Study design
Considerations for study design
When including pregnant women in a TB therapeutics trial, there are two main goals:
- demonstrating that drug exposures across all trimesters of pregnancy are comparable to those in non-pregnant adults and therefore can be assumed to be efficacious;
- describing (and comparing) maternal, pregnancy, and infant safety outcomes.
When including postpartum or lactating women in a TB therapeutics trial, there are two main goals:
- demonstrating the amount of drug in breastmilk and drug exposure in the infant;
- describing (and comparing) maternal* and infant safety outcomes.
*Note that collecting safety outcomes in postpartum women is important irrespective of whether they are lactating, as adverse events may be exacerbated by the recent pregnancy. For example, hepatotoxicity risk has been found to be higher in the postpartum period among women living with HIV (1, 2). This may require increased monitoring to ensure safety amongst all postpartum women.
Trials evaluating TB treatment and prevention will have different goals.
- TB treatment trials: Treatment is generally initiated at TB diagnosis, including during pregnancy. The risk–benefit balance typically favours maternal treatment to protect maternal health, fetal/infant safety, and prevent TB transmission to other household members and the infant. Here, many women may be willing to accept increased safety risk of treatment given the risk to mother and infant from TB without treatment. These trials will often have more limited sample sizes, as the population of interest tends to be smaller, and may have limited precision and/or power to evaluate PK and safety. This is further explored in pregnancy and lactation PK modelling and sample size sections below as well as the safety sample size section.
- TB prevention trials: Risk–benefit in prevention trials differs from treatment trials because the mother is otherwise well; studies must be adequately powered to assess fetal/infant safety and may warrant delaying treatment, especially depending on the exposure status and maternal risk level.
New TB drugs and regimens
For new TB drugs, PK and safety in pregnancy would ideally be addressed in a single protocol. One approach is a seamless design with two stages: the goal of the first stage is to assess PK to establish optimal dosing in pregnancy in a small non-randomized cohort (n=~20–25), followed by a second stage that enrols and randomizes a larger cohort and would be powered for maternal, pregnancy, and infant safety outcomes. These evaluations may be conducted as a standalone pregnancy study, but may also be embedded within a larger phase 3 (e.g. TBTC Study 37/NCT03474029) or phase 4 trial.
A second approach is to evaluate PK and safety among pregnant and lactating participants enrolled and randomized in the phase 3 trial (see case study on the BEAT-tuberculosis (NCT04062201) trial).
For women who become pregnant while on treatment, the pregnancy occurs post-randomization when they have already been assigned to a regimen, and the regimen may affect either the ability to become pregnant or the decision making around the pregnancy. Analysis of incident pregnancies will require adjustment for potential confounders and may not necessarily lead to the interpretation of causal effect when these data are analysed in isolation. Additionally, the sample size for such an approach is opportunistic, thus precision and/or power may be limited. The advantage of this approach is that outcomes may be evaluated for exposures during conception and the first trimester.
Existing TB drugs / regimens
For existing TB drugs and regimens, dedicated pregnancy studies will be needed to expand indicated populations to pregnancy as it is unlikely these regimens are still being tested in trials that enrol people who are not pregnant. Dedicated pregnancy studies will also be needed when specific safety questions need to be evaluated in pregnant women. This may be done using a randomized or observational design.
Types of study designs
- Randomized design: Randomized designs are most useful when comparing drugs or regimens with respect to safety. If sufficiently powered, then safety can be confirmed (e.g. IMPAACT P1078/NCT01494038) (1). If not sufficiently powered, these are not designed to confirm which regimen is safer, but they may still provide (i) important data on pharmacokinetics and dosing in pregnancy and (ii) estimates of safety and risk differences between groups (e.g. DOLPHIN Moms/NCT05122026) (3). While these examples are studies in pregnant women, this is also true for phase 3 and 4 studies that enrol lactating women.
