Methodological Challenges in Biomedical HIV Prevention Trials (2008)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Design Considerations: Pregnancy M any late-stage biomedical HIV prevention trials are conducted among sexually active women of reproductive age in areas with high fertility rates. Despite intensive counseling on family plan- ning, and provision of or access to contraceptives, a large percentage of women enrolled in biomedical HIV prevention trials become pregnant. Tri- als testing new products and devices (or new indications of existing drugs) restrict pregnant women from enrolling, and typically take women who become pregnant during the trial off the product, either permanently or for the duration of their pregnancy. High pregnancy rates and the product use implications for women who become pregnant have important implications for the design, conduct, and generalizability of biomedical HIV prevention trials, including loss in trial power when the occurrence of pregnancies has not been adequately accounted for when planning the size and duration of the trial, and potential interpretational problems when women who become pregnant are taken off product. Beyond the loss of study power and generalizability of results that the occurrence and handling of pregnancies can produce, removal of product in trial participants that become pregnant raises questions about the real-world use of approved products during pregnancy, because safety and efficacy data for pregnant women will not be available. That concern challenges the assumption that pregnant women should always be taken off the study product. The committee therefore considers variations on trial designs that may allow some pregnant women to remain on certain products to enable investigators to collect valuable information on their safety and efficacy. 104 

DESIGN CONSIDERATIONS: PREGNANCY 105 WHEN PREGNANCY OCCURS DURING TRIALS Women involved in clinical trials of any kind where the potential teratogenicity—or likelihood of birth defects—of the agent is unknown are usually counseled to employ at least two birth control methods to prevent pregnancy. As noted, studies of biomedical HIV interventions promote con- doms and provide them to participants, primarily to protect them from HIV infection, but also to prevent pregnancy while they are exposed to an inves- tigational drug or device whose risks to the fetus are unknown. Biomedical HIV prevention trials involving sexually active women of childbearing age also usually make other forms of contraception available, such as hormonal patches or injections, oral contraceptives, and to a lesser extent, intrauter- ine devices, through onsite provision of contraception or referrals to local family planning clinics. However, women are often unable to negotiate condom use with their sexual partners, and adherence to other contraceptive methods is less than universal (Raymond et al., 2007). Thus a significant number of women may become pregnant during a trial. Microbicide trials in West Africa observed pregnancy rates ranging from 32 to 76 per 100-person-years at sites in Ghana and Nigeria (Macqueen et al., 2007). Because most trials conduct frequent pregnancy testing with highly sensitive tests, these rates may over- estimate true pregnancy rates because they detect chemical pregnancies that never result in clinical pregnancies. Clinical pregnancy rates are nevertheless extremely high among participants in HIV prevention trials, even when reported condom use or oral contraceptive use is high (Raymond et al., 2007). Condoms are generally insufficient as the only contraceptive method for sexually active women with a high frequency of coital acts to result in a low pregnancy rate (Raymond et al., 2007). For example, Skoler et al. (2006) estimate that the 12-month cumulative probability that a woman engaging in 20 coital acts per month will become pregnant—given a 90 percent rate of condom use and no other contraceptive—is 51 percent. High pregnancy rates among participants in such trials reflect high background pregnancy rates in populations to which these products will ultimately be targeted. That is, pregnancy is not an incidental occurrence in HIV prevention trials, but rather a predictably common event. Because an approved prevention product will, on introduction into a community, be used by many women even after they become pregnant, it is important that trials obtain information on the safety and efficacy of a product when used during pregnancy. When an efficacy trial does not obtain such information, it is important that additional studies do so. Regulatory agencies and sponsors have required that women who become pregnant during trials of new products, including drugs and devices, without established benefit in humans discontinue the study product. Some 

106 METHODOLOGICAL CHALLENGES IN HIV PREVENTION TRIALS trials allow women to go back on the product once they are no longer preg- nant, provided they are not breast-feeding, while other trials require that pregnant women permanently discontinue use of the product. However, the occurrence of pregnancy during a trial, and the requirement that women discontinue use of a study product upon pregnancy diagnosis, have major statistical and ethical implications for the trial. Statistical Implications of PregnancIES occurring during Trials As discussed in Chapter 9, study results can be seriously biased if women in a trial who become pregnant are no longer followed for HIV infection. Thus, it is important that investigators continue to follow women who may discontinue their product owing to pregnancy, or any other rea- son, and use this information in analyzing trial results. As Chapter 9 also notes, discontinuing product use among pregnant women can diminish a study’s power to detect a beneficial product by attenuating the intervention effect. However, when investigators can esti- mate the pregnancy rate before the trial begins, they can increase the sample size or duration of follow-up of participants to compensate for this loss (see Chapter 2 on sample size). Data monitoring committees (discussed in Chapter 9) can also monitor actual pregnancy rates during a trial, and adjust its sample size and duration if these exceed expectations. Of course, this strategy may come at a significant cost in time (for events-driven trials) and money. Discontinuing product use among pregnant women also complicates the interpretation of trial results because agents that are proven efficacious will be used by women who will become pregnant. Thus, if the relative efficacy of the product in pregnant women differs from that in women who are not pregnant, or if the product affects the fetus, the results of the trial would not be representative of the impact if the product were introduced into the community. This concern makes it imperative that investigators gather information regarding product safety and efficacy as soon as possible. If they cannot do this during the registrational trial for licensing a product, then the product development plan devised before the trial should specify how investigators can reliably obtain such information after the trial.   ore frequent testing reduces the amount of time that a woman takes a product after she M becomes pregnant. Yet very early and very frequent pregnancy testing using a highly sensitive method may detect many “chemical” pregnancies—those that will not progress. Although women are required to terminate product use, many can resume product use shortly thereafter because of the short duration of chemical pregnancies. 

