Introduction Pregnancy In Liver
Introduction: Pregnancy in liver cirrhosis is rare, but is observed increasingly. There is a lack of large-scale comparative research of the complications and maternal and fetal outcome among women with liver cirrhosis. The objective of this systematic review was to analyze the maternal, fetal and obstetrical complications among pregnant women with liver cirrhosis.
Methods: A literature search without data or language restrictions was performed in the databases PubMed and EMBASE. Keywords of the search were pregnant women, liver cirrhosis, portal hypertension, esophageal varices, autoimmune hepatitis, chronic hepatitis B or C, primary biliary cirrhosis, congenital liver fibrosis, complications and outcome. Random-effects model was used for meta-analysis.
Results: Nine cohort studies were found to be eligible and five studies were included in the meta-analysis. In total 940 pregnancies were studied. Pregnant women with liver cirrhosis were more likely to deliver preterm (sOR 5.18 ). Obstetric complications that significantly occur more often in women with liver cirrhosis than in healthy pregnant women were postpartum hemorrhage (sOR 3.99 ) and preeclampsia (sOR 3.59 ). The maternal mortality rate was 2.7%. The most common cause of maternal mortality was variceal bleeding during vaginal delivery. Hepatic decompensation occurred in 5-21% of the pregnancies among cirrhotic women. Neonatal death in studies was between 0 and 6.5%.
Conclusion: The presence of liver cirrhosis in pregnant women leads to a remarkably high risk on maternal, fetal and obstetrical complications. This information is relevant for clinicians to be prepared for complications in the management of reproduction of cirrhotic women.
Table of contents
Pregnancy is a rare event in women with liver cirrhosis. The prevalence of cirrhosis in women of reproductive age is 0.045% and the incidence of cirrhosis in pregnancy is reported as 1 per approximately 4500 pregnancies . The most common etiologies of liver cirrhosis are alcoholic liver disease and viral hepatitis: hepatitis B virus as well as hepatitis C. Less frequent are congenital, autoimmune and metabolic diseases. Pregnancy in cirrhotic women is rare, in the first place because reproductive women are not the common patient group which suffers from liver cirrhosis. Often liver cirrhosis develops years after the reproductive phase and it is more prevalent in men than in women. Secondly, women with liver cirrhosis have decreased fertility as a result of anovulation and amenorrhea due to hypothalamic-pituitary-gonadal dysfunction . However, pregnancy and cirrhosis is a rare combination, it occurs occasionally and is becoming more common over the last decades owing to improvements in the treatment of liver cirrhosis. Issues that matter in such cases are twofold: (1) the effect of liver cirrhosis on the pregnancy and fetus, and (2) how will the pregnancy affect the health of the patient.
Previous studies have shown that pregnancy in cirrhosis is attended with high risk on obstetrical complications, including preterm delivery, caesarean section, postpartum hemorrhage and placental diseases . In addition, a frequently occurring and very severe maternal complication is variceal hemorrhage during the second or third trimester or during vaginal delivery . Studies concerning pregnancies with this health condition have suggested that obstetric complications occur often and that the maternal and fetal mortality and morbidity is higher than in the normal population . There are few reports among pregnancies in women with this health condition and the majority of studies are case series, case reports or studies with small subgroups of cirrhosis, for example . There is a lack of large-scale comparative research of the liver specific complications, obstetric complications and maternal and fetal outcome among women with liver cirrhosis. Therefore, there is need for a systematic review of the literature.
This study will be relevant for clinicians, especially for obstetricians and hepato-gastroenterologists, to be prepared for any complications in the management of pregnant patients with cirrhosis of the liver. Furthermore, cirrhotic women in the reproductive phase or during pregnancy will benefit from good counselling about the liver specific complications, such as hepatic decompensation and variceal bleeding, as well as the obstetric complications and the outcome of their infant. The objective of this systematic review was to analyze the maternal, fetal and obstetrical complications among pregnant women with liver cirrhosis.
