Executive Summary

The transition from the old alphabetic pregnancy risk categories (A, B, C, D, X) to the FDA’s Pregnancy and Lactation Labeling Rule (PLLR) narrative format represents a fundamental shift in how medication risks are communicated. Under the PLLR (effective June 2015), drug labels no longer use simplistic letter grades but instead provide detailed narrative summaries of the available data on pregnancy, lactation, and reproductive potential. This approach is intended to give healthcare providers and patients richer, contextual information about risks and benefits (^[1]) (^[2]).

However, crafting clear and useful Pregnancy Risk Summaries is especially challenging for complex biologic therapies (monoclonal antibodies, fusion proteins, enzyme therapies, cell and gene therapies). Biologics often have unique pharmacology (e.g. large size, placental transfer via Fc domains) and are frequently prescribed to women of childbearing age with chronic conditions like rheumatic or inflammatory diseases. Yet clinical trial data in pregnant women are almost always lacking. Consequently, pregnancy sections for biologics must often rely on animal reproductive studies, case reports, registry analyses, and expert judgment to convey risk.

This report examines how to draft clear, informative pregnancy risk summaries for biologics in the PLLR era. It provides historical context (the rationale for PLLR), details of the rule’s content requirements, and an analysis of issues unique to biologic products. We review empirical data about biologics in pregnancy – including registry outcomes – and how these translate into label language. Multiple perspectives are considered: regulatory guidance (FDA and international), industry feedback, clinician viewpoints, and patient information needs. Case examples (e.g. TNF–inhibitors in pregnancy) illustrate best practices and pitfalls in formulation of risk summaries. Finally, we discuss future directions: how evolving science, trial inclusion efforts (e.g. ICH E21 guidance on including pregnant women), modeling approaches, and health-literacy strategies might improve clarity and completeness of pregnancy labeling for biologics.

This comprehensive analysis underscores that while the PLLR model offers richer data than old category systems, implementing it well – especially for complex biologics with limited human data – requires careful synthesis of incomplete evidence, clear communication strategies, and ongoing data generation. Suggestions for improving clarity (such as consistent structure, plain-language explanations, and emphasis on known disease versus drug risk) are highlighted. Regulatory and clinical communities must collaborate to ensure pregnant and breastfeeding patients receive timely, understandable information to make informed treatment decisions.

Introduction and Background

Medications during pregnancy are common. In the United States, nine in ten women use at least one medication during pregnancy, and polypharmacy (≥4 drugs) has increased over recent decades (^[3]). Many pregnant patients have chronic diseases (e.g. asthma, diabetes, autoimmune disorders, hypertension) requiring ongoing pharmacotherapy (^[3]). Yet pregnant and breastfeeding women have historically been excluded from clinical trials (^[4]) (^[5]). The result is that most drugs enter the market with little or no direct human data on fetal or neonatal safety. Traditionally, labeling tried to fill this gap with teratogenicity categories (A, B, C, D, X) defined by combinations of animal and human evidence. Over time it became clear these letter categories were misleading and overly simplistic (^[1]) (^[2]).

For example, category D and X designations were widely misunderstood: both imply evidence of risk, but conceding only that one drug’s benefits might justify risk (D) versus another’s clearly not (X). Monotonic categories gave a false sense of precision despite vast variability in the data behind each classification (^[2]). As early as 2007, experts warned that clinicians often misinterpreted the letters as a simple grading of danger, and that important factors (like severity of maternal disease if a drug is withheld) were omitted (^[2]) (^[6]).

To address these problems, the FDA promulgated the Pregnancy and Lactation Labeling Rule (PLLR). Finalized in December 2014, it took effect for all new drug applications by June 30, 2015 (^[7]) (^[8]). The PLLR abolished the old A–X system and instead requires labels to contain three narrative sections: Pregnancy (with a "Risk Summary", discussion of data, and contextual considerations), Lactation (nursing mothers), and Females and Males of Reproductive Potential. New labels also include guidance on pregnancy testing and contraception recommendations where relevant. The goal was a more consistent and thorough way to communicate the actual evidence – both animal and human – about drug use in pregnancy/lactation, enabling more evidence-based decisions (^[9]) (^[2]).

Critics have noted that this new format, while data-rich, is also more complex. The PLLR mandates that labels be continually updated as new information emerges (^[7]). For products already on the market before 2015, a phased transition was required: drugs approved 2001–2015 had to revise labels by 2018–2020, and older drugs by 2018, whereas over-the-counter products remained exempt (^[7]). In implementing the PLLR, regulators stressed that each label should include a clear risk summary drawn from all known data (^[1]) (^[2]). In practice, however, manufacturers have found it challenging to condense fragmented evidence into concise paragraphs (^[10]) (^[2]). Early commentaries praised the PLLR’s intent – for example, dermatologists have noted it “facilitates prescribing by considering a variety of factors” rather than simplistic categories (^[11]) – but also warned of potential confusion if the narrative is too dense (^[10]) (^[2]).

Biologic therapies add an extra layer of complexity.These large-molecule drugs – including monoclonal antibodies, fusion proteins, enzymes, recent gene and cell therapies – often have unique safety considerations in pregnancy. Many are IgG antibodies that actively cross the placenta, especially late in gestation via the Fc receptor (^[12]) (^[13]). They may persist in the fetus or infant for weeks or months, and could theoretically alter immune development (e.g. live vaccines contraindicated for months in exposed infants (^[14])). Pharmacokinetics (half-life, absorption) differ from small molecules, and traditional animal reproductive toxicology models can be imperfect predictors for humans (monoclonals often cross species placenta poorly, etc). On the other hand, biologics often treat severe maternal disease whose activity itself can jeopardize pregnancy (“disease-associated maternal and fetal risk” (^[15])). Thus, risk summaries for biologics must address both the drug-related and disease-related risks.

Because pregnant patients usually cannot participate in randomized trials for ethical reasons, the pregnancy sections of biologic labels typically rely on observational data and registries when available (OTIS/MotherToBaby registries, postmarketing surveillance, patient registries). The amount and quality of such data varies widely. Some anti-TNF agents (e.g. adalimumab, infliximab) now have large registry cohorts; others have only case series or animal findings. For truly novel modalities (e.g. genome editing, CAR-T), human data are virtually nonexistent. Moreover, the pooled evidence base must be kept current: the PLLR expressly requires ongoing label updates as registries and other sources report new results (^[7]) (^[16]).

This report explores how to translate the PLLR mandates into clear pregnancy risk summaries for biologics. It covers the regulatory framework (US PLLR, plus international guidelines), the specific challenges of biologic drugs, and recommendations from the literature and stakeholders on effective labeling. We examine case studies (especially TNF-blockers and other immunomodulators commonly used by pregnant patients) to identify best practices in communicating limited and uncertain data. Perspectives from regulators, industry, clinicians, and patients are integrated, and we highlight initiatives (like increased inclusion of pregnant individuals in trials (^[5]) and modeling approaches (^[4])) that may improve the evidence base and labeling clarity in future. The ultimate aim is to ensure that labeling truly aids shared decision-making for pregnant and breastfeeding patients on biologic therapies – balancing transparency about risk with intelligibility and actionable guidance.

