6

Promising Innovative Approaches to Evidence Generation

Panelists discussed a range of technological and experimental approaches to closing the gaps in data on the safety and efficacy of drug products used during pregnancy and lactation.

INCORPORATING REAL-WORLD EVIDENCE

Real-world evidence, such as that found in electronic health records (EHRs), claims data, and pregnancy registries, can be used to support assessment of the safety and effectiveness of medication use during pregnancy and lactation, said Christina Chambers, professor of pediatrics at the University of California (UC) San Diego. She described interrelated research platforms at UC San Diego as examples of generating real-world evidence on medication use in pregnant and lactating persons.

One research initiative connects to potential participants through MotherToBaby,¹ a web-based resource that provides “evidence-based individualized risk assessment” to pregnant and lactating persons and their health care providers about current and potential exposures (e.g., drugs, vaccines, environmental agents). Chambers said the MotherToBaby counseling service is “a natural springboard for inviting pregnant [persons] to enroll in longitudinal prospective cohort studies, or a pregnancy registry, on the basis of … already having had a given exposure or exposures.” MotherToBaby is also used to enroll comparison populations of pregnant persons who have not had the exposure that is the subject of study but might have the same underlying conditions with different or no treatment. Chambers described the logistics of conducting studies with participants recruited across the nation, explaining that data on exposures and outcomes are collected via phone interviews with the pregnant participants

___________________

¹ See https://mothertobaby.org (accessed September 12, 2022).

and through medical records abstraction. In addition, most study participants are offered a specialized physical examination of their liveborn infant (generally via telemedicine). Pregnant persons and their infants are followed throughout pregnancy and for at least 1 year postpartum, with recontact for follow-up out to 10 years postpregnancy as needed. For example, many studies include neurodevelopmental follow-up of the children through the first year of life, and some to 5 years of age.

Current studies include federally funded projects and studies funded by product sponsors to fulfill postmarketing requirements and commitments. About 25 products are currently being studied. UC San Diego collaborates with the American Academy of Allergy, Asthma, and Immunology on the Vaccines and Medications in Pregnancies Surveillance System (VAMPSS).² Chambers explained that this project brings together real-world data from three parallel study arms: a pregnancy registry, a claims data study, and a case-control study.

Another UC San Diego research initiative is Mommy’s Milk,³ a human milk research biorepository established in 2014. The repository recruits lactating persons in the United States and Canada through the MotherToBaby counseling service and through direct-to-consumer and social media advertising, Chambers said. Lactating persons who enroll provide one or more breast milk samples and a detailed list of their medications and other exposures over the previous 2 weeks. Clinical data are gathered from interviews, and participants can also consent to release medical records for themselves and their child. Studies are funded in part by the National Institutes of Health (NIH), and Chambers noted that UC San Diego is a Maternal and Pediatric Precision in Therapeutics (MPRINT) Center of Excellence in Therapeutics. Mommy’s Milk is also involved in industry-sponsored studies (e.g., a study of remdesivir in breast milk), and in the Pediatric Trials Network’s CUDDLE Study.

Chambers said:

As an example, she said that having these infrastructures in place facilitated the rapid enrollment of several thousand pregnant and lactating persons who were infected with COVID-19 as well as uninfected pregnant and lactating participants for a comparison population. In December 2020, in the absence of data on the use of COVID-19 vaccines in pregnant

___________________

² See https://www.aaaai.org/about/strategic-relationships/vampss (accessed September 12, 2022).

³ See https://mommysmilkresearch.org (accessed September 12, 2022).

or lactating persons in the emergency use authorizations, these platforms enrolled several thousand COVID-19-vaccinated pregnant and lactating persons and a comparison population of pregnant and lactating persons who had declined to be vaccinated.

Chambers highlighted two main challenges in conducting studies with real-world evidence, including lack of awareness of these studies among pregnant and lactating persons and their providers and lack of diversity in enrolled populations.