DESIGN CONSIDERATIONS: PREGNANCY 107 Recommendation 4-1: Investigators should take several steps to mini- mize the loss of study power and potential biases in results that can occur when women become pregnant during a trial: • Before the start of the trial, investigators should attempt to accu- rately estimate the rate of pregnancy that will occur during participant follow-up, and use these estimates in calculating sample size and trial duration. • Data Monitoring Committees should monitor actual pregnancy rates during the trial, and recommend appropriate adjustments to sam- ple size and trial duration if these rates exceed expectations. • Investigators should continue to follow all women who become pregnant for HIV infection, regardless of whether they discontinue the study product. COLLECTING INFORMATION ON BENEFITS AND RISKS Ultimately, from a public health perspective, the goal is to develop products that people at high risk of HIV acquisition can use. And in most countries, women of childbearing age, with high background rates of preg- nancy, are one of the populations at highest risk. As noted, many women who become pregnant are likely to use a product with demonstrated effi- cacy in preventing HIV infection, regardless of recommendations on the product label. Therefore, investigators and sponsors are obligated to collect as much information, as early as possible, about potential risks and benefits of the product to pregnant women and their fetuses. As discussed later, relying only on posttrial pregnancy registries to col- lect information about the safety of a product in pregnant women is insuf- ficient, and perhaps unethical, in a case where a high proportion of likely users will become pregnant, and where the epidemiologic infrastructure is inadequate to support such registries. This raises important questions about the best ways to collect safety information, and whether there are cir- cumstances under which it would be ethical to allow women who become pregnant to continue receiving the study product. Historical Context In evaluating whether there are circumstances when women who become pregnant during a trial should have the option of remaining on product, it is important to understand the historical context, legal trans- formations, and ethical considerations that have shaped current policy and practice. (See Box 4-1 for more detail.) Some groups and ethicists have suggested allowing pregnant women 

108 METHODOLOGICAL CHALLENGES IN HIV PREVENTION TRIALS BOX 4-1 Historical Perspectives on Including Pregnant Women in Clinical Trials Because many developing countries do not have the regulatory capacity to process applications for new drugs, product approval in these countries is heav- ily influenced by approval of the FDA or regulatory agencies in other developed countries. Historically, two competing concerns drove the FDA’s policies regarding the inclusion of women in clinical trials: “(1) the need to protect research participants, and (2) the ‘rights’ of participants to gain access to clinical studies” (IOM, 1994, p. 36). Early FDA guidelines reflected a protectionist orientation. In 1977, the FDA issued guidelines that specified the conditions under which women of childbearing potential could participate in trials (FDA, 1977). However, in practice, such trials almost universally barred premenopausal women. These standards reflected a move toward greater protection for vulnerable populations in response to research abuses, and, most notably, the fetal injuries that occurred during the thalidomide disaster. Moreover, until the 1990s, developers of biomedical products rarely relied on animal studies to identify potential reproductive risks to women and the effects on their offspring until phase 3 clinical trials revealed a product’s efficacy. Because the potential reproductive toxicity of a drug was usually unknown before a trial began, women of childbearing potential were considered ineligible to participate, because they could not make a valid risk-benefit assessment during the consent process. The result of these exclusionary standards was that women were systematically underrepresented in U.S. federally funded research (IOM, 1994). During the 1990s, AIDS activists and women’s health groups called on the FDA to revise its protectionist regulations. Rather than excluding all fertile women from trials, they urged a standard of informed consent. In response, the Institute of Medicine convened the Committee on the Ethi- cal and Legal Issues Relating to the Inclusion of Women in Clinical Studies, which issued Women and Health Research: Ethical and Legal Issues of Including Women in Clinical Studies (IOM, 1994). The committee’s conclusions and recom- mendations represented a profound challenge to prevailing practice. Based on the urgent need to include women in trials, and the impor-tance of respecting their reproductive rights and autonomy, the committee concluded that the potential for becoming pregnant during a trial should not be used to exclude or limit a woman’s participation.a a The IOM concluded that “investigators and IRBs not exclude persons of reproductive age from participation in clinical studies. In the case of women of reproductive age, the po- tential or prospect of becoming pregnant during the study may not be used as a justification for precluding or limiting participation. Risks to the reproductive system should be considered in the same manner as risks to other organ systems. Risks to possible offspring of both men and women who are not pregnant or lactating should not be considered in the risk-benefit calculation. It is the responsibility of the investigators and IRBs to assure that the informed consent process includes an adequate discussion of risks to reproduction and potential 

