[page 45↓]

4.  Summary , discussion and perspectives

The standard management of pregnancies complicated by GDM is focused on the avoidance of maternal hyperglycemia. 1, 19, 79Diet education and self-glucose-monitoring is recommended in all women. Insulin therapy is added if the glucose values of the daily profiles exceed a certain threshold. 19 There is strong evidence that in GDM pregnancies maternal, fetal and neonatal morbidity increases with increasing maternal hyperglycemia. 43, 80-83 However, the relationship seems to behave in a continuous fashion 40, 41, 73, 74, 84, and diagnostic criteria for GDM as well as glycemic targets during pregnancy are rather based on expert opinion or consensus than on evidence. A great diversity exists regarding the criteria for the diagnosis of GDM. All threshold glucose values for the oral glucose tolerance test (oGTT) that had been used to define glucose intolerance are modifications of the original O’Sullivan criteria established in the early Sixties 21, and show a variation of up to 25 mg/dl for the post challenge glucose levels. 1, 19, 20, 85-87 Regarding glycemic control during pregnancy there is controversy if measurements at one hour postprandial are superior to the determination of two hour postprandial glucose values. 43, 88-90 The glucose targets show the same wide variation as the diagnostic criteria. Overall, over the years the rate of adverse outcome in pregnancies with GDM had been reduced, and stillbirth is now rare in treated women with GDM. However, despite good glucose control the rate of neonatal morbidity is still elevated compared to pregnancies of women with normal glucose tolerance. 1, 73, 91-93 The same phenomenon is seen in pregnancies of women with preexisting type 1 diabetes. 45, 94 Some groups aimed to solve the dilemma by applying very tight glucose control to all women with GDM. This approach could reduce the rate of accelerated growth but resulted in insulin therapy and the demand of a high frequency of glucose monitoring in the majority of the women. 57-59 This strategy appears to be cost intensive and questionable with respect to the occurrence of morbidity only in a minority of cases and lacks evidence based data indicating the necessity of intensive treatment in all women.

The primary goal of the present work was to determine to what extent maternal glycemia in GDM pregnancies treated according to the standard management predicts neonatal morbidity. In a second step, we investigated whether inclusion of fetal growth patterns improves the neonatal outcome and provides an additional clinical tool for antenatal risk assessment. Our studies covered four major questions. Are maternal glucose values helpful to assess the risk for morbidity in early pregnancy of women with GDM? Are the [page 46↓]existing glucose criteria for diagnosis and treatment of GDM reliable to identify pregnancies at increased risk for diabetic morbidity? What are the major determinants of accelerated growth in treated GDM pregnancies? Are concepts of management based on fetal growth criteria save and what are the advantages compared to the standard approach?

It is well known since the early Sixties that preexisting type 1 diabetes is associated with an increased risk for congenital anomalies in the offspring and that the rate of malformations is positively correlated with the degree of maternal hyperglycemia at time of embryogenesis. 23, 24, 26, 27, 29, 33, 95, 96 Up to 1995, there was a paucity of data investigating the occurrence of congenital anomalies in GDM. Malformations had not been considered as a problem in GDM pregnancies since pregnancy induced glucose intolerance is supposed not to occur before the second trimester. Thus, we retrospectively investigated the rate and risk factors of congenital anomalies in a large cohort of 3700 women with hyperglycemia first detected in pregnancy (i.e. GDM) who attended the Diabetic Prenatal Care Clinic of the Women’s Hospital of the University of Southern California in Los Angeles. The fasting glucose value at time of diagnosis was the strongest predictor for major malformations and we identified a glucose level of 120 mg/dl as threshold of an increased risk for anomalies. 97 The rate of major anomalies increased from 2.1% to 5.2 % for fasting glucose levels between 120-260 mg/dl. Interestingly, our identified threshold of 120 mg/dl corresponds to the glucose level in the first trimester previously identified for an increased risk for malformations in pregnancies with preexisting type 1 diabetes. 31 Furthermore, we investigated the pattern of congenital anomalies and their relationship to maternal fasting levels at diagnosis. We saw the same predominance of organ systems affected as in type 1 diabetes and a tight relation of the severity, i.e. the number of affected organ systems, and the level of fasting hyperglycemia. 98

