Association Between Pancreatitis and Subsequent Risk of Pancreatic Cancer : a Systematic Review of Epidemiological Studies

Pancreatic cancer (PC) has become one of the most serious diseases threatening human health and life all over the world. GLOBOCAN 2008 reported that there were about 277 thousand new PC cases (including 144 thousand males and 133 thousand females) and 266 thousand PC deaths (including 138 thousand males and 127 thousand females) worldwide (Ferlay et al., 2010). Although PC is less common than cancer of the lung, breast, colorectum, stomach and liver, etc., only ranked thirteenth (accounted for only 2.2%) of all new cancer cases; its mortality rate is quite high, ranked eighth (accounted for 3.5%) of all cancer deaths. Pancreatic ductal adenocarcinoma, the most common type of PC, was the fourth leading cause of cancer death in the United States with a median survival of <6 month and a 5-year survival rate less that 5% (Hezel et al., 2006). It is believed that risk factors of PC could


Introduction
Pancreatic cancer (PC) has become one of the most serious diseases threatening human health and life all over the world.GLOBOCAN 2008 reported that there were about 277 thousand new PC cases (including 144 thousand males and 133 thousand females) and 266 thousand PC deaths (including 138 thousand males and 127 thousand females) worldwide (Ferlay et al., 2010).Although PC is less common than cancer of the lung, breast, colorectum, stomach and liver, etc., only ranked thirteenth (accounted for only 2.2%) of all new cancer cases; its mortality rate is quite high, ranked eighth (accounted for 3.5%) of all

Association Between Pancreatitis and Subsequent Risk of Pancreatic Cancer: a Systematic Review of Epidemiological Studies
Gui-Xian Tong & , Qing-Qing Geng & , Jing Chai, Jing Cheng, Peng-Lai Chen, Han Liang, Xing-Rong Shen, De-Bin Wang* be divided into 2 categories including environmental factors (e.g., cigarette smoking, alcohol drinking, coffee intake, diabetic mellitus, cholelithiasis, H. pylor infection) and genetic factors (e.g., family history of PC and high expression of BRCA2).Of these, cigarette smoking has been regarded as the most important risk factor.Smokers have 2-fold to 3-fold increased risk of PC compared to non-smokers with cigarette smoking accounting for approximately a quarter of PC incidence (Howe et al., 1991;Silverman et al., 1994;Fuchs et al., 1996;Lowenfels et al., 2006;Ko et al., 2007;Bao et al., 2009;Talamini et al., 2010;Nitsche et al., 2011).Hereditary PC accounts for 5%-10% of all cases.Individuals with family history of PC, especially have first-degree relative(s) with PC, were consistently been found to be associated with added risk, and risk raised as the number of affected first-degree relative increased (Falk et al., 1998;Brentnall et al., 1999;Klein et al., 2004;Ko et al., 2007;Permuth-Wey et al., 2009).
The associations between medical conditions and the risk of PC have been explored extensively for the past several decades.Of these, pancreatitis (including its acute and chronic forms) is one of the most frequently studied diseases.Although the pathogenesis mechanisms underlying the role of pancreatitis in PC etiology are still unclear, the molecular pathway for this association had been put forward by several hypotheses (Sakorafas et al., 2012;Gukovsky et al., 2013;Pinho et al., 2014;Kolodecik et al., 2014).Just like other benign diseases are associated with an increased cancer risk in the target organs (e.g., hepatitis and liver cancer, gastritis and gastric cancer), increased cell turnover and defective DNA repair in pancreatitis cases could lead to the occurrence of PC.Rosty and colleagues found a significant minority of pancreatic intraepithelial lesions (PanINs) arising in chronic pancreatitis cases showed loss of p16 expression (Rosty et al., 2003), a common precursor of cancer (Lowenfels et al., 2006).A meta-analysis of 15 studies documented that K-ras mutations, which play an important role in the evolution of PC, had been detected in subjects with chronic pancreatitis (Löhr et al., 2005).Past decades witnessed a variety of epidemiological studies on this issue, those including case-control, retrospective cohort and prospective cohort studies of patients affected by pancreatitis (Lowenfels et al., 1993;Malka et al., 2002).A meta-analysis published in 2010 documented a 5.1fold risk of developing PC in patients with unspecified pancreatitis, 13.3-fold in patients with chronic pancreatitis and 69.9-fold risk for hereditary pancreatitis (Raimondi et al., 2010).In recent years, several new and large sample size studies conducted in Australia, China, Canada, Europe and United States, Denmark and Taiwan have been published.Thus, an updated comprehensive assessment may provide more accurate and detailed information.This study aims at summarizing published epidemiological evidence and producing an updated pooled evidence of relationship between pancreatitis and PC.

