Introduction

First-trimester screening in pregnancy has traditionally focused on the fetus: aiming to define risks of chromosomal abnormality and to identify fetuses affected by structural anomalies . Screening for chromosomal abnormality at 11 to 13 ⁺⁶ weeks’ gestation has proven to be very effective. Predictive algorithms have gradually become more complex and refined, and it is now possible to detect >90% of fetuses with Down syndrome for <5% false-positive rate . The development of first-trimester screening for chromosomal abnormality has also helped clinicians recognize the importance of standardization of measurement for quality-assured risk analyses. The current algorithms for Down syndrome screening combine maternal history, biochemical parameters (free beta human chorionic gonadotropin (ßhCG), pregnancy-associated plasma protein A (PaPP-A) and placental growth factor (PlGF)) and biophysical parameters (ultrasound-assessed fetal nuchal translucency (NT) and nasal bone or ductus venosus flow) to assess risk . The use of this combination of factors recognizes the heterogeneity of the condition being assessed, a principle that is readily transferred when developing screening tools for the great obstetric syndromes, pre-eclampsia (PET) and gestational diabetes (GDM).

Whilst maternal and infant mortalities continue to be a significant challenge in many countries, the rates of maternal and perinatal mortality are significantly lower in industrialized nations . In this environment, it seems appropriate to review the strategic role of obstetric care and consider whether this should be expanded to include a more detailed assessment of a woman’s risk of morbidity in pregnancy. Diseases such as PET and GDM not only present significant risk to mother and fetus during pregnancy but are also recognized as having a longer-term impact on maternal and infant health . By focusing on early prediction of these conditions, it may be possible to prevent the development of symptomatic disease later in pregnancy and we hypothesize that this may be of value to not only short-term but also long-term outcomes for mother and fetus.

A number of research groups are currently attempting to develop predictive algorithms for the risk of PET and GDM. The rationale for PET screening is more clear-cut, as there are meta-analyses that demonstrate the value of early therapeutic intervention in the prevention of disease. Whilst there is currently no recognized intervention available to prevent GDM, the prevalence of this disease is increasing and early prediction may provide a means of improving the morbidity associated with this condition.

Prediction and prevention of PET

PET is recognized as one of the leading causes of maternal and perinatal mortality and morbidity, accounting for 16% of maternal deaths in industrialized nations . The gestation of onset and course of this disease is unpredictable, leading, in the 1920s and 1930s, to the development of a programme for increasingly regular clinical assessment during the antenatal period . Together with the use of anti-hypertensive medications and neuroprotective agents to prevent eclampsia, this has led to a significant reduction in maternal mortality in developed countries . The aetiology of PET is complex and involves multiple pathways, but these can be grouped into processes related to the primary development of poor placentation and a cascade of secondary responses, which lead to the symptoms and signs of this disease . Current surveillance and therapeutic interventions focus on responding to the second stage of this disease process. An alternative strategy would be to attempt to recognize the primary pathology at an earlier gestation, through identification of biomarkers associated with failure of trophoblast invasion, and to take steps to avoid the progression of disease.

All clinicians involved in antenatal care are aware of the risk of hypertensive disease in pregnancy and take a detailed medical history from women seeking obstetric care to try and define whether they are at risk . This strategy has not proven to be useful in reducing the prevalence of disease – in part because maternal history is an ineffective screening tool but also because this assessment has often occurred too late (>16 weeks’ gestation) to allow timely intervention. The majority of women who develop PET do have risk factors that will be revealed through such a booking assessment but this is not a specific enough tool to act as the reliable basis for any intervention . The standard clinical assessment involves a binary approach (risk factors are either present or absent) and computerization of this, applying odds ratios to the presence or absence of each factor, improves both sensitivity and specificity . Nevertheless, several studies have shown that screening on the basis of maternal history alone is insensitive when measured at a fixed 90% level of specificity .