Ideally, there would be a single, population-wide pregnancy exposure registry in the United States, Chambers said. This model would be more cost-efficient, and less complicated with regard to identification of participants for enrollment in studies. The registry would capture data on the use of all new medications during pregnancy as soon as they are approved for use in the general population and could also be used to prioritize studies of the use of existing medications during pregnancy. Real-world evidence studies are complementary to randomized controlled trials, she said, and can help to address gaps in trial data on risks or safety (e.g., first trimester exposures, rare outcomes). Chambers also agreed with others on the need for improved collaboration across data sources and geographies, improved processes for identifying priorities and sharing data, and standardization of data collection to facilitate cross-platform and cross-study comparisons. She mentioned the EU ConcePTION project as a model for data collection, analysis, prioritization, and sharing that the United States could follow.⁴

Engaging Pregnant and Lactating Persons in Real-World Evidence Studies

Tolúwalàṣé Ajayi, director of clinical research and diversity initiatives at Scripps Research Translational Institute and assistant professor of pediatrics at UC San Diego and Rady Children’s Hospital San Diego, described her personal experience being pregnant during the COVID-19 pandemic and how her obstetrician did not have enough information about the use of the vaccine in pregnant persons to give her any advice about getting it. After discussions with colleagues at American College of Obstetrics and Gynecology (ACOG) and reviewing the data herself, Ajayi decided to receive the vaccine. In addition, she documented the rest of her pregnancy and childbirth on social media, sharing her experiences and her personal medical information. Ajayi said that these experiences affected how she approached becoming the lead researcher for PowerMom.⁵

___________________

⁴ See https://www.imi-conception.eu (accessed September 12, 2022).

⁵ See https://powermom.scripps.edu (accessed September 12, 2022).

PowerMom is a mobile bilingual research platform that aims to create a large diverse research community by engaging one million pregnant and postpartum persons. Participants contribute their data by completing surveys and providing data from wearable sensors and can also share their EHR data. Participants own their data and choose what they share with the researchers. This robust, diverse database is used for observational studies, and information learned is shared with participants. She noted that this existing infrastructure (the platform and the engaged population) can also be used to enable participants to opt-in to other studies (e.g., pharmacological studies).

Drawing on her experiences as both a clinical researcher and a pregnant person, Ajayi understood that “we have to create and develop the research with the people that we want in mind,” and therefore “PowerMom is directly driven by and informed by our participants,” she said. For example, participants provide input on elements of the mobile research platform, such as the timing of the surveys and how information is deployed. Obstetricians are engaged to ensure that the data and studies have clinical relevance. Other stakeholders who provide input include community advocacy groups, institutions, companies, legislators, and international partners (e.g., March of Dimes, Microsoft).

A key aspect of this initiative is bringing the research to the participants, Ajayi said. As noted throughout the workshop, pregnant people are just like other people and have busy lives on top of being pregnant. The research needs to be designed to fit their lives, and PowerMom makes research accessible. Participants can take surveys and contribute data when it best fits their schedule within a set timeframe.

The PowerMom mobile app is also used to disseminate information that is learned. Participants always have access to their own data, but they have access to real-time information from the community and can see how the data they enter contributes to advancing science, noted Ajayi. For example, when they answer a question about having received the COVID-19 vaccine, they might then be able to see data on how many other participants like them also received the COVID-19 vaccine (or a particular medication, etc.). This builds community in real time, she said.

PHARMACOLOGIC APPROACHES TO EVIDENCE GENERATION

There are five key considerations for the use of approved drug products to treat pregnant persons: efficacy in the pregnant person, safety in the pregnant person, safety in the fetus, neonatal safety, and efficacy in the fetus in special circumstances, said Raman Venkataramanan, professor of pharmaceutical sciences and pathology at the University of Pittsburgh.

Studies in Pregnant Persons

The goal when delivering pharmacotherapy is to optimize the dose required to achieve a given response. In general, this involves titrating the dose and measuring the response; however, response to a drug is not easily measured in most of the cases, Venkataramanan said. In cases when the response can be measured, it is often too late to be useful. The solution is to use exposure to the drug as a surrogate marker for response. “If we can make the assumption that similar exposure leads to similar response, then we can look at the exposure and infer what is likely to happen to the response,” he explained.

The area under the plasma drug concentration-time curve, or AUC, reflects the exposure of the body to a drug over a given time interval. The limitation of this approach, Venkataramanan said, is that it requires the pregnant person remain in the clinic for 8 to 12 hours for the collection of multiple blood samples. This is not always practical, and addressing this limitation could involve providing support such as having someone accompany the pregnant person, or facilitating assistance with the care of other children, transportation, or costs associated with participation.