Regulatory Context and Evolution of Pregnancy Labeling

Limitations of the Old System

From 1979 until 2015, the U.S. FDA required a pregnancy category (A–X) in the prescribing information of every drug. These categories combined animal and human data to “quantitatively” grade fetal risk. Category A meant controlled human studies show no risk (e.g. levothyroxine); B indicated no adverse animal findings despite no human data (e.g. amoxicillin); C signified adverse animal data or no studies (e.g. sertraline); D signified evidence of human fetal risk that might be outweighed by benefits; and X meant contraindicated in pregnancy (e.g. methotrexate) (^[17]) (^[18]). However, researchers and practitioners raised two major criticisms of this scheme (^[6]) (^[2]):

• Over-simplification and Misinterpretation: The letter (and color-coding) gave a false precision. Drugs in the same category could differ greatly in actual risk, yet were lumped together. Moreover, clinicians often took the letters as an ordinal ranking of safety (A safest, X worst) rather than a nuanced summary. As the Teratology Society and other experts noted, category D and X “may pose similar risks but are classified based on different risk/benefit considerations” (^[19]). This bred confusion: one author commented that the letter system was “misinterpreted as a grading system, which gave an over-simplified view of the product risk” (^[9]).

• Lack of Clinical Context: The old labels failed to communicate key information that providers need. Notably absent was any guidance on the consequences of withdrawing a needed medication in a pregnant patient. For chronic diseases like epilepsy or lupus, stopping therapy could harm mother and fetus, but the letter category could not convey this. Similarly, letter categories did not address male reproductive potential or provide any lens on contraception. Stakeholder input before PLLR emphasized these gaps: pregnancy drug labeling “lacked clarity” and key factors like maternal health status were not adequately covered (^[2]).

In practice, these shortcomings affected prescribing. Many physicians admitted difficulty interpreting the old categories. The American College of Obstetricians and Gynecologists (ACOG) and rheumatology associations called for more concrete data on specific drugs rather than categorical letters. A survey in dermatology found providers needed to explain the letters’ meaning to patients, effectively performing additional risk counselling beyond the label. Industry also recognized the issues: in comments on the proposed PLLR, the Biotechnology Industry Organization (BIO) lauded the removal of the letters and focus on factual data, but warned that the new narrative format could paradoxically overwhelm prescribers with complexity (^[20]) (^[21]).

The PLLR Framework

In response, the FDA released the final rule “Content and Format of Labeling for Human Prescription Drug and Biological Products; Requirements for Pregnancy and Lactation Labeling” (receipt in the Federal Register, Dec 3, 2014) (^[8]). This Pregnancy and Lactation Labeling Rule (PLLR) removed the requirement for letter categories and introduced the current subsections. Key elements of the rule include (^[1]) (^[8]):

• Narrative Risk Summary: Each label must have a concise “Pregnancy” subsection including a Risk Summary. This must briefly state the risks (if known) of drug use during pregnancy. It should base conclusions primarily on human data when available, supplemented by animal data, and must also describe limitations of the evidence. The summary should also address disease state risks and any FDA-required statements (boxed warnings).

• Clinical Considerations: Under the Pregnancy section, manufacturers should discuss how the drug might affect pregnancy (e.g. recommendations on monitoring, dose adjustments, and known pregnancy outcomes). Crucially, this includes both maternal and fetal considerations. For example, if the maternal condition itself poses a risk (e.g. uncontrolled rheumatoid arthritis increases preterm birth), this should be noted. If the drug could cause neonatal complications (e.g. bleeding risk, immunosuppression), that too falls here.

• Data: A narrative summary of relevant evidence (human or animal) is required. For biologics, this would include registry data, clinical study outcomes, and any pertinent case series. The goal is to give readers a sense of how robust the data are. When human data are very limited, the label should transparently say so. Animal toxicology results (teratogenicity studies) are still reported here, including maternal dosage and species studied.

• Lactation: A parallel format covers breastfeeding. The summary states if the drug or its metabolites appear in breast milk and what is known about infant exposure. Clinical considerations address potential infant effects and recommendations on nursing (e.g. “incompatible with breastfeeding” or “compatible if benefits outweigh risks”). Data on excretion and infant monitoring are included.

• Reproductive Potential: A new section (added to US labeling in 2018 per PLLR extensions) covers pregnancy testing, contraception, and infertility. Notably, it includes information on whether the drug can affect fertility in either men or women. For biologics, some labels note immunogenicity effects on sperm and eggs or advise contraception duration after therapy ends.

Importantly, the PLLR applies equally to biologic products (BLAs) as to small-molecule drugs (^[22]). The labeling format is the same, though FDA’s Center for Biologics Evaluation and Research (CBER) and CDER share oversight (^[22]). The rule underscores reliance on “human data when available, particularly data generated from pregnancy registries” (^[23]). Indeed, one strength is that it explicitly encourages development of pregnancy exposure registries and inclusion of registry results in the label. The final rule states that where such registries exist, a brief description of key findings should be included (^[24]).

Transition and Current Status

Since PLLR’s enactment, all new drugs (and new approvals of existing drugs) have used the narrative format. Older products were converted on set schedules (^[7]). FDA guidance documents (including a Small Business Compliance Guide) provide details on format and content, aiming to help manufacturers translate sometimes sparse data into balanced text (^[22]). However, these guidelines are fairly high-level; manufacturers and reviewers still exercise judgment in phrasing.

Academic assessments suggest variable progress. For newer drugs with active registries (e.g. TNF inhibitors, some cancer therapies), revised labels tend to be detailed and data-heavy (^[16]) (^[25]). But for older drugs or biologics with little post-market data, the pregnancy sections often end up short and equivocal (“The effects are unknown; animal studies show no adverse...” etc.). Even ten years after PLLR implementation, concerns were voiced that many labels still lack substantive human data commentary and that clinicians find some summaries lengthy and hard to parse. Surveys of provider familiarity (outside scope of this review) hint that not all clinicians fully trust or understand the narratives, especially for complex therapies (^[26]) (^[6]).

Internationally, approaches have evolved in parallel. The European Medicines Agency (EMA) never used A–X letters in its official Summary of Product Characteristics (SmPC), but certain EU practice (and some member states) had categorized risk informally. However, EMA died data-guidance stipulates much the same information: risk statements regarding use in pregnancy, pertinent study data, and a harm-lesion rationale. In 2008 the EMA issued guidance on “Reproductive Toxicity” which advises marketing authorization holders to include clear pregnancy risk information based on all data (^[27]) (with EU update consultations ongoing). In 2024, EMA joined global regulators (ICH) in drafting formal guidelines on including pregnant people in clinical trials, underlining the commitment to generating better pregnancy data (^[5]) (^[28]). Canada and other regulators are also aligning more with narrative summaries and encouraging real-world studies in pregnancy.

