Neural tube defects (NTDs), the most common of which are anencephaly, spina bifida, and encephalocele, result from multifactorial disturbances in embryonic neurulation. 1 Studies have established an extensive list of NTD risk factors, including race/ethnicity of parents, maternal age, socioeconomic status, place of residence, gravidity, 2 exposure to antiepileptics or antifolates, 3,4 and maternal disorders such as functional vitamin deficiencies (for example, folic acid). 5,6 Both non-insulin-dependent diabetes mellitus (NIDDM) 7 and obesity, 8–10 two interrelated disorders, increase NTD risk twofold. The similarity of risk conferred by NIDDM and obesity raises the issue of whether they are actually independent risk factors or perhaps two manifestations of one underlying metabolic disorder. One metabolic defect that can coincide with or precede both NIDDM and obesity is hyperinsulinemia. 11
Compared with non-Hispanic populations, the prevalences of NIDDM, 12–14 obesity, 15–17 and NTDs 18,19 are elevated in Mexican Americans. In Texas, the prevalence of NIDDM is two to three times higher among Mexican-American women than among non-Hispanic white women. 12,14 Nearly half of Mexican-American women are overweight [a body mass index (BMI) ≥27.3 kg/m2]. 15 Studies in California and Texas show that the risk of NTDs is 50–200% higher in Mexican-American women than in non-Hispanic white women. 18,19 In 1993–1995 along the Texas-Mexico border, the NTD prevalence was 14.9 per 10,000 livebirths for Mexican Americans and 10.6 for non-Hispanic whites. 19
More than 90% of livebirths along the Texas-Mexico border are Mexican American. 19 This Mexican-American border population, with its high prevalence of NTDs, NIDDM, and obesity, offers a unique opportunity to study the relation among these risk factors. We conducted a case-control study to determine whether maternal hyperinsulinemia is a risk factor for NTDs independent of obesity and hyperglycemia. Obesity was assessed from self-reported prepregnancy height and weight; hyperglycemia was assessed by measuring fasting blood serum glucose levels and through the medical history.
Subjects and Methods
Data for this study came from the Texas Department of Health’s Neural Tube Defect Project, which was conducted along the Texas-Mexico border. The project included multisource active surveillance, a folic acid intervention, and a case-control study. Cases (infants or fetuses) had a diagnosis of anencephalus (International Classification of Diseases, 9th revision, Clinical Modification, code 740), spina bifida (code 741), or encephalocele (code 742.0). Through surveillance, we identified all Mexican-American women with NTD-affected pregnancies who resided and delivered in one of the 14 Texas-Mexico border counties (study area) from June 1995 through May 2000 for participation in the case-control study. Mexican-American control women were identified from study-area residents delivering normal livebirths during the same time period. Control women were randomly selected annually in proportion to the number of livebirths that occurred during the previous year in a given facility. Facilities included hospitals and midwife-attended birthing centers.
About 5–6 weeks postpartum, case and control women were interviewed in person in either English or Spanish using a questionnaire modeled after the Mother Questionnaire (Centers for Disease Control and Prevention, Birth Defect Risk Factor Surveillance; unpublished document, 1992). At this time, fasting whole blood and serum specimens were collected from the participants. Before the interview and blood sample collection, written informed consent was obtained in the subject’s preferred language. Before each interview, the staff physician and interviewer established the date of conception for the index pregnancy using gestational age from the medical record. The interviewer used a personalized calendar to focus the respondent’s attention on exposures and events during the 6-month interval beginning 3 months before and ending 3 months after conception. Through the questionnaire, we assessed maternal age at conception, use of medications, years of schooling, household annual income, reproductive history, and preconceptional use of vitamin supplements. We calculated BMI (kg/m2) from self-reported prepregnancy height and weight. Diabetes history was based on the question, “Were you ever told by a doctor or other health care provider that you had diabetes or high blood sugar?” A positive response would have included insulin-dependent, non-insulin-dependent, and gestational diabetes. We defined diabetics as those with any history of diabetes or a postpartum fasting serum glucose level greater than 125 mg/dl.
