Regulatory Impact of Selected U.S. FDA Postmarketing Safety Registries Conducted for Drugs Used to Treat Inflammatory or Autoimmune Conditions

ABSTRACT

Purpose

Assess the regulatory impact of selected FDA postmarketing safety registries on drug product labeling updates.

Methods

Postmarketing safety studies were identified in internal record repositories for the Center for Drug Evaluation and Research, U.S. FDA, in March and September 2021. Studies eligible for review included prospectively enrolling patient registry studies conducted to assess the safety of drug products used to treat inflammatory or autoimmune conditions. These studies were requested between 1999 and 2011.

Results

This paper analyzed 10 safety (non‐pregnancy) registries and four pregnancy registries (n = 14). Only four safety registry studies were successful in reaching their targets for both patient enrollment and patient follow‐up or drug exposure. These registries were either multi‐center, multinational studies or studies using participants from a health insurance or health maintenance organization. None of the safety registries led to safety labeling updates, regardless of targets' achievement for study enrollment and follow‐up: six did not detect a new safety signal and four provided inconclusive results. Two pregnancy registries reached their targets for patient enrollment, and all four resulted in safety labeling updates, as required by the Pregnancy and Lactation Labeling Rule guidance.

Conclusions

While six non‐pregnancy registries did not detect a new safety signal, four did not produce safety results considered sufficiently robust to warrant specific regulatory action including safety‐related labeling updates. The lack of safety signal detection in these observational studies should not imply the absence of safety signals. Appropriately designed, prospective, randomized controlled safety studies are the most reliable way to obtain interpretable safety data.

Summary.

• Main limitations include insufficient subject enrollment and/or follow‐up, absence or inadequacy of control groups, and lack of adjustment for important confounders.

• Only four of the 10 safety non‐pregnancy registries were successful in reaching their targets for both patient enrollment and patient follow‐up or drug exposure.

• Two of the four pregnancy registries reached their targets for patient enrollment.

• None of the non‐pregnancy registries led to safety labeling updates: six did not detect a new safety signal and four provided inconclusive results.

• The lack of safety signal detection in these observational studies should not imply the absence of safety signals.

1. Introduction

The U.S. Food and Drug Administration (FDA)'s mission is to ensure that safe and effective new drugs are available as quickly as possible and remain safe for as long as the drugs remain on the U.S. market [1]. To fulfill this mission, the FDA reviews data submitted by drug manufacturers with New Drug Application (NDA) or Biologics License Application (BLA) to ensure medical products are effective and safe. Since pre‐approval development programs involve only several hundred to several thousand subjects, all possible safety concerns cannot be assessed or anticipated during premarket review [2, 3, 4]. Therefore, postmarketing surveillance and risk assessment programs are often deemed necessary to identify new safety issues that may occur when a treatment is used by a larger population and for longer durations in the postmarket period [2, 3, 4]. Newly identified safety information may be used to update drug labeling, and, on rare occasions, to reevaluate the marketing decision [2, 5, 6]. New safety information may concern a serious risk, expected or unexpected, associated with the use of the drug. Several postmarketing surveillance tools have been used by the FDA for drug product safety signal detection including the FDA Adverse Event Reporting System (FAERS) [7], the active postmarket risk identification and analysis (ARIA) [8], the FDA's sentinel initiative [8, 9], and postmarketing requirements (PMRs) and postmarketing commitments (PMCs) [9, 10].

The Food and Drug Administration Amendments Act of 2007 (FDAAA), signed into law H.R. 3580 on September 27, 2007, gives authority to the FDA to require postmarketing studies or clinical trials as PMRs to investigate serious potential risk associated with the use of the drug (section 505(o) (3) of the Act) [1, 11]. FDAAA went into effect on March 25, 2008 [12]. PMRs and PMCs are observational studies or clinical trials that are conducted by the drug manufacturers after the FDA has approved or licensed a product for marketing [13], and aim to answer particular questions about a product's safety or efficacy [14]. PMCs are studies or clinical trials not required by statute or regulation, that the drug manufacturers agreed to conduct, whereas PMRs are studies or clinical trials that the drug manufacturers are required to conduct by statute or regulation [9, 10].

One type of drug safety studies conducted under PMRs and PMCs are registries, including pregnancy registries. According to FDA's Draft Guidance for Industry [15], a registry is defined as an organized system that collects clinical and other data in a standardized format for a population defined by a particular disease, condition, or exposure [15]. There are three types of registries: disease registries that use as inclusion criteria the state of a particular disease or condition; health services registries that use as inclusion criteria the exposure to a specific health care service; and product registries that use as inclusion criteria the exposure to a specific health care product [15].

A pregnancy exposure registry is a prospective observational study that actively collects information on medical product exposure during pregnancy and associated pregnancy and infant outcomes [16].

The main goals of requesting PMR and PMC safety studies are to assess the known and unknown risks, or adverse events, associated with the use of the drug product in a routine clinical practice setting. The evaluation of final study reports helps the FDA to make regulatory decisions such as updating the labeling to inform providers and patients about potential risks associated with the product use. However, these PMR and PMC safety studies are expensive and time‐consuming. Estimating 100 sites, a postmarketing safety study can easily cost a drug manufacturer $50 million [17]. Previous research conducted by other investigators have assessed postmarketing studies required to evaluate (1) drug exposure during pregnancy [18], (2) methods used to conduct prospective pregnancy exposure registries [19], and (3) postmarketing study completion rates, timeliness, results reporting, and study outcomes [14, 20, 21].

Our study assesses the regulatory impact of long‐term PMR and PMC safety studies on safety labeling update, for drug products used to treat inflammatory or autoimmune conditions, approved by Divisions of Dermatology, Rheumatology, and Gastroenterology.

2. Methods

2.1. Data Sources and Search Strategy

We identified 1125 PMRs and PMCs by conducting a database search in the Center for Drug Evaluation and Research (CDER), U.S. Food and Drug Administration (FDA), Document Archiving, Reporting and Regulatory Tracking System (DARRTS). This search was conducted on March 20, 2021, using a customizable report programmed by the CDER Office of Business Informatics (OBI) in SAP Enterprise Business Informatics. We also identified 522 completed PMR and PMC safety studies and trials by separate database search conducted by the CDER's Analytics and Data Services Staff (ADSS), in September 2021.

2.2. Data Selection

The criteria used to identify eligible PMR and PMC studies included: (1) biologic product, used for treatment of inflammatory or autoimmune condition, approved by Divisions of Dermatology, Rheumatology, or Gastroenterology or a non‐biologic product of special interest to the above divisions; (2) completed postmarket registry with a final study report reviewed by the FDA; and (3) registry with patients enrolled prospectively.

All registries meeting the above inclusion criteria were included, regardless of the countries where they have been conducted and regardless of the time period when they have been conducted.

2.3. Data Abstraction and Analysis

We abstracted and summarized the following information from study reports, protocols, statistical analysis plans, FDA reviews, product labeling, and related internal and external communications: study objectives and outcomes; study design; total patient enrollment, duration of drug exposure, and patient follow‐up; study findings and conclusions; and FDA assessment of the study findings, strengths or limitations, as well as recommendations for safety labeling update.

3. Results

Database searches identified 14 registry studies conducted as either a PMR or PMC that met the inclusion criteria.