In summary, the regulatory landscape now requires pregnancy labeling to be data-driven and contextual, rather than categorical. But the practice of drafting these narratives — especially for biologics with limited data — remains challenging. The following sections delve into the specific issues faced when assembling pregnancy risk summaries for biological products, drawing on case studies and guidelines to illustrate principles of best practice.

Biological Therapies: Unique Considerations in Pregnancy Labeling

Biological products (“biologics”) include a wide variety of complex therapies: monoclonal antibodies and receptor-Fc fusion proteins, therapeutic proteins (e.g. enzymes, growth factors), blood-derived products, vaccines, gene therapies, and cell therapies. This report focuses mostly on protein biologics, as vaccines (live virus contraindications are usually obvious) and cell therapies (scarce data) are distinct topics. In the context of PLLR pregnancy labeling, biologics differ from small-molecule drugs in several key ways that affect risk summaries:

• Pharmacokinetics and Transfer: Most therapeutic mAbs are IgG1 subtype or IgG-based formats. Normal physiology transfers maternal IgG to the fetus primarily via the placenta’s FcRn receptor. This transfer is minimal in the first trimester but increases steadily, peaking in the third trimester (^[12]) (^[13]). Thus an antibody given late in pregnancy can lead to substantial drug levels in the newborn at birth. Etanercept, a TNF inhibitor fused to an Fc but somewhat smaller, has appreciably lower placental transfer (likely due to molecular differences) (^[12]). Certolizumab pegol, being Fc-free, shows minimal placental crossing (^[29]). For all mAbs, the label must typically note this physiologic fact (as seen in Humira’s label: “Transplacental transfer in the third trimester” (^[30])) and warn of possible infant immunosuppression.

• Extended Half-Lives: Biologics generally have long half-lives (often weeks), so fetal exposure can linger postpartum. Labels may need to advise on timing of conception in relation to last dose. For example, the infliximab label specifically recommends that women avoid pregnancy for 6 months after the last infusion (^[31]). These kinds of recommendations provide concrete guidance in spite of limited direct evidence.

• Mechanism Overlap with Disease: Many biologics suppress immune or inflammatory pathways. This means the medication’s mechanism often overlaps with the disease being treated (e.g. cytokine blockade in autoimmune disease). As a result, disease activity itself is a crucial factor in pregnancy outcomes. Label narratives should therefore consider the risk of not treating as well as dosing effects. For instance, the Humira label’s Clinical Considerations explains that active rheumatoid arthritis or IBD is linked to higher rates of preterm birth and low birth weight (^[15]). This context helps clinicians weigh the drug’s theoretical fetal risk against the known harm of disease flare.

• Animal Study Limitations: When first-in-class biologics are developed, the traditional teratology studies in rodents or rabbits can be hard to interpret. Mice, rats, etc. may not express the human target or may handle proteins differently. Firms often must conduct developmental toxicity studies in non-human primates instead. Even so, a negative primate study (no malformations) does not strongly prove safety, given species differences. Labels acknowledge this uncertainty by summarizing animal data carefully (e.g. “No fetal harm was observed in primates at doses up to X-fold human exposure” (^[30])) and noting that human data are deficient.

• Data Sources: Because pregnant women on biologics are rarely studied pre-approval, much information comes post-authorization. Sources include pregnancy exposure registries (drug companies and academic collaboratives enroll women prospectively), voluntary reporting (pharmacovigilance), and observational research (cohort studies). For example, a prospective registry might compare outcomes in women with rheumatoid arthritis on a drug versus those not on it. However, registries often have small numbers, self-selection bias, and lack randomization. Therefore, labels usually state that estimates of major malformation rates come from these registries, with caution about interpretation. Humira’s label, for instance, cites the MotherToBaby registry: 8 of 80 live births (10.0%) in exposed women vs 8 of 106 (7.5%) in disease-matched unexposed women had a major birth defect (^[32]) (^[33]). It then notes that the “lack of pattern of major birth defects is reassuring” but that the difference could be due to study limitations (^[32]) (^[33]).

• Special Cases – Advanced Therapies: For newer modalities (gene therapy, cell therapy), pregnancy data are virtually nonexistent by design. Instead, labels must rely heavily on nonclinical risk and cautionary language. For example, the onasemnogene (Zolgensma) gene therapy label carries a pregnancy category (now obsolete) of C, with a statement of unknown risk in humans, possible effects on germline, and advice to use contraception (^[34]). As gene/cell therapies for diseases like hemophilia or SMA expand, any pregnancy labeling will likely emphasize the unknowns and strongly recommend avoiding pregnancy. Some gene products have organ system targets (e.g. liver-directed AAV vectors), and animal teratology could reveal organ toxicity that must be described. These cases underscore that lack of human data must be honestly conveyed, usually in terms such as “the background risk of adverse outcomes in the treated population is not well established”.

In summary, biologics’ complexity arises from their size, mechanism, and scarcity of data. Drafting their pregnancy labels thus requires integrating basic science (placental physiology), clinical context (disease vs therapy risks), and all available evidence streams (animal and human). The next sections discuss how to perform this integration in label language, illustrated by real-world examples, and how to do so in the clearest possible way.

Key Elements of a Pregnancy Risk Summary for Biologics

A well-crafted pregnancy risk summary under PLLR typically contains three interlocking parts: 1. Pregnancy Risk Summary, 2. Clinical Considerations, and 3. Supporting Data (both human and animal). For biologics, certain issues tend to recur in these sections. Below we analyze each component and highlight strategies and pitfalls, using examples from actual labels and guidance.

1. Pregnancy Risk Summary: High-Level Conclusions

The Risk Summary is ideally a concise (one to two paragraph) statement in plain language that encapsulates what is known about the drug’s safety in pregnancy. It should answer: “Based on the evidence, what are the likely risks if a woman takes this drug during pregnancy?”

For biologics, Risk Summaries often include phrases such as “no evidence of increased birth defects” or “it is not known whether [drug] is safe.” Effective summaries typically:

• Balance evidence and uncertainty. If substantial registry data exist, the summary can lean on that. For example, Humira’s summary begins: “Available studies with use of adalimumab during pregnancy do not reliably establish an association between adalimumab and major birth defects.” (^[32]). This acknowledges that, to date, the data do not show a clear risk. It then briefly reports the registry findings (10.0% vs 7.5% defects) to give context (^[32]). Importantly, it adds that this “lack of pattern...” is reassuring. This phrasing helps clinicians understand that no specific malformation signal has appeared and that differences may be due to chance or confounding（e.g. disease activity） (^[32]) (^[33]).

• Mention placental transfer/time-dependence. Almost all biologics that are IgG mention that they cross the placenta, especially in late pregnancy. For instance, Humira’s summary explicitly states: “Adalimumab is actively transferred across the placenta during the third trimester and may affect immune response in the in-utero exposed infant” (^[35]). This alerts prescribers that if they give the drug late in pregnancy, the newborn will have measurable drug levels. Often the advice to avoid live vaccines (given in Clinical Considerations) stems from this fact.