A study protocol was written to describe the methods of the study in detail. This protocol includes the search strategy and method of screening, the criteria to include and exclude studies, the methodological quality assessment of included studies, the outcome domains and measures of interest and at least the method of data extraction. A short version of the study protocol was published on the international prospective register of systematic reviews, PROSPERO, in April 2018.
Search strategy and screening
A literature search without data or language restrictions was performed in the databases PubMed and EMBASE on the 16th of February 2018. Preceding, the search was checked by the clinical librarian of our academic center to generate the search as sensitive as possible. Keywords of the search were pregnant women, liver cirrhosis, portal hypertension, esophageal varices, autoimmune hepatitis, chronic hepatitis B or C, primary biliary cirrhosis, congenital liver fibrosis, complications and outcome. For the complete search strategy see appendix 1
After deduplication, the publications describing pregnancy in cirrhotic women or describing subgroups of women with liver diseases which could end up in liver cirrhosis were sought by two reviewers (LS and BT). Title and abstract of all publications were screened independently by the two reviewers using Covidence, a tool for screening and data extraction of systematic reviews developed by Cochrane. Relevant publications that potentially met the inclusion criteria were extracted. Subsequently, the full text was assessed for eligibility by two reviewers before including studies. In case of disagreement in any stage of the study selection process the reviewers resolved this by discussion. If necessary, a third reviewer was consulted (RP).
Inclusion and exclusion criteria
The inclusion criterion of patient’s population were pregnant women with liver cirrhosis due to one of the etiologies described below of pregnant women with liver cirrhosis complicated by portal hypertension or esophageal varices. The liver cirrhosis should have been diagnosed by biopsy or imaging combined with laboratory findings. The etiologies included autoimmune hepatitis, viral hepatitis including chronic hepatitis B or C, alcoholic liver disease, primary biliary cirrhosis, congenital liver fibrosis, Wilson’s disease or more uncommon causes. In case of studies describing a subgroup of liver cirrhosis, the subgroup should be larger than five patients. Results have to contain at least one item of each interested outcome domain. Outcome domains and measures are clarified in ‘outcome domains/measures and data extraction’. Studies with the designs cohort study or case-control study were included.
Exclusion criteria were non-human studies, studies outside pregnancy and case reports, case series, opinions or reviews.
For quality assessment, we used the Newcastle Ottawa scale for cohort studies (appendix 2). This is a representative tool developed for meta-analysis of observational studies. In this scale studies are assessed by their methodological quality on three domains. These domains are selection, comparability and ascertainment of outcome . Studies were assessed by reading the full text. Retrospective studies, per definition, had the outcome of interest present at start study. The minimum follow-up time to encounter all outcomes was set on 9 months. The highest score of stars in the Newcastle Ottawa scale to aware per study is 9 points. Since a standard criterion for what constitutes a high, medium or low-quality study has not yet been universally established, we decided to score studies with 7 or more stars as high quality. Studies with less than 4 stars were classified as low quality and studies with stars between 4 and 6 were classified as medium quality. A study with low quality was no reason for exclusion, although it was taken into account interpreting the results.
Outcome domains/measures and data extraction
The outcomes of interest are divided into three domains, maternal complications, obstetrical complications and fetal outcome or complications. Maternal complications include maternal mortality, defined as maternal death during pregnancy or within 42 days of termination of pregnancy, maternal morbidity and hepatic decompensation, defined as presence of jaundice, ascites, hepatic encephalopathy or gastrointestinal bleeding. Interested maternal complications are as well variceal bleeding, intrahepatic cholestasis of pregnancy and splenic artery aneurysm. Obstetric complications include miscarriage (pregnancy loss before 20th week, include elective terminations), caesarean delivery, preterm delivery (birth 37 weeks), postpartum hemorrhage (loss of 500 ml blood within 24 hours during or after delivery), puerperal infections, hypertensive disorders of pregnancy, include PIH (systolic blood pressure ≥90 mmHg or diastolic blood pressure ≥140 mmHg), preeclampsia (hypertension and proteinuria) and HELLP syndrome (diagnosed by laboratory abnormalities). Placental diseases are of interest as well, including abruption placentae, placenta previa and placental insufficiency. Fetal outcome or complications include live births, neonatal mortality (death within 28 days after pregnancy), intrauterine fetal demise (pregnancy loss at or after 20th week of gestation; also known as stillbirths or fetal death), admission at neonatal intensive care unit, intrauterine growth restriction (include neonate small for gestational age (SGA) or low birth weight (2500 g)) and congenital abnormalities. As well of interest are the mean gestational age at delivery and the mean birth weight of infants.
