Intakes of citrus fruit and risk of esophageal cancer: A meta-analysis.

Esophageal cancer (EC) is the eighth most common cancer and the sixth most frequent cause of cancer death in the whole world. Many studies have investigated the association between citrus fruit intake and the risk of EC, but the results are inconsistent and not analyzed by category. We aimed to perform a meta-analysis of studies to evaluate the incidence between citrus fruit consumption and subtypes of esophageal cancer and derive a more precise estimation.Through searches of PubMed, OVID, and Web of Science we updated 1988 systematic review up to April 2016. Based on an inclusion and exclusion criteria, conventional meta-analysis according to DerSimonian and Laird method was used for the pooling of the results. Random-effect models were used to calculate subgroups.Twenty-five English articles (20 case-control studies and 5 cohort studies) comprising totally 5730 patients of esophageal cancer would be suitable for use in this study. The result indicated the inverse associations between intakes of citrus fruit and EC (relative risk [RR] = 0.65, 95% confidence interval [CI] 0.56-0.75, I  = 51.1%, P = .001), Esophageal squamous cell carcinoma (ESCC) (RR = 0.59, 95% CI 0.47-0.76, I  = 60.7%, P = .002), no significant relationship between citrus fruit and esophageal adenocarcinoma (EAC) (RR = 0.86, 95% CI 0.74-1.01, I  = 0.0%, P = .598).This meta-analysis indicates that intakes of citrus fruit significantly reduce the risk of ESCC and is no obvious relationship with EAC. Further studies about constituents in citrus fruit and its mechanism are warranted.

Introduction

Esophageal cancer (EC) is the eighth most common cancer and the sixth most frequent cause of cancer death in the whole world.[ These cases account for >400,000 deaths worldwide and most having occurred in the southern and eastern Africa and eastern Asia.[ In some parts of China, the number of deaths from esophageal cancer is 10 to 100 times that of the United States. The 5-year survival outcomes of patients with esophageal cancer only provided between 15% and 25%.[ It is still unclear today with the high mortality and morbidity. Many previous studies have suggested risk factors for esophageal cancer, including cigarette smoking, alcohol drinking, and body mass index,[ whereas the consumption of vegetables and fruits play an essential role in reducing the incidence rate of esophageal cancer.[

Citrus fruits, mainly including oranges, mandarins, limes, lemons, grapefruits, and citrons, belong to the genus Citrus of the family Rutaceae. Citrus fruits contribute to cancer prevention, because they are being loaded with vitamin C and other antioxidant, antimutagenic, and antiproliferative constituents.[ Available experiments showed that vitamin C destroys cancer cells, suppresses tumor growth, and is cytotoxic to cancer,[ which might reduced the risk of gastric cancer, breast cancer, lung tumorigenesis, colonic tumorigenesis, hepatocarcinogenesis, bladder cancer, and pancreatic cancer.[ Since 1980s, many epidemiologic studies indicated a potential association between citrus fruit consumption and risk of esophageal cancer, but there are still conflicting results.

Esophageal cancer is divided into esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC) by histological classification. There is a clear difference in 2 subtypes, including incidence, mortality, risk factors, clinical features, genetic susceptibility, and pathogenesis.[ In the past decades, compared with the rising rates of EAC, the ESCC incidence trends have been decreasing or stabilizing among men.[ With respect to esophageal cancer, recent experiment found different effects on risk of EAC and ESCC, but in a recently published meta-analysis took them into account without detailed statistical analysis and comprehensive original articles.[ We therefore conducted a meta-analysis of cohort and case-control studies to evaluate the incidence between citrus fruit and subtypes of esophageal cancer.

Materials and methods

Search strategy

A systematic literature search of the related English articles was carried out using MEDLINE database (PubMed) and journals@Ovid Full Text database and BIOSIS previews databases (OVID) by 2 independent researchers (WZ and LL). The following Medical Subject Headings (MeSH) search terms were used: (Citrus [MeSH] OR orange [MeSH] OR food OR diet OR fruit) AND (esophagus OR esophageal) AND (adenocarcinoma OR cancer OR neoplasm OR carcinoma OR tumor). We limited our search to reports on human subjects in the English language. We searched databases for articles published between 1983 (Because Brown et al published their result on the first results of EC in 1988) and April 2016. The reference lists of all retrieved articles in English were also searched to find relevant publications.