3.1. Distribution of Registries and Safety Outcomes of Interest

Ten of the fourteen registries were safety (non‐pregnancy) registries and four were pregnancy registries. The PMRs/PMCs included in this assessment were issued between 1999 and 2011 for safety registries and between 2002 and 2005 for pregnancy registries. Nine of the fourteen registries (five safety registries and four pregnancy registries) addressed the safety of Tumor Necrosis Factor Alpha (TNF‐α) Inhibitors. The remaining safety registries addressed the safety of a T Lymphocyte Activation Inhibitor (n = 2), Interleukin (IL) 12 and 23 Antagonist (n = 1), Calcineurin Inhibitor (n = 1), and Proton Pump Inhibitor (n = 1).

The most common safety outcomes of interest in safety registries were malignancies (eight studies), followed by infections (seven studies), cardiovascular events (four studies), and autoimmune diseases (four studies). A total of eight drug products were studied in these 14 registries, including: Etanercept (3 registries), Infliximab (3 registries), Abatacept (2 registries), Adalimumab (2 registries), Certolizumab Pegol (1 registry), Pantoprazole Sodium (1 registry), Tacrolimus (1 registry), and Ustekinumab (1 registry). The approved indications for each of these drug products are provided in Appendix A. None of these eight drug products has a Risk Evaluation and Mitigation Strategy (REMS). Table 1 provides the description of the 14 registries, including: study purpose, objectives, safety outcomes of interest, study time‐period (date of first patient enrollment to the end of subject follow‐up or end of data collection), and source of data collection.

TABLE 1.

Description of studies: Purpose, objectives, safety outcomes of interest, study time‐period, and source of data collection.

