[page 10↓]

2.  Own contributions

2.1. The influence of maternal glycemia on embryogenesis

2.1.1. Introduction and summary

Preexisting diabetes is a well known risk factor for congenital anomalies since maternal hyperglycemia during time of embryogenesis has a teratogenic effect of the development of the embryo. 23-28 Data from animal and clinical studies have demonstrated a correlation of the degree of maternal hyperglycemia during early pregnancy and the occurrence of malformations in the embryo. 29-31 Preconceptional care and optimizing of maternal glucose control can reduce the rate of anomalies to the level of the normal population. 27, 32, 33 In women with preexisting diabetes, a great body of data is available to assess the risk for diabetes based on the level of maternal glucose values. Congenital anomalies typical for diabetes affect primarily the heart, central nervous system, kidneys and the axial skeleton.

Figure 3: Newborn with caudal regression syndrome - the most specific but rare congenital anomaly in pregnancies complicated by diabetes (from Smith's Recognizable Patterns of Human Malformation )

Disturbance of fetal development causing these disorders must occur during the first 8 weeks of pregnancy. Thus, GDM is not considered as risk factor for congenital malformations because it typically develops not before the late second trimester coincident with the decreasing insulin sensitivity at this time. But the definition of GDM as any glucose intolerance diagnosed first in pregnancy, comprises a wide range of metabolic [page 11↓]decompensation - from mild intolerance to overt hyperglycemia. It can be speculated that women with severe hyperglycemia at time of diagnosis of GDM might have had hyperglycemia in early pregnancy high enough to impart a risk for malformation to their children. So far, there had been very few data to quantify the risk of malformations in these heterogeneous population of women with GDM.

Congenital malformations in offspring of women with GDM

Our study aimed to determine the incidence of congenital anomalies in women with hyperglycemia diagnosed first in pregnancy and to identify clinical predictors for an increased risk for anomalies. A total of 3743 infants of mothers with GDM who attended the Diabetic Clinic of the Los Angeles County-University of Southern California Women’s and Children’s Hospital between 1987 and 1995 could be analyzed. Infants with genetic syndromes and aneuploidy were excluded. The incidence of was 2.9% for major congenital and 2.4% for minor anomalies. There was no difference in maternal historical or glycemic parameters between mothers of pregnancies with normal infants and infants with minor anomalies thus we combined them for the further analysis. The multivariate analysis revealed that the fasting glucose concentration at time of diagnosis was the only independent predictor (Odds ratio 1.13 for each 10 mg/dl increase, 95% CI 1.08-1.16) for major malformations. The fasting glucose is an easy accessible clinical parameter since it is part of the diagnostic procedure. Thus we examined if there is a threshold glucose value for an increased risk. The population was divided into subgroups according their fasting glucose values using strata of 20 or 40 mg/dl respectively. There was an abrupt increase in the risk for major congenital anomalies at a serum glucose concentration > 120 mg/dl and again at >260 mg/dl (table 1).

Table 1: Rate of and odds ratio for the risk of major congenital malformations in a population of women with hyperglycemia diagnosed first in pregnancy

 

Total population

Subgroup with normal

oGTT 1- 4 month postpartum

Fasting glucose (mg/dl)

Rate

OR (95%CI)

Rate

OR (95%CI)

< 120

62 (2%)

1.0

22 (1.7%)

1.0

121-260

39 (5%)

2.6* (1.7 - 3.8)

8 (3.7%)

2.1* (1.1-4.82)

> 260

7 (30%)

20.5* (8.5-50.7)

0

 

* p< 0.05


[page 12↓]

The majority of our study population belonged to the ethnic group of Mexican-American who are characterized by a high degree of glucose intolerance due to insulin resistance. To exclude that our results were valid only for women with presumably undiagnosed preexisting type 2 diabetes, we repeat the analysis in a subgroup of women with normal oGTT (n= 1600) 1- 4 month postpartum. We identified the same cut-off at 120 mg/dl but there was no case of fasting glucose > 260 mg/dl in this subgroup.