Study selection

Inclusion criteria

Two authors (WZ and LL) independently scanned all the relevant studies retrieved according to the prespecified criteria as the following: were original articles; used a case-control or cohort study design; presented the odds ratios (ORs)/relative risk (RR) with 95% confidence intervals (CIs).

Exclusion criteria

Studies were excluded with the content as the following: studies reported the same population, we chose the most complete one; combined cancer that included upper aerodigestive tract cancer; mechanistic studies, non–peer-reviewed articles, abstracts, and editorials; used other diet simultaneously that influence the result.

Data extraction

Two of the authors (WZ and LL) independently extracted the data from the primary sources. Any discrepancies between the 2 were resolved by consensus. Abstracted information included study design, institution database, date published, etc.

The following predefined variables were recorded electronically: first author, publication year, study design, the number of cases and controls or cohort size, type of citrus fruit and consumption categories, dietary assessments, the OR, or RR with the corresponding 95% CI for the highest versus lowest level of intake, the risk estimates were extracted with the greatest number of adjustments for potential confounders.

Statistical analysis

The statistical analysis was based on the Cochrane System Manual. Continuous outcome measures were used standardized mean difference and corresponding 95% CIs and results of categorical data were used odds ratio (OR) and corresponding 95% CIs. The statistical inference of the combined effect was tested by Z test and results are considered to be significant for P < .05. We estimated between-study heterogeneity in each meta-analysis by the Q test and I2 statistics. A fixed effect model was performed using Mantel-Haenszel fixed effects model[ if I2 < 50% and P > .10, which means no heterogeneity; otherwise a random effects model as described by DerSimonian and Laird[ was used. We performed separate meta-analyses for EAC and ESCC.

Sensitivity analysis was performed to evaluate the effect of individual study and the stability of our results by omitting each study or some studies and summarizing the remaining. Funnel plot and Egger test[ were performed to assess publication bias. Significant publication bias was indicated when P values were <.05. Most of the statistical analysis was conducted using spreadsheet software (Microsoft Excel 2007, Microsoft, Redmond, WA) and statistical software (STATA, version 14.0, StataCorp, College Station, TX). A P < .05 was considered statistically significant.

Results

Studies selected and general demographics

The search strategy generated 4814 citations, of which 84 were considered of potential value. Among these 84 articles, 67 articles were further selected out by following various reasons: 48 articles did not evaluate the association of interest, 9 articles were combined cancer, 6 were published without RR/OR or 95% CI, and 4 articles studied on the same population. Eight additional articles were included from the reference review.

Thus, 25 articles, including 20 case-control studies and 5 cohort studies, were incorporated in this meta-analysis[ (Fig. 1). The Newcastle-Ottawa scale was used to evaluate randomized studies in this meta-analysis from selection, comparability, and outcomes, including 22 articles of high quality (final score >6). Two authors (Z.W. and L.L.) scored for each articles independently and negotiated to get final score.

Figure 1

Flow diagram. CI = confidence interval, RR = relative risk.

Studies were published between 1983 and 2015, which consist of 2456 ESCC (range 47–395), 1284 EAC (range 67–282), and 1990 EC (range 53–1246). Of these 25 studies, 10 were from Europe, 6 from Asia, 6 from the United States, and 3 were from the South America (Table 1 ). The Newcastle-Ottawa Quality Assessment Scale (NOS) scores of 25 clinical trials range from 5 to 9, with an average of approximately 7. The median score was 6.75 for case-control studies and 8 for cohort studies (Table 1 ).

Table 1

Characteristics of studies of citrus fruit intake and the risk of esophageal cancer.

Quantitative data synthesis

ESCC group

Totally, 10 case-control studies and 3 cohort studies were included in the pooled analysis between citrus fruit consumption and ESCC risk (Table 2). The pooled RR was 0.59 (95% CI, 0.47–0.76) with significant heterogeneity (I2 = 60.7%, P = .002). The analyses revealed a significant association for citrus fruit intake with a reduced risk of ESCC. There was no evidence of publication bias; the P values were .059 for Begg test and .064 for Egger test.

Table 1 (Continued)

Characteristics of studies of citrus fruit intake and the risk of esophageal cancer.

EAC group

Five case-control studies and 3 cohort studies were included in the pooled analysis between citrus fruit consumption and EAC risk. The pooled RR was 0.86 (95% CI, 0.74–1.01) with no heterogeneity (I2 = 0.0%, P = .598) (Table 3). The Egger test (P = .071) and the Begg test (P = .063) showed that there was no evidence of publication bias.

