Never Say Never in Medicine

F or over two decades I have religiously clung to the notion that maternal hyperglycemia is the root of all evil in pregnancies complicated by diabetes (1). To this end, I have valiantly waved the flag for universal screening for gestational diabetes (2), intensified glucose monitoring (3,4), and intensified insulin delivery for all women with hyperglycemia in pregnancy (5). In addition, I championed absolute normalization of blood glucose to mimic the levels documented in a normal population of pregnant women, as recently reported by Parretti et al. (6). Despite the fact that I have been a coauthor on articles that suggest that normalization of blood glucose may be associated with an increased risk of small-forgestational-age infants (7), I have tended to ignore or explain away this finding even when the evidence has been convincing. This old dog became an ostrich. Her head was in the sand despite examples of definitive studies that showed that there is an increased prevalence of smallfor-dates ( 10th percentile birth weight) infants in programs of “tight control.” The landmark article of Langer et al. (8) showed the relationship between optimal levels of glycemic control and perinatal outcome in a prospective study of 334 gestational diabetic women and 334 control subjects matched for obesity, race, and parity. Three groups were identified on the basis of mean blood glucose level throughout pregnancy (low, 86; mid, 87–104; and high, 105 mg/dl). The low group had a significantly higher prevalence of small-for-gestational-age infants (20%). In contrast, the prevalence of large-for-gestational-age infants was 21fold higher in the high mean blood glucose category than in the low mean blood glucose category (24 vs. 1.4%, P 0.0001). In the control group, the overall prevalence was only 11% for small-forgestational-age infants and only 12% for large-for-gestational-age infants. They concluded that a relationship exists between level of glycemic control and neonatal weight. “Too tight control,” defined as a mean capillary glucose 87 mg/dl, is associated with a higher risk of intrauterine growth retardation in offspring of gestational diabetic women. If one did not have an emotional attachment to this debate, then when the first report by Buchanan et al. (9) was published, which compared management based on maternal glycemic criteria with management based on fetal abdominal circumference measurements by ultrasound to select gestational diabetic women for insulin treatment, the obvious interpretation would have been that using ultrasound may be the ideal means to identify the women who would benefit from an intensive insulin protocol. Instead, I interpreted the results with caution (10). Buchanan et al. (9) showed in 98 gestational diabetic women with fasting plasma glucose concentrations of 105–120 mg/dl that the women randomized to begin insulin when self-monitored capillary glucose levels were elevated fared no better than the women randomized to receive insulin only when the abdominal circumference, measured monthly, was 70th percentile. Birth weights (3,271 458 vs. 3,369 461 g), frequencies of birth weights 90th percentile (6.3 vs. 8.3%), and neonatal morbidity (25 vs. 25%) did not differ significantly between the standard and experimental groups, respectively. However, the cesarean delivery rate was significantly lower (14.6 vs. 33.3%, P 0.03) in the group that was managed by blood glucose monitoring and initiation of insulin when blood glucose became elevated (the standard group) when compared with the group managed by ultrasound abdominal circumference measured monthly. This difference in cesarean delivery rate was not explained by birth weights. They concluded that in women with gestational diabetes, fetal abdominal circumference measurements identified pregnancies at low risk for macrosomia and resulted in the avoidance of insulin therapy in 38% of patients without increasing rates of neonatal morbidity. My argument against the use of ultrasound as the sole guide for insulin therapy rather than relying on the maternal glucose concentrations was that we may be missing the opportunity to prevent fetal macrosomia (10–13). In our Latino population in California, which has such a high risk of macrosomia if hyperglycemia is left untreated, I became an even stronger advocate for intensified insulin delivery for all documented hyperglycemia in pregnancy (14). However, the study by Schaefer-Graf et al. (15), in this issue Diabetes Care, on German women with gestational diabetes confirms the original report. This study may have changed my mind. They showed that when the approach to insulin therapy was determined by monthly fetal growth patterns as evidenced by ultrasound, the outcome improved with a lower cesarean section rate and no increase in macrosomia fetal morbidity as compared with a group of gestational diabetic women treated with insulin based solely on maternal glycemia. They also showed that, much like the study from California (9), this approach reduced the number of women with mild gestational diabetes who required selfmonitoring of glucose and/or exogenous insulin therapy, thereby providing the potential to improve the cost-effectiveness of antepartum management of gestational diabetes. Before this present study, I felt that saving treatment for only those mothers whose babies were already big and sick seemed to go against the notion that fetal macrosomia is the origin of adult type 2 diabetes (16–18). It was not rational, but despite the evidence, I did not believe that we could ignore women who had documented hyperglycemia. In all fairness, I was ignoring the other end of the fetal growth curve. The “Barker Hypothesis” (19) suggests that low birth weight predicts subsequent physiological disturbances in adult life. Small-for-gestational-age infants and/or intrauterine growth– retarded fetuses have been reported (20– E D I T O R I A L ( S E E S C H A E F E R G R A F E T A L . , P . 2 9 7 )