Hemoglobin A1c Levels
Whole blood was collected in a 2-ml purple-cap tube, inverted ten times, and shipped daily on wet ice to a SmithKline Beecham clinical laboratory in Dallas for hemoglobin A1c testing (refrigerated overnight if pickup time was missed). The percentage of hemoglobin A1c was measured by ion-exchange high-performance liquid chromatography using the Bio-Rad VARIANT method (Bio-Rad Diagnostics Group, Hercules, CA).
Serum Insulin and Glucose Levels
Whole blood was also collected into a serum separator tube. This blood was allowed to stand at room temperature 30 minutes before being centrifuged at 1,500 rotations per minute for 15 minutes. One milliliter of the serum was transferred to a plastic vial and frozen on dry ice within 30 minutes. These samples were shipped on dry ice overnight and stored at −20°C until the insulin levels were run by the Diabetes Division, Department of Medicine, University of Texas Health Science Center at San Antonio. Serum insulin concentration was measured by a solid-phase 125I radioimmunoassay (Coat-A-Count; Diagnostic Products, Los Angeles, CA). Another milliliter of serum was transferred to a plastic vial, put on wet ice or refrigerated within 30 minutes, and sent to the Dallas SmithKline Beecham clinical laboratory. Serum glucose was measured spectrophotometrically via the Olympus glucose procedure with an Olympus AU5200 analyzer (Olympus America, Melville, NY).
Of 225 Mexican-American case women and 378 Mexican-American control women identified for study, 184 (82%) case women and 225 (60%) control women completed interviews. We did not obtain information on 26 (12%) case women and 101 (27%) control women who refused to be interviewed or on 15 (7%) case women and 52 (14%) control women who had moved out of the study area. We excluded the one woman who reported taking steroids, thyroid medications, amphetamines, or thiazides during the 3 months before and after conception, as these pharmaceuticals may affect blood glucose levels. We restricted the analysis to the 149 (81%) Mexican-American case women and 178 (79%) Mexican-American control women who had values for both serum insulin level and BMI.
To evaluate the association of maternal serum glucose, hemoglobin Alc, and insulin levels, and BMI with NTDs, we categorized women using the laboratory clinical guidelines and for BMI, clinical guidelines from the National Heart, Lung, and Blood Institute. 20 Categories were defined as follows: for glucose, low, <70 mg/dl; normal, 70–115 mg/dl; borderline, 116–126 mg/dl; and diabetic, >126 mg/dl; for hemoglobin A1c, low, <4.4%; normal 4.4–6.1%; and high, >6.1%; for insulin, low, 1–5 micro-international units (μIU)/ml; normal, 6–10 μIU/ml; high, 11–15 μIU/ml; and very high, >15 μIU/ml; and for BMI, underweight, <18.4 kg/m2; normal, 18.5–24.9 kg/m2; overweight, 25.0–29.9 kg/m2; and obese, ≥30.0 kg/m2.
For each measure, we used the normal category as the referent to calculate crude odds ratios and 95% confidence intervals (CIs). Because few women fell into the above-normal categories for glucose (only eight) and for hemoglobin A1c (only nine), we did not examine these variables further. In addition to the univariate crude analysis, we calculated crude odds ratios for the single and joint effects of insulin and BMI using simple stratification. For this and the multiple logistic regression analysis, we combined categories for insulin (low or normal, ≤10 μIU/ml; hyperinsulinemia, >10 μIU/ml) and for BMI (underweight or normal, <30.0 kg/m2; obese, ≥30.0 kg/m2) on the basis of imprecise and similar odds ratios. In multiple logistic regression, we adjusted odds ratios first for hyperinsulinemia or obesity and then for maternal age at conception, education, preconception use of folic acid-containing vitamins, and previous stillbirth or miscarriage. We conducted these analyses with and without the inclusion of known diabetics (those with a history of diabetes mellitus or with serum glucose greater than 125 mg/dl).
Table 1 shows the demographic characteristics of study participants. About half of case and control women had less than 12 years of education and were born in Mexico. More than half of both case and control women had annual incomes of $15,000 or less, and about a quarter were under 20 years of age.
Table 2 shows crude odds ratios for categories of insulin level and BMI. Compared with women with normal insulin levels, NTD risk for women in the high or very high category was similarly elevated with odds ratios of 1.76 (95% CI = 1.01–3.08) and 1.87 (95% CI = 1.07–3.25), respectively. Women who were overweight (25–29 kg/m2) had nearly the same risk as normal women, but obese women (≥30 kg/m2) carried an increased risk (OR = 1.64, 95% CI = 0.94–2.86).
