One Step Closer to a Definition of Early Gestational Diabetes Mellitus: Secondary Analyses From the Treatment of Booking Gestational Diabetes Mellitus (TOBOGM) Randomized Trial

The diagnosis and treatment of gestational diabetes mellitus (GDM) in late pregnancy has been widely accepted as standard of care. However, until recently, there were no large-scale randomized trials demonstrating a benefit of treatment of GDM in early pregnancy (i.e., prior to 20 weeks’ gestation). Despite this, the practice of screening for, diagnosing, and treating hyperglycemia in people without preexisting diabetes in early pregnancy has become widespread, albeit with extremely variable diagnostic procedures applied across the U.S. and the world. In the absence of high-quality evidence, pre-2024 statements from major professional societies provide little direction or have conflicting guidance on how and whether glycemic abnormalities shy of overt diabetes should be treated in early pregnancy.

The multicenter Treatment of Booking Gestational Diabetes Mellitus (TOBOGM) trial, the primary analysis of which was published in June 2023, enrolled individuals with diabetes risk factors who met International Association of the Diabetes and Pregnancy Study Groups (IADPSG) criteria for GDM at <20 weeks’ gestation in Australia, Austria, Sweden, and India (1). Those with fasting glucose ≥110 mg/dL or 2-h postload glucose level ≥200 mg/dL on an oral glucose tolerance test (OGTT) were excluded from further study. Included participants with early GDM (N = 802) were randomized to early pregnancy treatment versus deferred treatment. The deferred treatment group had a repeat OGTT at 24–28 weeks’ gestation and only received treatment if GDM was still apparent at that time. Treatment of GDM in both the early and deferred treatment groups (when indicated) included self-monitoring of blood glucose, dietary advice, and pharmacologic glucose lowering as needed to meet conventional glycemic targets. Cleverly, the study team masked the diagnosis of those with early GDM by including as “decoys” participants who did not meet IADPSG criteria on their early pregnancy 75-g OGTT but were disguised as study participants (1).

In the TOBOGM trial, a 5.6% absolute risk reduction was observed (24.9% vs. 30.5%) in the primary neonatal composite outcome (preterm birth, respiratory distress, macrosomia, birth trauma, shoulder dystocia, hyperbilirubinemia requiring phototherapy, or fetal/neonatal death) with early pregnancy treatment of GDM as compared with deferred treatment (1). Perhaps surprisingly, the reduction in this composite outcome appeared to be largely driven by a reduction in neonatal respiratory distress, defined as a need for ≥4 h of respiratory support within 24 h of birth (early treatment 9.8% vs. deferred treatment 17.0%, adjusted relative risk 0.57 [95% CI 0.41–0.79]). In prespecified subgroup analyses, the risk reduction in the primary outcome appeared greatest for those with glucose in the range that served as the “higher band” (to use the phrasing of Simmons et al. [2], below) (fasting glucose ≥95 mg/dL, 1-h postload glucose ≥191 mg/dL, 2-h postload glucose ≥162 mg/dL) and those diagnosed prior to 14 weeks’ gestation. In the lower band (fasting glucose 92–94 mg/dL, 1-h postload glucose 180–190 mg/dL, or 2-h postload glucose 153–161 mg/dL) there was a signal for harm of early treatment in the form of a greater risk of small-for-gestational-age birth weight (early treatment 12.9% vs. deferred treatment 7.4%, adjusted odds ratio 1.75 [95% CI 1.05–2.92]) (1).

In this issue of Diabetes Care, five secondary analysis papers appear, providing additional results that should inform clinical guidelines on early pregnancy hyperglycemia diagnosis and treatment.

In a brief report, “Regression From Early GDM to Normal Glucose Tolerance and Adverse Pregnancy Outcomes in the Treatment of Booking Gestational Diabetes Mellitus Study,” TOBOGM investigators examine the phenomenon of early GDM regression, defined as meeting IADPSG criteria on the OGTT at <20 weeks’ gestation but no longer meeting those criteria on the OGTT at 24–28 weeks’ gestation (2). The TOBOGM study design facilitated investigation of GDM regression in the deferred treatment group, who were not treated unless results of a repeat OGTT at 24–28 weeks’ gestation also met criteria for GDM. Notably, 32% of these participants who met criteria for GDM at <20 weeks’ gestation regressed (2). GDM regression likely represents a combination of intraindividual variability in OGTT results and true physiologic regression of hyperglycemia in some individuals. Despite never receiving treatment, this group of individuals who regressed had outcomes similar to those of the group who had normal glucose tolerance throughout pregnancy, suggesting that early treatment would not have been beneficial. Combined with the concern from the primary study that treatment may lead to small-for-gestational-age birth weight for those with lower glucose levels, these findings argue against using only the IADPSG thresholds to delineate who should qualify for early pregnancy treatment. The investigators valiantly attempted to identify features that would predict GDM regression. Higher glucose levels and polycystic ovary syndrome emerged as risk factors for persistence of the GDM diagnosis at 24–28 weeks’ gestation, but only 38% of regressors could be identified with use of a model incorporating these factors. In the lower glucose band group with milder hyperglycemia, approximately one-half of early cases of GDM regressed. In contrast, only approximately one in five cases regressed in the higher glucose band group (2). Therefore, it would be reasonable to consider early pregnancy treatment only in those who meet the higher glucose band thresholds.

