Association between Liver Fluke Infection and Hepatobiliary Pathological Changes: A Systematic Review and Meta-Analysis

Objective To provide information about the role of liver fluke infection as a risk factor for hepatobiliary pathological changes and promote awareness among the people living in endemic areas, a systematic review and meta-analysis based on published studies was conducted to examine the association between liver fluke infection and hepatobiliary pathological changes. Methods Relevant original literature was searched in multiple literature databases, including PubMed, Cochrane, Clinical Evidence, Trip Database, Clinical Trials, Current Controlled Trials, Web of Science, the China National Knowledge Infrastructure (CNKI) database, and the Wanfang academic journal full-text database. Studies were selected based on strict screening with inclusion and exclusion criteria. Tests of heterogeneity, sensitivity and publication bias were performed with the Review Manager software, version 5.3, and meta-regression analyses were performed with the Stata software, version 11.0 (Stata Corporation, College Station, TX, USA). Pooled risk ratios (RRs) and odds ratios (ORs) with their 95% confidence intervals (95% CIs) were calculated and used to evaluate the risk of hepatobiliary pathological changes resulting from liver fluke infection. Linear trend analyses were conducted to determine the dose-response relationship using IBM SPSS Statistics 20.0. Result A total of 36 studies were included in the meta-analysis. Significant associations were found between liver fluke infection and cholangitis or cholecystitis (RR: 7.80, P<0.001; OR: 15.98, P<0.001), cholelithiasis (RR: 2.42, P = 0.03; OR: 4.96, P = 0.03), hepatocellular carcinoma (OR: 4.69, P<0.001) and cholangiocarcinoma (RR: 10.43, P<0.001; OR: 4.37, P<0.001). In addition, heavier infection was significantly associated with higher incidence of hepatobiliary pathological changes (P<0.05). However, cirrhosis was not significantly associated with liver fluke infection (RR: 3.50, P = 0.06; OR: 5.79, P = 0.08). The statistical heterogeneity was significant, no significant difference was observed in the sensitivity analysis, and no publication bias was found. Conclusion The meta-analysis found that liver fluke infection was significantly associated with cholangitis, cholecystitis, cholelithiasis, hepatocellular carcinoma and cholangiocarcinoma and that more severe infection was associated with higher incidence. However, the association between liver fluke infection and cirrhosis was not significant.

(S1 Table). Relevant literature that reported an association between liver fluke infection and hepatobiliary pathological changes was identified and screened from databases, including PubMed, Cochrane, Clinical Evidence, Trip Database, Clinical Trials, Current Controlled Trials, Web of Science, the China National Knowledge Infrastructure (CNKI) database, and the Wanfang academic journal full-text database. The following Medical Subject Heading (MeSH) terms were used individually and in combination in the search: "Fasciola hepatica," "Clonorchis sinensis," "Opisthorchis," "Case-Control Studies," "Cohort Studies," "Cross-Sectional Studies," "Hepatobiliary pathological changes," "Cholangitis," "Cholecystitis," "Cholelithiasis," "Cirrhosis," "Hepatocellular Carcinoma" and "Cholangiocarcinoma." The literature search was not limited by language or geographical region. The references in all of the retrieved articles were reviewed to search for additional relevant studies.

Criteria for inclusion and exclusion
The inclusion criteria were as follows: (1) published full text available; (2) an observational study (a cohort study or a case-control study); (3) sufficient data reported to calculate the odds ratio (OR) with its 95% confidence interval (CI); and (4) the diagnosis of liver fluke infection based on (a) microscopy of liver fluke eggs in stool samples; (b) detection of worm-specific antibodies in serum samples or worm-specific antigens in serum or stool samples; (c) skin test with an intradermal injection of diluted crude live fluke antigen in veronal-buffered saline [12]; (d) observation of liver fluke eggs or parasites from bile, gallstones or intramural stones; (e) detection of diffuse dilatation of intrahepatic bile ducts in abdominal computed tomography (CT) or cholangiography; (f) results of molecular techniques such as polymerase chain reaction (PCR); or (g) history of liver fluke infection that could be confirmed by medical records. Studies were excluded if they were (1) comments, congresses, abstracts, reviews, or editorials without raw data or control subjects or (2) studies that included fewer than 10 participants.