- Single arm design: Single arm trials are usually designed to focus on PK of novel or existing drugs and to provide initial estimates of safety. Some single arm PK studies are opportunistic, taking advantage of the fact that participants are already, or will be, on study drug as part of standard of care (rationale for this design choice may be cost savings in that study does not need to supply drug). Scientifically, an interventional PK study, where study drug is provided by the study, is usually better than an opportunistic PK study (even for existing drugs) since the safety data are less prone to selection bias (drug is introduced at entry and not before, and clinical parameters are assessed before (baseline) and after drug initiation.
Considerations for study population
Women of reproductive age. When preclinical DART data are favourable, being a woman of reproductive age should not be an exclusion criterion; inclusion should be the norm unless there is a concerning safety signal for fetus or mother (e.g. teratogenic risk).
Pregnant women. When preclinical DART data are favourable, pregnancy should not be considered an automatic exclusion or treatment discontinuation criterion. When insufficient data are known about dosing in pregnancy, PK assessments should be included to assess adequacy of dosing in the second and third trimesters, as described below.
Lactation. When preclinical lactation studies show minimal transfer of drug into breastmilk and do not present any significant infant safety concerns, lactation should not be considered an automatic exclusion or treatment discontinuation criteria. Notably, while drugs with teratogenic effects may not be evaluable in pregnancy, they are often evaluable during lactation. When insufficient data are known about transfer of drug in breastmilk, paired maternal plasma, breastmilk and infant PK should be included to assess relative infant dose, as described below.
Lack of guidance in interpreting preclinical data can be challenging, and may require complex assessment of the risks and benefits of inclusion versus exclusion from the trial. As further guidance from regulators is awaited, early meetings with regulators and inclusion of maternal/infant researchers on the study team and a multidisciplinary steering committee that includes these areas of expertise may be considered.
Timing of drug exposure in relation to fetal development. Exposures during the first trimester can affect organogenesis whereas exposures during the second and third trimesters more commonly affect fetal growth and birth outcomes. Researchers may consider not enrolling women in the first trimester due to (i) concern for potential for teratogenicity during organogenesis and (ii) concern for miscarriage and adverse fetal development being misattributed to drug in the setting of relatively high background rates.
| Is it enough to collect data on women with incident pregnancy while enrolled in drug trials to inform PK and safety of TB drugs? |
| Allowing women with an incident pregnancy to reconsent and remain on study is an important tool to gain information on the effect of exposures early in pregnancy on maternal, pregnancy, and infant safety outcomes, including organogenesis. These data, however, do not provide sufficient information to allow the approval of a drug for use during the second and third trimesters of pregnancy or replace the need for evaluations to be done later in pregnancy. Though these data are opportunistic, they can be critical to understanding human risk, particularly when preclinical DART data are unfavourable (given the imperfect correlation between animal DART studies and human-specific toxicity). Though the sample size may not be sufficient to inform drug labels, the data may provide important information for clinicians counselling their pregnant patients until additional data become available (4, 5). |
Table 5.1 summarizes information that can be collected through enrolment at different times during pregnancy.
Pregnancy stage | PK | Safety |
Incident pregnancy | PK is often not needed during the first trimester as drug exposures during early pregnancy are often similar to exposures in non-pregnant women. | Important to understand first trimester risk to organogenesis. |
First trimester | Planned enrolment during the first trimester is often not recommended given potential risk during the period of organogenesis and high background rates of miscarriage which could be misattributed to drug. | |
Second and third trimesters | Physiologic changes during pregnancy that affect drug exposures become pronounced in the second and third trimesters. These changes differ in the second versus third trimester; PK sampling from both trimesters may be informative for dosing recommendations during pregnancy. | Critical for infant growth and birth outcomes. |
Incident pregnancy. When preclinical data around teratogenicity are favourable, a study may allow women with an incident pregnancy to remain on study drug with reconsent, even with minimal human data available. When preclinical data around teratogenicity are unfavourable, a study may consider allowing women with an incident pregnancy to reconsent and remain on study drug once sufficient data are available from incident pregnancies that occur on study drug without neonatal anomaly from clinical trials and/or observational studies, including surveillance studies.