Data of includes studies was extracted by using a predefined and structured data extraction form especially made and bespoken for the systematic review. At least extracted of each included study was the first author, year and journal of publication, inclusion and exclusion criteria, method of data collection, number of patients with cirrhosis and total number of patients, number of pregnancies, maternal characteristics, fetal characteristics and already describes complications of interest. See appendix 3 for the data extraction form. In case of studies missing data or studies with no access to the full text, first authors were contacted at least two times.
The odds ratios (OR) and 95% confident intervals (CI) comparing women with liver cirrhosis and controls were presented in a forest plot using the RevMen 5.3 software . Meta-analysis was performed using the generic inverse variance method. A p-value 0.05 was considered statistical significant. The random effects model was used as analysis model, since we expected some heterogeneity inter studies. The I2 was used for testing statistical heterogeneity. An I2 value greater than 75% was considered an indication of heterogeneity between studies. Statistical heterogeneity implies variation in the result which cannot be accounted for by chance.
Results are presented in three categories, fetal outcome and complications, obstetric complications and maternal complications. At the end of the results section table 2 is inserted and contains all findings per study. Description of general information per study and the quality assessment is inserted prior to the outcome measures of this systematic review.
The literature search through databases Pubmed and EMBASE resulted in 2651 publications. Nine studies are included in this systematic review . Four studies did not have a healthy pregnancy control group, therefore these studies could not be involved in the meta-analysis , but are included in the systematic review. Figure 1 shows the flow chart of study selection. Twelve studies were eligible to be included in this systematic review, but the study of Heneghan et al. and two other studies of Westbrook et al. were excluded of this systematic review. Reason for that was to avoid analyzing the same population two times, even though these studies contain relevant information and meet the eligibility criteria. The study of Westbrook et al. was decided to include since this one contained the most useful information for this systematic review. Murthy et al. and Shaheen et al. extracted their data from the same database over the same period, but it is unclear whether or not they studied the same patient population, so both studies are included.
Table 1 shows the study characteristics and maternal and fetal characteristics of the included studies. Eight studies were retrospective cohort studies and one study was a prospective cohort study. The information of five studies was based on a cohort that entered the specific care institution. Besides, the information of three other studies was obtained from large databases and one study sent a questionnaire and analyzed this. The total number of studied pregnancies in cirrhotic women is 940 in nine studies. Studies took place between 1982 and 2016. The most common etiologies of liver cirrhosis were autoimmune hepatitis, chronic hepatitis B and C and alcoholic liver disease. Shaheen et al., Murthy et al. and Rasheed et al. studied the largest patient populations in their studies. The mean maternal age was 30 years (range 16-40). The mean gestational age of the infants at delivery was 36 weeks. Only 2 studies, Jena et al and Palatnik et al, reported median birth weight, respectively 2.4±0.65 and 2.7±0.8 kg. The total number of pregnancies included in the meta-analyses is 724 in the liver cirrhosis group and 2954022 in the control group.
