Many types of exposures during pregnancy have been shown to be harmful to the exposed fetus (Table 1). Unfortunately, “new” teratogens, such as mycophenolate mofetil (Carey et al., 2009), continue to be identified. This underscores the importance of health care professionals being aware of the characteristics of a human teratogen, and the recognized teratogens, current controversies, and unanswered questions.
|Example: anticoagulant (warfarin), anticonvulsant drugs, e.g., valproate, phenobarbital, retinoic acid (Accutane) virilizing androgens|
|2. Maternal conditions|
|Examples: insulin-dependent diabetes mellitus; excessive alcohol intake; cigarette smoking|
|3. Intrauterine infections|
|Examples: toxoplasmosis; rubella; varicella|
|4. Heavy metal|
|Examples: mercury; lead|
|Example: cancer therapy (not diagnostic x-rays)|
|6. Procedures during pregnancy|
|Examples: chorionic villus sampling (CVS); dilation and curettage (D&C); intracytoplasmic sperm injection (ICSI)|
|Examples: heat; hypotension; gasoline sniffing (excessive)|
For this discussion, a teratogen is defined as an exposure in pregnancy that has a harmful effect on the fetus. To evaluate systematically the information available on an exposure in pregnancy, the characteristics of human teratogens should be considered. These have been developed by several basic scientists, clinicians, epidemiologists, and teratologists (Hill, 1965; Lenz, 1971; Wilson, 1973; Brent, 1978; Khoury et al., 1991; Shepard, 1994; Webster and Freeman, 2001).
Characteristics of a Human Teratogen
Significant increase in specific outcomes
There is a significant increase in a specific outcome, such as a malformation, growth restriction, or IQ deficit, in the exposed fetuses in comparison to the fetuses in an unexposed, comparison group.
Shepard (1994) suggested that the standard should be that the positive association has been consistent in “two or more epidemiologic studies of high quality”.
There should be a relationship between the “dose” or the amount of the exposure and the outcome. Three examples are: (1) the blood levels of phenylalanine in pregnant women with phenylketonuria, that is maternal phenylketonuria (Lenke and Levy, 1980); (2) the correlation of the IQ deficit in children with the number of cigarettes smoked during pregnancy (Fried et al., 1997); and (3) the dose of the anticonvulsant drug valproate taken by the pregnant woman (Omtzight et al., 1992; Mawer et al., 2002).
To illustrate the spectrum for the woman with maternal phenylketonuria, a blood level of over 1200 micromoles of phenylalanine was associated with a 92% frequency of mental retardation and 12% frequency of heart defects in comparison to 21% mental retardation and 0% heart defects, if the maternal level was 200 to 600 micromoles (Lenke and Levy, 1980).
A corollary to the dose-response relationship is the concept that, for any teratogen, there is a threshold of exposure to the fetus below which there is no harmful effect. While this concept is reasonable, this has not been confirmed with measurements of exposure levels for any medication shown to be teratogenic in human pregnancies.
One theoretical example of the threshold effect is the fact that exposure of the fetus through the mother's skin to the retinoic acid tretinoin (all-trans retinoic acid) in creams (Retin-A), is not teratogenic (De Wals et al., 1991; Shapiro et al., 1997), whereas exposure of the fetus to 13-cis-retinoic acid (Accutane, Roche Pharmaceuticals, Inc., Nutley, NJ), taken by mouth in the first trimester, has been associated with a 35% absolute rate of major malformations (Lammer et al., 1985). Both of these retinoic acid compounds are metabolized to all-trans-retinoic acid, which, at high enough levels in the human pregnancy, can be very harmful to the exposed fetus. The difference in the risks to the fetus is due to the low amount of absorption of tretinoin into the blood stream of the adult who has applied it to her/his skin (Clewell et al., 1997).
Period of greatest sensitivity
There will be a period during pregnancy when the exposure is more harmful to the exposed fetus. In general, exposures in the first trimester of pregnancy are associated with an increased rate of occurrence of malformations. By comparison, teratogenic exposures in the second and third trimesters of pregnancy could be associated with an increased risk for growth restriction and/or lowered intelligence of the exposed fetus.
