Dietary fruit, vegetable, fat, and red and processed meat intakes and Barrett’s esophagus risk: a systematic review and meta-analysis

The relationships between dietary fruit, vegetable, fat, and red and processed meat intakes and Barrett’s esophagus (BE) risk remain inconclusive. We conducted a systematic review and meta-analysis to summarize the available evidence on these issues. PubMed, EMBASE and the Cochrane Library were searched for studies published from inception through October 2015. A total of eight studies were included in this analysis. Fruit intake was not associated with BE risk (OR = 0.65, 95% CI = 0.37–1.13), but vegetable intake was strongly associated with BE risk (OR = 0.45, 95% CI = 0.29–0.71). Saturated fat, red meat and processed meat intakes were not associated with BE risk with OR = 1.25 (95% CI = 0.82–1.91), OR = 0.85 (95% CI = 0.61–1.17) and OR = 1.03 (95% CI = 0.73–1.46), respectively. Dietary vegetable not fruits intake may be associated with decreased BE risk. Fat and red and processed meat intakes may not contribute to an increased BE risk. Well-designed, large prospective studies with better established dose-response relationships are needed to further validate these issues.

1. Studies unrelated to our topics were excluded. 2. Case-control studies and cohort studies were included. 3. The patients with BE were all diagnosed via endoscopy and biopsy. The histological feature, which was not consistent with the diagnostic gold standard, was excluded. 4. Data that were incomplete or could not be combined were excluded. 5. Narrative reviews, systematic reviews and meta-analyses were excluded. 6. Cases in which only comments, case reports, editorials, letters and the abstract could be obtained were excluded. 7. The most recent studies, the most samples, and the quality studies that included reports with the same patients were selected. 8. Red and processed meats in this study included beef, corned beef, beefburgers, veal, bacon, bacon rashers, luncheon meat, lamb, mutton, ham, sausage, salami, hot dogs, souse meat, smoked meat, salted meat and barbecued meat. 9. Poultry, chicken, fish and other white meats were excluded.
Search strategy. We searched PubMed, EMBASE and the Cochrane Library for studies of fruits, vegetables, fat, and red and processed meats and BE published from inception through October 2015. The following search terms were used: "fruit", "fruits", "vegetable", "vegetables", "fat", "meat", "red meat", "processed meat", "beef ", "corned beef ", "beefburgers", "veal", "bacon", "bacon rashers", "pork", "lamb", "mutton", "luncheon meat", "ham", "sausage", "salami", "hot dogs", "sauce meat", "smoked meat" "salted meat", "barbecued meat" in combination with "Barrett's esophagus/oesophagus", "gastrointestinal tract", "metaplasia", "precancerous/neoplasia/canceration/cancerization", "carcinogenesis/tumorigenesis", "esophageal cancer/carcinoma/adenomas/adenocarcinoma". Reference lists in the included studies were also searched manually to identify additional literature. The two sets of keywords were combined individually, and the eligibility criterion was independently judged by two authors (Zhanwei Zhao and Zhongshu Pu). In case of disagreement between the two authors, the third author (Qingchuan Zhao) made a consensus decision. The language of all studies was limited to English only. The studies were also limited to only those in humans. Barrett's esophagus (BE). BE is defined as the presence of columnar-type mucosa in the esophagus on endoscopy and pathology. Subsequently, BE is also recognized as a specialized intestinal metaplasia by histology of the esophagus in most studies 1,5,8 . Barrett's columnar epithelium is considered to be a marker for severe reflux and a precursor to adenocarcinoma of the esophagus 5,19 .
Study quality. Study quality in this meta-analysis was assessed using the Newcastle-Ottawa Scale (NOS) 20 . The range of NOS was 0-10 stars, which was judged in three parts, including the elucidation of exposure or the outcomes of interest for case-control or cohort studies, the selection of the study populations and the comparability of the populations. Two authors (Zhanwei Zhao and Zhongshu Pu) independently assessed the quality of the studies, and a consensus decision was made regarding any discrepancies in interpretations by the third author (Qingchuan Zhao).
Data Extraction. A data extraction sheet was set up to enter data from each study, including the first author, year of publication, country, study type, case/control, study population, method of dietary assessment, type of dietary exposure measured, meat exposure categories, NOS score and controlled variables ( Table 1). The controlled variables were specifically listed in Table 2.
Statistical analysis. The data were collected and extracted using SPSS 17.0 (Chicago, Illinois, USA). RevMan 5.3 (The Cochrane Collaboration, Oxford, UK) and STATA, version 12.1 (STATA Corporation, College Station, TX) software were used for data synthesis and analysis. Heterogeneity was detected using the I 2 statistic (25%, 50% and 75% meant low, moderate and high heterogeneity, respectively; in this analysis, I 2 < 50% was considered as having low heterogeneity among studies, while I 2 > 50% was considered as having high heterogeneity) 21 . A fixed-effects model was used if there was no heterogeneity among the studies, and a random-effects model was used if there was heterogeneity among the studies. Publication bias was tested using Begg's test.

