Prevalence of Splanchnic Vein Thrombosis in Pancreatitis: A Systematic Review and Meta-Analysis of Observational Studies

Splanchnic vein thrombosis (SVT) may be negatively associated with the prognosis of pancreatitis. We performed a systematic review and meta-analysis of literatures to explore the prevalence of SVT in pancreatitis. All observational studies regarding the prevalence of SVT in pancreatitis were identified via PubMed and EMBASE databases. The prevalence of SVT was pooled in the total of patients with pancreatitis. And it was also pooled in the subgroup analyses according to the stage and causes of pancreatitis, location of SVT, and regions where the studies were performed. After the review of 714 studies, 44 studies fulfilled the inclusion criteria. Meta-analyses showed a pooled prevalence of SVT of 13.6% in pancreatitis. According to the stage of pancreatitis, the pooled prevalence of SVT was 16.6% and 11.6% in patients with acute and chronic pancreatitis, respectively. According to the causes of pancreatitis, the pooled prevalence of SVT was 12.2% and 14.6% in patients with hereditary and autoimmune pancreatitis. According to the location of SVT, the pooled prevalence of portal vein, splenic vein, and mesenteric vein thrombosis was 6.2%, 11.2%, and 2.7% in pancreatitis. The prevalence of SVT in pancreatitis was 16.9%, 11.5%, and 8.5% in Europe, America, and Asia, respectively.


Introduction
Splanchnic vein thrombosis (SVT) is one such vascular complication of pancreatitis. SVT involves the portal vein (PV), splenic vein (SplV), and mesenteric vein (MV) with the occurrence in combination or separately [1]. The consequence of pancreatitis-induced SVT may generate a localized form of portal hypertension and then splenoportal or gastroepiploic systems burden following localized venous hypertension and lead to gastric, oesophageal, or colonic varices. It may also cause the liver failure, bowel ischemia, and gastrointestinal bleeding [2]. Patients suffer from great pain and potentially lethal threaten. The limited literature cannot acquire an exact prevalence of SVT in pancreatitis. The aim of the present systematic review and meta-analysis is to obtain the prevalence of SVT in pancreatitis by resolving the following three questions. (1) What is the prevalence of SVT in pancreatitis, including PVT, SplVT, and MVT? (2) What is the prevalence of SVT in different types of pancreatitis?
(3) What is the prevalence of SVT in pancreatitis in different regions? This study is conducted according to the guidelines for the reporting of meta-analysis of observational studies, which were published by the Meta-analysis of Observational Studies in Epidemiology Group in 2000 [3].

Methods
As a systematic review and meta-analysis were planned, we reviewed standard guidelines to conduct meta-analysis studies according to a protocol determined before the study, including study objectives, prespecified eligibility criteria, and methods of statistical analysis.
(2) All cohort and case-control studies were eligible, regardless of the retrospective or prospective nature of the study; case reports were excluded.
(3) Reviews, comments, or letters on the relationship of pancreatitis and SVT were excluded.
(5) There was no publication date or publication status restrictions.
(6) There were no language restrictions. (7) The number of participants in any included study was beyond 10.

Search Strategy.
The MEDLINE and EMBASE databases were searched using a search strategy from their inception to July 2014. Search items combined keywords and medical subject heading terms (MeSH) were listed as follows: ("pancreatitis" (MeSH Terms) or "pancreatitis" (All Fields)) and (("portal vein" (MeSH Terms) or ("portal" (All Fields) and "vein" (All Fields)) or "portal vein" (All Fields)) and ("thrombosis" (MeSH Terms) or "thrombosis" (All Fields))) or (("splenic vein" (MeSH Terms) or ("splenic" (All Fields) and "vein" (All Fields)) or "splenic vein" (All Fields)) and ("thrombosis" (MeSH Terms) or "thrombosis" (All Fields))) or (("mesenteric veins" (MeSH Terms) or ("mesenteric" (All Fields) and "veins" (All Fields)) or "mesenteric veins" (All Fields) or ("mesenteric" (All Fields) and "vein" (All Fields)) or "mesenteric vein" (All Fields)) and ("thrombosis" (MeSH Terms) or "thrombosis" (All Fields))). The last search was performed on July 29, 2014. The reference lists of the included articles were further hand-searched to identify any additional relevant studies. When the same data were found in more than one publication, only the studies with more complete data and more extensive interval of enrolment were included in the meta-analysis. Full-texts were found by three investigators (Wenda Xu, Xingshun Qi, and Chunping Su).

