Ursodeoxycholic acid in intrahepatic cholestasis of pregnancy: a systematic review and individual participant data meta-analysis

Abstract

Background: Ursodeoxycholic acid is commonly used to treat intrahepatic cholestasis of pregnancy, yet its largest trial detected minimal benefit for a composite outcome (stillbirth, preterm birth, and neonatal unit admission). We aimed to examine whether ursodeoxycholic acid affects specific adverse perinatal outcomes.

Methods: In this systematic review and individual participant data meta-analysis, we searched PubMed, Web of Science, Embase, MEDLINE, CINAHL, Global Health, MIDIRS, and Cochrane without language restrictions for relevant articles published between database inception, and Jan 1, 2020, using search terms referencing intrahepatic cholestasis of pregnancy, ursodeoxycholic acid, and perinatal outcomes. Eligible studies had 30 or more study participants and reported on at least one individual with intrahepatic cholestasis of pregnancy and bile acid concentrations of 40 μmol/L or more. We also included two unpublished cohort studies. Individual participant data were collected from the authors of selected studies. The primary outcome was the prevalence of stillbirth, for which we anticipated there would be insufficient data to achieve statistical power. Therefore, we included a composite of stillbirth and preterm birth as a main secondary outcome. A mixed-effects meta-analysis was done using multi-level modelling and adjusting for bile acid concentration, parity, and multifetal pregnancy. Individual participant data analyses were done for all studies and in different subgroups, which were produced by limiting analyses to randomised controlled trials only, singleton pregnancies only, or two-arm studies only. This study is registered with PROSPERO, CRD42019131495.

Findings: The authors of the 85 studies fulfilling our inclusion criteria were contacted. Individual participant data from 6974 women in 34 studies were included in the meta-analysis, of whom 4726 (67·8%) took ursodeoxycholic acid. Stillbirth occurred in 35 (0·7%) of 5097 fetuses among women with intrahepatic cholestasis of pregnancy treated with ursodeoxycholic acid and in 12 (0·6%) of 2038 fetuses among women with intrahepatic cholestasis of pregnancy not treated with ursodeoxycholic acid (adjusted odds ratio [aOR] 1·04, 95% CI 0·35-3·07; p=0·95). Ursodeoxycholic acid treatment also had no effect on the prevalence of stillbirth when considering only randomised controlled trials (aOR 0·29, 95% CI 0·04-2·42; p=0·25). Ursodeoxycholic acid treatment had no effect on the prevalence of the composite outcome in all studies (aOR 1·28, 95% CI 0·86-1·91; p=0·22), but was associated with a reduced composite outcome when considering only randomised controlled trials (0·60, 0·39-0·91; p=0·016).

Interpretation: Ursodeoxycholic acid treatment had no significant effect on the prevalence of stillbirth in women with intrahepatic cholestasis of pregnancy, but our analysis was probably limited by the low overall event rate. However, when considering only randomised controlled trials, ursodeoxycholic acid was associated with a reduction in stillbirth in combination with preterm birth, providing evidence for the clinical benefit of antenatal ursodeoxycholic acid treatment.

Funding: Tommy's, the Wellcome Trust, ICP Support, and the National Institute for Health Research.

Figure 1

Study selection Adapted from PRISMA and PRISMA individual participant data. ICP=intrahepatic cholestasis of pregnancy. PRISMA=Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 2

Kaplan–Meier plots of the prevalence of spontaneous preterm birth by gestational week of birth, according to ursodeoxycholic acid use and disease severity at randomisation Analyses were done by use of individual participant data from women with singleton pregnancies participating in randomised controlled trials. Kaplan–Meier plots for all women (A), women with baseline bile acid concentrations less than 40 μmol/L (B), women with baseline bile acid concentrations between 40 μmol/L and 100 μmol/L (C), and women with baseline bile acid concentrations of 100 μmol/L or more (D). HRs compare women randomly assigned to ursodeoxycholic acid treatment with those randomly assigned to placebo. (E) All women by baseline bile acid concentration; HRs compare women with baseline bile acid concentrations of 40·0–99·9 μmol/L and 100·0 μmol/L or more to those with baseline bile acid concentrations less than 40·0 μmol/L. HR=hazard ratio.

Figure 3

Stillbirth and preterm birth in women in RCTs of ursodeoxycholic acid treatment (A) Stillbirth. (B) All preterm births before 37 gestational weeks. (C) Spontaneous preterm births before 37 gestational weeks. Aggregate published data were used. Stillbirth was analysed by number of fetuses, except for Glantz and colleagues (in which stillbirth was analysed by number of pregnancies). The black diamonds are the individual study point estimates, the grey boxes reflect the weight of the individual study, the horizontal lines represent the CIs of the effect estimates, the white diamonds represent the pooled ORs and CIs, and the vertical dotted line represents the pooled OR. Study weight was calculated from the inverse variance. Preterm birth was analysed by number of fetuses, except for Glantz and colleagues, Kondrackiene and colleagues, Roncaglia and colleagues, and Palma and colleagues (in which preterm birth was analysed by number of pregnancies). OR=odds ratio.