Preterm Birth and in Utero Exposure to Corticosteroids Are Associated With Increased Infection Risk in Children of Mothers With IBD

Abstract

Background

Corticosteroids, thiopurines, and biologics may come into play during pregnancy in women with inflammatory bowel disease and potentially impact the developing fetal immune system. We aimed to assess the risk of serious infections in children stratified by in utero exposure to biologics and immunomodulators or concomitant treatment with corticosteroids.

Methods

All singleton IBD pregnancies between 2008 and 2022 at a tertiary IBD center in Denmark were included. Maternal and offspring demographics, maternal disease activity, antenatal medical treatment, and infant infections resulting in hospital admission were recorded after review of medical records.

Results

In 602 live births (99.0%), we registered exposure to antenatal treatment as follows: biological monotherapy (n = 61, 10.2%), thiopurines (n = 110, 17.9%), biologics and concomitant thiopurines (n = 63, 10.3%), and controls (ie, no treatment with biological and/or thiopurines; n = 369, 60.6%). Preterm delivery (<37 gestational weeks) and systemic steroid administration during the third trimester were associated with an increased risk of serious infection in the offspring immediately after birth (relative risk = 17.5; 95% confidence interval, 7.8-39.8; P < .001, and relative risk = 4.8; 95% confidence interval, 1.5-12.7; P = 0.003, respectively).

Intra-uterine exposure to biologics or combination treatment were not associated with a statistically significant higher risk of serious infections compared with controls; however, combination treatment showed an inclination towards an increased risk across analyses.

Conclusion

Preterm birth and systemic corticosteroid administration late in pregnancy are significant risk factors for serious infections in the offspring of IBD mothers.

Introduction

Remission in IBD is of pivotal importance for a favorable birth outcome, and nowadays women may receive potent immunosuppressive treatment during pregnancy as standard of care.¹

Placental transfer of biologics increases as the pregnancy progresses, and it has been speculated that biologics may remodulate the development of the immature immune system of the fetus and render the offspring more susceptible to infections.²

Long-term effects of intrauterine exposure to anti tumor necrosis factor (anti-TNF) and/or thiopurines have been investigated with conflicting conclusions; several prospective or retrospective studies have found no significant increased risk of infant infections,^2–4 whereas others have reported a significantly increased risk of infections after exposure to anti-TNF and/or thiopurines in utero.^5–8

Preterm birth is an important factor in the risk assessment of infections in children born to women with IBD, regardless of antenatal treatment.⁹ Indeed, preterm birth is a risk factor for serious infections leading to hospitalization, prescribed antibiotic treatment administered at home, and simple self-limiting infections.^4,9

In the PIANO cohort, Odufalu et al found an association between steroid exposure in the second and/or third trimester and an increased risk of serious infection at 9 and 12 months after birth.¹⁰ These observations are compatible with the concept that intrauterine exposure to immunosuppressive treatment may exert a lasting effect on the infant’s immune system.^10,11

The present study aimed to investigate the risk of serious infections in children stratified by exposure to corticosteroids, biologics, and immunomodulators from 3 angles: (1) hospital admission due to serious infections at birth; (2) the rate of infections requiring hospital admission; and (3) the number of serious infections during the follow-up time.

Materials and Methods

Study Population

This cohort study was conducted at the Department of Hepatology and Gastroenterology outpatient clinic, Aarhus University Hospital (AUH), a tertiary IBD center. The study period was from January 1, 2008, to June 1, 2022.

From an administrative hospital database, we identified every woman with a referral diagnosis of either Crohn’s disease (CD, DK50) or ulcerative colitis (UC, DK51) including all subtypes of IBD. From review of medical records, we excluded women examined for but not diagnosed with IBD, and in patients diagnosed with both CD and UC at one point, the most recent code was used. Among these, we restricted to those giving birth after the IBD diagnosis, ascertained by a mandatory international classification of diseases tenth edition (ICD-10) diagnosis code (DO8) assigned when giving birth to a child. All types of singleton births, including vaginal delivery and cesarean sections (CS) with different indications were included. Multifetus pregnancies were not included because women who are pregnant with more than 1 fetus may have more complicated pregnancies.¹² Both mothers and children participating were identified by the civil registration number assigned to every individual resident of Denmark, enabling linkage to medical records and birth record data. Follow-up was conducted from the time of birth until June 1, 2022, and the participating children thus contributed to different lengths of follow-up time.

Controls were defined as women with IBD who did not receive treatment with thiopurines, biologics, or the 2 combined.

