Malformations in Infants of Diabetic Mothers

Abstract

Maternal insulin-dependent diabetes has long been associated with congenital malformations. As other causes of mortality and morbidity have been eliminated or reduced, malformations have become increasingly prominent. Although there is not universal agreement, the great majority of investigators find a two- to threefold increase in malformations in infants of insulin-dependent diabetic mothers. This increase is not seen in infants of gestational diabetics. It probably is not present in women whose diabetes can be controlled by diet or oral hypoglycemic agents. The risk does not appear to be primarily genetic since diabetic fathers do not have an increased number of malformed offspring. Most studies show a generalized increase in malformations involving multiple organ systems. Multiple malformations seem to be more common in diabetic than non-diabetic infants. Caudal regression has the strongest association with diabetes, occurring roughly 200 times more frequently in infants of diabetic mothers than in other infants. The teratogenic mechanism in diabetes is not known. Hyperglycemia may be important but human studies focusing on the period of organogenesis are lacking. Hypoglycemia has also been suggested based mainly on animal experiments. Insulin appears unlikely. Numerous other factors including vascular disease, hypoxia, ketone and amino acid abnormalities, glycosylation of proteins, or hormone imbalances could be teratogenic. None has been studied in sufficient detail to make a judgment. A large-scale prospective study is required to determine early fetal loss rates, correlate metabolic status during organogenesis with outcome, and assess the effect of diabetic control on malformation rates.

An association between diabetes mellitus in women and congenital malformations in their offspring has been suspected since the nineteenth century. In 1885, LeCorché reported two infants of diabetic mothers with hydrocephalus. The prognosis for diabetic women prior to the discovery of insulin was poor, however, and few women delivered successfully. It was not until better control of hyperglycemia, close monitoring in the last months of pregnancy, and early delivery for fetal distress were instituted that salvage rates in diabetic pregnancies improved substantially. At this point, the full impact of congenital malformations was appreciated.

Despite better management, the incidence of congenital malformations has not decreased over the past 25 years (Soler, ’76). Congenital malformations have now replaced respiratory distress syndrome as the leading cause of death in some diabetes centers (Soler, ’76). This has stimulated investigators to examine the relationship between maternal diabetes and malformations.

EVIDENCE THAT INFANTS OF DIABETIC MOTHERS HAVE HIGHER MALFORMATION RATES

Evidence that infants of diabetic mothers have higher malformation rates has accumulated over the last several decades. Initially, centers reporting their experience with diabetic pregnancies noted high malformation rates in the infants of diabetic mothers. The author is aware of more than a dozen studies reporting malformation rates of 6% or more. Since malformations were not the primary focus of most of these studies, they were frequently uncontrolled and, hence, useful mainly to direct attention to the question of teratogenesis.

Numerous animal studies (to be discussed in detail later) were then performed to define the relationship between diabetes and malformations. Rats and mice made diabetic by alloxan or streptozotocin consistently produced more malformed offspring than expected. One recent experiment (Sadler, ’79) took serum from diabetic rats and injected it into mouse embryo cultures. The injected (but not control) embryos developed a dose-related increase in malformations, suggesting that some diabetic factor, not the alloxan or streptozotocin, is teratogenic.

The definitive study relating diabetes to malformations in humans has not yet been reported. Before discussing specific studies comparing malformation rates in infants of diabetic mothers with control infants, it would be wise to consider some of their methodologic weaknesses. The diabetic group may not be representative of all diabetics. This is particularly true at university hospitals with referral populations, since those with more severe disease or complications are more likely to be referred. If, for example, vasculopathy were responsible for the increased malformation rate, the university hospitals would be likely to see a larger percentage of malformed infants, since those who have vascular disease are more likely to be cared for at a university hospital than those who do not. To overcome this problem, a population-based study in which all diabetic women from a given area are included would be needed.

The source of the control population is just as important as the source of the diabetic population. Control subjects chosen from a different area or bearing children at a different time may be totally inappropriate. To use an extreme example, suppose control subjects were chosen from a time and place where thalidomide was being used during pregnancy. The control population would show a large number of limb defects. If the diabetic group were from a non-thalidomide-exposed area or time period, they would not show the same thalidomide-induced defects. If diabetes produced limb defects, the association could be masked by the high rate in the control population. On the other hand, if the control group were contemporaneous, both groups would show the thalidomide effect and the diabetes effect would show up as an added risk.

