Suggested Changes and Redesign of the Pregnancy Label for Human Drugs

November 12, 1997

J. DeGeorge, A. Ellis, J. Farrelly, E. Fisher, G. Fitzgerald, A. Jacobs, M. McNerney, L. Meyers, D. Morse, H. Sheevers, S. Williams

Introduction

This document describes an approach for pregnancy labeling of human drugs. This approach has been developed by an ad hoc group within The Center for Drug Evaluation and Research (CDER) of the FDA. This document does not represent official CDER opinion, but is presented to the docket in order to elicit discussion and comment and to encourage an environment of open dialogue on a prominent issue.

In brief, we propose that important experimental animal and human study data, the risks, and the management of drug use in pregnancy be compiled into a concise table at the beginning of the pregnancy section of the label. Following this table, a narrative will follow that discusses in greater detail the results and recommendations found in the table. This format serves the information needs of those who need to gather data quickly, as well as those who need as much information as possible to make a therapeutic decision.

The table is divided into three sections. The significant findings from experimental animal data and human study data or experience are described (section 1). These significant findings are then characterized by the extent of the risk posed to humans. The risks are based primarily on comparative exposure data and mechanistic considerations (section 2). The last section of the table (section 3) presents pregnancy risk management information for the use of a drug for an approved clinical indication(s). The section uses standardized language to express clinical implications for drug use during pregnancy. Because this "last" section represents the most important information needed for a therapeutic decision, the section is presented first in the table.

The narrative section expands upon sections 1 and 2 of the table. A description of the information to be included in the three sections of the table, as well as a discussion of findings to be included in the detailed narrative is given below. At the end of this document, we have provided a template of this proposed labeling for reproductive toxicity data.

Section 1

Experimental and Study Results from Animals and Humans

This section of the table summarizes important experimental findings that are scientifically robust and of potential clinical significance. It is intended to catalog reproductive and developmental hazards that have been identified in animal studies. If human data are available, they would also be included in this section of the table. Findings of more uncertain relevance may be included in the narrative section, accompanied by a discussion of study limitations. This section of the table gives a synopsis of experimental findings and is intended to catalog reproductive and developmental hazards that have been identified in animal studies. If human data are available, they would also be included in this section of the table. Only data that are scientifically robust and of potential clinical significance are conveyed in the table. Studies of lesser quality may be included in the narrative section with a discussion of the study limitations. If the available human data alone are adequate to fully characterize a particular reproductive hazard, those data may replace the animal findings. The animal findings could then be moved to accompanying text.

Findings are organized into two categories in the table, developmental and reproductive drug effects. Endpoints describing developmental toxicities are organized according to four generally accepted subcategories of manifestation: embryo fetal death, structural alteration, growth perturbation and functional deficits. Relevant positive animal developmental findings are listed in the table and scientifically compelling human data are also presented. A complete description of dose, exposure, and the period of drug administration (which may include postpartum dosing) that is associated with such findings is included in the accompanying textual narrative. Negative findings in each species are listed as "no response." Examples of endpoints in each subcategory are provided below.

1. Embryo-fetal death: Effects on the developing conceptus that are incompatible with survival, represented as spontaneous abortion or stillbirth in humans; or in resorptions, spontaneous abortions, or stillbirths in animals.

2. Structural Alteration: Malformations and significant variations of specific organs or organ systems should be included. Citations should follow the MARTA (Middle Atlantic Regional Teratology Association) nomenclature.

3. Irreversible Growth Retardation: Sublethal developmental toxicity may cause intrauterine growth retardation or retarded growth of specific organs. [Significant findings that are reversible would be discussed in the narrative, but may be factored into the evaluation of risk (e.g., if reversible growth retardation is seen in one species and fetal death or irreversible growth retardation is noted in a second species)].

4. Functional Deficit: Sublethal toxicity may be associated with functional abnormalities, including neurobehavioral and other peripheral organ system deficits, as well as postnatal cancer which may not be apparent as gross structural alterations.

Reproductive toxicities in adult humans or animals are organized into subcategories by effects on fertility (including gametogenesis), pregnancy and parturition, and lactation. Relevant positive findings are described for each species (including human) in which they are observed. Brief examples of endpoints in each subcategory are included below.

1. Fertility (including gametogenesis): Reduced fertility may be mediated through changes in count, morphology, motility and viability of sperm in males, and through follicular atresia, alterations to follicular development, failure to ovulate and alterations to the integrity of the corpus luteum in females. Changes in fertility are conveyed through specific indices in humans and animals (the pregnancy rate per thousand persons or the successful mating index in animal species.)