The development of a more complex algorithm for first-trimester risk assessment for PET is attractive for a number of reasons. First, there is now evidence that early (<16 weeks’ gestation) intervention, through the prescription of aspirin, can reduce the prevalence of early-onset disease . Second, first-trimester screening for aneuploidy has led to earlier attendance for antenatal care. Third, some factors, such as PaPP-A, measured for aneuploidy screening are recognized as being altered in women that subsequently develop PET . Fourth, screening tools that are predictive of PET at later gestations, such as the uterine artery Doppler waveform, can be reproduced at earlier gestations and can therefore be incorporated into the 12-week scan .

Poon et al. (2009) described a multiple logistic regression model screening for PET at 11–13 ⁺⁶ weeks’ gestation . This model defined a risk for all women based on demographic factors, maternal mean arterial pressure, uterine artery Doppler pulsatility index (PI) and the biochemical markers PAPP-A and PlGF. The model was most effective in screening for PET that required delivery <34 weeks’ gestation (‘early’ pre-eclampsia or ePET), with >90% sensitivity for 95% specificity. The model was less effective in identifying women with late-onset PET or those who developed gestational hypertension later in pregnancy. This is significant, because <20% of women who develop PET have the severe, early-onset form of the disease; whilst they are more likely to suffer significant maternal and fetal mortality and morbidity, we need to recognize that much of the burden of disease seen in developing nations is, in fact, of late rather than early onset. Despite this, effective treatment that reduces the prevalence and/or severity of ePET would be a significant development in the management of this condition.

The work of Poon et al. has been validated in other settings. In our own group, we applied the Fetal Medicine Foundation (FMF) algorithm in an observational study of 3000 women attending for routine aneuploidy-based first-trimester screening . In this cohort, the algorithm performed as predicted; using a combination of maternal history, mean arterial pressure, uterine artery Doppler and PaPP-A, >90% of ePET cases were predicted for a 90% specificity. Other combinations of markers have also been described: A Spanish group used an algorithm including the biochemical markers free ßhCG and PaPP-A in a series of >5000 low-risk women with 81% sensitivity for 90% specificity . A Chilean group reported an algorithm involving demographic and biophysical but no biochemical parameters with a sensitivity of 75% for 90% specificity , whilst a Dutch nested case control study suggested that a combination of mean arterial pressure and biochemical markers (PAPP-A, ADAM12 and PlGF) was similarly effective (sensitivity 72%; specificity 90%) . In each of these studies, prediction was far superior for ePET than for disease leading to delivery >34 weeks’ gestation.

Recently, a US-based group has attempted to retrospectively validate these algorithms in their own population, and found performance to be lower than that that had been anticipated . Whilst the authors correctly identify that this may be due to differences between stuffy populations, it is interesting that some markers, such as uterine artery Doppler parameters, which have been quite robust in other data sets, were not of significant value in risk prediction here. In our own practice, we have recognized the potential to underperform by poor measurement of uterine artery PI and have spent considerable time developing local normal ranges and audit techniques for this parameter .

Other groups have also had difficulty in generating algorithms that are effective as screening tools. The US-based National Institute of Child Health and Human Development (NICHD) maternal–fetal medicine (MFM) network described a sensitivity of 46% for 80% specificity – but it did not stratify these data for the gestation at delivery of the pre-eclamptic group . Interestingly, a more recent publication from the same group demonstrated significantly better detection for ePET than for either PET leading to delivery >34 weeks or severe PET developed at any gestation . The Screening for Pregnancy Endpoints (SCOPE) consortium, who reported serum sampling at 14–16 weeks’ gestation rather than 11–13 weeks’ gestation, found that PlGF was associated with PET (leading to delivery <37 weeks’ gestation), but was not sufficiently strong to be used as an isolated marker for prediction of risk .

The National Institute for Health and Care Excellence (NICE) guideline suggests that women should be screened for PET and that those at high risk should be offered aspirin from an early (<16 weeks) gestation . Groups that have comprehensive programmes for combined first-trimester screening for aneuploidy are in a position to change imaging protocols to include measurement of uterine artery Doppler indices, and therefore to offer a more effective screening test for ePET than a review of medical history alone. It is important to recognize the limitations of screening though – and that these tools are not as effective at defining pregnancies at risk of either severe or late (delivery >34 weeks) PET . Similarly, it is important to recognize that no screening algorithm has been described for women with multiple pregnancies. Whilst incorporation of PET screening into the combined first-trimester test presents some difficulties for counselling, expedience dictates this is both the most appropriate gestational point and cost-effective approach for screening . There is no evidence that women defined as being at high risk for ePET have higher levels of anxiety later in their pregnancy .