DESIGN CONSIDERATIONS: PREGNANCY 109 BOX 4-1  Continued More significantly, the committee called for allowing pregnant women to participate in trials based on informed consent. In the face of uncertainty about potential fetal risks, the IOM committee concluded that the pregnant woman should be able continue to participate. The committee also concluded that only when a clinical investigation had no direct clinical benefit to the woman herself and “a risk of significant harm to potential offspring is known or can be plausibly inferred” was it permissible to exclude women from trials. These proposed policy changes would have dramatically increased the burden of proof required to ex- clude pregnant women. Some of the IOM committee’s conclusions mirrored those of the FDA itself, which had abandoned its earlier position of excluding women of childbearing potential from trials in 1993. FDA guidelines recommended screening women for pregnancy before enrollment, and counseling them about the use of contracep- tion. The guidelines also recommended that investigators complete reproductive toxicity studies before phase 2 or phase 3 trials, and the informed-consent process was to include all available information about potential reproductive toxicities. In contrast to the IOM committee’s position, however, FDA guidelines said that the potential risk to the fetus took precedence over the choice of the woman who had become pregnant in trials before efficacy had been established. During this time trials in pregnant women showed that antiretroviral therapy was effective in interrupting mother-to-child transmission of HIV. This was a some- what different case, since the trials were designed to benefit newborns. Still, sponsors continued to prohibit HIV treatment trials from enrolling pregnant women and to require investigators to discontinue study drugs when pregnancy occurred until after phase 3 trials had demonstrated both safety and efficacy. The FDA also recommended the establishment of pregnancy registries to record the outcomes in women exposed to antiretroviral product during pregnancy, as phase 4 clinical trials in pregnant women were usually not conducted. In 2001, the U.S. Department of Health and Human Services modified its regulations and abandoned the universal exclusion of pregnant women, in part because “information on maternal safety and efficacy and fetal safety can be collected in well-designed research settings” (Uhl et al., 2004). However, these regulations have important caveats that limit the circumstances in which pregnant women can participate in research. Specifically, “pregnant women or fetuses may be involved in research only if a trial meets all the following conditions: • Where scientifically appropriate, preclinical studies, including studies on pregnant animals, and clinical studies, including studies on nonpregnant women, offspring, including, where appropriate, an adequate discussion of relevant considerations of birth control. The committee recommends that the participant be permitted to select voluntarily the contraceptive method of his or her choice where there are no relevant study-dependent, scientific reasons for excluding certain contraceptives (e.g., drug interaction). The committee recommends that pregnancy termination options be discussed as part of the consent process in clinical studies that pose unknown or foreseeable risks to potential offspring” (IOM, 1994, p. 15). continued 

110 METHODOLOGICAL CHALLENGES IN HIV PREVENTION TRIALS BOX 4-1  Continued have been conducted and provide data for assessing potential risks to pregnant women and fetuses; • The risk to the fetus is caused solely by interventions or procedures that hold out the prospect of direct benefit to the woman or the fetus: or, if there is no such prospect of benefit, the risk to the fetus is not greater than minimal and the purpose of the research is the development of important biological knowledge that cannot be obtained by any other means; • Any risk is the least possible for achieving the objectives of the research; • If the research holds out the prospect of direct benefit to the pregnant woman, the pros-pect of a direct benefit both to the pregnant woman and the fetus, or no prospect of benefit for the women nor the fetus when risk to the fetus is not greater than minimal and the purpose of the research is the development of important biomedical knowledge that cannot be obtained by any other means, her consent is obtained in accord with the informed consent provisions of subpart A of this part (45 CFR 46.204).”b More recently, several groups have recommended that women who become pregnant be given the option to remain in the trial based on an informed consent standard. A recent report by UNAIDS and the World Health Organization, Ethical Considerations in Biomedical HIV Prevention Trials, recommends a more inclu- sionary approach (2007, p. 37): Although the enrollment of pregnant or breast-feeding women