Regression From Early GDM to Normal Glucose Tolerance and Adverse Pregnancy Outcomes in the Treatment of Booking Gestational Diabetes Mellitus Study

TOBOGM, a multicenter RCT of immediate or deferred treatment of eGDM, used World Health Organization (WHO) 2-h 75-g OGTT criteria (3) (fasting glucose 92–125 mg/dL [5.1–6.9 mmol/L], 1-h glucose 180 mg/dL [≥10.0 mmol/L], and 2-h glucose 153–199 mg/dL [8.5–11.0 mmol/L]) across 17 hospitals in Australia, Sweden, Austria, and India, as previously described (1,2). Women had singleton pregnancies at 4–19⁺⁶ weeks (superscript values indicate the number of days; e.g., 19⁺⁶ indicates 19 weeks plus 6 days), were aged ≥18 years, had ≥1 risk factor for GDM, and had no significant relevant medical disorders.

Pregnancy details were collected from case records. Women with eGDM were randomly allocated 1:1 within site by minimization using an electronic randomizer. Randomization was stratified into a lower band (fasting glucose 92–94 mg/dL [5.1–5.2 mmol/L] and/or 1-h glucose 180–189 mg/dL [10.0–10.5 mmol/L] and/or 2-h glucose 153–161 mg/dL [8.5–8.9 mmol/L]) and an upper band (fasting glucose 95–109 mg/dL [5.3–6.0 mmol/L] and/or 1-h glucose ≥190 mg/dL [≥10.6 mmol/L] and/or 2-h glucose 162–198 mg/dL [9.0–11.0 mmol/L]). All enrolled women without treated eGDM had a repeat OGTT at 24–28 weeks. All women with GDM on the 24–28 weeks’ OGTT were treated. Neonates of women within the RCT and a randomly selected proportion of women without eGDM underwent heel prick glucose measurement within 1–2 h. Clinic, study staff, and participants were blinded to OGTT outcome unless treated for GDM.

The primary outcome was a composite of birth <37⁺⁰ weeks, birth weight ≥4,500 g, birth trauma, neonatal respiratory distress, phototherapy, stillbirth/neonatal death, and/or shoulder dystocia, as previously defined (1). Large-for-gestational-age infants (≥90th centile) and small-for-gestational-age infants (≤10th centile) were defined using ethnicity and sex-adjusted, customized centiles for birth weight (www.gestation.net). Neonatal hypoglycemia was defined as 1- to 2-h postbirth heel prick glucose ≤2.2 mmol/L (≤40 mg/dL).

Analyses were performed in SPSS, version 29.0 (IBM, U.S.). Women within the RCT with treated eGDM were excluded, as were those without eGDM but who had GDM on the second OGTT at 24–28 weeks. Figure 1 shows the study design for the current analyses. Descriptive analyses summarize demographic characteristics. Group comparisons were adjusted for age, prepregnancy BMI, ethnicity, current smoking, primigravity, and tertiary or higher education and for site/site clusters (grouped into 4 clusters as described in Simmons et al. [1]). Generalized linear modeling was used to adjust continuous outcomes and logistic regression binary outcomes. Gestational weight gain (GWG) was log transformed, and geometric means are shown. Stepwise logistic regression was performed within the untreated eGDM group to predict GDM maintenance regression at 24–28 weeks. Besides the prespecified factors, other variables entered into the stepwise regression included duration of gestation at diagnostic OGTT, history of polycystic ovary syndrome, first-degree relative with diabetes, previous GDM (coded as nulliparous, no, yes, or maybe), HbA_1c, initial fasting, and 1-h and 2-h glucose at the OGTT. A second model was then run that included GWG. Cox and Snell R² is shown for variance. All tests were two-sided, with P < 0.05 predetermined as significant.

The protocol was approved by local ethics committees in each country, commencing with the South Western Sydney Local Health District Research and Ethics Office, Australia (15/LPOOL/551).

Overall, 2,218 (85.5%) women were in the No-GDM group, 254 (9.8%) in the GDM-Maintained group, and 121 (4.7%) in the GDM-Regression group. In the GDM-Regression group, 75 (62.0%) were in the lower glucose band and 46 (38.0%) were in the upper band. In the GDM-Maintained group, 79 (31.1%) were in the lower band and 175 (68.9%) were in the upper band.