To examine the independent and joint effects of insulin and BMI on NTD risk, we stratified the data as shown in Table 3. Compared with women who were neither hyperinsulinemic nor obese, women with high or very high insulin levels showed a consistent twofold risk in the absence of obesity. Although numbers were small, obese women appeared to have an additional increased risk independent of insulin level.
Table 4 summarizes the independent effects of insulin and obesity on NTD risk determined through multiple logistic regression. Because the high and very high insulin categories had odds ratios of similar magnitude, we dichotomized insulin levels at 10 μIU/ml for this analysis. As in the previous analysis, we dichotomized BMI at 30.0 kg/m2. For comparison, we also present the crude odds ratios for these categories. Adjustment for obesity slightly reduced the odds ratio for hyperinsulinemia, from 1.91 to 1.75, with 95% CIs that exclude the null. Alternatively, adjustment for hyperinsulinemia reduced the odds ratio for obesity slightly more, from 1.73 to 1.45, to an estimate compatible with the null. Additional adjustment for other potential confounders (age, education, previous miscarriage or stillbirth, and folic acid vitamin use) had no appreciable effect on either estimate. Excluding the 23 women considered to be diabetics (women with a history of diabetes mellitus or a fasting glucose level greater than 125 mg/dl) did not materially change adjusted odds ratios for hyperinsulinemia or obesity. If diabetics were excluded, the adjusted odds ratios for hyperinsulinemia and obesity were 1.65 (95% CI = 1.01–2.68) and 1.54 (95% CI = 0.86–2.74), respectively. Examination by type of NTD showed that the odds ratio estimates for hyperinsulinemia (adjusted for obesity) were similar for anencephaly (1.80) and spina bifida (1.90).
With adjustment for BMI, insulin levels in Mexican Americans were related to NTD risk, regardless of diabetic history. The effect of BMI as a risk factor for NTDs was attenuated, but not eliminated, by adjustment for insulin levels. These results suggest that hyperinsulinemia and obesity are related processes, part of the same pathway that leads to an increased NTD risk among Mexican Americans.
In addition to conferring risk of maternal and perinatal mortality, diabetes mellitus in the mother has long been known to increase the risk for a variety of birth defects in offspring, including NTDs. 2 Using 1968–1980 birth data from the Metropolitan Atlanta Congenital Defects Program, investigators found that the risk for major central nervous system defects among insulin-dependent diabetics was 16 times that of nondiabetics. 7 Among other diabetics who required insulin during their pregnancy, the risk was twofold. Although maternal and perinatal mortality dropped precipitously with the availability of insulin after 1920, 21 it was not until the 1980s that it was realized that strict control of glucose metabolism during organogenesis can decrease, 22 and perhaps normalize, 23 the rate of birth defects among babies born to diabetic women.
Since the mid-1990s, at least five case-control studies have reported an increased risk of NTDs among obese mothers. 8–10,24,25 Except for one case-control study from Sweden, 25 estimated relative risks for women with BMIs of greater than 29 or 31 kg/m2 have been remarkably consistent, ranging from 1.8 to 2.0. In the Swedish study, the relative risk estimate for BMI greater than 29 kg/m2 was only 1.3, but it is likely that the average BMI in this obese group was much lower than those for the U.S. studies. 25 Studies that included data on diabetes history uniformly reported that excluding the relatively few women with diabetes mellitus before or during pregnancy did not appreciably change estimated risks for obese women. 8–10,24
The finding of postpartum hyperinsulinemia in case women is highly suggestive of periconceptional hyperinsulinemia in these same women. There is strong evidence that insulin resistance and hyperinsulinemia, independent of obesity, precede the development of NIDDM by many years. 26 In one of the Boston Joslin Diabetes Center studies, 16% of children born to two parents with NIDDM developed the same disease after being traced an average of 13 years. Hyperinsulinemia at study onset was a far more important predictor of NIDDM than initial BMI or fasting glucose. 27 As conceptualized in an extensive review article by DeFronzo, 26 the cascade of events leading to overt NIDDM appears to begin at birth with impaired tissue sensitivity to insulin. This effect is followed years to decades later by a compensatory increase in insulin secretion, which may be aggravated by obesity to yield consistent hyperinsulinemia. Many women who reach this point go on to exhibit impaired glucose tolerance and gestational diabetes in their 20s and 30s, and frank diabetes by their 40s or 50s.