Study	Purpose	Objectives	Safety outcomes of interest (SOI)	Study Time‐period a	Source of data collection
Non‐pregnancy registries (n = 10)
Abatacept (Orencia)—IM101045A	Quantify the incidence of pneumonia, opportunistic infections, and all hospitalized infections in abatacept exposed rheumatoid arthritis patients.	Estimate the incidence rates of infection in three groups of subjects who initiate: abatacept other Biologic Disease‐Modifying Drugs (BDMs) non‐biologic Disease‐Modifying Anti‐Rheumatic Drugs (DMARDs).	Hospitalized pneumonia or IV antibiotic‐treated pneumonia; tuberculosis; herpes zoster; opportunistic infection (with and without herpes zoster); hospitalized infection.	December 1, 2005 to April 30, 2015	Administrative database from a commercial health insurance: the Optum Research Database (ORD)
Abatacept (Orencia)—IM101045B	Assess risks of malignancies and infection in patients with rheumatoid arthritis (RA) treated with Abatacept in comparison to other DMARDs within an existing registry containing RA patients.	Estimate and compare: Risk of malignancies among a cohort of RA patients receiving abatacept and those receiving other RA treatments. Risk of infections among a cohort of RA patients receiving abatacept and those receiving other RA treatments. Risk of autoimmune disease among a cohort of RA receiving abatacept and those receiving other RA treatments.	Malignancies (all malignancies, lung cancer, lymphoma, breast cancer, non‐melanoma skin cancer); infections (hospitalized infections, pneumonia, opportunistic infections, and tuberculosis); autoimmune diseases (lupus, Multiple Sclerosis and psoriasis).	July 1, 2005 to June 4, 2016	National Data Bank for Rheumatic Diseases (NDB)
Adalimumab (Humira)—P06‐134	Evaluate long‐term safety and effectiveness of adalimumab (in routine clinical practice) in subjects with moderate to severe Crohn's disease (CD), who have had an inadequate response to conventional therapy or have lost response to or are intolerant to infliximab.	Evaluate the long‐term safety of Humira in adult patients with CD who were treated as recommended in the local product label.	Infections, malignancies, immune reactions, demyelinating disorders, interstitial lung disease, cardiovascular events, gastrointestinal events, hematologic disorders, hepatic events, injection site reactions, skin/subcutaneous disorders, pulmonary embolism, nervous system disorders and medication administration related medication errors. Treatment Emergent malignancy adverse events of special interest (AESIs) other than non‐melanoma skin cancer (NMSC) and NMSC AESIs	September 5, 2007 to February 4, 2016	Adult patients with moderately to severely active Crohn's disease (CD), from centers in the United States, Canada, the European Union, South Africa, Australia, and New Zealand, who were newly prescribed Humira, as well as those who were enrolled in CD clinical trials for Humira.
Certolizumab Pegol (Cimzia)—SECURE (C87075)	Evaluate known safety risks associated with TNF blocker class in CD patients: serious infections, including opportunistic infections; lymphoma and other malignancies; demyelinating disorders and autoimmune disorders.	Measure the long‐term safety outcomes of Certolizumab Pegol (CZP) compared to other Crohn's disease (CD) treatments when used in customary clinical practice.	Malignancy events including malignant tumors and malignant or unspecified tumors. Serious infections, opportunistic infections, serious cardiovascular events, congestive heart failure, demyelinating‐like disorders, serious bleeding events, lupus and lupus‐like illness, serious skin reactions, aplastic anemia, pancytopenia, thrombocytopenia, neutropenia, and leucopenia.	January 30, 2009 to August 28, 2020	Patients (in the US) with medically documented CD who were prescribed CZP or other CD treatments in accordance with customary clinical practice.
Etanercept (Enbrel)—20 021 626	Need to evaluate longer‐term post‐approval safety and effectiveness of etanercept in children with polyarticular or systemic Juvenile Idiopathic Arthritis (JIA).	Determine the long‐term safety of etanercept (with or without other DMARDs) in pediatric patients with polyarticular course or systemic JIA compared to a control cohort of patients with polyarticular course or systemic JIA receiving methotrexate (MTX) (with or without other DMARDs).	Adverse events Serious adverse events Autoimmune diseases Malignancies Medically important infections Growth rates Tanner scores	April 1, 2000 to January 31, 2008	Multicenter study of subjects (2 to 18 years old) with polyarticular course or systemic JIA receiving etanercept alone or in combination with methotrexate or other DMARDs, and those receiving methotrexate alone or in combination with other DMARDs
Etanercept (Enbrel)—20 040 210	Proactively monitor the potential long‐term risk associated with etanercept therapy in a large number of subjects with plaque psoriasis	Assess the long‐term safety of subjects with plaque psoriasis treated with Enbrel as determined by evaluation of incidence rates of serious adverse events and serious infectious events.	All cancers excluding non‐melanoma skin cancer, lymphomas, non‐melanoma skin cancer, hospitalized infectious events.	May 31, 2006 to December 4, 2012.	Multi‐center, observational, surveillance registry of U.S. (338 centers) and Canadian (37 centers) subjects with chronic plaque psoriasis treated with etanercept in routine clinical setting.
Infliximab (Remicade‐ C0168Z02‐ Dose Escalation	Evaluate whether trough concentrations at the time of loss of clinical response can be used to identify pediatric ulcerative colitis (UC) and CD patients who have low infliximab exposures and would benefit from a dose increase above that approved without increasing risk of serious adverse events.	Evaluate the efficacy, safety, and pharmacokinetics of Remicade among pediatric patients with CD or UC who had an escalated dose from the currently approved dose of 5 mg/kg q8w to 10 mg/kg q8w (defined as a dose ≥ 9 to ≤ 11 mg/kg).	All adverse events, including those of special interest, such as dysplasias and malignancies of all types, serious infections, and new autoimmune disease.	May 31, 2007 to June 30, 2018	Sub‐study of DEVELOP (multicenter, prospective, long‐term observational registry of the safety and clinical status of pediatric patients with inflammatory bowel disease (IBD) treated with Remicade and/or other medical therapies.
Pantoprazole Sodium (Protonix)—3001A1‐100 034	Evaluate whether long‐term use of pantoprazole is associated with an increased risk of cancer. [In rodents, all proton pump inhibitors (PPIs) were carcinogenic and caused rare types of gastrointestinal tumors. Overall, some studies suggested pantoprazole produced a somewhat greater incidence and variety of tumors than did the other PPIs.]	Compare incidence rate of gastric cancer after long‐term exposure to pantoprazole with the incidence rate of cancer after comparable exposure to other PPIs adjusting for age, sex, and other factors.	Gastric cancer, certain digestive system cancers (colon, small intestine, pancreas, and liver), overall cancer (excluding non‐melanoma skin cancers (NMSC)).	June 21, 2004 to March 1, 2013	Patients from the Kaiser Permanente Northern California (KPNC) health maintenance organization.
Tacrolimus (Protopic)—APPLES	Answer a research question: Does topical treatment of children with tacrolimus ointment increase the long‐term risk of malignancy?	Assess the long‐term safety of tacrolimus ointment (0.03% or 0.1%) for the treatment of children (under 16 years old) with atopic dermatitis under actual use conditions, including the risk of developing cutaneous or systemic malignancies.	Hodgkin and non‐Hodgkin lymphoma; cutaneous malignant melanoma; non‐melanoma skin cancer; all cancers	March 28, 2005 to January 31, 2019	APPLES (A Prospective Pediatric Longitudinal Evaluation to Assess the Long‐Term Safety of Tacrolimus Ointment for the Treatment of Atopic Dermatitis)
Ustekinumab (Stelara)‐ EDMS‐ERI‐70976995	Update analyses on Major Adverse Cardiovascular Event (MACE) and thrombotic event profile of STELARA. This study addressed a concern of cardiac adverse events with a drug of the same class: Briakinumab, IL‐12 and IL‐23 antagonist.	Evaluate the safety of Ustekinumab for three outcome Cardiovascular (CV) categories (MACE, other thrombotic events, and other CV events) in three data sources: clinical studies (pooled data from 14 clinical trials), the Psoriasis Longitudinal Assessment and Registry (PSOLAR), and Janssen's global postmarket safety data.	MACE: CV death, nonfatal myocardial infarction [MI], and nonfatal stroke. Other nonfatal thrombotic events: unstable angina, venous thromboembolic disease, and peripheral arterial thrombosis. Other nonfatal CV events: heart failure (HF) and arrhythmia requiring intervention.	March 24, 2009 to August 23, 2013	The data source considered for this paper was the ongoing Psoriasis Longitudinal Assessment and Registry (PSOLAR) study.
Pregnancy registries (n = 4)
Infliximab (Remicade‐C0168T71	Assess pregnancy outcomes in women who were exposed to Infliximab during pregnancy relative to background risk in similar patients not exposed to Infliximab.	Primary objectives Collect and analyze information on: Pregnancy outcomes in women exposed to Remicade during pregnancy, relative to the background risk in non‐biologic exposed patients. Health status during the first year following delivery, of infants born to women following prenatal exposure to Remicade and their non‐biologic exposed counterparts.	Pregnancy outcomes: live birth, normal; live birth, abnormal: preterm birth, small for gestational age/intrauterine growth retardation, malformations, perinatal morbidity; fetal death: stillbirth versus neonatal death Infant outcomes: Congenital malformations; infections; neonatal/infant neutropenia; hospital care encounters; one‐year mortality.	January 2000 to December 2014	Data captured from the Swedish Medical Birth Register (SMBR), the Danish Medical Birth Register (DMBR), and the Finnish Medical Birth Register (FMBR).
Infliximab (Remicade‐ PRIORITY	Established with the aim to complement data obtained from the infliximab C0168T71 registry study, which has low enrollment in the infliximab exposed cohort.	Primary objectives Collect and analyze information on: Pregnancy outcomes in women with IBD who are exposed to Remicade during pregnancy, relative to the background risk in women with IBD who are exposed to nonbiologic systemic therapy during pregnancy, and Health status during the 4 years following delivery, of infants born to women following prenatal exposure to Remicade and their nonbiologic systemic therapy exposed counterparts.	Nonlive birth outcomes: induced abortion, spontaneous abortion, still birth, neonatal death; Pregnancy outcomes: low birth weight, small for gestational age (SGA), preterm birth [< 37 weeks] very preterm birth [< 32 weeks], congenital malformations; Selected infant outcomes: hospitalization for sepsis, hospitalization for antibiotics.	January 1, 2007 to March 19, 2013	Used data PIANO (Pregnancy in Inflammatory Bowel Disease And Neonatal Outcomes) pregnancy registry for all women with IBD (including CD, UC, and indeterminate colitis [IC]) conducted by the Crohn's and Colitis Foundation of America (CCFA) Clinical Alliance.
Etanercept (Enbrel)—OTIS‐20040246	Proactively monitor the potential risk associated with etanercept therapy in pregnant women and pregnancy outcomes	Evaluate the effect of etanercept when used in the first trimester of pregnancy with respect to: Major structural birth defects of newborns (primary objective). Potential minor malformations, spontaneous or elective abortion or stillbirth, preterm delivery, pre‐ and post‐natal fetal and infant growth, developmental milestones, malignancies, and serious or opportunistic infections reported in the first year of life (secondary objectives).	Overall proportion of major structural defects, frequency and patterns of minor malformations, spontaneous and elective abortion or stillbirth, preterm delivery, pre‐ and postnatal fetal and infant weight, length, and head circumference, and in the first year of life, developmental milestones, malignancies, and serious or opportunistic infections.	April 1, 2005 to June 13, 2014	‐ Women with RA, juvenile RA, ankylosing spondylitis (AS), psoriatic arthritis (PsA), and psoriasis (PsO) recruited from the United States and Canada. ‐ Non‐diseased controls from historical data on pregnancies followed through the California Teratogen Information Service's Clinical Research Program between August 1993 and December 2008.
Adalimumab (Humira)—OTIS‐M03604	Monitor planned and unplanned pregnancies in women exposed to adalimumab (ADA) to evaluate the possible effects of ADA on pregnancy outcome. The lack of human fetal safety data for ADA made this monitoring system an important component of epidemiologic research on the safety of this drug.	Estimate the risk of major birth defects and other adverse pregnancy outcomes in offspring of women exposed to ADA during pregnancy; Detect any pattern of these outcomes compared to study‐generated disease‐matched and non‐diseased unexposed groups; Compare the rate of birth defects overall to external data from the CDC Metropolitan Atlanta Congenital Defects Program (MACDP) population‐based birth defects surveillance program.	Primary SOI: Major Birth Defect (MBD). Secondary SOI: Minor malformation. Spontaneous abortion (fetal loss at less than 20 completed weeks' gestation). Preterm delivery (live born infant prior to 37 completed weeks' gestation). Birth size and postnatal growth. Serious and opportunistic infections.	January 1, 2004 to August 15, 2016.	Pregnant women with RA or CD, exposed to ADA between the date of conception and the end of the 12th week post‐conception and resided in the US or Canada. Registry conducted by the Organization of Teratology Information Specialists (OTIS).

^a Date of first patient enrollment to the end of subjects' follow‐up or end of data collection.

