Women in Clinical Trials of New Drugs -- A Change in Food and Drug Administration Policy

Responses to drugs are influenced by many factors, including age, sex, ethnic background, metabolic phenotype, body-fat content and distribution, and body size^12,13. The presence of diseases other than the one for which a study drug is being tested and the use of concomitant therapies are also relevant. Such factors, either singly or in combination, can influence a drug's pharmacokinetics (the concentration of the drug in the blood or other tissues over time) or its pharmacodynamics (the body's response to a given concentration of a drug). When such differences are recognized, adjustments in the dose or dose interval, choice of drug, monitoring procedures, or other aspects of drug administration can improve outcomes for patients.

Sex-related differences in pharmacokinetics or response have been identified for a number of drugs^14,15. Propranolol, for example, is metabolized more slowly in women than in men; it has been suggested that sex hormones regulate some of the enzymes that metabolize this drug¹⁶. The half-life of theophylline is significantly shorter in female nonsmokers and smokers than in male nonsmokers and smokers, presumably because of differences in hepatic metabolism¹⁷. Lower rates of clearance of acetaminophen,¹⁸ several benzodiazepines,^19,20 lidocaine,²¹ aspirin,²² ondansetron,²³ and mephobarbital²⁴ have also been described.

The most likely causes of differences in pharmacokinetics between women and men and among women are variations in body size and composition and the effects of hormones. The usually smaller body size of women and their higher body-fat content may influence the pharmacokinetics of drugs even if there are no differences in metabolism. For example, a smaller body size results in relatively higher blood concentrations after a given dose of ethanol²⁵. In addition, the higher body-fat content of women and their lower body-water volume contribute to higher blood alcohol concentrations. These distributional aspects of pharmacokinetics are magnified by metabolic differences. Recently, it has been demonstrated that the gastric mucosa of women elaborates less alcohol dehydrogenase than that of men²⁶.

The hormonal environment could affect both the pharmacokinetics of drugs and their pharmacodynamic effects. Four factors appear to be relevant to women: (1) the effects of levels of gonadotropins and circulating steroidal hormones, notably estradiol and progesterone, during the menstrual cycle; (2) the differences between premenopausal and postmenopausal women, including the effects of hormone-replacement therapy after menopause; (3) the effects of drastically different hormone levels during pregnancy and the metabolic consequences of pregnancy itself; and (4) the effects of steroidal contraceptives on the metabolism of drugs taken concomitantly and, conversely, the effects of other drugs on the efficacy of contraceptives.

An example of the influences of the varying levels of sex hormones during the menstrual cycle is insulin binding. In one study, insulin binding to monocytes and erythrocytes was higher in the follicular phase than in the luteal phase²⁷. There is an inverse relation between the binding of insulin to monocytes and levels of estradiol and progesterone. This correlation may result in an exacerbation of hyperglycemia during the luteal phase in some women with insulin-dependent diabetes mellitus²⁸.

The differences in hormonal patterns between premenopausal and postmenopausal women and the use of exogenous hormones may affect pharmacokinetics. For example, the half-life of prednisolone is significantly longer in young women taking oral contraceptives than in women of the same age who are not taking such agents. A similar increase also occurs in postmenopausal women who are receiving conjugated estrogens, as compared with women not taking hormones²⁹.

Approximately 10 million women in the United States are currently taking steroidal contraceptives, and the possibility that concomitant drug therapy could decrease the effectiveness of these contraceptives is of serious concern. Griseofulvin increases the hepatic metabolism of contraceptive steroids, thus lowering blood levels³⁰. Plasma contraceptive-hormone concentrations may also be lowered by broad-spectrum antibiotics such as tetracycline,³⁰ the antitubercular agent rifampin,³¹ and some anticonvulsant agents, including carbamazepine and phenytoin^30,32. Susceptible women may experience breakthrough bleeding or even become pregnant when given these agents in conjunction with oral contraceptives, especially the low-estrogen oral contraceptives commonly used today.

As specified in the FDA's 1977 “General Considerations for the Clinical Evaluation of Drugs,” drugs are tested in three phases before a sponsor submits the new drug application required for marketing approval^33,34. Phase 1 studies are the initial studies in humans and generally involve small numbers of healthy volunteers or patients treated over a short period of time. These studies assess individual tolerance of the drug and examine its metabolism and short-term pharmacokinetics. They may also provide preliminary pharmacologic information related to clinical effectiveness.

Phase 2 studies, which normally involve a few hundred patients, are the earliest controlled trials designed to demonstrate effectiveness and relative safety. During phase 3, the final testing phase before a marketing application is submitted to the FDA for review, as many as several thousand patients are studied. These studies provide additional evidence regarding safety and effectiveness, including data on long-term exposure; refine information on dose-response and concentration-response relations; and identify relatively rare adverse effects. The inclusion of a broad sample of the population in phase 3 trials and the examination of the data for differences in response make it possible to identify demographic, pathophysiologic, and other characteristics that affect patients' responses to the drug.