Our study findings on hyperinsulinemia and obesity are tempered by limitations inherent to the case-control study design. First, we measured insulin levels postpartum and cannot be completely positive that these levels reflect the metabolic status of early pregnancy when NTDs occur. Although Reece and Coustan 21 describe normal pregnancy “as a progressive condition of insulin resistance, hyperinsulinemia, and mild postprandial hyperglycemia,” these metabolic changes are not clinically significant until the third trimester, and screening for gestational diabetes is not recommended until 24–28 weeks. 21 Approximately 7% of the women in our sample were classified as diabetics. With the exception of a small number of women with gestational diabetes (2–18%), 28,29 insulin levels normalize well before 5–6 weeks postpartum. 21 Moreover, excluding women with diabetes mellitus from analysis did not change the overall results for hyperinsulinemia and obesity. For all these reasons and because we measured insulin levels 5–6 weeks postpartum, we believe that postpartum levels reflect periconceptional levels, nondifferentially.
Our second study limitation is that prepregnancy weights were based on the recall of mothers an average of 10 months later. Any errors in recall would be expected to distribute equally between case and control women. Nevertheless, the gestational difference between case livebirths and the control normal livebirths lengthened the recall period for control women over that of case women. We argue that this difference was not large; the average recall periods for case and control mothers were 37 and 43 weeks, respectively. Nonetheless, prepregnancy weights were likely to be lower than postpartum weights when insulin levels were measured. Thus, some nondifferential misclassification of BMI was unavoidable, possibly resulting in attenuated effect estimates for obesity.
A final concern was the differing participation rates between case and control mothers (82% and 60%, respectively). The demographic profile of participating control mothers (52% U.S.-born and 51% with education 12 years or more), however, was similar to that of all border Hispanic women giving birth during the study years (50% U.S.-born and 50% with education 12 years or more). This similarity would seem to diminish the chance of any severe selection bias.
By what mechanism could hyperinsulinemia produce NTDs? One of two general mechanisms might be proposed; either maternal hyperinsulinemia is in some way teratogenic or an inherited embryonic insulin/insulin receptor defect is teratogenic. Let us turn first to how maternal hyperinsulinemia might give rise to NTDs. The central nervous system begins to develop (day 16 of gestation) before the pancreas and endocrine hormones, liver, circulatory system, and placenta. Although the placenta is impermeable to insulin later in development, it is unclear whether insulin crosses the placenta at this early stage. Thus, to interfere with normal development, maternal insulin (monomer molecular weight, 5,800 daltons) would need to (1) cross the rudimentary placenta and exert a direct teratogenic effect or (2) exert an indirect teratogenic effect by altering the maternal concentration of some smaller, more permeable molecule that could cross the placenta to harm the developing embryo. The first possibility, that insulin crosses the placental barrier and then exerts a direct teratogenic effect, is unlikely. Experimental data show that 9–10-day rat whole-embryo cultures exposed to insulin levels as much as 250 times physiologic levels (10,000 μU/ml) exhibit no congenital deformities. 30 Therefore, the more plausible of the two scenarios involves a messenger molecule rather than insulin itself. Although the most obvious candidate for this messenger role is glucose, the concentration of a number of other maternal molecules that cross or deliver fuels to the placenta can also be affected by hyperinsulinemia. These include ketone bodies (β-hydroxybutyrate and acetoacetate) 30; free fatty acids 31; triglycerides, very low-density lipoproteins, and various high-density lipoprotein subfractions 32; and amino acids. 33 Although there is some evidence from animals that ketone bodies may be teratogenic, 34,35 a detailed discussion of the potential teratogenic effects of most of these noncarbohydrate substrates is beyond the scope of both this study and these authors.
