Portal de Periódicos da CAPES

Update on diagnosis of hyperglycemia in pregnancy and gestational diabetes mellitus from FIGO’s Pregnancy & Non‐Communicable Diseases Committee

2021; Elsevier BV; Volume: 154; Issue: 2 Linguagem: Inglês

10.1002/ijgo.13764

1879-3479

David McIntyre, Jeremy Oats, Anne‐Beatrice Kihara, Hema Divakar, Anil Kapur, Liona C. Poon, Moshe Hod,

Bariatric Surgery and Outcomes

The optimal process for detecting and treating gestational diabetes mellitus (GDM) remains controversial, with wide variances in practice within and between countries. A recent pragmatic randomized controlled trial (RCT) undertaken by Hillier et al.1 in the USA has raised renewed worldwide interest in and discussion of GDM diagnosis. The study reported no benefit at a population level in frequency of large-for-gestational-age (LGA) infants, hypertensive disorders of pregnancy, primary cesarean delivery, or a predefined composite perinatal outcome using a "one-step" diagnostic process and diagnostic criteria (primary diagnostic 75 g oral glucose tolerance test [OGTT]) as suggested in 2010 by the International Association of Diabetes in Pregnancy Study Groups (IADPSG)2 as compared to current dominant USA-based criteria,3 which employ a "two-step" process (initial nonfasting glucose challenge test [GCT] followed by a full OGTT if threshold glucose values are exceeded on the GCT). It should be noted that the USA criteria use very similar OGTT diagnostic thresholds to the IADPSG criteria, but require two elevated values for a GDM diagnosis. This time-honored practice lies contrary to the large body of evidence, confirmed in a recent USA-based systematic review, demonstrating that women with one abnormal value on their diagnostic OGTT carry substantially higher risks of LGA babies, cesarean delivery, pregnancy-induced hypertension, Apgar score <7, neonatal intensive care unit admission, and respiratory distress syndrome.4 Since the IADPSG criteria classified 16.5% of women as GDM compared to an 8.5% frequency using the dominant USA criteria in the Hillier et al. study,1 an accompanying editorial by Casey5 concluded that the IADPSG approach offered "no apparent maternal or perinatal benefit overall." While we applaud the efforts of Hillier et al.1 in conducting such an important study, we consider that the interpretation of the trial's results merits detailed scrutiny. The study by Hillier et al.1 was designed and conducted as a "pragmatic RCT." The major advantages of this approach were that the study was integrated into standard care across the two Kaiser Permanente health systems involved and recruited a very substantial number of women (23 792 randomized). The investigators obtained a waiver of individual consent, arguing that both methods of GDM screening were current, as recommended in the USA by various bodies. However, the trial design allowed providers the option of using a diagnostic strategy of their own choice, rather than complying with the randomly allocated strategy. Furthermore, care providers were aware of the screening strategy used for individual patients, although trial investigators and study staff remained blinded. There appeared to be substantial provider bias toward "business as usual" in the actual diagnostic strategy used, as only 66% of women in the one-step group received their allocated testing strategy, as compared to 92% in the two-step group. The authors attempted to adjust for this difference using inverse probability weighting, but residual bias remains likely. Considering the study results as published,1 we note that overall, 16.5% of women were classified as GDM with the one-step IADPSG approach compared to 8.5% with the two-step USA approach.3 Thus, in each group, approximately 83.5% received no GDM treatment and 8.5% received GDM treatment, irrespective of randomization. Thus, the actual therapy received differed only for 8% of participants, which clearly makes demonstration of any difference in outcomes at the population level difficult. Those women identified as having GDM in each group were managed in a similar manner. Under similar protocols for clinical care, the rates of pharmacotherapy were 43% (one step) and 46% (two step). This clearly suggests that the additional GDM women detected with the one-step IADPSG approach did not have trivial hyperglycemia. It is overwhelmingly likely that day-to-day hyperglycemic excursions, rather than mere glucose levels observed during the diagnostic OGTT, are related to overall pregnancy outcomes. Thus, it could be argued that although the more inclusive one-step strategy almost doubled GDM frequency, it also allowed identification of women with an almost equal risk of having hyperglycemia severe enough to merit pharmacotherapy. These women would have remained untreated under the two-step strategy. Although