Association between vitamin C intake and the risk of pancreatic cancer: a meta-analysis of observational studies

Quantification of the association between the intake of vitamin C and risk of pancreatic cancer is still conflicting. We therefore conducted a meta-analysis to assess the association between them. Pertinent studies were identified by a search of PubMed and Web of Knowledge throughSeptember of 2014. A random effects model was used to combine the data for analysis. Sensitivity analysis and publication bias were conducted. Data from 17 studies including 4827 pancreatic cancer cases were used in this meta-analysis. Pooled results suggested that highest vitamin C intake amount versus lowest amount was significantlyassociated with reduced the risk of pancreatic cancer [summary relative risk (RR) = 0.705, 95% CI = 0.612–0.811, I2 = 42.3%]. The associations were also significant both in Caucasian [summary RR = 0.741, 95% CI = 0.626–0.876], Asian [summary RR = 0.455, 95% CI = 0.275–0.754] and Mixed population [summary RR = 0.677, 95% CI = 0.508–0.901]. No publication bias was found. Our analysis suggested that the higher intake of vitamin C might reduce the risk of pancreatic cancer.

'lifestyle, ' 'vitamin C, ''vitamins' or 'ascorbic acid' . Two investigators searched articles and reviewed all the retrieved studies independently. Disagreements between the two investigators were resolved by consensus with a third reviewer. Study Selection. For inclusion, studies had to fulfill the following criteria: (1) have a prospective or case-control or retrospective cohort studies; (2) vitamin C intake was the independent variable of interest; (3) the dependent variable of interest was pancreatic cancer; (4) relative risk (RR) or odds ratio (OR) or hazardratio (HR) with a 95% confidence interval (CI) was provided (we presented all results with RR for simplicity). If data were replicated in more than one study, we included the study with the largest number of cases. Accordingly, the following exclusion criteria were also used: (1) reviews; (2) repeated or overlapped publications. Data extraction. Two researchers independently extracted the following data from each study that met the criteria for inclusion: the first author's last name, year of publication, geographic locations, study design, sample source, the age range of study participants, duration of follow-up, the number of cases and participants, and RR (95% CI) for vitamin C intake and pancreatic cancer risk. From each study, we extracted the RR that reflected the greatest degree of control for potential confounders. If there was disagreement between the two investigators about eligibility of the data, it was resolved by consensus with a third reviewer.

Quality assessment.
To determine the quality score of included studies, two reviewersindependently performed the quality assessment by using the Newcastle-Ottawa Scale 28 , which is a validated scale for non-randomized studiesin meta-analyses 29 . The Newcastle-Ottawa Scale is anine-point scale that allocates points based on the selection process of cohorts (0-4 points), the comparability of cohorts (0-2 points), and the identification of the exposure and the outcomes of study participants (0-3 points). We assigned scores of 0-3, 4-6, and 7-9 for low, moderate, and high quality of studies, respectively. Statistical analysis. The pooled measure was calculated as the inverse variance-weighted mean of the logarithm of RR with 95% CI, to assess the association between vitamin C intake and pancreatic cancer risk. Random-effects model was used to combine study-specific RR (95% CI), which considers both within-study and between-study variation 30 . The I 2 was used to assess heterogeneity, and I 2 values of 0, 25, 50 and 75% represent no, low, moderate and high heterogeneity 31 , respectively. Meta-regression with restricted maximum likelihood estimation was performed to assess the potentially important covariates that might exert substantial impact on between-study heterogeneity 32 . Publication bias was evaluated using Egger's regression asymmetry test 33 . The Duval and Tweedie nonparametric trim-and-fill method was performed to further assess the potential publication bias 34 . A study of influence analysis 35 was conducted to describe how robust the pooled estimator was to removal of individual studies. An individual study was suspected of excessive influence if the point estimate of its omitted analysis lay outside the 95% CI of the combined analysis. All statistical analyses were conducted with STATA version 11.0 (StataCorp LP, College Station, Texas, USA). Two-tailed p-value ≤ 0.05 was accepted as statistically significant.

Results
Search results and study characteristics. The search strategy identified 158 articles from PubMed and 243 from the Web of Knowledge, and 28 articles were reviewed in full after reviewing the title/ abstract. In total, 17 articles 11-27 (4 cohort studies and 13 case-control studies) involving 4827 pancreatic cancer cases were used in this meta-analysis after reviewed in full articles. The detailed steps of our literature search are shown in Fig. 1. Four studies were conducted in the United States, 2 in the Canada, 9 in the Europe, 1 in the Japan and 1 in the Australia. The characteristics of these studies are presented in Table 1. The quality of studies was generally good, with results of study quality assessment yielded a score of 6 or above for all included studies, with an average score of 7.2.
High versus low analyses. Eight of the studies included in our analysis reported an inverse association of vitamin C intake with the risk of pancreatic cancer, while no significant association was reported in 9 studies. Our pooled results suggested that the highest vitamin C intake amount compared to the lowest amount was significantly associated with the risk of pancreatic cancer [summary RR = 0.705, 95% CI = 0.612-0.811, I 2 = 42.3%] (Fig. 2).
When the studies were stratified bystudy design, the associations were also found in the case-control studies [ Sources of heterogeneity and meta-regression. As shown in the pooled results, moderate heterogeneity (I 2 = 42.3%, P heterogeneity = 0.034) was found in the analysis. In order to explore the moderate to high between-study heterogeneity founded in several analysis, univariate meta-regression with the covariates of publication year, ethnicity, study design (case-control or prospective), number of cases and source of controls was performed. No significant findings were found in the above-mentioned analysis. Considering the adjustment of individual studies is heterogeneous, we then provide the original unadjusted relative risks and pooled them together to derive an effect size estimate. The pooled RR was 0.771 (95% CI = 0.685-0.868) for vitamin C intake and pancreatic cancer risk. Low heterogeneity was found (I 2 = 12.1%, P heterogeneity = 0.331).