As is well known, maternal and fetal glucose levels are highly correlated. 36 Both hyper- and hypoglycemia can be harmful to the developing embryo and fetus. Maternal hyperglycemia is associated with spontaneous abortions, 37 congenital malformations including NTDs, 38 and large-for-gestational-age babies. 39 Hypoglycemia is also harmful; in humans it may lead to small-for-gestational age babies, 40 and in mice it appears to be teratogenic. 30,41,42 The hypothesis that embryos deprived of glucose (or for that matter, other fuels such as free fatty acids) may simply lack the energy necessary for development of the neural tube or repair of developmental errors was termed “fuel-mediated teratogenesis” by Norbert Freinkel in 1980. 43
Now let us turn to the possibility that a genetic defect leading to a dysfunctional embryonic insulin (or a biologically active precursor such as proinsulin or preproinsulin) or insulin receptor (or precursor) could be causing NTDs. Insulin and/or its receptor are appealing candidates for this etiologic role, as they seem to be in the right spot at the right time. For instance, in the chick embryo, the neural tube appears to be one of the structures in which prepancreatic preproinsulin and its receptor are localized during neurulation (before the pancreas and circulatory system form). 44 Unlike pancreatic insulin, which is partially responsive to glucose levels, the level of prepancreatic preproinsulin and its receptor do not appear to be regulated by glucose in chick embryos. 44 That their levels are independent of both maternal insulin and glucose levels bolsters the arguments for an intrinsic embryonic genetic defect and detracts from the arguments that an abnormal concentration of an extrinsic maternal metabolite is responsible for the defects. Additional experiments by the same authors suggest a possible cellular mechanism. Administration of insulin to chick embryos during neurulation seems both to prevent apoptosis and to stimulate growth. 44
In conclusion, we observed a strong association between hyperinsulinemia and NTD risk in a Mexican-American population at high risk for NTDs, diabetes mellitus, and obesity. Although these findings have important implications for the prevention of NTDs, they should be confirmed in non-Hispanic ethnic groups. In addition to the intriguing benchtop issues of embryonic and molecular mechanisms, two questions of particular relevance to public health policy remain. First, for the prevention of NTDs, should populations at high risk for diabetes mellitus be screened for hyperinsulinemia before conception? Our population tended to be hyperinsulinemic; 54% of control women had elevated insulin levels (>10 μIU/ml). A cohort study looking for the development of NTDs and other birth defects in other hyperinsulinemic populations would help answer this question. Second, does a population that is genetically disposed to diabetes mellitus require higher than recommended levels of folic acid to prevent NTDs? At present, this Mexican-American border population has far lower intakes of supplemental folic acid than the general population (3–9% compared with 25%). 45 This observation and the apparent additional risk associated with a prediabetic state imply that supplementation with folic acid would have a relatively large beneficial effect in this Mexican-American population.
We thank the following NTD Project team members for their crucial role in interviewing case and control women: in El Paso, Hilda Chavarria, Maria Torres, Carmen Ramos, Donna Brom, and Patricia Velázquez; in Harlingen, Oralia Villafranca, San Juana Thompson, Graciela Rubio, Manuela Flores, Rene Rodríguez, Sara Mungia, and Jorge Trevino; and in Laredo, Ricardo Treviño, Miguel Madrigal, Olivia Macias Gutiérrez, Cynthia Medina de Llano, Jackie Bassini, and Armandina Ortiz. We also thank Rich Ann Roche, Kelly Johnson, Hermia Brooks, Billie Woullard, Jennifer Tisch, John Dunn, and Jackie Stroupe for their careful work that insured the accuracy of the data; the University of Texas Health Science Center Research Laboratory and Ralph De Fronzo; E. Albert Reese for reviewing the penultimate draft of the manuscript; Tom Sadler for his thought-provoking discussions; and Scott Simpson for his early and unflagging support of the insulin hypothesis.
1. Copp AJ, Brook FA, Estibeiro JP, Shum AS, Cockroft DL. The embryonic development of mammalian neural tube defects. Prog Neurobiol 1990; 35: 363–403.
2. Elwood JM, Little J, Elwood JH. Epidemiology and Control of Neural Tube Defects. Oxford, UK: Oxford University Press, 1992.
3. Nau H. Valproic acid-induced neural tube defects. In: Bock G, Marsh J, eds. Neural Tube Defects. Ciba Foundation Symposium No. 181. Chichester, UK: John Wiley and Sons, 1994; 144–160.
