Prescribing in pregnancy - The effects of drugs on the fetus - technical
A drug can harm the fetus only if it crosses the placenta, but most drugs do. The placenta offers a lipid barrier to the transfer of drugs, and the rate at which a drug crosses from mother to baby will depend on its lipophilicity and polarity. However, with the exception of drugs administered acutely around the time of delivery, the rate of transfer is of little importance, and for any course of drug treatment it should be assumed that transfer will occur. The only notable exceptions are heparin—including low molecular weight preparations—and insulin.
Drugs can adversely affect the developing fetus in different ways depending on the gestation at which exposure occurs. For this reason it is appropriate to consider organogenesis, fetal growth and development, the breastfed infant, and childhood growth and development separately.
Prescribing in the first trimester
Organogenesis occurs between 18 and 55 days of gestation and it is during this time that drugs can cause anatomical defects. A drug can cause a teratogenic effect only if it is present in the embryo during organogenesis, and even a definite teratogen will not cause a structural defect if it is given following this period. These seemingly obvious statements become relevant in prepregnancy counselling and in providing advice when exposure to a possible teratogen has occurred during pregnancy.
Being present in the embryo during organogenesis is not necessarily synonymous with being prescribed during this period. The retinoids are stored in adipose tissue and released slowly, so a teratogenic effect can occur long after the course of treatment has been completed. It is important to recognize that teratogenic effects are not seen in all cases: on the contrary, most first-trimester exposures to teratogenic drugs will not harm the baby. Clearly there is more to drug-induced fetal abnormality than simply the drug: the genetic make-up of the baby is important too. Some drugs that are definitely teratogenic in humans, together with approximate risks, are listed in Table 1.
Anticonvulsants, and the difficulties of establishing good data on drug toxicity in pregnancy
Anticonvulsants have been extensively studied over the last 40 years and illustrate well some of the principles and pitfalls associated with producing robust information. Many of the studies have serious weaknesses, involving design, ascertainment, classification, and statistical power. However, the advent of epilepsy-pregnancy registers from the mid 1990s has improved the quality of data available. Some themes are remarkably consistent. Polytherapy is accompanied by a greater risk of fetal abnormality.
The United Kingdom Epilepsy and Pregnancy Register found that 6% of babies whose mothers had taken more than one anticonvulsant had a major abnormality, compared to 3.7% following monotherapy (except for valproate, which carried a risk of 6.2%) and 1 to 2% in the unaffected population. However, women taking several drugs are likely to have more severe disease, hence cause and effect is hard to establish, and interestingly the abnormality rate in epileptic women who had not taken any drugs was 3.5%, still higher than the general population.
This finding is consistent and has generated much debate about the relative roles of drug, disease and inheritance. Several studies have suggested that the teratogenic effects of valproate increase above 800 to 1000 mg daily, and lamotrigine may also show dose-dependent teratogenicity.
A further very important theme that runs through the consideration of anticonvulsant teratogenicity is that new drugs are often welcomed as being safer, until experience demands otherwise. Valproate was heralded as the drug of choice for women of child-bearing age in the early 1980s. Lamotrigine enjoyed similar claims but is now known to increase the risk of oral clefts.
|Table 1 Some commonly used drugs that are known to be teratogenic|
|Drug||Main abnormality||Approximate risk (%)|
|Sodium valproate||Neural tube; possible neurodevelopmental||6|
|Lamotrigine||Oral cleft; possibly others||3|
|Warfarin||Chondrodysplasia punctata||Up to 25|
|Lithium||Cardiac (Ebstein complex)||2|
|Danazol||Virilization of female fetus||Uncertain|
CNS, central nervous system.
Another aspect to the consideration of anticonvulsants in pregnancy is the extent to which a ‘fetal anticonvulsant syndrome’ exists. The term was first coined many years ago to describe a constellation of features including major malformations, microcephaly, hypoplasia of the midface and fingers, and growth retardation. More recently there have been several reports of dysmorphic features, particularly with valproate. There is, however, considerable overlap between the appearances claimed for individual drugs and indeed with those seen in the children of epileptic women who have not had in utero drug exposure.
Preventing drug-induced teratogenesis
It is difficult to prevent drug-induced teratogenesis, short of the obvious solution of not taking the drug. Where there is a risk of neural tube defect, folic acid 5 mg daily should be prescribed from the time that pregnancy is planned. There is no direct evidence to support this approach, but it is logical: folic acid is known to be effective in the secondary prevention of naturally occurring neural tube defect, and anticonvulsants lower folate levels.