3.2. Study Enrollment

The number of subjects exposed to the treatment drugs ranged from 262 to 34 178 in safety registries and from 257 to 370 in pregnancy registries.

Overall, the enrollment goals were pre‐specified in eight of the 10 safety registries and all four pregnancy registries. Six safety registries and two pregnancy registries reached their pre‐specified enrollment goals. Table 2 provides detailed information on enrollment, including enrollment targets, actual enrollment, and reasons for not reaching the enrollment target.

TABLE 2.

Evaluation of study enrollment, drug exposure, and study follow‐up.

Study	Target enrollment (actual enrollment)	Target drug exposure in PYs (actual exposure)	Target follow‐up in PYs (actual follow‐up)	Reasons for not reaching study targets (Enrollment—Drug exposure—Follow‐up)
Non‐pregnancy registries (n = 10)
Abatacept (Orencia)—IM101045A	Abatacept initiators: 1200 (1268)		Abatacept: 2000 (2869)
Abatacept (Orencia)—IM101045B	Abatacept: 5000 (1496) DMARD a : 7500 (1520) BDM b : 7500 (3490)	Abatacept: (2502) DMARD: (4330) BDM: (9658)	Abatacept: 25000 (4896) DMARD a : 37500 (4816) BDM b : 37500 (11777)	Enrollment: Lack of prescribing by rheumatologists due to increasing choices of biologics and other treatments; lack of insurance coverage; lack of support and resources to enroll patients when first prescribed or when obtaining the medication.
Adalimumab (Humira)—P06‐134	5000 (5025)	15 180 (16680)
Certolizumab Pegol [CZP] (Cimzia)—SECURE (C87075)	Certolizumab: 2000 (1131) Other treatments: 2000 (1666) Treatment gap: (275)		Certolizumab: 20000 (5204.2) Other treatments: 20000 (8209.9) Treatment gap: (1118.0)	Enrollment challenges: Few CZP patients, closure of recruitment sites due to lack of CZP‐treated patients, staff availability and resources, difficulty finding patients meeting the inclusion/exclusion criteria, and lack of insurance coverage (patients have to fail other medications before insurance will cover CZP). Enrollment and Follow‐up: The Sponsor, in agreement with the FDA, ceased enrollment earlier than scheduled, and modified the follow‐up period to a minimum of 2 years. Due to challenges with enrollment, high patient discontinuation rate, and frequent treatment switching, enrollment of additional patients will not lead to an improvement in the malignancy assessment. In the original protocol, each patient was to be followed for a minimum of 10 years.
Etanercept (Enbrel)—20 021 626	Etanercept arm: 400 (397) Methotrexate arm: 200 (197)	Etanercept arm: (859) Methotrexate arm: (388)
Etanercept (Enbrel)—20 040 210	2500 (2510)	(6167.4)	7564 to 9725 (8912.8)
Infliximab (Remicade)—C0168Z02‐Dose Escalation	(262)		(493)	No sample size calculations were performed: This was a dose escalation study, and all patients meeting the pre‐defined dose escalation criteria were included in the study.
Pantoprazole Sodium (Protonix)—3001A1‐100 034	Pantoprazole: 24000 (34178) Other PPI c : 20000 (27686)		[primary outcome of gastric cancer] Pantoprazole: 163755 (206000) Other PPI c : 52488 (59000)
Tacrolimus (Protopic)—APPLES	8000 (7954) [8071 subjects enrolled, with 7954 meeting inclusion criteria.]		80 000 (45279)	Follow‐up: Earlier discontinuation of study due to the futility of gaining additional meaningful information from the cohort in the allotted timeframe. A large number of subjects was lost to follow‐up, and the Applicant did not expect that continuing the study would provide additional information to determine the risk of cancer in the pediatric population. End of subject follow‐up was modified from December 2022 to January 31, 2019.
Ustekinumab (Stelara)—EDMS‐ERI‐70976995	Ustekinumab (1493) Other Biologic (1010) Non biologic (2152)	Ustekinumab (2453) Other Biologic (1548) Non biologic (5172)	Ustekinumab (2453) Other Biologic (1548) Non biologic (5172)	No sample size calculations were performed: the study used subjects already enrolled and followed‐up in the ongoing Psoriasis Longitudinal Assessment and Registry (PSOLAR) study.
Pregnancy Registries (n = 4)
Infliximab (Remicade)—C0168T71	Infliximab: 370 (270) Non‐biologic: 2000 (6460)			Enrollment in infliximab cohort: The European Summary of Product Characteristics (EU SmPC) has stricter language on the use of Remicade during pregnancy as compared to other regions. Physicians in the Nordic countries tend to adhere strictly to the prescribing information and discontinue treatment with Remicade before pregnancy. Remicade is a hospital administered drug that is not well captured in the prescribed drug registries, making it difficult to identify patients exposed to Remicade. Through 2010, only Sweden and Denmark agreed to participate. Although Finland was added in 2011, the number of Finnish Remicade exposed cases through October 15, 2012 has been only 4.
Infliximab (Remicade)—PRIORITY	Infliximab cohort: 230 (294) Non‐biologic cohort: 506 (515)
Etanercept (Enbrel)—OTIS‐20040246	Etanercept cohort: 300 (370) Diseased control cohort: 300 (164) Non‐diseased unexposed historical cohort: 300 (296)			Enrollment in diseased control cohort: With the proportion of women using a tumor‐necrosis factor inhibitor increasing throughout the years of the study, it became challenging to enroll women without use of a TNF‐inhibitor within 3 months of the last menstrual period into the comparison group.
Adalimumab (Humira)—OTIS‐M03604	Adalimumab‐exposed group: 120 (257) Diseased unexposed group: 100 (120) Non‐diseased unexposed group: 120 (225)

^a DMARD: non‐biologic disease‐modifying anti‐rheumatic drug.

^b BDM: other biologic disease‐modifying drug.

^c PPI: proton pump inhibitor.

3.3. Study Follow‐Up

Four safety registries lasted 2 to less than 5 years; four safety and three pregnancy registries lasted 5 to less than 10 years; and two safety registries and one pregnancy registry lasted for at least 10 years.

All 10 safety registries reported the total person‐years (PYs) of follow‐up time (eight studies) and/or drug exposure (five studies). Seven safety registries had pre‐specified targets for PYs of follow‐up time (six studies) or drug exposure (one study), and only four of them reached their targets for PYs of follow‐up time (three studies) or drug exposure (one study). Table 2 provides detailed information on study enrollment, drug exposure, and study follow‐up.

3.4. Regulatory Action Regarding Safety Labeling Update

Final reports from 14 registry studies fulfilled FDA PMRs and PMCs. However, none of the 10 safety registries reports led to safety‐related labeling update, while all four pregnancy registries resulted in labeling update (labeling section 8.1 Pregnancy). Six of the 10 safety registries did not detect a new safety signal and four provided inconclusive results. The information included in the labeling section 8.1 pregnancy, for each of the four pregnancy registry studies, is provided in Appendix B.

FDA assessments of final reports of studies are provided in Tables 3 and 4. Table 3 includes four safety registries that achieved their targets for both patient enrollment and study follow‐up or drug exposure, and two pregnancy registries that achieved their targets for patient enrollment. Table 4 includes safety registries that did not achieve their targets for both patient enrollment and study follow‐up or drug exposure (n = 3), safety registries without pre‐specified targets (n = 3), and pregnancy registries that did not achieve their targets for patient enrollment (n = 2).

TABLE 3.