2. Pregnancy and parturition: Significant changes in the course and outcome of pregnancy and parturition should be presented. These may include clinical signs in the pregnant woman, extensions or reductions in the duration of pregnancy, and evidence of dystocia. In animals, these changes may be represented by changes such as death during parturition, changes in gestational period, or prolonged delivery.

3. Lactation: Derived from animal or human data, this section of the label should include effects on the quality and quantity of milk production, drug secretion into milk, extent of oral absorption and drug levels in the newborn (if known and relevant), and associated toxicities of concern. A full description of toxicities of concern should be presented in the text, including the animal or human data, standard toxicology findings, and pharmacology or reproductive toxicity study outcomes.

An example of this section of the table is presented below:

Section 2

Exposure and Characterization of Probable Risk

This section of the "Reproductive Toxicity" portion of a drug label characterizes the extent of risk associated with administration of the pharmaceutical to humans, based on the findings discussed above. All relevant human and animal data (see section 1) should be factored into the assessment of risk and determined for each of the major endpoints.

Comparisons of the exposures in animals having the toxicities described in Section 2, with exposure of the pharmaceutical measured in patients at relevant clinical doses, should be conducted separately for each indication that results in more than one exposure. Comparisons of exposure are based on pharmacokinetic (or toxicokinetic) parameters whenever available. When there is evidence that a particular parameter presents the best comparitor for risk, that parameter is used; otherwise AUC or Cmax (whichever results in the lowest multiple of systemic exposure) is preferred. For example, when data demonstrate that a reproductive toxic effect occurs at a lower daily dose when the drug is given once daily than if the same total dose is divided, Cmax data would be used in the comparison, although this may decrease the animal to human dose multiple compared to AUC data. Use of parent drug or metabolite data in the comparison should be based on information available about the activity of individual components. If adequate pharmacokinetic data are not available, comparisons of exposure should usually be made based upon doses normalized to body surface areas. Comparison of exposure based on a nominal dose should be justified. Risk categorization also should factor in the types of effects observed, the magnitude of the effects, and whether the mechanism that elicits the effect is potentially relevant to humans.

Categorical language is used to describe each reproductive risk for the drug. Proposed categories include:

1. Insufficient or no data

2. No apparent risk

3. Low risk

4. Significant risk

The stage of pregnancy at which the risk of an adverse effect occurs should also be described. (For example, ACE inhibitors would be placed in the "low risk" pregnancy category for first trimester exposure and "significant risk" for second and third trimester exposure.) Potential risks to humans based solely upon animal data are stated in a general fashion (e.g., "significant" or "low risk of developmental toxicity"). The category is followed by the exposure multiple above which the effect is observed, e.g., "Greater than 15X the human exposure at the therapeutic dose-equivalent on an AUC basis." If human risks are known based upon clinical data, these are stated specifically (i.e., " low risk of spina bifida" or "significant risk ofretinoid syndrome which is manifested by . . . "). The portion of the table dealing with exposure and risk characterization should include the sort of information presented in the table below.

In summary, this section of the label should characterize the risks to fertility, developmental or reproductive toxicity by integrating by the results from Section 1, human exposure data, and relevant pharmacokinetic, embryological, and mechanistic information . This section does not deal with risk/benefit considerations for specific indications for which the pharmaceutical is being administered (see section 3).

Section 3

Risk Management

The section on risk management is presented in the table first, because the section describes the clinical implications of drug use that are directly relevant to the health care practitioner and the patient who is pregnant or planning to become pregnant. The risk management section is written for each approved indication of a drug, and factors in the potential benefits of the therapy, the risk of failure to treat the disease, and the specific risks for pregnancy associated with use of the drug. It is a general guidance for these considerations and is not intended to replace the individual risk benefit decision in the physician-patient interaction. For example, use of an orally administered antifungal product may have different implications for women with systemic infections compared to localized nail fungal infections. Standardized language is used to describe the risk management consideration. The standardized language is intended to provide clear advice that is straightforward and easy to interpret, and allows for direct comparisons among various drugs for one indication. Further, this approach ensures consistency in the types of advice given for similar risk and use settings.

As described in sections 1 and 2 above, the data may be organized into two main types of experimental results (reproductive or developmental toxicity), and several basic categories of risk. (We have suggested insufficient or no data, no apparent risk, low risk, and significant risk for each type of finding in the experimental results section). The final section, risk management, will interpret the information in sections 1 and 2 for relevance to the clinical situation.

Four possible pregnancy situations can be described that require a decision about drug therapy when a potential hazard has been identified.