Collection of Data Through Review of Medical Records

All data were obtained through review of medical records, and all data from both mothers and children were recorded prospectively in medical records from routine clinical care.

Maternal data regarding the year of IBD diagnosis and previous IBD-related operations were registered. The Montreal classification system served as IBD disease classification.

For each pregnancy, we registered maternal age, maternal height, and prepregnancy weight, smoking status during pregnancy (yes/no), parity, and intraabdominal operations 6 months before or during pregnancy. Medication requiring a prescription was registered, as well as comorbidities that might require medical treatment involving steroids or immunosuppressants and/or affect the birth weight or birth outcome through separate pathology. Concurrent usage of biologics and thiopurines was termed combination treatment.

For the duration of each pregnancy, fetal growth restriction (FGR) was monitored by ultrasound in the second trimester at gestational week 20 and in the third trimester at varying time points between gestational weeks 28 and 38. Fetal growth restriction was defined as a fetal weight deviation greater than 15% below the expected weight at the gestational age concerned.¹³

When assessing the presence of disease activity in the mother during pregnancy, we, through review of medical records, employed the Physician’s Global Assessment (PGA) to define dichotomously whether or not a woman experienced disease activity. The PGA was based on the attending gastroenterologist’s assessment incorporating escalation in medical treatment, patient completion of validated disease activity indices (Harvey-Bradshaw or Simple Clinical Colitis Activity Index), and if available endoscopic evaluation and/or biomarker of inflammation (C-reactive protein [CRP] and fecal calprotectin [FC]). The PGA was performed by only 1 gastroenterologist to ensure consistency in all assessments and was applied in each trimester of pregnancy. In pregnancies with limited contact with the Department of Gastroenterology, observations registered by midwives in the context of antenatal care were readily retrieved from the hospital’s shared medical records, strengthening the validity of the PGA.

Regarding the children, we registered gestational age at delivery, birth weight, length, APGAR score at 5 minutes, congenital malformations, caesarian section and the indication for CS. A low APGAR score was defined as an APGAR score <7 at 5 minutes. Stillbirths were registered if the woman went into labor after gestation week 23 and gave birth to a deceased child. Births before gestational week 23 were excluded, as they were classified as miscarriages. Preterm birth was defined as spontaneous preterm labor before 37 full weeks of gestation.

Small for gestational age (SGA) was defined as a birth weight below the tenth percentile of newborns of the same gestational age. Low birth weight (LBW) was defined as a birth weight below 2500 grams, irrespective of gestational age at birth.¹³

In accordance with WHO criteria, major malformations were defined as malformations that create significant medical problems for the individual or require specific medical treatment or surgery. Minor malformations were defined as unusual external physical features that do not impair a major body function.¹⁴

During follow-up time, we registered all admissions due to serious infections. Serious infections were stated as infections that required hospital admission. For each admission, we recorded the date of admission, number of days of admission, and location of infection. We also noted if antibiotics were administered.

Statistics

Characteristics of participating women and their pregnancy outcomes were tabulated in contingency tables and stratified by antenatal treatment.

Data on birth outcomes were tabulated stratified by antenatal treatment received by the mother during pregnancy.

Stillbirths were excluded when calculating specific end points concerning birth outcomes, but the pregnancies they resulted from were included in the remaining analyses concerning maternal characteristics and pregnancy outcomes.

All continuous covariates are presented as medians with interquartile range (IQR), as only birth weight followed a Gaussian distribution.

Groups of unpaired data were compared using simple t tests if equal variance and otherwise using Welch’s correction to obtain P values. When comparing groups of unpaired data of non-Gaussian distribution, nonparametric Mann-Whitney U tests were applied. The χ² test, Fisher’s exact test, or Wilcoxon rank test were used to compare nominal variables, depending on sample sizes and data distribution.

When assessing severe infection in connection with birth, we performed a univariate logistic regression to investigate the association with different relevant covariates. We subsequently performed a multivariate logistic regression analysis to adjust for relevant covariates. The relevance of covariates was assessed by creating a directed acyclic graph (DAG) of all variables involved.

We performed Cox regression analyses to assess the rate at which the first severe infection occurred in the distinct treatment groups.

When assessing the number of serious infections over time, we created quasi-Poisson regression models with the log of the follow-up time in days as offset. We also accounted for the random effect of individual mothers giving birth to up to 4 of the children in the total cohort. All models were adjusted for treatment groups.

