Primary biliary cholangitis in pregnancy: A systematic review with meta-analysis

Hydr El Jmly , , , Guy D Eslik , Mrtin Weltmn ,

a Department of Gastroenterology and Hepatology, Nepean Hospital, Penrith, New South Wales, Australia

b Nepean Clinical School, The University of Sydney, Penrith, New South Wales, Australia

c NHMRC Centre for Research Excellence in Digestive Health, Hunter Medical Research Institute, The University of Newcastle, Newcastle, New South Wales,Australia

Keywords:Pregnancy Fetal Maternal Primary biliary cholangitis Miscarriage

ABSTRACT

Introduction

Primary biliary cholangitis (PBC) is a rare chronic autoimmune liver disease characterized by progressive inflammatory processes targeting the small caliber intrahepatic bile ducts. Proposed predisposing factors to the disease include genetic susceptibility, epigenetic changes and environmental risks [1] . PBC, like many autoimmune diseases, has a female predominance and it primarily affects middle-aged patients [1] . Histologically, inflammation is marked by biliary epithelial cell senescence, cellular apoptosis and duct loss [2] . In early population studies, over 25% of patients diagnosed with PBC developed cirrhosis within the next subsequent decade [3] . The rollout and therapeutic use of ursodeoxycholic acid (UDCA) since 1987 for PBC led to a substantial improvement in prognosis, delaying histological progression and lowering the risk of complications such as the development of esophageal varices [4] .

The predisposition of PBC prevalence in middle-aged female patients and the presence of cirrhosis in a subset of those patients justify a thorough assessment of the risk of these pregnancies.Many of the studies investigated pregnancy outcomes during PBC based on broad cross-sectional analysis that also assessed various risk factors associated with the condition from personal medical history, family history and reproductive history and were occasionally done retrospectively. The latest American Association for the Study of Liver Diseases (AASLD) practice guidelines for PBC [5] attested to the limited data regarding “infant outcomes” in pregnancy. Based on existing observational study findings, we hypothesized that most PBC pregnancies without cirrhosis do not have a significantly increased risk of maternal or fetal complications.We also expected an increased risk of patients having personal or family history of other autoimmune diseases. The aim of this meta-analysis was to assess the clinical impact of PBC on pregnant females, lifestyle associations and any link to other medical conditions in order to provide more information that can educate this subset of patients and help relay fears during their journey to motherhood.

Methods

Study protocol

We followed the Preferred Reporting Items for Systematic Reviews (PRISMA) guidelines [6] . A systematic search was conducted using MEDLINE (Ovid), PubMed, Google Scholar, Embase,and the ClinicalTrials.gov database up to February, 2022 to identify relevant articles. PubMed search items included primary biliary cholangitis/cirrhosis AND pregnancy, OR primary biliary cholangitis/cirrhosis AND gestation, OR primary biliary cholangitis/cirrhosis AND birth, OR primary biliary cholangitis/cirrhosis in pregnancy. Furthermore Embase, MEDLINE (Ovid) and Cochrane library searches utilized MeSH terms. The reference lists of relevant articles were also searched for appropriate studies. The search included an assessment of unpublished literature.

Study selection

We included studies that met the following inclusion criteria: 1) immunological criteria [positive antimitochondrial antibody (AMA)] and any of histological (liver biopsy findings) or a cholestatic picture [elevated alkaline phosphatase (ALP)] with a chronic picture in the absence of other disease; 2) female of any age group with at least one pregnancy. Eligible study designs consisted of observational studies and case series, both published and non-published projects in the English language. The primary outcome was a comparison between PBC maternal all-cause mortality and neonatal mortality to the non-PBC pregnancies and the occurrence of common adverse neonatal outcomes. Secondary outcomes included the prevalence associations of lifestyle and other medical comorbidities with PBC. Exclusion criteria were studies with a pediatric population, case reports, post publication peer reviews, data pertaining to male subjects and studies with autoimmune hepatitis(AIH)/PBC overlap syndrome.