Another advantage to contemporaneous control subjects is that diagnosis or definition of congenital malformations is more likely to be consistent. Ideally, the same person should examine all subjects using an agreed upon plan for diagnosing malformations. Agreement on what constitutes a malformation is important because, if cosmetic problems such as nevi or skin tags are counted as malformations in one group but not the other, inaccurate comparisons of rates will result.

The question of how malformations are ascertained is a particularly thorny one in studies of infants of diabetic mothers. Since these children are at high risk for respiratory distress syndrome, hypoglycemia, and hypocalcemia, it is not surprising that they receive a good deal of medical attention in the neonatal period. This attention could result in the discovery of malformations which would go unnoticed in the control population. The infants of diabetic mothers would then appear to have more malformations when, in fact, they merely had more complete ascertainment of their malformations. Fortunately, this is not as great a problem with major malformations as it is with minor malformations, since most (although not all) are readily diagnosable. All of these potential biases must be kept in mind when evaluating a report of malformations associated with diabetes.

Nonetheless, nearly all studies to date are consistent in finding a significantly increased risk for major malformations in the infants of diabetic mothers (Molsted-Pedersen et al., ’64; Soler et al., ’76; Chung and Myrianthopoulos, ’75; Yssing, ’75). Now let us review the evidence offered by some of the better studies.

Moltsed-Pedersen et al. (’64) examined 853 consecutive infants of diabetic mothers in Denmark and compared them to 1212 infants of nondiabetic mothers. As shown in Table 1, the frequency of major (including fatal) and total malformations was three times higher in the diabetic group. Fatal malformations and malformations involving several organ systems were approximately six times more frequent in the diabetics. All subjects came from the same departments of the hospital. Most of the infants of diabetic mothers were examined by the same person. Unfortunately, the control subjects were obtained over a six-month period while the diabetics spanned 37 years. Data were obtained retrospectively. This study is particularly noteworthy because of the large number of diabetic pregnancies studied and because 791 of the subjects were insulin-dependent before pregnancy.

TABLE 1.

Congenital malformations in 853 infants of diabetic mothers and 1,212 control infants (over 1000 gm)

	Total	Number with congenital malformations
		Total	Major	Fatal
Infants of diabetic mothers	853	55 (6.4%)	44 (5.2%)	18 (2.1%)
Infants of nondiabetic mothers	1,212	26 (2.1%)	14 (1.2%)	4 (0.3%)

Pedersen et al., Lancet, 1964.

The Collaborative Perinatal Project (Chung and Myrianthopoulos, ’75), a prospective study, also found that infants of insulin-dependent diabetic mothers were more likely to have malformations. They examined 48,437 infants, diagnosing malformations throughout the first year of life. White infants of insulin-dependent diabetic mothers had a 17.7% major malformation rate compared to 8.3% in non-diabetics. Black infants of insulin-dependent mothers showed a risk of 17.3 versus 8.5% for nondiabetics. The increased rate in the diabetic groups was statistically significant in both whites and blacks. The Collaborative Perinatal Project obtained subjects from 14 geographically dispersed institutions. Its strengths include the prospective approach, careful diagnosis of diabetes and malformations, and a very large, well-studied, normal control population. Because of the rarity of diabetes, the Collaborative Perinatal Project added patients not selected according to the protocol procedure from the Joslin Clinic to increase the sample size. While these subjects were not, strictly speaking, from the same population as the remainder, they were followed in the same fashion.

Kucera’s (’71) analysis of the world’s literature, while not as well designed as the Collaborative Perinatal Project, is interesting nonetheless. Combining all reported case series of infants of diabetic mothers, he found an overall malformation rate of 4.8%. He selected data from the World Health Organization for comparison. This control group showed a malformation rate of 1.65% (See Table 2). The value of this study lies in its summary of reports of all types of malformations. Since the World Health Organization data also specify type of malformation, it is possible to look at specific malformations, albeit crudely, with this report. Despite the major limitations of this method (the use of literature reports for rates and different definitions of malformations), this report is consistent with other studies showing that diabetic women are two to three times more likely to produce malformed infants.

TABLE 2.

Summary of the world literature on malformations in infants of diabetic mothers (1930–1964)

	Total	Number with malformations	Percentage with malformations
Infants of diabetic mothers	7,101	340	4.8
W.H.O. world survey of malformation rates in “normals”	431,764	7,124	1.65

Kucera, J. Reprod. Med., 1971.