Very specific periods of greatest sensitivity have been postulated for three teratogens: thalidomide, warfarin, and methotrexate. Based on his findings in interviews with the mothers of infants damaged by thalidomide, Lenz (1966) established the period between day 20 and 34 postfertilization as the period of greatest sensitivity to this exposure. The pattern of abnormalities in the exposed fetus differed depending on the specific days when the exposure occurred. Recent follow-up studies by Miller et al. (2009) confirmed this period of greatest sensitivity in thalidomide-exposed individuals, and extended the correlations to additional effects on cranial nerve function.
The period of greatest sensitivity for the anticoagulant warfarin, postulated by Hall, Pauli, and Wilson (1980), was based on a case series of 24 infants. These warfarin-damaged infants had been identified by several different clinicians. The periods of exposure for each affected infant were compiled and showed overlap in menstrual weeks 6 to 9 during pregnancy, which is 6 to 9 weeks after her last menstrual period. This common period of exposure was interpreted to be the period of greatest sensitivity of the fetus to warfarin or Coumadin (Bristol-Myers Squibb, New York). Based on this hypothesis, the pregnant woman who discontinues treatment with warfarin before the sixth week of gestation can prevent the occurrence of the warfarin embryopathy. There has been some clinical confirmation of this hypothesis (Iturbe-Alessio et al., 1986).
Feldkamp and Carey (1993) compiled the reported information on six infants with the methotrexate embryopathy. They postulated that the critical period of exposure in each was weeks 6 through 8 of gestation, that is 6 to 8 weeks after her last menstrual period.
It is very helpful in the identification of a human teratogen to have an animal model in which exposure by the same route, such as oral exposure, has produced a harmful effect on the exposed fetus. In an animal model, a wide range of doses can be used from a dose high enough to cause maternal toxicity to a dose low enough to have no apparent fetal effect.
The effects of very high dose exposures in animal models are usually not relevant to the much lower dose exposures in human pregnancies. Consider, for example, the medication acetazolamide (Diamox, Lederle, Philadelphia, PA). In susceptible strains of mice, a very high dose (1000 mg/kg maternal weight) injected intraperitoneally on day nine of gestation have produced a distinctive pattern of limb deficiencies (Holmes et al., 1988). That dose (1000 mg/kg maternal weight) is much, much higher than the recommended dose in a young woman: 250 to 1000 mg given orally each day. The total dose in an adult woman, weighing 60 kg, would be 500 to 1000 mg/60 kg, which is 16 mg/kg, whereas the amount (1000 mg per kg maternal weight) injected into the pregnant C57Bl/6J mouse weighing only 25 grams would be much higher. In the limited number of human pregnancies exposed to acetazolamide, there has been no increase in the frequency of major malformations.
The alleged teratogenic effect makes sense biologically
Teratogenicity from an exposure is often postulated in medical journals (or in the lay press) before a systematic assessment has been carried out with regard to its characteristics. In that systematic review, the question should be asked whether the alleged teratogenicity “makes sense biologically”. This characteristic was a major reason for doubting the alleged teratogenicity of the antinausea drug Bendectin (Merrell Laboratories, Cincinnati, Ohio) (Brent, 1995) and the herbicide benomyl (Bianchi et al., 1994). Ultimately, studies failed to show a biologic basis for the alleged fetal effects of either of these exposures.
There is a genetically more susceptible group of individuals
There are very few examples of genetic susceptibility to human teratogens, reflecting the limited number of studies and the lack of plausible hypotheses. One exception is the studies of the effect of maternal cigarette smoking on the birth weight of the exposed fetus (Wang et al., 2002). For example, the presence of two metabolic genes, the MSP1 polymorphism (Aa and aa) in the CYP1A1 gene and the deletion of the GSTT1 gene, was found in 741 mothers, 174 of whom smoked. When the smoking mothers were found to have the CYP1A1 (Aa) or (aa) genotypes and had the GSTT1 deletion, the birth weight of the exposed fetus was 467 grams less than the birth weight of unexposed infants (p < 0.001) among white infants and 475 grams less among smoking-exposed African American infants.