Our findings were opposing the hypothesis that the physiologic decrease of glucose tolerance in pregnancy does reach a significant level far beyond embryogenesis. We were faced to the argument that a fair amount of the women in our population presumably had undiagnosed type 2 diabetes before pregnancy. The ethnic background of our predominantly Mexican-American women allows speculations about a high rate of undiagnosed type 2 diabetes although a fasting of 120 mg/dl does not even fulfill the recently lowered criteria for diabetes outside pregnancy. 37 However, when we excluded women with persisting diabetes postpartum diagnosed by an oGTT 4-16 weeks postpartum, we still confirmed our results. Considering the course of the development of pregnancy induced glucose intolerance, the women presumably have had even lower glucose values in the first trimester. Another aspect [page 47↓]was the high rate of obesity in the study population that is common in women with GDM. Obesity and the associated hyperinsulinemia by itself is a risk factor for malformations. 99, 100 Experimental studies demonstrated that diabetic embryopathy is associated with an excess of radical oxygen species. It had been shown that hyperglycemia leads to increased embryonic levels of the products of lipid peroxidation. 101 The physiologic increase of lipoxygenase activity is normally counterbalanced by an increase in the antioxidant system activity. 101, 102 It might be that in obese women with increased number of adipocytes a lower degree of hyperglycemia is sufficient to disturb this counterbalance. For obstetrical clinical care during pregnancy, the speculation of an underlying undiagnosed type 2 diabetes is of minor relevance and per definition the diagnosis of GDM comprises a wide range of severity of glucose intolerance. 1In conclusion, we could positively answer our question that maternal glucose values are helpful to assess the risk for morbidity in early pregnancy of women with GDM. High fasting glucose levels at diagnosis clearly indicate a risk for congenital malformations and should prompt an intensive specified fetal ultrasound examination. Our findings had been confirmed by recent studies. 103-106

In contrast to the obvious tight relation between maternal glucose values and embryonic morbidity in early pregnancy, we are faced with a different situation in the later course of pregnancy. The data of our studies indicate that neither the current diagnostic criteria nor the glucose values during therapy reliably predict neonatal morbidity. The diagnosis of GDM requires at least two pathologic glucose values independently of the applied criteria - either O’Sullivan, Carpenter and Coustan, ADA or WHO. We investigated specific parameters of diabetic morbidity in pregnancies of women with only one elevated glucose value in the oGTT according to O’Sullivan criteria, defined as impaired glucose tolerance (IGT). The rate of elevated amniotic fluid insulin (> 7 µU/ml), hypoglycemia (< 30 mg/dL), neonatal obesity, LGA and severe immaturity of the placenta was significantly higher in newborns of women with IGT compared to those from women with a normal oGTT. 73, 107 Neonatal obesity was defined as the sum of skinfold thickness measurements obtained at three sites of the body above the 90th percentile of gender-specific percentile rankings which were previously determined by measurements of skinfolds in 250 consecutive newborns with gestational age > 37 weeks. Hyperinsulinism, hypoglycemia and neonatal obesity were virtually even more frequent in IGT than in treated GDM pregnancies.

We also confirmed the impact of borderline glucose intolerance in a study that was primarily designed to determine the incidence and timing of and risk factors for hypoglycemia [page 48↓]in large-for-gestational-age (LGA) newborns of non-diabetic mothers. Excessive fetal growth may indicate fetal hyperinsulinism that exposes the newborn at high risk of hypoglycemia when the glucose supply suddenly drops after delivery. Therefore, frequent glucose testing is recommended in LGA newborns. 108, 109 Hypoglycemia occurred in 16% of the infants and the only predictor for hypoglycemia was the 1-hour glucose value of the maternal antenatal oGTT. 110 A threshold glucose level of 180 mg/dl revealed to be a good discriminator for an increased risk of hypoglycemia. The incidence sharply rises to 25% compared to 2.5% for 1-hour glucose value < 120 mg/dl and 9.3 % for 120-179 mg/dl. Interestingly, the glucose level of 180 mg/dL corresponds to the 1-hour glucose threshold of the Carpenter and Coustan criteria for the oGTT. 20 Thus, a majority of the mothers of newborns with high risk for hypoglycemia had untreated IGT.