Data extraction
The following information was independently extracted from all of the included studies: the name of the first author, publication year, country or geographical area, liver fluke species, diagnostic methods for liver fluke infection, sample size, the number of the exposure or outcome of interest for case-control or cohort studies, respectively, and the quality of each study.

Quality assessment
The quality of all of the included studies was assessed using The Newcastle-Ottawa Scale (NOS) (S2 Table). This scale involves a "star system" in which a study is judged on three broad perspectives: the selection of the study groups, the comparability of the groups and the ascertainment of either the exposure or outcome of interest for case-control or cohort studies, respectively. Studies having more stars are considered to be of higher quality.

Statistical analysis
Statistical heterogeneity among studies was calculated using the χ 2 test, P values, and I 2 statistics [13]. A random-effects model was used to estimate the overall relative risk (RR) or overall odds ratio (OR) when heterogeneity was significant (Q: P<0.1, or I 2 >50%); if the reverse was true, a fixed-effects model was used (Q: P>0.1, or I 2 >50%). The overall RRs and ORs and their 95% CIs were estimated (P<0.05 was considered significant), and forest plots were generated for each disease associated with liver fluke infection. A sensitivity analysis was conducted, and publication bias was evaluated using funnel plots [14]. Meta-regression analyses were generated to explore possible sources of heterogeneity (adjusted R 2 >50% and P<0.05 were considered significant.) [15,16], such as geographical area, decade of publication, liver fluke species, diagnostic methods and study sample size. Linear trend analyses were performed to determine the relationship between infection intensity and incidences of hepatobiliary pathological changes. Risk estimates, tests of heterogeneity, sensitivity calculations and publication bias analyses were performed using the Review Manager software, version 5.3; meta-regression analysis was performed using the Stata software, version 11.0 (Stata Corporation, College Station, TX, USA); and linear trend analyses were performed using IBM SPSS Statistics 20.0.
The risk of hepatobiliary pathological changes associated with liver fluke infection Cholangitis or cholecystitis. Several studies have reported a close association between liver fluke infection and cholangitis or cholecystitis [34,52]. The overall RR with its 95% CI was extracted from the 3 included cohort studies [17,18,41], and the overall OR with its 95% CI was extracted from the 4 included case-control studies [22,23,43,44]. The statistical heterogeneities of both the cohort studies and case-control studies were significant (I 2 = 95%, P<0.001 and I 2 = 55%, P = 0.08, respectively); hence, the overall RR for the cohort studies and the overall OR for the case-control studies were estimated using a random-effects model. The analysis of the cohort studies and case-control studies revealed that liver fluke infection was significantly associated with cholangitis and cholecystitis. Cholelithiasis. Liver fluke infection has been investigated as a risk factor for cholelithiasis [21]. In total, 8 cohort studies [17, 19-21, 32-34, 41] and 5 case-control studies [22,23,33,35,43] were used to perform the respective meta-analyses using a random-effects model (I 2 = 97%, P<0.001 and I 2 = 75%, P<0.001, respectively). The analyses yielded an RR of 2.42 (95% CI: 1.07-5.46, P = 0.03) and an OR of 4.96 (95% CI: 1.19-20.56, P = 0.03), indicating that infection with liver flukes is a risk factor for cholelithiasis and that the association is significant (Figs 2 and 3).
Cirrhosis. In total, 3 cohort studies [17,33,42] and 3 case-control studies [22,23,45] on cirrhosis and liver fluke infection were identified and used to perform meta-analyses. A random-effects model was applied to the analyses (I 2 = 98%, P<0.001 and I 2 = 74%, P = 0.02, respectively). However, the result did not reveal a significant association between liver fluke infection and cirrhosis. For cohort studies, the overall RR of cirrhosis between infection with liver fluke and without infection was 3.50 (95% CI: 0.95-12.89, P = 0.06); for case-control studies, the overall OR of exposure to liver fluke infection between the case group and control group was 5.79 (95% CI: 0.83-40.28, P = 0.08) (Figs 2 and 3).
Cholangiocarcinoma. The association between cholangiocarcinoma and liver fluke infection has been identified in articles over the last several decades [54,55]. In our meta-analysis, 3 cohort studies [17,33,41] and 19 case-control studies [23, 27-31, 36-40, 43, 44, 46-51] were included. A fixed-effects model was used in the analysis of the cohort studies (I 2 = 41%, P = 0.19), and a random-effects model was used (I 2 = 77%, P<0.001) in the analysis of the case-control studies. The overall RR for the association between liver fluke infection and cholangiocarcinoma was 10.43 (95% CI: 2.90-37.47, P<0.001), and the association was significant. The overall OR for the association of cholangiocarcinoma with liver fluke infection was 4.37  (95% CI: 2.84-6.72, P<0.001), which indicated that liver fluke infection was a risk factor for cholangiocarcinoma (Figs 3 and 4).