In all regression analyses, covariates affected by collinearity were excluded from the multivariate regression analysis. The statistics and analyses were made using R for statistical computing. All figures were created using biorender.com.

Ethical Considerations

The study was approved by the Danish Data Protection Agency (j.nr: 1-16-02-370-18) and the Ethics Committee of the Central Region of Denmark (j.nr: 1-10-72-289-18).

Results

Between 2008 and 2021, 378 women contributed with a total of 609 pregnancies. In total, 6 children were excluded due to stillbirth (1.0%), and 1 (0.2%) child was put up for adoption right after birth and lost to follow-up. The remaining 602 children contributed with a total of 3441 years of follow-up time. Among women on biologics (n = 126), most were treated with anti-TNF (n = 108, 85.7%). Most patients treated with thiopurine and/or biologics had CD (n = 78, 61.9%), and the controls primarily had UC (n = 261, 70.2%). Overall, controls experienced less disease activity and had less extensive and less complex disease than women treated with thiopurine and/or biologics. Maternal characteristics are shown in Table 1.

Maternal Pregnancy Outcome

Disease activity

Women treated with biological monotherapy (n = 33, 53.2%) or combination treatment (n = 28, 43.8%) more frequently experienced disease activity within 6 months of conception than did controls (n = 86, 23.1%) and women treated with thiopurines (n = 22, 19.8%).

Women treated with biological monotherapy (n = 44, 71.0%) more frequently experienced disease activity anytime during pregnancy than did controls (n = 176, 47.3%) and women in combination treatment (n = 32, 50.0%). Details regarding disease activity can be seen in Table 2.

Birth Outcomes

The prevalence of preterm birth was significantly higher in children born to women treated with biological monotherapy (n = 14, 23.0%) or combination treatment (n = 7, 11.1%) than in children born of controls (n = 20, 5.4%) or women on thiopurine monotherapy (n = 7, 6.4%). Table 3 presents birth outcomes.

In contrast to women treated with thiopurine or combination therapy (biologics and thiopurine), women on biological monotherapy gave birth to infants with a median 290 grams lower birthweight than controls (P = .001), however, still within normal range. There was no significant variation concerning the risk of SGA, LBW, APGAR score <7, or congenital malformations between controls and women treated with biological monotherapy, thiopurines, or combination therapy, respectively.

Serious Infections Requiring Hospital Admission

Neonatal infections

The relative risks of serious infection in connection with birth are presented in Table 4 as univariate and multivariate logistic regressions.

In univariate analysis, being treated with concomitant systemic steroids during pregnancy (relative risk [RR], 3.2; IQR, 1.2-7.5; P = .01) or just having systemic corticosteroids administered in the third trimester (RR, 4.8; IQR, 1.5-12.7; P = .003) were significantly associated with an increased risk of serious infection, as was preterm birth (RR, 17.5; IQR, 7.8-39.8; P < .001). When performing multivariate logistic regression analysis to adjust for relevant covariates, preterm birth maintained a statistically significant risk factor for serious infection after birth (adjusted relative risk [aRR], 18.0; IQR, 7.2-45.4; P < .001), while administration of systemic corticosteroids in the third trimester lost statistical significance.

In the univariate and the multivariate analyses, we found no significant association between risk of serious infant infections and in utero exposure to biological monotherapy, thiopurines, or combination therapy.

Time to first hospital admission due to infection

The hazard ratios (HRs) for the first serious infection after birth are presented in Table 5 as univariate and multivariate Cox regressions.

In the univariate analysis preterm birth, systemic corticosteroid usage anytime during pregnancy or administration of systemic corticosteroids in the third trimester significantly increased the hazard ratio of severe infections. When adjusting for relevant covariates in a multivariate Cox regression analysis, only preterm birth remained of statistical significance in increasing the hazard of severe infection (HR, 2.5; 95% confidence interval [CI], 1.7-3.8; P < .001). Of note, neither in the univariate nor the multivariate analysis did exposure to biological monotherapy (aRR, 0.8; 95% CI, 0.5-1.3; P = .3), thiopurine (aRR, 0.9; 96% CI, 0.6-1.4; P = .7), or combination therapy (aRR, 1.3; 95% CI, 0.9-2.0; P = .2) significantly influence the time to first serious infection after birth.