• State lack of data when applicable. If human data are very sparse, the summary should say so. A typical line might be “There are no well-controlled studies of [drug] in pregnant women”, followed by a brief note of any animal findings or registry absence. Vague language like “use only if clearly needed” may appear when risk is unknown; regulators generally discourage simply saying “safety not established” without further context.

• Address disease vs drug risk. In some biologic labels, the summary may mention that risks from the underlying disease complicate interpretation. For example, belimumab (a lupus biologic) notes that active systemic lupus itself increases miscarriage risk. This helps users not over-attribute outcomes to the drug.

• Include recommendations if any. Though recommendations more often go in “Clinical Considerations,” some Pregnancy sections still say things like “individualize therapy balancing the benefits of control of [disease] against potential fetal risk.” If a drug is known teratogenic (rare for biologics), it should say so bluntly (e.g. “contraindicated in pregnancy”). But most biologic labels do not declare pregnancy contraindications; instead they emphasize caution and require contraception discussions in the reproductive potential section.

Example: Consider the Risk Summary for certolizumab pegol (Cimzia), an Fc-free TNF inhibitor. The label notes: “No birth defects or miscarriages have been seen in 1137 certolizumab-exposed pregnancies, indicating rates similar to the general population (^[25]).” It explains that because certolizumab lacks an Fc region, placental transfer is minimal (^[29]). Consequently, the summary concludes no drug-related risk is apparent and suggests it is “a preferred choice during pregnancy” (^[25]) (^[29]). This is unusually positive, but it reflects robust registry evidence and a favorable pharmacokinetic profile.

In contrast, a newer biologic with scant data (e.g. an IL-17 inhibitor like secukinumab) might have a summary like: “Data on use in pregnancy are limited. A pregnancy registry is ongoing. Animal studies did not show malformations, but human effects are not known.” Such wording neither overstates safety nor ignores uncertainty.

Clarity Principles: Reviewers and labels strive to avoid jargon in the risk summary, though naming specific study data (registry results) is common. Plain-language phrases (“no evidence of” rather than “not associated with”, “safety not established” rather than “Category C”) help clinician comprehension. One expert suggested FDA should “test its new label with actual practitioners and consumers to make sure the communication is clear” (^[6]). That sentiment underscores that an ideal risk summary is succinct, trustable, and free of contradictory tones.

2. Clinical Considerations: Contextual Guidance

Following the summary, the Clinical Considerations section provides practical interpretation of the risks to guide patient care. For biologics, this often includes separate subheadings for Disease-associated Risks, Maternal Risks, Fetal/Neonatal Risks, and any dosing or monitoring recommendations.

Key points commonly addressed:

• Disease Control vs. Drug Risk: A prominent feature is description of how the underlying condition itself affects pregnancy. For instance, Humira’s label explicitly states that active rheumatoid arthritis or inflammatory bowel disease is linked to adverse outcomes (preterm birth, low birth weight) (^[15]). This signals to clinicians that uncontrolled disease is dangerous, so cautioning all use without acknowledging this would be incomplete. The label implies that “the risk of discontinuing treatment may outweigh the unknown risk to the fetus,” a message often conveyed by patient registries and expert consensus, even if not stated word-for-word.

• Vaccination and Immune Effects: Because many biologics modulate the immune system, the label typically advises on infections and vaccination for the neonate. For example, Humira’s Clinical Considerations note that maternal IgG can cross late in pregnancy and suggest weighing the timing of live vaccines in the newborn (^[14]). Some labels also mention maternal risk of infections (since many biologics increase infection risk), but more often that is in Warnings sections rather than pregnancy-specific context.

• Dosage Adjustments or Timing: Rarely, labels may state known pharmacokinetic changes (e.g. volume expansion in pregnancy) if studied. Some biologics recommend continuing the last dose into pregnancy or spacing it differently; these suggestions usually come from medical society guidelines rather than label mandates. Labels seldom provide a precise prescribing plan, reflecting a lack of evidence.

• Paternal Exposure: Some biologics comment on partners’ exposure. For example, a few TNF inhibitor labels (Humira, Enbrel) state if the father is on therapy at conception, no increase in birth defects has been observed. This provides reassurance but is often buried in fertility sections rather than pregnancy sections.

• Advice on Pregnancy Management: Occasionally, a label may urge contraception during use and breastfeeding discussion postpartum. The fertility/reproductive potential section is usually where birth-control recommendations are formalized (e.g. “Women of childbearing potential should use effective contraception during treatment and for 5 months after the last dose” (^[32])). Sometimes the pregnancy section reiterates it: Humira’s summary mentions “Manufacturer recommends contraception during therapy and for 5 months after” (^[12]), highlighting persistent presence of drug.

The Clinical Considerations should be action-oriented. For example, if animal data showed an effect size at high dose, a label might still say “No specific actions are recommended based on animal data alone” but suggest standard prenatal monitoring. If disease control is paramount, the label may explicitly state that treatment should not be stopped abruptly (sometimes under a subheading like “Disease-associated maternal and embryo/fetal risk” (^[15])). It rarely gives tactical guidance on delivery or breastfeeding, leaving that to individual clinician judgement or specialty guidelines.

One instructive example is from the MTP (Teratology Society) guidance draft: it suggests labels should remind providers that unresolved maternal disease itself can cause pregnancy loss or congenital anomaly, and that stopping drugs can cause severe flares. Humira’s label reflects this: by describing the maternal disease risks, it implicitly endorses continuing therapy to avoid flares (^[15]). Such balancing statements are crucial for clarity: without them, a risk summary that sounds benign might be misread as permission, or a cautious summary might scare providers into unnecessary drug holidays.

3. Data (Human and Animal): Evidence Synopsis

The Data section typically follows Clinical Considerations (for many FDA labels it is integrated within the Pregnancy section). This part lists supporting evidence, often organized by human vs animal:

• Human Data: For most biologics, available human data come from pregnancy registries, observational cohorts, or case reports. Labels summarize the size and outcomes of these sources. This may be in narrative text or a tabular format. For example, Humira’s label details the MotherToBaby registry: “In a prospective cohort, major birth defects occurred in 10% (adalimumab) vs 7.5% (control) of infants; no birth defects pattern was seen” (^[33]). It also notes study limitations (voluntary enrollment, small numbers) when drawing conclusions (^[33]). Many labels cite the underlying study references for these data (often not hyperlinked on the label but available in FDA review packages).

Good practice is to highlight sample sizes and limitations clearly. Instead of vague “no evidence of harm”, stating “X number of pregnancy exposures have been reported” with references makes the summary transparent. “No pattern of abnormalities has emerged among ~N infants” is a common phrasing. Labels often include statements about disease-matched controls, to put numbers in context (^[32]) (^[33]). If a registry (e.g. for rheumatoid arthritis patients) shows outcomes comparable to background rates after adjusting for maternal age and disease, that is reassuring and emphasized. However, labels correctly point out that registry-derived rates cannot prove safety (e.g., “cannot reliably establish whether there is an association…because of methodological limitations” (^[36])).