Patterns of congenital anomalies and their relationship to initial maternal hyperglycemia

In pregnancies with type 1 diabetes there is a predominance of organ systems which are frequently affected by congenital anomalies. In this second study we aimed to investigate 1.) the types of malformations in infants of mothers with GDM or type 2 diabetes and 2.) whether the types of anomalies occurring are related to the level of maternal hyperglycemia at entry to care. We hypothesize that some organ systems are more susceptible to hyperglycemia than others . Study subjects were again retrieved from our database of diabetic women attending the Diabetes Clinic of the Los Angeles County women’s hospital. Diagnosed major congenital malformations were categorized by the number and type of affected organ systems.

In a total of 3764 women with GDM and 416 women with known type 2 diabetes mellitus. Maternal historical (age, prepregnancy BMI, prior pregnancy with macrosomia, stillbirth or anomalies) and clinical parameter (gestational age at first prenatal visit, first trimester exposure to sulfonylurea agents) and value of the initial fasting glucose and HbA1c were investigated regarding their relation to anomalies. 143 infants (3.4%) with major anomalies were identified, with a prevalence of 2.9% in GDM and 8.9% in type 2 diabetes. The most frequently affected organ systems were cardiac (37.6%), musculo-skeletal (16%) and central nervous system (9.8%).

In 16% of the infants multiple organ systems were affected. There was no predominance seen of any organ system affected with increasing fasting glucose values (figure 4). But major anomalies involving multiple organ systems were associated with significantly higher glucose levels (166 ± 64 mg/dl) than malformations which were limited to one affected organ system (141 ± 55 mg/dl, p=0.006.)

2.1.2. Discussion

In this first large-scale study in women with GDM, we could confirm the tight relationship between maternal glycemia and the rate of congenital anomalies that had been


[page 13↓]

Figure 4: Distribution of affected organ systems in infants with major anomalies according to the fasting glucose value at entry to care

demonstrated in women with preexisting type 1 diabetes. Additionally, we saw the same predominance of anomalies affecting the heart, skeleton and central nervous system in a mixed population of women with GDM and known type 2 diabetes as reported from pregnancies with type 1 diabetes.

The overall rate of major anomalies was slightly higher than in non-diabetic women of our population. Our women with GDM had an unusual wide range of the degree of glucose intolerance partly due to the high level of insulin resistance in the Mexican-American population in Los Angeles. 34 There is a great chance that women with fasting glucose values above our second identified threshold for an increased risk for anomalies (260 mg/dl) had undiagnosed type 2 diabetes before pregnancy. But already glucose levels below the level required for the diagnosis of diabetes outside pregnancy at the time of study (fasting glucose 140 mg/dl) were related with in increased rate for anomalies. 35 A second analysis in a subgroup of women with normal postpartum oGTT and therefor little chance of having preexisting diabetes confirmed that women who develop severe glucose intolerance first in pregnancy are also at risk for an infant with congenital anomalies. We have no information about the glucose values in early pregnancy since in general screening for GDM is [page 14↓]recommended not before 24-28 weeks of gestation. Thus, we could only speculated about the degree of hyperglycemia during embryogenesis which is required to cause anomalies. Interestingly our identified threshold of a fasting glucose value >120 mg/dl was identical with the threshold which was reported for 1st trimester glucose measurement in women with type 1 diabetes. 31 Regardless the final classification of diabetes after pregnancy, our data provide a useful tool to counsel women with hyperglycemia diagnosed first in pregnancy about their risk for major anomalies based on their fasting glucose levels at time of diagnosis. Ultrasound examination on a high level of expertise should de offered to women with fasting glucose > 120 mg/dl with special attention to the most frequently affected organ systems. Additionally, the high chance for anomalies involving multiple organ systems have to be considered since increasing glucose levels had been associated with a higher number of affected organ systems in the infants of our population. Further prospective studies are needed to develop strategies to identify preconceptionally women without overt diabetes but glucose intolerance sufficient to cause congenital anomalies. A minority of the women would have qualified for routine diabetes testing which is limited to women with age > 45 or other risk factors like prior GDM. 36 But even with routine testing the women who appeared to be at risk would not have been detected considering the existing diagnostic criteria for diabetes outside pregnancy at the time of the study. The rate of major anomalies was more than double in women with fasting glucose > 120 mg/dl, a level which was classified as normal by the recommendations of the American Diabetes Association from 1996. 36 Meanwhile the diagnostic criteria for diabetes outside pregnancy had been modified and a repeat fasting glucose measurement of 125 mg/dl in venous plasma and 110 mg/dl in capillary blood qualifies for the diagnosis of diabetes. 37


[page 15↓]

2.2.  The influence of maternal glycemic values on fetal growth and neonatal morbidity

2.2.1. Studies in pregnancies with borderline glucose intolerance

2.2.1.1. Introduction and summary

The diagnosis of GDM is based on the glucose values obtained by an oral glucose tolerance test. There is an ungoing discussion for three decade about the thresholds for defining maternal glucose intolerance in pregnancy which resulted in a great variation of glucose values used for the definition of GDM (table 2).

