group) had normal glucose tolerance (plasma glucose <140 mg/dL at 2 h after ingesting 75 g of glucose [20]) at 15 weeks. However, plasma glucose at 1 h after glucose ingestion was ∼20 mg/dL higher and total plasma glucose area under the curve between 0 and 120 min (AUC_0–120) was higher in the GDM than in the No-GDM group (Table 1, Fig. 1A).

Fasting plasma glucose and both the 2-h plasma glucose value and the plasma glucose AUC_0–120 during the OGTT decreased between 15 and 35 weeks in the No-GDM group and increased in the GDM group (Table 1, Fig. 1A). Plasma glucose at 1 h after glucose ingestion was not different at 15 and at 35 weeks in the No-GDM group, but increased in the GDM group (Table 1, Fig. 1A).

Fasting plasma FFA concentration at 15 weeks was not different between the GDM and the No-GDM groups and did not change between 15 and 35 weeks in either group (Table 1, Fig. 1B). Plasma FFA decreased at 1 h and 2 h after glucose ingestion in both the GDM and No-GDM groups at 15 weeks, without a difference between groups (Fig. 1B). However, the decrease in FFA concentration after glucose ingestion was blunted at 35 weeks compared with 15 weeks in both the No-GDM and the GDM groups, and both the 1-h and 2-h values were greater in the GDM group than in the No-GDM group (Fig. 1B).

Fasting plasma insulin and C-peptide concentrations, basal ISR, basal ICR, and the AUC_0–120 for plasma insulin ISR and ICR at 15 weeks were not different between the GDM and No-GDM groups (Table 1 and Fig. 1C–E). All of these outcomes were higher at 35 than 15 weeks in both the No-GDM and GDM groups, but the increases in insulin and C-peptide concentrations and ISR were greater in the GDM than in the No-GDM group (Table 1 and Fig. 1C–E). β-Cell function, assessed as β-cell responsiveness to glucose (Fig. 1F) and β-cell responsiveness to glucose adjusted for IS (Supplementary Fig. 2), which evaluate ISR in the context of the prevailing plasma glucose concentration and, additionally, IS, were lower in the GDM than in the No-GDM group at 15 weeks (P < 0.05). β-Cell responsiveness to glucose and β-cell responsiveness to glucose adjusted for IS increased in both the No-GDM and the GDM groups between 15 and 35 weeks, but the increases were greater in the No-GDM than in the GDM group (P < 0.05) (Fig. 1F and Supplementary Fig. 2).

IS with respect to glucose metabolism, assessed with the Matsuda-IS index, at 15 weeks was lower in the GDM than in the No-GDM group (Table 1). Moreover, IS decreased between 15 and 35 weeks in both the No-GDM and the GDM groups, and the decrease was greater in the GDM than in the No-GDM group (Table 1). The Adipo-IR index at 15 weeks was not different between the No-GDM and the GDM groups (Table 1). The Adipo-IR index increased in both the No-GDM and the GDM groups between 15 and 35 weeks, and the increase was greater in the GDM than in the No-GDM group (Table 1).

There were significant differences in several metabolic outcomes between the GDM and No-GDM groups at baseline. The ROC analysis found fasting plasma glucose, plasma glucose at 1 h during the OGTT, and the glucose concentration AUC_0–120 during the OGTT at 15 weeks, but not the Matsuda-IS index at 15 weeks, predicted GDM with ≥95% confidence (Fig. 2). Moreover, this analysis showed a fasting plasma glucose concentration >82 mg/dL and a 1-h plasma glucose concentration >150 mg/dL during the OGTT were the best predictors of GDM with ∼65% certainty (true positive) and ∼35% false-positive rates and positive predictive values of 24% (fasting plasma glucose) and 29% (1-h glucose).

