Biologic therapy and pregnancy outcomes in women with rheumatic diseases

Introduction

Most of the autoimmune rheumatic diseases are more common in women than in men. Because many of these conditions (e.g., rheumatoid arthritis [RA]) affect women of childbearing potential, reproductive issues related to disease management often emerge. Although some autoimmune rheumatic diseases tend to improve during pregnancy, treatment is sometimes required throughout the pregnancy and/or during the postpartum period. Unfortunately, some important therapeutic agents (e.g., methotrexate [MTX] in RA and cyclophosphamide in vasculitis) are teratogenic; therefore, treatment options during pregnancy are limited, and newer treatment options would be welcome. However, none of the newer biologic therapies (anti–tumor necrosis factor [anti-TNF] agents, anakinra, rituximab, or abatacept) are classified by the US Food and Drug Administration (FDA) as safe to use during pregnancy (1). This lack of safety classification is mainly attributable to the lack of adequate and well-controlled studies. Still, in the last decade, numerous case series and case reports of pregnant patients exposed to biologic therapy have accumulated in the literature. Because biologic agents may constitute an important therapeutic alternative in pregnant women experiencing persistent or increased disease activity, we present a comprehensive review of the relevant data.

Search strategy.

We performed a systematic literature review to identify all studies with original human data on fetal and/or child outcomes following exposure to biologic agents during pregnancy and/or lactation. We searched the following electronic databases for primary studies: PubMed (1950 to October 2008), EMBase (1996 to October 2008), and Web of Science (1991 to October 2008). Our search strategy was restricted to articles published in English, French, or Spanish and included the following medical subject headings (MeSH) index terms: “infliximab,” “adalimumab,” “abatacept,” “rituximab,” “pregnancy,” “lactation,” and “breastfeeding,” combined with the non-MeSH terms “etanercept,” “anakinra,” and “teratogenicity.” We hand-searched the reference lists of the primary studies and review articles for relevant articles not already captured by the electronic searches mentioned above. Any pertinent secondary references, including meeting abstracts and product monographs, were also reviewed.

One reviewer (EV) screened the citations (titles and abstracts) identified from all of the sources. Subsequently, full-text articles of the studies selected in the initial screen were reviewed to identify the final set of relevant studies (Figure 1).

Figure 1. Flow diagram of article selection for literature review.

Anti-TNF agents

FDA category B: no definite harm but insufficient evidence

The anti-TNF agents effective in treating inflammatory arthritis include etanercept (a soluble TNF receptor fusion protein linked to the Fc portion of human IgG1), infliximab (a chimeric monoclonal IgG1 anti-TNF antibody), and adalimumab (a human monoclonal IgG1 anti-TNF antibody) (2).

Although little is known about the ability of engineered monoclonal antibodies to cross the placenta, there is substantial placental transfer of maternal IgG to the fetal circulation during normal pregnancy, particularly during the second and third trimesters. Notably, a recent case report described therapeutic levels of infliximab at 6 weeks of age in a child who was exposed to infliximab during pregnancy (3).

In normal lactation, the amount of IgG1 secreted in breast milk is scant (3). Infliximab has been reported as being undetectable in the breast milk of most (3) but not all treated patients (4). Etanercept is secreted in human breast milk, and there are no data on whether adalimumab is secreted (5). Therefore, despite the fact that anti-TNF antibodies are likely digested in the gastrointestinal tract of the child, at the present time, lactation during anti-TNF treatment is discouraged (6).

In a series of 96 pregnancies in women with Crohn's disease or rheumatic diseases exposed to infliximab at conception and/or during the first trimester (Table 1), there were 64 live births, 14 miscarriages, and 18 therapeutic abortions (7). These pregnancy outcomes did not differ significantly from outcome expectations based on general population data. One child, born at 24 weeks, experienced intracerebral and intrapulmonary bleeding and died. Another child experienced respiratory distress and seizures. Two infants presented with congenital malformations: 1 with tetralogy of Fallot and 1 with intestinal malrotation. Many of the women in this series had active disease and during pregnancy were concomitantly exposed to other drugs such as azathioprine (33% of the 96 pregnancies), metronidazole (14%), and MTX (8%). The investigators in that study concluded that their data did not suggest an increase in fetal anomalies due to infliximab exposure.