Limited sampling is a potential methodological solution. This involves collecting several samples in the early part of the plasma concentration-time profile and extrapolating a theoretical exposure AUC. Another option is to measure trough concentration, which requires a single sample, as a surrogate for AUC. He noted that it is necessary to document that limited sampling or trough concentration do reflect the entire AUC due to a dose.

Population pharmacokinetic models can be used when limited samples are available from a large number of individuals. These models use combined data to predict drug exposure in patients. This allows for evaluation of patient cofactors that result in data points that deviate from the model-predicted values and can be used to infer whether pregnancy is a cause of variability in the exposure. It allows for an understanding of how patient factors (e.g., race, body weight, gestational age) contribute to the observed exposure.

One novel approach to minimizing the inconvenience of blood sampling is volumetric absorptive microsampling, which is an advanced sample collection method based on dried blood spot sampling. Venkataramanan explained that individuals can collect a blood sample at home by pricking a finger, collecting the sample in a small cotton swab-like device, and mailing the sample to a laboratory. Microdosing can alleviate safety concerns when performing studies in pregnant subjects.

Venkataramanan highlighted the need to systematically study drug transport during pregnancy and said studies using a probe medication are valuable. Data can then be used in physiologically-based pharmacokinetic

models to improve predictive capabilities. Placenta-on-a-chip technology could also be used to simulate pharmacokinetic profiles and evaluate placental transfer of drugs.

In the absence of administering any medication, Venkataramanan suggested that endogenous biomarkers could provide information about metabolic and transport function of eliminating organs (e.g., liver), which could also inform dosing.

Venkataramanan pointed out that all of these analyses only provide the average change (e.g., twofold difference) and do not provide information on the exposure variability between individuals. He said that better understanding of the factors that contribute to interindividual variability (e.g., genotype, hormonal levels, cytokines and proinflammatory mediators that modulate the enzyme activity) is needed. Studies suggest that pharmacodynamics also change during pregnancy for certain drugs (e.g., insulin sensitivity is different in gestational diabetes compared to type 2 diabetes).

Studies in Lactating Persons

Venkataramanan emphasized that lactating persons and pregnant persons are physiologically quite different, and “we need to systematically start doing studies in this patient population, not only from the mother’s perspective but also from the baby’s perspective.” A pharmacologic study in lactating persons would ideally collect all breastmilk pumped over a dosing interval to assess the total amount of drug excreted. This would indicate the cumulative potential dose that the neonate could be exposed to. He noted that this approach is not very practical and added that many nursing parents would prefer to have that milk for their baby. The best alternative approach, he said, would be to collect blood and milk samples simultaneously. Otherwise, a minimum of one sample taken close to maximum drug concentration could be used to calculate the highest potential dose to which the baby can be exposed (based on assumptions of average milk production).

Venkataramanan said that the time span from research and data collection to clinical implementation is long and needs to be condensed, and he suggested that professional organizations should take a lead in promoting the translation of clinical research data and observations into practice.

IMPROVING SAFETY IN THE POSTMARKET SETTING: A REGULATORY PERSPECTIVE

The U.S. Food and Drug Administration (FDA) has the authority to require postmarket safety studies under section 505(o)(3) of the Food,

Drugs, and Cosmetics (FD&C) Act, said Wei Hua, acting deputy director in the Division of Epidemiology of the Office of Surveillance and Epidemiology at the Center for Drug Evaluation and Research (CDER) at FDA.⁶ She said:

Further, she said, “Such studies can also be required after approval, if FDA becomes aware of new safety information.”

In the postmarket setting, pregnancy outcomes associated with exposure to drug products can be assessed using data from pharmacovigilance activities, pregnancy registries, and complementary data sources, and Hua said that “rigorous studies are desired to inform regulatory decision-making.” Pregnancy registries are prospectively designed and collect detailed patient-level data, but Hua noted that patient recruitment and retention in registries is a barrier. Studies using complementary data sources such as electronic health care data (e.g., EHR data, claims data) can help to overcome the limitations of registry-based studies. However, the validity of studies with electronic health care data can be affected by a wide range of factors. Hua said that, although there are many publications describing methods that have been developed and validated, concerns persist because of suboptimal validation approaches and results. Data and methodological challenges for studies using electronic health care data also differ in regard to the purpose of the study, type of study, study question, type of medication, anticipated magnitude of the exposure during pregnancy, nature of the outcome being studied, characteristics of data source, and other elements.