• Postmarketing Surveillance: The PLLR expects post-marketing data to be ongoing in labels. Companies are required to report pregnancy exposures to the FDA. If substantial adverse findings emerge, labels should be updated. One example: certolizumab’s label (as noted above) cites a pharmacovigilance database analysis by Clowse et al. (2018) showing no rises in malformations (^[25]). It explicitly states registry data of 1137 patients showed “no teratogenicity or increased risk of fetal death.” Presenting these figures convinces clinicians more than a blanket endorsement.

• Animal (Nonclinical) Data: The final part of the data section lists animal study results. For biologics, a typical summary reads: “Reproduction studies in animals (often in monkeys and one rodent) revealed no evidence of fetal harm at doses X times human exposure.” For instance, Humira’s label notes that pregnant monkeys given >300x the human dose had no malformations (^[30]). If any adverse findings occur (e.g. embryo-fetal death or growth restriction at high dose), these are mentioned. It is important to include species, dose multiples, and specific endpoints (teratology, embryo-fetal death, etc.). However, such findings are often minimized in writing when human data are reassuring.

Biologics usually have similar nonclinical profiles: wide margin in animals, or no signal. When drafting labels, sponsors must avoid over-generalizing animal safety to humans. Thus, labels often conclude the animal data are “not predictive of human outcomes” unless human data confirm an effect.

Summarizing Evidence: The overall picture is built by connecting the narrative from Risk Summary through the data. For example, Humira’s label uses the data to explain the earlier statements about birth defect rates and transfer (and then goes on to “Infant Data” and “Labor or Delivery” as subsections). A label must carefully ensure that the risk summary and clinical considerations do not contradict the data. Over time, as new registry updates are published, companies revise the data section. Ghalandari et al. (2020) analyzed EMA documents and found that many biologic labels had not been updated with new pregnancy information even decades after approval (^[16]), highlighting the need for vigilance.

Tabular Data (optional): Some companies use tables to present complex registry results or animal study details. For instance, a table of pregnancy registry outcomes by trimester, or a table listing animal study designs and findings, can make the data more transparent. Euradvice suggests that including either tabular or bulleted data summaries can aid comprehension. We propose in this report (see Table below) a schematic way to present limited data for selected biologics (human registry size, major outcome rates, etc.) – though actual labels rarely include such tables, using bullet points instead.

Example Reference (Human Data): A meta-analysis (Bržica et al. 2020) of pregnancy outcomes with TNF inhibitors found no significant increase in congenital malformations overall, reinforcing label messages of safety (^[37]) (^[38]). Such systematic reviews often appear in label reference lists but are generally not quoted directly (labels prefer manufacturer data or controlled registries). Nonetheless, mentioning systematic review findings (with citation) in a discussion or guidance document would bolster confidence that the label reflects current knowledge.

Challenges and Opportunities in Clarity

While the PLLR structure is logically sound, pitfalls arise if the text is too technical, too verbose, or too cautious:

• Excess Complexity: As the BIO trade association warned, physicians might become “confused by the complexity” of narrative labels (^[21]). Dense paragraphs of data can overwhelm a busy prescriber. Breaking up text into short paragraphs or bullets for key points (benefit vs risk, study highlights) improves readability. Some industry labels incorporate subheadings or italics (where allowed) to guide readers: e.g. “Background Risk of Birth Defects: 2–3% in general population” as a separate note to contextualize findings.

• Negative Framing: Physicians have reported that negative statement (“safety not confirmed”) without follow-up advice can induce anxiety. Labels should strive for balanced framing. Instead of “no data exist” possibly say “Limited data are available” followed by “Physicians should weigh this against the known risk of disease flare”. The label alone cannot solve all patient anxiety, but helpful context can prevent misunderstandings (e.g. patients thinking narrative sections mean something is definitely dangerous when evidence is really lacking).

• Inconsistency Across Labels: Feedback suggests labels sometimes differ in style, making cross-product comparison hard. For instance, one anti-TNF label might completely omit paternal effects, while another covers it. Similarly, some labels quantify background risk and others do not. Standardization (consistent headings and phrasing) would help HCPs quickly find key info. The FDA and professional societies have called for greater harmonization of language and emphasis in pregnancy sections.

• Patient-Facing Communication: Although PLLR text is written for clinicians, pregnant patients also encounter FDA labels (often via internet searches). There is a push to include patient summaries or at least plain-language explanations alongside the technical label (^[39]). Currently, a layperson reading a risk summary about biologics may be confused by terms like “NODAL, PPK mice model embryotoxicity” or unfamiliar drug names. Ideally, manufacturers and regulators could supplement labels with brochures or web content specifically for patients, translating the same findings into understandable language.

• Keeping Current: Because biologics are relatively new to many indications, evidence evolves fast. The PLLR requires annual reviews of the literature for new pregnancy data. However, in practice many labels lag behind. As case in point, Ghalandari et al. found that by 2020 only 4 of 16 biologic GI/arthritis drugs had updated post-market pregnancy data, even though some had been on market >5 years (^[16]). Companies must proactively update risks summaries or at least note when data are insufficient (e.g. “A registry in [disease] is active; no results yet”).

• Global Harmonization: Even within the US, differences in formatting have emerged. Moreover, in Europe or other markets, labels may emphasize different aspects. For multinational companies, aligning label content across regions is important but not always straightforward (e.g. EMA SmPC versus FDA USPI phrasing). Ideally, the same key messages (e.g. safety signals or lack thereof) would be consistent, with only language style varying. Coordination via ICH or WHO guidelines on pregnancy labeling could improve this.

Conclusion of this section: A clear Pregnancy Risk Summary for a biologic is one that succinctly states what is known and what is not, quantifies any known risks, and advises on management. Achieving this requires judicious editorial choices: include necessary data points without excessive detail, emphasize meaningful conclusions, and avoid contradictory or ambiguous language. The examples above (Humira, Cimzia, etc.) illustrate successful approaches. The rest of the report will analyze real-world data and stakeholder opinions to identify best practices for future biologic labels.

Evidence Review: Case Studies and Data Analysis

To illustrate the principles, we examine specific biologics and relevant studies on pregnancy outcomes. These serve as case studies for how evidence translates into risk summaries. We focus on biologic immunomodulators (a large class with relatively more data) and note where biosimilars or newer biologics fit in.

TNF-α Inhibitors

TNF-α inhibitors (including infliximab, adalimumab, etanercept, certolizumab) are among the most extensively studied biologics in pregnancy, because they are common treatments for rheumatoid arthritis, Crohn’s disease, and psoriasis – conditions often affecting women of reproductive age. Key findings from the literature include:

• Birth Defects: Multiple registry analyses and meta-analyses consistently show that TNF inhibitors do not significantly increase major congenital malformation rates above background. For instance, Humira’s label reported ~10% defect rate in exposed vs 7.5% in controls (^[32]) (^[33]); though numerically higher, this difference was not statistically significant and no pattern of anomalies emerged. Weiss et al. (2020) systematically reviewed TNF inhibitors in pregnancy and likewise found no significant rise in major anomalies (^[37]) (^[38]). Thus, labels commonly convey “no apparent increase” in defects, mirroring human data.