FDA assessment of final reports of studies that achieved their targets (n = 6): Main limitations and conclusions.

Study	Main limitations from FDA review	FDA review conclusions
Non‐pregnancy registries (n = 4)
Abatacept (Orencia)—IM101045A	Potential outcome misclassification, possible surveillance bias, possible residual confounding in estimating potential treatment effects, Relative Risks (RRs) not estimated for the specified autoimmune disorders.	Within the study limitations, the data suggest that the treatment drug is not associated with an increased risk of hospitalized infections including pneumonia or the specified autoimmune disorders.
Adalimumab (Humira)—P06‐134	Absence of control groups; lacking information from long‐term follow‐up; rates of lymphoma and malignancy derived from a sample of patients with 57% having at least one prior TNFα/biologic use and 53% using concomitant immunosuppressants or systemic corticosteroids at baseline; no distinction made between concomitant use of thiopurine at baseline versus thiopurine use in the past; only 25% of subjects were U.S. patients.	Within study limitations, the results align with information presented in the approved label, with respect to the frequency of malignancy in patients treated with tumor necrosis factor alpha (TNF‐α) inhibitors.
Etanercept (Enbrel)—20 040 210	No internal comparator (study not designed to estimate risk of adverse outcomes relative to non‐treatment groups); use of retrospective external claims data for comparison to prospective registry; enrollment of prevalent users; high dropout rate of 42%; registry outcomes not verified (potential outcome ascertainment bias).	No new safety concerns identified; however, the study limitations diminish the generalizability of the results and limit the results for labeling.
Pantoprazole Sodium (Protonix)—3001A1‐100 034	Lacked ability to detect due to the absence of a non‐proton pump inhibitor (PPI) comparator; results susceptible to exposure misclassification biases both toward or away from the null; results susceptible to detection bias as clinical practices changed; failed to adequately describe regression analysis covariate inclusion criteria; differences observed in the 1% manual chart abstraction of potential risk factors that may confound the results; unclear definition for censoring based on membership coverage; information lacking on data transformations that were done and on what variables.	While the study did not find an increased risk of cancers, it was difficult to conclusively determine whether the treatment does not increase the risk of cancers due to limitations.
Pregnancy registries (n = 2)
Infliximab (Remicade)—PRIORITY	The lack of adjustment for important confounders such as concomitant use of other medications or treatments, body mass index (BMI), and disease severity may have biased the analyses for congenital malformations; the study involved multiple comparisons, for multiple subgroups and multiple outcomes, but there was no control for multiple comparisons; it is not possible to know whether knowledge about the viability of pregnancy influenced enrollment in the study (volunteer bias); the completeness of the data on pre‐pregnancy and early pregnancy remains unclear since there is no information on the gestational age at registry enrollment (selection bias); and it is not possible to comment on the validity of the data, because outcomes were based on patient reports.	The study may have lacked power to detect a statistically significant difference in the risk of low birth weight, as the study OR estimate was only 1.4. Many of the non‐biologics therapies that form the active control are known to be teratogenic; thus, a statistically significant lower risk of congenital malformations relative to non‐biologic may not demonstrate that the drug is safe for use in pregnancy. The methodological limitations hinder interpretation of the study results.
Adalimumab (Humira)—OTIS‐M03604	The study was underpowered for all study endpoints, except for head circumference at 1 year; for the primary endpoint of Major Birth Defects (MBDs), the study was only powered to detect a statistical significance for an odd ratio (OR) of 3 or higher; therefore, for other clinically relevant risk estimates of < 3.0, the absence of a statistically significant result cannot be used to infer the absence of risk; the differences in baseline characteristics between the two groups were too large for reliable comparisons; the exposure definition is subject to misclassification; and possibility of selection bias due to differences in gestational age at enrollment between the exposed and unexposed groups.	Overall, the study's methodology to investigate the association between drug exposure and adverse pregnancy and infant outcomes is reasonable. The registry was well‐conducted with numerous design strengths, including rigorous ascertainment of malformations; however, there are several concerns related to the interpretation of the results.

TABLE 4.

FDA assessment of final reports of studies that did not achieve their targets (n = 8): Main limitations and conclusions.

Study	Main limitations from FDA review	FDA review conclusions
Non‐pregnancy registries (n = 6)
Abatacept (Orencia)—IM101045B	Hazard Ratios (HRs) not estimated for lymphoma, breast cancer, lung cancer, and multiple sclerosis.	Within study limitations, the data suggest that the treatment drug is not associated with an increased risk of hospitalized infections including pneumonia, overall malignancies, non‐melanoma skin cancer, other specific cancers, or the specified autoimmune disorders.
Certolizumab Pegol (Cimzia)—SECURE (C87075)	Frequent treatment switching and use of multiple therapies; high discontinuation rate & limited follow‐up; inclusion of subjects who failed to meet inclusion/exclusion criteria and subjects who did not receive treatment within 2 months of study enrollment; inclusion of prevalent users; inclusion of patients with a prior cancer history without accounting for their increased susceptibility for the primary outcome of interest; not accounting for the correlation that occurred in the Incidence Rate Ratio (IRR) calculation; Enrolled Set populations containing patients less relevant to the study; potential Class Effects (patients who received other TNF‐α in combination with the treatment drug could have been at a greater risk than reported).	Study could not provide definitive evidence nor support manufacturer's conclusion of no evidence of an increased risk of malignancies with long‐term drug exposure due to the descriptive nature of the study and the numerous limitations arising from the study conduct to the analysis.
Etanercept (Enbrel)—20 021 626	Assuming power was 0.8 and alpha was 0.05, the hypothetical cohort study would require 103 000 pediatric patients (over 51 000 patients per group) to detect a twofold increase in cancer incidence in this population. However, only ~10 000 pediatric patients have received the treatment drug since its initial marketing (over a 10‐year period).	Study was not designed specifically for the purpose of studying the risk of cancer, and a post‐marketing study to rule out a greater risk of malignancies would be highly impractical.
Infliximab (Remicade)—C0168Z02‐ Dose Escalation		No new safety signals or increased risk of adverse events/serious adverse events were associated with dose escalation, and the data suggest that dose escalation from 5 mg/kg to 10 mg/kg may be safely performed.
Tacrolimus (Protopic)—APPLES	Combination of attrition bias and early termination resulted in an analysis that lacked the power to examine potential relationships in more detail and the inability to capture cancers with longer latency periods. 45 279 patient‐years (PYs) of follow‐up time reached versus 80 000 PYs expected; attrition bias: 18.3% of patients withdrew consent from the study and another 54.9% were lost to follow‐up; the use of national cancer registries to calculate the expected number of malignancies instead of enrolling an unexposed control group limited the ability to account for potential confounding beyond age, sex, and race, and that may bias the results away or towards the null.	While the study did not detect an excess cancer risk, it has not ruled it out and provides limited contributions to the overall understanding of the drug on cancer risk, due to the lack of the power to examine potential relationships in more detail, and the inability to capture cancers with longer latency periods.
Ustekinumab (Stelara)—EDMS‐ERI‐70976995		Study results indicate no statistically significant risk, and the benefit/risk profile remains favorable.
Pregnancy registries (n = 2)
Infliximab (Remicade)—C0168T71	Some potential important confounders were not available in the data: disease severity, alcohol or recreational drug use, concomitant medications use, and socioeconomic status; in the U.S. study, patient reported outcomes were collected without clinical validation; the study involved multiple comparisons, for multiple subgroups and multiple outcomes, but there was no control for multiple comparisons.	The study findings indicate no increased risk for several outcomes including major malformations; however, this study is not definitive in terms of ruling out risk. Many of the non‐biologics therapies that form the active control are known to be teratogenic; thus, a statistically significant lower risk of congenital malformations relative to non‐biologic may not demonstrate that the drug is safe for use in pregnancy. The methodological limitations hinder the interpretation of the study results.
Etanercept (Enbrel)—OTIS‐20040246	Because teratogen exposure information was obtained via patient interview, a contribution in the observed trend due to recall bias cannot be ruled out; the low enrollment rate impaired the ability to determine stratified risk estimates by primary disease, and the small event rate may have resulted in wide confidence intervals for the OR estimates in every analysis method and large variability in OR estimates across different statistical methods.	The study was well‐conducted; however, FDA disagreed with the manufacturer's adjustment for covariates. Even though the results may raise concern that the treatment drug is associated with Major Congenital Malformation (MCM), the totality of the data do not reliably support that the treatment drug causes MCM. The study's results, albeit statistically insignificant, should be included in the product labeling with the addition of the limitations of the registry approach.