Sensitivity analysis
A sensitivity analysis was performed to identify whether the results of the meta-analysis were significantly affected by the exclusion of any individual study or the study with the highest quality or the greatest weight in the results. There was no significant impact observed in the overall ORs and 95% CIs.

Publication bias
Funnel plots of the studies in the meta-analysis were generated to evaluate publication bias (Figs 5 and 6). For both cohort studies and case-control studies, the plots approximately resembled a symmetrical funnel, and no publication bias was found.

Meta-regression analyses
Meta-regression analyses were generated to explore possible sources of heterogeneity. Our meta-regression showed that geographical area, decade of publication, liver fluke species and diagnostic method did not contribute significantly to the heterogeneity (Adjusted R 2 <50% or P>0.05) for either cohort studies or case-control studies. In contrast, for cohort studies only, the study sample size did have a contribution (Adjusted R 2 = 73.13%, P<0.001). The results of the meta-regression analyses are shown in Table 3.

Linear trend analyses of the dose-response relationship
In total, 2 cohort studies [19,41] and 3 case-control studies [25,43,50] with intensity groups (3) of liver fluke infection were included in the linear trend analysis to examine the relationship between infection intensity and incidences of hepatobiliary pathological changes ( Table 4). The results revealed a significant trend toward increasing incidences of hepatobiliary pathological changes with increasing intensity of liver fluke infection (P<0.05).