Table 2

Forest plots investigating the association of citrus fruit intake and the risk of esophageal squamous cell carcinoma (ESCC).

EC group

Twenty-one case-control studies and 7 cohort studies were included in the pooled analysis between citrus fruit consumption and EC risk. The pooled RR was 0.65 (95% CI, 0.56–0.75) with low heterogeneity (I2 = 51.1%, P = .001) (Table 4). The significant inverse associations between citrus fruit intake and the risk of EC were observed. The Egger test (P = .018) showed that there was no evidence of publication bias.

Table 3

Forest plots investigating the association of citrus fruit intake and the risk of esophageal adenocarcinoma (EAC).

Subgroup analysis

We performed the subgroup analysis by study design, population, number of cases, alcohol, Food Frequency Questionnaire (FFQ), study quality, smoking and cigarette, body mass index (BMI), red meat, physical activity, and fruit and vegetable to find the source of the heterogeneity. As shown in Figure 2 , the pooled ORs of EAC and ESCC for prospective cohort studies, population-based case-control studies and hospital-based case-control studies were 0.78 (95% CI: 0.56, 1.10), 0.89 (95% CI: 0.71–1.11), 0.90 (95% CI: 0.61–1.23), 0.66 (95% CI: 0.49, 0.88), 0.82 (95% CI: 0.62, 1.09), and 0.50 (95% CI: 0.33, 0.75), respectively. Moreover, significant inverse associations between citrus fruit intake and the risk of ESCC were observed in cohort study (OR = 0.66, 95% CI: 0.49–0.88) and hospital-based study (OR = 0.82, 95% CI: 0.33,0.75), but not in population-based study (OR = 0.82, 95% CI: 0.62,1.09), in >7 scores study (OR = 0.56, 95% CI: 0.43,0.72), but not in <7 scores study (OR = 0.70, 95% CI: 0.36, 1.37), in FFQ study (OR = 0.57, 95% CI: 0.44, 0.75), but not in without FFQ study (OR = 0.76, 95% CI: 0.51, 1.13), in nondrinkers (OR = 0.59, 95% CI: 0.45, 0.76), not in drinkers (OR = 0.55, 95% CI: 0.28, 1.08). In addition, a significant inverse association between citrus fruit intake and the risk of EAC were observed in adjusted red meat study (OR = 0.63, 95% CI: 0.41, 0.96), but not in nonadjusted red meat study (OR = 0.91, 95% CI: 0.77, 1.07).

Figure 2

Subgroup analyses of citrus intake and risk of esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC), sensitivity analysis, and meta-regression analysis. BMI = body mass index.

Sensitivity analysis and publication bias

However, no variable were found in univariate and multivariate meta-regression analysis. In addition, we performed the sensitivity analysis by omitting 1 study at a time and further indicated that our results were stable. Egger test showed no evidence of significant publication bias for the meta-analysis on the association between citrus fruit intake and risk of ESCC, EAC, and EC, as provided in Figures 3 to 5. The Begg funnel plots did not reveal apparent asymmetry. Therefore, there were no obvious bias in these studies.

Figure 2 (Continued)

Subgroup analyses of citrus intake and risk of esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC), sensitivity analysis, and meta-regression analysis. BMI = body mass index.

Figure 4

Begg funnel plot of studies evaluating the association between citrus fruit intake and esophageal squamous cell carcinoma (ESCC) risks with 95% confidence limits. OR = odds ratio, SE = standard error.

Begg funnel plot of studies evaluating the association between citrus fruit intake and esophageal adenocarcinoma (EAC) risks with 95% confidence limits. OR = odds ratio, SE = standard error.

Begg funnel plot of studies evaluating the association between citrus fruit intake and esophageal cancer (EC) risks with 95% confidence limits. OR = odds ratio, SE = standard error.

Discussion

This meta-analysis, including overall 25 studies, suggested that intakes of citrus fruit reduced the risk of total ESCC, as well as ESCC and EAC combined, especially in quality scores (NOS score >6) ≥7 studies. The inverse association with citrus fruit was esophageal cancer, possibly because of most of clinical trials were from Asia.[ Consumption of citrus fruit was not associated with EAC, although the estimate was based on few studies.