Data extraction

The data extraction was performed by two authors (El Jamaly H and Eslick GD) using a standardized data extraction form, collecting information on the following maternal outcomes: pregnancyrelated outcomes of miscarriage, abortion and fertility history.Lifestyle factors included smoking status, the use of the oral contraceptive pill (OCP) and hormonal replacement therapy (HRT). Fetal outcomes of neonatal mortality, premature birth, congenital defects and the mode of delivery (vaginal birth vs. caesarean section)were extracted. Data were also collected for lifetime exposure to genito-urinary infections. All available data relating to maternal or neonatal outcome measures in PBC pregnancies were coded. All data analysis was based on the number of patients rather than the number of pregnancies. Additional information was also collected on the publication year, study design, number of cases, number of controls, total sample size, temporal direction, and continent.We contacted one author from one study to request further information on pregnancy related outcomes and the total number of pregnancies though we did not receive a reply. The literature search was conducted from June 2020 to February 2022. Studies published at any time point were included up to the latter date of the search.

Fig. 1. ROBINS-I risk of bias tool in non-randomized studies.

Quality assessment

Quality assessment of the included observational studies was performed using the Newcastle-Ottawa Scale and case series using the MOGA scale [ 7 , 8 ]. These scales assess each article on selection of the study groups, confounding, and comparability of the groups and ascertainment of the outcome of interest. The risk of bias was summarized using the risk-of-bias ROBINS-I tool for nonrandomized studies of intervention [9] ( Fig. 1 ). Finally, the quality of the evidence for the outcomes was rated using the GRADE system [10] . Agreement regarding the marking of each study was reached by consensus.

Statistical analysis

We used a random-effects model for this analysis, using prevalence rates [event rates (ER)] and odds ratios (ORs) with 95% confidence intervals (CIs) [11] . A random-effects model was used due to varying effect sizes across the included studies. We tested heterogeneity using theI2statistic, which represents the percentage of the total variability across studies which is due to heterogeneity.I2values of 50% or above correspond to a substantial degree of heterogeneity [12] . We assessed publication bias using the Egger’s regression model [13] . All analyses were performed with Comprehensive Meta-analysis (Version 3.0, Biostat, Englewood, NJ, USA).Pooled ORs were based on two-armed case control and crosssectional studies, whilst prevalence data were obtained from all included studies.

Results

Search strategy

A total of 920 publications were searched. Of these, 784 publications were eliminated based on title and abstract search. The remaining 136 publications were independently assessed and 9 fulfilled the inclusion criteria. Of those remaining 9 articles, a detailed search in their reference lists yielded two further studies which met the study’s criteria for inclusion. In total, 11 articles [14–24] were included in this meta-analysis ( Fig. 2 ). The quality assessment and the quality of the evidence for outcomes were shown in Tables 1 and 2 .

Study characteristics

Fig. 2. Flowchart of the search strategy used. AIH: autoimmune hepatitis; PBC: primary biliary cholangitis.

The 11 studies included a total of 2179 female patients. Five of the larger studies reported pregnancies in mean or median per patient and the total pregnancies in the analysis were estimated at 6073. However, due to the consistent reporting of outcomes per patient in all the included studies, this meta-analysis also reported pregnancy outcomes in such manner. Three studies were case-control studies, four were cross-sectional studies, two were cohort studies and two were case series. The study designs included six retrospective studies, one prospective study and four questionnaire-based studies. Participants were enrolled between 1979 and 2019. The length of maternal follow-up ranged from 12 to 17 months (n= 2). Majority of the studies (6 of 11) were from Europe, three were from the United States whilst one each was from Canada and Turkey. The studies were conducted in different databases and timeframes, hence it was extremely unlikely for the same patient to be enrolled in multiple studies. Five studies assessed concurrent autoimmune diseases. Less than half of the studies (5 of 11) reported on biochemical parameters of liver function,while two studies reported serological parameters used in PBC.Table 3 presents a summary of individual study characteristics. The cohort studies results were reported in Tables 3 and 4 .