Our observations of the impact of IGT on the neonatal outcome are in agreement with other studies using either one abnormal oGTT glucose value or increasing glucose values below the diagnostic criteria for GDM to define borderline glucose intolerance. 38-42, 92, 111 In contrast to these studies, we included highly specific parameters like amniotic fluid insulin that had not been investigated in IGT before. The data of others and our group reflect the major deficit of all currently applied criteria. They are all derived from the original O’Sullivan criteria that had been design to investigate the relation of antenatal oGTT values and the risk for maternal diabetes in later life. They did not consider the risk for neonatal morbidity. The multicenter HAPO-study involving 25000 pregnancies has been started in the year 2000 (HAPO= hyperglycemia adverse pregnancy outcome) to finally determine oGTT thresholds for short and long term morbidity of the offspring. 112 The results will not be available before 2004/5. When currently available criteria were applied, based on our data the requirement of at least two pathologic values in the oGTT to initiate therapeutic intervention does not appear to be justified and should be reconsidered. The actual German guidelines for diagnosis and therapy of GDM reflect the impact of IGT and recommend diet education and glucose control similar to GDM pregnancies. 87

As mentioned before, good glycemic control during pregnancy according to the recommended glycemic goals is not able to normalize the morbidity rate in GDM pregnancies. In our own population, we could lower the macrosomia rate of 24% in the early Nineties 73 by the implementation of self-glucose monitoring in all women. However, the current rate is still about 18%. 89, 113Thus, we retrospectively investigated the relation of maternal glucose values during the course of pregnancy and fetal growth pattern in a [page 49↓]population of 400 women with IGT or GDM. Since the fetal abdominal circumference (AC) is known to be a good predictor for LGA at birth 47, 55, 56(own unpublished data 5.1), we have chosen a fetal AC < and ≥ 90th percentile to discriminate between pregnancies with normal and accelerated growth. 48 We compared pre-and postprandial glucose values of the daily profiles at 5 different periods of pregnancy with the AC measurements at corresponding gestational ages. There was no difference in glucose values between pregnancies with AC < and ≥ 90th percentile either at diagnosis or later in pregnancies, with the exception of the fasting glucose values between 32-35 weeks of gestation. In contrast, there was a tight relation to the maternal BMI. The rate of fetal AC ≥ 90th percentile and LGA at birth was significantly higher in women with BMI ≥ 30 kg/m2 compared to lean women (28% vs 14% at entry). 113

In a second step, we evaluated various parameters influencing fetal growth and determined independent predictors for an AC ≥ 90th percentile at diagnosis, during pregnancy and for LGA at birth. A history of LGA and obesity with BMI ≥ 30 kg/m2, either alone or combined, were independent predictors for an AC ≥ 90th percentile at entry, at 24-27 (history of LGA) and 28-31(both) weeks of gestation and for LGA at birth. In contrast, the fasting glucose at 32-35 weeks proved to be the only predictors at 32-35 and 36-40 weeks. 89 The identified predictors and the periods of their major influence on fetal growth reflect the primary impact of genetic and epigenetic factors in previous pregnancies (history of LGA) in the early third trimester (obesity), of recent maternal parameters in the mid third trimester and the influence of the stimulation of growth by elevated maternal glucose in the late third trimester. The majority of the women (> 90%) had good glucose control which might explain the limited influence of maternal glucose values. But this also implicates that tighter glucose control might not be efficient because of the strong contribution of other factors. e.g. maternal obesity. Maternal obesity is known to be an independent risk factor for macrosomia in pregnancies with impaired 114 as well as normal glucose tolerance 115-117 due to peripheral hyperinsulinism 118 and increased levels of serum lipids and amino acids in face of normal maternal glucose values. 50, 51 Macrosomic infants of obese mothers show the same disturbances in lipids profiles as their mothers 119 and the neonatal fat mass determines 43% of the variance in birth weight although it accounts only for 14% of the total body weight. 120, 121

In conclusion, the existing glucose criteria for diagnosis and treatment of GDM do not appear to be reliable to identify pregnancies at increased risk for neonatal morbidity. Thus, a tailored management concept which considers the individual maternal conditions, might [page 50↓]improve the efficacy and outcome of care in GDM. In obese women, prepregnancy counseling about the adverse effects of obesity on pregnancy outcome and a moderate caloric restriction might reduce additional obesity associated morbidity. It was shown that a 33% caloric restriction lowers triglycerides without marked ketonuria. 122 However, the effect of interventions during pregnancy seems less promising than prepregnancy counseling since the majority of obese women demonstrate low weight gain in pregnancy.

