Continuous Glucose Monitoring Profiles in Pregnancies With and Without Gestational Diabetes Mellitus

Gestational diabetes mellitus (GDM) affects approximately 14% of pregnancies worldwide (1). Traditional screening strategy for diagnosing GDM is an oral glucose tolerance test (OGTT) administered at 24–28 weeks’ gestation. GDM is associated with perinatal morbidities such as hypertensive disorders of pregnancy, fetal overgrowth, neonatal hypoglycemia, and neonatal intensive care unit admissions (2–5). Recent findings suggest that earlier identification and treatment of dysglycemia prior to 24 weeks (6) can mitigate some of these risks. Screening strategies for early GDM and thresholds for a screen positive test vary among organizations and clinical practices, and the diagnostic thresholds for identifying early GDM may differ from those used in the third trimester (7–9). Early glucose screening using the OGTT validated for GDM diagnosis in the third trimester yields inconsistent results in early pregnancy, leading to both missed cases and overdiagnosis of GDM (6,10). In the Treatment of Booking Gestational Diabetes Mellitus (TOBOGM) study, up to a third of individuals diagnosed with early GDM by standard OGTT failed to test positive on repeat OGTT in the third trimester of pregnancy (6).

Evidence from the Hyperglycemia and Adverse Pregnancy Outcomes (HAPO) study demonstrates that even milder degrees of hyperglycemia below the OGTT thresholds for GDM diagnosis increase the risk of adverse perinatal outcomes such as large-for-gestational-age infants (11). There are limited data on glucose levels in pregnant individuals with normal glucose tolerance, yet what is known suggests mean glucose levels in individuals without GDM are well below current therapeutic targets for GDM management (12), averaging 70–110 mg/dL.

Continuous glucose monitoring (CGM) is a practical and convenient way to capture glucose data (13–15). There are limited data on the utilization of CGM-measured glycemic profiles in pregnant individuals without diabetes and whether these profiles are associated with GDM. The objective of our study was to compare CGM-derived glycemic patterns between participants with GDM as diagnosed by standard OGTT at 24 to 34 weeks’ gestation versus those without GDM.

This prospective observational, nonintervention study was conducted at two academic-based clinical sites: University of Pennsylvania and International Diabetes Center at Park Nicollet. The protocol and informed consent forms were approved by the Jaeb Center for Health Research Institutional Review Board, Tampa, FL.

Individuals were eligible for participation if they had a confirmed singleton viable pregnancy determined by ≤12-week ultrasound, were at least 18 years old, and had an initial hemoglobin A1c (HbA_1c) of <6.5% (<48 mmol/mol). Enrollment occurred as early as possible based on eligibility, with all participant enrollment occurring prior to 17 weeks’ gestation. Individuals were excluded if they were non-English speaking, had a diagnosis of type 1 or type 2 diabetes, had prior gastric bypass surgery, used medications intended to lower blood glucose or oral systemic steroids at the time of screening, or were unwilling to wear a CGM sensor during pregnancy (see Supplementary Table 1 for a complete listing of inclusion and exclusion criteria). Each participant provided electronic informed consent.

During the enrollment visit, a blinded Dexcom G6 Pro CGM System sensor was placed, and the participant was instructed on its care and mailing of the transmitter to the coordinating center after 10 days. Subsequent sensor placements occurred at standard care office visits and/or at home. Training for sensor insertion was performed virtually or in person. Participants were encouraged to wear the CGM sensor continuously, as long as they were willing, throughout pregnancy until the day of delivery. Routine prenatal care and delivery management followed the standard obstetrical practice of the respective institutions. Clinical characteristics of the participant and neonate were abstracted via individual chart review.

GDM status was determined through either the two-step method including a 50-g glucose challenge test (GCT) followed by a 100-g OGTT, a one-step 75-g OGTT, or a two-step 75-g OGTT. The GCT and OGTT procedures and criteria are detailed in Supplementary Table 1, and the number of participants completing each OGTT type are listed in Supplementary Fig. 1. For the remainder of this article, the term “OGTT” will be used to describe the overall testing procedures involving both the GCT and/or any follow-up 75-g or 100-g OGTTs, unless a specific dose (i.e., 50 g, 75 g, or 100 g) or the term “GCT” is used. The term “GDM group” will be used to refer to participants with GDM as assessed by the OGTT, and “No GDM” group will be used to refer to participants who had a negative OGTT result.