One final aspect of analysis of first-trimester screening for PET should involve an economic evaluation of the cost benefit of universal screening. Obstetric services that have already provided access for combined first-trimester screening will face a minimal increase in expenditure to screen for PET at 12 weeks of pregnancy. The costs of the intervention, aspirin, are minimal and the potential savings – with a reduction in significant length of neonatal admissions – are very significant. There are no published data reviewing the cost-effectiveness of this intervention, but in our own service, making a number of assumptions about the likely performance of both the screening and preventative parts of this process, we have estimated cost savings of >$700,000/year by screening and intervening in the 5000 pregnancies attending our obstetric unit.

Prediction and prevention of PET

PET is recognized as one of the leading causes of maternal and perinatal mortality and morbidity, accounting for 16% of maternal deaths in industrialized nations . The gestation of onset and course of this disease is unpredictable, leading, in the 1920s and 1930s, to the development of a programme for increasingly regular clinical assessment during the antenatal period . Together with the use of anti-hypertensive medications and neuroprotective agents to prevent eclampsia, this has led to a significant reduction in maternal mortality in developed countries . The aetiology of PET is complex and involves multiple pathways, but these can be grouped into processes related to the primary development of poor placentation and a cascade of secondary responses, which lead to the symptoms and signs of this disease . Current surveillance and therapeutic interventions focus on responding to the second stage of this disease process. An alternative strategy would be to attempt to recognize the primary pathology at an earlier gestation, through identification of biomarkers associated with failure of trophoblast invasion, and to take steps to avoid the progression of disease.

All clinicians involved in antenatal care are aware of the risk of hypertensive disease in pregnancy and take a detailed medical history from women seeking obstetric care to try and define whether they are at risk . This strategy has not proven to be useful in reducing the prevalence of disease – in part because maternal history is an ineffective screening tool but also because this assessment has often occurred too late (>16 weeks’ gestation) to allow timely intervention. The majority of women who develop PET do have risk factors that will be revealed through such a booking assessment but this is not a specific enough tool to act as the reliable basis for any intervention . The standard clinical assessment involves a binary approach (risk factors are either present or absent) and computerization of this, applying odds ratios to the presence or absence of each factor, improves both sensitivity and specificity . Nevertheless, several studies have shown that screening on the basis of maternal history alone is insensitive when measured at a fixed 90% level of specificity .

The development of a more complex algorithm for first-trimester risk assessment for PET is attractive for a number of reasons. First, there is now evidence that early (<16 weeks’ gestation) intervention, through the prescription of aspirin, can reduce the prevalence of early-onset disease . Second, first-trimester screening for aneuploidy has led to earlier attendance for antenatal care. Third, some factors, such as PaPP-A, measured for aneuploidy screening are recognized as being altered in women that subsequently develop PET . Fourth, screening tools that are predictive of PET at later gestations, such as the uterine artery Doppler waveform, can be reproduced at earlier gestations and can therefore be incorporated into the 12-week scan .

Poon et al. (2009) described a multiple logistic regression model screening for PET at 11–13 ⁺⁶ weeks’ gestation . This model defined a risk for all women based on demographic factors, maternal mean arterial pressure, uterine artery Doppler pulsatility index (PI) and the biochemical markers PAPP-A and PlGF. The model was most effective in screening for PET that required delivery <34 weeks’ gestation (‘early’ pre-eclampsia or ePET), with >90% sensitivity for 95% specificity. The model was less effective in identifying women with late-onset PET or those who developed gestational hypertension later in pregnancy. This is significant, because <20% of women who develop PET have the severe, early-onset form of the disease; whilst they are more likely to suffer significant maternal and fetal mortality and morbidity, we need to recognize that much of the burden of disease seen in developing nations is, in fact, of late rather than early onset. Despite this, effective treatment that reduces the prevalence and/or severity of ePET would be a significant development in the management of this condition.