At 15 weeks, 18 of the 164 participants in the No-GDM group and 11 of the 29 participants in the GDM group met GDM criteria (Supplementary Fig. 1). Between the No-GDM group and the GDM group, age (mean ± SD, 29 ± 6 and 29 ± 6 years), weight (mean ± SD, 93 ± 18 and 94 ± 19 kg), BMI (mean ± SD, 34 ± 5 and 34 ± 6 kg/m²) and plasma glucose and insulin concentrations during the OGTT (Supplementary Fig. 3) were not different. Participants who met GDM criteria at 15 but not 35 weeks gained ∼45% less weight between 15 and 35 weeks than participants who met GDM criteria at both 15 and 35 weeks (met criteria at 15 weeks, 6.2 ± 4.1; met criteria at both weeks, 11.1 ± 7.3 kg). Plasma glucose concentration during the OGTT decreased in those who met GDM criteria at 15 but not 35 weeks, whereas plasma glucose concentration did not change, or they increased in participants who met GDM criteria at both 15 and 35 weeks (Supplementary Fig. 3). Plasma insulin concentration at 35 weeks were not different between participants who no longer met GDM criteria at 35 weeks and participants who met GDM criteria at both 15 and 35 weeks (Supplementary Fig. 3).

The results from our study demonstrate that fasting and 1-h OGTT plasma glucose values at 15 weeks’ gestation were higher and dynamic measures of IS with respect to glucose metabolism and β-cell responsiveness to glucose were worse in women who had GDM at 35 weeks’ gestation compared with those who did not have GDM at 35 weeks. Moreover, between 15 and 35 weeks’ gestation, women with GDM had a greater decline in IS and a blunted increase in β-cell function than did women without GDM, which resulted in an increase in fasting and postprandial plasma glucose in those with GDM compared with a decrease in those without GDM. In addition, at 35 weeks’ gestation, the decline in plasma FFA concentration after glucose ingestion was blunted and the increase in Adipo-IR was greater in the GDM group than in the No-GDM group, demonstrating GDM is also associated with an increase in Adipo-IR. A ROC analysis showed fasting plasma glucose and plasma glucose concentration at 1 h after glucose ingestion or total postprandial plasma glucose concentration (AUC_0–120) during the OGTT at 15 weeks’ gestation were predictors of GDM at 35 weeks’ gestation. Together, these data demonstrate that abnormalities in insulin action and β-cell function are present very early during pregnancy in women who develop GDM, and data from previous studies show these abnormalities are even present before pregnancy (8). The alterations in glycemic control that are present early during pregnancy, in conjunction with a greater decline in IS and impaired increase in β-cell function during continued gestation, are responsible for the development of GDM.

The reason(s) for the greater decrease in IS and the blunted increase in β-cell function during the course of pregnancy in the GDM group compared with the No-GDM group is not clear but likely involves differences in placental hormone secretion and the response to placental hormones. Placental hormones (primarily growth hormone, lactogen, and sex steroids) trigger changes in maternal physiological functions (e.g., mammary enlargement, plasma volume expansion, reduced IS, increased β-cell function due to enhanced β-cell viability, hyperplasia, hypertrophy) that are necessary to support fetal growth and prepare for lactation. The increase in plasma FFA in both the No-GDM and the GDM groups at 35 weeks compared with 15 weeks’ gestation could have also contributed to the decrease in IS in both groups (9). The decrease in IS attenuates the risk of hypoglycemia induced by increased β-cell insulin secretion, plasma volume expansion, and fetal glucose use during pregnancy; there is normally a small decrease in fasting and postprandial blood glucose concentration during pregnancy (6,7,11,21,22). Although GDM is associated with alterations in placental structure (e.g., increased weight, increased angiogenesis, delayed villous maturation) and hormone gene expression profiles, circulating placental hormone concentrations are typically not different in women with and without GDM (23,24). However, GDM is associated with impaired pregnancy-induced β-cell adaptations, presumably because of an inadequate response to hormonal stimuli (25). GDM is also associated with single nucleotide polymorphisms within prolactin receptors that are involved in regulating glucose homeostasis in pregnancy (26,27).

Both β-cell responsiveness to glucose and IS during early (15 weeks’ gestation) and late (35 weeks’ gestation) pregnancy were worse in the GDM group than in the No-GDM group. In addition, the increase in β-cell function between 15 and 35 weeks was less, whereas the decrease in IS was greater, in the GDM compared with the No-GDM group. Therefore, the insulin secretory response in relation to IS was worse in the GDM than in the No-GDM group, resulting in higher fasting and OGTT plasma glucose concentrations and a greater increase in fasting and OGTT plasma glucose concentrations between 15 and 35 weeks in the GDM than in the No-GDM group. In contrast, the increase in β-cell function at 35 weeks compensated for the decrease in IS in the No-GDM group, so fasting and OGTT glucose concentrations at 35 weeks were lower than at 15 weeks. The reason why some women met GDM criteria at 15 but not at 35 weeks is unclear but could be related to ∼45% less gestational weight gain than in women who met GDM criteria at both 15 and 35 weeks.