Many studies have investigated the mechanism that high intake of citrus fruit may reduce the risk of esophageal cancer.[ Citrus fruits have such a protective effect for cancer, because they are rich in vitamin C and secondary metabolites, such as flavonoids, alkaloids, coumarins, limonoids, carotenoids, phenol acids, and essential oils.[ These components show the effect of antioxidant, antimutagenic, and antiproliferative by inhibiting oxidation, protecting DNA from damage and stimulating apoptosis to cancer line.[ Recent study found that the antitumor effects of citrus fruit were derived from inhibitory effect on epithelial-to-mesenchymal transition and interfering with the canonical transforming growth factor-β1-drosophila mothers against decapentaplegic protein-Snail/Slug axis.[ But some studies indicate that too much antioxidant supplementation increased overall mortality from cancers, which may explain the result of the nonsignificant inverse associations between EAC and citrus fruit. Based on the completely distinctive in etiological and pathological characteristics,[ we can come to conclusion that different subtypes of esophageal cancer have been effected distinctively by citrus fruit.

In view of the increasing incidence of EC and the difficulty of operation, prevention of esophageal cancer is particularly important. The existing methods to prevent esophageal cancer mainly include higher intake of vitamin C, antioxidant, anthocyanidin, and dietary fiber.[ The factors that increased the risk of esophageal cancer included alcohol intake, red meat, and diabetes mellitus.[ Citrus fruit are rich in vitamin C, vitamin E, antioxidant, and flavonoids that make people more clearly aware of how to intake above substances in daily life.

As for confounding, only 2 studies included in the meta-analysis did not control for smoking and alcohol drinking,[ the major risk factors for esophageal cancer. However, the risk estimates for the relation of citrus fruit intake with the risk of cancer from such studies did not materially differ from studies that were properly adjusted. Because dietary intake is basically constant, desirable eating habits seem particularly important. Studies have linked Mediterranean diet with a reduced risk of cancer,[ which may guide people a reasonable diet. From a public health perspective, increasing consumption of citrus fruit was very effective for esophageal cancer prevention. In addition, present researches indicated that citrus fruit benefits on endothelial function, protection against the risk of renal stone formation, and reduction in risk of cardiovascular disease and stroke.[

Several limitations exist in this meta-analysis described as follows. First, the original studies use the different diet assessment methods, which cannot compare the result with each other in some degree and unavoidably cause measurement errors. The measurement errors occurred when some original studies used a nonvalidated FFQ to evaluate the consumption of citrus fruit and esophageal cancer, which may lead to decrease the relativity of dietary habit with cancer risk. In fact, the relationship between citrus fruit and esophageal cancer is stronger than the result. Second, publication bias was observed across studies, because some original studies with negative result were more difficult to publish and all studies included are in English. Some studies, furthermore, did not provide the raw data, which is partly responsible for bias. In this meta-analysis, little publication bias was found by Egger weighted regression method about EAC and consumption of citrus fruit because there are not enough EAC cases included. In addition, a significant heterogeneity was observed in ESCC, which influence reliability of summary RR estimates. But we had no way to unite the duration of follow-up, types and quality of original articles, study populations, confounders, and so on. Moreover, most original studies only provided frequency of consumption, but not provided quantitative assessment of citrus fruit intake, with no information on the portion size. Therefore, we could not analyze the dose-risk relation between citrus fruit and esophageal cancers, given the lack of information in the categorization in some studies, and the heterogeneity in the categorization across some others. Finally, the inadequate adjustment for confounders in observational studies was another limitation, which included less intakes of smoking, alcohol drinking, higher intake of vegetable, less BMI, and better physical activity. The confounders may make us overestimate the relationship between citrus fruit and esophageal cancer. However, most of articles in this meta-analysis restricted these confounders.

This meta-analysis shows that citrus fruit intake is associated inversely with the risk of EC and ESCC, no significant association with EAC. However, the association should be considered with some caution because of the measurement errors, confounders, and heterogeneity. Further studies are warranted to find which constituents in citrus fruit prevent esophageal cancer and its mechanism.

Author contributions

Conceptualization: W. Zhao.

Data curation: W. Zhao.

Formal analysis: L. Liu, S. Xu, W. Zhao.

Funding acquisition: L. Liu, S. Xu.

Methodology: W. Zhao.

Project administration: W. Zhao.

Resources: W. Zhao.

Software: W. Zhao.

Supervision: W. Zhao.

Validation: W. Zhao.

Visualization: W. Zhao.

Writing – original draft: W. Zhao.

Writing – review & editing: W. Zhao.

Table 4

Forest plots investigating the association of citrus fruit intake and the risk of Esophageal cancer (EC).