The work of Poon et al. has been validated in other settings. In our own group, we applied the Fetal Medicine Foundation (FMF) algorithm in an observational study of 3000 women attending for routine aneuploidy-based first-trimester screening . In this cohort, the algorithm performed as predicted; using a combination of maternal history, mean arterial pressure, uterine artery Doppler and PaPP-A, >90% of ePET cases were predicted for a 90% specificity. Other combinations of markers have also been described: A Spanish group used an algorithm including the biochemical markers free ßhCG and PaPP-A in a series of >5000 low-risk women with 81% sensitivity for 90% specificity . A Chilean group reported an algorithm involving demographic and biophysical but no biochemical parameters with a sensitivity of 75% for 90% specificity , whilst a Dutch nested case control study suggested that a combination of mean arterial pressure and biochemical markers (PAPP-A, ADAM12 and PlGF) was similarly effective (sensitivity 72%; specificity 90%) . In each of these studies, prediction was far superior for ePET than for disease leading to delivery >34 weeks’ gestation.

Recently, a US-based group has attempted to retrospectively validate these algorithms in their own population, and found performance to be lower than that that had been anticipated . Whilst the authors correctly identify that this may be due to differences between stuffy populations, it is interesting that some markers, such as uterine artery Doppler parameters, which have been quite robust in other data sets, were not of significant value in risk prediction here. In our own practice, we have recognized the potential to underperform by poor measurement of uterine artery PI and have spent considerable time developing local normal ranges and audit techniques for this parameter .

Other groups have also had difficulty in generating algorithms that are effective as screening tools. The US-based National Institute of Child Health and Human Development (NICHD) maternal–fetal medicine (MFM) network described a sensitivity of 46% for 80% specificity – but it did not stratify these data for the gestation at delivery of the pre-eclamptic group . Interestingly, a more recent publication from the same group demonstrated significantly better detection for ePET than for either PET leading to delivery >34 weeks or severe PET developed at any gestation . The Screening for Pregnancy Endpoints (SCOPE) consortium, who reported serum sampling at 14–16 weeks’ gestation rather than 11–13 weeks’ gestation, found that PlGF was associated with PET (leading to delivery <37 weeks’ gestation), but was not sufficiently strong to be used as an isolated marker for prediction of risk .

The National Institute for Health and Care Excellence (NICE) guideline suggests that women should be screened for PET and that those at high risk should be offered aspirin from an early (<16 weeks) gestation . Groups that have comprehensive programmes for combined first-trimester screening for aneuploidy are in a position to change imaging protocols to include measurement of uterine artery Doppler indices, and therefore to offer a more effective screening test for ePET than a review of medical history alone. It is important to recognize the limitations of screening though – and that these tools are not as effective at defining pregnancies at risk of either severe or late (delivery >34 weeks) PET . Similarly, it is important to recognize that no screening algorithm has been described for women with multiple pregnancies. Whilst incorporation of PET screening into the combined first-trimester test presents some difficulties for counselling, expedience dictates this is both the most appropriate gestational point and cost-effective approach for screening . There is no evidence that women defined as being at high risk for ePET have higher levels of anxiety later in their pregnancy .

One final aspect of analysis of first-trimester screening for PET should involve an economic evaluation of the cost benefit of universal screening. Obstetric services that have already provided access for combined first-trimester screening will face a minimal increase in expenditure to screen for PET at 12 weeks of pregnancy. The costs of the intervention, aspirin, are minimal and the potential savings – with a reduction in significant length of neonatal admissions – are very significant. There are no published data reviewing the cost-effectiveness of this intervention, but in our own service, making a number of assumptions about the likely performance of both the screening and preventative parts of this process, we have estimated cost savings of >$700,000/year by screening and intervening in the 5000 pregnancies attending our obstetric unit.