Whereas numerous other studies have also found higher malformation rates in infants of diabetic mothers, one well-performed study did not. Farquhar (’59, ’69) reported that infants of diabetic mothers in Edinburgh did not show higher malformation rates than other infants. The reason for this discrepancy is not clear from Farquhar’s data; he does not specify actual malformations seen. He does note, however, that the diabetic group had more severe malformations. This is consistent with others’ observations that it is in major malformations rather than minor that the increase is found. Farquhar’s work not withstanding, it is generally accepted that infants of insulin-dependent diabetic mothers have malformation rates two to three times higher than normal, and that major malformations account for most of the increase.

WHO IS AT RISK?

As the increased risk of malformations became evident, investigators turned their attention to determining which diabetics were at risk.

An obvious question was, do diabetic fathers have more malformed children? This question has important etiologic implications because a link between various HLA types and insulin-dependent diabetes has been established in many different populations. If this “diabetes-associated gene” were responsible for the higher malformation rate, infants of diabetic fathers would be likely to show an increase as well. Bennett and associates (’79) have reviewed the evidence from their own studies of Pima Indians and from the Collaborative Perinatal Project. As shown in Table 3, diabetic (and prediabetic) fathers do not produce more malformed children than nondiabetics. This suggests that the teratogenic effect of diabetes is not a result of genetic factors related to diabetes-associated genes. It does not, however, rule out the possibility that women with the genetic marker for diabetes might have an increased susceptibility to some potentially teratogenic metabolic derangement such as hyperglycemia.

TABLE 3.

Incidence of malformations and paternal diabetes

	Number of pregnancies	Malformations				Significance
		Number	Percentage
Collaborative Perinatal Project
Nondiabetic father	51,803	4,298	8.28			P > 0.05
Diabetic father	253	23	9,09
Pima Indians
Nondiabetic father	395	13	3.3			P > 0.05
Prediabetic father	197	7	3.6		3.7
Diabetic father	44	2	4.5

Bennett, Pregnancy Metabolism Diabetes and the Fetus, 1979.

Although the increased risk for producing malformed infants lies with the diabetic mother, not the diabetic father, it does not appear that women with all types of diabetes are at equal risk.

There is very little to suggest that “prediabetic” women are at increased risk of producing malformed infants. Only two highly controversial studies have found an increased risk of malformations in the “prediabetic” woman. In one study, 34.3% of women with normal carbohydrate tolerance during pregnancy who delivered malformed infants had developed diabetes 25 years later. Only 1% of control women were found to have diabetes (Navarette et al., ’70). A second investigator (Hagbard, ’58) found an increased incidence of malformations in infants delivered up to 20 years before clinically apparent maternal diabetes. These reports suggest that the “prediabetic” woman may already be at increased risk before she develops diabetes. There are, however, compelling reasons to question these findings. Selection bias may be a problem in this type study. A prevalence of diabetes 30 times higher in the case group than in the control group suggests either a very strong exposure effect (which in this case has not been confirmed) or a major selection bias. Without additional information on how diabetic and control women were selected for study and how malformations were searched out, these studies cannot be accepted. In fact, it is reasonable to conclude that the “prediabetic” woman is not at risk, since if she were, the woman who actually develops diabetes during pregnancy should certainly show an increased risk. This, as we shall see, does not appear to be the case.

Malformation rates in infants of gestational diabetic women (White Class A) have been published by many centers (Soler et al., ’76; Chung and Myrianthopoulos; ’75; Bennett et al., ’79; Day and Insley, ’76; Karlson and Kjellmer, ’72). There is general agreement that malformation rates are not increased. The findings of the Collaborative Perinatal Project are typical. Among 392 gestational diabetics, the malformation rates were 15.3% for whites and 13.7% for blacks. The corresponding rates for nondiabetics were 14.6 and 17.0%, respectively. The differences were not significant. This study is very strong on several counts. The number of diabetics and normal controls is large. It is prospective, making good documentation of gestational diabetes possible, and all participants were carefully screened for malformations. This study clearly demonstrates that those who do not have diabetes prior to pregnancy are not at increased risk for producing malformed infants.

