Critical periods in human fetal development

The human gestation period is approximately 40 weeks from the first day of the last menstrual period (38 weeks’ post-conception) and is conventionally divided into the first, second and third trimesters, each lasting 3 calendar months. Another method for classifying the stage of pregnancy is according to the stage of fetal development. This is a more useful approach when assessing the potential risks associated with drug use in pregnancy.

Principles of teratogenesis

In 1959, James Wilson, co-founder of The Teratology Society, proposed several principles of teratogenesis that have since been expanded and modified but remain fundamental in assessing whether a drug or chemical exposure during pregnancy is likely to be associated with reproductive or developmental toxicity. A basic understanding of these factors is essential to both the interpretation of preclinical (animal) reproductive toxicity studies and to enable accurate risk assessment in clinical practice. A subset of Wilson’s principles are discussed as follows.

Pharmacological effect

Pharmacological effects on the fetus are by far the most common drug effects during pregnancy, and the consequences are often minor and reversible compared to the idiosyncratic effects that can lead to major irreversible anomalies. Pharmacological effects are usually dose related and to some extent predictable. Drugs may adversely affect the fetus via effects on the maternal circulation or they may cross the placenta and exert a direct pharmacological effect on the fetus. Equally, drugs are sometimes administered to pregnant women in order to treat fetal disorders; for example, flecainide has been used to resolve fetal tachycardia.

The neonate can also be adversely affected by maternal drug therapy (see Table 47.1). It is generally only at birth that signs of fetal distress are observed due to in utero drug exposure or the effects of abrupt discontinuation of the maternal drug supply. The capacity of the neonate to eliminate drugs is reduced, and this can result in significant accumulation of some drugs, leading to toxicity. Neonatal withdrawal effects may require treatment.

Table 47.1 Examples of drugs with pharmacological effects on the fetus or neonate

(adapted from Schaefer et al., 2007)

Idiosyncratic drug effects in the fetus and neonate are possible but occur rarely compared with pharmacological effects.

Maternal pharmacokinetic changes

There are a number of maternal changes which occur during pregnancy and are summarised in Table 47.2.

Table 47.2 Summary of pharmacokinetic changes during pregnancy (adapted from Schaefer et al., 2007)

Distribution

The volume of distribution of drugs may be altered because of an increase of up to 50% in blood (plasma) volume and a 30% increase in cardiac output. Renal blood flow increases by up to 50% at the end of the first trimester and uterine blood flow increases and peaks at term (36–42 L/h). There is also a mean increase of 8 L in body water (60% to placenta, fetus and amniotic fluid and 40% to maternal tissues). As a consequence, there may be increased dosage requirements for some drugs to achieve the same therapeutic effect, provided these effects are not offset by other pharmacokinetic changes. Both the total plasma and the free-drug concentrations of phenytoin, carbamazepine and valproic acid decrease during pregnancy, but the free-drug fraction (ratio of free to total plasma concentration) may increase (Pavek et al., 2009).

Fetal-placental transfer

Most drugs diffuse easily across the placenta and thus enter the fetal circulation to some extent. In general, the ratio fetal:maternal drug concentration is less than 1. Drugs differ in the extent to which they bind to fetal and maternal proteins. For example, fetal and newborn plasma proteins appear to bind ampicillin and benzylpenicillin with less affinity and salicylates with greater affinity, than maternal proteins. Maternal albumin gradually decreases during pregnancy and fetal albumin concentrations increase, so different fetal: maternal albumin concentrations occur at different stages of pregnancy. The degree of protein binding of any drug is an important determinant of its movement across the placenta. Drugs which are highly protein bound tend to achieve higher maternal and lower fetal concentrations. Drugs with very large molecular weights such as insulin and heparin have negligible transfer. Lipophilic, un-ionised drugs cross the placenta more easily than polar drugs, and weakly basic drugs may become ‘trapped’ in the fetal circulation due to the slightly lower pH compared with maternal plasma. Some other factors such as enzymes or drug efflux transporters in the placenta may facilitate or restrict the transfer of a drug to the fetus.