The limited predictive ability of maternal glycemia for neonatal outcome is not only caused by the impact of maternal fuels besides glucose but also by the alterations in placental transport and consumption of fuels associated with diabetes, 123-125 and the individual susceptibility of the fetus to oversupply. 78, 126Fetal-based management strategies aimed to avoid these uncertainties by concentrating directly on the target, the fetus. The measurement of fetal insulin was one approach established by Weiss. 65 The level of fetal insulin can be determined by amniotic fluid insulin (AF insulin) secondary to the urinary excretion of insulin. 68, 127 Diabetic fetopathy is causally related to fetal hyperinsulinism. 14, 44, 128 The Weiss approach directed intensive insulin therapy to women with elevated AF insulin levels without respect to maternal glycemia. The disadvantage of his strategy was that amniotic fluid is accessible only by an invasive amniocentesis. Another approach used fetal overgrowth identified by ultrasound as a clinical marker for presumed fetal hyperinsulinism 60-62 , and directed insulin therapy to pregnancies with an accelerated growth of the fetal AC. Determination of fetal growth is an indirect approach to assess hyperinsulinism and there was concern about over- or under-treatment when therapy is predominately based on fetal growth. Thus, we evaluated our data from amniocenteses in women with diabetes derived from times when determination of AF insulin was part of our routine management. We could demonstrate a weak but significant correlation between the fetal AC at time of amniocentesis and the level of AF insulin. Moderately elevated levels (90th percentile = 7 µU/ml) were poorly identified but a level of 16 µU/ml was excluded by an AC < 75th percentile with a negative predictive value of 100%. 129Kainer et al 70, demonstrated similar results in a population consisting exclusively of women with type I diabetes. Interestingly, the identified AC threshold corresponds to the AC percentile that is used in the fetal-growth-based approach to initiate insulin therapy. The level of 16 µU/ml is still below the levels reported by other researchers to be associated with short 66, 71 and long term morbidity. 72 After establishing a correlation between the fetal AC and AF insulin we were additionally interested to determine the power of the AC to predict LGA at birth. Although an AC > 90th percentile at entry was the strongest [page 51↓]independent predictor, the predictive power of the ultrasound was only slightly higher than that of maternal history predictors like BMI, history of GDM or LGA. A second ultrasound slightly added to the predictive power (own unpublished data in 2.3.1) when evaluated in a single fashion. But when all predictors were combined for a score, the second ultrasound did not improve the predictive power. Similar to the predictive power for AF insulin, ultrasound seemed to be more helpful to exclude than to predict LGA at birth.