The overall message—not just for anticonvulsants but for any drug treatment in pregnancy—is to use no drug if you can, but if you must then use the smallest number of drugs in the lowest possible doses.
Many of the abnormalities caused by drugs can be detected by detailed ultrasound scanning at 18 to 20 weeks gestation, but the defects caused by warfarin involve mainly soft tissue and do not fall into this category.
Table 1 (above) is not comprehensive and includes only those drugs commonly encountered in general medical practice. Some drugs used in specialist areas are teratogenic, e.g. several agents used in cancer chemotherapy. Many more drugs may be teratogenic in a small percentage of exposures, but definitive information is not available because both prediction and detection of human teratogens is difficult. Predicting the effect of a drug in humans usually depends on studying its pharmacology in experimental animals.
However, this is not fruitful in the area of teratogenesis because species variation is so great: e.g. thalidomide causes phocomelia only in primates, while lithium causes cardiac abnormalities in humans at doses that produce no effect in the rat. Detecting teratogenic effects is complicated by the normal occurrence of fetal abnormalities, hence if a drug is teratogenic very occasionally it can be exceedingly difficult to distinguish its effects from those arising naturally.
Balance of benefits and risks
Even if a drug is a teratogen, the balance of benefits and risks may still be in favour of its use. For example, chloroquine and proguanil are indicated for malarial prophylaxis in areas where Plasmodium falciparum remains sensitive. Currently available evidence suggests that chloroquine may cause a very small increase in birth defects: in one study 169 infants whose mothers took chloroquine base 300 mg once weekly were compared with 454 children whose mothers took no drug.
Abnormal babies were born to 1.2% of the treated group, compared to 0.9% of the controls: not a significant difference, but the study was too small to detect anything less than a fivefold increase in abnormality rate. By contrast to the possibility of this small increase in risk, malaria presents a major risk to the health and life of both mother and baby, particularly when an expatriate woman is travelling in an endemic area. The argument in favour of using prophylaxis is therefore overwhelming—but not so overwhelming as the advice for pregnant travellers to avoid malarial areas!
Similar arguments apply to corticosteroids, which have acquired a reputation for causing oral cleft defects. The evidence in support of this effect is at best conflicting and is easily outweighed by the benefits of steroids in conditions such as severe asthma, inflammatory bowel disease, systemic lupus erythematosus, or organ transplantation. The placenta inactivates around 90% of prednisolone, but corticosteroids such as betamethasone, which are used to accelerate fetal lung maturity, have much greater penetration to the fetus.
Prescribing later in pregnancy
Beyond organogenesis, the fetus undergoes growth and development. The scope for producing anatomical defects has largely passed, exceptions being premature closure of the ductus arteriosus caused by indometacin and bleeding into the fetal brain produced by warfarin. Growth and function tend to be the targets of drug adverse effects for the remainder of the pregnancy.
The possible effects of some commonly used drugs later in pregnancy are shown in Table 2.
|Table 2 Some drugs that can cause harm later in the pregnancy|
|Antithyroid drugs||Fetal hypothyroidism if used in excessive dose|
|Aspirin||Analgesic doses associated with neonatal bleeding; not seen with low-dose aspirin|
|β-Agonists||Pulmonary oedema can occur in management of preterm labour, particularly when combined with excessive fluids and/or corticosteroids|
|β-Blockers||Use throughout pregnancy associated with around 25% risk of intrauterine growth retardation; not seen with short-term use in third trimester|
|Benzodiazepines||Drug dependence in the fetus|
|Corticosteroids||Claims that these drugs cause intrauterine growth retardation, or suppress the fetal adrenal, are not supported by the available evidence|
|Heparin||Maternal osteoporosis: risk increases with dose and duration of exposure|
|Indomethacin||Multiple neonatal morbidity; premature closure of ductus|
|Phenytoin||Neonatal haemorrhage accompanied by low levels of vitamin K-dependent clotting factors|
|Tetracyclines||Tooth discoloration. No evidence of harm following limited exposure in first trimester|
|Warfarin||Fetal cerebral haemorrhage—can occur with therapeutic INR in the mother|
ACE, angiotensin-converting enzyme.