outcomes did not differ between the randomization groups at a population level, the question of whether the additional 8% of women considered GDM under the one-step approach derived benefit from their treatment cannot be determined from the data as presented. This is an important question and deserves specific consideration in a formal RCT. The publication by Hillier et al.1 focused primarily on USA-based clinical practice for GDM management and included predominantly white and Asian women. However, the issues related to GDM diagnosis are even more pressing on a global scale, especially in low- and middle-income countries (LMICs).6 Hyperglycemia in pregnancy affected 21.4 million live births in 2013, with more than 90% of cases occurring in LMICs.7 Populous LMICs (accounting for 90 million pregnancies worldwide) have limited resources for and access to prenatal care. It has been well documented that great challenges exist in systematic testing for and treatment of GDM in such countries,8 with uptake of universal screening—even when recommended—varying between 10% and more than 90%. Given these background problems, we consider that the increased complexity of a two-step approach to GDM diagnosis will lead to poorer uptake for the second visit because of multiple personal and health system factors, including scheduling issues, additional travel time, and cost. Such problems have been identified even in well-resourced countries such as Canada.9 In LMICs, the likely result would be that the majority of at-risk pregnancies would be effectively untested for GDM. This will result in a missed opportunity to manage those with hyperglycemia in pregnancy and reduce both immediate adverse pregnancy outcomes and address long-term risk of complications for both mother and baby. On this basis, we should not move back to two-step testing for such large populations with substantial cost implications in addition to issues of accessibility and availability of human resources. The two-step procedure fails to identify women primarily manifesting fasting hyperglycemia as they do not even qualify for the OGTT. In the Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study, the frequency of GDM diagnosed by the fasting glucose value ranged from 24%–26% in Thailand and Hong Kong to 73%–74% in the USA and Caribbean.10 Thus, the efficiency of postglucose load screening varies greatly across different geographic areas and ethnic groups. Moreover, a significant proportion of women fail to complete the evaluation as they do not turn up for the OGTT. This approach therefore may miss many women with hyperglycemia in pregnancy. Explicit in the Casey editorial5 and implicit in many other critiques of the IADPSG approach is the suggestion that "detection of gestational diabetes in almost one in five pregnant women" is implausible. In fact, as previously noted,11 the inverse appears true. The most recent report from the USA-based National Health and Nutrition Examination (NHANES) notes a frequency of 4.5% overt diabetes and 29.3% prediabetes in adults aged between 20 and 44 years in the USA.12 This age group also displays the highest prevalence of undiagnosed diabetes.13 Even younger population groups, for example adolescent females in the USA, display a reported prevalence of 0.6% overt diabetes and 13.2% prediabetes at 12–19 years. Thus, we would suggest that, if systematic prepregnancy testing were effectively instituted, the rates of hyperglycemia complicating pregnancy would be even greater than those reported by Hillier et al.1 Indeed, many cases currently identified as GDM due to pregnancy-based testing likely represent undiagnosed hyperglycemia that predates the onset of pregnancy.14 Other USA-based studies such as the CARDIA cohort15 have demonstrated that abnormal lipid and glucose parameters are present for many years before women are eventually diagnosed as having abnormal glucose metabolism through systematic GDM testing. Instead of focusing on "excessive" GDM diagnoses, a coordinated approach that aims to improve overall glucometabolic health in women at all stages of the life cycle appears warranted. The lack of an overall effect at the population level noted by Hiller et al.1 is not surprising, as 92% of women in each arm of this pragmatic RCT were treated in a similar manner. In fact, the small reduction in LGA frequency seen with one-step (8.9%) versus two-step (9.2%) testing is entirely consistent with predictions from previous studies,16, 17 allowing for the dilution of any treatment effect due to analysis at the population level. The study by Hillier et al.1 deals only with immediate pregnancy outcomes, whereas much of the personal and health economic "value" of GDM detection lies in the opportunity to identify women who carry long-term risks of developing overt diabetes, cardiovascular