All exposed pregnancies should be followed through to delivery even if the participant does not continue to take the study drug. This is especially important for drugs with long half-lives, where exposure continues even after drug discontinuation (e.g. bedaquiline). Even if discontinued, it can provide an opportunity to do terminal elimination PK in pregnancy, which can be informative.
In all scenarios, a rigorous informed-consent process is essential. It must clearly communicate what is known and unknown about a drug and its potential effects on the pregnant or lactating woman and fetus/infant. Women must be given this information and trusted to make informed decisions for themselves and their infants, as they already do in routine clinical care, though with less comprehensive information than a research consent process offers.
Contraception. Decisions regarding contraception should align with how incident pregnancy is being managed in the study. Investigators should carefully consider whether contraception is required or offered. When available preclinical and/or clinical data do not support reconsenting participants with incident pregnancy to remain on the study and study drug, contraception may be appropriate as an eligibility requirement. Conversely, when data support reconsenting participants with an incident pregnancy to continue the study and study drug, contraception may be offered but should not be required as an eligibility criterion.
Note: Contraception requirements should be proportionate to actual risks and respect participants’ reproductive autonomy. Contraception should be offered and discussed in a way that supports informed choice. It should not be imposed in ways that restrict access to research unnecessarily or conflict with personal, cultural, or clinical considerations.
Maternal age. Adolescence (age < 20 years) is a high-risk time for both TB and pregnancy. While obtaining both parental consent and adolescent assent may be challenging, this group is important for TB therapeutic trials because (i) both pregnancy and TB are highly relevant to this population who demonstrate poorer TB treatment outcomes (6-9), and (ii) PK data from adult pregnant women may not be extrapolated to adolescents who may metabolize drugs differently than adults (10, 11), although further data are needed to fully understand drug metabolism in girls versus women during pregnancy. TB-CHAMP is an example of a trial that allowed adolescents who became pregnant to reconsent and remain on trial and on study drug (6).
Considerations for the control group
In studies where there is no control group (such as single arm trials of novel regimens), external controls may be useful. Several types of external controls are possible. Historical external controls are often problematic due to a population profile that may differ from the trial participants’ profile, changing external factors (secular trends), and if data arise from observational or surveillance data, differences in data collection (e.g. passive versus active). Nevertheless, historical data may be helpful for sample size planning and would already be available at the time of analysis. These challenges can be mitigated by the following:
- concurrent data in similar populations with the same collection method available (e.g. data from another trial);
- trial data sufficiently harmonized with external control data in terms of outcome definitions and targeted events for maternal–infant populations;
- application of trial emulation methodology to address potential observed confounding and alignment of target population (13);
- data on known important confounders of outcomes of interest are collected from external controls and the trial. This may be achieved with a well-designed surveillance programme or a large ongoing observational cohort, preferably in coordination with clinical trialists engaged in maternal–infant TB research.
Considerations for schedule of evaluations (SOE)
A schedule of evaluations for a pregnant participant will include evaluations based both on the study intervention timeline and the peripartum timeline. While the schedule of evaluations for the study intervention is defined by the trial protocol, it may not coincide with an individual participant’s peripartum course. Therefore, the timing of the intervention relative to pregnancy, delivery, and the immediate postpartum period may vary across participants. To ensure that clinically relevant information is captured, targeted evaluations should be added at key peripartum time points. For pregnant women, their pregnancy-specific evaluations are measured during various trimesters. For a live-born infant, evaluations are measured from birth. Pregnancy and birth outcomes are typically captured at the first post-delivery visit, which should occur soon after delivery.
To help visualize these needs in an SOE, the following approaches have been used in TB protocols:
- adding flexibility of maternal and infant visits for PK, ultrasound, etc. throughout the schedule of evaluations (e.g. TB prevention protocols where there is a directed sample size for pregnant women); or
- not directly including these evaluations as lines within the SOE due to their relatively limited frequency (e.g. TB treatment trials where enrolment may be more limited).
Considerations for outcomes and interpreting drug toxicity
Investigators should use standardized outcomes so that data may be more easily pooled across studies and across disease areas (e.g. TB and HIV). Standardized outcomes for surveillance have been developed in general and expanded for TB and can be adapted for this purpose. In addition, drug-specific adverse effects should be interpreted in the context of physiological changes during pregnancy and lactation, as these populations may require special considerations.