complicates the analy- sis of risks and benefits, because both the woman and the fetus or infant could be benefited or harmed, such women should be viewed as autonomous decision makers, capable of making an informed choice for themselves and for their fetus or child. In order for women to be able to make an informed choice for their fetus/breast-fed infant, they should be duly informed about any potential for teratogenesis and other known or unknown risks to the fetus and/or the breast-fed infant. Similarly, the Council for International Organizations of Medical Sciences’ In- ternational Ethical Guidelines for Biomedical Research Involving Human Subjects (CIOMS, 2002) states that: The justification of research involving pregnant women is complicated by the fact that it may present risks and potential benefits to two beings—the woman and the fetus—as well as to the person the fetus is destined to become. . . . Even when evidence con- cerning risks is unknown or ambiguous, the decision about acceptability of risk to the fetus should be made by the woman as part of the informed consent process. bA description of this section of the Common Rule can be found at http://www.bioethics. gov/reports/reproductionandresponsibility/chapter5.html. The regulation itself is found at http:// a257.g.akamaitech.net/7/257/2422/13nov20061500/edocket.access.gpo.gov/cfr_2006/octqtr/ pdf/45cfr46.203.pdf. 

DESIGN CONSIDERATIONS: PREGNANCY 111 to enter trials, or allowing women who become pregnant during a trial to continue to receive a product by disclosing possible reproductive risks dur- ing the informed-consent process (IOM, 1994; CIOMS, 2002; UNAIDS, 2007). However, historically regulators and sponsors have tended to adopt a more protectionist view (FDA, 1977; IOM, 1994). This approach involves removing pregnant women from exposure to an experimental drug or device unless earlier research has established some potential benefit to study subjects. In the absence of knowledge about such benefit, the prevailing understanding of research ethics dictates that a woman’s right to autonomy should not take precedence over the potential hazard to the fetus. Accord- ing to this perspective, the principle of justice also dictates that the vulner- able fetus be protected from unfair risks. If the benefits of the product to women are clearer, the balance between maternal rights and fetal concern shifts, on the grounds of both autonomy and justice. In practice, however, investigators do not know the potential efficacy, benefits, and harm of a product for the mother and fetus with certainty, making it more difficult to weigh risks and benefits. Several sources can provide information on the potential benefits and risks of an HIV prevention product or device. These include preclinical studies in animals, experience with a product in other clinical trials or applications, follow-up of women who are exposed to the product and become pregnant during the trial, and research and monitoring during the posttrial phase. Preclinical Studies Several types of preclinical tests performed on animals are particularly important in assessing the potential risks of a new product to women and their fetuses (see Box 4-2). Reproductive toxicity studies assess the product’s effects on conception, gonadal function, birth defects, and off- spring development. Pharmacokinetic studies, which evaluate absorption, distribution, excretion, and metabolism of a product, look for evidence of carcinogenicity and systemic toxicity. These studies are important even in the case of topically applied agents (such as vaginal microbicides), where the degree of systemic exposure must be established. Preclinical reproductive toxicity studies can help researchers and policy makers establish the potential risks against which to weigh the potential benefits to pregnant women who participate in clinical trials, although they cannot absolutely rule out the risk of harm. These studies, conducted in more than one mammalian species, can screen for potential human teratogenicity—or birth defects—although they are imperfect predictors of teratogenicity in humans (Ward, 2001; Kennedy et al., 2004). For example, aspirin is teratogenic in mice but not in humans (Corby, 1978). And while 

112 METHODOLOGICAL CHALLENGES IN HIV PREVENTION TRIALS BOX 4-2 Preclinical Testing of HIV Prevention Compounds Before human dosing of HIV prevention compounds begins, preclinical stud- ies are conducted in animals to identify a safe starting dose in humans, and to identify the types of expected toxicity for which human trials will require meticulous monitoring. In 1997, the International Conference on Harmonisation (ICH) issued guideline M3, which was adopted by the United States, European Union (EU), and Japanese drug regulatory authorities, on the type and extent of preclinical safety studies needed for the conduct of clinical trials of investigational pharmaceuti- cals in humans (ICH, 2000). This document provides guidance on the timing of multiple types of preclinical studies (including safety pharmacology, toxicokinetic and pharmacokinetic, single dose toxicity, repeated dose toxicity, local tolerance, genotoxicity, carcinogenicity, reproductive toxicity studies and supplementary stud- ies) with respect to clinical testing. The guidelines also identify areas that are not harmonized among the three regions. As