Table 2: Different diagnostic criteria used for diagnosis of GDM

 

1964

O'Sullivan

& Mahan

1979

NDDGa

1982

Carpenter

& Coustan

WHO

1998

ADAb

Germany

Until

2001

2001

DDGc

Load (g)

100

100

100

75

100

75

75

Method

SM-Nd

GO/HKe

GO/HK

GO/HK

GO/HK

GO/HK

GO/HK

Medium

VBf

VPg

VP

VP

VP

CBh

CB

VP

Fasting

(mg/dl)

90

105

95

126

95

90

90

95

1H

165

190

180

 

180

165

180

180

2H

145

165

155

140

155

145

155

155

aNational Diabetes Data Group
bAmerican Diabetes Association
cGerman Diabetes Association
dSomogy-Nelson
eHexokinase/Glukoseoxigenase,
fWhole venous blood
gVenous plasma
hCapillary blood

All used definitions were derived from the original criteria from O’ Sullivan and Mahan from 1964. 21 But these were based on the subsequent maternal risk for diabetes and did not investigate the risk for fetal or neonatal morbidity. Furthermore there is more and more evidence that the relationship between maternal glycemia during pregnancy and neonatal morbidity behaves more like a continuum, with no precise threshold to discriminate between high and low risk fetus. 38 39 For termination of thresholds for increased morbidity an untreated population of women with glucose intolerance in pregnancy would be needed. Since it is unethical to withhold a therapy that had been shown to reduce morbidity, the studies addressing this issue were limited to women with glucose intolerance below the existing [page 16↓]thresholds for gestational diabetes. They investigated the outcome either in women with positive glucose challange test but negative oGTT or in women with only one pathologic value in the oGTT.

Fetal hyperinsulinism, neonatal obesity and placenta immaturity

The aim of the present study was to determine the impact of borderline glucose intolerance on diabetic fetopathy indicated by neonatal obesity, fetal hyperinsulinism and placenta immaturity. Our study was performed between 1992 and 1993 at the Department for Obstetrics at the Vivantes Medical Center Neukoelln in Berlin. We involved 325 women with risk factors for GDM who were tested for glucose intolerance by a 75 g oGTT. The O’Sullivan criteria were applied for diagnosis of GDM. Diabetes care consisting of diet education and frequent glucose profiles was limited to women with GDM, defined as usual by two pathologic values in the oGTT. The study population was divided into women with normal oGTT, women with one abnormal value (IGT= impaired glucose tolerance) and women with GDM and neonatal outcome was compared between the groups. Neonatal parameters tested were as followed: birth weight, large-for.gestational-age birth weight (LGA), skinfolds at three sites of the newborn (figure 5), amniotic fluid insulin at time of delivery, cord blood glucose and insulin, neonatal glucose and villous maturation of the placenta. Neonatal obesity was defined according percentile rankings obtained by skinfold measurements that had been previously performed in 250 consecutively born infants.

Figure 5: Skinfold measurement at the triceps in a newborn


[page 17↓]

Women with one abnormal value had significantly higher rates of LGA infants and infants with central obesity, of hyperinsulinism and neonatal hyperglycemia compared to women with normal oGTT with a rate similar to women with GDM. Central obesity and hyperinsulinism with consecutive neonatal hypoglycemia was even more frequent than in pregnancies with GDM. Severe placental immaturity was seen most frequently in GDM pregnancies but again the rate in IGT was significantly higher compared to normal pregnancies.