To be included in the analyses, participants must have completed an OGTT between 24 and 34 weeks’ gestation and have at least 14 days of CGM data during their pregnancy with at least 10 days of CGM data in the second trimester. For first-trimester tabulations, a minimum of 10 days of CGM data were required. CGM data were included through the time of the OGTT; CGM data on or after the OGTT date were excluded, except when participants did not meet the minimum CGM requirements before the OGTT date, in which case CGM data after the OGTT date were included until adequate CGM data were met. This occurred in eight participants for the overall gestational period, three participants in the first trimester, and seven participants in the second trimester.

CGM-measured glycemic metrics were also summarized by the third trimester and by 2-week intervals for participants with at least 10 days of CGM data during the respective period. Glucose metrics were summarized as means and SDs or summary statistics appropriate to the distribution. Throughout, to convert glucose to millimoles per liter, multiply by 0.0555. CGM-measured glycemic metrics for the GDM group were compared with the No GDM group using two-sample t tests for normally distributed variables and Mann-Whitney U tests for skewed distributions. Post hoc two-sample t tests and Mann-Whitney U tests were conducted to test the GDM versus No GDM difference in mean glucose and hyperglycemia metrics during daytime (6 a.m. to before midnight), nighttime (midnight to before 6 a.m.), and during each 2-week gestational period. Post hoc two-sample t tests and Mann-Whitney U tests were also conducted to test the GDM versus No GDM difference in glycemic metrics within each trimester.

Second trimester and gestational week 13–14 percent times >140 mg/dL were used to predict GDM. The area under the receiver operating characteristic curve (AUROC) was calculated by determining the sensitivity and specificity at each distinct percent time >140 mg/dL cut point. AUROC was used to evaluate predictive performance, with higher values representing greater predictive accuracy.

P values are two-sided, and multiple comparisons were adjusted for using the Benjamini-Hochberg adaptive false discovery rate correction procedure (16). Analyses were performed with SAS software, version 9.4 (SAS Institute).

Data will be made available on a publicly available website (www.jaeb.org) at a later date.

Between June 2020 and December 2021, a total of 937 eligible individuals were enrolled, with 768 meeting the inclusion criteria for this analysis (Supplementary Fig. 1). The mean age at enrollment was 33 ± 4 years (range, 18–54 years), with 592 (77%) self-reported as White, and 49 (6%) self-reported as being of Hispanic or Latino ethnicity. Mean HbA_1c was 5.2 ± 0.3% (33 ± 3.3 mmol/mol), mean BMI was 26.8 ± 6.1 kg/m², and 3% had a history of GDM in a prior pregnancy. Five participants with GDM and no participants without GDM had an HbA_1c ≥6.0% (≥48 mmol/mol).

There were 58 (8%) participants with GDM during the study, based on OGTT results. Clinical characteristics of individuals stratified according to GDM status are listed in Table 1. The GDM group had a higher mean BMI and were more likely to have had a prior diagnosis of GDM compared with the No GDM group. The overall duration of CGM wear before the OGTT was a median of 67 days (interquartile range [IQR]: 50 to 82).

In comparison with the No GDM group, the GDM group had a higher mean glucose (109 ± 13 vs. 100 ± 8 mg/dL [6.0 ± 0.7 vs. 5.6 ± 0.4 mmol/L], P < 0.001), greater glucose SD (23 ± 4 vs. 19 ± 3 [1.3 ± 0.2 vs. 1.1 ± 0.2 mmol/L], P < 0.001), a higher coefficient of variation (21% ± 4% vs. 19% ± 3%, P < 0.001), and lower percent time 63–140 mg/dL (3.5–7.8 mmol/L) (87% ± 11% vs. 94% ± 4%, P < 0.001) throughout the gestational period prior to completion of the OGTT (Table 2). Similarly, percent time spent in 63–120 mg/dL (3.5–6.7 mmol/L) was also lower throughout gestation (70% ± 17% vs. 84% ± 8%, P < 0.001). Expanded glycemic profiles during the 24-h day, including additional glycemic metrics, are summarized in Supplementary Table 2.

When analyzing glucose trends according to time of day, the GDM group had consistently higher mean glucose levels during both the daytime and overnight periods in comparison with the No GDM group (Fig. 1). Mean daytime glucose was 110 ± 12 mg/dL (6.1 ± 0.7 mmol/L) in the GDM group compared with 101 ± 8 mg/dL (5.6 ± 0.4 mmol/L) in the No GDM group (P < 0.001). Likewise, the mean overnight glucose (midnight to before 6 a.m.) was 107 ± 16 mg/dL (5.9 ± 0.9 mmol/L) for the GDM group versus 98 ± 9 mg/dL (5.4 ± 0.5 mmol/L) for the No GDM group (P < 0.001).