Data Extraction.
Using a predefined protocol, two investigators (Wenda Xu and Xingshun Qi) independently reviewed the titles and abstracts of all references to identify studies for inclusion in the analysis. Dealing with disagreement between the two reviewers, a consensus was achieved through discussion among all of the reviewers. A schematic diagram depicting reference flow is shown through the systematic review process. Additionally, a data extraction sheet was generated that included authors, publication year, study design, country where the study was conducted, period of enrolment, inclusion and exclusion criteria, type of diseases (AP, CP, HP, or AIP), total sample size, demographic data (age and gender), number of patients with PVT, SplV, or MVT, and proportion of patients with PVT, SplV, or MVT, respectively.

Evaluation of Study Quality.
Quality assessment of studies was carried out independently by two reviewers (Wenda Xu and Xingshun Qi). Discrepancies of interpretation and comprehension were resolved by consensus. The higher quality studies should fulfill the following predetermined criteria.
(1) Country where the study was conducted, interval of enrolment, inclusion and exclusion criteria, and participant characteristics (age, gender) were clearly recorded.

Data Synthesis and Statistical
Analysis. The proportion of pancreatitis patients with SVT in each study was combined to give a pooled prevalence of SVT for all studies. With this method, the pooled prevalence of PVT, SplV, and MVT was calculated. Furthermore, according to the type of pancreatitis and continents (Europe, America, and Asia), the pooled prevalence of SVT, PVT, SplV, and MVT was also created. The number and crude proportion of participants with SVT recorded by each study were used to pool the overall proportion, using the DerSimonian-Laird random-effects method. Between-study heterogeneity was assessed by using the 2 index ( 2 > 50% was considered having substantial heterogeneity) and the Chi-squared test ( < 0.01 was considered representing significant statistical heterogeneity) [4]. Individualized random effects meta-analyses were performed to estimate percentages and 95% confidence intervals (CIs) for all endpoints queried. Analyses were conducted using StatsDirect statistical software version 2.7.8 (StatsDirect Ltd, Sale, Cheshire, UK).

Description of the Included Studies.
In the initial search strategy, a total of 947 studies were selected. Among them, 714 studies were retrieved by removing duplicate research. Two additional unique references were found through reference lists of an original article and a review, respectively [5,6].

Quality of the Included Studies. The involved countries
of all studies could be found. Interval of enrolment was unavailable in ten of the 44 studies. Twenty-one studies had no eligibility criteria; eight of the remaining 23 studies had detailed inclusion and exclusion criteria. Demographic data were completely recorded in nineteen studies. The pancreatitis diagnostic criteria were not elaborate in nine studies only with abstracts and one study with full-test.

Prevalence of SVT in HP and AIP.
Only two studies reported the prevalence of SVT in HP, ranging from 7% to 19% (Figure 6(a)). The pooled prevalence was 12.2% (95% CI: 3.0%-26.5%) and there was no statistical heterogeneity between the two studies ( = 0.0914). Two studies reported the prevalence of PVT in HP, ranging from 2.8% to 11.5% (Figure 6(b)). The pooled prevalence was 6.6% (95% CI: 0.7%-17.8%) and there was no statistical heterogeneity between the two studies ( = 0.1056). Two studies reported the prevalence of SplVT in HP, ranging from 4.2% to 7.7% ( Figure 6(c)). The pooled prevalence was 5.8% (95% CI: 2.1%-11.2%) and there was no statistical heterogeneity between the two studies ( = 0.437). There was no data on the prevalence of MVT in HP.