noted, two types of studies are particularly important to understanding the effects of investigational drugs on pregnant women and their fetuses: toxico- kinetic and pharmacokinetic studies and reproductive toxicity studies. Toxicoki- netic and pharmacokinetic studies consist of animal pharmacology studies, and absorption, metabolism, distribution and excretion studies (ADME) and systemic absorption studies. These are usually completed prior to phase 1 clinical trials of an investigational drug (ICH, 2000). Reproductive toxicity tests examine three “segments” of mammalian repro- duction. Segment I (fertility and general reproductive studies) examines fertility and reproductive function in male and female animals, which reveals the drug’s effects on an animal’s ability to get pregnant. Segment II studies (developmental studies) examine developmental toxicity and malformation, to reveal the effects of the drug on fetal development in animals. Segment III studies (late-gestation and lactation studies) examine fertility, growth, and development to examine the drug’s postnatal and perinatal toxicity, to determine its effects during gestation, and on newborn animals during birth or breast-feeding (Ponce and Faustman, 1998). The type and timing of reproductive toxicity studies depends on many factors, including the toxicity profile of the drug and the reproductive potential of the popu- lation in clinical trials (see ICH Guideline S5A/B). There are regional differences in the timing of reproductive toxicity tests, particularly with respect to drugs being tested in women of childbearing potential on highly effective birth control (ICH, 2000). The guidelines also recommend that all female reproductive toxicity studies and the standard battery of genotoxicity studies be completed prior to inclusion in any clinical trial for women of childbearing potential who are not using highly effec- tive birth control or whose pregnancy status is unknown and for women who are pregnant. Prior to inclusion of pregnant women in trials, safety data from previous human exposure are also generally needed (ICH, 2000). 

DESIGN CONSIDERATIONS: PREGNANCY 113 thalidomide is a potent human teratogen, studies in mice and rats did not observe limb malformations (McBride, 1961). Animal studies with positive findings (for example, an agent tests positive as a developmental toxin) tend to be more predictive of human effects than animal studies showing negative effects (Rogers and Kavlock, 1998). Concordance between animal and human effects is strongest when there are positive findings in multiple animal test species, although certain effects do not extrapolate well across species (Rogers and Kavlock, 1998). Absorption, distribution, metabolism, and excretion (ADME) studies can provide information on the potential of a compound to produce sys- temic toxicity. Compounds that show systemic toxicity in animal tests— that is, drugs with hepatic, gastrointestinal, hematological, pulmonary, neurological, or cardiac toxicity—would create concern if intended for widespread use as HIV prevention interventions. Any compounds showing systemic toxicity would be poor candidates for use during pregnancy, as such effects would be unwelcome in both mother and fetus. If a topical compound is designed to be nonabsorbable, systemic absorption studies in animals can provide information on whether, in fact, the compound is absorbed through the skin or vaginal mucosa into the bloodstream, and how long it persists. Products with systemic absorption carry a higher risk of systemic drug-related toxicity for both mother and fetus during pregnancy, and potentially for the infant while breast-feeding. Topical agents with no systemic absorption may have less capacity to pro- duce teratogenic effects during pregnancy. Although ADME studies must be completed before phase 1 testing in humans, reproductive toxicity tests for new products, particularly those without commercial sponsors, are sometimes not completed until the prod- ucts are approved for licensure. This is a problem when the ultimate target population for these products has a high rate of pregnancy. Although there is no mandate to complete all segments of reproductive toxicity studies at any specific time, the International Conference on Harmonization (ICH) has issued guidelines on the timing of such studies. The guidelines on tim- ing of reproductive toxicity studies vary by region, particularly with respect to trial populations of women of childbearing potential on highly effective birth control (ICH, 2000). For women of childbearing potential who are not on highly effective birth control or whose pregnancy status is unknown, and for women who are pregnant, ICH guidance recommends that all female reproductive toxicology studies be completed before enrolling these populations in any clinical trial (ICH, 2000). (See Box 4-2.) Guidance issued by FDA for topical microbicides indicates that all reproductive toxic- ity studies should be completed prior to Phase 2/3 trials (FDA, 2005). The first trial of a candidate microbicide in pregnant women (MTN 002) is expected to start in early 2008. This study, involving 16 pregnant 