Neonatal Hypoglycemia in LGA newborns

Neonatal glucose testing is routine part of neonatal care in infants of mothers with known diabetes. Additionally, neonatal glucose testing is recommended in all LGA infants (birth weight> 90th percentile) independently of the diabetic status of the mother. Excessive growth is the major clinical sign of fetal hyperinsulinism due to maternal hyperglycemia in pregnancy. Macrosomic newborns are at increased risk for neonatal hypoglycemia when after delivery the insulin secretion has to be adapted to the sudden drop in glucose supply. Universal testing in all LGA newborns implicates unnecessary diagnostic in infants at low or no risk for hypoglycemia since only a minority of macrosomia is caused by diabetes. Therefor we investigated the rate of hypoglycemia in LGA newborns of non-diabetic mothers and whether maternal or neonatal risk factors for hypoglycemia could be identified. In 887 LGA infants, we observed hypoglycemia within the first day of life in 16% of the infants with a steep decrease of the incidence after the first two hours.

There was no clinical useful predictor for hypoglycemia unless glucose values of an oGTT in pregnancy were available. In the subgroup of infants of mothers with oGTT the 1 hour-glucose value was an excellent discriminator between infants at low, intermediate or high risk for hypoglycemia. Three cutoff points with stepwise increase in the rate of hypoglycemia were identified. The rate of hypoglycemia was 2.5% for glucose values ≤ 120 mg/dl, increased to 9.3% for values of ≥ 120 and ≤ 179 mg/dl and further to 22% for ≥ 180 and ≥ 239 mg/dl.

2.2.1.2. Discussion

Several groups have been shown that women with glucose intolerance below the existing thresholds for gestational diabetes have a higher rate of macrosomia, cesarean delivery and preeclampsia. 38-42 The best evidence comes from the Toronto Tri Hospital study, a prospective study that involved 3600 women with normal oGTT, patients and care [page 18↓]givers were blinded to the glucose values. They could demonstrate a graded increase in adverse maternal-fetal outcome with increasing maternal carbohydrate intolerance. 38 In addition to other studies which were limited to clinical complications known to be increased in diabetes, we could confirmed the influence of borderline glucose intolerance on very specific parameters for diabetic fetopathy. Beside macrosomia and hypoglycemia also hyperinsulinism, trunk obesity and placenta immaturity were significantly more frequent in untreated women with IGT than in normal women. Our second work related to this topic concentrated on macrosomic infants of non-diabetic mothers. We could show that the risk of neonatal hypoglycemia in these infants is tightly related to the 1-hour oGTT value of the mother. Interestingly, the identified threshold corresponds to the threshold for an abnormal 1-hour value according to the Carpenter and Coustan criteria for GDM. 20 Thus, all mothers of the infants at greatest risk for hypoglycemia had IGT that was not treated because the oGTT did not fulfill the criteria for GDM. Secondly, our data support the clinical importance of a general screening for GDM since without available oGTT values a risk assessment for hypoglycemia in LGA newborns seems not be possible.


[page 19↓]

2.2.2.  The impact of maternal obesity

2.2.2.1. Introduction and summary

The hypothesis of Pedersen proposed that glucose from the maternal circulation is a major regulator for fetal growth. 14 A large body of clinical and experimental studies supported that maternal hyperglycemia enhances fetal growth by an excessive glucose supply to the fetus at a time when the fetal pancreas is able to respond by increasing its production of insulin. 43-45 Although the stimulation of insulin secretion starts with 11-15 weeks of gestation 46 , accelerated growth due to maternal diabetes occurs at around 28 weeks, presumable because of the fetal capacity to store triglycerides at that time. 47 Thus, it is obvious that maternal hypergylcemia is a risk factor for macrosomia, however the regulation of fetal growth is far more complex and is influenced by many factors. The clinical experience indicates that despite of tight glucose control neonatal macrosomia occurs. On the other side, normosomic infants are born to mothers with hyperglycemia.