Prediction and prevention of GDM

GDM, defined as carbohydrate intolerance resulting in hyperglycaemia with onset or first recognition during pregnancy (World Health Organisation 1999), is one of the most common complications of pregnancy and is associated with significant adverse maternal and neonatal sequelae . The prevalence of GDM is rising internationally and parallels the increased prevalence of advanced maternal age, obesity and type 2 diabetes mellitus (T2DM) in pregnant women. Diagnosis and treatment in the latter part of the second trimester of pregnancy have been shown to improve pregnancy outcomes . Universal diagnostic testing is recommended as pre-identifiable risk factors are only recognizable in 50% of women with GDM .

Normal pregnancy is characterized by progressive insulin resistance and hyperlipidaemia arising from placental secretion of counter-regulatory hormones including growth hormone, corticotrophin-releasing hormone, placental lactogen, progesterone, oestrogen and prolactin. These physiological changes, that ensure adequate fetal growth and development and that the additional energy demands of delivery and lactation are met, are evident by 16–18 weeks’ gestation and are most pronounced late in the third trimester . GDM occurs when maternal insulin secretion is unable to compensate for increased insulin resistance in pregnancy. Maternal hyperglycaemia leads to increased transplacental glucose transport, fetal hyperglycaemia and hyperinsulinaemia and accelerated fetal growth .

Interpretation of the literature related to GDM is confounded by the lack of international consistency in its definition. Although controversy remains, there is evolving consensus towards international standardization of the diagnosis of hyperglycaemia based on the findings of the Hyperglycaemia and Adverse Pregnancy Outcome (HAPO) study . Whilst the lower glycaemic thresholds described in this protocol increase the prevalence of disease (potentially as high as 15%), this avoids the lower sensitivity of a two-step process that relies on the results of a glucose challenge test (GCT) to generate a high-risk cohort for a formal oral glucose tolerance test (OGTT) .

The rationale for first-trimester screening for GDM is twofold: first, that prediction may allow the prevention or development of milder maternal hyperglycaemia, which is associated with a lower risk of adverse outcomes ; and, second, that prevention of GDM will reduce the risk of maternal T2DM in later life . Neither of these hypotheses has been proven, although there is some evidence of fetal benefit to early intervention . Early diagnostic testing in high-risk groups identifies significant numbers (29–66%) of women who already have evidence of GDM . There is also evidence that an earlier diagnosis of GDM is associated with worse maternal and fetal morbidity . The potential for screening for GDM in the first trimester is also demonstrated by the fact that a number of maternal or placental biomarkers involved in recognized pathophysiological pathways appear to be altered at this early stage.

Observational studies have identified a variety of both modifiable and non-modifiable risk factors associated with GDM. These include advanced maternal age, increasing parity, ethnicity , maternal obesity , high gestational weight gain , physical inactivity , low-fibre high-glycaemic-load diets , history of previous macrosomia or GDM , family history of DM and history of polycystic ovarian syndrome (PCOS) . Medications such as glucocorticoids and antipsychotic therapy are also associated with GDM . A history of previous GDM appears to be the strongest predictor of subsequent GDM (associated with a 16-fold increased risk); however, the increasing prevalence of primiparity and advanced maternal age underscores the limitations of the current risk factor approach to GDM prediction . Studies assessing the predictive value of traditional clinical risk factor models show variable success in sensitivity with poor specificity and low positive predictive value . Most recently, likelihood ratios have been published that allow incorporation of historical factors into algorithms for risk assessment that can be developed in a similar way to those used for aneuploidy and PET screening .

The International Association of the Diabetes and Pregnancy Study Groups (IADPSG) guidelines do not currently advocate universal first-trimester screening, rather early testing in high-risk pregnancies (defined on the basis of maternal history) . A number of studies have examined markers of hyperglycaemia including using fasting blood glucose levels, the GCT and the OGTT. One recent study compared the screening efficacy of these three approaches in a population with 10.9% prevalence of GDM, describing sensitivities of 47.2%, 68.2% and 87.1% for 77.4%, 100% and 100% specificity, respectively . The authors concluded that an OGTT was the best option for testing a high-risk population. This would be difficult to incorporate in a universal 12-week screening programme, where women are seen for combined first-trimester screening at various times of the day (making fasting difficult) and where multiple points of sampling and the additional time taken in first-trimester screening would make compliance difficult.