Unfortunately, most potentially teratogenic exposures during pregnancy have not been studied thoroughly for each of these six characteristics. This means that the clinician counseling a pregnant woman about her exposures must make value judgments as to whether there is definite evidence of potential teratogenicity. If systematic studies have not been carried out, the counselor must decide how to describe the risks of a harmful fetal effect. To prepare for that counseling session, the health care provider will rely on one or more information sources.
The possibilities include:
- 1Literature searches.
- 2Information obtained from postgraduate courses, symposia, or periodic guidance documents prepared by professional societies, such as the Technical Bulletins prepared by the American College of Obstetricians and Gynecologists.
- 3Attending the annual meeting of relevant organizations, such as the Teratology Society, David W. Smith Workshop on Morphogenesis and Malformations, the Society of Perinatal Epidemiology, the Society of Maternal Fetal Medicine, and the International Society of Pharmacoepidemiology.
- 4Reviewing counseling resources online.
- a)Reprotox (http://reprotox.org).
- b)Catalog of Teratogenic Agents. Shepard TH, Lemire RJ. Twelfth edition. The Johns Hopkins University Press, Baltimore, 2007.
- c)TERIS: Teratogenic effects of drugs in resource for clinicians (TERIS), Friedman JM, Polifka J. Second edition. The Johns Hopkins University Press, Baltimore, 2000.
- 5Reference books:
- a)Briggs GG, Freeman RK, Yaffe SJ. Drugs in Pregnancy and Lactation. Eighth edition. Lippincott Williams & Wilkins, Philadelphia, 2008.
- b)Schaefer C, Peters P, Miller RK. Drugs during pregnancy and lactation. Second edition. Elsevier, Amsterdam, 2007.
- 6Telephone-based counseling resources. For the health care provider who does not have the time required for reviewing these sources or who has concerns that these sources may not contain the most recent reports, he/she can consult telephone-based counseling programs:
Major Limitations in 2010
In consulting publications, attending conferences, and reviewing the database resources, it is clear that there are major deficiencies in the information available in 2010. These include:
- a)There is a lack of knowledge of the molecular basis for the harmful fetal effects of most teratogens. There are a few exceptions:
- b)Even when the molecular basis has been determined, the cellular basis for the distinctive fetal effects have not been delineated. For example, why does warfarin (Coumadin) cause dramatic hypoplasia of the nose? Another unanswered question is the mechanism by which valproate produces myelomeningocele and not anencephaly? Are there any anatomic differences between the myelomeningocele that is more common in infants of diabetic mothers in comparison to the defects associated with exposure to valproate? Why does exposure to phenytoin cause distinctive changes in the fingers and fingernails (Lu et al., 2000), but many fewer changes in the toes of the same individuals (Bokhari et al., 2002)?
- c)Epigenetic effects: the potential effects of human teratogens on epigenetic mechanisms should be explored. For example, assisted reproductive technology has been postulated to produce an increased frequency of malformations, in general, and errors in imprinting, such as Angelman Syndrome, in particular (Cox et al., 2002). Studies of infants with Angelman Syndrome have shown that 25% had an imprinting defect with silencing of the maternal UBE3A gene (Ludwig et al., 2005).How often do defects in imprinting occur in an in vitro fertilization pregnancy, in general, or in a pregnancy following the technique of intracytoplasmic sperm injection?
- d)Initial studies of human teratogens have often focused on the increased rate of occurrence of malformations or patterns of major and minor anomalies. Equally significant are effects on learning skills and intelligence, but these studies are labor intensive, if extended to an informative sample size. For any teratogen known to cause an increased frequency of malformations, an increased frequency of problems with learning or a deficit in intelligence should be ruled out with well-designed, controlled studies. These studies should include concern about the age of the child, the instruments used in the evaluation of cognitive function, matched controls, and an assessment of the level of education, occupation, and intelligence of each parent of both the drug-exposed and control children (Adams et al., 2006).For any drug taken during pregnancy, the counselor should discuss whether or not studies of the effect on learning skills and intelligence have been carried out.