Discussion
The previous published meta-analysis that assessed the prevalence of SVT in patients with pancreatitis almost focused on AP, CP, HP, or AIP alone without systematic search. The prevalence varied from 1% to 24% in previous studies according to the type of pancreatitis and imaging technique used (US, EUS, CT, MRI, or ERCP) [30,40,55]. However, our study was remarkably different from the previous ones, as follows. (1) In the previous studies, most of them were singlecenter studies with a limited enrollment period and target population. By comparison, our meta-analysis included all studies conducted from 1958 to 2014 and 10560 patients with pancreatitis. Furthermore, we made a distinction on the source of patients according to the continents. (2) As we have known, SVT involves PVT, SplVT, and MVT. In contrast to the previous studies, we have paid more attention on the prevalence of PVT, SplVT, and MVT in pancreatitis. (3) Our research reported the prevalence of SVT on the basis of the different types of pancreatitis including AP, CP, HP, and AIP. However, there was no similar record in the previous studies.
Pancreatitis is associated with a variety of vascular complications including SVT. Clinically, SVT in pancreatitis is becoming common with the advancement of imaging technique. In the current systematic review and meta-analysis, we demonstrated that 13.6% of pancreatitis had SVT, 6.2% had PVT, 11.2% had SplVT, and 2.7% had MVT. The prevalence of SVT in pancreatitis showed some regional differences. We found that the prevalence of SVT in pancreatitis in Europe reached 16.9%, which was the highest among the three continents. Moreover, the prevalence of PVT, SplVT, and MVT in pancreatitis in Europe was higher than that in America or in Asia, respectively. We can find from the involved studies that the results came from nine countries of Europe, which were more than those from four of Asia and two of America. Compared with Asia, the prevalence of SVT in pancreatitis was 11.5% higher than that of 8.5% in Asia, but the prevalence of SplVT was a little lower in America.
Previous studies that reported the vascular complications of pancreatitis showed different prevalence of SVT in pancreatitis regardless of the type of pancreatitis. This meta-analysis demonstrated that the prevalence of SVT in AP was 16.6% higher than previous studies with a reported incidence of 1-2% [1]. The reason was that previous studies took PVT, SplVT, and MVT inclusion of SVT as numerator.
The controversy also exists about the prevalence of SplVT in CP. The old series reported that the prevalence of SplVT in patients with CP varied between 2.5% and 25% [6,51,56,57]. Our results showed 12.8% CP with SplVT, which was similar to that from a recent report by Butler et al. [58]. However, there was no detailed description on the prevalence of SVT in HP and AIP. Our results showed that the prevalence of SVT was 12.2% and 14.6% in HP and AIP, respectively. All of these could enrich the whole research on the prevalence of SVT in pancreatitis.

Limitation
Our study has several limitations. Firstly, the heterogeneity of available data from various years' studies was significant. Depending on advanced imaging technique, the majority of asymptomatic patients with SVT could be involved in research groups. Therefore, only the random-effects model was applied in our meta-analysis to generate a more conservative estimate of the proportion. Secondly, there are different results between Western and Asian countries. Besides the difference in the population race, lifestyle, and diagnostic level, the number of countries involved in the research is different. America included two countries and four countries were from Asia. Thirdly, there is no related research on the prevalence of SVT in HP and AIP. The final results may be not so typical.

Conclusion
This systematic review and meta-analysis attempt to quantify the incidence of SVT in pancreatitis according to the different forms of pancreatitis and regional distribution. Further studies are needed to analyze the relationship between natural history, clinical significance, long-term outcomes, and the prevalence of SVT in pancreatitis. The rate of SVT associated gastrointestinal bleeding and the security and reasonability of anticoagulation therapy on thrombosis are all needed to develop complete, large, multicentre, and collaborative studies.