Drug dosing in pregnancy

As a general principle, the dose of a drug given at any stage of pregnancy should be as low as possible to minimise potential toxic effects to the fetus. Drug therapy that is considered essential can be tapered to the lowest effective dose either before conception (ideally) or at the time the pregnancy is diagnosed. Where drug exposure during the third trimester is predicted to have an adverse effect on the neonate postpartum, consideration may be given to slowly reducing the dose towards term to minimise the risks to the baby. Such decisions are, however, not always straightforward. Recommendations to wean an expectant mother off antidepressants and antipsychotics to reduce the likelihood of neonatal withdrawal syndrome (characterised by jitteriness, altered muscle tone, poor feeding and irritability) and, in the case of the SSRIs, avoid the possible increased risk of persistent pulmonary hypertension of the newborn (PPHN) are now being challenged. Not only are there insufficient data to conclusively demonstrate neonatal benefit or an optimal time of weaning, but also the increased risk of psychiatric problems and relapse in the immediate postpartum period needs to be taken into account.

Many physiological changes occur during pregnancy which may affect the way the body handles drugs. Knowledge of these changes can allow some prediction of the impact on pharmacotherapy while remaining aware that there is variability in the extent of these changes during the pregnancy, and high inter-individual variability. The need for changes in dosages is influenced by whether the drug is excreted unchanged by the kidneys or which metabolic isoenzymes are involved in its elimination. Women taking drugs with enhanced clearance and for which a good correlation between plasma levels and therapeutic effect exist, should have their plasma concentrations closely monitored and dose adjusted to reduce the risk of suboptimal therapy, for example, phenytoin, carbamazepine, lithium and digoxin. Similarly, highly protein-bound drugs may require free-drug concentration monitoring. However, there is no clear guidance for adjusting doses during pregnancy, and for most drugs, the concentration of free drug, and therefore the effect of that drug, is unchanged.

Pregnancy itself can cause a temporary worsening or improvement of some diseases and in that way influence drug dosages.

Drug selection in pregnancy

Although there are few, if any, drugs for which safe use in pregnancy can be absolutely assured, only a handful of drugs in current clinical use have been conclusively shown to be teratogenic. In general, drugs that have been used extensively in pregnant women without apparent problems are recommended in preference to new drugs for which there is less experience of use. For example, methyldopa is used rarely to treat hypertension in the non-pregnant state but has historically been preferred in pregnancy because of a long history of safe use (Schaefer et al., 2007). However, older drugs may be less effective in terms of controlling maternal illness and are often associated with an increased side-effect risk profile.

In most cases, the decision as to whether to commence or continue with a medication in pregnancy will depend on the risk–benefit analysis for that specific mother–infant pair. A frequent error made by health professionals is to apply the U.S. Food and Drug Administration (FDA) pregnancy risk categories (A (no demonstrable risk), B, C, D and X (teratogenic agents that are considered to be completely contraindicated in pregnancy) when considering whether or not to prescribe a drug in pregnancy. It is now widely accepted that these categories are oversimplified and are of little practical help in a clinical setting. The FDA has proposed that the existing categories be replaced with more detailed information sheets containing a summary of the fetal risk and the additional maternal factors that need to be taken into consideration. Importantly, the need for a detailed section discussing the available data including observed human versus animal data, the study design, dose exposure, and reported congenital malformations and/or adverse events has been emphasised (see http://www.fda.gov for up-to-date information).

It is worth noting that standard literature sources often contain unhelpful information such as ‘do not use in pregnancy unless the benefits outweigh the risks’. This is understandable from a medico-legal point of view but offers little in terms of risk assessment. The primary literature is frequently inadequate because ethical and legal restraints mean that randomised controlled trials are rarely undertaken in pregnant women. Often, the only information that is available is confined to retrospective studies, voluntary reporting schemes and/or animal studies. The rate of anomalies in retrospective studies and voluntary reporting databases may be erroneously elevated due to preferential reporting of abnormal outcomes. Individual case reports are also difficult to interpret as the denominator of drug exposure is unknown, and an abnormal outcome may be coincidental to the drug exposure. More recently, prospective controlled trials have been utilised where the pregnancy outcomes of a defined cohort of women exposed to the drug are compared with outcomes of a matched control group. Complete follow-up of each pregnancy and post-natal monitoring is an essential feature of this type of investigation.