Examples include (14):
- Liver function tests: Normal pregnancy causes changes in liver laboratory values, specifically alkaline phosphatase. Alkaline phosphatase rises significantly due to placental production. Conversely, alanine aminotransferase and aspartate aminotransferase are usually unchanged (in over 90% of cases) or only slightly altered. This can complicate interpretation of drug-induced hepatotoxicity, especially with hepatotoxic TB drugs.
The postpartum period represents a vulnerable window for liver enzyme abnormalities due to rapid physiologic and metabolic transitions after delivery, including shifts in hepatic drug metabolism, hormonal regulation, and systemic haemodynamics. In addition, pregnancy is characterized by relative immune tolerance, and the postpartum immune rebound can unmask or exacerbate inflammatory liver conditions, including viral hepatitis flares or autoimmune-mediated injury. In the setting of TB treatment, these postpartum changes may increase susceptibility to drug-induced liver injury with hepatotoxic agents such as isoniazid, rifampicin, and pyrazinamide. Thus, even when baseline liver enzymes are normal, postpartum women may be at increased risk for subsequent transaminase elevations and warrant careful monitoring. - Cardiac electrophysiology (QTc): Pregnancy leads to increased heart rate and cardiac output, as well as changes in autonomic tone and blood volume that can affect ECG parameters. These hemodynamic and autonomic changes may influence QTc measurement and susceptibility to QT prolongation with drugs such as bedaquiline, clofazimine, or fluoroquinolones.
- Electrolyte shifts: Hormonal and renal adaptations in pregnancy alter fluid and electrolyte balance (e.g. increased aldosterone activity, changes in potassium and other electrolytes). These shifts, especially in the setting of vomiting or pre-eclampsia, may influence arrhythmia risk and interact with QT-prolonging agents.
- Haematologic changes: Pregnancy is characterized by expanded plasma volume and a relative dilutional anaemia, as well as changes in white blood cell count and coagulation factors (physiologic hypercoagulability). These changes can confound attribution of drug-related cytopaenias.
- Renal clearance: Pregnancy increases renal blood flow and glomerular filtration rate, leading to lower serum creatinine and increased clearance of many drugs. However, decreased renal reserve late in pregnancy or with complications (e.g. pre-eclampsia) can increase vulnerability to toxicity and modify drug handling.
Two case studies are shared below of trials that have successfully included pregnant and lactating women and/or allowed adolescents who became pregnant on study drug to reconsent and remain on drug: BEAT-TB and TB-CHAMP.
CASE STUDY 5.1: BEAT-TB (link to clinical trial)
Example: Inclusion of pregnant and breastfeeding women in a phase 3 trial evaluating a 6-month regimen to treat rifampicin-resistant TB
Status: Completed in August 2023.
Overview
The BEAT Tuberculosis (BEAT-TB) trial was a pragmatic, phase 3, open-label, multicentre randomized controlled trial conducted in South Africa. It compared the efficacy and safety of a 6-month regimen (BDL600 LfxC) for the treatment of rifampicin-resistant TB against the minimum 9-month standard of care outlined by the South African National TB Programme.When the COVID-19 pandemic halted global study activities in March 2020, the study team used the pause to reassess the study protocol. Although pregnant and breastfeeding women had initially been excluded, the team recognized that these populations were already receiving the same drugs in routine care—despite a lack of robust evidence on their safety or efficacy. To address this critical evidence gap, the protocol was amended to include them.
We knew we might only enrol a few women, but even a small dataset was better than nothing
Study Design
Eligibility: Women over 18 who were pregnant (in any trimester) or breastfeeding could be enrolled into the study. A live intrauterine pregnancy was confirmed via ultrasound, causing about a one-day delay in treatment initiation. A fetal anatomy ultrasound scan was performed after 16 weeks’ gestation. Women who were planning a pregnancy were excluded.