Data sources and search strategy
We utilized two approaches to locate as many relevant papers as possible.First, we searched the literatures in Medline and Embase available by February 5th, 2014 using the following search terms "(pancreatitis) AND (cancer OR neoplasm* OR tumour OR tumor OR carcinoma OR malignanc* OR adenocarcinoma) AND (pancreatic OR pancreas)", where * represents wildcard characters.Second, we reviewed the bibliographies of relevant review papers for additional articles.This process was conducted iteratively until no new papers were identified.

Inclusion criteria
The inclusion criteria of were paper: 1) written in English; 2) cohort or case-control study investigating the relationships between PC and medical conditions that include pancreatitis; and 3) provided adequate original data for the recalculation of odds ratio (OR) or relative risk (RR) of PC.

Data extraction and analysis
Descriptive data about the included studies were extracted from the papers identified using a data-extracting form, comprising first author, year of publication, country of study, recruitment period, type of pancreatitis, study design, sample size, source of controls and attained age of subjects (the age of subjects when they were studied).All data were extracted independently by two researchers and discrepancies were solved by consensus.Stata 11.0 (StataCorp, College Station, TX, USA) was used to perform statistical analyses and examine publication bias.Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated using the random effects model.

Quality assessment
Newcastle-Ottawa Scale (NOS) (Wells et al., 2011) was used to assess the methodological quality of included studies.The tool provides a comprehensive score system with 8 items for both case-control and cohort studies.Quality items of case-control studies include adequate definition of patient cases (0-1 point), representativeness of patients cases (0-1 point), selection of controls (0-1 point), definition of controls (0-1 point); comparison controlled for important factor or additional factor (0-2 point), ascertainment of exposure (0-1 point), same method of ascertainment for participants (0-1 point) and nonresponse rate (0-1 point); while quality items of cohort studies include representativeness of the exposed cohort (0-1 point), selection of the non-exposed cohort (0-1 point), ascertainment of exposure (0-1 point), outcome of interest not presented at start of study (0-1 point), comparison based on the design or analysis (0-2 point), assessment of outcome (0-1 point), long enough follow-up for outcomes to occur (0-1 point) and adequate evaluation of follow-up of cohorts (0-1 point).Total score was calculated by adding up the points awarded to each item.Only studies scored 6 or higher were considered to be of high methodological quality.

Studies included
A total of 11563 articles were retrieved from Medline and Embase, of which 10507 articles were excluded on the basis of title and abstract.Of the remaining 51 articles, 43 were excluded after more detailed evaluation via full texts including 17 articles with irrelevant contents, 6 review articles, 13 studies without control group and 7 papers lacking original data for further analysis.After combining the 4 studies from reference lists, 17 studies finally met the inclusion criteria and included in this study (Figure 1).

Descriptive analyses
As shown in Table 1, the 17 articles documented 3 cohort and 14 case-control studies containing 14667 PC cases and 17587 pancreatitis cases from countries of Australia, Canada, China, Denmark, Finland, Greece, Italy, Netherlands, Poland and United States etc.The sample sizes of studies ranged from 218 to 779430 and the mean/median attained age of subjects ranged from DOI:http://dx.doi.org/10.7314/APJCP.2014.15.12.5029 Association Between Pancreatitis and Subsequent Risk of Pancreatic Cancer: a Systematic Review 54.5 to 66.1.Nine out of the 17 studies did not specify the type of pancreatitis; four studies included both acute and chronic pancreatitis and the remaining four, limited to chronic pancreatitis.Source of controls varied across studies including community population (n=10), hospital patients (n=3), visiting relatives of hospital patients (n=1), hospital patients and community population (n=1), hospital patients and visiting relatives of hospital patients (n=1), and visiting relatives of hospital patients and community population (n=1).