Pre-conception advice

Drug use during the first trimester, in particular, the embryonic stage, carries the greatest risk of malformation as this is when the fetal organs are being formed. Ideally, all unnecessary drug therapy should be discontinued prior to conception. However, inadvertent drug exposure frequently occurs, as approximately half of all pregnancies are unplanned. It is thus critical to make careful drug choices when prescribing for women of reproductive potential.

Women with chronic illnesses requiring drug treatment should be offered specialist counselling before conception, and the options explored to reduce or change drug therapy to a safer agent. Epilepsy is an example, in which, if continued drug treatment is necessary, attempts are made to stabilise treatment with a single drug at the lowest effective dose. It is also important to note that many pregnant women become less adherent to their drug therapy out of concern about possible harm to their infant. In many cases, such as asthma, inflammatory bowel disease, epilepsy, inadequate treatment of the underlying disease may be more detrimental to the mother–fetus pair than the drugs used to treat the condition. It is thus essential that women of reproductive potential are given clear and accurate information so that unrealistic fears about the risks to their baby do not result in unnecessary pregnancy termination or disease relapse.

All women planning a pregnancy should be offered general advice to minimise the risk of congenital anomalies. This includes avoidance of recreational drugs, ‘natural’ or herbal remedies, alcohol, smoking, vitamin A products, minimisation of caffeine consumption and beginning daily supplementation with at least 400 µcg of folic acid to reduce the risk of neural tube defects. It is recommended that the daily dose of folic acid be increased to around 4–5 mg daily in women who have epilepsy or who have had a previous child with a neural tube defect. Some infectious diseases may carry important fetal consequences if contracted during pregnancy. For example, rubella infection in the first 20 weeks of pregnancy is associated with an increased risk of miscarriage and a syndrome comprising problems such as deafness, cardiac defects and mental retardation in more than 20% of pregnancies. Women who lack immunity to rubella should be immunised prior to conception.

Post-conception advice

It is important to draw distinction between advice given to women pre-conceptually and that provided to a pregnant woman who has already been exposed to a drug. In the former setting, it may be recommended that an alternative preparation be considered or that a drug treatment be stopped where clinically appropriate. This advice often hinges on the lack of definitive safety data and does not automatically translate to exposure to that drug in pregnancy being an indication for discontinuing the drug, additional fetal monitoring or termination of the pregnancy on the basis of the exposure. Any change to the woman’s medication should be based on a careful and individual risk assessment and include a discussion with the woman to provide her with accurate up-to-date evidence-based advice. In many such cases, the woman can be reassured that a normal baby is the most likely outcome, or where appropriate be offered additional prenatal investigation to screen for congenital malformation where the risk to fetus is considered to be significant.

Teratology Information Services and Pregnancy Registries

It is difficult to keep up to date with the published literature. There is an increasing need for summary documents that include and critically appraise all available data, and which enable health care providers to have a balanced and informed discussion with patients regarding the risks and benefits of a certain therapy in pregnancy. This is evidenced by the current debate surrounding the teratogenic potential of various antidepressants with conflicting opinion even amongst experts in the field.

Teratology information services (TISs) have been established in several countries across the world and provide evidence-based, up-to-date information and individual case-based risk assessments. In addition to reviewing published literature on drugs, teratology services also have access to specialist online databases and discussion forums. A number routinely collect pregnancy outcome data on the women about whom they receive an enquiry, to enable surveillance for potential teratogens.

For some new drugs, pregnancy registries have been initiated that record all reported drug exposures and follow up the outcome of the pregnancy. These registries are cumulative and work on the basis that specific anomalies would be identified relatively quickly and that there will eventually be sufficient statistical power to detect the magnitude of any increased risk relative to the general population. These registries may be held by a TIS, or by independent groups with an interest in a defined area. The 2009 A/H1N1 influenza pandemic provided an example of teratology services across the globe responding to the urgent need for safety data by establishing registries on antiviral and pandemic vaccine exposure during the pandemic.

Transfer of drugs into breast milk

Most drugs pass into breast milk to some degree although transfer is usually low. The drug ‘dose’ ingested by the infant via breast milk only rarely causes adverse effects. Examples of adverse effects observed in breastfed infants exposed to medication via breast milk are given in Table 47.4, although not all of these are proven to be directly due to the drug ingested via breast milk.