First author Year
Place Study design and obtained data Study period Controls Number of controls Quality Number of patients with cirrhosis Etiology of cirrhosis Number of pregnancies with cirrhosis Mean/median age (years) Gestational age at delivery (wk) Mean/median birth weight (kg)
Quality assessment was performed and resulted in four studies of high quality (Palatnik et al. Puljic et al. , Rasheed et al. and Shaheen et al.), four studies of medium quality (Jena et al. , Britton et al. Westbrook et al. and Murthy et al.) and one study (Borssén et al ) of low quality. Since exclusively cohort studies were included in this systematic review, only the Newcastle Ottawa scale for cohort studies was necessary. Only the study of Rasheed et al. awarded a star at the point ‘outcome of interest not present at start study’, since only this study was prospective. In all studies, the diagnosis liver cirrhosis was based on pathological examination or imaging combined with laboratory findings in nearly all cases. All studies, excepted the one of Borssén et al., retrieved data on outcomes and complications from clinical records or databases. Studies could aware two points at comparability: one for no-disease pregnant controls and another if pregnant controls were also linked at age, body mass index, parity or year of delivery. Since most studies were retrospective, a star for ‘adequacy of follow up of cohorts’ was allocated if studies accounted for all know pregnancies in cirrhotic patients in the hospital/database the study was enrolled in. Most studies did not report over which time period they retrieved data. Only Puljic et al. and Rasheed et al. reported this, therefore they awarded a star at follow-up time at least 9 months. Complete quality adjustment of the included studies is enclosed in appendix 4.
Fetal outcome and complications
A meta-analysis of intrauterine growth restriction of four studies could be conducted. It can be seen from the forest plot in figure 2 that the summary odds ratio for intrauterine growth restriction is 4.23 ( P= 0.003), favoring the control group. Since the test for heterogeneity is 75%, we should be careful by interpreting these results. Two studies in the meta-analysis reported small for gestational age, both defined as birth weight less than 10th percentile for gestational age, and two studies reported intrauterine growth restriction. Intrauterine growth restriction in the included studies ranges from 16 to 36%. The studies of Jena et al. and Westbrook et al. which were not included in the meta analyses found percentages of intrauterine growth restriction of 41 respectively 23%. Three studies did not report intrauterine growth restriction.
Figure 2 forest plot of intrauterine growth restriction in liver cirrhotic pregnant women vs healthy pregnant women
Seven studies reported neonatal death. In two studies, neonatal death did not occur, therefore the range of neonatal death is 0-6.5% (mean 3.0%). Only two studies, Jena et al. and Westbrook et al., reported a live birth rate of 41% respectively 58%, this live birth rate represented the ratio of all conceptions that ended up in live births. Three studies reported a rate of admission on the neonatal intensive care unit of 3-50%. In the study of Jena et al. half of the neonates (n=6) need an admission. In the study of Palatnik et al. only one infant of the 31 was admitted to the intensive care. Congenital malformations were only reported in the study of Shaheen and colleagues. They found a percentage of 0.4%.
Not all studies reported intrauterine fetal demise in the same manner. Shaheen et al. found a percentage of 5.9%, but this includes miscarriages. Borssén and colleagues reported one fetal death in 43 pregnancies, but it is unclear whether or not this occurred in the cirrhotic subgroup. The remaining six studies were comparable and reported percentages of 0-6.8% (mean 3.6%).
The study of Murthy et al. reported no fetal outcomes, except fetal death, due to no accessibility in the Nationwide Impatient Sample database they extracted their data from.
Preterm delivery was reported in eight studies, of which five were included in the meta-analysis. The summary odds ratio was 5.18 ( P 0.00001) (figure 3). All studies by itself were significant and preterm delivery was defined as birth 37 weeks in each study. Ranges of preterm delivery in absolute terms were 26-45% in the studies in the meta-analysis. Three studies, not included in the meta-analysis, Jena et al., Borssén et al. and Westbrook et al., reported percentages of respectively 42, 23 and 64%. The rates of cesarean section were mentioned by seven studies, again five studies of which were included in the meta-analysis. The summary odds ratio of cesarean section was 2.35 ( P 0.00001) (figure 4). Overall the studies differ greatly in percentages and therefore the I2-test for heterogeneity in the meta-analysis was 89%. The lowest percentage of cesarean section, reported by Britton et al., was 13% and the highest, reported by Rasheed et al., was 81%. Postpartum hemorrhage, the odds ratios were reported by four studies, had a summary odds ratio of 3.99 ( P 0.00001) (figure 5). Studies comply with statistical homogeneity, since the I2-test is 14%.