Table 1. Summary of pregnancy outcomes in studies of women with rheumatic diseases and Crohn's disease exposed to anti-TNF therapy during pregnancy*

Author, (ref.)	Anti-TNF agents	Other drugs†	Pregnancies, no.	Live births, no.	Miscarriages, no.	Terminations, no.	Birth defects, no.	Other reported adverse events
* TNF = tumor necrosis factor; inflix = infliximab; etan = etanercept; ada = adalimumab; NS = not specified. † Methotrexate, leflunomide, or azathioprine. ‡ Trisomy 18. § Intestinal malrotation (1 patient), tetralogy of Fallot (1 patient). ¶ VATERL association (vertebral anomalies, anal atresia, esophageal atresia with or without tracheoesophageal fistula, renal anomalies, and preaxial defects of the upper limbs, including polydactyly of the thumb). # Tracheomalacia+bronchomalacia+patent ductus arteriosus+skeletal anomaly.
Cush, (10)	Inflix+etan+ada	NS	417	387	25	5	0	0
Hyrich et al, (13)	Inflix+etan+ada	Some patients	22	13	6	3	0	Low birth weight (1)
Joven et al, (15)	Inflix+etan+ada	NS	14	10	1	3	0	0
Chambers et al, (12)	Inflix+etan	No	33	28	4	1	1‡	0
Rosner et al, (18)	Inflix+etan	Some patients	4	4	0	0	0	0
Katz et al, (7)	Inflix	Some patients	96	64	14	18	2§	Cerebral bleeding + neonatal death (1), respiratory distress and seizures (1)
Mahadevan et al, (9)	Inflix	Some patients	10	10	0	0	0	Low birth weight (1), respiratory distress (1)
Lichtenstein et al, (8)	Inflix	Some patients	36	–	–	–	0	0
Tursi, (19)	Inflix	Yes	1	1	0	0	0	0
Angelucci et al, (20)	Inflix	Yes	1	1	0	0	0	0
Burt et al, (21)	Inflix	No	1	1	0	0	0	0
Kinder et al, (22)	Inflix	Yes	1	0	1	0	0	0
Vasiliauskas et al, (3)	Inflix	No	1	1	0	0	0	0
Stengel and Arnold, (23)	Inflix	No	1	1	0	0	0	0
Chakravarty et al, (24)	Etan	Some patients	7	6	1	0	0	0
Kosvik et al, (25)	Etan	No	5	3	2	0	0	0
Roux et al, (26)	Etan	Yes	2	1	0	1	0	0
Rump et al, (27)	Etan	No	1	1	0	0	0	0
Feyertag et al, (28)	Etan	No	1	1	0	0	0	0
Carter et al, (29)	Etan	No	1	1	0	0	1¶	0
Sinha and Patient, (30)	Etan	No	1	1	0	0	0	0
Micheloud et al, (31)	Etan	No	1	1	0	0	0	0
Otermin et al, (32)	Etan	No	1	1	0	0	0	0
Vesga et al, (33)	Ada	No	1	1	0	0	0	0
Kraemer et al, (34)	Ada	Yes	1	1	0	0	0	0
Mishkin et al, (35)	Ada	No	1	1	0	0	0	0
Coburn et al, (36)	Ada	Yes	1	1	0	0	0	0
Carter et al, (37)	Ada	NS	1	1	0	0	1#	0

A Crohn's disease registry reported outcomes in pregnant women exposed to infliximab at varying times and dosages (8). No birth defects were reported, and the rates of miscarriage and neonatal complications were not different between pregnant patients who were treated with infliximab (n = 36) and those who did not receive such treatment (n = 30). Mahadevan et al reported intentional treatment with infliximab during pregnancy in 10 women with severe Crohn's disease (9). Eight of these 10 women received infliximab throughout the pregnancy, 1 was exposed to infliximab during the first trimester only, and 1 began therapy in the third trimester. All 10 pregnancies ended in live births, without congenital malformations. Three of the births were premature, 1 child had low birth weight, and 1 child experienced respiratory distress.

With regard to studies of pregnant women with RA, one study of 417 pregnancies exposed to anti-TNF agents (etanercept was the treatment in 81% of cases), 387 normal deliveries were observed (10). There were 25 miscarriages, 5 pregnancy terminations, and 9 preterm births; all rates were comparable with those in the general population (11). No malformations or neonatal deaths were reported. The Organization of Teratology Information Services registry also collected information on 33 pregnant patients with RA who were exposed to etanercept (n = 29) or infliximab (n = 4), without concomitant MTX, during the first trimester (12). These 33 pregnancies were compared with 77 pregnancies in patients with RA who were treated with various agents, including disease-modifying antirheumatic drugs, but not anti-TNF agents, and with pregnancies in 50 healthy control subjects. Preterm deliveries were more frequent in women exposed to anti-TNF agents (28% of those treated with etanercept) and in control patients with RA (23.5%) compared with healthy subjects (4.3%). This difference was most likely attributable to the underlying disease and/or the concomitant use of systemic steroids. Spontaneous abortions occurred with both etanercept (3 events) and infliximab (1 event); in 1 spontaneous abortion occurring after etanercept exposure, a trisomy 18 chromosomal abnormality was observed. No difference in the rate of congenital malformations was seen between the 3 groups.