- e)Airborne and dermal exposures are a common cause for concern, but few exposures have been studied. Studies are needed to establish the amount of exposure, the amount absorbed, and whether or not there are fetal effects in comparison to unexposed matched controls.
- f)Pregnancy registries need to be held to higher standards. Several pregnancy registries have been established at drug companies (Shields et al., 2004), at coordinating centers (Covington et al., 2004), and at medical centers (Morrow et al., 2006; Holmes et al., 2008) over the past 15 to 20 years. The published findings show a wide variety of methodologies and a lack of internal comparison groups in all, except one (Smith and Holmes, 2008). There should be a national advisory center on pregnancy registries that, with support from the Food and Drug Administration (FDA), would stress the need for:
- i)Each pregnancy registry should recruit an internal comparison group; one successful approach has been to recruit controls among the pregnant friends and family members of the enrolled women (Smith and Holmes, 2008).
- ii)Establish the inclusion and exclusion criteria for exposed and unexposed women to be used in evaluating the findings in the infants.
- iii)Establish the time window for identifying a malformation, such as “at birth”, “between birth and 12 weeks of age”, or “between birth and 1 year of age”. The baseline malformation rate will be different for each time window.
- iv)Continue the enrollment until at least 1000 exposed women have enrolled, so it will be possible to begin to determine the frequency of some specific common malformations among the exposed infants.
The convening of a workshop on pregnancy registries at the annual meeting of the Teratology Society would provide an opportunity for issues in methodology and the assessment of outcomes to be discussed by the staff members from many pregnancy registries.
Issues in Counseling
The discussions with pregnant women about potential fetal effects from exposures during pregnancy may be conducted in person, by e-mail, and by telephone. This process is similar in some ways to genetic counseling.
The setting should be private and conducive to discussion.
The counselor should be well-informed about the risk factors being discussed.
These discussions should use terminology that is well understood by the pregnant woman and her partner.
Visual aids should be used to facilitate communication.
A written summary should be sent to the patient; this letter improves communication with her, her relatives, friends, and her doctors and nurses.
This counseling would be more effective in person, as the communication would be better. However, currently, many concerned pregnant couples rely on communication by e-mail because they do not have access to counseling “in person”.
Counseling about exposures in pregnancy differs from genetic counseling in several respects.
The exposure of concern could be removed in some pregnancies.
The risk to the fetus from continuing exposure can be reduced by decreasing the exposure (e.g., the dose administered).
A less harmful medication could be substituted.
The woman for whom the drug of concern is essential treatment could have her eggs fertilized and implanted into a surrogate mother to avoid the exposure.
Counseling about exposures in pregnancy is limited by several factors.
The inaccuracy of the drug categories A, B, C, D, and X remains a major cause of incorrect counseling. Unfortunately, these drug categories are often used to identify potentially harmful exposures by busy physicians, nurses, midwives, and other health professionals. Most health care professionals do not realize that systematic analyses (Friedman et al., 1990; Lo and Friedman, 2002) have shown that these categories are simply not accurate. Education campaigns among physicians, nurses, nurse midwives, and pharmacists are needed to alert them to this issue. When reminding health care providers of the inaccuracy of the drug categories, the plans by the FDA for new narrative summaries could be described (Kweder, 2008).
Common Exposures with Inconclusive Study Results
Example #1: SSRIs, including paroxetine, fluoxetine, sertraline, citalopram, and others. The first reports of potential teratogenicity were presented on the website of the manufacturer GlaxoSmithKline. This report has stimulated several studies and analyses. The postulated effects from exposure in the first trimester of pregnancy have included:
- a)an increased frequency of malformations, including omphalocele, craniosynostosis, anencephaly, and heart defects (Alwan et al., 2007); and
- b)an increased frequency of the rare, but potentially severe, right ventricle outflow tract obstruction (Louik et al., 2007).
These two sets of different findings in two large, well-designed case-control studies illustrate the uncertainty. Both studies included many exposed and unexposed infants, but their findings were quite different. Because of this, the possible increased risk of associated malformations, in general, or specific malformations, in particular, has not been determined.