A number of studies have focused on assessing serum concentrations of markers that are currently used in first-trimester screening. These placental proteins affect placental development and carbohydrate homeostasis and could therefore be of value in screening for GDM as well as in screening for chromosomal abnormality and PET. PaPP-A is a metalloprotease produced by trophoblasts. It modulates insulin-like growth factor 1 (IGF-1) bioavailability; low PaPP-A and IGF-1 levels could potentially lead to hyperinsulinaemia and increase insulin resistance . PaPP-A is routinely assessed as part of the first-trimester combined screening for aneuploidy, and its levels are reduced in trisomic pregnancies. A number of confounding factors are known to affect PaPP-A levels and raw data are typically adjusted for gestational age, maternal weight and ethnicity, maternal smoking and history of fertility treatment in order to produce a customized multiple of the median (MoM) value that can then be used in risk analysis . Spencer et al. (2013) reviewed a series of 27,660 pregnancies that attended for combined first-trimester screening, 8299 of whom had had a GTT later in pregnancy . They described a significant reduction in mean MoM in pregnancies that developed GDM (0.91 MoM) compared to controls (1.00 MoM). Another study reviewed the first-trimester PaPP-A results of 306 women who had subsequently developed GDM during their pregnancy and compared these to a similar-sized cohort of women matched as controls . They showed a significant reduction in median PaPP-A levels (0.9MoM in GDM pregnancies versus 1.3MoM in controls). This difference was used to develop a categorical process of risk assessment; a PaPP-A MoM <0.62 was associated with a 4.8-fold increase in the risk of developing GDM. Whilst this contributed to an improvement in the screening efficacy of a clinical risk scoring system, the test still performed relatively poorly with 81.4% sensitivity for 50.5% specificity. Other authors have reported that PaPP-A is not altered in pregnancies that subsequently develop GDM .

A second placental protein that is commonly measured during combined first-trimester screening is PlGF. This protein promotes placental vascular angiogenesis, and is reduced in trisomic pregnancies and in pregnancies that subsequently develop PET or fetal growth restriction. It is being introduced into first-trimester screening as it improves the positive predictive value for these conditions. In a series of 40 pregnancies that were affected by GDM matched to 94 controls, PlGF levels were found to be significantly increased . This correlated with PaPP-A levels, so these markers may not be independent of each other. Analysis by logistical regression (which would be limited by the small number of cases) was used to produce an algorithm combining maternal age, maternal weight and log ₁₀ PlGF to predict the development of GDM. This performed with 71.4% sensitivity for 75% specificity.

Moving beyond traditional first-trimester markers, researchers have examined the potential predictive value of markers for insulin resistance, anomalous lipid metabolism and of chronic inflammation. These markers are all recognized as being associated with the development of T2DM. Sex hormone-binding globulin (SHBG) is a glycoprotein that acts as a carrier protein for circulating oestrogen and testosterone. Insulin inhibits the synthesis of SHBG by the liver and a reduction in SHBG is a marker of hyperinsulinaemia and insulin resistance – and has the advantage that it can be measured in the non-fasting state – so is clinically applicable. A number of groups have examined the performance of SHBG as a first-trimester predictive marker for GDM. Smirnakis et al. (2007) described significantly lower levels of SHBG in pregnancies later affected by GDM (185.1 vs. 255.6 nmol/l), although others have suggested that the wide standard deviation of these distributions mean this will have minimal impact from a screening perspective . Other groups have also described a reduction in SHBG levels in pregnancies that go on to develop GDM, reporting 77.3% and 80.0% sensitivity for 43.5% and 84.5% specificity, respectively . SHBG levels measured at 13–16 weeks were lower and more highly predictive for GDM requiring insulin therapy (AUC 0.866). Nanda et al. (2011) developed an algorithm that incorporated the reduction seen in SHBG levels (mean MoM: 0.81 vs. 1.02 MoM) into an algorithm that defined an a priori risk based on multiple demographic factors . This had 69.9% sensitivity for 80% specificity.