They include:

1. Inadvertently pregnant and taking a non-essential drug

2. Planning pregnancy and taking a non-essential drug

3. Inadvertently pregnant and taking an essential drug for therapy

4. Planning pregnancy and choosing an essential drug therapy to be used throughout the pregnancy

A drug is considered essential when used for conditions that are considered life-threatening or result in significant morbidity. Not treating a medical condition with an essential drug generally would pose a significant risk to the mother and/or the fetus; examples include drugs for the treatment of diabetes and epilepsy.

A drug is considered non-essential if it is for the treatment of conditions that are not life threatening and are associated with limited morbidity. A non-essential drug would usually provide no known benefit to the fetus and the benefit to the mother would be relatively minor, such as medication for acne or mild allergies.

These pregnancy/pharmaceutical use situations can be evaluated using data from the sections 1 and 2 (experimental results and risk) and may then be presented with standardized language. Two examples of the type of standard language that can be used are presented below.

Consider a drug that appears to present a significant risk for embryo-lethal effects and structural abnormalities, and no risk for any other developmental effects or effects on fertility. For an essential drug used by a woman planning pregnancy, the standardized language could state:

"Use of {drug name} during {indicate sensitive period or trimester, if known} pregnancy poses an increased risk to the structure of one or more essential fetal organ systems and is also associated with embryo-fetal death. Consider the use of other, safer available therapies. If {drug name} is used during pregnancy, precautions such as fetal monitoring should be considered. No additional precautions appear warranted with {drug name} use during delivery or lactation."

As a second example, consider a drug that appears to present no risk during fetal development, but causes prolonged and difficult labor. In this example, we have added the realistic possibility that a patient may be planning a pregnancy, or may become inadvertently pregnant while taking a drug. For an essential drug, the standardized language could state:

"Use of {drug name} during pregnancy or parturition can result in a more difficult pregnancy and/or delivery. Discontinue therapy if possible, and consider other, safer available therapies. No additional precautions appear necessary with use during early stages of pregnancy or during nursing."

We believe that a matrix of standardized language could be created that would supply the standardized language for virtually any pregnancy and drug use scenario. The standard language could be appropriately combined to describe any set of risks for the indications and pregnancy management.

Narrative Discussion of Tabulated Findings and Other Relevant Concerns for Drug Exposure
in Pregnancy, for Fertility, and in Nursing

The narrative includes a more complete discussion of the experimental results in humans (if available) and animals. All of the results on reproductive toxicity are discussed in the text. The text should give important background information as necessary and elaborate on the recommendations in the table (e.g., clarify when findings were observed only at maternally toxic drug doses.). The text should delineate the process that ended in the final risk characterization conclusions (e.g., "Because a finding was seen in two species and the exposure at which these effects occurred in y species was only x times the exposure in humans at the therapeutic dose, there appears to be a significant increased risk for structural malformations in humans").

Experimental and Study Results: As noted above, only data that are scientifically robust and of potential clinical significance are included in the table. The textual discussion; however, will first present the experimental and study results from the human and animal studies. Study designs, including dose, species, strain, timing of drug exposure from animal studies, and general methodology should be included. Important limitations of the studies should be noted. Studies not directly relevant to the clinical route but that provide information relevant to the risk consideration (such as data from a dermal study for an intranasal drug) should also be included. When useful, a discussion of the findings should be in this section. Maternal effects that are observed concomitantly with developmental toxicity toxicity should be identified (e.g., biomarkers of exposure). Additionally, suspected syndromes would be discussed here. The reversibility of the effects seen, their expected lethality, compatibility with life, and potential morbidity should be included in the text. Additionally, as human data are collected and become more robust than the animal data, the animal data might be removed from the table and be limited to a description in the narrative section of the label.

Characterization of Risk: The discussion should present the basis for the choice of risk category. The source of the comparative exposure data and the basis for the comparison used should be described. Sources of exposure data may be from various sources, such as pregnant humans, reproductive toxicity studies in pregnant or non-pregnant animals, single dose or repeat dose studies, etc. The basis for comparison of exposure, be it AUC, Cmax, or mg/m² on parent drug or metabolite data, may be discussed in this section. More detailed information on exposure data that support the category selection should be included. Again, study limitations, relevant study design information, and other important information not included in the table should be included here. The narrative discussion should weigh the seriousness of effects, the incidence of effects, mechanistic data that are used to consider the level of risk, and any issues of clinical relevancy used to include or dismiss findings (e.g., that the only species showing an effect is the one making a toxic metabolite not found in humans). Whether the effect is reversible or potentially correctable, occurred only at maternally toxic doses of the drug, and occurred in multiple species is included in the narrative discussion.

Risk Management: No additional description in the text is planned for this section.