Results
Literature selection, study characteristics and quality scores. Figure 1 shows the flowchart of the search strategy for selecting eligible studies. A total of 109 studies were initially identified for this meta-analysis. Forty-two studies were excluded for duplication, and 67 studies were selected for further consideration. Among them, 55 studies were excluded after reviewing the titles and abstracts. Finally, 8 studies met the eligibility criteria after excluding the different studies (n = 4) and missing data (n = 1), and including one study from reference review.
Scientific RepoRts | 6:27334 | DOI: 10.1038/srep27334 The 8 selected studies included 6 case-control studies and 2 cohort studies with 858 cases of fruit intake, 858 cases of vegetable intake, 667 cases of fat intake, 412 cases of red meat intake and 633 cases of processed meat intake ( Table 1). The NOS scores of quality of the included studies are listed in Table 1. The controlled variables are specifically listed in Table 2.
Fruit intake and BE risk. The ORs have been adjusted for the potential multivariable confounders, including age, sex, ethnicity, energy intake, body mass index, waist-to-hip ratio, smoking, alcohol, education, medication use, gastro-esophageal reflux and vitamins. Figure 2 showed that the pooled OR was 0.65 (95% CI = 0.37-1.13) with significant heterogeneity (P = 0.004, I 2 = 77%). No statistical evidence of publication bias was observed from Begg's test (P = 0.089). These results suggest that fruit intake may be not significantly associated with BE risk.
Vegetable intake and BE risk. The ORs (95% CI) of each study and the pooled OR (95% CI) for the highest versus lowest category are listed in Fig. 3. The ORs were adjusted for the confounders listed above. The pooled OR was 0.45 (95% CI = 0.29-0.71) with heterogeneity (P = 0.05, I 2 = 61%) with no publication bias according to Begg's test (P = 0.308). The results indicate that vegetable intake may have a strong inverse association with BE risk (with a 55% decreased risk of BE).
Fat intake and BE risk. The ORs were adjusted for the confounders listed above and Helicobacter pylori infection. The pooled OR of saturated fat intake was 1.25 (95% CI = 0.82-1.91, Fig. 4a) with no significant heterogeneity (P = 0.69, I 2 = 0%) and no publication bias according to Begg's test (P = 0.540). There were two case-control studies for monounsaturated fat (Fig. 4b) and the pooled OR was 0.92 (95% CI = 0.32-2.64) with heterogeneity (P = 0.09, I 2 = 65%). Begg's test showed no statistical evidence of publication bias (P = 1.000). The pooled OR of polyunsaturated fat intake was 0.67 (95% CI = 0.35-1.26) with no heterogeneity (P = 0.28, I 2 = 16%, Fig. 4c) and no publication bias (Begg's test, P = 1.000). Additionally, OR = 1.03 (95% CI = 0.48-2.23) indicated that there was no association between cholesterol intake and BE risk (Fig. 4d). There was no analysis for other types of fat, such as trans fat and omega 3 because of limited data. These results suggest that fat intake may be not associated with BE risk.
Red and processed meat intake and BE risk. The ORs (95% CI) of each study and the pooled OR (95% CI) for the highest versus lowest category are listed in Fig. 5a,b. The ORs were adjusted for potential multivariable confounders, including age, sex, energy intake, body mass index, waist-to-hip ratio, physical activity, smoking, alcohol, education, medication use, gastro-esophageal reflux and H. pylori. A pooled analysis yielded the findings that red meat consumption was no associated with BE risk (OR = 0.85, 95% CI = 0.61-1.17) with no significant heterogeneity (P = 0.52, I 2 = 0%, Fig. 5a). Additionally, there was no association between processed meat consumption and BE risk (OR = 1.03, 95% CI = 0.73-1.46) with no heterogeneity (P = 0.25, I 2 = 27%). These results suggest that red and processed meat consumption may be not associated with BE risk. Dose-response meta-analysis. We also performed a dose-response meta-analysis to evaluate the dose-response associations between dietary fruit and vegetable 17,19,22 , saturated fat 23,24 , monounsaturated fat 23,25 and polyunsaturated fat 23,25 intakes and BE risk. It was not possible to analyze dose-response relationships for cholesterol, red meat and processed meat due to limited data. The dose-response analysis showed that BE risk decreased 16% (OR: 0.84, 95% CI 0.78-0.91) per unit increase (serving/day) with vegetable intake. There were no significant changes of BE risk per unit increase (serving/day) Table 2. Controlled variables of the included studies. N: Netherlands; BMI: body mass index (kg/m 2 ); WHR: waist-to-hip ratio; GER: gastro-esophageal reflux; V: vitamin.