114 METHODOLOGICAL CHALLENGES IN HIV PREVENTION TRIALS women, seeks to understand the extent of drug absorption during preg- nancy, and the degree to which the gel’s active ingredient may be transferred to the fetus. A single dose of tenofovir topical gel, an antiretroviral-based candidate microbicide, will be given prior to caesarean delivery (Microbi- cide Trials Network, 2007). Prior Experience with a Product Some agents that are now being tested for efficacy in preventing HIV infection are already licensed for other indications, and their use has shed light on their effects in pregnant women and fetuses. For instance, there is considerable evidence regarding the safety of tenofovir, an approved anti- retroviral drug for treating HIV-infected persons, and growing experience with the use of tenofovir during pregnancy. Oral tenofovir is now being tested as a preexposure prophylaxis agent to prevent HIV acquisition, and tenofovir-based microbicides are also in phase 2 testing. Other antiretrovi- ral agents (such as zidovudine or nevirapine) have been extensively studied in pregnant women, and have been found to be effective and safe when used during pregnancy to prevent mother-to-child transmission of HIV (Guay et al., 1999; Dorenbaum et al., 2002; Mofenson, 2002; Moodley et al., 2003; Dabis et al., 2005). For serodiscordant couples trying to conceive naturally, researchers have also suggested giving PrEP to the uninfected woman, to reduce her susceptibility to HIV infection while trying to get pregnant (Barreiro et al., 2006a,b; Vernazza et al., 2006). No data have yet been published to demonstrate the efficacy of this approach. Information Collected During a Trial Given high pregnancy rates during clinical trials of HIV interventions, some late-stage efficacy trials will accrue significant experience with prod- uct exposure early in the first trimester, one of the most vulnerable times for teratogenicity, even when the trial is designed to discontinue product upon detection of pregnancy. The length of exposure will depend on the frequency of pregnancy testing, which typically occurs monthly or every two months. Information on birth outcomes from women in clinical trials exposed to a product during pregnancy can thus provide information on possible adverse effects on the fetus. To assess such information, investigators can compare pregnancy out- comes in the control and intervention groups, and also examine these outcomes as a function of the duration of exposure to the product dur- ing pregnancy within the intervention group. For example, if a two-arm placebo-controlled microbicide trial of 2,000 women experienced a 50 per- cent pregnancy rate, investigators could obtain information on pregnancy 

DESIGN CONSIDERATIONS: PREGNANCY 115 outcomes on as many as 500 women in each of the placebo and microbicide groups. If pregnancy testing occurred monthly, women participating in the trial who became pregnant would be exposed to the product for up to one month after conception. Such a trial could thus provide valuable informa- tion about the effects of the product on birth outcomes. Indeed, investiga- tors could obtain useful information about the safety of the product while the trial is continuing. In that case, the risk-benefit ratio could change enough to justify use of the product during pregnancy. Posttrial Phase Many developed countries use pregnancy registries to collect informa- tion on birth outcomes of children born to women who were exposed to a product during pregnancy. For example, drug companies that produce antiretrovirals support a pregnancy registry that is intended to provide early signals of birth defects associated with prenatal exposure. Health care pro- fessionals may register pregnant women who have used the drugs, so infor- mation on pregnancy outcomes can be recorded. Registration of exposed women is entirely voluntary. However, it is not clear how effective such registries would be in detecting signals of teratogenicity from agents used to prevent HIV acquisition. Establishing new registries is also not likely to be feasible in developing countries that lack the needed epidemiologi- cal infrastructure and health care reporting. Regulators and others should therefore not rely on pregnancy registries as a major source of postmarket information. If a trial is likely to support market approval, then the “rollover” period between the availability of the findings and product approval pro- vides another opportune time to collect information on experiences during pregnancy. For example, because sponsors typically take 6 months to pre- pare an application for regulatory approval, and the U.S. Food and Drug Administration (FDA) and other agencies take at least 6 months to com- plete a review, trial participants who are still in follow-up might be allowed to remain on the product if they become pregnant during this period and no contravening safety issues arise. Alternatively, women from both the original intervention and control groups who have not yet developed HIV or become pregnant could be randomized to discontinue or not discontinue the study product upon pregnancy during this period. In both cases, the resulting data could provide useful insights into the efficacy and safety of the product during pregnancy. Investigators would therefore do well to prepare a protocol for studying women who become pregnant and consent to use a biomedical intervention during this time. Such a study could con- tinue past regulatory approval, to accrue more data on experiences with the product during pregnancy. 