Among women who already have diabetes at the time of conception, those who are noninsulin-dependent must be considered separately. Data on women treated with oral hypoglycemic agents are not clear cut. The Birmingham Maternity Hospital Series (Day and Insley, ’76) showed that 31 women treated by diet or hypoglycemic agents at the time of conception had no more malformed infants than control women. Since the percentage receiving oral hypoglycemic agents is not given, it is impossible to tell whether an increased risk in those on hypoglycemic agents was diluted out by a larger number in the diet group. The Collaborative Perinatal Project (Heinonen et al., ’77) examined the risk of malformations in offspring of 21 women receiving oral hypoglycemic agents. Their survival and race-standardized relative risk was 1.10, not significantly different from nondiabetic women. On the other hand, Soler et al., (76) reported that the offspring of 46 women receiving oral hypoglycemics in the first trimester had an 8.7% malformation rate. This was similar to the insulin-dependent diabetics’ rate (8.1 %) in their center. The authors estimated the rate among nondiabetics to be only 1.7%. They acknowledged that the rate among nondiabetics did not come from the same source and was not strictly comparable. They maintained, however, that their results suggested an increased risk for non-insulin-dependent diabetics receiving oral hypoglycemic agents.

It is difficult to draw any conclusions from these studies. The number of subjects in each is small, making their rates unstable. While most of the evidence suggests that women treated with oral hypoglycemic agents do not produce more malformed infants, the evidence is not conclusive.

As noted previously, the evidence that insulin-dependent diabetic women produce more malformed infants is very strong. Investigators have examined factors such as duration of disease and age of onset in insulin-dependent women to determine how they influence malformation rates. White Class, which is based on age of onset, duration of disease, and pressence of complications, correlates well with malformation rates. Those with earliest onset, longest duration, and with complications have the highest risk according to most authors (Molsted-Pedersen et al., ’64; Soler et al., ’76; Karlson and Kjellmer, ’72). Day and Insley (’76) are an exception. They found no correlation between malformations and age of onset, duration of diabetes, vascular complications, maternal age, or parity. It is likely that their failure to find an association is due to the small number of cases examined. It is reasonable that early onset and long duration of diabetes would be associated with higher malformation rates, since early onset and long duration would generally result in a more abnormal maternal milieu during pregnancy. Diabetic complications, particularly vascular, could be a reflection of this metabolic disorder as well as a direct mechanism for teratogenesis. Later, we shall consider how specific aspects of the disturbed metabolic environment and complications relate to malformations.

WHAT TYPES OF MALFORMATIONS ARE ASSOCIATED WITH DIABETES?

Multiple organ systems are susceptible to the teratogenic effects of diabetes. All studies of congenital anomalies in infants of diabetic mothers have shown cardiovascular, genitourinary, musculoskeletal, and other malformations. No single study has contained enough cases to determine whether or not the risk for each specific defect is significantly increased. The number of subjects required would be enormous. Kucera (’71) tried to estimate relative risks based on cases reported in the literature. His estimates are derived from malformation rates in published series of diabetic pregnancies (1945–1965) and from additional data obtained directly from the authors. He reviewed 7101 pregnancies, of which 340 produced malformed offspring. This is a rough estimate of relative risk because reported series may not be a good choice of a denominator for mothers at risk and because his comparison group, World Health Organization “normals,” does not come from the same population. Keeping these limitations in mind, it is interesting that he found that abnormalities of the skeleton, kidneys, heart, gastrointestinal system, and genitalia all occurred significantly more frequently in infants of diabetic mothers. An estimate of the risk ratio for infants of diabetic mothers calculated from Kucera’s data is shown in Table 4. Thus, the teratogenic action of diabetes is generalized.

TABLE 4.

Congenital anomalies most often seen in infants of diabetic mothers with risk ratios

Anomaly	Risk ratio
Caudal regression	252
Situs inversus	84
Ureter duplex	23
Renal agenesis	6
Cardiac anomalies (TGA, VSD, ASD)	4
Anencephalus	3

Mills, Diabetes, 1979.

It is also apparent that the teratogenic effect occurs early in gestation. The malformations Kucera found to be significantly more frequent were examined by the author and his associates (Mills et al., ’79), using a developmental morphologic approach. This method is based on the knowledge of when organ development occurs and on the assumption that malformations of an organ cannot occur after the organ has been completely differentiated. Thus, for each anomaly, a time was assigned before which the defect must have occurred. As shown in Table 5 all the anomalies occur before the seventh week of gestation.