The correlation ofmaternal obesity and high rates of fetalmacrosomia

Existing studies are limited to the investigation of the influence of maternal glycemia and LGA at time of birth. Our study aimed to examine the correlation of maternal glucose values and fetal growth at different gestational weeks of pregnancies in normal and overweight women with GDM. In 406 women with GDM or IGT a total of 919 serial ultrasound examinations was performed. A fetal abdominal circumference > 90th percentile according to gestational age 48 was defined as fetal macrosomia. Glucose values at diagnosis - oGTT , entry glucose profile and HbA1c – and the glucose values of the profiles performed at 5 different categories of gestational weeks were compared between pregnancies with and without fetal macrosomia diagnosed at correspondent gestational ages. Each analysis was adjusted for maternal obesity, defined as body mass index (BMI) ≥ 30 kg/m2. 49 There was no difference in glucose values either at entry or during pregnancy between pregnancies with or without fetal macrosomia either in lean nor in obese women. In contrast, the fetal macrosomia rate was significantly higher in obese compared to lean women at each category of gestational age and at birth.


[page 20↓]

Determinants for in utero macrosomia at different gestational ages

Our first work revealed the strong influence of maternal obesity on fetal growth. In a second step, we investigated the influence of other maternal parameters. We use the above described population to determine independent predictors for fetal macrosomia at different periods of pregnancies and at birth. We included maternal historical (prior pregnancy with LGA or GDM, prepregnancy BMI and parity) and glycemic parameters at entry (oGTT, HbA1c and mean fasting and postprandial glucose values of the daily profile) and the glucose values of the profiles at the different periods of pregnancy. We found different parameters univariately associated with accelerated growth at different times of pregnancy: LGA in a previous pregnancy, parity, prepregnancy obesity, fasting of the oGTT or fasting glucose at 32GA. The independent predictors are displayed in figure 6.

Figure 6: Timeline of independent predictors for an AC≥90th percentile at entry, in different gestational age categories and for large-for-gestational age birth weight in pregnancies with GDM and IGT. (Odds Ratio and 95%CI given). Obesity was defined as BMI ≥ 30 kg/m2. OR per 5 mg/dl increase of fasting glucose


[page 21↓]

2.2.2.2.  Discussion

The complexity of fetal growth occurs at several levels determined by the mother, the placenta and the fetus. Both of our presented studies revealed the strong influence of maternal obesity on the risk for accelerated growth. At no time in pregnancy, a higher rate of fetal macrosomia was associated with higher maternal glucose values but with obesity. Obesity is often associated with elevated lipids and proteins and peripheral hyperinsulinism which had been shown to be related to the risk for macrosomia. 50-52 Lipids and amino acid levels are influenced by the carbohydrate metabolism but there is no linear correlation between the elevation of glucose and non-glucose nutrients. Thus, the effect of hyperlipidemia and hyperacidemia on fetal growth cannot be eliminated solely by glucose control. When we looked for other maternal predictors in the second analysis, we found independent predictors that represent the three major determinants of fetal growth. A history of a prior LGA infant representing the genetic influence, maternal obesity reflecting genetic and non-glucose fuels and the fasting hyperglycemia indicating an increased glucose supply to the fetus. In the early pregnancy the influence of genetic factors predominates; about 15% of the variation in birth weight is due to genetic predisposition. 53 It could be shown that an early symmetric accelerated growth is not associated with fetal hyperinsulinism. 54 In the early third trimester maternal obesity became a strong predictor coincident with the time of fetal adipocyte proliferation and lipid storage. Maternal glycemia appears to have the strongest influence in the late trimester, the time when accelerated growth in diabetic pregnancies was described 55, 56. Maternal hyperglycemia leads via fetal hyperinsulinism to an increase of the insulin sensitive tissue, like the adipose tissue.

In summary, our GDM management that was focused on tight glucose control could not lower the macrosomia rate in obese women. Considering the strong influence of non-glucose related parameters, a modified approach in obese women might be more effective to lower the high rate of LGA infants in these women.


[page 22↓]