Hua described several ongoing efforts by FDA to improve capabilities for drug safety assessment in pregnancy. One area of focus is developing and refining methods to assess medical product use, safety, and effectiveness during pregnancy using FDA’s Sentinel system,⁸ a “congressionally mandated distributed data network for drug safety surveillance.” FDA is also working with external collaborators to develop approaches for using national Medicaid data for postmarketing active surveillance to assess

___________________

⁶ Hua noted that the opinions expressed are her own and do not necessarily represent an official position of FDA.

⁷ See FDA Guidance on Postmarketing Studies and Clinical Trials—Implementation of Section 505(o)(3) of the Federal Food, Drug, and Cosmetic Act at https://www.fda.gov/media/131980/ download and Draft Guidance on Postapproval Pregnancy Safety Studies at https://www.fda.gov/media/124746/download (accessed September 6, 2022).

⁸ See https://www.sentinelinitiative.org (accessed September 16, 2022).

the safety of drug products during pregnancy. Finally, as mentioned by Catherine Sewell (Chapter 3), FDA is working to fulfill its Prescription Drug User Fee Act pregnancy safety commitment to have improved analytic capabilities and a consistent approach to postmarket requirements and commitments addressing product safety in pregnancy by the fall of 2027.

FACILITATING THE IMPLEMENTATION OF INNOVATIVE APPROACHES TO EVIDENCE GENERATION

Tools and Approaches for Recruiting Study Participants

Participants further discussed the tools and approaches for reaching out to potential trial participants. Chambers reiterated that many of their enrollees are referred via the counseling service and are already interested in participating in a clinical trial. She added that, over the last decade, many participants also self-refer for studies they learn about over social media. Ajayi recommended using existing applications that are already popular among pregnant persons, such as WebMD or the Philips pregnancy app, to reach potential clinical trial participants. Chambers and Ajayi said social media and mobile applications are the modern version of the advertising circulars that used to arrive in every mailbox (e.g., Pennysaver). Venkataramanan said,

He said he tries to talk directly with pregnant persons who enroll in studies to learn what influences them to participate or to not participate in a study.

Approaches for Retention in Longitudinal Studies

The challenge of retaining participants for long-term studies postpregnancy was raised, perhaps following the child for as long as through puberty or to fertility. Chambers said that pregnant persons are concerned about the long-term outcomes for their child, such as the child’s performance in school. She noted that the pregnancy studies at UC San Diego have been able to maintain about a 95 percent retention out to 1 year, and she speculated they could maintain comparable rates for longer-term studies of neurodevelopment and other outcomes. Ajayi added that retention is promoted by returning information of value to participants. PowerMom regularly shares new information about topics

such as COVID-19 and pregnancy, breastfeeding, pumping, exercise, and returning to work. She said that “90 percent of people of childbearing age own a smartphone device” (Pew Research Center, 2021). She encouraged the use of smartphones as a tool to provide information that is timely and relevant to participants, which she said can help keep them engaged with the research platform and studies. Richard Kern, Bill & Melinda Gates Foundation, agreed that using smartphones as a platform can be a powerful tool for bringing the research to the patients.

Data Continuity Across Pregnant Person, Fetus, and Infant

Kern pointed out that, in the United States, the mother and newborn are dissociated in the EHR owing to Health Insurance Portability and Accountability Act (HIPAA) requirements. Participants discussed using research platforms to overcome this limitation and provide data continuity across pregnant person, fetus, and infant. Chambers explained that, for the studies done by UC San Diego, the pregnant person is the enrolled participant, and once the baby is born, the parents grant permission for the baby to be followed in the study as an individual participant. For example, an ultrasound finding of a potential pregnancy outcome would be reported under the pregnant person. The outcome confirming that finding as it presents in the newborn (or the products of conception) is associated with the record for the infant. A pregnant or lactating person who is recruited and enrolled provides clinical data about their exposures and provides milk samples if appropriate. The parent also gives permission for the newborn/child to be a participant and provides information about outcomes in the child (e.g., adverse events, neurodevelopmental information). Ajayi said the same approach is used for PowerMom studies. She noted that birth outcomes are recorded on the pregnant person’s record (e.g., small for gestational age, large for gestational age), while the results of newborn screening (e.g., hearing testing) are recorded under the infant’s record. A participant said they link safety data collected from the pregnant person to the infant at the time of birth.