• Pregnancy Outcomes: Some evidence suggests TNF inhibitors may improve pregnancy outcomes by controlling maternal disease. Uncontrolled IBD or RA is associated with preterm birth and low birth weight (^[15]); TNF therapy can reduce inflammation. A recent meta-analysis showed TNF-exposed pregnancies had similar or sometimes better outcomes than those with active disease/no therapy (^[40]) (^[38]). This nuance is sometimes mentioned in discussion even if not in the label text. Labels often list adverse outcomes (e.g. “no increase in stillbirth or neonatal death rates has been observed in exposed pregnancies” – a line present in multiple TNF labels).

• Placental Transfer: All IgG1 TNF antibodies cross readily in late pregnancy, exposing infants for months. Infliximab and adalimumab are fully IgG1; etanercept (a dimeric protein with Fc) crosses less efficiently (^[41]). Certolizumab (Fab-PEG) basically does not cross (^[29]). These facts appear in labels as: heavy emphasis on timing (suggesting final doses before third trimester if possible) and in some cases distinct pregnancy guidance. For example, label summaries often caution: “Given placental transfer, limit fetal exposure – e.g. avoid dosing after 30 weeks gestation”. However, this detail usually appears in clinical recommendations rather than risk summary.

• Lactation: TNF inhibitors are large proteins that have low oral bioavailability for infants, and small amounts are detected in breast milk. The AAP deems them “compatible with breastfeeding,” a view echoed in labels: e.g. “concentrations in breast milk were minimal; no adverse events reported” (^[41]) for etanercept, and similar wording for adalimumab. This positive language reassures providers that continuing therapy through nursing might be acceptable.

• Live Vaccines: An explicit consideration for all intrauterine TNF exposures is the live-attenuated rotavirus vaccine. The Humira label (under Clinical Considerations) advises delaying live vaccines in newborns exposed in utero for at least 3–6 months (^[14]). This guidance was not needed in older category labels.

Example: Adalimumab (Humira®)

Adalimumab’s U.S. label (8.1 Pregnancy section) exemplifies a transparent risk summary. It states up front that no causal link to birth defects is established by current data (^[32]). It then quantifies the MotherToBaby registry findings (10% vs 7.5% defects) and explicitly notes “the lack of pattern of major birth defects is reassuring” (^[32]). The wording carefully avoids giving a definitive “safe” rating, instead saying “do not reliably establish an association.” It also reports the monkey study at 373× dose, clearly indicating no malformations in animals (^[30]).

In Clinical Considerations, it adds that discontinuing RA or CD treatment may itself harm pregnancy (^[15]). Thus, the dual perspective is that adalimumab appears low-risk, but underlying disease must be managed. The Data section repeats the registry result with sample size (221 on drug vs 106 controls) and attributes the slight numerical difference to bias and limitations (^[33]). In total, a practitioner reading this would conclude: “Adalimumab is not proven to be teratogenic; keep the mother’s disease controlled, but also discuss fetal monitoring. Advise contraception through therapy and counsel about vaccination of baby.”

This thorough labeling contrasts with an older category label (pre-2015) which might have labeled Humira category B (if animals were fine) or C (if no human data). The narrative now gives much more nuance and actionable detail. Other TNF inhibitors have similar language: etanercept and infliximab labels describe registry rates and note no patterns of defects (^[41]) (^[42]).

Biosimilars

Biosimilars are effectively equivalent to their reference biologics, so their pregnancy risk profiles are expected to match. However, when new biosimilars enter the market, they often carry class-label sections. For example, biosimilar anti-TNFs (for infliximab or adalimumab) will generally cross-reference the innovator’s pregnancy data and include a statement like “data from reference product H apply”. A Drug Topics article reported the first clinical study specifically on TNF biosimilars in pregnancy: 18 women treated with infliximab/adalimumab biosimilars had no infant harm, and discontinuations (to avoid use) led to flares (^[43]). The authors concluded that stopping therapy led to worse maternal outcomes than continuing drug. This kind of finding – that real-world biosimilar use parallels innovator safety – suggests that pregnancy risk summaries for biosimilars may reasonably rely on those of the originator. Clear labeling should say so: e.g., “Available evidence for [biosimilar] is consistent with that of reference product [Drug], as reported above”.

Other Immunologic Biologics

Beyond TNF blockers, a variety of monoclonals for other autoimmune diseases and cancers (e.g. rituximab, ustekinumab, tocilizumab, IL-17/23 inhibitors) are used in women of childbearing age. Data for these vary widely:

• Rituximab: Small studies (and the ability to measure CD19+ B cells in cord blood) have shown that rituximab given in pregnancy can lead to transient B-cell depletion in the neonate, requiring caution about live vaccines (^[44]). However, major defects are not reliably increased. The FDA label notes limited numbers (e.g. <200 exposures in literature) and reports no pattern of malformations. It generally advises contraception during and for 12 months post-treatment.

• Ustekinumab (IL-12/23 inhibitor) and newer IL-17/23 inhibitors: Virtually no controlled human data exist. Small case series in psoriasis show healthy outcomes, but registries are tiny. Animal studies often show no teratogenicity. Labels say “limited human experience” and recommend avoiding live vaccines in exposed infants.

• Tofacitinib (JAK inhibitor, technically a small molecule): Its label is cautionary, citing animal teratogenicity (it was a pill known to be risky) (^[45]). Pregnancy is strongly discouraged and contraception is mandated. This shows that labeling can diverge: not all advanced therapies end up “probably safe.” Each product’s text must reflect its own evidence or risks.

• Vaccines: While beyond the main scope, it’s worth noting that FDA PLLR also applies to prophylactic biologics. The influenza and COVID-19 vaccines each have pregnancy sections, but these focus on population efficacy since those were tested in pregnant women. Monoclonal antibodies for COVID (e.g. REGN-COV2) have pregnancy labels based largely on animal data (no human trials), so they say “no human data available” and “it is not known if drug affects the fetus”.

Table: Selected Biologics and Pregnancy Evidence

Below is a summary table (adapted from various sources) illustrating how some representative biologics are handled in labeling. Each row lists the type of biologic, the nature of available pregnancy data, and how this translates into label language.

Table 1. Summary of Pregnancy Data for Selected Biologic Therapies

(Data sources: established pregnancy registries; labels; clinical studies referenced in text. Not all biologics are shown.)

This table exemplifies how the quantity and quality of data shape the pregnancy section. A product like certolizumab, with large registries, has a confident, positive narrative. Older or niche drugs with sparse data have more guarded statements. Regulatory reviewers look for consistency: e.g. if two TNF drugs both have hundreds of cases, one should not be labeled as safe while another is uncharacterized. Cross-label calibration is advised but not always achieved.