The British Society of Rheumatology Biologics Register reported 22 pregnancies in women with rheumatic diseases exposed to anti-TNF therapy at the time of conception (16 were receiving etanercept, 3 were receiving infliximab, and 3 were receiving adalimumab) (13). Nine of these women were being treated with concomitant MTX, and 2 were receiving leflunomide. All patients stopped receiving therapy during the first trimester, except 2 patients who continued receiving etanercept throughout the pregnancy. In this series of 22 pregnancies, there were 6 first-trimester miscarriages (3 in patients receiving concomitant MTX and 1 in a patient who was receiving leflunomide at the time of conception), 3 elective abortions, and 13 live births (1 preterm birth and 1 child with low birth weight). No birth defects occurred.

The Spanish Registry of Adverse Events of Biological Therapies in Rheumatic Diseases reported 14 pregnancies in women with inflammatory arthritis who were exposed to anti-TNF agents (8 were receiving etanercept, 4 were receiving infliximab, and 2 were receiving adalimumab) (14, 15). These pregnancies resulted in 7 uncomplicated live births in 4 women who were treated with etanercept and 3 women who were treated with infliximab. No outcome information was available for 3 pregnancies, and the remaining 4 pregnancies resulted in termination (2 in women who were treated with etanercept and 1 in a woman treated with adalimumab) and 1 spontaneous abortion (in a woman who was treated with infliximab). No congenital malformations were reported.

Fewer data on adalimumab exposure during pregnancy are available, since this agent was introduced relatively recently. In a study of 61 pregnant women with inflammatory arthropathies, 4 were exposed to adalimumab (2). No maternal or fetal adverse outcomes were reported.

Because elevated TNFα levels may be associated with recurrent spontaneous abortions (16), some investigators have used anti-TNF agents for the treatment of recurrent abortions. Winger et al retrospectively analyzed 75 pregnancies in women with a history of recurrent spontaneous abortions who were treated with combinations of anticoagulation therapy, intravenous immunoglobulin (IVIG), and anti-TNF agents (etanercept or adalimumab) (17). The anti-TNF agent was started prior to conception and continued until there was evidence of fetal cardiac activity. In women treated with a combination of anti-TNF therapy, anticoagulation therapy, and IVIG, the live birth rates (71%) were greater than those in women treated with anticoagulation therapy alone (19%) or in those receiving a combination of anticoagulation therapy and IVIG (54%). Fetal outcomes, including gestational age and birth weight, were similar across the groups, and no congenital anomalies were reported after anti-TNF agent exposure.

Additional experience with infliximab, etanercept, and adalimumab from case reports is summarized in Table 1 (18–37).

Rituximab, FDA category C: insufficient evidence

Rituximab is a chimeric anti-CD20 monoclonal antibody approved for the treatment of non-Hodgkin's lymphoma and RA. Very few data on rituximab treatment during human pregnancy have been reported. Only 9 women exposed to rituximab during pregnancy have been described in different case reports (Table 2) (38–46). These 9 women received rituximab at a dose of 375 mg/m², and repetition of rituximab administration varied from 4–8 times at 1– 4-week intervals. All of the pregnancies ended in a live birth; 5 of the 9 children were born at term (39, 40, 42, 43, 45), and 4 were born preterm. Of these 4 preterm neonates, 1 was delivered by caesarean section due to recurrent maternal seizures in the context of thrombotic thrombocytopenic purpura (41). For the 3 other cases, no etiology for the prematurity was described.

Table 2. Characteristics of the pregnancies exposed to rituximab*

Author, (ref.)	Diagnosis†	Exposure, trimester	Gestational age at birth, weeks	B cell depletion in the offspring‡	Response to immunization in the offspring
* TTP = thrombotic thrombocytopenic purpura; AIHA = autoimmune hemolytic anemia; ITP = idiopathic thrombocytopenic purpura. † In women with lymphoma, 5 received concomitant CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone) therapy. ‡ CD19+ B cell levels were measured at birth and/or at 1 week of life.
Herold et al, (38)	Lymphoma	2 and 3	35	–	–
Kimby et al, (39)	Lymphoma	1	40	Yes	Normal
Friedrichs et al, (40)	Lymphoma	2 and 3	41	Yes	Normal
Scully et al, (41)	TTP	2 and 3	30	–	–
Ojeda-Uribe et al, (42)	AIHA	1	38	No	–
Magloire et al, (43)	Lymphoma	2	39	–	–
Decker et al, (44)	Lymphoma	2	33	Yes	Normal
Klink et al, (45)	ITP	3	38	Yes	Normal
Rey et al, (46)	Lymphoma	2 and 3	33	–	–

Rituximab levels were measured at birth in 3 of the 9 neonates (40, 44, 46); all of the neonates had detectable levels of rituxmab, and 2 had levels that exceeded maternal levels. However, all of the neonates showed a decline in rituximab levels over the subsequent weeks, and this decline was expected given the known half-life of rituximab. Although there are few data on placental transfer of monoclonal antibodies, physiologic IgG subtype concentrations in the neonate usually exceed maternal levels at birth, due to active transport through specific Fc receptors located at the placental barrier (40). Because rituximab has a human IgG1 constant region, this might explain the increased rituximab levels in the offspring compared with the mothers (40).

