In gestational diabetes mellitus (GDM) hyperglycemia develops during pregnancy, and GDM usually resolves after birth. This common disorder is rising in prevalence, predominantly due to the global increase in obesity, including in low- and middle-income countries (1).
Understanding of the fundamental mechanisms underlying GDM has not progressed substantially in the last 20 years; indeed, the lack of understanding of the (patho)physiology has led to a stagnation of therapeutic advances or significant improvements in diagnosis or management.
The interesting studies by Thaweethai et al. (2) and Mittendorfer et al. (3) in this edition of Diabetes Care both address an important topic: the relative roles of pancreatic β-cell insulin secretion and maternal insulin resistance in the control of glucose homeostasis in normal and hyperglycemic pregnancies. The underlying interplay between these factors has been infrequently explored, other than in the innovative studies by Catalano and colleagues in the 1990s, with characterization, for the first time, of the insulin secretory and resistance profile longitudinally in normal and hyperglycemic pregnancies (4–6). Thaweethai et al. and Mittendorfer et al. build on this background research.
The two articles, while addressing the relative roles of insulin resistance and β-cell function in GDM, focus on differing but complimentary aspects; Thaweethai et al. explore these parameters in early-onset versus “classic” GDM (diagnosis 24–28 weeks’ gestation), whereas Mittendorfer et al. concentrate on changes in these parameters between early pregnancy and the third trimester in women diagnosed with the disorder.
Thaweethai et al.: A Brief Overview
More recently, the existence of GDM that develops early in pregnancy in contrast to the classic diagnosis at ∼24–28 weeks’ gestation has been recognized and clinically embraced, despite continuing discussion about appropriate diagnostic glucose thresholds when the standard oral glucose tolerance test (OGTT) is applied in early pregnancy, as well as unclear benefits of treating mild hyperglycemia if identified.
The novelty of this well-constructed study lies in the interrogation of the balance between insulin secretion and resistance in women who develop “early” GDM in contrast to those more frequently studied who develop classic GDM. In an attempt to describe this succinctly the authors also propose a novel β-cell functional index, the pregnancy insulin physiology (PIP) index, a pregnancy equivalent to the disposition index, which they suggest may facilitate first-trimester prediction of classic GDM.
To summarize, using validated measures, the authors evaluated insulin resistance (Matsuda index [7]) and insulin secretory response (Stumvoll first-phase estimate [8]) to an OGTT at three time points, early pregnancy, mid-late pregnancy, and postpartum in 166 women. This multiethnic cohort comprised women with one or more risk factors for GDM (previous GDM, family history of diabetes or GDM, BMI ≥25 g/m2, plus one additional risk factor as listed by the American Diabetes Association [9]). To guide the reader, we found Supplementary Fig. 2 and Supplementary Table 1 the most useful when interpreting the data.
Considering all women (GDM and no-GDM) together, the authors reported an early pregnancy increase in insulin secretory response to the glucose load when compared with the nonpregnant (postpartum) response. This was independent of insulin resistance and aligns with the original work of Catalano and colleagues, and of others since (10,11). The physiological stimulus for enhanced insulin secretion is often attributed to influences of maternal and placental hormones on the pancreatic β-cell, although this concept has been challenged (11). Of potential importance to the etiology of GDM, when the same analysis was confined to women with GDM (both early and classic) the augmentation of the insulin secretory response (adjusted for insulin sensitivity, and assessed versus postpartum values) was deficient in early pregnancy in comparison with women without GDM.
Inadequate secretion in the early GDM group was associated with markedly low nonpregnancy secretory function, which may indicate a preexisting habitual β-cell abnormality. Poor secretion in this “early” GDM group was also paralleled by an early drop in insulin sensitivity with ensuing hyperglycemia. This differed from the classic GDM group, who were insulin resistant compared with non-GDM control subjects throughout pregnancy and postpartum. In these women, a more conserved initial insulin secretory response perhaps contributes to the observed maintenance of glucose homeostasis until mid-late gestation, at which point both show compromise, leading to hyperglycemia. To discriminate between insulin secretion and sensitivity the authors used a covariate-adjusted linear effects model of the two insulin indices (Stumvoll and Matsuda), plus the PIP index (Stumvoll × [Matsuda0.4]).