116 METHODOLOGICAL CHALLENGES IN HIV PREVENTION TRIALS Overall, the committee believes that the current “one-size-fits-all” pol- icy requiring women to discontinue product use upon pregnancy is unnec- essary and potentially counterproductive. No single approach to product use during pregnancy should apply to all biomedical HIV prevention trials, partly because of the diversity of such interventions themselves. Recommendation 4-2: Although the current policy of excluding preg- nant women from biomedical HIV prevention and other trials stems from an historically protectionist orientation adopted by regulators, the principles of research ethics neither mandate nor preclude use of the product by pregnant women. Because any approved product subse- quently would likely be used by many woman who become pregnant, sponsors and investigators of a biomedical intervention should specify in advance of any late-stage trial how they will establish its safety and efficacy for pregnant women and their fetuses, based on information collected both during and after clinical trials. At a minimum, investiga- tors should take the following steps to collect such information. • Investigators should conduct appropriate preclinical tests in ani- mals, including reproductive toxicity and pharmacokinetic studies, to allow a more informed decision on whether to continue product use in pregnant women participating in late-stage trials. These tests would ideally be completed before the product or device enters phase 2 testing, but should be completed no later than phase 3 testing. • Investigators should routinely collect and analyze information about birth outcomes from women who become pregnant during a trial, regardless of whether a product is discontinued upon detection of pregnancy. • In trials that discontinue the use of a product by women who become pregnant, investigators should allow women who are no longer pregnant to have the choice of resuming the study medication. • Investigators should conduct observational or randomized stud- ies in pregnant women in the postapproval, premarketing, and posttrial periods, to provide additional information on the safety and efficacy of biomedical HIV prevention interventions for pregnant women. Recommendation 4-3: Regulators, sponsors, and investigators should evaluate the strength of the evidence on the beneficial and harmful effects to both a pregnant woman and her fetus on a product-by- product basis, and evaluate whether there are circumstances in which women who become pregnant can continue to receive the study prod- uct, based on what is known about its benefits and risks. 

DESIGN CONSIDERATIONS: PREGNANCY 117 Recommendation 4-4:  Trials using products with favorable risk- benefit profiles, but which are nonetheless discontinued upon preg- nancy, should monitor pregnancy outcomes during the interim analysis of trial results, as this information might alter the risk- benefit profile to allow continuation of the product during preg- nancy. Such trials might be modified to thereafter allow women who become pregnant to remain on product or offer them the opportu- nity to be randomized to remain on product versus to discontinue product. Recommendation 4-5: Regulatory agencies and institutional review boards (IRBs) should receive periodic safety updates during a trial that include experience with the product during pregnancy. When interim analyses provide evidence of fetal safety and potential benefit to women, regulators and IRBs should consider allowing women to stay on product while pregnant. REFERENCES Barreiro, P., J. del Romero, M. Leal, V. Hernando, R. Asencio, C. de Mendoza, P. Labarga, M. Nunez, J. T. Ramos, J. Gonzalez-Lahoz, and V. Soriano. 2006a. Natural pregnan- cies in HIV-serodiscordant couples receiving successful antiretroviral therapy. Journal of Acquired Immune Deficiency Syndromes 43(3):324-326. Barreiro, P., A. Duerr, K. Beckerman, and V. Soriano. 2006b. Reproductive options for HIV- serodiscordant couples. AIDS Reviews 8(3):158-170. CIOMS (Council for International Organizations of Medical Sciences). 2002. International eth- ical guidelines for biomedical research involving human subjects. Geneva, Switzerland. Corby, D. G. 1978. Aspirin in pregnancy: Maternal and fetal effects. Pediatrics 62(5 Pt 2 Suppl):930-937. Dabis, F., L. Bequet, D. K. Ekouevi, I. Viho, F. Rouet, A. Horo, C. Sakarovitch, R. Becquet, P. Fassinou, L. Dequae-Merchadou, C. Welffens-Ekra, C. Rouzioux, and V. Leroy. 2005. Field efficacy of zidovudine, lamivudine and single-dose nevirapine to prevent peripartum HIV transmission. AIDS 19(3):309-318. Dorenbaum, A., C. K. Cunningham, R. D. Gelber, M. Culnane, L. Mofenson, P. Britto, C. Rekacewicz, M. L. Newell, J. F. Delfraissy, B. Cunningham-Schrader, M. Mirochnick, and J. L. Sullivan. 2002. Two-dose intrapartum/newborn nevirapine and standard anti- retroviral therapy to reduce perinatal HIV transmission: A randomized trial. Journal of the American Medical Association 288(2):189-198. FDA (Food and Drug Administration). 1977. General consideration for the clinical evaluation of drugs. Rockville, MD: FDA. FDA. 2005. Points to consider in the nonclinical pharmacology/toxicology development of topical drugs intended to prevent the transmission of sexually transmitted diseases (STD) and/or for the development of drugs intended to act as vaginal contraceptives. http:// www.fda.gov/cder/guidance/ptcnonclinical.htm (accessed October 2007). 