Discussion
Several published studies [52,56,57] have reported an association between liver fluke infection and various hepatobiliary pathological changes, including cholangitis, cholecystitis, cholelithiasis, cirrhosis, hepatocellular carcinoma and cholangiocarcinoma. However, these published studies have not identified consistent findings for the risk of these hepatobiliary pathological changes and liver fluke infection. In this systematic review and meta-analysis of cohort studies and case-control studies, significant associations were found between liver fluke infection and The observed statistical heterogeneity was significant, although sensitivity analysis did not alter the overall RR, overall OR, or their 95% CIs, and there was no evidence of publication bias.
A random-effects model was used in all of the analyses (except the analysis of cohort studies in cholangiocarcinoma) because significant heterogeneity was observed. Meta-regression analyses showed that the study sample size contributed significantly to the heterogeneity of the cohort studies (Adjusted R 2 = 73.13%, P<0.001); as interpreted, the study sample size could explain 73.13% of the heterogeneity. In contrast, geographical area, decade of publication, liver fluke infection and diagnostic methods did not contribute to the heterogeneity. This result is most likely related to the limited information included in the studies, such as study design, the stages of pathological changes, and other demographic characteristics. In this systematic review and meta-analysis, we found that liver fluke infection was significantly associated with increased risk of cholangitis and cholecystitis. The liver fluke secretes  Association between Liver Fluke Infection and Hepatobiliary Diseases metabolites while invading, some of which are highly immunogenic, stimulating a strong humoral immune response that can be measured in the serum and bile [58]. Another study revealed that Opisthorchis antigens were observed along with an inflammatory cell infiltration, and the antigens were not only in the fluke itself but also in the biliary epithelium and surrounding tissue, which might then activate host immune responses [59].
Our study confirmed that liver fluke infection was significantly associated with cholelithiasis. The cause of clonorchiasis was most likely related to changes in the concentration of bilirubin, cholesterol, phospholipids, bile acid and the core of the gallstone formed from parasite debris or epithelial cells from the biliary ducts [60]. The metaplasia of bile duct epithelial cells into goblet cells and mucin secretion occurs in clonorchiasis and promotes a favorable environment for secondary bacterial infection [61]. A positive association was found between hepatocellular carcinoma and liver fluke infection. The mechanism of hepatocellular carcinoma in patients with liver fluke infection remains unknown. One possible mechanism is that epithelial ulceration and hyperplasia induced by the suckers of liver flukes induce stimulation of metabolites from the worms [62]. Secondary bacterial infection gives rise to periductal adenomatous hyperplasia and mucin secretion, which may result in hepatocellular carcinoma [62]. Another possible mechanism is that severin, a liver fluke excretory/secretory product, plays a key role in inhibiting apoptosis in human hepatocellular carcinoma cell lines and exacerbates hepatocellular carcinoma [63].
Our systematic review and meta-analysis confirm a significant relationship between infection with liver flukes and cholangiocarcinoma. The mechanisms by which liver flukes contribute to cholangiocarcinoma are multi-factorial [56] and include mechanical damage caused by the activities and movements of the worms, chronic inflammation, and the effects of parasite secretions [57].
This study confirms not only the relationship between liver fluke infection and various hepatobiliary pathological changes, such as cholangitis, cholecystitis, cholelithiasis, hepatocellular carcinoma and cholangiocarcinoma, but also the relationship between intensity of liver fluke infection and incidences of the hepatobiliary pathological changes. We found a significant trend toward increasing incidences of hepatobiliary pathological changes with increasing intensity of liver fluke infection. The ordinal intensity of liver fluke infection was analyzed by linear trend analyses instead of meta-analyses due to the limited sample size and the different ordinal scales used among the included studies. Additionally, information was too limited to generate analyses of the association between the intensity of liver fluke infection and the severity of pathological changes. However, in our included studies [41,43], most cases of cholangiocarcinoma were identified from heavily infected patients, which supports the hypothesis that high pathogenicity relates to heavy parasite infection. The pathogenesis is due to the mechanical irritation by the flukes and some toxic substances produced by them [64].
Although published studies provided evidence to support the hypothesis that liver fluke infection is associated with cirrhosis [45], our analysis failed to provide sufficient evidence for this association. This inconsistency likely occurred because the studies that identified a relationship between cirrhosis and liver fluke were limited to animals, such as cattle, goats and sheep [65,66]. In addition, most cirrhosis is associated with Fasciola hepatica infection [66,67], which was not included in our analysis because of the absence of eligible studies.
Several limitations of our study deserve mention. First, non-English, non-Chinese studies were not included in our meta-analyses, which might have an impact on the overall results. Second, because of the limited number of studies involved and limited information on the studies, our study was not powered to perform subgroup analyses, which might provide reasons for the significant heterogeneity as well.

Conclusion
In conclusion, our systematic review and meta-analysis found that liver fluke infection is associated with an increased risk of cholangitis, cholecystitis, cholelithiasis, hepatocellular carcinoma and cholangiocarcinoma, and more severe infection is associated with higher incidence. However, no significant evidence was found for the association between liver fluke infection and cirrhosis.
Supporting Information S1 Table. PRISMA checklist for this meta-analysis. (DOC) S2 Table. The Newcastle-Ottawa Scale (NOS) for quality assessment. (DOC)