Incidence rate and incidence rate ratios of number of infections within 4 months, 1 year, 3 years, and 5 years after birth

When performing regression analyses (Table 6), preterm birth increased the incidence rate ratio (IRR) of severe infection in both univariate and multivariate analyses and remained statistically significant at all checkpoints. Controlling for preterm birth, disease activity, systemic corticosteroid administration in the third trimester, and disease phenotype, the use of biological monotherapy, thiopurine, or combination therapy was not associated with increased risk of infections in the first 4 months, 1 year, 3 years, or 5 years of life, apart from combination therapy within 5 years after birth (aIRR, 2.1; IQR, 1.0-4.1; P = .04)

The maternal IBD phenotype of CD significantly increased the IRR of infant infections at both 3 and 5 years (Table 6).

The crude incidence rate (IR) of serious infections stratified by medical treatment can be seen in Supplementary Table 1, and the localization of severe infections is displayed in Supplementary Table 2.

Discussion

In the present study, we sought to elucidate the impact of antenatal exposure to biologics, corticosteroids, and immunomodulators on the incidence and location of serious infections equivalent to hospitalizations in children of mothers with IBD. We observed a remarkable impact of preterm birth and an indication of an increased risk of serious infection in children exposed to corticosteroid treatment in utero reported in previous studies.

Concerning preterm birth, our study found a significantly increased risk of severe infections in children born prematurely, maintained in both univariate and multivariate regressions. Prematurity was found to be associated with an increased risk of both serious infections subsequently after birth, as well as an increased hazard and incidence rate of severe infection during the 5-year follow-up time. These important findings are substantiated by both the TEDDY and PIANO studies, which after 1 year of follow-up reported increased risks of serious infection in children born preterm (TEDDY, n = 841, HR 2.5 95% CI [1.5; 4.3]; and PIANO, n = 1010, odds ratio [OR], 1.73 95% CI [1.2; 2.5], respectively).^2,4 This association underlines the importance of providing healthcare to ensure that women with IBD may carry their children to term, in accordance with both European and American guidelines.^9,15

Regarding systemic corticosteroid treatment, usage at any time during pregnancy was found to be of importance to the risk of serious infection subsequently after birth, especially regarding the children who were subjected to systemic corticosteroids during the third trimester of pregnancy. However, the statistical significance was not sustained when adjusting for relevant covariates, which may be due to the small sample size rather than actual lack of impact. Of note, our finding of increased risk of serious infections within the first year of life in children exposed to corticosteroids in utero are in line with a large Taiwanese register-based study comparing 45 232 exposed infants compared with 1 915 313 nonexposed infants (aHR 1.24; 95% CI, 1.16-1.32; P < .001), as well as the PIANO registry including offspring of IBD mothers, where Odufalu et al found an association between steroid exposure in the second and/or third trimester and an increased risk of serious infection at 9 (P = .03) and 12 months (P = .001) after birth.^10,16

Concerning the risk of infections following intrauterine exposure to biologics and thiopurines, we overall found no increased risk in children exposed to biological monotherapy; however, a trend towards an increased risk in children subjected in utero to combination therapy with biologics and thiopurines was present across analyses. This particular trend has also been observed in other studies,^2–4,7 and although the findings of previous studies has been toned down due to lack of statistical significance,^2,3 we advocate for the necessity to regard the propensity of the true risk rather than draw conclusions on P values alone.

A meta-analysis including 8490 infants of mothers with IBD suggested a nonsignificant increase in the odds ratio of severe infection incidence with anti-TNF exposure during pregnancy (OR, 1.33, 95% CI [0.95-1.86]);⁸ however, the study did not substratify for combination treatment and was performed before the conduction of several important studies.^2–7 The sizeable Danish register-based study by Nørgård et al thus reported significantly increased risk of infections within the first year of life (HR, 1.44; 95% CI, 1.19-1.74) in children subjected to biological monotherapy with anti-TNF adjusted for concomitant thiopurine usage (n = 493) compared with controls (n = 728 055).⁶ This was in accordance with a large Swedish study by Bröms et al that compared infants subjected in utero to combination therapy (n = 674) with untreated controls (n = 1 617 886) and likewise reported a significantly increased incidence rate of serious infections in infants exposed to combination treatment (aIRR, 1.24; 95% CI [1.02-1.51]).⁵ Both register-based studies, however, included several inflammatory diseases eligible for anti-TNF therapy in their treatment groups without stratification for IBD indication and employed the background population as controls, thereby not considering the impact of IBD presence in itself; and the Swedish study furthermore included varied immunomodulators in the combination treatment.