Table 1 shows the three groups’ baseline characteristics. The GDM-Regression group members were similar to the GDM-Maintained group members for age, previous GDM, and prepregnancy BMI. GDM-Regression glycemic measures were between those of the No-GDM and GDM-Maintained groups. Pregnancy outcomes were similar between the No-GDM and GDM-Regression groups (Table 2). GWG from prepregnancy to 24–28 weeks was lower in the GDM-Regression group than in the No-GDM group and was nonsignificant after adjustment for confounders (crude geometric means, 6.4 vs. 8.0 kg, P = 0.015; adjusted, 7.6 vs. 8.4 kg, P = 0.574). GWG to 36–38 weeks, gestation at birth, birth weight, and birth centile were similar between groups.

Compared with the GDM-Regression group, neonatal intensive care unit admission was higher in the GDM-Maintained group (Table 2). Table 3 shows the forward conditional logistic regression comparing GDM-Regression and GDM-Maintained groups. The 1-h glucose accounted for 9.0% of the variance; HbA_1c, fasting glucose, 2-h glucose, and previous polycystic ovary syndrome accounted for a further 8.0%. GWG was not a significant entrant into the model. After adjusting for confounders, the lower glucose band was 3.09 (95% CI 1.87–5.10) more likely to regress than the upper glucose band. Only 38.1% of those who regressed were predicted correctly by the full model.

In this large international, multicenter cohort with participants recruited prospectively, those in the GDM-Regression group had pregnancy outcomes similar to those of individuals who did not have GDM identified at any time. Under 40% of those who regressed could be identified using logistic regression, as the equation only accounted for 17.0% of the variance and did not include classic risk factors, e.g., BMI, indicating that other variables (e.g., new biomarkers) are needed to identify the GDM-Regression group. The GDM-Regression group was small (n = 121), as 50% of women in the treatment arm of the RCT were excluded; hence, these observations warrant caution in their interpretation.

Reasons for GDM regression include general OGTT variability, particularly given the physiologic differences in glucose metabolism between early pregnancy and midpregnancy, as well as regression to the mean. Some variation could have occurred through preanalytical loss, but this was minimized through standardized and careful handling of samples (spinning/separating within 30 min) (1,2). A further potential contributor is the physiological fall in fasting plasma glucose that occurs between early pregnancy and 24–28 weeks, with a median reported decrease in plasma glucose of 0.1 mmol/L (2 mg/dL) between gestational weeks 6 and 10 with little further glucose reduction through to the third trimester (4).

It is unclear whether participants in the GDM-Regression group were more likely to change their lifestyles (e.g., physical activity and diet) or experience nausea or hyperemesis. No data were collected to inform these unknowns. The women were recruits to an RCT for those at risk for GDM and hence potentially were more likely to choose behaviors facilitating regression. At least one meta-analysis of GDM prevention RCTs suggests that early lifestyle change may be the most effective method to reduce GDM risk (5).

This was the first major RCT to investigate the treatment of eGDM. Among nonpregnant adults, intervention studies have reported the characteristics of those whose impaired glucose tolerance or fasting glucose have regressed (6). The main factor contributing to dysglycemia regression has been weight loss, something that is avoided in pregnancy.

The mechanism behind the lack of difference in pregnancy outcomes between the No-GDM and GDM-Regression groups is unclear. One of the concerns was that the OGTT 24–28 weeks’ gestation would be normal due to “fetal-steal syndrome” (7), where fetal hyperinsulinemia creates a gradient to promote glucose flux toward the fetus. This might have been expected to lead to an increased risk of LGA and neonatal hypoglycemia, but the adjusted odds ratios were close to 1. Similarly, greater early glucose flux might have been expected to influence growth and fat deposition through the Pedersen hypothesis (8), but the influence of this is greatest in the third trimester, which in this study would have been after the normal late OGTT.

This study has a range of strengths, including its origin from a large well-designed RCT combined with a prospective cohort study of those recruited to but excluded from the RCT (1,2). The data were prospectively and systematically collected using the same methodology. A randomly selected subgroup without GDM was available with heel prick glucose. The data are multiethnic and international, with few dropouts and missing data. The analysis is the first of its kind, with important findings regarding pregnancy outcomes.

Weaknesses include the small size of the GDM-Regression group, which was underpowered for some measures. Conversely, the trends for the major neonatal outcomes were often slightly better in the GDM-Regression group than in the No-GDM group. The eGDM cohort is 50% smaller than expected on a population basis because of exclusion of those randomized to treatment, but excluded women should be comparable with those in this analysis, as they were selected through randomization. Women identifying as European with no risk factors were excluded from the study, few women identifying as African or Hispanic were available in the study catchment, and no additional biomarker measures have yet been performed (e.g., fasting insulin for measures of insulin resistance). No measures of lifestyle change were collected to identify any behavioral changes between first and second OGTT, but adjusted GWG was unchanged. Participants were those willing to participate in the TOBOGM RCT.