The 1977 guidelines stated that drugs should be studied in the population that would receive them and specifically stated that all age groups should be included. The guidelines were not explicit, however, about the need to study both sexes. Nevertheless, FDA surveys conducted in 1983 and 1988 found that, in general, both sexes had substantial representation in clinical trials conducted before FDA approval of drugs, in proportions that usually reflected the prevalence of the disease in the sex and age groups included in the trials³⁵ (and Temple R, FDA: personal communication). Women tended to predominate in studies of nonsteroidal antiinflammatory drugs, whereas men predominated in studies of coronary artery disease. Roughly equal numbers of men and women were included in trials of most antibiotics, antihistamines, and hypnotics. Despite adequate participation by both women and men, however, few analyses of the data were being conducted to detect possible differences in effectiveness or safety between men and women.

In an effort to stimulate the use of the collected data to learn about individual characteristics that affect the behavior of drugs, the FDA in 1988 specifically called for studies of whether safety and effectiveness were similar within population subgroups defined by such characteristics as sex, age, and race³⁶. Recent evaluations have shown that the requested analyses were not being carried out regularly. In consultation with the FDA, the General Accounting Office (GAO) reviewed the participation of women in phase II and III clinical trials of new drugs approved from 1988 through 1991³⁷. The GAO found that the recommended analyses were being carried out in only about 50 percent of the trials. Because the GAO survey included many applications submitted to the FDA before the 1988 guidelines were published, the FDA surveyed new drug applications submitted from June 1991 to July 1992. It found that safety data had been analyzed according to sex just 64 percent of the time and that data on effectiveness had been analyzed in this way just 54 percent of the time.

In the light of these findings, the FDA will review all new drug applications shortly after submission to ensure that they include appropriate analyses by sex. If such analyses are lacking, the FDA will call for their submission and may consider refusing to initiate review of the application if sex-specific analyses are not provided within a reasonable period (Temple R, FDA: personal communication).

In addition to reviewing new drug applications to see that analysis according to sex is included, the FDA is issuing new guidelines on the participation of women in drug evaluations³⁸. This document is similar in approach to one published in 1989 to ensure that elderly patients would be included in studies and evaluations³⁹. The guidelines urge that reasonable numbers of women be included in studies of new drugs. “Reasonable numbers” are not defined precisely; rather, the agency expects enough representation of both sexes so that significant differences can be detected. The guidelines stress the importance of assessing possible pharmacokinetic differences between women and men, either by formal studies or with use of pharmacokinetic screening.

Pharmacokinetic screening is an approach to assessing the full range of factors, such as demographic characteristics, underlying disease, and concomitant therapy, that can alter a drug's pharmacokinetics⁴⁰. It consists of obtaining a small number of steady-state blood-concentration measurements in most subjects in phase 2 and phase 3 trials and then analyzing them to detect relations between pharmacokinetics and particular characteristics of the subgroup, such as sex, age, renal or hepatic function, body size, muscle mass, and concomitant therapy. If the results suggest important differences, more formal pharmacokinetic studies can then be undertaken.

Few clinically important sex-related pharmacodynamic differences in clinical response have been documented up to now, and the guidelines do not call for separate clinical or pharmacodynamic studies in women in most cases. Instead, substantial representation of both sexes is expected in studies of safety and effectiveness, and the data should be examined for sex differences in the effectiveness, adverse-event rates, and dose response of drugs. If these analyses suggest differences between the sexes, or if the presence of such differences could be especially important, as in the case of drugs with a low therapeutic index, additional formal studies may be needed.

The FDA guidelines emphasize three pharmacokinetic issues: (1) the effects of the menstrual cycle and menopausal status on a drug's pharmacokinetics; (2) the effects of concomitant estrogen supplementation or use of systemic contraceptive agents, including both estrogen-progestin combinations and long-acting progesterones, on a drug's pharmacokinetics; and (3) the influence of a drug on the effectiveness of oral contraceptives.

Finally, the new guidelines recognize that although clinical or pharmacokinetic data collected from late phase 2 and phase 3 studies may provide evidence of differences between the sexes, these data may become available too late to affect the design and dose selection of the pivotal controlled trials. The FDA therefore encourages the inclusion of women in all age groups early in drug development. Thus, the agency no longer prohibits women with childbearing potential from participating in the earliest phases of most clinical trials.

The guidelines published in 1977³³ specifically stated that women with childbearing potential should be excluded from phase 1 and early phase 2 studies. Once some information about relative safety and effectiveness had been amassed in early phase 2 trials, and once preclinical data on teratogenicity and female fertility in animals had been obtained, women with childbearing potential (broadly defined as the “capacity” to become pregnant) could participate in later phase 2 and phase 3 studies.

In addition to premenopausal women who were sexually active and using no contraception, the exclusion applied to women who were unlikely to become pregnant, such as women using oral, injectable, or mechanical contraception, women whose partners had had vasectomies, women who were sexually inactive, and lesbians. The restriction arose, in part, from earlier discoveries of birth defects that followed exposure to specific drugs^41-43. It reflected the view that risk to the fetus was unacceptable in studies that were not intended to have important medical benefits for the subjects. It did not apply to women with life-threatening diseases. Thus, women with conditions such as cancer and, more recently, AIDS have been included in the earliest phases of drug trials, before the completion of animal-reproduction studies. In these situations, the potential risk to the fetus was balanced by the compelling possibility of prolonging the life of the mother.