Carmen Zorrilla, University of Puerto Rico School of Medicine, observed that some patients who join a registry have limited access to health care to address the conditions they might have or procedures they might need. She suggested that data on lack of needed services should be collected along with clinical data. Chambers said that people are enrolled in UC San Diego studies from across the country, including rural and other areas that might not have access to the level of services available at an academic medical center. Although the studies she discussed are observational and not interventional, she said they act as a resource for participants and try to help them identify necessary services near where they live.

Ajayi said that the PowerMom platform includes resources for patients to empower them to act on their data; however, the researchers do not directly act on information provided to the platform by participants. PowerMom also collects zip code data from participants. This information is used to identify health care deserts and to inform efforts to improve access to services. She said that collaboration and data sharing are essential if the data collected in these large repositories are to lead to change.

OPPORTUNITIES TO IMPROVE EVIDENCE GENERATION FOR PREGNANT AND LACTATING PERSONS: BREAKOUT DISCUSSION HIGHLIGHTS

To supplement the panel discussions, participants divided into breakout groups to consider opportunities to improve the generation of evidence on the safety and efficacy of drug products that will be used by pregnant or lactating persons.⁹ A summary of the points made by individual breakout group participants was provided in plenary session by a designated leader for each of the breakout groups.¹⁰

Opportunities to Improve the Generation of Clinical Trial Evidence During Pregnancy

Three groups considered opportunities to improve evidence generation during pregnancy. Highlights of their discussions were provided by Kavita Shah Arora, associate professor and division director of General Obstetrics and Gynecology at the University of North Carolina at Chapel Hill, Anup Challa, AstraZeneca, and Brittany Bettendorf, University of Iowa. Potential solutions discussed included:

• Raising awareness among all stakeholders. A key topic of discussion among participants in the three breakout groups was the importance of educating patients, their partners and families, and the public about the importance of participating in clinical trials, and educating clinicians about the importance of recruiting patients to trials. Several participants suggested that education about clinical trials should begin early in the training of physicians, nurses, and others, and that there should be continuing education for clinicians on the importance of including pregnant people in studies and in

___________________

⁹ For each of the topic areas, breakout groups were convened both in person and via Zoom for online attendees.

¹⁰ This section is the rapporteurs’ summary of the breakout group reports by the group leaders and should not be construed as reflecting any group consensus.

• registries, and of promoting diversity in trials. Engaging with professional boards in promoting education on this topic was also suggested. Patient education could be done by a trusted provider during routine office visits. Social media outreach, podcasts, and interviews with researchers might be used to engage the public and build trust. There were also suggestions that sponsors might engage with FDA in a preinvestigational new drug application meeting to discuss developing materials to recruit and educate clinical investigators, and that the development of such educational materials might be supported by NIH Research Education (R25) grant programs.
• Changing the expectations of when studies with pregnant persons are conducted. Participants discussed the need to conduct studies with pregnant persons in the preapproval stage, rather than automatically delaying them to the postmarket phase. Suggestions included conducting early subpopulation studies to collect pharmacokinetic data during pregnancy, reducing the requirements for contraception in later-phase studies so data can be collected from persons who become pregnant over the course of a trial, and conceptualizing outcomes longitudinally (including maternal health as well as child health and neurodevelopmental outcomes). Studies that have been done of drugs for HIV pre-exposure prophylaxis were mentioned as examples.
• Incentivizing inclusion and addressing liability concerns. Participants discussed developing incentives similar to those in place for increasing studies conducted with pediatric patients, such as patent extensions, monetary incentives, and reducing legal and regulatory barriers. The need to develop programs to support sponsors in addressing liability postmarket was also discussed, including the creation of global and no-fault injury compensation funds for harms incurred by pregnant persons during a trial (similar to the National Vaccine Injury Compensation Program).
• Optimizing the IRB process. Another major area of discussion among participants in the three breakout groups was addressing challenges throughout the institutional review board (IRB) process. Topics discussed included ensuring appropriate diversity within the IRB, instituting an IRB-wide policy to establish the expectation of inclusion of pregnant persons and requiring submissions to explain why pregnant persons would not be included (rather than requiring justification for their inclusion), creating dedicated IRB committees with obstetrician/gynecologist or maternal fetal medicine expertise that could be convened to review complex populations, raising awareness of the resources available and creating new resources to train IRB members about basic protections and study

