Prevalence of Budd-Chiari Syndrome during Pregnancy or Puerperium: A Systematic Review and Meta-Analysis

Women during pregnancy or puerperium are likely to develop Budd-Chiari syndrome (BCS). However, the reported prevalence of pregnancy-related BCS varied considerably among studies. Our study aims to systematically review this issue. Overall, 817 papers were initially identified via the PubMed, EMBASE, China National Knowledge Infrastructure, and Chinese Scientific and Technological Journal databases. Twenty of them were eligible. The prevalence of pregnancy-related BCS varied from 0% to 21.5%. The pooled prevalence was 6.8% (95% CI: 3.9–10.5%) in all BCS patients, 6.3% (95% CI: 3.8–9.4%) in primary BCS patients, and 13.1% (95% CI: 7.1–20.7%) in female BCS patients. Among them, one study was carried out in Africa with a prevalence of 10.6%; 14 studies in Asian countries with a pooled prevalence of 7.1% (95% CI: 3.1–12.6%); and 5 studies in European countries with a pooled prevalence of 5.0% (95% CI: 3.1–7.3%). The pooled prevalence was 6.7% (95% CI: 2.6–12.3%) in studies published before 2005 and 7.3% (95% CI: 4.2–12.5%) in those published after 2005. In conclusion, pregnancy is a relatively common risk factor for BCS, but there is a huge variation in the prevalence among studies. Physicians should be aware of pregnancy-related BCS.


Introduction
Budd-Chiari syndrome is characterized by hepatic venous outflow obstruction [1]. The main mechanism of obstruction is thrombosis of the hepatic veins (HV) or the terminal portion of the inferior vena cava (IVC) [2]. Recent studies have identified that many thrombophilic factors are associated with the development of BCS [3]. Common thrombophilic factors include inherited and acquired thrombophilia, such as antithrombin deficiency, protein C deficiency, protein S deficiency [4], heterozygous Factor V Leiden, prothrombin G20210A mutation [5], homozygous MTHFR mutation and hyperhomocysteinemia [6], and myeloproliferative neoplasms [7].
Pregnancy is a hypercoagulable state. The maternal hypercoagulable state is a physiological preparation for delivery; however, this hypercoagulability is associated with an increased risk of venous thromboembolism (VTE) [8][9][10]. The incidence of VTE in pregnant women, as derived from retrospective cohort studies, is estimated to be 5-12 events per 10,000 pregnancies antenatally (from conception to delivery), 7 to 10 times higher than the incidence in age-matched controls [10]. Clinical evidence also confirms that a pregnant woman's risk of VTE further increases immediately after the birth of the baby [11][12][13][14][15]. In a large population-based case control study from the Netherlands, a sixty-fold increase in the risk of VTE was detected in the puerperium compared with nonpregnant controls [16].

Search Strategy.
We searched the PubMed, EMBase, China National Knowledge Infrastructure (CNKI), and Chinese Scientific and Technological Journal databases (VIP) databases for studies that reported the prevalence of BCS during pregnancy or puerperium. The last search was performed on April 28, 2014. The search items were as follows: ("Budd-Chiari syndrome" or "hepatic venous outflow obstruction" or "hepatic venous outflow occlusion" or "membranous occlusion of inferior vena cava" or "membranous obstruction of inferior vena cava" or "hepatic vein thrombosis" or "hepatic vein occlusion" or "hepatic vein obstruction") and ("pregnancy" or "postpartum" or "puerperium" or "peripartum" or "perinatal" or "obstetric * "). The reference lists of the articles included were further hand-searched to identify any additional relevant studies.

Selection Criteria.
The following inclusion criteria were used to identify target studies. (i) All observational studies, including cohort and case-control studies, were included; case reports, reviews, comments, essays, and animal studies were excluded. (ii) All studies reported the proportion of pregnancy-related BCS. (iii) The number of participants in all included studies was greater than 10. Studies with fewer than 10 patients were excluded because a small sample size could maximize the impact of systematic error, yielding less than authentic results. (iv) There were no publication date or publication status restrictions. (v) There were no language restrictions.

Data Extraction.
All relevant studies were abstracted. A data extraction sheet was developed that included the authors, publication year, country, and affiliation where the study was carried out, study design, eligibility criteria, period of enrollment, number of patients with BCS, site of obstruction (IVC, HV, or combined obstruction), number of patients with pregnancy-related BCS, proportion of pregnancy-related BCS in total and female patients, and risk factors for the development of pregnancy-related BCS.

Quality Assessment.
The studies were considered to be of higher quality if they fulfilled all of the following predetermined criteria: (1) patients were consecutively admitted; (2) interval of enrolment and eligibility criteria were clearly recorded; (3) patients were diagnosed with primary BCS; and (4) the sites of BCS obstruction were clearly reported.