Table 47.4 Adverse reactions reported in breastfed infants

Almost all drugs enter milk by passive diffusion of un-ionised, unbound drug through the lipid membranes of the alveolar cells of the breast, according to the pH partitioning theory. Several factors influence the rate and extent of passive diffusion. These include maternal plasma level, physiological differences between plasma and milk and the physicochemical properties of the drug. Differences in composition between blood and milk determine which physicochemical characteristics influence diffusion.

Milk differs from blood in that it has a lower pH (7.2 vs. 7.4), less buffering capacity and higher fat content. The following drug parameters affect the extent of transfer into milk:

The profile of a drug that passes minimally into milk would therefore be an acidic drug that is highly protein bound and has low to moderate lipophilicity, for example, most NSAIDs. In contrast, a weakly basic drug that has low plasma protein binding and is relatively lipophilic will achieve higher concentrations in the milk compartment, for example, sotalol.

In the first few days of life, large gaps exist between the alveolar cells. These permit enhanced passage of drugs into milk. By the end of the first week, these gaps close under the influence of prolactin (Lawrence and Lawrence, 2011). There is greater passage of drugs into colostrum (early milk) than in mature milk as the former contains more protein and less fat. There is also some variation in fat and protein content of milk between the beginning and end of a feed, but these changes have less influence on drug passage than the physicochemical properties of the drug.

Another method by which a drug may enter milk is by a pumping system whereby energy is used to effect transfer into milk. The most important example is iodides which pass into milk in high concentrations (Hale, 2010).

Milk to plasma concentration ratio

Several methods have been proposed to determine the amount of drug transferred to breast milk. The milk to plasma (M/P) ratio is often used as a measure of the extent of drug transfer into breast milk. It is usually obtained from case reports or small clinical studies and may be based on paired concentrations or full area under the concentration–time curve (AUC) analysis. M/P ratios that are based on a pair of milk and plasma samples collected simultaneously may be inaccurate as they assume that the concentrations of drug in milk and plasma are in parallel, which may not be the case. It is better to collect multiple samples of plasma and milk across a dosing interval or until the drug is cleared from both phases after a single dose, for determination of an M/P ratio based on the respective AUCs (M/P_AUC). Figure 47.1 demonstrates the markedly different estimates of M/P ratio that can be obtained via both sampling methods. The true M/P ratio may vary significantly during the same episode of breastfeeding.

Fig. 47.1 Diagrammatic representation of a drug concentration–time profile in the maternal plasma and milk phases after a single oral dose. The points a, b and c illustrate simultaneous sampling times from both phases. Sampling at ‘a’ would yield an estimated M/P value of 0.2, at ‘b’ a value of 1.0, and at ‘c’ a value of 9.3. All can be misleading. The true M/P ratio calculated by the AUC method is 1.2 when extrapolated to infinity.

If human-derived M/P ratios are lacking for a particular drug, it may be possible to predict the extent of transfer using known physicochemical properties, for example, pK_a, and a published predictive model (Atkinson and Begg, 1990; Begg et al., 1992). M/P ratios obtained from animal studies should not be used for clinical decision making, as they may not correlate well with human M/P ratios.

Studies in humans demonstrate that most drugs have an M/P ratio less than 1.0, with the range of reported ratios being from around 0.1 to 5.0. It is often thought that drugs with high M/P ratios (e.g. 5.0) are unsafe because the concentration in milk exceeds that in plasma, while those with low ratios (<1.0) are believed to be safe. This is not always the case as the M/P ratio often fails to correlate with the ‘dose’ of drug the infant ingests via milk. The amount of drug transferred into milk is principally determined by the maternal plasma level. Thus, even where the M/P ratio is high, if the maternal plasma level is low, drug transfer is still low. Therefore, the M/P ratio must never be used as the sole measure of drug safety in breastfeeding. However, it can be used to estimate the ‘dose’ ingested via milk, which is a better predictor of safety.

Calculating the infant ‘dose’ ingested via milk

Infant plasma drug levels are the most accurate indicator of drug exposure, but these are seldom available.