4. Warkany J. Aminopterin and methotrexate: folic acid deficiencies. Teratology 1978; 17: 353–358.
5. MRC Vitamin Study Research Group. Prevention of neural tube defects: results of the MRC vitamin study. Lancet 1991; 338: 132–137.
6. Czeizel AE, Dudas I. Prevention of the first occurrence of neural-tube defects by periconceptional vitamin supplementation. N Engl J Med 1992; 327: 1832–1835.
7. Becerra JE, Khoury MJ, Cordero JF, Erickson DJ. Diabetes mellitus during pregnancy and the risks for specific birth defects: a population-based case-control study. Pediatrics 1990; 85: 1–9.
8. Waller DK, Mills JL, Simpson JL, Cunningham GC, Conley MR, Lassman MR, Rhoads GG. Are obese women at higher risk for producing malformed offspring? Am J Obstet Gynecol 1994; 170: 541–548.
9. Shaw GM, Velie EM, Schaffer D. Risk of neural tube defect-affected pregnancies among obese women. JAMA 1996; 275: 1093–1096.
10. Watkins ML, Scanlon KS, Mulinare J, Khoury MJ. Is maternal obesity a risk factor for anencephaly and spina bifida? Epidemiology 1996; 7: 507–512.
11. Plaisted CS, Istfan NW. Metabolic abnormalities of obesity. In: Blackburn GL, Hardens BS, eds. Obesity Pathophysiology, Psychology, and Treatment. New York: Chapman and Hale Inc, 1994; 79–97.
12. Hanis CL, Ferrell RE, Barton SA, Aguilar L, Garza-Ibarra A, Tulloch BR, Garcia CA, Schull WJ. Diabetes among Mexican Americans
in Starr County, Texas. Am J Epidemiol 1988; 128: 1302–1311.
13. Perez-Stable EJ, McMillen MM, Harris MI, Juarez RZ, Knowler WC, Haynes SG. Self-reported diabetes in Mexican Americans
: HHANES 1982–1984. Am J Public Health 1989: 79: 770–772.
14. Stern MP, Haffner SM. Type II diabetes and its complications in Mexican Americans
. Diabetes Metab Rev 1990; 6: 29–45.
15. National Center for Health Statistics. Health, United States: 1995. Hyattsville, MD: US Public Health Service, 1996; 183.
16. Hazuda HP, Mitchell BD, Haffner SM, Stern MP. Obesity in Mexican-American subgroups: findings from the San Antonio Heart Study. Am J Clin Nutr 1991; 53: 1529S–1534S.
17. Stern MP, Gaskill SP, Allen CR, Garza V, Gonzalez JL, Waldrop RH. Cardiovascular risk factors in Mexican Americans
in Laredo, Texas. I. Prevalence of overweight and diabetes and distribution of serum lipids. Am J Epidemiol 1981; 113: 546–555.
18. Harris JA, Shaw GM. Neural tube defects: why are rates high among populations of Mexican descent? Environ Health Perspect 1995; 103 (suppl 6): 163–164.
19. Hendricks KA, Simpson JS, Larsen RD. Neural tube defects along the Texas-Mexico border, 1993–1995. Am J Epidemiol 1999: 149: 1119–1127.
20. US Department of Health, Public Health Service. National Heart, Lung, and Blood Institute. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: The Evidence Report. NIH Pub. No. 98–4083. Bethesda, MD: National Heart, Lung, and Blood Institute, 1998.
21. Reece EA, Coustan DR, eds. Diabetes Mellitus in Pregnancy. 2nd ed. New York: Churchill Livingstone, 1995.
22. Mills JL, Knopp RH, Simpson JL, Jovanovic-Peterson L, Metzger BE, Holmes LB, Aarons JH, Brown Z, Reed GF, Bieber FR, Van Allen M, Holzman I, Ober C, Peterson CM, Withiam MJ, Duckles A, Mueller-Heubach E, Polk BF. Lack of relation of increased malformation rates in infants of diabetic mothers to glycemic control during organogenesis. N Engl J Med 1988; 318: 671–676.
23. Fuhrmann K, Reiher H, Semmler K, Fischer F, Fischer M, Glockner E. Prevention of congenital malformations in infants of insulin dependent diabetic mothers. Diabetes Care 1983; 6: 219–223.