Maternal outcomes

The prevalence rates for various maternal outcomes are shown in Table 4 . The mean maternal age during pregnancy in the PBC cohort was 24.64 years (95% CI: 22.37-26.91). There were no reported maternal deaths or decompensated cirrhosis events occurring within pregnancy and up to 6 months post-partum. Females with PBC had higher odds for abortion (OR = 1.50, 95% CI: 1.09-2.07;P= 0.01; three studies) [ 15 , 18 , 23 ] compared to the control group, and there was no evidence of heterogeneity (I2= 0%). The odds of gestational diabetes mellitus (GDM) were higher in PBC patients (OR = 1.71, 95% CI: 0.28-10.39;P= 0.56; one study) [17] but were not statistically significant. Pregnant females with PBC were significantly more likely to have a vaginal delivery (OR = 1.69,95% CI: 1.33-2.14;P＜0.001; two studies) [ 15 , 23 ] compared to the control group with low level of heterogeneity (I2＜0.001%),and slightly less likely to undergo a caesarean section delivery(OR = 0.85, 95% CI: 0.68-1.06;P= 0.15; three studies) [ 15 , 22 , 23 ]with low level heterogeneity (I2＜0.001%).

There was no difference in PBC females or controls in overall fertility history, with the odds for never conceiving before in PBC patients being slightly lower (OR = 0.89, 95% CI: 0.69-1.15;P= 0.36; four studies) [ 17 , 18 , 20 , 22 ] with low evidence of heterogeneity (I2= 8.67%). Similarly, the odds for ever conceiving in PBC patients were slightly higher (OR = 1.13, 95% CI: 0.87-1.46;P= 0.36; four studies) [ 17 , 18 , 20 , 22 ] with low evidence of heterogeneity (I2= 8.67%).

The odds for ever smokers were significantly elevated in PBC patients (OR = 1.95, 95% CI: 1.17-3.23;P= 0.01; one study) [17] from a single study. The odds for ever using OCP in PBC patients were not different to the control population (OR = 1.07,95% CI: 0.75-1.51;P= 0.72; four studies) [ 17 , 18 , 22 , 24 ] with a high degree of heterogeneity (I2= 78.41%). On the other hand, the odds for ever using hormone replacement therapy (HRT) were higher in PBC group without reaching statistical significance (OR = 1.18, 95%CI: 0.86-1.63;P= 0.31; four studies) [ 17 , 18 , 22 , 24 ] with a substantial level of heterogeneity (I2= 76.75%).

Compared to the control population, PBC patients had a nonsignificant trend towards an earlier age of menarche (standard difference mean = -0.052, 95% CI: -0.125 to 0.02;P= 0.16;four studies) [ 17 , 18 , 22 , 23 ] and menopause (standard difference mean = -0.044, 95% CI: -0.131 to 0.043;P= 0.32; three studies) [ 18 , 22 , 23 ]. The mean gravidity in PBC was higher than con-trols (standard difference mean = 0.232, 95% CI: -0.010 to 0.474;P= 0.06; five studies) [ 15 , 17 , 18 , 22 , 23 ] but did not reach significance. PBC patients also showed a trend towards an earlier age of conception without reaching statistical significance (standard difference mean = -0.123, 95% CI: -0.247 to 0.002;P= 0.053; five studies) [ 15 , 17 , 18 , 22 , 23 ].

Table 1 Individual study quality assessment.

Table 2 GRADE assessment.

Table 3 Individual study characteristics.

Pregnancy timing in relation to PBC diagnosis

Based on gravidity figures, the pooled prevalence of pregnancy related female patient data preceding a formal diagnosis of PBC was 84.1% (95% CI: 24.7%-98.8%; nine studies) [ 14-19 , 21-23 ] with a high level of heterogeneity (I2= 92.83%). However, the pooledprevalence of pregnancies occurring during or after a formal PBC diagnosis was notably lower at 15.9% (95% CI: 1.2%-75.3%;nine studies) [ 14-19 , 21-23 ] with a high degree of heterogeneity(I2= 92.83%).

Table 4 Prevalence data of PBC related outcomes.

Pruritus and pregnancy in PBC

Women with PBC had increased odds of pruritus or jaundice during pregnancy (OR = 3.37, 95% CI: 0.89-12.82;P= 0.075; three studies) [ 15 , 18 , 20 ] compared to the control group with substantial evidence of heterogeneity (I2= 77.01%). The prevalence of pruritus during pregnancy and a worsening of cholestasis in pregnancy in PBC patients were 35% and 39%, respectively ( Table 4 ). Furthermore, whilst the vast majority of pregnant women with PBC used UDCA at some point in pregnancy (86%), just over a half continued to utilize UDCA well throughout pregnancy (55%).