Two studies itself were significant and two were not. The studies included in the meta-analysis found ranges of postpartum hemorrhage in cirrhotic women between 5 and 16%. Another study, the one of Jena et al. reported a percentage of 17% in their study. Preeclampsia had a summary odds ratio of 3.59 ( P = 0.003) based on four studies (figure 6). Each study itself was significant. The percentages of preeclampsia in cirrhotic women of studies included in the meta-analysis ranges from 7 to 19%. Jena and colleagues found a percentage of 17%. Preeclampsia was the only hypertensive disorder of pregnancy that was reported in multiple studies. The syndrome of Hemolysis Elevated Liver Enzymes and Low Platelets (HELLP) was only reported by Rasheed et al. (4%) and pregnancy induced hypertension was only reported in the study of Murthy et al. (22%). No more data about hypertension disorders of pregnancy was described.
Five studies reported miscarriages and the range, including elective terminations, vary between 6 and 55% (mean 30%). Some studies included elective terminations in miscarriages, therefore it was not possible to make a distinction between miscarriages and elective terminations. Three studies reported placental diseases including placental abruption and placenta previa. Murthy et al., Shaheen et al. and Rasheed et al. found percentages of placental abruption of respectively 5.6, 7.1 and 10.1%. These findings were each significant compared with controls. Placenta previa was reported in two studies: Shaheen et al. and Murthy et al.. In the study of Shaheen and colleagues no case of placenta previa was occurred. Murthy et al. reported a percentage of 0.5%. Three studies reported puerperal infections varying from 1.6 to 2.1%. In one study this outcome was significant (Murthy et al.).
Maternal death was reported by six studies and ranged from 0 to 7.8%. Altogether there were 25 events of maternal death in 860 pregnancies (2.9%). To give some impression to these numbers, the maternal mortality rate (number of maternal death per 100.000 live births) in high-income regions in 2015 was 0.012% . See table 2 for detailed information about the causes of death in women with liver cirrhosis. Three of the five included studies with controls, did not report (odds ratios for) maternal death, therefore a meta-analysis could not be conducted.
Especially of our interest were hepatic decompensation and variceal hemorrhage among cirrhotic women. Variceal hemorrhage was reported in seven studies. In the study of Rasheed and colleagues 61/129 (47%) pregnancies were complicated by variceal hemorrhage. Other studies found lower percentages varying between 0 and 16%. A meta-analysis could not be conducted due to a lack of odds ratios. Hepatic decompensation or causes of hepatic decompensation was reported in six studies. Ranges of hepatic decompensations vary between 5 and 21%. Ascites was mentioned in five studies and ranges from 3 to 11% (mean 7%). Hepatic encephalopathy was also mentioned in five studies and ranges from 1 to 13% (mean 4%). Since Shaheen et al. reported more causes of hepatic decompensation, this higher than the rates of ascites and encephalopathy combined. See figure 7 for detailed information and causes per study. Splenic artery aneurysm was an outcome measure in the studies of Palatnik et al. and Rasheed et al. In both studies, no incidents occurred. None of the studies mentioned intrahepatic cholestasis of pregnancy.
The results of the present study demonstrate that women with liver cirrhosis have a remarkably higher risk on maternal, fetal and obstetrical complications during pregnancy and labor. They are five times more likely to deliver their infant before 37 weeks and four times more likely to deliver an infant small for gestational age. They have, as well, an almost four times higher risk of suffering from preeclampsia during pregnancy and a four times higher risk of postpartum hemorrhage during labor.
Limitations: The conclusions of the present study are based on retrospective cohort studies. These did not have the highest level of incidence.
Strength: We excluded case reports and case series to keep the evidence of high quality. Case reports and case series often describe the most complex cases and do not represent the whole cohort. Therefore, these studies have a high risk on biased patient population.
It should be noted that published reports may be biased toward complications, leading to an overestimation of the risk.