Contraception and incident pregnancies: Non-pregnant women were encouraged, but not required, to use contraception. Women who became pregnant during the trial could continue treatment if they reconsented.
Consent: Pregnancy-specific consent forms were developed and administered (see resources section)
Study Development
Funder and regulator support: BEAT-TB was funded by the United States Agency for International Development (USAID). The study team discussed with the funders the need to weigh the ethical and practical implications of including pregnant and breastfeeding women in the study and, ultimately, USAID strongly supported the decision
Community engagement: An active community advisory board supported BEAT-TB from the time of protocol development through to study implementation. The advisory board included drug-resistant TB survivors as well as physicians, nurses, and allied health workers. These individuals were acutely aware of the burden of disease among pregnant women, and some had experienced first-hand the frustration of being unable to provide these women with evidence-based counselling, treatment and care. As such, they were supportive of the proposed approach to the inclusion of pregnant women in the study.
Ethics / regulatory approval: The University of the Witwatersrand Human Research Ethics Committee (Wits HREC) raised no additional concerns, as the study involved no investigational drugs and included robust safety monitoring. The study was also approved by the South African Health Products Regulatory Authority without issue.
Insurance: The study team was able to increase their insurance coverage in a stepwise fashion in increments of ten women at a time. Given that they enrolled only ten pregnant women in the end, the cost increase was minimal. The insurance company was reassured by the safety measures in place, including the initial ultrasound and the anatomy scan.
Study implementation
Recruitment: Pregnant women were easily identified and recruited through routine antenatal TB screening. Of the ten approached, none declined. The option of being randomized to a shorter regimen was attractive to participants, as was the availability of additional monitoring, such as the anatomy ultrasound
Site capacity: Pregnant women were easily identified and recruited through routine antenatal TB screening. Of the ten approached, none declined. The option of being randomized to a shorter regimen was attractive to participants, as was the availability of additional monitoring, such as the anatomy ultrasound
Pharmacokinetics: 24-hour PK samples were collected in the third trimester and postpartum, along with breastmilk and infant blood samples. No cord blood samples were obtained as no study team member was present at delivery. Participants appreciated hospital admission during PK sampling for the opportunity to rest and receive prepared meals.
Infant follow-up: Babies were examined for anomalies and signs of TB at birth. Mothers consented to share medical records for six weeks post-delivery.
We felt we were truly contributing. We kept our women safe and watched their babies grow—that was rewarding.
Study Results
Of 170 female participants (42% of total), nine (6%) were pregnant at baseline: one in the first trimester, six in the second, and two in the third. One participant tested positive for pregnancy shortly after enrolment. Of the 10 pregnant women, six received the control regimen and four received the trial regimen. All pregnancies resulted in singleton live births with one premature delivery in the control arm, which was unrelated to TB therapy. One pregnant woman receiving the trial regimen experienced a recurrence of TB disease.
This small but significant inclusion of pregnant women in this study helped inform global policy. Based on the BEAT-TB trial data and existing experience using the component drugs of the regimen, WHO now recommends use of the 6-month BDL600 LfxC regimen for people with MDR/RR-TB, including for pregnant and breastfeeding women with MDR/RR-TB (15). As inclusion of pregnant women in research becomes more the norm, future recommendations will be informed by increasingly robust evidence.
Lessons Learned
Inclusion of pregnant women is possible with creative problem-solving Incremental insurance coverage and private-sector partnerships helped overcome logistical and cost barriers.
Pregnant women want to participate in research All 10 women approached agreed to join the study—highlighting strong interest among this population in research that may improve their care.
Even small datasets can inform global policy change Low enrolment will always be an issue in drug-resistant TB and pregnancy research, but these studies are essential to inform WHO recommendations.
Resources
Study resources
CASE STUDY 5.2: TB-CHAMP
Example: Allowing incident adolescent pregnancies to remain in a phase 3 trial evaluating a six-month TB preventive treatment regimen for child and adolescent contacts of people with multi-drug-resistant TB.
Status Completed in January 2023.
