Gestational Diabetes Mellitus: Is It Time to Reconsider the Diagnostic Criteria?

It has been 10 years since results from the landmark Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study were published (1) and 8 years since the recommendations of the International Association of Diabetes and Pregnancy Study Groups (IADPSG) appeared in this journal (2). The HAPO study demonstrated a continuous relationship between glucose levels on a 75-g oral glucose tolerance test (GTT) and the risks for selected fetal outcomes. Potential diagnostic criteria for gestational diabetes mellitus (GDM) were considered by a Consensus Panel of the IADPSG, which, after a close vote, decided to select an odds ratio (OR) of 1.75, relative to the mean, for selected outcomes. This corresponded to glucose levels of ≥5.1 mmol/L at fasting, ≥10.0 mmol/L at 1 h, and ≥8.5 mmol/L at 2 h on a 75-g GTT. Although the prevalence of GDM varied markedly between the 15 study centers when applying these criteria, the relationship between glucose and outcomes was consistent, and therefore it was considered appropriate for these criteria to be applied globally (3).

The IADPSG recommendations were initially generally greeted with enthusiasm. Finally, diagnostic criteria based on fetal outcomes were available and would replace criteria in the U.S., for example, that were derived from the prediction of subsequent maternal diabetes. Furthermore, given the ethnic diversity of the HAPO participating centers, there was a realistic prospect that their universal adoption was both feasible and practical. However, in the same issue of Diabetes Care that the IADPSG recommendations appeared in there was also a commentary (4) that asked whether the new criteria represented a “problem solved or a Pandora’s box?” Where have we progressed since 2010? Has the problem been solved or have we opened a Pandora’s box—a process that, once commenced, generates many complicated problems?

Questions were asked and comments, both constructive and critical, followed and have expanded with time. Could expert opinion replace scientific evidence (5)? The greatest concern was the lack of clinical trial evidence that treatment at these glucose thresholds, using a one-step diagnostic strategy, improved pregnancy outcomes, as well as the potential for collateral harm that might be caused by an increase in obstetric and neonatal interventions (5–7). Concerns were raised about the universally anticipated increase in the prevalence of GDM, and these have been borne out by reports of GDM being diagnosed in as many as 30% of some populations under the new criteria (8,9). Finally, the cost-effectiveness remains questionable, with retrospective observational studies giving conflicting results (9,10).

Further confusion, or lack of clarity, relates to the decisions and recommendations of professional groups and societies. The World Health Organization adopted the IADPSG diagnostic criteria in 2013 with the precaution that the quality of the evidence was very low and the strength of the recommendation was weak (11). The National Institutes of Health considered the potential new criteria and decided to remain with the status quo until comparative clinical trials could be conducted (12). The American Diabetes Association’s Standards of Medical Care in Diabetes—2018 (13) for the diagnosis of GDM has the option of either the one-step IADPSG criteria or a two-step procedure with three different postglucose-challenge action points and two different diagnostic options!

In this issue of Diabetes Care, the observations of McIntyre et al. (14) in a Danish cohort challenge the universal suitability of the IADPSG diagnostic thresholds. In a northern European Caucasian population, who would normally be considered low risk for GDM, 40% had fasting glucose ≥5.1 mmol/L and would have been so diagnosed. There was no evidence of excessive fetal growth (on the basis of large for gestational age or neonatal anthropometry) or hypertensive disease of pregnancy in this population until the fasting glucose level was ≥5.6 mmol/L, well above the IADPSG threshold of 5.1 mmol/L, and there was no relationship between cesarean section and fasting glucose at all. These findings are similar to a recent Spanish study that only demonstrated a sharper rise in the risk of large for gestational age and gestational hypertension when fasting glucose was above 5.3–5.5 mmol/L (15). Now, this variance, compared with the continuum of risk demonstrated by the HAPO study, may be related to the numbers involved or it could reflect true ethnic variation. There also may be something hitherto unrecognizably different about the Danish population, as the GDM prevalence for women with a BMI ≥29, at more than 50%, was the highest in a group of selected European countries (16). The McIntyre et al. study also needs to be interpreted with caution as only half the women in the cohort had a fasting glucose level performed and there were some differences in characteristics between these subjects and those who did not. Nevertheless, the authors justifiably ask the question of whether it would be appropriate to treat the 40% of women who would have been diagnosed on the basis of a fasting glucose level ≥5.1 mmol/L.

What are the implications of this study? Although international standardization of the glucose thresholds for the diagnosis of GDM is highly desirable, as McIntyre et al. (14) stated, perhaps “one size does not fit all.” Should each country undertake local studies and reach its own consensus regarding the most appropriate glucose thresholds for its population and health system? Or, perhaps, the selection by the IADPSG/World Health Organization of the OR of 1.75 should be reconsidered?

Clearly, using an OR of 1.75 for the diagnosis of GDM based on HAPO has not gained the acceptance anticipated for diverse reasons. However, it would be wise to retain the science of the HAPO study but consider shifting the diagnostic criteria up the curve to, say, an OR of 2.0. Applying an OR of 2.0 would increase the diagnostic glucose levels to ≥5.3 mmol/L at fasting, ≥10.6 mmol/L at 1 hr, and ≥9.0 mmol/L at 2 h. This increase would greatly reduce the prevalence of GDM (17) and place the critical fasting level for diagnosis in alignment with the established fasting level of the Carpenter and Coustan criteria. With a diagnostic fasting level of ≥5.3 mmol/L, the de facto treatment target would be to aim for a fasting level of <5.3 mmol/L, which would also agree with the evidence from the Maternal-Fetal Medicine Units (MFMU) Network randomized trial of treatment for GDM (18).

In the U.S., one of the key consequences of the two-step procedure is that the initial 1-h 50-g glucose challenge test effectively excludes women with fasting hyperglycemia alone from further testing, and therefore such women will not be diagnosed with GDM. By raising the fasting cutoff on the GTT to ≥5.3 mmol/L, there would be greater concordance between women diagnosed by the one-step and two-step strategies. Perhaps this will lead to greater acceptance of the one-step strategy.

With the confusion over international diagnostic criteria, the discordance between diagnostic criteria and treatment targets, and the inadequacy of economic data, we believe that it is now time for another international consensus meeting to discuss a revision to the criteria for the diagnosis of GDM. Until randomized controlled trials demonstrate improved outcomes treating GDM by a set of criteria derived from HAPO data, the decision regarding the glucose cutoffs to be applied will depend on what we now know about GDM prevalence, resource utilization and economics, and extrapolation of data from clinical trials using other criteria. The consensus in 2018 may well be that an OR of 2.0 should be applied to the diagnosis of GDM.