Exposure to Biological Therapies During Conception and Pregnancy

A Systematic Review

E. Pottinger; R.T. Woolf; L.S. Exton; A.D. Burden; C. Nelson-Piercy; C.H. Smith


The British Journal of Dermatology. 2018;178(1):95-102. 

In This Article


In this systematic review, we report a trend towards drug-specific harm with TNFi exposure in chronic inflammatory disease, which manifested as increased risk of congenital malformations and preterm birth. Notably this risk was low and did not reach statistical significance in all studies or across all outcomes. We highlight a critical gap in the evidence because no studies were identified that specifically address the risk of harm in women with psoriasis or longer-term outcomes in neonates/infants.

In the identified studies, major congenital malformations in TNFi-exposed women were reported at 3·6–5·0% with an incidence of 1·5–4·7% in comparison groups. This gave a potential increase in OR from 1·32 to 1·64. Reassuringly, if the observed trend across the studies towards potential harm to the fetus from biologics exposure is confirmed, the magnitude of the risk is low and the majority of women exposed to TNFi had a successful pregnancy. In addition, there was no distinct pattern of malformations observed (including VACTERL), which may be expected for a teratogenic medication.[25] VACTERL had previously been reported in association with TNFi therapy in a large observational study of self-reported pregnancy outcomes in a pharmacovigilance database;[26] however, the significance of this is uncertain, as the study did not adjust for potential confounders and there was potential for reporter/recall bias.

The mechanism by which TNFi may affect organogenesis is not clear, as direct embryonic exposure to biologics is negligible.[1,2] However, TNF-α is present in tissues of the female reproductive tract, the placenta and embryo; and TNF-α has been implicated in both embryonic defences during development and in the induction of embryo loss in the context of developmental damage.[27–29] Interruption of these processes may theoretically affect embryogenesis but this needs further investigation.

However, there are important limitations to these findings. Firstly, the studies were underpowered, which was reflected in wide confidence intervals that often included the null. The reported incidence of major congenital malformations in developed countries is 2–3%, and power calculations indicate a sample size estimate of 1000 women (intervention and control groups) to demonstrate a two-fold increased risk of harm.[24,25]

Secondly, the study by Weber-Schoendorfer et al., which reported the most significant effect size, was conducted in women who had consulted telephone teratology information services.[23] Although the comparison group included women who had not been exposed to known teratogens, the two groups were not disease matched. The prevalence of inflammatory diseases in the control group was not known and would likely be significantly lower than in the intervention group. This may introduce both a reporting and selection bias, as the effect attributed to TNFi exposure may be confounded by the coexistent disease status of the women in the intervention group, which were not adjusted for. Conversely, this may also account for the low incidence of congenital abnormalities in the control group, which might have a significantly lower incidence of coexisting medical conditions.

Thirdly, the studies did not always adjust for important confounders, including disease activity or concomitant disease-modifying medications (which may include known teratogens such as methotrexate and mycophenolate). In addition, these studies covered the period when biologics were newly introduced (1995–2004). It might be that, to justify the uncertain risk of exposure to a novel medication at this time, women who were exposed to biologics during conception and/or pregnancy had particularly difficult disease to manage (unstable and/or severe). Such unadjusted confounders were partially addressed by the use of disease-matched comparison groups;[20–22] however, these factors should be independently adjusted for.

Fourthly, heterogeneity between studies prevented meta-analysis of the data and may limit the generalizability of the findings. There was variation in the intervention (period of biologics exposure), the definition of adverse pregnancy outcomes (primarily around congenital malformations) and the follow-up period. In two studies this was limited to just 9–12 weeks postpartum, which may miss relevant adverse events that might present later, such as neonatal infection.[21,23] Many important outcomes that were predefined in our study protocol were not reported, such as spontaneous miscarriage, live birth rate, admission to the neonatal intensive care unit, neonatal infection and maternal death. Lastly, but importantly, the relevance of these findings to our patient cohort is especially compromised as most studies did not explicitly include women with psoriasis;[20,21,23] many included TNFi that are not licensed for the treatment of psoriasis (certolizumab and golimumab)[20,22,23] and did not identify any studies that included IL-12/23 and IL-17 antagonists. Such study heterogeneity and indirectness, with the imprecision (often small sample size) determined their 'very low' quality rating.[8]

An additional study, published subsequent to our review period, has reported outcomes for 74 pregnancies in women with rheumatoid arthritis who were exposed to adalimumab during pregnancy.[30] This study analysed data from the Adalimumab Pregnancy Registry and in all cases included drug exposure during the first trimester. In comparison with a small disease-matched comparison group (n = 80), and following adjustment for disease severity, the reported relative risk of major birth defects was 0·75 (95% CI 0·13–3·61), with no pattern of malformations evident, and a nonsignificant risk of spontaneous miscarriage (HRadj 2·06, 95% CI 0·53–7·98). In comparison to the trend of outcomes reported in the four reviewed studies, these findings are reassuring; however, the study by Burmester et al.[30] has important and similar limitations, principally the small sample size, limited multivariate adjustment and the indirect population studied.

This systematic review identifies possible harm associated with TNFi for inflammatory disease during conception but the risk estimate is very imprecise and quality of evidence low. Counselling women with psoriasis regarding the use of biologics during conception and pregnancy is challenging. Guidelines from the British Society for Rheumatology and European League Against Rheumatism for women with rheumatic diseases and an international position statement regarding women with IBD have indicated that TNFi should still be considered during pregnancy where there is clear clinical need.[6–8] These recommendations are largely based on consensus opinion and did not rigorously exclude low-quality data from unadjusted studies (including unmatched cohort studies and case series). These recommendations also reflect the known risk that poorly controlled IBD or inflammatory arthritis may have on pregnancy outcomes.[9–14] Currently, the impact of poorly controlled psoriasis on pregnancy is not known. Psoriasis is also a visible, chronic condition that can lead to feelings of low self-esteem, stigmatization and decreased confidence[31] – this alone may be a barrier to intimate relationships and justify the treatment of psoriasis during conception.

We conclude that the potential effect of biologics on pregnancy outcomes specifically in women with psoriasis has not been adequately studied to quantify accurately. Data on use in other indications is limited. Women of child-bearing potential should be advised routinely to use regular contraception; however, when planning conception the risks and benefits of continuing vs. stopping therapy should be discussed on a case-by-case basis. To address the clinical uncertainty, a large disease-matched cohort study is required, taking into account potential confounders such as disease activity, concomitant therapies and maternal demographics. Psoriasis-specific pharmacovigilance registries, for example the British Association of Dermatologist Biologics Intervention Register (BADBIR)[32] in collaboration with others via the European Psoriasis Registry Network (PSONET), provide opportunity to address this need. Clinicians and patients are strongly encouraged to participate in these registries.