Quality of studies and Sensitivity analysis
As shown in Table 2, the total quality scores of studies ranged from 5 to 9 with a mean score of 7.12.Quality of cohort studies was relatively better compared with case-control studies.In casecontrol studies, only 2 studies ascertained blinding of diagnosis of pancreatitis, and only 1 studies made it clear that the difference in non-response rate was not statistically significant (p >0.05) in both groups.After excluding 3 studies scored less than 6, the pooled OR increased slightly from 7.05 (95% CI: 6.42-7.75) to 7.13 (95% CI: 6.48-7.85).

Publication bias
A funnel plot was produced to detect the presence of publication bias and which revealed a relatively moderate degree of asymmetry.Begg's regression asymmetry test produced a p value of 0.171, indicating a low probability of publication bias.

Table 2. Quality of Included Studies Based on the Newcastle-Ottawa Scale
Compared with acute and chronic pancreatitis, certain rare types of pancreatitis seemed to be more closely associated with PC.Lowenfels and colleagues' cohort study of 412 hereditary pancreatitis patients in the United States showed that the risk of PC was approximately 50 to 60 times greater than background population (Lowenfels et al., 2000).A cohort study of 418 hereditary pancreatitis cases from 14 European countries revealed that the cumulative risk of PC was 44.0% (95% CI: 8.0%-80.0%)at 70 years from symptom onset, and the standardized incidence ratio reached 67.0% (95% CI: 50.0%-82.0%)(Howes et al., 2004).A recent published study of France cohort documented that the relative risk of PC for the whole population, men, and women were 87.0 (95% CI: 42.0-113.0),69.0 (95% CI: 25.0-150.0),and 142.0 (95% CI: 38.0-225.0),respectively (Rebours et al., 2008).Chari and colleagues' study reported that subjects with tropical pancreatitis, a form of pancreatitis found primarily in southern Indian and in parts of sub-Saharan Africa, appeared to have a significantly increased risk of PC when compared with the background pancreatic cancer rate with an RR of 100.0 (95% CI: 37.0-218.0)(Chari et al., 1994).
Evidences regarding the associations between PC and other medical conditions were summarized in previous papers.Ben and colleagues performed a meta-analysis of cohort studies investigating the relationship between diabetes and PC and found that diabetes was associated with an increased risk of PC with a pooled RR of 1.94 (95% CI: 1.66-2.27)(Ben et al., 2011).Li's meta-analysis suggested that chronic hepatitis B virus infection was linked with increased risk of PC; the pooled OR of PC for overall, case-control and cohort studies were 1.40 (95% CI: 1.14-1.73),1.43 (95% CI: 1.06-1.94)and 1.31 (95% CI: 1.00-1.72)respectively (Li et al., 2013).Recently published meta-analysis also found that cholecystectomy and gastrectomy were associated with 23% and 54% excess risk of PC respectively (Lin et al., 2012;Gong et al., 2012).While Olson et al' s pooled analysis of 10 case-control studies revealed that allergy was associated with reduced risk of PC with the pooled ORs of 0.79 (95% CI: 0.62-1.00)for all types of allergy, 0.74 (95% CI: 0.56-0.96)for hay fever and 0.62 (95% CI: 0.41-0.94)for allergy to animals (Olson et al., 2013).
Although there is a strong link between pancreatitis and PC, screening is not recommended for subjects with pancreatitis for the long time lag between the diagnosis of benign diseases and the occurrence of cancer.Future studies should focus on strategies using pancreatitis as an alarm sign and finding additional indications including medical conditions (e.g., diabetes mellitus, cholelithiasis and obesity), risk behaviors (cigarette smoking, alcohol drinking and coffee intake) and family history (e.g., first-degree relatives with PC) etc; and perform more sophisticated analysis of the risk of multi-indications or build multi-variable models (e.g., score systems, regression models).Future researchers should also have the vision and courage in conducting long-term, large sample size prospective cohort studies with adequate attention being paid to the study quality.

Figure 1 .
Figure 1.Flow Diagram of Study Selection Process

Figure
Figure 2. Forest Plot of ORs between Pancreatitis and Pancreatic Cancer

Figure
Figure 3. Forest Plot of ORs between Pancreatitis and Pancreatic Cancer Stratified by type of Pancreatitis

First
Figure 4. Forest Plot of ORs between Pancreatitis and Pancreatic Cancer Stratified by Duration since Diagnosis of Pancreatitis