Fetal outcomes

The prevalence rates for various fetal outcomes are shown in Table 4 . Below, we present the odds ratio for some of these outcomes where applicable. Fetal mortality was notably elevated (OR = 21.68, 95% CI: 1.19-394.10;P= 0.04; one study) [15] whereby all events occurred with pregnancies that preceded a formal PBC diagnosis. Pregnant females with PBC were significantly more likely to have a miscarriage (OR = 1.27, 95% CI:1.02-1.58;P= 0.03; four studies) [ 15 , 17 , 18 , 23 ] with no evidence of heterogeneity (I2= 0.0%). The odds for congenital defects were also lower (OR = 0.75, 95% CI: 0.21-2.72;P= 0.66; one study) [17] . PBC patients had almost 50% increased odds of premature birth compared to the control population without reaching statistical significance (OR = 1.48, 95% CI: 0.84-2.60;P= 0.18; two studies) [ 15 , 17 ]with no evidence of heterogeneity (I2= 0%) ( Fig. 3 ).

Cirrhosis and transplant

The occurrence of cirrhosis in PBC patients was 13% (95%CI: 8%-20%, four studies with low level of heterogeneityI2= 15.93%) [ 14-16 , 19 ]. There was a paucity of data which precluded an assessment of pregnancy related outcomes between cirrhotic and non-cirrhotic patients with PBC. There were no female patients with known baseline liver transplants. However, in one study [18] , a 6% prevalence of baseline liver transplantations was present, from both sexes.

Subgroup analysis: Pregnancies during or after PBC diagnosis

An analysis of the maternal and fatal outcomes of pregnancies occurring during or after PBC formal diagnosis was conducted as a small subset of the overall study. Overall, 5 studies were included with a total 98 pregnancies. The design of the studies and available data did not allow for direct group comparison and we present prevalence rates below. There was neither maternal nor neonatal mortality. The prevalence for miscarriage is 20% (95% CI: 6%-48%;four studies) [ 14-16 , 21 ] with a moderate degree of heterogeneity(I2= 48.82%). The prevalence for premature birth was 14% (95%CI: 4%-35%; five studies) [ 14-16 , 19 , 21 ] with a substantial degree of heterogeneity (I2= 61.95%). The prevalence for congenital defects was 3% (95% CI: 0.6%-13%; two studies) [ 16 , 17 ] with no evidence of heterogeneity (I2= 0%). The prevalence of prepartum, antenatal and postpartum biochemical disease flares in this subgroup were 11% (95% CI: 4%-25%,I2= 0%; two studies) [ 14 , 16 ], 12% (95% CI:2%-44%,I2= 60.43%; three studies) [14–16] , and 57% (95% CI: 45%-58%,I2= 0%; three studies) [14–16] , respectively.

Lifetime genital infection risk

The odds for lifetime events of vaginal infections were higher in PBC patients without reaching significance (OR = 1.48, 95% CI:0.80-2.74,P= 0.21; three studies) [ 17 , 22 , 23 ] with a substantial degree of heterogeneity (I2= 83.91%). The odds for lifetime events of pelvic inflammatory disease were also higher without reaching significance in the PBC group (OR = 1.23, 95% CI: 0.75-2.01;P= 0.41;two studies) [ 17 , 22 ] with no evidence of heterogeneity (I2= 0%).

Discussion

This meta-analysis provides collective evidence that pregnant females with PBC have significantly high rates of lifetime miscarriage, abortion and lifetime smoking compared to matched controls from national healthcare registries. The odds for GDM and premature birth were also elevated without reaching statistical significance. The elevated odds for premature birth and miscarriage may relate to the underlying state of chronic liver disease or to active PBC disease amongst other related established factors, including malnutrition and genitourinary infections [25] . UDCA utilization in pregnancy is essential for the stabilization of disease activity at a period of a substantial risk of biochemical flares which may progress to hepatic decompensation. In fact, outside of pregnancy, inconsistent treatment or incomplete UDCA response is a known predictor of decompensation in PBC [26] , whilst overall,UDCA treatment significantly reduces all-cause mortality [27] . That correlates to the strong recommendation by the American College of Gastroenterology (ACG) Liver Disease and Pregnancy Guidelines on continuation of UDCA throughout pregnancy [28] . These data are important for clinicians managing these patients before, during and after pregnancy.