2.3.  The importance of the fetal abdominal circumference in pregnancies with diabetes

2.3.1. Introduction and summary

As demonstrated above, the reliability of maternal glycemic values to predict diabetic morbidity in the newborn is limited. Normalization of maternal hyperglycemia could lower the rate of adverse outcome in pregnancies with GDM but the rate of macrosomia and neonatal morbidity is still elevated compared to the normal obstetrical population. 1 In studies with very strict control the macrosomia rate had been lowered to 10% but this management required intensive insulin therapy in 66-100% of the women. 57-59 Furthermore, in gestational diabetes aggressive lowering of the maternal glucose levels may lead to an increased rate of intrauterine growth retardation and an adverse perinatal outcome for small-for-gestational-age newborns. Attainment of strict control in all women with GDM might result in unnecessary treatment in low-risk pregnancies and absorption of limited resources needed for intensive therapy in high-risk pregnancies. Therefor some researchers were looking for other predictors besides maternal glycemia to identify pregnancies at high risk for morbidity. One approach is based on fetal growth 60-62 and limits intensive insulin therapy to pregnancies with accelerated growth of the fetal abdominal circumference (AC). Diabetes associated macrosomia is characterized by an asymmetric growth of the fetal abdomen versus head and long bones due to the stimulation of the insulin sensitive fat tissue by fetal hyperinsulinism (fig.1 ). 63, 64 In diabetic pregnancies, the fetal AC ( fig. 7) measured in the early third trimester revealed to be a good predictor for a LGA newborn. 47, 55, 56

Figure 7: Measurement of the fetal abdominal circumference


[page 23↓]

The second approach uses amniotic fluid insulin to diagnose fetal hyperinsulinism. 65-68 The level of fetal insulin is supposed to correspond to the level of insulin in the amniotic fluid secondary to the urinary excretion. When insulin levels are elevated, insulin therapy is either initiated or intensified. Although this approach offers a direct estimation of the fetal reaction on maternal glycemia, it is not widely accepted because it requires an amniocentesis as an invasive procedure to obtain amniotic fluid.

Fetal abdominal circumference as predictor for neonatal macrosomia

Existing data demonstrate the tight relation of the fetal AC in the third trimester with the LGA status at birth but there is a paucity of data investigating the predictive power of the fetal AC compared to maternal parameters known to influence fetal growth. Therefor we determined independent predictors for LGA and their predictive power. Secondly, we aimed to create a score of the discriminatory parameters and quantitated the predictive power by receiver operator characteristics (ROC) curves analysis. In 728 women treated for GDM four independent predictors could be identified: a history of GDM, prior delivery of an infant > 4000 g birth weight, prepregnancy BMI ≥ 30 kg/m2 and fetal AC ≥ 90th percentile at entry with the fetal AC being the strongest predictor (OR 3.9) (unpublished data 5.3). None of the glycemic parameters revealed to be predictive. The area under ROC curve of a score based on the three historical risk factors was 0.66, which could be increased to 0.71 by inclusion of fetal AC at entry to diabetes therapy (p>0.05). The negative predictive value (NPV) for women with no risk factor was 0.90 and improved to 0.93 when the fetal AC was considered as well. Subsequent ultrasound examinations did not improve predictive power of the score.

Correlation ofamniotic fluid insulin levels and fetal abdominal circumference at time of amniocentesis

The fetal AC measurement is an indirect approach to assess fetal morbidity in pregnancies with diabetes based on a clinical manifestation of fetal hyperinsulinism. But since fetal growth is influenced by many other factors beside the fetal insulin levels there is still a concern of over- or under-treatment when insulin therapy is administered solely depending on the fetal AC. Thus, we investigated the correlation between amniotic fluid insulin (AF insulin) and fetal AC percentiles at time of amniocentesis performed in the third trimester in 121 diabetic women. In a second step, we aimed to find a threshold for fetal AC measurements that identifies low vs high risk levels of AF insulin without performing an amniocentesis. We could show that AF insulin levels were significantly correlated with the [page 24↓]AC percentiles (r=0.3, p=0.0005) by linear regression. Division of the cohort according to AC percentiles revealed a significant stepwise increase in AF insulin ≥ 7 µU/ml at the 80th percentile of the AC. An amniotic fluid insulin >7 µU/ml was previously defined as 90th percentile of a normal obstetrical population by our group. 69 But the negative predictive value (NPV) was low (77.6 %) and the ROC curve confirmed that there was no good threshold of the fetal AC to identify an AF insulin ≥7 µU/ml (fig 8). In contrast an AC threshold ≥ 75th percentile could reliably identify fetal hyperinsulinism with an AF insulin >16 µU/ml. All 10 cases of AF insulin >16 µU/ml were identified with a NPV of 100%

(74/74).