When using quantitative data from milk analyses, the most accurate estimation of the infant ‘dose’ is from studies in which the milk is collected over a complete dose interval at steady state and the total dose is calculated (Fig. 47.2). Unfortunately, these studies are seldom performed. Therefore, information must be obtained from less than ideal conditions.

Fig. 47.2 The cumulative dose received by the infant is plotted against time. The arrows represent the maternal dose times. This type of study is undertaken by the mother expressing all the milk, from both breasts, at usual feeding times. An aliquot of milk is taken and assayed for the drug. The volume of milk is measured and the total amount of drug at each time is calculated as the product of concentration and volume. This type of study must be undertaken at steady state.

If the M/P ratio is known from published studies, the likely infant dose (D_inf) can be calculated as follows, with some assumptions:

where Cp_mat is the average maternal plasma concentration. M/P_AUC is used in preference to a ratio based on paired concentrations when available, but this is seldom the case. The volume of milk ingested (V_milk) is not known but is generally assumed to be around 150 mL of milk per kilogram of body weight per day. The above equation simplifies if the actual milk concentration data are available:

The likely infant plasma drug concentration (Cp_inf) can be calculated by:

where F is oral availability and Cl_inf is the infant clearance. Unfortunately, neither F nor Cl_inf is known accurately for infants, so estimation of the likely steady-state average plasma drug concentration will be very approximate. Weight-adjusted Cl_inf values, that is, L/h/kg, are often significantly less than adult values in the early stages of life (Table 47.5).

Table 47.5 Approximate drug clearance by age as percentage of maternal value (Begg, 2000)

Given the difficulty in estimating infant plasma drug concentrations, the relative infant dose, for example, compared with a therapeutic infant dose, is often used as a surrogate of exposure. To give some basis for comparison, the likely infant dose from milk can be compared with an infant therapeutic dose. This is reasonable for drugs such as paracetamol that are usually administered to infants but is unsuitable for drugs such as antidepressants that are not. In the absence of a clearly defined range of infant doses, the weight-adjusted maternal dose expressed as a percentage (% dose) is widely used to indicate infant drug exposure.

For the great majority of drugs, this calculation yields infant doses in the order of 0.1–5.0% of the weight-adjusted maternal dose expressed as a percentage (% dose). It is generally thought that relative infant dose values of less than 10% of the maternal dose are probably safe. However, the inherent toxicity of the drug should be taken into account when using this figure.

To calculate the daily infant drug intake via milk, the standard milk intake of 150 mL/kg/day is multiplied by the concentration of the medication in milk:

Estimated daily infant intake = Drug concentration in milk (μcg/L) × 0.15/infant weight (kg)

Some authors use the peak concentration in milk to indicate the maximum infant intake.

Variability

To complicate matters further, there will be significant variability between and within individuals in the values used to estimate infant exposure (i.e. D_inf, F, Cl_inf, where D_inf is itself a function of the estimated parameters Cp_mat, M/P, volume of milk). Some of this variability will change over time due to developing organ function in the maturing baby and part will be unexplained variability. In addition to this pharmacokinetic variability, there will be variability in response of the infant to any given concentration of the drug. It is fortunate that most drugs seem to fall readily into safe (RID <10%) based on expected exposure. However, care should be taken when using these values to assess drug safety in lactation, when variability in the estimates of the parameters used may impact on the accuracy of prediction of their safety. This is especially true for those circumstances when initial estimates of these parameters are less precise, for example, in neonates.

Recently, attention has been focused on the possible role of pharmacogenetic factors in affecting the safety of breastfed infants exposed to drugs via milk (Madadi et al., 2009). Sedation (and one death) occurred in infants of mothers with rare genotypes of cytochrome P450 2D6 leading to ultra-rapid metabolism of codeine to morphine. The incidence of these genotypes varies amongst different populations. The overall percentage of Western Europeans with the CYP2D6 ultra-rapid metaboliser phenotype is 5.4% (Ingelman-Sundberg, 2005). Higher percentages have been reported in populations from northeast Africa and the Middle East.

Assessing the risk to the infant

Many factors must be considered when assessing the risk of maternal drug therapy to the breastfeeding infant (Box 47.2).