Perspectives and Stakeholder Input

Developing effective pregnancy risk summaries requires aligning regulatory intent, industry capability, clinician needs, and patient communication. Key perspectives include:

• Regulatory/Government: FDA officials have emphasized the PLLR’s purpose to improve safety decision-making. In guidance meetings, FDA staff (e.g. Sandra Kweder) stated that categories were “misinterpreted…which gave an over-simplified view of product risk” (^[9]). Thus the FDA expects summaries to be grounded in data and to include contrasts of risk versus benefit. The use of human data is stressed, even if limited, and pregnancy exposure registries (often FDA-postmarketing commitments) are explicitly encouraged (^[23]). The Office of Women’s Health (OWH) tracks research on pregnancy drug safety and often calls for more funding to study medication use in pregnancy (^[46]). Internationally, EMA and ICH are coordinating on guidelines requiring inclusion of pregnant women in research, acknowledging the historical gap (^[5]). Collectively, regulators push for more evidence generation, not merely better label writing.

• Industry (Manufacturers): Biotech and pharma sponsors have mixed views. The 2008 BIO comments to the FDA on the proposed rule praised many aspects of PLLR (removal of categories, use of factual statements) but cautioned that the narrative requirement might confuse clinicians if the prose is too technical (^[20]) (^[21]). Their feedback suggested standardized templates or examples could help. The industry also faces economic and logistical issues: collecting pregnancy data (via registries) is costly and voluntary. Without incentives, companies may only update labels sporadically. In practice, some manufacturers have added plain-language brochures or online Q&A to supplement product labeling for patient education.

• Clinical Practitioners: Clinicians have welcomed the richer information but sometimes struggle with interpreting it. Surveys (in specialties like dermatology and rheumatology) indicate that doctors appreciate knowing the underlying data (e.g. “registry of size n X found Y outcomes”) but still sometimes default to category-like rules of thumb (“I know TNF blockers are mostly safe”). There is a call for decision support tools and helplines (like MotherToBaby or Reprotox) to help navigate label nuances. Pharmacists often use tertiary compendia or guidelines (e.g. rheumatology recommendations) which distill label info; if the label is unclear, these experts try to fill the gap. Training is needed so providers can explain PLLR narratives to patients effectively.

• Patients and Advocacy: Patients, especially those with chronic illness, often find the old categories mystifying. The new narrative labels are theoretically more informative, but not directly written for lay readers. Patient advocacy groups (e.g. March of Dimes, pregnancy centers) have advocated for transparency of studies and for FDA to communicate findings in understandable formats. Anecdotally, many pregnant women search online for drug safety info; encountering the PLLR text may be daunting. Some experts propose adding patient “Infographic Appendix” that summarizes pregnancy safety in bullet points. As one case story, a pregnant woman on adalimumab recounted how her doctor described the label data as “so much text, it just said we kept reassuringly find nothing new”, reflecting that clinicians themselves sometimes truncate the label content when counseling.

Discussion: Implications and Future Directions

The shift to PLLR has advanced pregnancy labeling fundamentally, but the journey to optimal risk communication remains ongoing. For biologics, several implications and future trends emerge:

1. Increasing Data Through Inclusion: A major theme is the need for more direct evidence. The upcoming ICH guideline (E21) and similar initiatives encourage deliberately including pregnant and lactating women in clinical studies when feasible (^[5]). This paradigm shift could, over time, populate labels with real trial data rather than just post-market anecdotes. In addition, improved surveillance (national pregnancy registries, electronic health record analyses, birth defect monitoring) may generate large datasets. The U.S. National Institutes of Health and FDA Office of Women’s Health have funded pragmatic pregnancy studies; results from these could feed future label updates.

2. Model-informed Predictions: Advances in physiologically based pharmacokinetic (PBPK) modeling can help fill gaps. For many biologics, models of maternal-fetal antibody transfer (incorporating placental induction, FcRn levels) have been developed (^[4]). Such models might predict fetal drug levels without direct human data. EMA has endorsed model-informed strategies for pregnancy dosing recommendations (^[4]). If validated, these could be cited in Clinical Considerations (e.g. “Modeling predicts minimal fetal exposure when dosing is held in third trimester”). Modeling future risk might reduce uncertainty and lead to more specific advice on dose timing.

3. Cross-Product Harmonization: Regulators and experts note the benefit of consistent terminology. For instance, if two products have similar safety profiles, their labels should convey this similarly. Inconsistent wording can mislead – e.g. one label says “no risk determined due to insufficient data” while another says “no increased risk observed”. The FDA pregnancy labeling rule itself lacks a rigid template (unlike the highly structured US product label in pharmacology); manufacturers have flexibility. In coming years, FDA might issue more detailed guidance or examples (possibly casebooks or test label templates) to reduce inter-product inconsistency.

4. Improving Readability: Given the complexity, using user-friendly formatting could help. The FDA’s Patient Labeling Initiative and Plain Writing Act suggest that drug information should be accessible. Although the official Pregnancy section is for providers, adding summary bullet points or patient handouts could bridge the gap. Joint efforts between FDA, industry, and patient groups (such as the Coalition of Obstetric, Maternity, and Pediatric Pharmacists) could produce standardized patient fact sheets.

5. Global Coordination: As biologics often have global markets, aligning country labels would aid multinational practitioners. ICH Guideline E2D (Pharmacovigilance planning) and related efforts might push for shared content on pregnancy. WHO and global teratology networks could maintain centralized drug safety databases to inform labeling worldwide, reducing duplicative work for companies and harmonizing messages across regions.

6. Case-by-Case Advisory: In practice, healthcare providers often consult Teratology Information Services (e.g. MotherToBaby, embrotox databases) for complex cases. These experts use evidence beyond the label. Recognizing that labels cannot cover every nuance, there may be increased integration whereby label summaries reference respected external resources (some labels now have footnotes like “see [MotherToBaby Fact Sheet]”). This hybrid approach acknowledges that providers will seek additional counsel beyond the label text.

7. Ethical and Legal Dimensions: FDA’s shift from categorization to narrative was partly motivated by ethical concerns – ensuring women had better-informed choices. Lawsuits and public pressure (such as after thalidomide and other harms) have underscored the duty to update labeling as new evidence emerges. Going forward, if pregnancy warnings are incomplete and harm occurs, companies could face liability. This legal incentive may drive more proactive updating and clearer language. Transparency (open-label databases, FDA advisory reports) will likely increase scrutiny of pregnancy sections.

Conclusion

The PLLR framework aims to put “reality” into pregnancy labeling, but translating that reality into clear guidance is a nuanced task. For complex biologics, where randomized trial data in pregnant women are virtually nonexistent, achieving clarity requires careful balance:

• Summarize what is known (e.g., no signal of harm in hundreds of cases) while candidly stating uncertainties.
• Emphasize maternal health: e.g., “keeping the mother well is often the best way to protect the fetus” (^[15]).
• Use data from registries and literature to justify statements (“no pattern of birth defects”, “reassuring findings in preclinical studies”).
• Adopt consistent, plain language and structure so key points stand out for hurried clinicians.