4. Discussion

In the current study, except for four pregnancy registry studies that resulted in labeling update, none of the 10 safety non‐pregnancy registry studies led to safety‐related labeling updates, regardless of the targets' achievement status for study enrollment and follow‐up: six did not detect a new safety signal and four provided inconclusive results.

4.1. Study Enrollment and Follow‐Up

Recruitment and retention of participants are critical elements in postmarketing drug safety studies [3, 22], and the number of subjects enrolled and followed‐up will determine the statistical power of these studies [23]. A sufficient number of subjects should be enrolled and followed‐up for a sufficient amount of time to make the safety findings statistically relevant [3, 17, 22, 23].

In the current assessment, one‐third of the registries (two safety and two pregnancy) with pre‐specified enrollment targets did not reach their targets due to challenges with enrollment. Specific enrollment challenges included: difficulty finding patients meeting the inclusion/exclusion criteria, lack of insurance coverage, lack of prescribing due to increasing choices of biologics and other treatments for the conditions of interest, lack of support and resources to enroll patients, hospital‐administered drug not well captured in the prescribed drug registries. Challenges with enrollment were also highlighted by manufacturers, in the literature, as their participating sites faced challenges with recruiting sufficient numbers of patients [4].

Two safety studies did not have pre‐specified enrollment targets because sample size calculations were not performed for those studies. One study was a dose escalation study in which all patients meeting the pre‐defined dose escalation criteria were included. The other study used subjects already enrolled and followed‐up in an ongoing study, the Psoriasis Longitudinal Assessment and Registry (PSOLAR) study.

In this analysis, 10 of the studies (six safety and four pregnancy registries) lasted for at least 5 years. However, only four safety registries reached their targets for PYs of follow‐up time or drug exposure. These four studies also reached their targets for patient enrollment. They were either multicenter, multinational studies (two studies) or studies using participants from health insurance or health maintenance organization (two studies). In the current analysis, three safety registries did not reach their pre‐specified targets for PYs of follow‐up time. The main reasons included: low patient enrollment, high patient discontinuation or lost to follow‐up rate, and early termination of study due to the futility of gaining additional meaningful information from the cohorts in the allotted timeframe. A concern with patient follow‐up in postmarketing safety studies has also been reported in the literature, as keeping patient participation and sustaining studies over a long enough time period, to confirm our understanding of or to better characterize a product safety profile, have been challenging for manufacturers [17].

4.2. Regulatory Action Regarding Safety Labeling Update

In the current assessment, none of the eight drug products of interest has a Risk Evaluation and Mitigation Strategy (REMS). None of the safety non‐pregnancy registries led to safety‐related labeling updates, regardless of the targets' achievement status for study enrollment and follow‐up: six of the studies did not detect a new safety signal and four provided inconclusive results. Two of the six safety registries that did not detect new safety signals had major limitations, even though they have achieved their targets for both patient enrollment and study follow‐up time or drug exposure. The limitations in one of these registries included: absence of a control group, use of concomitant treatments in 53% of the study subjects, and lacking information from long‐term follow‐up. The major limitations in the other registry study included: inadequacy of a control group (use of retrospective external claims data for comparison to the prospective exposure registry), high dropout rate of 42% over the study period, and lack of verification of the registry outcomes.

Four of the safety registries were inconclusive regarding malignancy risks. The main reasons were: (1) lack of power to detect cancers of intertest due to insufficient subjects receiving the treatment drug; (2) lack of power to examine potential relationships in more detail and inability to capture cancers with longer latency periods due to attrition bias and insufficient follow‐up; (3) inadequate comparator, exposure misclassification, detection bias due to changes in clinical practices, and lacking information from the analyses; and (4) insufficient patient enrollment and follow‐up, inclusion of subjects less relevant to the study, use of concomitant treatments, and not accounting for a prior cancer history for the primary cancer of interest.

Overall, the main limitations in safety non‐pregnancy registries were the studies' inabilities to recruit and/or retain a sufficient number of relevant subjects. In fact, two of the eight safety registries with pre‐specified enrollment targets did not reach their targets, and three of the seven safety registries with pre‐specified targets for PYs of follow‐up time or drug exposure did not reach their targets. Not reaching the targets for patient enrollment and targets for patient follow‐up time or drug exposure are not the only deficiencies that may preclude studies from providing robust data. The absence or the inadequacy of a comparator was another critical limitation that precluded reliable interpretation of data from these safety registries. Four safety registries did not include either a control group (one study) or an adequate control group (three studies). The other important limitations in non‐pregnancy registries included: frequent treatment switching and use of concomitant treatments, exposure or outcome misclassification, confounding, and selection bias. Four of the safety non‐pregnancy registry studies were successful in reaching their targets for both patient enrollment and patient follow‐up or drug exposure; however, none of them led to a safety‐related labeling update.

In pregnancy exposure registries, data from all four registries' final reports had been included in the labeling. Since pregnant women are excluded from clinical trials that support approval of most drugs and biologics, the pre‐approval data contain no or limited information on the safety of the drugs or biologics use during pregnancy in humans, except those developed to treat conditions unique to pregnancy [23, 24, 25, 26]. Therefore, pregnancy registry data have contributed to labeling changes and clinical guidelines [24]. According to the FDA Pregnancy, Lactation, and Reproductive Potential: Labeling for Human Prescription Drug and Biological Products Guidance (draft, July 2020), “when a registry is closed or there are changes to the contact information of an existing registry, the labeling must be updated” [27].

Like safety non‐pregnancy registries, the inability to recruit or retain a sufficient number of relevant subjects was an important limitation in pregnancy registries, as half of these pregnancy registries did not reach their enrollment targets due to challenges with enrollment. Other significant limitations in pregnancy registries included: concomitant use of other medications or treatments; lack of power to detect a statistically significant risk difference; small event rate; lack of adjustment for important confounders such as disease severity, body mass index (BMI), alcohol or recreational drug use, and socioeconomic status; lack of control for multiple comparisons; and selection bias.