TABLE 5.

Latest date at which common anomalies in infants of diabetic mothers could occur

Anomaly	Gestational age after ovulation in weeks
Caudal regression	3
Situs inversus	4
Ureter duplex	5
Renal agenesis	5
Cardiac anomalies
T.G.A.	5
V.S.D.	6
A.S.D.	6
Anencephalus	4

Mills, Diabetes, 1979.

While infants of diabetic mothers are at risk for a wide variety of malformations, one syndrome seems particularly strongly associated with diabetes. Caudal regression is a condition in which agenesis or hypoplasia of the femorae occurs in conjunction with agenesis of the lower vertebrae (see Figs. 1 and 2). It is very rare in the general population. In Kucera’s (’71) review of the literature, he found 9 cases in 7,101 infants of diabetic mothers. Comparing the incidence with data from the World Health Organization, he concluded that the relative risk of caudal regression syndrome in infants of diabetic mothers was over 200. This estimate is unstable, yet it seems conservative in comparison with Pedersen’s (’77) relative risk estimate of 600 based on 4 cases in his series.

Figs. 1–2.

Infant with caudal regression syndrome: Sacral dysgenesis and shortening of the femorae. (Photographs courtesy of Lewis Holmes, M.D.).

As previously stated, the number of subjects available is too small to estimate the relative risk of most specific anomalies in the infant of the diabetic mother. Data from the Collaborative Perinatal Project (Mitchell et al., ’71) did, however, provide an estimate of relative risk of congenital heart disease. Infants of diabetic mothers had an incidence of 25.4 per 1000 compared to 8.1 per 1000 in infants of normal mothers (relative risk 3.1). Defects of the great vessels and septal defects accounted for virtually all cases. These findings are particularly important. They came from a prospective study where the special attention paid to diabetic pregnancies would not have biased the results, since all infants were screened for congenital heart disease.

While the relative risk of caudal regression is many times higher than the relative risk of cardiac anomalies, the magnitude of the problem is actually much greater for cardiac anomalies because their incidence rate is so much higher. One infant in 40 will have congenital heart disease, whereas, even using the highest estimate, only 1 in 350 will have caudal regression (Pedersen, ’77).

In summary, there is good evidence that maternal diabetes is associated with a wide range of malformations in the offspring. Multiple malformations are common in infants of diabetic mothers. While no single syndrome is specific for diabetes, caudal regression occurs far more often than in the general population.

POSSIBLE TERATOGENIC MECHANISMS IN DIABETES

The search for a teratogenic agent in diabetes is complicated by the fact that diabetes is not simply a disorder of carbohydrate regulation. Diabetes is responsible for a loss of normal homeostasis not only of carbohydrate but of fat and protein metabolism as well. Vascular complications may lead to additional metabolic changes such as hypoxia or impaired renal clearance of toxins. In short, there are multiple factors in the disordered milieu of the pregnant diabetic which could be teratogenic.

In addition to considering all the metabolic changes in diabetes, care must be taken to relate these events to the appropriate period during gestation. Unfortunately, most studies to date have not addressed this problem. Either because the investigators did not appreciate how early in pregnancy organogenesis occurs, or because the women did not come to medical attention early enough, few studies have examined diabetic control during the critical period. Because of this, most evidence on possible teratogenic agents comes from retrospective analysis or from animal studies.

A number of animal studies have suggested that hyperglycemia could be teratogenic. Mice made diabetic with alloxan produced offspring with abnormally high malformation rates (Watanabe and Ingalls, ’63). Rats given alloxan or streptozotocin to make them diabetic had three times as many malformed offspring as control animals (Deuchar, ’77). These studies could be criticized on the grounds that the alloxan or streptozotocin could have caused the malformations. However, Horii and associates (’66) reduced the increased malformation rate produced by alloxan in mice to baseline by treating the mice with insulin, strong evidence that it was the uncontrolled diabetes which was responsible. Since insulin corrects a number of metabolic abnormalities in diabetes, not just hyperglycemia, this experiment does not establish hyperglycemia as the teratogenic agent. It does indicate that appropriate treatment, including the prevention of hyperglycemia, might reduce the risk of congenital malformations.