The utility of the fetal AC to guide metabolic therapy in women with GDM was first investigated in a pilot study by the group of Buchanan and Kjos from Los Angeles. Their study population was limited to women with glucose values that would not have prompted insulin therapy based on a standard guidelines. Insulin was given in pregnancies with a fetal AC ≥ 75th percentile diagnosed by a single ultrasound at entry to care. 60, 61 The macrosomia rate could be reduced by 3 fold compared to pregnancies with AC ≥ 75th percentile but no insulin therapy (13 vs 45%) and reached the level of pregnancies with AC < 75th percentile at entry. This first study addressed the question whether morbidity could be lowered by targeted intervention in pregnancies identified as high risk based on fetal growth. A second study was designed to investigate whether the ultrasound-based approach also allows to avoid intensive intervention, i.e. insulin therapy and intensive glucose monitoring, in pregnancies at low-risk despite maternal hyperglycemia that would have required insulin therapy. 62 Insulin was not given when the fetus did not demonstrate accelerated growth in serial ultrasound examinations during pregnancy. Women with severe hyperglycemia had been excluded. Indeed, insulin therapy could be avoided in 38% of the women in the ultrasound-guided group without adverse neonatal or maternal outcome. Both studies were performed in predominantly Mexican-American women who are genetically determined to have a high rate of insulin resistance and obesity that might have had an important impact on the results. Therefore the wide applicability of the therapy concept was questioned. However, we could confirm the benefit and safety of the fetal-growth-based approach in a Caucasian population of 200 women included in a study in Berlin 135 .We combined the approaches of both pilot studies and included both women who presented euglycemia under diet therapy and those with hyperglycemia. We did not find any adverse outcome either in the mother or in the offspring in pregnancies guided predominately by fetal growth compared to those guided solely by maternal glycemia. When we divided the population according to the selection that was done in the two previous pilot studies, we found the same reduction of LGA rate in women with euglycemia but accelerated fetal growth and save of insulin therapy in women with hyperglycemia but normal fetal growth. In the ultrasound-guided group, maternal [page 52↓]hypoglycemia prompting clinical intervention occurred in no case when insulin therapy was applied based on accelerated growth despite maternal glucose levels below the standard thresholds for the initiation of insulin therapy. Maternal jeopardy by hypoglycemia had been a serious concern when we transferred the fetal-growth-based approach to a population with milder glucose intolerance than normally seen in Mexican-Americans. Insulin therapy was given exclusively based on a fetal AC ≥ 75th percentile. No women developed a level of severe hyperglycemia which was included as additional criteria for insulin therapy in the protocol.

Our studies in pregnancies complicated by GDM demonstrated that in women presenting high fasting glucose values at diagnosis the maternal glucose values are extremely reliable to assess the risk for disorders in embryogenesis resulting in congenital anomalies. In the majority of the women with GDM congenital malformation are not a concern. In the later course of pregnancy, maternal glucose values are of limited use to predict neonatal morbidity and a new management approach that is predominantly based on fetal growth instead of maternal glycemia showed promising results. Considering the knowledge about glucose metabolism in pregnancy and the numerous factors influencing the fetal development, we were not surprised about the low predictive power of maternal glucose values in treated GDM pregnancies. The normalization of maternal glucose values is indisputably beneficial for mother and offspring in pregnancies with a high level of hyperglycemia like in preexisting type 1 and 2 diabetes. However, the majority of women with GDM demonstrate only moderately elevated values. In these women, insulin therapy often results only in a reduction of fasting glucose values by 10 mg/dl or postprandial glucose by 20-30 mg/dl. There is evidence that the thresholds for the initiation of insulin therapy are still higher than the glucose values of the normal obstetrical population. Parretti et al77 reported a 1-hour postprandial glucose value of 114 mg/dl as 97th percentile even at 38 weeks of gestation. If we attempt to reach this level, we will end up with an unacceptable high rate of insulin use. No doubt that maternal glucose supply is an important factor but it is only one parameter in the whole cascade that determines the outcome. A management solely based on maternal glycemia does not consider other maternal fuels or characteristics, placental function and the individual susceptibility of the fetus for disturbances. Twin studies demonstrated impressively the occurrence of fetal hyperinsulinism in one and normal insulin values in the other fetus. 78 These data support our clinical experience. We saw fetuses obviously presenting disproportional growth with an AC > 90th percentile and the mother had glucose values far [page 53↓]below the thresholds for insulin therapy 135. On the other side, we had normally grown fetuses with maternal glycemia that would require insulin therapy when we followed the recommendations. In extreme cases, we caused growth restriction. We had 6 small-for -gestational-age (SGA) newborns in the standard group of the Berlin study when we had to give insulin in fetuses with normal AC (table 3, Box B, 6.1). Additionally we should keep in mind that we based our decision for insulin therapy on self-glucose-monitoring. This implicated considerable problems like the accuracy of reflectance meters, the individual technique of the women and an improved compliance in diet at days selected for glucose profiles.