The FDA “Fiscal Year (FY) 2021” report in the Federal Register regarding the performance of postmarket studies and clinical trials required or requested by the FDA, indicated that 324 PMRs and PMCs were closed (fulfilled or released) within the FY 2021 (October 1, 2020 to September 30, 2021) [13]. Our findings from March 2021 and September 2021 PMRs/PMCs searches (n = 14) do not reflect the FDA annual reporting in the Federal Register. In fact, reports from PMRs and PMCs are frequently delayed; and the PMRs/PMCs included in the FDA annual report in the Federal Register are combined data from the Center for Drug Evaluation and Research (CDER) and the Center for Biologics Evaluation and Research (CBER), while our study included only a subset of data from the Office of Immunology and Inflammation (OII). OII is one of the Clinical Offices of the Office of New Drugs (OND), and OND is one of the Super Offices within the CDER.

Even though the European Medicines Agency (EMA) publishes the protocols, abstracts, and final study reports of their post‐authorization safety studies (PASS) on the Heads of Medicines Agencies (HMA)‐EMA catalogue of real‐world data studies [28], we found very few publications on the regulatory impact of those studies, pertaining to safety‐related labeling update. Bouvy, et al., in their assessment of the safety‐related labeling changes, identified 33 PASS requested for centrally approved new active substances (NAS) (n = 47) in Europe in 2007 [29]. They reported an average of 8.1 safety‐related labeling changes per NAS after a follow‐up period of 6 years after marketing [29]. They found that requested PASS were the source of about 4% of all cases of new safety information identified [29]. However, the authors did not provide any details regarding the type of PASS responsible for those safety‐related labeling changes, such as pregnancy versus non‐pregnancy studies. Regarding our current study, we cannot draw any conclusions pertaining to the overall regulatory impact of FDA PMRs/PMCs since our study is not representative of PMRs/PMCs issued by the FDA in any given year, and our study included only a very small subset of data from the FDA's PMRs/PMCs database.

4.3. Strengths and Limitations of the Current Study

The strengths of the current study include the use of information from study reports, protocols, statistical analysis plans, product labeling, in addition to related communications including drug manufacturers cover letters, FDA review of final study reports and additional information requests, and drug manufacturers responses to those FDA information requests. The limitations include the relatively small sample (n = 14). The PMRs/PMCs included in the current assessment have been issued more than 10 years ago (from 1999 to 2011). That can be explained by the study inclusion criteria which required not only the submission of final studies' reports to the FDA but also the completion of FDA reviews of those final studies' reports. In addition, studies included in the current assessment were restricted to those studies involving drug products used to treat inflammatory or autoimmune conditions, approved by Divisions of Dermatology, Rheumatology, and Gastroenterology. Therefore, the findings from this study may not be generalizable to registries requested for different indications and/or registries requested in more recent years.

5. Conclusion

In the current study, all four pregnancy registry studies resulted in safety labeling update, as required by the FDA Pregnancy and Lactation Labeling Rule (PLLR) guidance. None of the 10 non‐pregnancy registry studies led to safety labeling updates, regardless of the targets' achievement status for study enrollment and follow‐up. Six of them did not detect a new safety signal, and the remaining four studies did not produce safety results considered sufficiently robust to warrant specific regulatory action including safety‐related labeling updates.

Multi‐center multinational registry studies and registries using participants from a health insurance or health maintenance organization may have greater chance for reaching their targets for both patient enrollment and patient follow‐up or drug exposure. However, besides recruiting and retaining a sufficient number of relevant subjects, other postmarketing safety registries challenges are to: include an adequate control group; account for concomitant treatments, frequent treatment switching, and confounding; and minimize exposure and outcome misclassification, and selection bias.

Some of the limitations may be inherent to these long‐term observational safety studies, such as: consent withdrawal and lost to follow‐up; insurance change and insurance denial or copay difficulty; frequent treatment switching and use of multiple therapies; and potential surveillance biases among treating physicians.

The lack of safety signal detection in these long‐term observational safety studies should not imply the absence of safety signals. Appropriately designed, prospective, randomized controlled safety studies are the most reliable way to obtain interpretable safety data.

5.1. Plain Language Summary

This paper assessed the regulatory impact of selected long‐term postmarketing safety registries (observational studies conducted by the drug manufacturers after the FDA has approved or licensed a product for marketing) on drug product labeling updates. The analysis focused on studies conducted for drug products used to treat inflammatory or autoimmune conditions. A total of 14 safety registries were included in the current analysis, 10 non‐pregnancy registries and four pregnancy registries. Only four safety non‐pregnancy registry studies were successful in reaching their targets for both patient enrollment and patient follow‐up or drug exposure, and two of the four pregnancy registries reached their targets for patient enrollment. None of the non‐pregnancy registries led to safety labeling updates: six of them did not detect a new safety signal and four provided inconclusive results. The lack of safety signal detection in these long‐term observational safety studies should not imply the absence of safety signals. Appropriately designed, prospective, randomized controlled safety studies are the most reliable way to obtain interpretable safety data.

Conflicts of Interest

The authors declare no conflicts of interest.

Disclaimer

This article reflects the views of the authors and does not necessarily represent FDA's views or policies.

Appendix A. Approved Indications for the Eight Drug Products

A.1.