The less percent time spent in the range of 63–140 mg/dL (3.5–7.8 mmol/L) and 63–120 mg/dL (3.5–6.7 mmol/L) in the GDM group was primarily attributed to increased hyperglycemia rather than decreased hypoglycemia. Compared with the No GDM group, the GDM group spent more time above 120 mg/dL (6.7 mmol/L) (median 23% vs. 12%, P < 0.001) and above 140 mg/dL (7.8 mmol/L) (7.4% vs. 2.7%, P < 0.001), and spent less time below 63 mg/dL (3.5 mmol/L) (0.7% vs. 1.3%, P = 0.003) and below 54 mg/dL (3.0 mmol/L) (0.3% vs. 0.5%, P = 0.01) (Table 2). This trend persisted when evaluating overnight glucose levels specifically, with a percent time >120 mg/dL (>6.7 mmol/L) of 11% in the GDM group compared with 6% in the No GDM group (P < 0.001), and a percent time >140 mg/dL (>7.8 mmol/L) of 2.1% in GDM compared with 0.7% in No GDM (P < 0.001). The GDM group exhibited a lower area over the curve 63 and 54 mg/dL (3.5 and 3.0 mmol/L), and a higher area under the curve for glucose levels of 120 and 140 mg/dL (6.7 and 7.8 mmol/L), in contrast to the No GDM group (Supplementary Table 2). Glycemic metrics comparing the GDM and No GDM groups separated by OGTT type (i.e., two-step 75-g OGTT and two-step 100-g OGTT) are summarized in Supplementary Table 3.

These patterns in glycemic trends—higher mean glucose, increased glycemic variability (SD), less time in range 63–140 mg/dL (3.5–7.8 mmol/L), and greater hyperglycemia—remained consistent across all three trimesters of pregnancy before completion of the OGTT (Table 2 and Supplementary Fig. 2). The GDM group had higher mean glucose, percent time >120 mg/dL (>6.7 mmol/L), and percent time >140 mg/dL (>7.8 mmol/L) as early as 13–14 weeks of gestation, and was consistently higher during subsequent 2-week periods (Fig. 2). Other glycemic trends were also consistent by 2-week periods (Supplementary Table 4). Limiting this analysis to only those with first trimester CGM data yielded similar glycemic trends.

CGM metrics obtained prior to OGTT had the ability to predict GDM. The AUROC was 0.81 when using second trimester percent time >140 mg/dL to predict GDM. Using percent time >140 mg/dL as early as 13–14 weeks of gestation to predict GDM yielded an AUROC of 0.74. This provides further evidence that CGM-measured hyperglycemic metrics can achieve high precision in predicting GDM.

This prospective observational study identified distinct CGM glycemic patterns associated with a diagnosis of GDM at 24–34 weeks’ gestation. All glycemic metrics, including mean glucose and the percent time spent exceeding both 120 and 140 mg/dL (6.7 and 7.8 mmol/L), were found to be significantly higher as early as 13–14 weeks of gestation among individuals in the GDM group compared with the No GDM group. Considering the recruitment of a predominantly healthy cohort during early pregnancy, we chose to assess the percent time spent above 120 mg/dL (6.7 mmol/L), revealing a greater percentage in the GDM group. The percent time spent within the ranges of 63–120 mg/dL (3.5–6.7 mmol/L) and 63–140 mg/dL (3.5–7.8 mmol/L) was observed to be lower in the GDM group compared with the No GDM group. Notably, the GDM group consistently displayed higher mean glucose at each 2-week interval leading up to the completion of the OGTT.

Our findings suggest that aiming for a higher percent time in range 63–140 mg/dL (3.5–7.8 mmol/L) might be beneficial in individuals with gestational diabetes, as this would be more reflective of the 94% time in range reflective of the glucose levels in individuals with normal glucose tolerance in our study. Additionally, consideration of a narrow glycemic window, such as 63–120 mg/dL (3.5–6.7 mmol/L), may be more suitable for a healthy pregnant cohort without a diagnosis of diabetes. Although the use of tighter glycemic targets for fingerstick glucose testing did not reduce the rate of large-for-gestational-age infants in the Tight or Less Tight Glycaemic Targets for Women with Gestational Diabetes Mellitus for Reducing Maternal and Perinatal Morbidity trial, serious infant outcomes were lower. In this stepped wedge cluster trial, four times per day fingerstick glucose testing was used, which may not be comparable to the large amount of data obtained in a 24-h daily profile using CGM (17). Further study is needed before definitive guidance can be recommended. Our results are consistent with those of Hernandez et al. (12) showing a population mean of 110 mg/dL in pooled data from 12 studies including 255 pregnant individuals with normal glucose tolerance. In more detailed analysis, the authors demonstrated fasting mean glucose levels of 70.9 mg/dL and postprandial mean glucose levels of 108.9 and 99.3 mg/dL for 1 and 2 h postmeal, respectively. Both our findings and those of Hernandez et al. underscore that glycemic levels in pregnant individuals with normal glucose tolerance fall well below the current therapeutic thresholds used in the management of GDM during pregnancy. These results are also consistent with those of the Hyperglycemia and Adverse Pregnancy Outcomes study, which showed an increased risk of large-for-gestational-age infants with as little as a 1-SD increase in glucose level, similarly below current therapeutic targets.