Overview
The TB Child Multidrug-Resistant Preventive Therapy (TB-CHAMP) trial was a phase 3 double-blind, cluster-randomized, placebo-controlled trial assessing the efficacy and safety of a six-month regimen of daily levofloxacin (6Lfx) as TB preventive treatment (TPT) in child contacts of people with MDR-TB.
Initially, TB-CHAMP enrolled only children under 5 years of age. In the last year and a half of the study, the protocol was amended to include children and adolescents aged 5–17 years with a positive interferon-gamma release assay (IGRA). This decision was informed by a 2020 meta-analysis showing TB incidence is 3–4 times higher in IGRA- or tuberculin skin test (TST)-positive children ≥5 years compared to younger children in household contacts of people with drug-susceptible TB (16).
The amendment raised two key questions: Should pregnant adolescents be allowed to enrol in the trial, and should those who become pregnant during the trial remain in the study?
Study Design
Eligibility: Adolescents who were already pregnant were excluded due to limited safety and PK data on levofloxacin in pregnancy available at the time (February 2021) and the preventive nature of the trial. Adolescents ≤6 months postpartum were also excluded due to the risk these adolescents may be breastfeeding and limited data on the transfer of levofloxacin via breastmilk.
Contraception and incident pregnancies: Adolescents were counselled on contraception but were not required to use it. Those who became pregnant during the trial were allowed to reconsent and remain enrolled, based on the rationale that levofloxacin is routinely used to treat drug-resistant TB in pregnant women in South Africa and data from a retrospective cohort study which suggests that levofloxacin is not independently associated with any adverse pregnancy or infant outcomes (other than low birth weight) (17, 18).
Consent: Legal guardian consent was required for minors (defined as <18 years in this trial). Children and adolescents over 7 years of age provided assent. Assent from the fetus’ biological father was not required.
Study Development
Funder and regulator support: Stellenbosch University (sponsor) and funders, including Unitaid, strongly supported the decision to allow adolescents with incident pregnancies to remain in the trial.
Community engagement: The community advisory board at Stellenbosch University advocated for the inclusion of pregnant adolescents in the trial.
Ethics / regulatory approval: The trial was approved by the Human Research Ethics Committees of Stellenbosch University, the University of Witwatersrand, the WHO Research Ethics Committee, and the South African Health Products Regulatory Authority. No delays or issues arose related to inclusion
Insurance: Stellenbosch University secured additional insurance coverage to cover the inclusion of adolescents with incident pregnancies without prohibitive cost or delays.
Study implementation
Site capacity and training: Stellenbosch University trial sites have a family-centred approach to care and research and are involved in numerous studies involving populations at high risk of TB disease. As such, their sites were well equipped to support inclusion of pregnant adolescents in TB-CHAMP. Some study-specific training was provided, including around how to provide contraception counselling to adolescents.
Infant follow-up: Any adverse events in infants born to trial participants within 28 days of delivery were reported.
Study Results
Because the inclusion of adolescents began late in the enrolment period, over 90% of TB-CHAMP participants were under 5 years of age. One 16-year-old adolescent became pregnant during her third month of levofloxacin treatment; she completed her TPT course without complications and delivered a healthy full-term infant.
Current WHO guidelines state that TPT with levofloxacin in pregnancy requires a risk–benefit assessment and informed choice by the pregnant woman on whether to take TPT or defer until after pregnancy (19).
Including pregnant and breastfeeding women in research should be the default. Priority populations should only be excluded for a clear, compelling reason. Don’t look for obstacles; look for solutions.
Lessons Learned
Don’t overlook pregnant adolescents: Even with a low number of adolescents enrolled into TB-CHAMP, an incident pregnancy occurred, highlighting the need to anticipate and plan for pregnancy in this population.
Allowing women and adolescents with incident pregnancies to remain on study fills critical data gaps This approach prevents loss of valuable safety data when allowing enrolment of women who are already pregnant is not scientifically justified.
Investing in site capacity makes inclusive research possible Trial sites with strong experience in family-centred care and research involving priority populations were able to support the inclusion of pregnant adolescents with minimal additional training.
Resources
Study resources
Resources