We truly understand the concerns raised by our study (1) investigating a mainly fetal growth–based management strategy of gestational diabetes mellitus (GDM), since our data questioned the maternal glycemia-based management that we have used for decades. We are grateful for the opportunity of discussion.
Kitzmiller (2) is right that two profiles per week may not be considered as intensified management and that we might have missed occasional events of hyperglycemia. The large-for-gestational-age (LGA) rate in the standard group might have been slightly lower with more insulin therapy. But then we would have ended up with a rate of LGA newborns lower than the normal 10%. The tight blood glucose treatment in the ultrasound group is part of the concept of ultrasound-based management that concentrates intensive intervention on fetuses at risk (3,4).
In contrast to most areas of medicine, a major dilemma in obstetrics is that there is no direct approach to the fetus. We have to rely on methods of surveillance that give us only indirect information about the well-being of the fetus. This implicates, per se, a lack of reliability. We all know about the high false-positive rate of fetal heart tracing. We are faced with the same problem in the management of GDM. From the Pedersen hypothesis, we know that severe maternal hyperglycemia is harmful for the fetus. Fasting glucose values >95 mg/dl and 2-h postprandial glucose >120 mg/dl are considered as an indication for insulin therapy. But where is the evidence for these values to be accepted as normal? When we look at the glucose values from pregnant women with normal glucose tolerance from Parretti’s data (5), our thresholds seem to be too high. Where is the evidence that the degree of placental glucose to the fetus is equal in each woman? What about the studies of twins demonstrating that each fetus reacts individually to an increased glucose supply?
Other than these uncertainties in assessing the consequences of maternal hyperglycemia on the fetus, we are faced with the problems of home glucose monitoring. A variability of ±10% from the reference value is accepted, per se. Additionally, we have to consider the skills and compliance of our patients because the predominate source of variability is the user. In our population in Berlin, and I know that it is also true for California and many other areas of the world, many women have a low degree of education, do not speak our language, or do not perform glucose profiles competently.
No doubt, we also have to be concerned about the reliability of fetal ultrasound. But I do not agree with Kitzmiller that the reliability is dependent on the medical degree of the ultrasonographer rather than on experience. I was impressed with the skills of the well-trained technicians at the University of Southern California at Los Angeles.
It is true that we (6) and Kainer et al. (7) reported that the fetal abdominal circumference (AC) measurement has a low sensitivity to detect a moderate fetal hyperinsulinism (cutoff >7 μU/ml). But according to our data, we seem to be able to exclude an amniotic fluid insulin level >16 μU/ml with 100% accuracy. In accordance to the long-term follow-up data from Metzger and Freinkel (8), even Weiss (9) reported that the morbidity of the children affected is limited to high amniotic fluid insulin (∼>17 μU/ml). And to answer Kitzmiller’s concern that we might miss a fetus with morbidity despite normal growth, at least from our data, there was no case of relevant hyperinsulinism with AC <75th percentile. This might be different in women with preexisting diabetes and vascular complications causing growth retardation.
We share Kitzmiller’s concern that we may overtreat fetuses with an AC >75th percentile that is not due to hyperinsulinism. We definitely have to develop more specific ultrasound parameters to distinguish between diabetes and genetic-relat-ed macrosomia. But we face the same problem with our glycemia-based strategy: Weiss (9) reported that 50% of the fetuses had normal amniotic fluid insulin in women with a mean glucose value of 100 mg/dl, which is more or less equivalent to the above-mentioned cutoffs for insulin therapy. Thus, we already seem to overtreat 50% of the women when applying the present practice using solely glycemic criteria.