• design considerations for including pregnant persons in research, and encouraging researchers to engage with FDA prior to submitting R01 grant requests to anticipate and address concerns prior to study section and IRB review. Participants also discussed that the consent process should be more user friendly in how information is imparted to pregnant participants (e.g., using pregnancy-centered language and limiting the use of legal language), and should take participant context into account (e.g., cultural differences globally in giving one’s consent).
• Promoting data standardization and harmonization. The need for standardized definitions of pregnancy-associated conditions and adverse outcomes (e.g., preterm birth) was highlighted by several participants. Participants noted that data standardization and harmonization could improve the collection and usability of data captured in registries and real-world data sets (e.g., to compare rates of adverse outcomes experienced by different populations). It was also suggested that clinical trial databases (e.g., ClinicalTrials.gov) could allow for the reporting of the number of pregnant people in a trial, and better clarify the number of pregnant persons and the number of infants corresponding to those pregnant persons, so that cohorts reported in clinical studies could be more readily identified for use in secondary analyses.
• Enhancing trust between patients and investigators. Participants discussed the need to build trust in the research enterprise. Ideas proposed included having a trusted health care provider discuss potential trials with their patient and engage in shared decision-making, using trusted proxies such as community-based organizations to conduct outreach, not rushing through the consent process, having clinical investigators that are representative of the community and population being enrolled, committing to share the results of the study with participants and the community where the research is being done, setting expectations about whether or when participants will be told if they were in a control or test arm, and addressing patient fears about legal implications (e.g., undocumented participants, studies involving treatments for illegal substance use in pregnant persons).

Opportunities to Improve the Generation of Clinical Trial Evidence During Lactation

Two groups focused on opportunities to improve evidence generation during lactation. Highlights of their discussions were provided by Diane Spatz, professor of perinatal nursing and the Helen M. Shearer profes-

sor of nutrition at the University of Pennsylvania School of Nursing and Danielle Askin, a mother and patient.

Participants discussed:

• Clarifying the distinction between lactation and pregnancy. Although often discussed together, several participants emphasized that pregnant and lactating persons are distinct populations and solutions for inclusion in clinical trials and evidence generation during lactation need to be separated from those for pregnancy.
• Considering the risks associated with not breastfeeding. Concerns were raised about the current practice of stopping breastfeeding when a lactating person is taking a drug “to be on the safe side.” Several participants emphasized that infants who are breastfed are benefiting from being breastfed, and decisions to stop should consider the risks of not breastfeeding for the lactating person and for the baby.
• Designing trials of new compounds to include lactating persons. It was noted that data on the use of many marketed drugs in lactating persons are available (e.g., collected in registries, as discussed by Thomas Hale in Chapter 2). Some products are clearly contraindicated (e.g., genotoxic chemotherapy, radiation treatments). Other categories of drug products require study (e.g., drugs that are extremely lipophilic). Some participants suggested that trials for any new-to-market compounds should be designed to include lactating persons.
• Improving patient centeredness. Several participants highlighted the need for lactation support, appropriate compensation and incentives for participation in clinical trials, improving access to clinical trials, and using understandable language.
• Addressing inequities. Concerns about inequities in access to information about breastfeeding while taking a drug were also raised. Several participants suggested that lactating persons who are better educated or resourced are more likely to seek out additional sources of information and guidance to make an informed choice if they are told to stop breastfeeding because they are taking a medication. Lower-resourced families and underserved populations often do not have the same opportunities. Equity concerns need to be addressed to move the science of treatment during lactation forward.
• Training clinicians and researchers. Lack of knowledge about how to conduct studies with lactating persons was identified as a barrier. Participants discussed the need for comprehensive training programs for clinicians and researchers on the benefits of including lactating persons in clinical trials and on procedures and logistics.