There are several important issues to be resolved, as enumerated by (Scialli, 2010):
Judging a group of drugs has not been productive. Data are needed on each SSRI separately.
It is notable that the heart defects reported in the original report of the manufacturer were primarily ventricular septal defects and atrial septal defects. How many were muscular ventricular septal defects, which could be considered a transient finding, and not a malformation (Roguin et al., 1995)? How many of the atrial septal defects were tiny openings identified by echocardiogram and were not significant clinically?
Is there a dose-response relationship?
No animal model has been established that has shown an effect on specific organs, such as the developing heart. In an animal model, doses higher than those used in human pregnancies will be used. These studies would be expected to show a dose-response correlation with the occurrence of malformations, in general, and heart defects, in particular. Those findings have not been reported.
Many human teratogens produce a pattern of multiple anomalies in the exposed fetus. It is notable that no pattern of anomalies has been reported from systematic, controlled studies in SSRI-exposed infants.
Is there biologic plausibility for the alleged fetal effects of SSRIs?
There is more of a consensus that there is an association between exposure in the third trimester to SSRIs and an increased risk for the exposed infant to develop a “neonatal withdrawal syndrome” (Sanz et al., 2005). More information is needed on the risk for specific SSRIs and to determine whether there is a dose-response relationship for this outcome.
Another postulated fetal effect from exposure to SSRIs in the third trimester of pregnancy is an increased risk for the rare, but very serious, condition known as persistent pulmonary hypertension of the newborn (Chambers et al., 2006a). Confirmation of this association is needed from other studies.
Example #2: Exposure during pregnancy to the plasticizer compounds, the phthalates, have been shown in experimental animals (Parks et al., 2000) and in newborn infants (Swan et al., 2005) to be associated with significant effects on the male fetus. At birth, there was a significant correlation between urine levels of phthalates during pregnancy and genital changes in exposed male infants, including a decreased anogenital distance and decreased size of the penis (Swan et al., 2005). The significant correlations were with four phthalate metabolites: monoethyl phthalate, mono-n-butyl phthalate, monobenzyl phthalate, and monoisobutyl phthalate. These compounds and their metabolites are present in baby care products, which makes it possible that the exposure is widespread (Sathyanarayana et al., 2008).
More information is needed to determine: (1) the safe levels of exposure to specific phthalates in drinking water during pregnancy; (2) the practical guidelines for how to avoid exposure; (3) whether there is a dose-response relationship between the level of exposure to specific phthalates during pregnancy and the physical features of the exposed male infant; and (4) if there are any risks for malformations in the exposed female fetus.
The social ramification of these concerns about phthalates should be recognized. Sometimes the fear among pregnant women and the parents of young children is much greater than is warranted by the information available.
Example #3: The antagonists to tumor necrosis factor-alpha (TNF-α) are used to treat chronic diseases, such as psoriasis, Crohn's disease, and rheumatoid arthritis. A case report of exposure to the TNF-α inhibitor etanercept throughout pregnancy described an infant with multiple anomalies (Carter et al., 2009). A well-designed cohort study is needed. The possible teratogenicity of TNF-α antagonists is being evaluated through the OTIS network of collaborating counseling centers (Chambers et al., 2006b). This approach could provide the first definitive evidence as to whether or not the infants exposed in utero to TNF-α inhibitors have a significant risk for malformations.
Limitations of Trying to Identify a “New” Teratogen from the Food and Drug Administration's Adverse Event Reports
The Adverse Event Reporting System at the FDA was established to make it possible to detect, as early as possible, a dramatic increase in the occurrence of severe malformations. It could: “detect the next thalidomide”. To date, the analysis of the exposures and associated malformations in the Adverse Event Reports has not identified a “new” human teratogen.
At least two have been postulated:
- a)that exposure to Retin-A produces holoprosencephaly (Rosa et al., 1994);
- b)that exposure to the statins produces a recognizable pattern of malformations consistent with the VACTERL Association of vertebral anomalies (Edison and Muenke, 2004).