Influence analysis and publication bias.
Influence analysis showed that no individual study had excessive influence on the association of vitamin C intake and pancreatic cancer risk. The trim-and-fill funnels (Fig. 3) and Egger's test (P = 0.414) showed no evidence of significant publication bias between vitamin C intake and pancreatic cancer risk.

Discussion
Finding from this meta-analysis suggested that the higher intake of vitamin C could reduce the risk of pancreatic cancer. The associations were also found in subgroups of Caucasian, Asian and Mixed population for vitamin C intake and pancreatic cancer risk.
Vitamin C is one of the most common antioxidants in fruits and vegetables, and it may exert chemopreventive effects 36 . It has generally been acknowledged that vitamin C protects cells from oxidative DNA damage, thereby blocking carcinogenesis 37 . A second mechanism for antioxidants is their effect on the inflammatory process, and chronic inflammation may play a role pancreatic carcinogenesis 38 . The results of our analysis had verified this hypothesis.
Munafo and Flint reported that between-study heterogeneityis common in meta-analyses 39 . Exploring potential sources of between-study heterogeneity is therefore an essential component of meta-analysis. We found a moderate degree of heterogeneity (I 2 = 42.3%, P heterogeneity = 0.034) in our pooled results. This might havearisen from publication year, study design, geographic location, and sources of controls or number of cases. Thus, we used meta-regression to explore the causes of heterogeneity for covariates. However, no covariate having a significant impact on between-studyheterogeneity was found among those mentioned above. We then performed subgroup analyses by the type of study design (prospective or case-control studies) and ethnicityto explore the sourceof heterogeneity. However, between-study heterogeneity persisted insome of the subgroups, suggesting the presence of other unknown confounding factors. Considering the adjustment of individual studies is heterogeneous, we then provide the original unadjusted relative risks and pooled them together to derive an effect size estimate. The pooled RR was 0.771 (95% CI = 0.685-0.868) for vitamin C intake and pancreatic cancer risk. Low heterogeneity was found (I 2 = 12.1%, P heterogeneity = 0.331). This may be because the adjustment of individualstudies for confounding factors is different.
As a meta-analysis of published studies, our findings showed some advantages. First, this is the first comprehensivemeta-analysis of vitamin C intake and pancreatic cancer risk based on highestamountversus lowest amount analysis. Second, large number of cases and participants was included, allowing   Table 2. Summary risk estimates of the association between vitamin C intake and pancreatic cancer risk.
a much greater possibility of reaching reasonable conclusions between vitamin C intake and pancreatic cancer risk. Third, no significant publication bias was found, indicating that our results are stable. There were also some limitations in this meta-analysis. First, a meta-analysis of observationalstudies is susceptible to potential bias inherent in the original studies, especially for case-control studies. Overstated association may be expected from the case-control studies because of recall or selection bias. Although case-control studycan allow a recall or selection bias, case-control studyis an important method in etiology research. In order to find whether the study design is a key contributor to the between-study heterogeneity, univariate meta-regression with study design (case-controlor cohort) was performed. No significant finding (P = 0.39) was found in the above-mentioned analysis. However, significant associations were found both in case-controls studies and in cohort studies. More studies with prospective design are wanted in the future studies while only 4 studies included in this meta-analysis were prospective design. Second, measurement errors are important in the assessment of dietary intake, which can lead to overestimation of the range of intake and underestimation of the magnitude of the relationship between dietary intake and cancer risk 40,41 . Third, in our meta-analysis, we used 'highest versus lowest vitamin C intake' . Although some references use quartiles and some use quintiles to partition vitamin C intake considering that differing definitions can be a source of heterogeneity, Egger's test (P = 0.414) showed no evidence of significant publication bias was found suggesting that our results are stable. Fourth, there seems to be a big gap in published material between 1993 and 2002 in our meta-analysis. For this reason, we have searched the databasecarefully and by hand-searching the reference lists of the retrieved articles again, and did not find any related articles. Fifth, there appears to be a large variability in the baseline riskdue to the limitation of published material used, for whichwe cannot change. Finally, between-study heterogeneity was found in the pooled analysis, and the between-study heterogeneity was successfully explained by themeta-regression.
In summary, results from this meta-analysis suggested that the higher intake of vitamin C might reduce the risk of pancreatic cancer.