Figure 8: ROC curves for the fetal abdominal circumference to identify amniotic fluid insulin ≥ 7 µU/ml and ≥ 16 ≥ 7 µU/ml

2.3.2. Discussion

Both our studies support the importance of the fetal AC in the management of pregnancies with diabetes. The percentiles of the fetal AC corresponded to the level of the fetal insulin indirectely determined by the AF insulin. The fetal AC ≥ 90th percentile was the strongest predictor for an LGA infant within a wide selection of tested parameters. In contrast, as we expected from our previous studies, the maternal glycemic values in this treated population were not predictive. We can only speculate if maternal glycemia would be more discriminative in an untreated population with a wider range of glycemic values. [page 25↓]Surprisingly, the predictive power of a score created from historical data was only slightly improved by inclusion of the fetal AC. Easy obtainable historical data by itself seem to provide enough information for clinicians to antenatally estimate the risk for an LGA newborn. All predictors in a single or combined fashion are superior in identifying an infant at low risk for excessive growth (NPV) while the sensitivity and specificity did not exceed 77% or 53%, respectively. Similar, the identified threshold fetal AC for an increased risk for severe fetal hyperinsulinism was highly reliable in excluding hyperinsulinism but weak in predicting elevated insulin levels > 16 µU/ml. For moderately elevated insulin levels the fetal AC offers no reliable tool for risk assessment. Almost 50% of the cases would have been missed by the identified AC threshold for AF insulin > 7 µU/ml. Kainer et al, the only group so far that investigated the relation of amniotic fluid insulin and the fetal AC also found the AC measurement to be useful only in identifying high levels of insulin. 70 Our finding corresponds to the data of Weiss et al who had demonstrated that neonatal morbidity was mostly limited to AF insulin levels which were increased 2 - 3 fold above normal. 66There is evidence that excessive birth weight is limited to markedly AF insulin levels about ≥ 20 µU/ml 66, 71. Similarly, long term effects of fetal hyperinsulinism like an increased rate of childhood obesity also appear to be restricted to the levels of AF insulin ≥ 20 µU/m. 72 These insulin levels correspond closely to the insulin level of ≥ 16 µU/ml that according to our data can be identified by an AC ≥ 75th percentile. Interestingly, the AC threshold of the 75th percentile found by our study to identify severe hyperinsulinism was identical to the AC threshold which has been recommended for initiating insulin therapy in GDM. 60-62


[page 26↓]

2.4.  Intervention studies – management of GDM based on fetal growth

2.4.1. Introduction and summary

Fetal growth in pregnancies complicated by diabetes is related to maternal glycemia but it is controversial to what extend hyperglycemia determines morbidity. 1 In agreement with other groups, we could show that the relation of neonatal morbidity and maternal glucose values seems to behave in a continuous fashion. 38, 42, 73-76Thus, the glucose targets that we aim to achieve during pregnancies are arbitrary and consensus based. Recently it had been shown that the 97th percentile of the 1-hour postprandial glucose value of women with normal glucose tolerance is far below the recommended cutoff for insulin therapy of 140 mg/dl. 77 The best evidence that the same maternal glucose values may result in different outcome comes from observations in twins. 78 Applying the strategy of tight glucose control on all women misses the change to target intervention on pregnancies with high risk for morbidity. In a pilot study limited to women with normoglycemia it was demonstrated that a single measurement at entry to therapy could identify a fetus at risk for macrosomia. Intensive insulin therapy could lower the macrosomia rate by 3 fold in this high risk population compared to those who were treated with diet only. The overall macrosomia rate of the study population was reduced without applying insulin to the majority of the women. Our subsequent studies which will be presented in the following aimed (1) to extend this approach to women with hyperglycemia and (2) to proof the applicability of this strategy in a population with a different ethnic background and without prior stratification according to the maternal glycemia status.