118 METHODOLOGICAL CHALLENGES IN HIV PREVENTION TRIALS Guay, L. A., P. Musoke, T. Fleming, D. Bagenda, M. Allen, C. Nakabiito, J. Sherman, P. Bakaki, C. Ducar, M. Deseyve, L. Emel, M. Mirochnick, M. G. Fowler, L. Mofenson, P. Miotti, K. Dransfield, D. Bray, F. Mmiro, and J. B. Jackson. 1999. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to- child transmission of HIV-1 in Kampala, Uganda: HIVnet 012 randomised trial. Lancet 354(9181):795-802. ICH (International Conference on Harmonisation). November 9, 2000. Guideline M3 (Step 4 version): Non-clinical safety studies for the conduct of human clinical trials for pharma- ceuticals. http://www.ich.org/LOB/media/MEDIA506.pdf (accessed February 4, 2007). IOM (Institute of Medicine). 1994. Women and health research: Ethical and legal issues of including women in clinical studies, Volume 1. Washington, DC: National Academy Press. Kennedy, D. L., K. Uhl, and S. L. Kweder. 2004. Pregnancy exposure registries. Drug Safety 27(4):215-228. Macqueen, K. M., L. Johnson, P. Alleman, B. Akumatey, T. Lawoyin, and T. Nyiama. 2007. Pregnancy prevention practices among women with multiple partners in an HIV preven- tion trial. Journal of Acquired Immune Deficiency Syndromes 46(1):32-38. McBride, W. B. 1961. Thalidomide and congenital abnormalities. Lancet 1358(2). Microbicide Trials Network. 2007. MTN-002: Phase I study of the maternal single-dose pharmacokinetics and placental transfer of tenofovir 1% vaginal gel among healthy term gravidas. http://www.mtnstopshiv.org/files/attachments/MTN%20002%20Final%20 Version%201%200%20082907%20(2).pdf (accessed March 25, 2008). Mofenson, L. M. 2002. U.S. Public Health Service task force recommendations for use of anti- retroviral drugs in pregnant HIV-1-infected women for maternal health and interventions to reduce perinatal HIV-1 transmission in the United States. MMWR Recommendations and Reports 51(RR-18):1-38; quiz CE31-34. Moodley, D., J. Moodley, H. Coovadia, G. Gray, J. McIntyre, J. Hofmyer, C. Nikodem, D. Hall, M. Gigliotti, P. Robinson, L. Boshoff, and J. L. Sullivan. 2003. A multicenter randomized controlled trial of nevirapine versus a combination of zidovudine and la- mivudine to reduce intrapartum and early postpartum mother-to-child transmission of human immunodeficiency virus type 1. Journal of Infectious Diseases 187(5):725-735. Ponce, R., and E. Faustman. 1998. Reproductive and developmental toxicity of metals: issues for consideration in human health risk assessment. In Reproductive and developmental toxicology, edited by K. S. Korach. New York: Marcel Dekker, Inc. Pp. 449-474. Raymond, E. G., D. Taylor, W. Cates, Jr., E. E. Tolley, D. Borasky, A. Cancel, B. Masse, E. A. Bukusi, and B. R. Johnson. 2007. Pregnancy in effectiveness trials of HIV preven- tion agents. Sexually Transmitted Diseases 34(12):1035-1039. Rogers, J. M., and R. J. Kavlock. 1998. Developmental toxicology. In Reproductive and devel- opmental toxicology, C. D. Klaassen, ed. New York: Marcel Dekker, Inc. Pp. 47-72. Skoler, S., L. Peterson, and W. Cates, Jr. 2006. Our current microbicide trials: Lessons learned and to be learned. Microbicide Quarterly 4(1):6. Uhl, K., M. A. Miller, and D. L. Kennedy. 2004. Complex clinical, legal, and ethical issues of pregnant and postpartum women as subjects in clinical trials. Journal of Womens Health (Larchmt) 13(6):743-744. UNAIDS. 2007. Ethical considerations in biomedical HIV prevention trials. Geneva, Switzerland. Vernazza, P. L., L. Hollander, A. E. Semprini, D. J. Anderson, and A. Duerr. 2006. HIV- discordant couples and parenthood: How are we dealing with the risk of transmission? AIDS 20(4):635-636. Ward, R. M. 2001. Difficulties in the study of adverse fetal and neonatal effects of drug therapy during pregnancy. Seminars in Perinatology 25(3):191-195.