The randomized trial of Schaefer-Graf et al. (1) evaluating a fetal growth–based strategy to guide the management of gestational diabetes mellitus (GDM) in Berlin requires comment. The authors used measurement of fetal abdominal circumference (AC) at 20–35 weeks’ gestation taken by only three ultrasonographers (physician investigators) to determine the need for insulin therapy (36 of 90 subjects) compared with self-monitored fasting blood glucose (>90 mg/dl) and 2-h postprandial blood glucose (>120 mg/dl) values obtained 2 days per week, which indicated the need for insulin treatment (27 of 97 subjects). The authors found that the ultrasound-based strategy provided outcomes (12% large for gestational age, 17% neonatal hypoglycemia, 14% transfer to the neonatal intensive care unit [NICU]) that were no different from the glycemic criteria strategy. However, the study design set up the glycemic control group for only partially intensified management by not requiring daily self-monitoring of fasting blood glucose (SMBG), which many other investigators have found to produce lower perinatal morbidity rates than in this study (2–6), treating GDM subjects with higher glucose diagnostic criteria than in Germany. The German authors also used tighter blood glucose treatment targets in the ultrasound-based than in the SMBG-based strategy.
The authors claimed that any physician certified for obstetrical ultrasound may be expected to produce reliable fetal AC measurements with standard equipment, since their interobserver coefficient of variance for the AC measurements was <7%. However, in most clinical settings in North America, technicians make the AC measurements and physicians trained in imaging read the films. As a result, there is greater variance in the accuracy and repeatability of fetal AC measurements than in this study. This reduces the utility of this measurement in standard clinical practice as a criterion for insulin treatment.
The concept that some marker of fetal hyperinsulinemia should be the indicator for aggressive therapy in GDM is an old idea (7,8). The authors stated that fetal AC <75th percentile for gestational age as a marker of minimal adiposity reliably excludes elevated amniotic fluid insulin levels on the day of amniocentesis. However, Kainer et al. (9) found that the sensitivity of excess AC to predict pathological amniotic fluid insulin was only 67% and the specificity was 53%, and Schaefer-Graf et al. (10) originally found that 11 of 74 fetuses with AC <75th percentile had amniotic fluid insulin >97th percentile for 822 nondiabetic control subjects in the landmark study of Weiss (11). Schaefer-Graf et al. also originally showed that the majority of fetuses with AC >75th percentile had normal amniotic fluid insulin. The variance between fetal AC and amniotic fluid insulin is not surprising, since amniotic fluid insulin can vary with time of day and maternal-fetal glycemia in metabolically unstable women.
One wonders why 75 infants of GDM mothers in the German study (1), with normal fetal AC and no hyperglycemia on 2 days/week, had a 19% rate of neonatal hypoglycemia and 17% transfer to NICU. Perhaps it was related to the inclusion of subjects who smoked up to five cigarettes per day, or perhaps it was related to hyperglycemia on the nonmonitored days. The concept that all we have to worry about in GDM is fetal adiposity ignores the long-established fact that nonmacrosomic infants of diabetic mothers can have neonatal morbidity. Furthermore, recent data show that maternal-fetal hyperglycemia can affect fetal/placental gene expression, which might be important in short- and long-term outcomes. Finally, measurement of glycemic responses to daily food intake in GDM is attractive for its educational value in women at high risk of developing type 2 diabetes.