In conclusion, in women with GDM risk factors, pregnancy outcomes are similar with a normal OGTT at 24–28 weeks with and without prior eGDM. However, further work is required to improve prediction of GDM regression, possibly by adding new biomarkers. Follow-up studies are crucial to assess whether forgoing early treatment in the GDM-Regression group contributes to the long-term metabolic risk to the offspring.

Acknowledgments. We thank the study participants involved in TOBOGM, the trial coordinators, research midwives and nurses at each site, the maternity service and laboratory staff who assisted the project, the trial coordination staff, and the personnel at Roche for the donation of the glucose-monitoring meters used for the trial participants.

Funding. This work was supported by the National Health and Medical Research Council (grants 1104231 and 2009326), the Region Örebro Research Committee (grants Dnr OLL-970566 and OLL-942177), the Medical Scientific Fund of the Mayor of Vienna (project numbers 15205 and 23026), the South Western Sydney Local Health District Academic Unit (grant 2016), and a Western Sydney University Ainsworth Trust grant (2019).

Neither the funding sources nor the author-affiliated institutions took part in the trial design, the collection, analysis, and interpretation of the data, manuscript writing, or the decision to submit it for publication.

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

Author Contributions. D.S. conceived the project and drafted the manuscript. J.I. assisted with the initial manuscript draft. All authors contributed to the writing and reviewed and edited the manuscript. All authors approved the final version of the manuscript. D.S. 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 analysis.

Members of the TOBOGM Research Group include David Simmons (Western Sydney University, Campbelltown, New South Wales, Australia), N. Wah Cheung (Westmead Hospital, Sydney, New South Wales, Australia), Jincy Immanuel (Western Sydney University, Campbelltown, New South Wales, Australia), William M. Hague (Robinson Research Institute, The University of Adelaide, Adelaide, South Australia, Australia), Helena Teede (Monash Health and Monash University, Melbourne, Victoria, Australia), Christopher J. Nolan (Canberra Hospital and Australian National University, Canberra, Australian Capital Territory, Australia), Michael J. Peek (Australian National University, Canberra, Australian Capital Territory, Australia), Jeff R. Flack (Bankstown-Lidcombe Hospital, Sydney, New South Wales, Australia), Mark McLean (Blacktown Hospital, Sydney, New South Wales, Australia), Vincent Wong (Liverpool Hospital, Sydney, New South Wales, Australia), Emily Hibbert (Nepean Clinical School, University of Sydney and Nepean Hospital, Sydney, New South Wales, Australia), Emily Gianatti (Department of Endocrinology, Fiona Stanley Hospital, Murdoch, Western Australia, Australia), Arianne Sweeting (Department of Endocrinology, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia), Suzette Coat (Robinson Research Institute, The University of Adelaide, Adelaide, South Australia, Australia), Raiyomand Dalal (Campbelltown Hospital, Campbelltown, New South Wales, Australia), Georgia Soldatos (Monash Health and Monash University, Melbourne, Victoria, Australia), Suja Padmanabhan (Westmead Hospital, Sydney, New South Wales, Australia), Rohit Rajagopal (Campbelltown Hospital, Campbelltown, New South Wales, Australia), Victoria Rudland (Westmead Hospital, Sydney, New South Wales, Australia), Jürgen Harreiter (Gender Medicine Unit, Division of Endocrinology and Metabolism, Department of Medicine III, Medical University of Vienna, Vienna, Austria), Alexandra Kautzky-Willer (Gender Medicine Unit, Division of Endocrinology and Metabolism, Department of Medicine III, Medical University of Vienna, Vienna, Austria), Herbert Kiss (Division of Feto-Maternal Medicine, Department of Obstetrics and Gynecology, Medical University of Vienna, Vienna, Austria), Helena Backman (Department of Obstetrics and Gynecology, Faculty of Medicine and Health, Örebro University, Örebro, Sweden), Erik Schwarcz (Department of Internal Medicine, Faculty of Medicine and Health, Örebro University, Örebro, Sweden), Glynis Ross (Royal Prince Alfred Hospital, Sydney, New South Wales, Australia), Viswanathan Mohan (Dr. Mohan’s Diabetes Specialities Centre and Madras Diabetes Research Foundation, Chennai, India), Ranjit Mohan Anjana (Dr. Mohan’s Diabetes Specialities Centre and Madras Diabetes Research Foundation, Chennai, India), and Uma Ram (Seethapathy Clinic & Hospital, Chennai, India).