Inherent toxicity of the drug will be a main factor in determining infant safety. Thus, antineoplastic drugs, radionuclides and iodine containing compounds would be of concern. Multiple maternal therapy with drugs with similar side-effect profiles, for example, psychotropic drugs or anticonvulsants is likely to increase the risk to the infant. Oral bioavailability is an indicator of the drug’s ability to reach the systemic circulation after oral administration. Drugs with a low oral bioavailability are either poorly absorbed from the gastro-intestinal tract, broken down in the gut or undergo extensive ‘first pass’ metabolism in the liver before entering plasma.

The presence of active metabolites, for example, desmethyldiazepam, may prolong infant drug exposure and lead to drug accumulation, especially where drug clearance is low such as in the neonatal period. Similarly, drugs with long half-lives, for example, fluoxetine, may be problematic at this time. Drug clearance by the infant does not reach adult values until 6–7 months (Table 47.5). A premature infant of 30 weeks’ gestational age has a drug clearance value of about 10% of the maternal value. It is important to distinguish between gestational age and time after delivery. A 2-week-old infant born at 28 weeks will have a gestational age of 30 weeks.

The maternal drug regimen can affect infant risk. Single doses or short courses seldom present problems, whereas chronic therapy can be problematic. Topical or inhalation therapy usually results in much lower plasma drug levels and therefore lower passage into milk. Multiple maternal medications increase the risk to the infant.

Reducing risk to the breastfed infant

A number of strategies may be adopted to reduce the risk of drug-related side effects in the breastfed infant. One technique that has been recommended for reducing infant exposure is to give the maternal dose immediately after the infant has been fed with the aim of avoiding feeding at peak milk concentrations. However, this is often impractical, especially where young infants are feeding frequently up to 2 hourly. In addition, accurate data on times of peak levels in milk are often unavailable, and it cannot be assumed that times of peak milk levels mirror those in plasma. This technique should be used selectively, that is, where the drug has a short half-life and where peak and trough levels are predictable, for example, antibiotics, anaesthetics.

Where a single dose of a drug known to be hazardous is given to a breastfeeding mother, for example, a radiopharmaceutical, it will usually be possible to resume breastfeeding after a suitable washout period, calculated as five times the half-life. Where the half-life is very long, the washout period necessary to avoid hazardous exposure to the infant may exceed the period of sustainable lactation.

Breastfeeding mothers should be advised to avoid self-medication. Where drug use is clearly indicated, the lowest effective dose should be used for the shortest possible time. Use of topical therapy such as eye/nasal drops for hay fever would reduce drug exposure in comparison to oral antihistamines.

The maternal regimen should be simplified wherever possible. A review of therapy before delivery will help to reduce risks to the neonate. New drugs are best avoided if a therapeutic equivalent is available for which data on safe use in lactation exists. All infants exposed to drugs via breast milk should be monitored for any untoward effects. Measures to ensure the safety of the breastfed infant are summarised in Box 47.3. Some commonly used drugs thought to be safe to use in mothers of full-term healthy infants are listed in Table 47.6.

Table 47.6 Examples of commonly used drugs thought to be safe for use in breastfeeding mothers of full-term healthy infants^a

a This table is to be used as a guide only. Expert advice is required when the maternal dose is high, if the infant is premature, has renal or hepatic disease or G6PD deficiency.

Special situations

Drug effects on lactation

Drugs that affect dopamine activity are the main cause of effects on milk production, mainly mediated by effects on prolactin. Early postpartum use of oestrogens may reduce the volume of milk, but the effect is variable and depends on the dose and the individual response. Progestogen contraceptives are preferred.

Drugs may occasionally be used therapeutically for their effect on lactation. Dopamine agonists such as cabergoline decrease milk production, and this effect may be utilised, for example, after an infant death. Dopamine agonists should not be used routinely for relief of the symptoms of postpartum pain or engorgement which can be managed with simple analgesics or breast support. Dopamine antagonists such as domperidone may be used in cases of inadequate lactation which have not responded to first-line methods such as improved technique or milk expression by hand or pump.

Other drugs may affect lactation as an unwanted side effect, for example, diuretics. When these are used on a long-term basis, infant weight gain should be monitored.