Fetal growth based approach applied in Latino women limited to women with maternal hyperglycemia

Eighty-nine women with GDM and venous fasting glucose levels > 105 < 200 mg/dl after a 1 week trial of diet were randomized to a standard and an experimental group. The standard group was treated with insulin due to maternal hyperglycemia. In the experimental group insulin therapy was limited to pregnancies with a fetal AC > 70th percentile at entry or in one of the subsequent monthly ultrasound examinations. Additionally, insulin was applied when maternal glucose exceeded fasting > 120 mg or postprandial 200 mg/dl. There was no difference in maternal characteristics at entry between the two study groups. According to the protocol, the glucose values during pregnancy were lower in the standard group compared to the experimental group. In the experimental group (n=48), insulin therapy was applied in 27 [page 27↓](56%) women because of fetal AC > 70th percentile, in 3 women because of fasting glucose > 120 mg/dl or non-compliance and in 18 (=38%) insulin could be withhold. The neonatal outcome did not differ between the groups with overall low rates of LGA newborns (6.3 vs 8.3 % for standard versus experimental group). Delivery by Cesarean section was performed more frequently in the experimental group but this could not be explained by complications related to diabetes. Despite intensive insulin therapy the LGA rate in women with fetal AC > 70th percentile at entry was higher than in women with normal fetal growth.

Evaluation of the fetal growth based approach in a Caucasian population without respect to maternal glycemia status

In this study women with diagnosis of GDM according O’Sullivan criteria were enrolled who attended the Diabetic Prenatal Care Clinic either of the Charité or of the Vivantes Medical Center in Berlin. Both institutions take care of a multiethnic population with a rate of approximately 40% women from Turkey, Arabian countries or East Europe. Women from Latin-America are rare. The women were randomized to a standard (n=100) and an ultrasound group (US-group, n=99) when fasting glucose < 120 mg/dl and postprandial values < 200 mg/dl in the glycemic profiles after one week of diet. In the standard group, women stayed on diet unless fasting glucose > 90 mg/dl and/or postprandial glucose >120 mg/dl. The US-group was started on insulin if fetal AC exceeded the 75th percentile at entry or at any examination thereafter corresponding to a 4 week examination schedule at 20, 24, 28, 32 weeks of gestation. Additional, insulin was recommended in case of severe maternal hyperglycemia defined as fasting glucose > 120 mg/dl or postprandial > 200 mg/dl. The two groups were similar regarding historical data, glycemic data and the rate of fetal AC at entry. In the US-group insulin was given exclusively based on AC > 75th percentile. Neonatal outcome was not significantly different in both groups. When we analyzed a subgroup of women with GDM according to Carpenter and Coustan criteria (n=161) the results were identical with the exception of a higher rate of insulin use in the standard group. In a secondary analysis in women with euglycemia and AC >75th percentile (n=34) the rate of LGA, C-section and neonatal hypoglycemia was lower in the insulin treated US-group compared to corresponding women in the standard-group. In those pregnancies with maternal hyperglycemia but AC< 75th percentile (n=35) there was no adverse outcome in the US-group although insulin was withheld.


[page 28↓]

2.4.2.  Discussion

Both our studies demonstrated that a management based on relaxed glycemic criteria combined with fetal AC measurements is a safe approach for mother and child independent on the ethnic background of the study population. The measurement of the fetal AC with ultrasound reliably identified fetuses at low risk for accelerated growth. This supports the prior work of Bochner, Landon and Ogata. 47, 55, 56 In GDM managed by a fetal-growth based approach the neonatal outcome was similar to pregnancies guided solely by maternal glycemia even in selected women with hyperglycemia from a population of Mexican-Americans that is known to have a high rate of severe glucose intolerance. 34 Insulin therapy could be avoided in 38 % of these women and in 43% of Caucasian women with hyperglycemia investigated in Berlin. The “Berlin study” combined the two pilot studies from Los Angeles and included both women with normoglycemia and hyperglycemia. The overall rate of insulin use was slightly higher in the US-group compared to the standard approach. This reflectes the mild degree of glucose intolerance when diagnosis of GDM is based on the low diagnostic criteria for GDM of O’Sullivan. 21 When we excluded women who did not fulfill the Carpenter and Coustan criteria that require higher post challenge glucose values, the insulin use in both study arms was similar. When we looked at the women who were treated differently in the US-group compared to the standard group we realized a better outcome in the US-group: a tendency toward a lower SGA rate in women with hyperglycemia but normal fetal growth and a lower LGA rate in women with euglycemia but accelerated growth. The last finding confirmed the study of Buchanan. 60 Interestingly, in both studies the LGA rate was unexpectedly low in the standard group. It can be only speculated about the influence of increased attention and motivation under the conditions of a clinical trial and the frequent demonstration of fetal growth by serial ultrasound examinations.


© 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