Our glucose data confirm the presence of variation in glucose levels throughout the course of pregnancy. These findings support previous studies demonstrating that fasting glucose levels decline between 10 and 28 weeks of gestation, with the most significant decrease as early as 6–10 weeks of gestation (18,19). Although we have limited data this early in pregnancy for our study population, we observed slightly higher glucose levels early in pregnancy, followed by a subsequent decline corresponding to increased insulin sensitivity during midgestation (16–20 weeks’ gestation) in both the GDM and No GDM group. Following this midgestation nadir, glucose levels in the GDM group progressively increase in the latter half of the second trimester and continue to rise throughout the third trimester, while the No GDM group maintained their glucose levels. The observational nature of our study provides insight into glucose physiology in low-risk pregnancies.

These findings raise questions about the most suitable time for screening for early GDM or dysglycemia in pregnancy. The results suggest that early glucose screening prior to 16 weeks’ gestation may be optimal, and that some of the discrepancies in early GDM screening may be attributable to the timing of the test. Early GDM screening might yield less accurate results during gestational weeks 17–18, a time period that seems to align with the physiologic decline in glucose we observed in our study group. Screening during this time may miss a higher proportion of individuals who are at risk for having GDM later in pregnancy.

Early screening for hyperglycemia is recommended for pregnant individuals with risk factors, aiming to identify previously undiagnosed type 2 diabetes (20). There is also compelling evidence that individuals with lesser degrees of hyperglycemia in early pregnancy below current diagnostic thresholds for diabetes are more likely to have GDM (21,22). Approximately 15–70% of individuals diagnosed with GDM have evidence of hyperglycemia prior to 24 weeks’ gestation (18). In a randomized clinical trial, early GDM diagnosis and immediate treatment demonstrated a modest decrease in a primary composite neonatal outcome that was primarily driven by a reduction in neonatal respiratory distress (6). Mean birth weight and rate of fetal overgrowth were not different between those undergoing immediate treatment and usual care. In this trial, approximately one third of individuals diagnosed with early GDM by standard OGTT in the usual care group failed to test positive on repeat OGTT in the third trimester of pregnancy, highlighting the inconsistencies that exist with our current strategy of point-of-care testing using an OGTT. As Simmons et al. (6) highlight, it is unclear whether criteria previously established for third trimester positivity can be applied to testing in early pregnancy. This may limit the utility of an OGTT in early pregnancy for diagnosis of dysglycemia. Early and late GDM screening may differ in the strategies used for diagnosis as well as varying thresholds for positivity. These challenges underscore the need to reevaluate our current screening strategies using the OGTT for early pregnancy dysglycemia.

In our cohort, CGM measured percent time >140 mg/dL as early as 13–14 weeks and, throughout the second trimester, showed high precision in predicting GDM. These findings support the idea that CGM could be used in addition to or instead of OGTT to screen individuals at risk for hyperglycemia during pregnancy, even as early as the first trimester. Additional modeling approaches utilizing various CGM metrics should be explored to further assess the value of CGM in predicting GDM.

CGM has been demonstrated to improve glycemic levels and reduce adverse perinatal outcomes in pregnant individuals with type 1 diabetes (13). The use of CGM in pregnancy is gaining acceptance and becoming more prevalent for diabetes management during this critical period (23). The pursuit of an individualized, patient-centered approach to glucose screening using CGM has the potential to yield accurate and consistent results, benefiting both pregnant individuals and clinicians. Leveraging an individual’s distinct glycemic patterns could potentially offer a more precise assessment of glucose levels than point-of-care testing through an OGTT. This personalized glycemic profile takes into account various clinical factors, encompassing body composition, metabolism, activity, and dietary habits, thus providing deeper insights into the intricacies of early dysglycemia in pregnancy. This focused and personalized approach holds promise for identifying individuals at risk for GDM later in pregnancy. The utility of CGM in early identification of individuals with adverse perinatal outcomes associated with hyperglycemia warrants further investigation.