It seems likely that the lethal and most handicapping malformations will be overrepresented among the cases described in the Adverse Event Reports. Because these are a series of case reports, the only associations that will be detected from these reports will be distinctive, rare, and severe phenotypes. For example, this makes it more likely that the phenotype associated with mycophenolate mofetil, an immunosuppressant agent, would be detected. That phenotype has included microtia (bilateral), cleft lip and palate, coloboma of the iris, microphthalmia, and heart defects, in varying combinations (Sifontis et al., 2006; Carey et al., 2009). By contrast, the alleged association of a doubling of the frequency of cleft lip or cleft palate in women who smoke cigarettes is not likely to be identified in the Adverse Event Reports. It is more likely that the infant with an isolated cleft lip or cleft palate would not be reported.
Being on the “Horns of a Dilemma”
Example #1: Consider the pregnant woman who is taking 1500 mg per day of valproate to prevent seizures from her underlying epilepsy and learns at 12 weeks of gestation that this exposure could be very harmful to the exposed fetus. The issues include:
- i)About 10% of the valproate-exposed fetuses have major malformations (Wyszynski et al., 2005). But, these malformations could be present but not visible in the prenatal screening by ultrasound.
- ii)The woman could resist changing the medication during pregnancy, as she has found that this medication is the only one which has been able to control her seizures.
- iii)Treating neurologists have found it very difficult to change medication during pregnancy. Yet, avoidance of this exposure could lessen significantly the exposure to valproate in the second and third trimesters of pregnancy and could lower the risk for developmental delay and possibly autism.
Example #2: One characteristic of a human teratogen is that there is a significant increase in the occurrence of specific outcome(s) among exposed infants. It has been suggested by Shepard (1994), that for the increase to be considered definite, it should have been observed in at least two different, large well-designed studies.
That philosophy applies to the recent report (Crider et al., 2009) that infants exposed during pregnancy to either sulfonamide or nitrofurantoin had a significant increase in the frequency of several different major malformations. The data are from the case-control study organized by the Centers for Disease Control which include many well-characterized infants with malformations and a large comparison group. For example, exposure to sulfonamides was associated with the occurrence of anencephaly (adjusted odds ratio [OR], 3.4; 95% confidence interval [CI], 1.3–8.8). Exposure to nitrofurantoins was associated with an increased risk for hypoplastic left heart syndrome (OR, 4.2; 95% CI, 1.9–9.1). Are these findings correct? As these drugs have previously been found in smaller studies not to be teratogenic, these findings were a surprise. These drugs are used frequently in pregnant women. Hopefully, additional studies are underway now.
Impact of Litigation
Reports in the lay press of the possibility that an exposure that may cause fetal damage can encourage the parents of an exposed, malformed child to sue the physician who prescribed the medication. The potential sad consequences for the family from litigation have been described by Brent (1977).
This decision to sue the prescribing doctor may reflect anger and frustration, rather than conclusions based on a careful review of the scientific literature. The fact that the scientific information available is not definitive and that major questions of causality have been raised does not mean that lawsuits will not be filed. The news reports of multimillion dollar settlements against the manufacturer attracts the interest of both attorneys and the parents of a malformed infant who was exposed to the same drug during pregnancy.
In the midst of the current uncertainty as to the potential teratogenicity of the SSRI paroxetine, advertisements by attorneys recruiting families have been televised. Meanwhile, the busy clinician, investigator, and counselor can hope that the limitations in the information available will be addressed in on-going, well-designed research studies.
The challenge for clinicians and scientists interested in the identification and study of human teratogens is to educate, educate, and educate. Too few physicians, nurses, and nurse-midwives in training and in practice have been taught about this subject. Too many continue to provide counseling to pregnant women based on inaccurate information, such as the use of the drug categories to estimate risks from exposures to medication during pregnancy.
Part of the solution is to encourage more lectures to students in training. Workshops on “human teratogens” can be encouraged at the annual meetings of busy clinicians.
Raising the level of awareness among health care providers will have a significant impact on health care for pregnant women and their unborn infants.