The population in this analysis had a low prevalence of cirrhosis(ER = 13%). Due to the cross-sectional nature of various included studies and the typical onset of PBC in middle age, the vast majority of patients reported on pregnancies that preceded the formal diagnosis of PBC. In the subgroup analysis, pregnancies during or after a PBC diagnosis were overall uneventful for maternal and fetal outcomes [ 17 , 18 , 22 , 23 ]. Conversely, the odds for miscarriage in all PBC patients, irrespective of timing of PBC diagnosis compared to the control population, were significantly elevated(OR = 1.27). One major factor in the relatively safe pregnancy outcomes, apart from the increased association with miscarriage, was the low prevalence of cirrhosis in our population. This reassuring information was important to provide as a clinician to prospective mothers who may have concerns about undergoing a pregnancy with PBC.

Our analysis found modestly elevated odds of lifetime smoking risk with PBC, reaching statistical significance congruent with literature findings [ 17 , 18 , 22 , 29 ]. It has been shown that there is a dose-dependent relationship between active smoking and the degree of fibrosis in established PBC without a histological inflammatory relationship [29] . Educating patients with PBC, including patients undergoing pregnancy, about healthy lifestyle and PBCspecific management choices should focus on smoking cessation and its importance in halting the progression to chronic liver disease. It is well known that active smoking has its own deleterious effects on the fetus, including preterm birth, low birth weight and birth defects [30] . Smoking is an established risk factor for preterm birth [31] . The higher odds for premature birth in our analysis could be partially explained by potentially higher cumulative smoking in the PBC group during pregnancy, which was suggested by the higher lifetime ever-smoking data in PBC patients compared to the control population.

Whilst the exact cause of PBC has not been established, susceptibility environmental factors which have been suggested include infections and smoking. The meta-analysis results support these susceptibility factors by demonstrating a significant association with lifetime smoking and a trend towards a higher risk of lifetime vaginal infections. The association between urinary tract infection and PBC is well-established [ 32 , 33 ]. In a prospective study, elevated odds between PBC and various forms of urinary tract infection (UTI) were replicated though, more importantly, it was demonstrated that those patients who developed UTI in the pre-clinical course of PBC years went on to develop liver condition subsequently, suggesting causation [32] . Various pathogens in particularE.coli,N.AromaticivoransandLactobacillusdelbrueckiihave been implicated in UTI and bacterial vaginitis occurring before or after the diagnosis of PBC. Of importance, the notion of molecular mimicry with cross reactivity is the most popular, in this case suggesting infection triggered autoimmune phenomena [34] . Whilst the strongest infectious link to PBC is of genitourinary origin [32] ,it is feasible that PBC patients have increased risk of infection due to immunosuppressive treatments of various well associated autoimmune conditions, such as type 1 DM, AIH and rheumatoid arthritis, that share the same risk loci in genomic studies [32] . Our analysis was restricted to female subjects and did not demonstrate urinary tract infections as a risk factor, but various included studies within the analysis demonstrated a significant UTI risk when including PBC patients from both genders [ 17 , 18 , 22 ].

The fertility profile of PBC patients was compared to the control population and they were found to have non-significant differences in menarche, menopause, gravidity and the proportion who ever conceived or never conceived. This pooled information is important, as two prior studies [ 15 , 17 ] have shown a statistically significant increase in gravidity in PBC, yet, our analysis combining three further studies [ 18 , 22-23 ] showed a trend towards higher gravidity without reaching statistical significance. Despite no significant difference in menopause age between the groups, there is notably a close to 20% increased odds of ever using HRT by PBC patients.Osteoporosis is associated with PBC with an estimated four-fold higher prevalence compared to the general population [35] . It is feasible that PBC patients in our analysis were commenced earlier on HRT as a treatment or prevention strategy for osteoporosis. A second likely explanation for the utilization of HRT in PBC is the established autoimmune association with other conditions such as primary ovarian insufficiency [36] that warrants early institution of lifelong HRT. However, the presence of primary ovarian insufficiency in significant numbers in this analysis would have been expected to yield an earlier menopause age.