At 15 weeks’ gestation, fasting plasma glucose and plasma glucose concentration at 1 h, but not at 2 h, during the OGTT were higher in the GDM group than in the No-GDM group. These findings are consistent with observations in the general population that demonstrate 1) progressive increases in fasting plasma glucose concentration are associated with progressive increases in the risk of developing type 2 diabetes (28–30); and 2) plasma glucose at 1 h during an OGTT is a stronger predictor of developing type 2 diabetes than is the 2-h value (31–34). Furthermore, the risk for future type 2 diabetes in the mother is more closely associated with increased fasting and 1-h than 2-h OGTT glucose concentration during pregnancy (35). Our ROC analysis found that both fasting and 1-h OGTT plasma glucose concentration can predict GDM, but the predictive power was weak, thereby limiting its clinical utility.

Our study has some limitations. We did not evaluate metabolic function before conception. Therefore, we do not know whether the metabolic dysfunction observed in the GDM group at 15 weeks’ gestation was induced by pregnancy or was already present before conception, as in studies that included women with a family history of type 2 diabetes and/or a history of GDM (8,19,36). Our study included only Black women with overweight or obesity who were Medicaid recipients and/or living in a high-poverty zip code, so our findings might not translate to other populations. Our study represents a secondary analysis of data from a randomized controlled trial that involved lifestyle counseling. However, the incidence of GDM was not different between the intervention and control groups and equal proportions of women in the GDM and No-GDM groups were assigned to the intervention. We used the Matsuda-IS index to evaluate IS, which is not as precise as measuring IS by using the hyperinsulinemic-euglycemic clamp procedure. Nonetheless, we still found significant differences between our study groups. We identified the GDM group on the basis of an OGTT at 35 weeks’ gestation, whereas an evaluation for GDM in clinical practice is typically conducted between 24 and 28 weeks. Finally, we did not measure glucagon; previous studies have shown that fasting plasma glucagon and plasma glucagon concentration after glucose ingestion are 10–15% higher in women with GDM than in those without GDM (39,40). Therefore, hyperglucagonemia could have contributed to the higher plasma glucose concentration in the GDM than the No-GDM group.

In summary, women with overweight or obesity who have GDM late during pregnancy, compared with those who do not, demonstrate abnormalities in major factors that regulate glucose homeostasis (IS and β-cell function) early during pregnancy and have a greater decrease in IS with respect to both fatty acid and glucose metabolism and a blunted increase in β-cell responsiveness to glucose during pregnancy than women who do not develop GDM. Although both fasting plasma glucose and plasma glucose concentration at 1 h during an OGTT were higher during early pregnancy in women who had GDM late during pregnancy compared with those who did not, the predictive power of plasma glucose concentration was weak.

Acknowledgments. The authors thank the staff of the Center for Human Nutrition and the Department of Obstetrics and Gynecology at Washington University School of Medicine and the Clinical and Translational Research Unit for assistance in conducting the metabolic studies and their technical assistance in processing the study samples. The authors also thank the study participants for their participation.

Funding. This study was supported by National Institutes of Health (NIH) grants U01 DK094416, P30 DK056341 (Washington University Nutrition and Obesity Research Center), P30 DK092950 (Washington University Center for Diabetes Translation Research), P30 DK20579 (Washington University Diabetes Research Center), and UL1 TR002345 (Washington University Clinical and Translational Science Award). The study was part of the Lifestyle Interventions For Expectant Moms (LIFE-Moms) consortium, which is supported by NIH grants U01 DK094418, U01 DK094463, U01 DK094416, U01 DK094466, U01 HL114344, U01 HL114377, U01 HD072834; the National Center for Complementary and Integrative Health; the NIH Office of Research in Women’s Health; the Office of Behavioral and Social Science Research; the Indian Health Service; and the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. B.M., D.H.-J., A.G.C., and S.K. contributed to the study concept and design. B.M., B.W.P., and S.K. contributed to data collection, analysis, and interpretation. B.M. drafted the manuscript, and all authors critically revised the manuscript. D.H.-J., A.G.C., and S.K. obtained funding for the study. B.M. and S.K. are the guarantors of the study 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.