Drugs and breastfeeding
Most women who breastfeed their babies will take a drug during this time. Iron, mild analgesics, antibiotics, laxatives, and hypnotics are the most commonly used. Much work has been performed on the pharmacokinetic aspects of breastfeeding, but systematic studies on the effect of drug ingestion by the mother on her breastfed baby are lacking.
Milk consists of fat globules suspended in an aqueous solution of protein and nutrients. Drugs move from plasma to milk by passive diffusion of the unionized and non-protein-bound fraction. Since breast milk has a slightly lower pH than plasma, drugs that cross most extensively into breast milk are lipid-soluble, poorly protein-bound, weak bases. However, even for drugs that cross readily into breast milk, considerable dilution has already occurred in the mother. Thus, when the concentration of a drug in breast milk and the volume of the milk consumed by the baby are translated into a dose, it is often the case that the baby receives too little drug to have any detectable pharmacological effect.
Some of the more commonly used drugs that, on the basis of experience, have a good safety record in breastfeeding mothers are listed in Table 3. It will be seen from this that many of the drugs that would be indicated for common medical problems in this context are safe to use. However, some qualification is needed about two of the drugs listed in Table 3 (below). Oestrogen-containing oral contraceptives may suppress lactation if they are taken before the milk supply is well established, and in some women may do so even after this time: progestogen-only contraceptives do not influence lactation at any stage. Metronidazole is not harmful to the baby but is said to make the milk taste bitter and may therefore interfere with feeding.
|Table 3 Some drugs that have been used in breastfeeding women without evidence of harm to the baby|
|Carbimazole||High doses may suppress the neonatal thyroid|
|Propylthiouracil||High dose may suppress neonatal thyroid|
Some drugs have been shown to affect the baby when ingested in breast milk: these are listed in Table 4 (below). There are several other drugs for which theoretical risks exist, or for which isolated reports of serious adverse consequences have appeared. For example, aspirin is contraindicated in young children because of the possible association with Reye’s syndrome, and some authorities consider that the drug should therefore be avoided in women who are breastfeeding.
No evidence is available to support this view, but unless the use of aspirin is considered essential in a breastfeeding woman (and such an eventuality must be rare), then it is probably best avoided. Similarly, indometacin has been associated with one case of neonatal convulsion when used during lactation: a decision with regard to its appropriateness in any given patient would depend on the likelihood of real benefit accruing from its use.
The most obvious consequences of a drug-induced fetal abnormality occur at or shortly after birth in the form of anatomical defects, and studies in teratology have largely concentrated on immediate pregnancy outcome. However, drugs can, on occasion, cause problems that become manifest only after several years. The most striking example is diethylstilbestrol which, when given during early pregnancy, can lead to adenocarcinoma of the vagina in teenage offspring. In addition to late morphological effects, concern has been expressed that drugs given during pregnancy can influence behavioural development.
Several studies have claimed that the use of anticonvulsants during pregnancy is associated with impaired intellectual development of the children, but findings have been conflicting. It is difficult to carry out studies in this area and the choice of control group is crucially important. In general, all studies have suffered from small size and many from possible selection bias. When all children of treated epileptic mothers in a single hospital in Finland were studied prospectively, using the offspring of untreated epileptic women and age-matched children of the same social class as controls, no difference was found in intellectual development at the age of 5.5 years.
Similarly, children exposed to carbamazepine in utero have been found in two studies to be of normal intelligence. In contrast, there have been several reports suggesting that valproate causes neurodevelopmental delay. The absence of methodologically sound research makes it difficult to draw reliable conclusions.
|Table 4 Some drugs that may be harmful to the baby when used in breast-feeding women|
|Benzodiazepines||Lethargy and weight loss|
|Iodine||Risk of neonatal hypothyroidism|
|Lithium||Hypotonia, lethargy, cyanosis|
|Sulphonamides||Risk of kernicterus in preterm, ill, or stressed babies, but safe in healthy term infants|
One of the earliest trials on the treatment of hypertension during pregnancy involved a comparison of methyldopa with no treatment. The children underwent physical and psychomotor assessment at 4 and 7.5 years. The 4-year-old children from the treatment group had a slightly smaller head circumference than their untreated controls, but there were no other physical or psychomotor differences. The evaluation at 7.5 years revealed no differences between the two groups. The reputation of methyldopa as a safe drug in pregnancy is largely based on this very well-conducted study.
The effects on childhood development of atenolol vs placebo have similarly shown no detrimental effects, a wide range of physical and psychomotor tests being performed on the children at the age of 1 year.