Products	Indications
Abatacept (Orencia)	Treatment of adult patients with moderately to severely active rheumatoid arthritis (RA). Treatment of patients 2 years of age and older with moderately to severely active polyarticular juvenile idiopathic arthritis (pJIA). Treatment of patients 2 years of age and older with active psoriatic arthritis (PsA). Prophylaxis of acute graft versus host disease (aGVHD), in combination with a calcineurin inhibitor and methotrexate, in adults and pediatric patients 2 years of age and older undergoing hematopoietic stem cell transplantation (HSCT) from a matched or 1 allele‐mismatched unrelated‐donor.
Adalimumab (Humira)	Rheumatoid Arthritis (RA): reducing signs and symptoms, inducing major clinical response, inhibiting the progression of structural damage, and improving physical function in adult patients with moderately to severely active RA. Juvenile Idiopathic Arthritis (JIA): reducing signs and symptoms of moderately to severely active polyarticular JIA in patients 2 years of age and older. Psoriatic Arthritis (PsA): reducing signs and symptoms, inhibiting the progression of structural damage, and improving physical function in adult patients with active PsA. Ankylosing Spondylitis (AS): reducing signs and symptoms in adult patients with active AS. Crohn's Disease (CD): treatment of moderately to severely active Crohn's disease in adults and pediatric patients 6 years of age and older. Ulcerative Colitis (UC): treatment of moderately to severely active ulcerative colitis in adults and pediatric patients 5 years of age and older. Limitations of Use: Effectiveness has not been established in patients who have lost response to or were intolerant to TNF blockers. Plaque Psoriasis (Ps): treatment of adult patients with moderate to severe chronic plaque psoriasis who are candidates for systemic therapy or phototherapy, and when other systemic therapies are medically less appropriate. Hidradenitis Suppurativa (HS): treatment of moderate to severe hidradenitis suppurativa in patients 12 years of age and older. Uveitis (UV): treatment of non‐infectious intermediate, posterior, and panuveitis in adults and pediatric patients 2 years of age and older.
Certolizumab Pegol (Cimzia)	Reducing signs and symptoms of Crohn's disease and maintaining clinical response in adult patients with moderately to severely active disease who have had an inadequate response to conventional therapy. Treatment of adults with moderately to severely active rheumatoid arthritis. Treatment of adult patients with active psoriatic arthritis. Treatment of adults with active ankylosing spondylitis. Treatment of adults with active non‐radiographic axial spondyloarthritis with objective signs of inflammation. Treatment of adults with moderate‐to‐severe plaque psoriasis who are candidates for systemic therapy or phototherapy.
Etanercept (Enbrel)	Adult patients with: Rheumatoid Arthritis (RA). Psoriatic Arthritis (PsA). Ankylosing Spondylitis (AS). Plaque Psoriasis (PsO). Pediatric patients with: Polyarticular Juvenile Idiopathic Arthritis (pJIA), 2 years of age or older. Juvenile Psoriatic Arthritis, 2 years of age or older (JPsA). Plaque Psoriasis, 4 years of age or older.
Infliximab (Remicade)	Infliximab is a tumor necrosis factor (TNF) blocker indicated for: Crohn's Disease: Reducing signs and symptoms and inducing and maintaining clinical remission in adult patients with moderately to severely active disease who have had an inadequate response to conventional therapy. Reducing the number of draining enterocutaneous and rectovaginal fistulas and maintaining fistula closure in adult patients with fistulizing disease. Pediatric Crohn's Disease: reducing signs and symptoms and inducing and maintaining clinical remission in pediatric patients 6 years of age and older with moderately to severely active disease who have had an inadequate response to conventional therapy. Ulcerative Colitis: reducing signs and symptoms, inducing and maintaining clinical remission and mucosal healing, and eliminating corticosteroid use in adult patients with moderately to severely active disease who have had an inadequate response to conventional therapy. Pediatric Ulcerative Colitis: reducing signs and symptoms and inducing and maintaining clinical remission in pediatric patients 6 years of age and older with moderately to severely active disease who have had an inadequate response to conventional therapy. Rheumatoid Arthritis in combination with methotrexate: reducing signs and symptoms, inhibiting the progression of structural damage, and improving physical function in adult patients with moderately to severely active disease. Ankylosing Spondylitis: reducing signs and symptoms in adult patients with active disease. Psoriatic Arthritis: reducing signs and symptoms of active arthritis, inhibiting the progression of structural damage, and improving physical function in adult patients. Plaque Psoriasis: treatment of adult patients with chronic severe (i.e., extensive and/or disabling) plaque psoriasis who are candidates for systemic therapy and when other systemic therapies are medically less appropriate.
Pantoprazole Sodium (Protonix)	Short‐Term Treatment of Erosive Esophagitis Associated with Gastroesophageal Reflux Disease (GERD) Maintenance of Healing of Erosive Esophagitis Pathological Hypersecretory Conditions Including Zollinger‐Ellison (ZE) Syndrome
Tacrolimus (Protopic)	Second‐line therapy for the short‐term and non‐continuous chronic treatment of moderate to severe atopic dermatitis (AD) in non‐immunocompromised adults and children who have failed to respond adequately to other topical prescription treatments for atopic dermatitis, or when those treatments are not advisable.
Ustekinumab (Stelara)	Adult patients with: Moderate to severe plaque psoriasis (PsO) who are candidates for phototherapy or systemic therapy. Active psoriatic arthritis (PsA). Moderately to severely active Crohn's disease (CD). Moderately to severely active ulcerative colitis. Pediatric patients 6 years and older with: Moderate to severe plaque psoriasis, who are candidates for phototherapy or systemic therapy. Active psoriatic arthritis (PsA).

Appendix B. Language Included in the Labeling Section 8.1, Pregnancy for the Four Pregnancy Registry Studies

B.1.

• Adalimumab (Humira) OTIS‐M03604, PMC #2700‐6: Humira (Adalimumab) Pregnancy Exposure Registry OTIS Rheumatoid Arthritis in Pregnancy Project.

Language included in the labeling Section 8.1, Pregnancy:

A prospective cohort pregnancy exposure registry conducted by OTIS/MotherToBaby in the U.S. and Canada between 2004 and 2016 compared the risk of major birth defects in live‐born infants of 221 women (69 RA, 152 CD) treated with adalimumab during the first trimester and 106 women (74 RA, 32 CD) not treated with adalimumab.

The proportion of major birth defects among live‐born infants in the adalimumab‐treated and untreated cohorts was 10% (8.7% RA, 10.5% CD) and 7.5% (6.8% RA, 9.4% CD), respectively. The lack of pattern of major birth defects is reassuring and differences between exposure groups may have impacted the occurrence of birth defects. This study cannot reliably establish whether there is an association between adalimumab and major birth defects because of methodological limitations of the registry, including small sample size, the voluntary nature of the study, and the non‐randomized design.

• Etanercept (Enbrel) OTIS‐20040246, PMC #2541‐4: Rheumatic Diseases and Psoriasis Pregnancy Registry OTIS Autoimmune Diseases in Pregnancy Project.

Language included in the labeling Section 8.1, Pregnancy:

A prospective cohort pregnancy registry conducted by OTIS in the US and Canada between 2000 and 2012 compared the risk of major birth defects in liveborn infants of women with rheumatic diseases or psoriasis exposed to etanercept in the first trimester. The proportion of major birth defects among liveborn infants in the etanercept‐exposed (N=319) and diseased etanercept unexposed cohorts (N=144) was 9.4% and 3.5%, respectively. The findings showed no statistically significant increased risk of minor birth defects and no pattern of major or minor birth defects.

• Infliximab (Remicade) C0168T71, PMC #2598‐1: Exposure to REMICADE® (Infliximab) during Pregnancy in Patients with Inflammatory Bowel Disease, Rheumatoid Arthritis, Psoriatic Arthritis, Ankylosing Spondylitis and Psoriasis: A Review and Analysis of Birth Outcomes from the Swedish, Danish, and Finnish Medical Birth Registers.

Language included in the labeling Section 8.1, Pregnancy:

The second study among IBD and non‐IBD patients in Sweden, Finland, and Denmark compared 97, 7, and 166 women exposed to Infliximab to 2,693, 2,499 and 1,268 women on non‐biologic systemic therapy, respectively. In this study, comparing pooled data across the three countries, exposure to Infliximab was not associated with increased rates of congenital anomalies or infant death. Infliximab in combination with immunosuppressants (mainly systemic corticosteroids and azathioprine) was associated with increased rates of preterm birth, small for gestational age, low birth weight, and infant hospitalization for infection compared with non‐biologic systemic treatment. Although the study did not show any associations with Infliximab monotherapy, the analyses could have been underpowered to detect an association.

There were additional methodological limitations with these studies that may account for the study findings in both studies: the concomitant use of other medications or treatments was not controlled, and disease severity was not assessed; in the U.S. study, patient reported outcomes were collected without clinical validation. These methodological limitations hinder interpretation of the study results.

• Infliximab (Remicade) PRIORITY, protocol REMICADEIBS4013: Pregnancy and Infant Outcomes Research Initiative (PRIORITY) Utilizing Data from the United States Based PIANO Registry on Inflammatory Bowel Disease.

Language included in the labeling Section 8.1, Pregnancy:

The first study was conducted in an IBD pregnancy registry in the United States and assessed pregnancy outcomes in 294 women with inflammatory bowel disease exposed to Infliximab during pregnancy compared with 515 women on a non‐biologic treatment. Infliximab exposure was not associated with increased rates of major congenital malformations, miscarriage/stillbirth, infants of low birth weight, small for gestational age, or infection in the first year of life.

There were additional methodological limitations with these studies that may account for the study findings in both studies: the concomitant use of other medications or treatments was not controlled, and disease severity was not assessed; in the U.S. study, patient reported outcomes were collected without clinical validation. These methodological limitations hinder interpretation of the study results.