In 5 of the 9 neonates, CD19+ B cell levels were measured at birth and/or at 1 week and were either undetectable or severely decreased in all but 1 child, who was exposed only during the first trimester (39, 40, 42, 44, 45). In those neonates with decreased or undetectable B cell levels, exposure occurred mainly during the second and/or third trimester, although 1 child was exposed only during the first trimester. B cell levels returned to normal by ∼3– 6 months in all children who had followup testing. Although a low granulocyte count was observed in 2 children during their first weeks of life, no infection developed in either child (39, 41). In fact, no serious infectious complications have been documented in neonates exposed to rituximab. Moreover, an adequate response to standard immunization was evaluated and found to be normal in 4 of these 9 children (39, 40, 44, 45).

One unpublished report on rituximab safety described 24 pregnancies in women with lymphoma who were exposed to rituximab 1–8 months prior to conception (47). Ten of these pregnancies had a known outcome, which was the delivery of a live infant without congenital anomalies or serious infections. However, 2 of the children presented with low granulocyte counts, and 1 also exhibited anemia and lymphopenia. Finally, Ostensen et al reported 3 women with systemic lupus erythematosus who were exposed to rituximab 12, 6, and 4 months before a pregnancy (47). One woman underwent a therapeutic abortion, while the others delivered healthy preterm neonates.

Abatacept, FDA category C: insufficient evidence

Abatacept belongs to a new class of agents developed for the treatment of RA. Abatacept is a recombinant fusion protein that selectively modulates the costimulatory signal required for full T cell activation (48). No data on exposure to abatacept during human pregnancy have been published. In addition, it is not known whether abatacept is excreted in human breast milk or is absorbed systemically after ingestion.

Anakinra, FDA category B: no definite harm but insufficient evidence in humans

Anakinra is a recombinant form of the human interleukin-1 receptor antagonist (IL-1Ra). IL-1Ra is a naturally occurring IL-1R blocker that inhibits the proinflammatory effects of IL-1. Anakinra is approved for the treatment of RA (49), but currently there are no published data regarding treatment with anakinra during human pregnancy. Moreover, it is unknown whether anakinra is excreted in human breast milk.

Discussion

The currently available data do not seem to support a large excess risk of adverse pregnancy and/or fetal outcomes in women exposed to anti-TNF therapy at some point during pregnancy. However, data are sparse, and conclusions are limited by the relatively small number of published case series and reports, differences in the type and dosage of agents, use of concomitant teratogens (such as MTX and leflunomide), as well as variability in the timing of exposure during pregnancy. Furthermore, because therapeutic levels of anti-TNF agents have been observed in a child exposed to one of these agents late in pregnancy (3), more information is needed regarding long-term adverse outcomes.

Experience with rituximab during pregnancy is very limited. Neonatal B cell depletion has been reported following exposure at any stage of pregnancy. More studies are needed to assess potential neonatal complications that could be caused by this transient B cell depletion, as well as the long-term impact of rituximab exposure in utero on immune function. Because of this lack of data, rituximab treatment should be stopped before conception. For increased safety, maternal serum levels should probably be measured and pregnancy attempted only after serum levels become negative (47).

At the present time, there are no data on exposure to abatacept or anakinra during human pregnancy. Therefore, no comment can be made on the safety of these 2 agents during pregnancy. For both of these agents, toxicology studies in animals have not demonstrated an increased risk of congenital abnormalities (50, 51). However, the potential for harm remains, because both drugs have been shown to cross the placenta (52), and immune dysfunction was observed in rodent offspring exposed to abatacept. IL-1Ra likely has pleiotropic effects in pregnancy, with the potential to inhibit embryonic implantation in mice (53, 54) as well as the potential to improve outcomes in animal neonates with experimentally induced sepsis and hypoxia/ischemia (55).

With the emergence of postmarketing surveillance, increased collection of prospective data is expected over the next 5 years. This will allow assessment of the number of adverse events that occur after in utero exposure to biologic agents compared with the number of such events in the general population or in other women with rheumatic diseases who were not treated with biologic agents. It may also provide more data on related issues, including the use of biologic agents in both men and women during the conception period. Furthermore, ongoing studies that link data from drug and birth registries will help to address these important issues (2).

AUTHOR CONTRIBUTIONS

All authors were involved in contributions to study conception and design, acquisition of data, or analysis and interpretation of data, and drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Vinet had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.