Most human studies have compared hyperglycemia and other measures of diabetic control in the third trimester with malformations, an inappropriate comparison. None as yet has correlated blood sugar (or other metabolic measurements) during the first six weeks of pregnancy with malformations. Some have argued that hyperglycemia is teratogenic, since those in worse White Classes (who are more likely to have hyperglycemia) have more malformations. The flaw in this argument is that those in worse White Classes have more severe metabolic derangements of other types as well as more hyperglycemia. The most important human data on hyperglycemia and malformations come from recent reports of hemoglobin A_1c levels in early pregnancy. Since hemoglobin A_1c represents an integrated measure of blood glucose over the preceeding weeks, elevation of hemoglobin A_1c in the second or third month of pregnancy is a fair indicator of hyperglycemia during organogenesis.

Miller and co-workers (’81) examined 116 infants of insulin-dependent diabetic women, 15 with major anomalies. Hemoglobin A_1c levels drawn on the mothers before the fourteenth week of gestation were strongly correlated with malformation rates. The malformation rate in infants whose mothers’ hemoglobin A_1c had been 8.5 or less was 3.4%; the malformation rate in infants whose mothers’ hemoglobin A_1c had been over 8.5 was a remarkable 22.4%. A hemoglobin A_1c determination sometime before the fourteenth week of pregnancy hardly tells us all we would like to know about the diabetic state during organogenesis. Indeed, it may reflect hyperglycemic episodes occurring after the critical period. Nonetheless, it is an important first step toward relating diabetic events during organogenesis to malformations.

A second, more modest study also found an association between high hemoglobin A_1c and malformations. Leslie and associates (’78) found that three of five insulin-dependent diabetic mothers with elevated hemoglobin A_1c levels at presentation for prenatal care had malformed infants. None of 18 babies born to diabetic mothers with hemoglobin A_1c in the range reflecting good diabetic control had malformations. Unfortunately, this paper does not provide enough detail to relate the hemoglobin A_1c determinations to organogenesis. “Presentation” may have occurred far enough beyond the critical period that the hemoglobin A_1c measurements are not relevant.

Both of these studies offer suggestive evidence that diabetic control is important in preventing malformations. It is not, however, possible to say that hyperglycemia is responsible for the malformations associated with high hemoglobin A_1c. While elevated hemoglobin A_1c certainly indicates past hyperglycemia, it tells us nothing about other events which could also have been present, such as ketosis or rapid fluctuations in blood sugar level.

Hypoglycemia is another state which has been said to cause malformations. Malformations have been reported in animals when their mothers were made hypoglycemic during pregnancy either by insulin or fasting (Duraiswami, ’50; Smithberg and Runner, ’63; Kalter and Warkany, ’59). It appears that hypoglycemia rather than insulin was the likely cause, both because the same effect was produced by fasting and because the malformation rate was reduced when insulin and adequate food to prevent hypoglycemia were administered simultaneously (Hannah and Moore, ’71).

Data on hypoglycemia in pregnant human beings are scarce. Malformations have been reported in infants whose mothers received insulin-shock treatment during pregnancy (Wickes, ’54; Impastato et al., ’64). The number of cases is, however, very small. It is very likely that these cases represent nothing more than coincidental occurrence of malformations in women receiving shock therapy.

Weak evidence against hypoglycemia as a teratogen comes from two sources. Rowland and his colleagues (’73) obtained a history of hypoglycemia during pregnancy nearly four times as often from diabetic women with normal infants as from diabetic women whose offspring had congenital heart disease. Most of these women had not come to medical attention until after the critical period of gestation, and it is not clear whether questions about hypoglycemia were related to the period of organogenesis. Molsted-Pedersen et al. (’64) reported that only 8 of 55 mothers with malformed infants had noted insulin reactions in the first trimester. They concluded that hypoglycemia could not, therefore, be an important cause of malformations. Here again, it is not clear that questions about hypoglycemia were directed to the appropriate period. In any such retrospective study, patient recall is a serious problem; moreover, some hypoglycemic reactions (particularly during sleep) go totally unrecognized.

Since congenital malformations occur most frequently in infants of insulin-dependent diabetic women, it is not surprising that insulin has been proposed as a teratogen. Early evidence supporting this contention came from animal studies. Injecting chicken eggs (Landauer, ’45) with insulin caused “rumplessness,” a condition tantalizingly like caudal regression. It was soon discovered, however, that this was not unique to insulin. In fact, merely shaking the egg could produce the same result (Landauer, ’45). As noted previously, insulin produced malformations in other animals as well, but the role of insulin was put in doubt when it was found that if glucose was given simultaneously the malformation rate returned to normal. In addition, diabetic mice treated with insulin had fewer malformations than those whose diabetes remained untreated.