Statistical
Analysis. The prevalence of pregnancy-related BCS in each study was combined to yield a pooled prevalence with a 95% confidence interval (CI) for all studies. Data were pooled using a random-effects model to generate a more conservative estimate of the prevalence. Heterogeneity between studies was assessed using the 2 statistic and the 2 -test ( 2 > 50% or < 0.10 was considered to indicate statistically significant heterogeneity). Publication bias was measured using Egger's test ( < 0.05 represents statistically significant publication bias). To explore the cause of heterogeneity among studies, subgroup analyses were performed according to the continents, publication years, and obstruction sites. Statistical analyses were carried out using the StatsDirect statistical software version 2.8.0 (StatsDirect Ltd, Sale, Cheshire, UK).
One Japanese study was carried out between 1975 and 1989 and reported a prevalence of 0% [29].
One Iranian study was carried out between 1984 and 1994 and reported a prevalence of 18.2% [27].
Two French studies were carried out between 1994 and 2005 and reported a prevalence of 4.8% and 7.3% [18,30]. One English study was carried out between 1976 and 1990 and reported a prevalence of 7.7% [24]. One Netherlands study was carried out between 2003 and 2005 and reported a prevalence of 2.6% [23]. One study was conducted by the Europe Network for Vascular Disorders of the Liver (EN-Vie) project between 2003 and 2005, which was composed of 9 European countries (France, Spain, Italy, Great Britain, Germany, Belgium, Netherlands, Portugal, and Switzerland); the reported prevalence was 3.7% [3].
In only one study, the percentage of patients with HV obstruction alone was >70%, with a prevalence of 10.6% [31].

Discussion
Pregnancy is a hypercoagulable state with increased serum clotting factors, predominantly originating from the liver [36][37][38]. Thus, the concentrations of HVs and IVC might be higher in pregnant or puerperal women, resulting in their increased chance of developing thrombosis within the hepatic venous outflow. To our knowledge, this study is the first systematic review and meta-analysis to evaluate the prevalence of BCS during pregnancy or puerperium. An important finding of our study is that the pooled prevalence of pregnancy-related BCS is 6.8% in all BCS patients and 13.1% in female BCS patients, suggesting that pregnancy might be a relatively common etiology of BCS. It appears that the prevalence of pregnancy-related BCS (6.8%) is similar to that of other common risk factors, such as inherited antithrombin (2.3%), protein C (3.8%), and protein S (3.0%) deficiencies [4], prothrombin G20210A mutation (3.0%) [5], and factor V Leiden mutation (12.0%) in BCS [3]. Therefore, pregnancy must not be neglected as the etiology of BCS is being assessed. The heterogeneity remained statistically significant among studies. To explore the causes of heterogeneity, subgroup analyses were conducted according to the following factors: continents, publication years, and sites of obstruction. This behavior of subgroup stratification could be explained by the following reasons. (i) The distribution of the BCS etiologies might be different due to geographic regions. Indeed, an observational study has found that common thrombophilia is less frequently found in Chinese BCS patients than in western BCS patients [39]. This discrepancy may account for the lower prevalence of pregnancy-related BCS in Chinese studies than in European studies (1.8% versus 5.0%). (ii) The courses of puerperium may be variable among countries. In some Asian countries, prolonged rest postpartum is followed by late and slow mobilization after childbirth. According the report by Dilawari et al., in rural India, puerperal women often experience up to 30 or 40 days of confinement and fluid restriction [19]. Thus, the combination of increased clotting factors, lack of activity, and dehydration may constitute a condition conducive to venous thrombotic complications, potentially leading to a higher prevalence of pregnancyrelated BCS in Indian studies (13.1%). (iii) Our understanding of the relationship between pregnancy and BCS has improved over time. Thus, differences in the prevalence of pregnancyrelated BCS might be observed based on publication years.  [19,30]. Thus, performing a subgroup analysis stratified according to the sites of obstruction seemed to be worthwhile. Indeed, our study found a relatively higher prevalence of pregnancy-related BCS in studies with a larger proportion of patients with HV obstruction alone than in those with a larger proportion of patients with IVC obstruction. Still, it should be noted that this difference was slight, as the 95% CI values overlapped between the subgroups. Several strengths can be considered from this study. First, the prevalence of pregnancy-related BCS was systematically reviewed in this paper for the first time. The increasing available data lends support that improved outcome of BCS may result from deep understanding and advanced management of underlying etiology [1,2]. Based on this consideration, we conducted the study to further replenish the series of studies regarding the etiology of BCS [4][5][6][39][40][41][42][43]. Second, an exhaustive search strategy was used to maximize the likelihood of identifying all relevant literature. In our previous studies, we had systematically searched the literature regarding BCS in PubMed, EMBASE, and Cochrane library databases [44,45]. However, the data in Chinese BCS patients was extremely lacking in English-language literatures. To update and enrich the data, we further searched PubMed, EMBASE, and two comprehensive Chinese-language databases (i.e., VIP and CNKI databases) to systematically review the prevalence of pregnancy-related BCS. Third, a random-effects model was used to pool the data to provide a more conservative estimate of the prevalence.
Our study had several limitations. First, despite the extensive search strategy, the number of studies included in our meta-analysis was insufficient, thereby precluding the achievement of adequate statistical power. Second, the heterogeneity among the included studies was significant. Although we attempted to explore the causes of heterogeneity by performing subgroup analyses, the heterogeneity among the studies persisted. Therefore, the present findings should be explained with caution. Third, there were no sufficient data provided to perform a meta-analysis of the potential risk factors for pregnancy-related BCS. Of the 20 included studies, 19 had no information on this subject. Only Rautou et al. described any risk factor for pregnancy-related BCS other than underlying thrombophilia [30]. Future studies should explore this aspect more thoroughly.
In conclusion, this systematic review and meta-analysis demonstrated a global prevalence of pregnancy-related BCS of 6.8%. Accordingly, we recommended that the physicians should be highly vigilant against pregnancy-related BCS. Still, it should be noted that this prevalence varied considerably among studies. Despite the subgroup analyses performed according to the geographic regions, publication years, and obstruction sites of BCS, the sources of heterogeneity remained unclear. Further research should explore the risk factors for pregnancy-related BCS and their potential mechanisms.