In the prospective ERA study among 80 infants exposed to anti-TNF, the risk of infection within the first year of life was significantly greater among infants exposed to combination therapy compared with monotherapy (RR, 2.7; 95% CI, 1.09-6.78; P = 0.02).⁷

A Dutch retrospective study including 1000 children born to mothers with IBD likewise hinted at an increased risk of severe infections requiring hospital admission, though not statistically significant, in infants exposed to anti-TNF and thiopurine combination treatment (n = 60; IRR, 1.65 95% CI [0.71-3.82]; P = 0.24).³ The PIANO registry (n = 1010 at 1-year follow-up) had an alternative stance and reported a negligible risk of serious infections within the first year of life when comparing unexposed controls to children subjected to biological monotherapy (RR, = 0.92; 95% CI, 0-70; 1.20], thiopurine monotherapy (RR, = 0.90; 95% CI, 0.64-1.28) or combination therapy (RR, = 0.93; 95% CI, 0.66-1.33) in utero.⁴

In the present study, we also report a significantly increased risk of severe infection within 5 years of birth in the combination treatment group but believe the difference to be fueled by a few children with many events rather than an actual increased risk that presents itself after several years of stability.

The present study has important strengths and distinguishes itself by regarding the risk of serious infections in offspring exposed to different medical treatments in utero from various angles, thereby allowing comparison to most other large studies, regardless of measure of association or exposure. The study cohort comprises very well-described pregnancies and thus time of exposure in the children concerning antenatal medical treatment and disease activity in the mother. As all data were collected prospectively; there is no recall bias.

We acknowledge several limitations of the study. First, although the cohort size was large, the number of events in the form of severe infections was limited. Secondly, we did not know the vaccination status of the mothers, which may, in turn, cause a bias due to an a priori increased risk of infections in children of unvaccinated mothers. Thirdly, we did not know the breastfeeding status of the children, and thus breastfeeding could not be assessed as a possible protective factor. Conversely, a slightly increased frequency of smoking in the treatment groups may have caused an increased risk of infections in the children, both during pregnancy and by continued passive smoking postpartum. Lastly, there was a higher incidence of extensive disease and disease activity in the treatment groups compared with the controls; and although we adjusted for the latter in the statistical analyses, we could not assess the effect of more severe disease in the mother.

In summary, our research reinforces the importance of disease management and full-term pregnancies, given that preterm birth heightens the risk of severe infections in both newborns and young children. Additionally, our findings raise concerns about elevated serious infection risks in children exposed to systemic steroids during the third trimester. Our study did not confirm a significant risk of infections after in utero exposure to either mono- or combination therapy with azathioprine. However, given the number of newly highly relevant studies published in recent years, the majority of which show a tendency towards an increased risk of serious infections in children subjected to combination treatment in utero, we suggest the conduct of a meta-analysis elucidating the risk of infant infections after exposure to biologics in 3 different categories: (1) monotherapy vs controls, (2) combination therapy vs controls, and (3) monotherapy vs combination therapy.

Supplementary Data

Supplementary data is available at Inflammatory Bowel Diseases online.

Author Contributions

T.V. was responsible for the study concept and design, obtaining funding, study supervision, acquisition of data, statistical analysis, interpretation of data, drafting of the manuscript, and critical revision of the manuscript for important intellectual content. J.K. and M.J. were responsible for the study concept and design, study supervision, interpretation of data, manuscript drafting, and critical revision of the manuscript for important intellectual content. I.H.M. was responsible for the acquisition of data and critical revision of the manuscript for important academic content. R.B.H. and S.F. were responsible for interpretation of data and critical revision of the manuscript for important intellectual content. All authors approved the final version of the article.

Funding

The study was supported by unrestricted grants by Aarhus University, Denmark, the Colitis-Crohn Organization Denmark, the Danish Rheumatism Association, Aase & Ejnar Danielsen’s Foundation, Louis-Hansen’s Foundation and the A.P. Moeller Foundation of the Advancement of Medical Science. The external funders had no involvement in any aspect of the study and writing of the report.

Conflicts of Interest

M.J. has received research grants for other investigator-driven studies from Takeda and Novo Nordisk Foundation (grant no. NNF23OC0081717), has received consultation fees from Ferring and Takeda, and has received speaker’s fees from MSD, Ferring, Tillotts Pharma, and Takeda. J.K. has served on the advisory board of Gilead, Takeda, and Janssen and has received speaker’s fee from Pfizer. The remaining authors disclose no conflicts.

Data Availability

Deidentified individual participant data can be made available upon request to the corresponding author with investigator support after approval of a new study proposal, a signed agreement from each participant, and a signed data access agreement with the participating sites.

References