Among experts, there is consensus that PLLR has overall improved the depth of information available (^[1]) (^[2]). However, many also contend that some labels have become overly verbose. Ongoing feedback loops are needed: user testing of label formats, and research (surveys, focus groups) on how physicians interpret the narratives. Tools or guidelines that apply health literacy principles to label writing (such as those advocated in regulatory science literature) could further improve communication (^[39]) (^[47]).

Case analyses (like those above) show that judiciously-worded biologic labels do a reasonable job of stating that with current evidence, most biologics do not appear grossly teratogenic while still urging informed decision-making. For instance, pregnant patients on TNF blockers can be counseled with the actual comparative figures from registries (^[32]) (^[33]), rather than a blank category “Unknown”. But much work remains to ensure all relevant products follow this model, and to gather the real-world data that will make future labels richer and more confident.

Looking ahead, the future of pregnancy risk summaries for biologics will likely be shaped by both science and communication practice. Scientifically, including pregnant populations in trials and harnessing big-data surveillance promise more robust evidence. Societally, increasing demand for patient-centered information will press for labels (and companion educational materials) that lay out both risks and context in understandable terms. Regulators, industry, and clinicians must collaborate—agreeing on best practices and sharing data—to make pregnancy labels genuinely helpful. In doing so, they can fulfill PLLR’s goal: not just replacing letters, but enabling better decisions for mothers and their babies in the real world.

References

• FDA. Pregnancy, Lactation and Reproductive Potential: Labeling for Human Prescription Drug and Biological Products – Content and Format. Guidance for Industry, Dec 2014. Link (^[1]) (^[23]).
• Pernia SL, DeMaagd GA. The new pregnancy and lactation labeling rule. Pharmacy Today. 2016 Nov;41(11):713–715. DOI:10.1016/j.clpt.2016.08.005 (^[2]) (^[17]).
• Blattner CM, Danesh M, Safaee M, et al. Understanding the new FDA pregnancy and lactation labeling rules. Int J Womens Dermatol. 2016 Feb 28;2(1):5–7. DOI:10.1016/j.ijwd.2015.12.005 (^[1]) (^[3]).
• Koh YP, Tian EA, Oon HH. New changes in pregnancy and lactation labeling: Review of dermatologic drugs. Int J Womens Dermatol. 2019 May;5(4):216–226. DOI:10.1016/j.ijwd.2019.05.002 (^[11]) (^[12]).
• Ghalandari N, Dolhain RJ, Hazes JM, et al. Pre- and post-authorization data on biologics in pregnancy and lactation (EMA analysis). Br J Clin Pharmacol. 2020 Jan;86(3):580–590. DOI:10.1111/bcp.14145 (^[16]).
• Manolis EM, Musuamba FT, Vries CS, et al. EMA perspective on model-informed drug development for pregnancy/breastfeeding. CPT Pharmacometrics Syst Pharmacol. 2024 Aug;13(11):1820–1823. DOI:10.1002/psp4.13214 (^[4]) (^[28]).
• EMA. Draft Guidelines on Inclusion of Pregnant and Breastfeeding Subjects in Clinical Trials. ICH E21 addendum (for consultation, 2024). EMA News Release (^[5]) (^[28]).
• BIO (Biotechnology Innovation Organization). Comments on Proposed Rule: Pregnancy, Lactation Labeling. May 2008. (^[20]) (^[21]).
• Nowosielski B. Time is now for new pregnancy labeling, says committee. Drug Topics. Nov 17, 2007. Link. (^[6]) (^[19]).
• Smith M, Rathach MR, Kenney MC. Medication use in pregnancy and the pregnancy and lactation labeling rule. Ther Innov Regul Sci. 2017 Apr;52(2):189–193. DOI:10.1177/2168479016674212.
• Dathe K, Schaefer C. Review of pregnancy and lactation labelling: is the current system adequate? **Pharmaceut **. 2014;129:70–74. DOI:10.1016/j.pharmthera.2010.11.011.
• Scoggins CR, Zhu Y, Seng KE, et al. Pharmacy labeling for pregnant and lactating women: a content analysis. Ther Innov Regul Sci. 2018 Jun;52(3):351–359. DOI:10.1177/2168479017749136.
• Lee S, Holden D, Webb R, Ayers S. Pregnancy-related risk perception in women, midwives & doctors: a survey. BMC Pregnancy Childbirth. 2019 Sep 27;19(1):335. DOI:10.1186/s12884-019-2467-4.
• Jacobson KL, Faughnan J, Myers L, et al. An innovative health literacy approach to improve patient understanding of medication labeling. Ther Innov Regul Sci. 2021 Aug;55(6):1180–1192. DOI:10.1007/s43441-021-00325-5 (^[39]).
• Gonalves S, Fonseca JE, Farach S, et al. Use of anti-TNF therapy during pregnancy – conclusions from a European Consortium. Rheumatology (Oxford). 2022 Jul;61(7):pub ahead of print.
• Chambers CD, Johnson DL, Dick LM, et al. Biologic agents for the treatment of rheumatic disease and pregnancy outcomes: a multicenter prospective cohort study. Arthritis Rheum. 2019 Oct;71(10):1689–1697.
• Scott MR, Moore S, Burt M, et al. TNF-inhibitor biosimilars in pregnancy: a retrospective study. Obstet Med. 2022 Nov;15(3):158–163 (^[43]).
• Chtanova T, Kennedy LG. Biologics and pregnancy: Informed decision-making benefits mothers and unborn children. Med J Aust. 2020 Mar;212(5):210–211.
• Bržica J, Najjar-Debbiny R, Lushchinina N, et al. Fetal and maternal outcomes after maternal biologic use in pregnancy: systematic review and meta-analysis. Clin Gastroenterol Hepatol. 2020 Feb;18(3):558–569. DOI:10.1016/j.cgh.2019.07.034 (^[37]) (^[38]).
• Clowse ME, Magder LS, Bingham CO 3rd, et al. Infliximab for Crohn’s and rheumatic diseases: updated pregnancy registry. Arthritis Rheumatol. 2018 Sep;70(Suppl):1399 (abstract).
• Eworuke E, Shin J, Radhakrishnan RK, et al. Placental transfer of antibodies during pregnancy. J Clin Pharmacol. 2019;59(7):856–862.
• FDA. Product Label: HUMIRA® (adalimumab) prescribing information. U.S. DailyMed, updated July 2025. (^[32]) (^[15]).
• FDA. Product Label: ENBREL® (etanercept) prescribing information. DailyMed.
• FDA. Product Label: REMICADE® (infliximab) prescribing information. DailyMed.
• FDA. Product Label: CIMZIA® (certolizumab pegol) prescribing information. DailyMed.
• Sarntivijai S, Le D, Shebelut C, He Y, Tohge T, Lee D. ReproTox Knowledge Graph: informing on-drug pregnancy and birth outcomes. NPJ Syst Biol Appl. 2024 Nov 14;10(1):71. Link.
• FDA. Biologics Regulatory Letter on Stem Cell and Gene Therapies and Pregnancy. 2021.
• Add references as needed for any quoted passages above.