Following critical data analysis and present knowledge it is obvious that an additional tool beside maternal glucose is needed to identify GDM pregnancies at risk for neonatal morbidity. Why do we not include the target - the fetus himself ? The approach based on the identification of fetal hyperinsulinism via measurement of amniotic fluid insulin presents the most reliable method but can not be widely recommended because of the requirement of an amniocentesis. However, the fetal-growth-based management uses a method to target GDM pregnancies with need for intensive intervention which is part of the routine in prenatal care. However, there are two concerns that should be addressed when we discuss the utility of this approach. First, the unsatisfying accuracy of ultrasound to predict birth weight or the development of LGA, especially in diabetic pregnancies 130 or extremely overweight fetuses. 131 This is caused by the limitation of the technique itself combined with the individual capacities of the ultrasonographers and the high rate of obesity in GDM women. In fetuses with accelerated growth due to maternal diabetes, we are additionally faced with an increase in adipose tissue that is less dense than fat-free tissue, e.g. muscle and bone. Thus the use of growth percentiles derived from an unselected patient population may lead to sonographic overestimation of fetal weight. 121, 132This might explain our finding of a high false positive rate of the AC in predicting LGA babies which is consistent with the reports of others. 55, 56 The specificity of the fetal AC to predict an LGA newborn was about 80% thus 20 of 100 women would have been treated with insulin without having a fetus at risk. On the other side, data from Weiss showed that only 50% of the fetuses presented elevated insulin levels with a maternal mean blood glucose level of 100 mg/dl. 66 This mean glucose level corresponds to the recommended fasting and postprandial glucose levels for initiation of insulin therapy. 19 Thus, using the standard glycemia-based approach 50 of 100 women are treated with insulin without having a fetus at risk.


[page 54↓]

What are the perspectives in the care of women with GDM? The increase of obesity in young women accompanied with a high risk for GDM will face us with rising numbers of women with GDM seeking care. Official surveys in Germany estimate a rate of 19% obese teenagers for 1995 which increased up to 32 % for the year 2000. 133, 134 Thus, we have to optimize the allocation of resources for treatment by targeting high-risk pregnancies for intensive intervention. Strict glucose control in all women without additional risk assessment might cause avoidable financial and emotional costs and will eventually consume valuable resources. The fetal-growth-based concept of care includes antenatal risk assessment and therefore might not only improve the efficacy of care but also help to improve cost-efficient treatment. In our study, almost 60% of the women in the ultrasound group identified to be at low-risk performed glucose profiles without clinical consequences. We neither discussed the values, nor were the women encouraged to observe their diet. They were not aware of the thresholds for insulin therapy. We can only speculate about the effect of self-glucose-monitoring in these women since they could not be blinded to the numbers showing up on the reflectance meters. In future, we might be able to adjust the frequency and intensity of glucose control based on the risk assessment by fetal growth, resulting in a considerable reduction of monitoring in low-risk women. Intensive glucose monitoring is the most cost intensive parameter in the diabetic management of GDM pregnancies and adds an unpleasant diagnostic technique for the mothers. In contrast, ultrasound is well accepted by the women and part of the routine prenatal care. Based on recent costs and diagnosis of GDM estimated at 28 weeks of gestation, the calculated expenses were 160 Euro for glucose monitoring consisting of 2 profiles per week and 220 Euro if 3 profiles would be performed. Our data indicate that 90% of the cases with AC > 75th percentile are diagnosed by a 1st or a 2nd ultrasound examination (unpublished data 6.1) and that a 3rd examination will add only little predictive power (unpublished data 5.1). When we assume 2 ultrasound examinations and e.g. 2 profiles per month, we will end up with a total amount of 37 Euro. The first ultrasound does not add additional cost since one exam of fetal growth is part of the German prenatal care protocol. Thus, we estimate that our approach will cut down the costs for diabetic care in GDM by 4 or 6 fold, respectively.

Our studies indicate that GDM therapy assignment based on fetal ultrasound assessment in addition to limited baseline maternal glycemic values is a safe, more pleasant and likely less cost intensive approach. Still, several topics need further investigations. We will have to determine to what extend glucose monitoring could be reduced in low-risk [page 55↓]pregnancies with moderately GDM without increasing risk of adverse neonatal and maternal outcome. Second, whether this strategy is not only safe but also represents a significant improvement of the outcome remains to be proven in larger cohorts. Last, the establishment of more specific, and easily reproducible sonographic measures of evolving diabetic fetopathy in utero as part of the routine ultrasound examination would further enhance the clinical value of involving fetal ultrasound.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 3.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
15.06.2004