In two original articles, TOBOGM investigators excluded those randomized to early treatment and conducted observational studies that may guide diagnostic criteria and inform prognosis. In “Relationship Between Early-Pregnancy Glycemia and Adverse Outcomes: Findings From the TOBOGM Study,” the investigators examine associations between outcomes and glucose levels that remained untreated in early pregnancy (3). The analysis included both untreated individuals with early GDM and those without early GDM during the screening OGTT at <20 weeks’ gestation (N = 3,188). Investigators found a continuous positive association between both early pregnancy 1- and 2-h postload OGTT values and the composite adverse perinatal outcome (as defined in the primary study). In analyses of categorical glucose values, only 1-h glucose >190 mg/dL (from the higher band category) was significantly associated with the composite perinatal outcome. The relationship between 1-h glucose and outcomes may have been stronger than those observed for fasting and 2-h glucose in part because the study excluded individuals with fasting glucose ≥110 mg/dL and 2-h glucose ≥200 mg/dL. In contrast, as there were no exclusion criteria for high 1-h glucose, the upper bound of this variable was less constrained (3). The authors also conducted an analysis that included the group randomized to treatment in early pregnancy, but its interpretation is complicated. In general, results in this article support that early pregnancy treatment should only be considered for those in the higher glucose band.

In the other original article, “Perinatal Outcomes in Early and Late Gestational Diabetes Mellitus After Treatment From 24–28 Weeks’ Gestation: A TOBOGM Secondary Analysis,” the investigators ask whether treatment starting at 24–28 weeks’ gestation normalizes pregnancy outcomes both in patients with early GDM (present on OGTT at <20 weeks’ gestation) and in patients with late GDM (not present on OGTT at <20 weeks’ gestation but appearing on OGTT at 24–28 weeks’ gestation) (4). In early GDM and late GDM treated after 24 weeks’ gestation, there was no excess risk of large-for-gestational-age birth weight; indeed, mean absolute birth weight was similar in pregnancies with early GDM treated late, those with late GDM treated late, and unaffected pregnancies. However, residual increased risk for early GDM treated after 24 weeks’ gestation was observed for preterm birth, neonatal respiratory distress, hyperbilirubinemia, and neonatal intensive care unit (NICU) admission. Notably, some residual risk for these outcomes (with the exception of preterm birth) was also observed for late GDM treated after 24 weeks’ gestation (4). In many studies of GDM treatment strategies, birth weight–related measures have been used as a primary outcome; the results from this TOBOGM secondary analysis draw attention to other hyperglycemia-associated pregnancy outcomes and suggest that our current treatment paradigm of glucose lowering after 24 weeks’ gestation may not fully address these non–birth-weight-related risks.

Two observation letters complete the TOBOGM collection and provide a nidus for further research. TOBOGM investigators report the results of glucose testing on samples collected in four different collection tube types simultaneously, in “Diagnosis of Gestational Diabetes Mellitus: How Should We Measure Glucose?” (5). In accord with previous suggestions outside the context of pregnancy, the highest glucose levels were observed in the tubes with citrate (NaF-EDTA-citrate), even when every type of tube was placed on ice and rapidly processed. Use of citrate tubes in diagnosis would have resulted in a 9.1% higher GDM prevalence rate. Although citrate tubes are an attractive option to mitigate in vitro glycolysis that leads to inaccuracy and imprecision in clinical glucose measurements, accepted GDM diagnostic criteria are based on glucose values measured in other tube types—presenting a barrier for implementation.

In “Association Between Immediate Treatment of Early Gestational Diabetes Mellitus and Breastfeeding Outcomes: Findings From the TOBOGM Study,” the investigators examine short-term breastfeeding outcomes for those with early GDM randomized to early versus deferred treatment (6). The investigators find that among those with early GDM, breastfeeding as the first feeding method was reduced in the group randomized to deferred treatment (versus early treatment). This difference was attenuated when birth and neonatal characteristics were accounted for, but some of these factors (e.g., gestational age at birth, birth weight, neonatal intensive care unit admission) could be mediators on the causal pathway between GDM treatment and breastfeeding (rather than confounders). Breastfeeding outcomes have seldom been examined rigorously in large, randomized trials of GDM management, so this secondary analysis of the TOBOGM randomized trial is a welcome contribution. While this analysis did not include investigation of more meaningful long-term breastfeeding outcomes, these findings should spur additional studies on the impact of GDM diagnosis and management on lactation.

In summary, this article collection from TOBOGM extends the findings of the primary study. These secondary analyses reinforce that benefits of early GDM treatment might be limited to those with glucose levels in a range higher than that delineated by IADSPG thresholds. In the case of early pregnancy glucose levels that only modestly exceed IADPSG thresholds (fasting glucose 92–94 mg/dL, 1-h postload glucose 180–190 mg/dL, or 2-h postload glucose 153–161 mg/dL), TOBOGM findings suggest potential for unnecessary treatment (high rate of GDM regression) and potential harm (small-for-gestational-age birth weight). Thus, providers currently initiating GDM therapy (blood glucose monitoring with potential for pharmacotherapy) in patients with early pregnancy glucose levels in TOBOGM’s lower glucose band should reconsider this practice. Primary and secondary TOBOGM analyses also draw attention to effects of early pregnancy hyperglycemia on outcomes other than birth weight. Finally, observations from TOBOGM highlight the need for further GDM research on technical laboratory issues in glucose measurement and breastfeeding outcomes.