Figure 2. Estimates (95% CI) of fruit intake (highest versus lowest category) and BE risk.
There was no association between fruit intake and BE risk (P = 0.13).

Discussion
Studies have explored and reported the possible relationships and mechanisms supporting the idea that high fruit and vegetable intakes reduce BE risk. Dietary fruits and vegetables may be inversely associated with BE risk, which may influence the early carcinogenesis of EAC 19 . Fruit intake is inversely associated with BE risk and may influence the process of carcinogenesis of EAC 19 . Keszei et al. 26 conducted a large Dutch cohort study and found that vegetable consumption may prevent the risk of BE, and intriguingly, this effect was different in men and women. However, findings from another study supported the idea that increased intake of fruits and vegetables is associated with a lower BE risk in both men and women 27 . Polyunsaturated fat and omega 3 intakes show inverse associations with BE risk that are stronger for long-segment BE 23 ; however, higher total fat and saturated fat intakes have been reported to be associated with significantly increased BE risk (3 rd compared with 1 st ) 24 . In contrast, some studies have reported inconsistent findings in terms of associations between total fat 23,25 and BE risk. Thus, these issues need to be examined by further investigations.  Recently, meats, particularly red and processed meat intakes, have been reported as risk factors for oral cavity and oropharynx cancer 28,29 , EAC 30-32 , gastric cancer 33,34 , colorectal cancer [35][36][37] , pancreatic cancer 38 , hepatocellular carcinoma 39 , breast cancer 40,41 , lung cancer 42,43 , renal cancer 44 , bladder cancer 45,46 , ovarian cancer 47 , brain tumors 48 , glioma 49 , non-Hodgkin lymphoma 50 , type 2 diabetes 51 , stroke 52 and coronary heart disease 53 . Red and processed meats are one of the major sources of nitrate and N-nitroso compounds, which are considered to be carcinogenic in humans and risk factors of BE 54 ; however, it is worth mentioning that not all cooked meats increase BE risk. Existing studies have suggested that there was no obvious association between well-cooked meats and BE risk 55,56 . Moreover, the results of some studies have shown that red and processed meat intakes were not associated with BE risk 18,25 .
Although some studies have reported that dietary fruit, vegetable, fat, and red and processed meat intakes were associated with BE risk, the related evidence is sparse and inconsistent, and there had been no published meta-analysis. This study focuses on these issues for the first time and provides reliable evidence to date.
Eight hundred fifty-eight cases of fruit intake, 858 cases of vegetable intake, 667 cases of fat intake, 412 cases of red meat intake and 633 cases of processed meat intake were included in this analysis. Our findings indicated that dietary vegetable rather than fruit intake was associated with significantly reduced BE risk (with 55% lower risk). Furthermore, the dose-response analysis suggested that the increases in vegetable [per unit increase (serving/ day)] intake was significantly associated with a 16% decreased risk of BE.
Figures 4a-d indicated that fats, including saturated fat, monounsaturated fat, polyunsaturated fat and cholesterol intakes, were not associated with BE risk. The dose-response analysis also yielded similar results. Other types of fat intake were not analyzed due to the limited data.
Although previous studies have reported that red and processed meat intakes were associated with a significantly higher incidence of BE 54 , our analysis suggested that there are inconsistent results (Fig. 5a,b). In addition, the dose-response analysis also did not indicate significant associations between red and processed meat intakes and BE risk.
Overall, these results suggest that essential moderate vegetable intake was associated with decreased BE risk, which may be important for the prevention of EAC. Dietary fruit, fat and red and processed meat intakes may be not associated with BE risk. Nonetheless, these results should be treated with caution, and more high-quality designs are needed to further validate these findings because of the limited studies included, the heterogeneity among the studies and the influence of potential confounders (listed in the following section on limitations).