In 1993, protecting the fetus from unanticipated exposure to potentially harmful drugs remains critically important, but the ban on women's participation in early clinical trials no longer seems reasonable for several reasons. First, there are notable scientific benefits to including women with childbearing potential in the early phases of trials. If important sex differences can be identified during phase 1 and early phase 2 studies, later phase 2 and phase 3 trials can be designed more suitably to further clinical understanding of the appropriate use of drugs in women.

Second, from an ethical perspective, the restriction on women with childbearing potential implies a lack of respect for their autonomy and decision-making capacity. The ethical principles articulated in the Belmont Report⁴⁴ -- respect for persons, beneficence, and justice -- as well as recent actions of the Congress and decisions of the Supreme Court suggest that women should have the right to make their own risk-benefit choices about their pregnancies. For example, the Pregnancy Discrimination Act, as interpreted by the Supreme Court in the landmark case of United Automobile Workers v. Johnson Controls, prohibits the blanket exclusion of pregnant women from jobs they are qualified to perform solely because working conditions pose potential risks to exposed fetuses^45,46. Although the purposes of clinical trials are manifestly different from the purposes of employment, the Court's emphasis on a woman's right to participate in decisions about fetal risk underscores the principles of autonomy and informed consent. Consistent with regulations issued in their present form by the FDA in 1981, subjects in clinical trials are expected to be fully informed in an unbiased manner about findings from animal-reproduction studies, to the extent that they have been completed, and to be reminded of the uncertainties inherent in experimental therapies⁴⁷.

Third, it is possible to reduce the risk of fetal exposure through protocol design. Since early clinical studies are typically of very short duration, often involving a single dose of medication, one approach is to administer a drug during or immediately following a woman's menstrual period, after a negative result from a pregnancy test that detects the beta subunit of the human chorionic gonadotropin molecule. For longer studies, trial subjects are expected to be counseled about the need to use reliable forms of contraception. Local institutional review boards will be expected to undertake careful reviews of investigational protocols that involve potential risks to the fetus from known or probable teratogens, in order to determine whether the trials should proceed. The FDA also reviews the risks and benefits of such protocols.

Whether removal of the impediments to their participation will increase the number of women in early trials depends partly on drug companies' concerns about liability. A review of case law suggests that manufacturers have not faced substantial litigation by clinical-trial participants⁴⁸. Liability litigation occurs mostly when an approved drug has been used in a population in whom it has not first been systematically studied⁴⁹. The future legal climate cannot be fully anticipated, however, and many states have ruled that once children reach the age of majority, those adversely affected by the medical decisions of a parent have the necessary cause of action to allow a lawsuit to proceed. If we are to achieve broader participation of women in all phases of clinical trials, legitimate issues such as liability will have to be addressed as part of ongoing dialogue among drug developers, scientists, policy makers, health advocates, and women's groups.

The FDA and the scientific community are concerned about the difficulty of testing drugs and biologic agents in pregnant women and the lack of systematic procedures for assessing postmarketing exposure. Maximizing protection of fetuses from potentially toxic therapies is prudent, and fear of liability is understandable, but the result is that many drugs are ultimately used during pregnancy without reliable data on their maternal and fetal effects. The potential risk of current practice was highlighted by the recent discovery, after the product was marketed, of sometimes fatal neonatal renal problems associated with pregnant women's use of angiotensin-converting-enzyme inhibitors during the second and third trimesters for the treatment of hypertension⁵⁰. This finding, which emerged from scattered clinical reports and the experience of a particular physician, underscores the need for a more formal mechanism for the pre-approval study of drugs that are likely to be used in pregnancy and for the systematic collection of postmarketing exposure data.

When a clinical trial represents the only source of a promising experimental therapy for a life-threatening condition, it is more obviously essential to include pregnant women. Thus, the agency has advocated that pregnant women who are infected with the human immunodeficiency virus be included in the early testing of new therapies for AIDS. Even in less urgent cases, the participation of women in formal studies may be appropriate when the drugs' use in pregnancy is likely. The FDA intends to explore further the complex issues of including pregnant women in clinical trials of new drugs and biologic agents and of improving the collection of postmarketing data in a series of public workshops and conferences similar to those that led to the policy changes highlighted in this article.

The members of the Working Group on Women in Clinical Trials are as follows: Grant Bagley, M.D., Ilisa Bernstein, Pharm.D., James Bilstad, M.D., Bruce Burlington, M.D., Linda Carter, Mary Ann Danello, Ph.D., Mary Gross, Jane Henney, M.D., Patricia M. Kuntze, Eileen Leonard, M.D., Catherine Lorraine, J.D., Thomas Ludden, Ph.D., Diane Murphy, M.D., Carl Peck, M.D., Mary Pendergast, J.D., Margaret Porter, J.D., Carol Scheman, Elyse Summers, J.D., Michael Taylor, J.D., Theresa Toigo, R.Ph., Gloria Troendle, M.D., Suzanne White, Janet Woodcock, M.D., and Nancy Yeates.