Pruritis is a common presenting feature of PBC that is often precipitated by triggers including clothing, heat and pregnancy [5] .An assessment of the latest Asian Pacific Association for the Study of the Liver (APASL) clinical practice guidelines for PBC was made and that did not specifically address pregnancy; hence the latest AASLD and European Association for the Study of the Liver (EASL)guidelines were considered [ 5 , 37 ]. In our analysis, the progression of cholestasis in pregnancy was indeed demonstrated to occur in almost 40% of cases. As per the 2018 EASL guidelines [37] , UDCA treatment is recommended to be continued throughout pregnancy,owing to its therapeutic efficacy in the general PBC population,and its safety in the puerperium. The data for its safety are limited to case series and expert opinion [37] . The course of pregnancy is generally well-tolerated in non-cirrhotic PBC patients, yet cirrhotic pregnancies take a higher risk of maternal and fetal complications [37] . In our analysis, the studies included were not powered and were provided with specific analyses of outcomes for the cirrhosis subset of patients. This is a recommended area for further research that would allow prognostication and more research informed future guidelines. Pre-conception counselling and planning are recommended for all female PBC patients, especially those with cirrhosis [37] .

PBC has the highest female to male ratio of the autoimmune liver diseases, has a peak incidence in the 5th decade, is rare in those under 25 years of age, and yet, up to 25% of patients are of childbearing age when PBC is diagnosed [38] . Although a subanalysis of pregnancy outcomes in those with established PBC diagnosis during or prior to pregnancy was not possible, it is feasible that those patients with an earlier onset of this autoimmune disease may eventually have a more severe form of PBC, owing to the chronic progressive nature of the condition. A proposed link between earlier onset disease and severity is supported by research evidence that younger PBC patients tend to be more symptomatic with lower perceived quality of life [39] .

The limitations of this analysis included a scant number of studies with pregnant PBC. This is an uncommon condition (incidence: 0.3 to 5.4/100 000; reported prevalence: 1.9-40.2/100 0 0 0) [40–42] . However, the comparative observational studies that were included were methodologically appropriate. The low case numbers and the lack of pregnancy related outcomes across different subpopulations with PBC such as cirrhosis patients or patients on UDCA treatment precluded subgroup analyses of pregnancy related outcomes. Four of the studies were cross-sectional,introducing a potential risk of recall bias. Additionally, the majority of patients underwent pregnancy prior to a formal diagnosis of PBC, although this does not negate the presence of the condition at the time, or a subclinical form of the disease. Finally, we cannot be assured that in the individual studies all negative outcomes were reported accurately.

The strengths of this meta-analysis included a comprehensive literature search strategy drawn from several electronic databases to ensure a broad range of literature from different regions of the world, MeSH terms inclusion and the inclusion of grey literature sources to reduce publication bias. An important strength was the lack of statistical heterogeneity found in various studied outcomes in the analysis. This highlights the consistency of the individual results for each of the various outcomes assessed in this study.

In conclusion, PBC is associated with a significant elevation in the risk of miscarriage, abortion, and vaginal birth compared to controls, and pregnant females with all-time smoking use had high risk for PBC. We found a notable increase in the odds of pruritis or jaundice during pregnancy in PBC. The utilization of UDCA pre-pregnancy, throughout pregnancy and in the early postpartum period and the correlation of this treatment and PBC disease activity with maternal jaundice or pruritis or other maternal and fetal outcomes would be a recommended area of interest for further investigation.

Acknowledgments

None.

CRediT authorship contribution statement

Hydar El Jamaly: Conceptualization, Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Guy D Eslick: Data curation, Formal analysis, Supervision, Writing – review & editing. Martin Weltman: Conceptualization, Supervision,Writing – review & editing.

Funding

None.

Ethical approval

Not needed.

Competing interest

No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.