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


Identified Studies

The systematic search identified 3383 eligible reference titles. Following the screening protocol, 27 full-text articles were assessed for inclusion; however, many articles were excluded for reasons that included no comparator group (n = 6) or failure to undertake multivariate adjustment (n = 5) (Figure 1). Four studies met our inclusion criteria: three retrospective cohort studies with disease-matched comparison groups,[20–22] and one prospective cohort study that compared biologics exposure with a population of pregnant women who had spontaneously contacted a teratology phone service during pregnancy or conception (Table 1).[23]

Figure 1.

Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flow diagram showing study selection process.

The identified studies were conducted in populations of pregnant women with largely noncutaneous chronic inflammatory diseases – most had a diagnosis of IBD or inflammatory arthritis. Two studies did include a small percentage of women with psoriatic arthritis,[21,23] and the study by Bröms et al. included a small percentage of women with psoriasis (4·5% in the intervention group, 25·7% in the disease-matched control group).[22] There was some variation in the size of the cohorts studied – Casanova et al. and Cooper et al. had relatively small intervention and control groups (56 and 171, 66 and 318, respectively),[20,21] whereas Weber-Schoendorfer et al. and Bröms et al. had larger intervention and control groups, using a telephone teratology information network and Scandinavian population-based health registers (495 and 1532, 683 and 21 549, respectively).[22,23]

The included studies investigated only exposure to TNFi during pregnancy (adalimumab, certolizumab, etanercept, golimumab and infliximab). No appropriate studies were identified for IL-12/23 antagonist or IL-17A antagonist use during conception and/or pregnancy. Biologics exposure in all studies included the first 12 weeks of pregnancy, and some also included the 3 months prior to conception,[20,22] and/or exposure throughout pregnancy.[20,23] The follow-up period also varied across studies, with a range of 9 weeks to 1 year postpartum (Table 1).

All four studies reported congenital malformations, but with some variation in the definition used (Table 1); Bröms et al. defined these as 'major birth defects' whereas Casanova et al. combined congenital malformations with other adverse outcomes to create a 'global pregnancy outcome' measure (Table 1).[20,22] Reporting of other outcome measures such as preterm birth, fetal death and life-threatening neonatal complications was not consistent across all four studies (Table 1).

The Safety of Biologics Therapies in Pregnancy

Major congenital malformations. Major congenital malformations were reported in 3·6–5·0% of women exposed to TNFi during the first trimester compared with 1·5–4·7% in control groups (Figure 2).[21–23] All three studies that reported this outcome measure found increased odds of potential harm to the fetus associated with TNFi [odds ratios (OR) 1·32–1·64] (Figure 2). However, this was statistically significant in only one of the studies: Weber-Schoendorfer et al. used a strict definition of major birth defects [European Surveillance of Congenital Anomalies (EUROCAT) network][24] and reported a 5% incidence in the TNFi-exposed group compared with 1·5% in the comparison group [OR 1·69, 95% confidence interval (CI) 1·10–2·60] (Figure 2).[23] Weber-Schoendorfer et al. reported a higher incidence of cardiac defects, which were in most cases associated with other birth defects; however, they found no distinct pattern of birth defects among TNFi-exposed infants, nor increase in VACTERL (vertebral defects, anal atresia, cardiac defects, tracheo-oesophageal fistula, renal anomalies and limb abnormalities).[23]

Figure 2.

Forest plot summarizing adverse pregnancy outcomes in comparison between TNFi therapy and control groups. CI, confidence interval; SE, standard error; TNFi, tumour necrosis factor (TNF)-α inhibitor.

Casanova et al. reported a potential increased risk of 'unfavourable global pregnancy outcome' in women with IBD who were exposed to TNFi therapy during the first trimester of pregnancy (OR 1·62, 95% CI 0·92–2·85) (Figure 2).[20] This broad outcome measure included congenital malformations, preterm delivery, low birth weight, spontaneous/elective abortion and adverse maternal measures (Table 1). This may account for the high incidence of 'unfavourable global pregnancy outcomes' (12·3%) and wide confidence intervals.

Other pregnancy outcomes. Weber-Schoendorfer et al. reported a significant association with a number of other adverse pregnancy outcomes: preterm birth (OR 1·69, 95% CI 1·10–2·60), low birth weight (median 3125 g in the TNFi-exposed group, median 3350 g in the control group, P = 0·02 following adjustment for infant age and sex) and elective termination of pregnancy [adjusted hazard ratio (HRadj) 1·69, 95% CI 1·0–2·9], but not with spontaneous miscarriage (HRadj 1·06, 95% CI 0·7–1·7).[23] The smaller study by Cooper et al. reported a nonsignificant trend towards increased risk of life-threatening neonatal complications (preterm infants OR 1·74, 95% CI 0·54–5·61; term infants OR 2·60, 95% CI 0·19–36·27) but no fetal deaths in the TNFi group compared with a disease-matched comparison group.[21]

Quality of Evidence

These studies were largely carried out in indirect populations of women who did not have a diagnosis of psoriasis. This indirectness and/or determined serious risk of bias and imprecision (using an adapted GRADE tool kit)[19] led to the overall quality of evidence being classified as 'very low' with respect to the review question. In addition, there was significant heterogeneity in study design across the identified studies – from women identified in the hospital setting to women self-reporting to a telephone-based teratology service.[23] Heterogeneity in study design, and in the definition of the intervention (biologics exposure) and outcome measures, prevented their meta-analysis (Table 1).

When considering the confounders included within the multivariate adjustment carried out by each study, only one paper adjusted for disease activity (Table 1).[20] Furthermore, only two studies adjusted for exposure to concomitant disease-modifying medications related to the inflammatory condition (Table 1).[20,21] Overall, there was a relatively low degree of overlap in the confounders considered by each study.