Hyperinsulinaemia increases free fatty acid (FFA) production, leading to increased triglyceride (TG) synthesis and very low-density lipoprotein (vLDL) secretion. Increased FFA levels further inhibit maternal glucose uptake and oxidation and stimulate hepatic gluconeogenesis, exacerbating maternal peripheral insulin resistance. Both fasting insulin levels and fasting FFA are independent predictors of GDM . Brisson et al. (2013) showed that high TG levels at 11–14 weeks’ gestation were associated with a sixfold increase in the risk of developing GDM later in pregnancy.

Adiponectin is an adipocyte derived polypeptide with insulin sensitizing properties and appears to be the best marker of anomalous lipid metabolism associated with insulin resistance and B islet cell dysfunction . Reductions in serum adiponectin are recognized as being associated with T2DM . Several studies have examined the predictive value of first-trimester measurement of adiponectin for GDM. In one study, the mean adiponectin levels were almost halved in pregnancies subsequently affected by GDM (4.1 vs. 8.1 μg/ml); an adiponectin level <6.4 μg/ml was associated with a 4.6 × risk of GDM and had a sensitivity of 73% and specificity of 67% as a single screening tool . This level of difference in mean adiponectin levels was reported in two other studies – which also reported low standard deviation in measurement – suggesting that this would be an ideal marker for screening . There are, however, some conflicting data – with studies that show only a small, albeit significant difference between GDM and control pregnancies but with large standard deviation making incorporation in screening algorithms more difficult . Interestingly, another publication by the latter group suggests that converting adiponectin data to MoMs reveals a significant difference (mean 0.66 MoM GDM; 1.02 MoM controls) that is of value from a screening perspective . When this is added to the maternal demographic/SHBG model described earlier, the final sensitivity is 78% for 80% specificity.

An association between T2DM and inflammatory markers/markers of oxidative endothelial stress is also well established. There are publications that support this association in pregnancies that go on to be affected by GDM, although this may not remain significant once maternal obesity has been adjusted for . In a series of 300 pregnancies with a 13% prevalence of GDM, Berggren et al. (2014) log ₁₀ high-sensitivity C-reactive protein (hsCRP) was significantly increased, but became non-significant after adjustment for body mass index (BMI) . A similar finding was reported in a second series, although a significant elevation in tumour necrosis factor alpha (TNFα) remains after accounting for BMI .

There are very few data describing the prevention of GDM through interventions instituted in early pregnancy. Findings in management of T2DM may, however, be instructional. ‘Lifestyle changes’ have been shown to be effective in preventing/delaying the progression of T2DM . Lifestyle interventions target established modifiable risk factors including obesity, excessive gestational weight gain and physical inactivity, and encourage ingestion of a high-fibre, low-glycaemic diet. The most effective nutritional preventive strategies for GDM have not been established, but once GDM has been diagnosed there is evidence that focusing on sustainment of low postprandial glucose concentrations is effective in reducing fetal macrosomia . Moderate caloric restriction of one-third of total energy has been shown to improve glucose metabolism without causing ketonaemia in obese women with GDM .

In non-pregnant individuals, physical activity has been shown to reduce the risk of and delay the onset of T2DM . Physical activity improves insulin sensitivity via a number of mechanisms and reduces the risk of gestational weight gain . Even engagement in normal household activities appears to be associated with a reduced risk of GDM . There is, however, little compelling evidence to advocate increasing exercise in pregnancy at this time .

Alternative, therapeutic interventions that reduce the prevalence of GDM in women deemed to be at high risk also need further evaluation. A recent Finnish randomized controlled trial demonstrated that maternal ingestion of specific probiotics altered the composition of the gut microbiome in normal-weight women in pregnancy and that a combined dietary/probiotic intervention was associated with 50% reduction in the rate of GDM from 34% to 13% . The potential utility of such an approach in obese women is being investigated in the SPRING study (the Study of Probiotics IN the prevention of GDM), coupling a formal risk assessment based on BMI with a therapeutic intervention designed to reduce the later prevalence of disease .