Mittendorfer et al.: A Brief Overview
This study population differs from that of Thaweethai et al. in that eligibility was confined to overweight or obese women with Black ethnicity, with no history of preexisting diabetes or GDM. Using multiple sampling at OGTT in both early (15 weeks) and late (35 weeks) pregnancy they compared fasting and post-load plasma glucose, free fatty acid (FFA) concentrations, and insulin indices including β-cell function between those defined as having GDM or not at 35 weeks. Diagnosis of OGTT is not currently clinically used or validated at this gestation, the classic window of testing being 24–28 weeks. Inclusion of these “classic” test data for the insulin indices would have aided interpretation; they were presumably absent because of practical limitations (this was a randomized controlled trial secondary analysis). An added complication of this study (see Supplementary Fig. 1) was the changing participant grouping due to fluctuating GDM diagnosis across gestation, with the result that individuals hyperglycemic at 24 weeks, but not at 35 weeks, were categorized as “No-GDM”; conversely, most women who had GDM at 35 weeks did not fulfill diagnostic criteria at 24 weeks. Some insight into the latter may be gained from our previous report illustrating how women with obesity with a negative “classic” OGTT (24–28 weeks) result then gained more weight than those with a positive test, possibly reflecting less restraint in dietary behaviors after the “reassuring” test result, and subsequent transition to hyperglycemia later in pregnancy (12).
Hyperglycemia was relatively mild in this study, as all individuals were managed with diet therapy. In consideration of the whole group, insulin sensitivity decreased, β-cell function increased, and postprandial suppression of plasma FFA decreased as pregnancy progressed from 15 to 35 weeks. At 15 weeks, individuals with GDM (diagnosed at 35 weeks) had lower insulin sensitivity and β-cell function but higher fasting and stimulated glucose compared with their non-GDM counterparts. Similarly, by 35 weeks, individuals with GDM experienced a greater drop in insulin sensitivity, a blunted rise in β-cell function, and a reduced decline in FFA following the glucose load in comparison with those without GDM. Differences in fasting and 1-h glucose at 15 weeks were found to be poor predictors of GDM using receiver operating characteristic analysis.
The blunting of FFA response to glucose, presumably as a result of tissue insulin resistance, is a welcome addition to substantive literature that typically describes glucose, insulin, and C-peptide responses at OGTT.
Both studies have approached a complex issue in assessment of the physiology and pathophysiology of glucose homeostasis in pregnancy. The standard clinical OGTT provides little insight into the intricacies of insulin secretion and insulin resistance, and insulin secretion is seldom evaluated in contemporaneous cohorts. A strength lies in the frequent sampling throughout the OGTT for better understanding of post–glucose load profiles of insulin and glucose to calculate insulin secretion and resistance and other indices. Furthermore, both studies included longitudinal profiling of these measures. Mittendorfer et al. provided added value through the measurement of FFAs.
The necessary statistical modeling in both articles as well as infrequently reported (or novel) indices such as PIP and adipose tissue insulin resistance (Adipo-IR) may prove difficult to interpret for the general reader. Despite this, both studies have described potentially important and novel insights into mechanisms of hyperglycemia in pregnancy.
Of note, both study populations unexpectedly showed no difference in BMI in early pregnancy between GDM and no-GDM groups; the literature generally reports a higher BMI in those who develop GDM. This may impact generalizability. Conversely, the choice of highly selected groups in both studies relates to specific groups of interest in clinical practice.
Specific limitations of the study of Thaweethai et al. include the use of postpartum data as a surrogate for prepregnancy insulin indices and glucose homeostasis. Furthermore, the authors recommend that the novel PIP index, described for the first time, might be a useful predictor of classic GDM but it first requires validation in other cohorts. The persistent reduction in insulin secretion in the early GDM group may provide a clinically useful marker of impending hyperglycemia in early pregnancy, especially if measured prepregnancy. The issue is how this would be practically measurable.
Specific limitations of Mittendorfer et al. include the diagnosis of GDM at 35 weeks’ gestation and the lack of inclusion of equivalent data at ∼24 weeks as well as the fluctuation of gestational glucose homeostasis among the women studied.
Both studies, albeit including only small groups of women with GDM (n = 45, n = 29), add important novel information to GDM research in better characterizing early and longitudinal pregnancy changes in insulin resistance and β-cell function, as well as profiling FFAs. However, they highlight in numerous ways the heterogeneity of this disorder (13). A comprehensive understanding of the etiology of GDM remains elusive and the quest continues. Novel and imaginative approaches, together with harmonization of protocols and key indices of physiological/pathophysiological function to facilitate meta-analysis and interrogation of key phenotypic subtypes, would take our understanding of this complex disorder to a higher-level enabling precision diagnosis and management.
Article Information
Funding. S.L.W. is funded by the Medical Research Council and King’s College London. L.P. is funded by Tommy’s Charity and King’s College London.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.