Study strengths and limitations.
Because there is no previously published meta-analysis that has evaluated the overall effects of fruit, vegetable, fat, and red and processed meat intakes on BE risk, a quantitative synthesis of the eligible data from included studies was required to provide important evidence on the associations. The dose-response analysis was also conducted to further assess these associations.
However, the limitations of the present meta-analysis must be taken into consideration. First, various potential factors may contribute to the heterogeneity of the included observational studies. However, the ORs of all these studies were adjusted for potential multivariable confounders, including age, sex, ethnicity, energy intake, body mass index, waist-to-hip ratio, physical activity, smoking, alcohol, education, medication use, gastro-esophageal reflux H. pylori and vitamins. Second, the quality of some studies was not high, despite meeting the criteria, and the sample size was not large. Third, there were five studies performed in the USA, two studies performed in the Netherlands and one study performed in the UK (Table 2), which may be related to the heterogeneity of statistical generalizability to some degree. Thus, more multicenter studies should be performed in other countries and regions. Fourth, because the results in this meta-analysis were only based on the diagnosis of BE, the limited data available should not be used to infer conclusions about other gastrointestinal lesions, particularly EAC. Fifth, because of the small number of studies, our analysis did not perform any subgroup analyses on the types of common fruits, such as apple, pear, orange and banana, and vegetables, such as dark green vegetables, leafy vegetables, starchy vegetables, allium, garlic, pepper and legumes 17,26 . Sixth, our study did not investigate the associations between BE risk and other dietary factors, such as fish, poultry meat 18 , other white meat 25 , cooking techniques, nitrates from pesticides on fruits and vegetables, heme iron from meat, dairy products, sugar, protein, dietary antioxidants and mineral intake. Further studies with multifactorial subgroup analyses are needed to provide more complete data. Lastly, the study type, publication year, geographic location, sample size, type of dietary exposure, dietary exposure category and quality of the studies may lead to bias.

Conclusions
This is the first systematic review and meta-analysis that examined the associations between fruit, vegetable, fat, and red and processed meat intakes and BE risk. Dietary vegetable intake may be significantly associated with a decreased risk of BE. Dietary fruit, fat and red and processed meat intakes do not contribute to an increased BE risk. Well-designed, large, prospective studies with better established dose-response relationships are needed to further validate these findings.