disease, and premature mortality.18-23 Data from the HAPO Follow Up Study (FUS) demonstrate that untreated IADPSG GDM women have an overall 41.5% risk of prediabetes and a 10.7% risk of overt diabetes after 11.4 years of follow-up.24 In the HAPO-FUS, women diagnosed post hoc with GDM using IADPSG criteria only (considered non-GDM on Carpenter–Coustan criteria closely aligned to the two-step criteria, as in Hillier et al.1), still had a 39.4% rate of prediabetes and a 7.9% rate of overt diabetes at 11.4 years. If the more stringent Carpenter–Coustan criteria were met,24 the relevant frequencies were 48.4% prediabetes and 20% overt diabetes. Thus, although the "IADPSG only" group do carry a somewhat lower long-term risk, we still consider that a 47.2% risk of abnormal maternal glucose metabolism after 11.4 years merits both detection during pregnancy and long-term follow-up to prevent progression to overt diabetes and allow early intervention. Further, Retnakaran et al.25 have demonstrated that, although the lifetime risk of cardiovascular disease following GDM is most marked in those women who progress to overt type 2 diabetes (hazard ratio [HR] 2.82; 95% CI, 2.241–3.30), an increased risk (HR 1.41; 95% CI, 1.11–1.80) is present even for those women who do not show such progression. This clearly underlines the importance of a GDM diagnosis in predicting future maternal health outcomes. Beyond the mere documentation of risk, it is clear that both effective lifestyle interventions26 and the use of metformin27 are able to reduce the risk of progression to overt diabetes following a GDM diagnosis. Thus, appropriate diagnosis of GDM identifies long-term remediable maternal health risks, which we consider deserving of attention both in terms of immediate pregnancy complications and long-term maternal health. From the offspring viewpoint, a maternal diagnosis of IADPSG GDM is also clearly associated with increased risks of adiposity and impaired glucose metabolism at 11.4 years of age, as again demonstrated by the HAPO FUS.24, 28 Interestingly, similar findings were reported by Hillier et al.29 in their 2007 publication. Like the more recent paper, this cohort included women from the Kaiser Permanente health system in Northwest USA and Hawaii.1 The 2007 publication noted that the offspring of women with less severe hyperglycemia in pregnancy (left untreated due to higher GDM diagnostic criteria used at Kaiser Permanente at that time) were more likely to be obese at age 5–7 years than the offspring of women with more severe hyperglycemia (who were actively treated during pregnancy). Hillier et al.29 concluded in 2007 that "GDM may be a modifiable risk factor for childhood obesity." Unfortunately, despite this publication, the efficacy of GDM treatment during pregnancy in preventing long-term offspring complications remains unproven, with negative results reported on follow-up of the landmark ACHOIS study by Gillman et al.30 and US MFMNU trial by Landon et al.31 RCTs of GDM treatment. Although both of these reports have some methodologic limitations, there is currently no evidence that maternal GDM treatment according to current protocols during an index pregnancy prevents later offspring effects. Although both arms of the Hillier et al.1 trial used a strategy involving some form of glucose testing for all pregnant women, approaches to GDM detection vary widely across the world. FIGO (the International Federation of Gynecology and Obstetrics),6 the IADPSG consensus group,2 and the American College of Obstetricians and Gynecologists (ACOG)3 have all favored universal testing, on the basis that GDM is generally without symptoms and that the associations of hyperglycemia with adverse pregnancy outcomes are independent of other risk factors. However, the UK National Institute for Health and Care Excellence (NICE) still promotes biochemical testing only in women with identifiable demographic or clinical risk factors,32 basing this argument primarily on health economic considerations, despite another UK report that concluded that GDM screening was cost-effective if population GDM frequency was greater than 4.2%.33 When selective screening is undertaken, compliance with the risk factor-based protocols is generally poor, estimated at 31% in Sweden34 and 61% in the UK.35 In countries with a high diabetes prevalence, such as Sri Lanka, it has been reported that only 13% of GDM cases would be missed by risk factor-based screening, but the high prevalence of known risk factors means that only 20% of women would not require testing.36 Thus, on a practical level, we continue to strongly endorse universal screening. The other major controversy in GDM detection, which forms the major focus of the Hillier et al.1 paper, is whether women should be "screened" or "tested" for GDM. In general, use of the "screen first" approach implies that the initial test will be both less burdensome than full diagnostic testing and will also have a high negative predictive value, so that few cases will be missed. In the case of GDM, the two-step approach relies on initial detection of postglucose load hyperglycemia after a nonfasting test, which can be conducted at a routine clinic visit. This offers some streamlining of testing processes, but on the other hand many women need to return for a second test. The two-step strategy has been estimated to miss approximately 25% of women who would be diagnosed on a full OGTT.37 Further, women who were screen positive with the first step (GCT) of the two-step strategy may have modified their dietary intake and other lifestyle factors prior to their diagnostic OGTT, which may also account for some of the difference in prevalence. By contrast, the one-step approach offers a single diagnostic test, so all women receive a definitive "Yes/No" GDM diagnosis. Given the known variability in OGTT results due to clinical, preanalytical, and analytical factors,38, 39 this does mean that a woman, if tested on more than one occasion, may receive a different diagnostic classification. However, since the initial management of GDM involves primarily dietary modification and enhanced physical activity, with pharmacotherapy instituted only if glycemic targets are not met, the risk involved with a false positive GDM diagnosis appears less than the risk of failing to detect significant maternal metabolic risk. As noted above, we consider that the study by Hillier et al.1 makes an important contribution to our understanding of GDM and its relationship to population level outcomes. The major pregnancy outcomes considered in Hillier et al.1 relate to and may be caused by multiple factors, not simply relating to a Yes/No dichotomous GDM diagnosis. As previously noted, a "risk engine" approach may give a truer outline of the relationship between maternal demographic and biologic risk factors and the frequency of adverse pregnancy outcomes.40, 41 In particular, in situations where maternal obesity is highly prevalent, its effects on pregnancy complications may predominate, especially on a population level.42 Thus, in contrast to an optimistic early report from Duran et al.43, the use of IADPSG criteria rather than Carpenter–Coustan criteria for GDM diagnosis does not appear likely to reduce the overall rates of pregnancy complications seen in routine obstetric practice. The study by Hillier et al.1 does strongly suggest that detection and treatment of GDM per se will not provide a solution to the vicious cycle of intergenerational obesity and metabolic disturbance, nor to the worldwide "slow motion disaster" of obesity and diabetes.44 Nonetheless, use of more inclusive criteria for GDM diagnosis may benefit individual women and their offspring in terms of immediate pregnancy outcomes and it also, very importantly, gives a vital opportunity to improve lifetime maternal metabolic and cardiovascular health. This is a vital question for future research, which should focus on determining whether the substantial group of women (and their offspring) who are "one-step positive/two-step negative" in terms of GDM diagnosis achieve a reduction in adverse outcomes from the identification and treatment of maternal GDM, both from the maternal and offspring perspective and both in the short and long term. The FIGO pragmatic guidelines for hyperglycemia in pregnancy/GDM6 provide a workable framework for accommodating differing diagnostic and therapeutic approaches to GDM across a variety of countries and healthcare systems. A FIGO group including many of the authors of the current article have also recently published a proposed strategy for moving toward a global consensus regarding GDM diagnosis, accommodating many of the fine print differences currently seen between the USA and other countries.45 We believe that this pluralistic and inclusive approach, adapted to local circumstances, continues to provide the best opportunity to effectively manage GDM in the current global context, whilst hoping for a more consensus-driven process in the future.45, 46 As a wise woman has stated: "Let not the ideal become the enemy of the good." H. David McIntyre reports payment or honoraria for lectures, presentations, speakers' bureaus, manuscript writing or educational events from Mead Johnson China and from the International Association of Diabetes and Pregnancy Study Groups. Anil Kapur reports a leadership role in the World Diabetes Foundation, and stocks or stock options from Novo Nordisk A/S. All other named authors have no conflicts of interest to declare. HDM was responsible for drafting the manuscript. All authors contributed to editing the manuscript and all authors approved the final version.