Our findings also demonstrate willingness among individuals without a diagnosis of diabetes to wear CGMs as a tool to monitor their glucose levels during pregnancy. In our study, participants were blinded to their glucose data to ensure an accurate representation of their daily glucose patterns without altering their diet or activity in response to their data. The average duration of CGM wear prior to the OGTT was 68 days, signifying a high tolerability in using wearable devices to monitor glucose levels. This observation carries profound implications for the potential integration of CGM into clinical care guidelines for glucose screening recommendations. In contrast, the OGTT has several disadvantages, including patient acceptance, tolerability, and compliance (10). Furthermore, being a point-of-care test, it only offers a snapshot of glucose levels for a specific day and time, thereby lacking the comprehensive and extended view of glycemic trends provided by CGM wear. This broader perspective afforded by CGM usage significantly enhances the chances of identifying elevations in glucose levels that could evade detection through a standard glucose tolerance test.

However, our findings are not without limitations, one of which includes the allowance of up to 14 days of CGM data to be included after the OGTT in the overall gestational period and up to 10 days in the first and second trimesters in order to allow participants with a possible early GDM diagnosis to be included in the analysis. Removing this requirement would remove few participants from the analysis, and, therefore, conclusions would remain unchanged. In addition, we have a limited amount of first trimester (<12 weeks’ gestation) data, which limits our ability to conduct a detailed analysis of glycemia at the earliest point in pregnancy. Although our study enrollment was large, we recognize that only 58 participants were diagnosed with GDM using the OGTT, and the demographics of the enrolled cohort may not be reflective of the true population with gestational hyperglycemia.

Besides the large sample size, our study has several other significant strengths, such as detailed CGM profiles across the entirety of gestation prior to administration of the OGTT, sophisticated CGM analyses conducted by our statisticians, and inclusion of centers that use both of the recognized diagnostic criteria (Carpenter and Coustan and The Internatioanl Association of Diabetes and Pregnancy Study Groups) for GDM diagnosis. Detailed analyses of various CGM glycemic metrics and their ability to predict both GDM and adverse perinatal outcomes will be the focus of future investigation.

There exists a pressing need to comprehensively understand and characterize dysglycemia in early pregnancy, particularly its correlation with GDM and the associated adverse perinatal outcomes. The current strategies for glucose screening exhibit significant variability, yielding inconsistent results. These inconsistencies may contribute to the challenge of achieving substantial progress in mitigating the adverse perinatal outcomes linked to hyperglycemia. The identification of a precise, widely acceptable approach for diagnosing early pregnancy dysglycemia holds significant promise. Such an approach could facilitate earlier interventions, which may improve perinatal outcomes associated with GDM. CGM could potentially play a pivotal role in providing timely identification of distinct glycemic patterns indicative of early dysglycemia. CGM has the potential to enable early detection and intervention, which may aid in the improvement of perinatal outcomes associated with GDM.

Acknowledgments. Study center staff and other individuals who participated in the conduct of the trial are listed in the Supplementary Material.

Funding. Study funding was provided by Leona M. and Harry B. Helmsley Charitable Trust and UnitedHealth Group. Dexcom provided the devices used in the study.

Duality of Interest. C.D. reports advisory work for Dexcom for GDM patient-facing materials and system implementation. R.W.B. reports no personal financial disclosures but reports that his institution has received funding on his behalf as follows: grant funding and study supplies from Dexcom. R.M.B. has received research support, has acted as a consultant, or has been on the scientific advisory board for Abbott Diabetes Care, Ascensia, Bigfoot Biomedical, Inc., CeQur, DexCom, Eli Lilly, Embecta, Hygieia, Insulet, Medtronic, Novo Nordisk, Onduo, Roche Diabetes Care, Tandem Diabetes Care, Sanofi, UnitedHealthcare, Vertex Pharmaceuticals, and Zealand Pharma. R.M.B.’s employer, nonprofit HealthPartners Institute, contracts for his services, and he receives no personal income for any of these activities.. A.L.C. reports no personal financial disclosures but reports that his institution has received funding on his behalf as follows: research support from Medtronic, Tandem, Insulet, Abbott, Dexcom, Eli Lilly, Novo Nordisk, Sanofi, and UnitedHealth Group and consultancy fees from Mannkind Corporation and Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. C.D. and Z.L. wrote and edited the manuscript. R.W.B., E.N., R.M.B., M.J., S.D., M.B., K.K., J.S., P.C., and A.L.C. reviewed, contributed to discussion, and edited the manuscript. A.L.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analyses.

Handling Editors. The journal editor responsible for overseeing the review of the manuscript was Steven E. Kahn.