24. Werler MM, Louik C, Shapiro S, Mitchell AA. Prepregnant weight in relation to risk of neural tube defects. JAMA 1996; 275: 1089–1092.
25. Kallen K. Maternal smoking, body mass index, and neural tube defect. Am J Epidemiology 1998; 147: 1103–1111.
26. DeFronzo RA. Pathogenesis of type 2 diabetes: metabolic and molecular implications for identifying diabetes genes. Diabetes Metab Rev 1997; 5: 177–269.
27. Warram JH, Martin BC, Krolewski AS, Soeldner JS, Kahn CR. Slow glucose removal rate and hyperinsulinemia precede the development of type II diabetes in the offspring of diabetic parents. Ann Intern Med 1990; 113: 909–915.
28. O’Sullivan JB, Mahan CM. Criteria for the oral glucose tolerance test in pregnancy. Diabetes 1964; 13: 278–284.
29. Coustan DR, Lewis SB. Insulin therapy for gestational diabetes. Obstet Gynecol 1978; 51: 306–310.
30. Sadler TW, Horton WE Jr. Effects of maternal diabetes on early embryogenesis: the role of insulin and insulin therapy. Diabetes 1983; 32: 1070–1074.
31. Hendrickse W, Stammers JP, Hull D. The transfer of free fatty acids across the human placenta. Br J Obstet Gynaecol 1985; 92: 945–952.
32. Knopp RH, Bergelin RO, Wahl PW, Walden CE. Relationships of infant birth size to maternal lipoproteins, apoproteins, fuels, hormones, clinical chemistries, and body weight at 36 weeks gestation. Diabetes 1985; 34 (suppl 2): 71–77.
33. Kalkhoff RK. Impact of maternal fuels and nutritional state on fetal growth. Diabetes 1991; 40 (suppl 2): 61–65.
34. Horton WE Jr, Sadler TW. Mitochondrial alterations in embryos exposed to B-hydroxybutyrate in whole embryo culture. Anat Rec 1985; 213: 94–101.
35. Hunter ES III, Sadler TW, Wynn RE. A potential mechanism of DL-beta-hydroxybutyrate-induced malformations in mouse embryos. Am J Physiol 1987; 253 (pt 1):E72–E80.
36. Pedersen J. Weight and length at birth of infants of diabetic mothers. Acta Endocrinol 1954; 16: 330–342.
37. Miodovnik M, Skillman C, Holroyde JC, Butler JB, Wendel JS, Siddiqi TA. Elevated maternal glycohemoglobin in early pregnancy and spontaneous abortion among insulin-dependent diabetic women. Am J Obstet Gynecol 1985; 153: 439–442.
38. Reece EA, Hobbins JC. Diabetic embryopathy: pathogenesis, prenatal diagnosis and prevention. Obstet Gynecol Surv 1986; 41: 325–335.
39. Pedersen J, Osler M. Hyperglycemia as the cause of characteristic features of the foetus and newborn of diabetic mothers. Dan Med Bull 1961; 8: 78–83.
40. Langer O, Levy J, Brustman L, Anyaegbunam A, Merkatz R, Divon M. Glycemic control in gestational diabetes mellitus—how tight is tight enough: small for gestational age versus large for gestational age? Am J Obstet Gynecol 1989; 161: 646–653.
41. Smithberg M, Runner MN. Teratogenic effects of hypoglycemic treatments in inbred strains of mice. Am J Anat 1963; 113: 479–489.
42. Cole WA, Trasler DG. Gene-teratogen interaction in insulin-induced mouse exencephaly. Teratology 1980; 22: 125–139.
43. Metzger BE. Biphasic effects of maternal metabolism on fetal growth. Diabetes 1991; 40 (suppl 2): 99–105.
44. Perez-Villamil B, de la Rosa EJ, Morales AV, de Pablo F. Developmentally regulated expression of the preproinsulin gene in the chicken embryo during gastrulation and neurulation. Endocrinology 1994; 135: 2342–2350.
45. Centers for Disease Control and Prevention. Neural tube defect surveillance and folic acid intervention, Texas-Mexico border, 1993–1998. MMWR 2000; 49: 1–4.