Developmental morphologic dating provided additional evidence that insulin is not responsible for malformations. The malformations, as noted previously, occur before the seventh week of gestation. Beta cells have been reported to differentiate in the fetus at 10.5 weeks (Like and Orci, ’72). Maternal insulin does not cross the placenta at 16 to 20 weeks (Adam et al., ’69) or at term (Kalhan et al., ’75). Unless maternal insulin can reach the embryo earlier in gestation, there is no insulin exposure during the critical period.

Many other abnormalities of the diabetic milieu have been mentioned as possible teratogens. The apparent association between White Class and malformation rates has generated several hypotheses. Women in White Classes D and F (those women with vascular disease) have the highest malformation rates, leading to speculation that vascular disease could be teratogenic. Whether it is vascular disease per se or some concomitant of it remains to be determined. Hypoxia, also associated with poor White Class and vascular disease, has been considered a possible teratogen. There are no data in humans to support or refute this theory.

Still other metabolic abnormalities created by diabetes could cause malformations. Maternal ketonuria has been associated with reduced I.Q. in infants of diabetic mothers (Churchill et al., ’69; Stehbens et al., ’77), raising speculation that ketones might interfere with normal development. Proteins may be glycosylated in the presence of high glucose concentration (as in hemoglobin A_1c). This could result in abnormal structural proteins in the fetus. Abnormal protein synthesis might also result from a lack of glucogenic amino acids or an overabundance of branched-chain amino acids as is seen in poorly controlled diabetes (Felig et al., ’70).

As anyone who has cared for diabetic patients well knows, diabetes changes much more than carbohydrate, insulin, fat, and protein metabolism. It has an enormous impact on the patient’s life style as well. Some of the changes in life style dictated by diabetes and its management ought to be considered in our discussion of teratogenesis. Diabetic women are advised to restrict their diets, generally in terms of caloric intake and free sugars. At the same time, they may be encouraged to use artificial sweeteners. They are often told to eat, sleep, and exercise on schedule. It is not difficult to imagine that some of these changes in life style might be teratogenic, nor is it difficult to see how complex the search for a teratogenic mechanism in diabetes could become given all these additional factors. Fortunately, one piece of evidence clarifies the situation. As noted previously, women with non-insulin-dependent diabetes do not produce more malformed infants than normal women. They do experience most of the changes in life style that insulin-dependent diabetics undergo. This suggests that the teratogenic effect of diabetes is not related to these changes; if it were, both groups should show a similar effect.

Finally, the possibility that diabetes-induced changes in other endocrine systems might influence fetal development needs to be considered. Jovanovic and co-workers (’80) were able to study diabetic women before the tenth week of gestation. Women whose diabetes was not well controlled had estradiol, prolactin, and human chorionic gonadotropin levels below the range associated with normal pregnancy. In their small sample, there were no major malformations in any of the offspring. To say for certain that these subtle endocrine changes are not deleterious will require a larger study. Jovanovic’s work shows that other endocrine organs may be affected by diabetes during pregnancy.

FUTURE INVESTIGATIONS

One important area which remains unexplored is early fetal losses in diabetic pregnancy. Is the early loss rate higher in diabetics than nondiabetics? If so, is the increase due to higher malformation rates? Conversely, could the rate be lower due to failure of diabetic women to abort malformed fetuses? If this is the case, it might explain the excess malformation rate among live births.

Both the question of early fetal losses and the question of teratogenic factors in diabetic pregnancies could best be addressed by a prospective controlled study beginning before pregnancy. Due to the rarity of malformations, a large sample size would be required. The National Institute of Child Health and Human Development is currently performing a collaborative study to examine the problems of malformations in infants of diabetic mothers. In addition, the study will attempt to assess the role of good diabetic control during organogenesis in reducing malformations.

Malformations associated with maternal diabetes mellitus present an unusual opportunity to the teratologist. If insulin can be excluded as the teratogen, some endogenous factor created by the diabetic state must then be teratogenic. What is even more remarkable is the possibility that appropriate medical management can eliminate the risk.