Dietary flavonoid intake and the risk of digestive tract cancers: a systematic review and meta-analysis

Several epidemiological studies have investigated the association between dietary flavonoid intake and digestive tract cancers risk; however, the results remain inconclusive. The aim of our study was to evaluate this association. PubMed and the Web of Knowledge were searched for relevant publications from inception to October 2015. The risk ratio (RR) or odds ratio (OR) with the 95% confidence interval (95% CI) for the highest versus the lowest categories of flavonoid intake were pooled using a fixed-effects model. A total of 15 articles reporting 23 studies were selected for the meta-analysis. In a comparison of the highest versus the lowest categories of dietary flavonoid intake, we found no significant association between flavonoid intake and oesophageal cancer (OR = 0.91, 95% CI = 0.75–1.10; I2 = 0.0%), colorectal cancer (OR = 1.02, 95% CI = 0.92–1.14, I2 = 36.2%) or gastric cancer (OR = 0.88; 95% CI = 0.74–1.04, I2 = 63.6%). The subgroup analysis indicated an association between higher flavonoid intake and a decreased risk of gastric cancer in the European population (OR = 0.78, 95% CI = 0.62–0.97). In conclusion, the results of this meta-analysis do not strongly support an association between dietary flavonoid intake and oesophageal or colorectal cancer. Furthermore, the subgroup analysis suggested an association between higher dietary flavonoid intake and decreased gastric cancer risk in European population.

Statistical analysis. The pooled ORs with 95% CIs (highest compared to the lowest category of flavonoid intake) were computed from the adjusted ORs and RRs to measure the association between dietary flavonoid intake and the risk of digestive tract cancers.
The extent of heterogeneity across studies was determined using a chi-square test and an I 2 test; if P < 0.05 and/ or I 2 > 50%, indicating significant heterogeneity, a random-effect model was selected. Otherwise, a fixed-effect model was applied 11 . Meta-regression and subgroup analyses were performed to explore the possible source of heterogeneity, such as geographic region, experimental design, sample size and publication year 12,13 . Subgroup analyses were also performed to evaluate the potential effect of the modification of variables, including the study design, geographic region, cancer subtype, and dose. Begg's funnel plots and Egger's linear regression test were performed to assess the publication bias. A value of P < 0.05 was considered statistically significant 14,15 .
All analyses were conducted using STATA software (version 12.0; StatCorp, College Station, TX, USA) and a value of P < 0.05 was considered as statistically significant.

Results
Literature search and study characteristics. The electronic search of PUBMED and web of knowledge identified a total of 1595 potentially relevant articles. Fifty-one articles were reviewed in full after reviewing the title and abstract. Among them, 7 articles were reviews, 27 articles reported sub-class flavonoids, and 2 articles reported flavonoid supplements. As a result, 15 articles reporting 23 studies, encompassing 312,734 digestive tract cancer cases and 1,142,276 controls were selected for the meta-analysis [16][17][18][19][20][21][22][23][24][25][26][27][28][29][30] . The detailed steps of our literature search are shown in Fig. 1, and the main characteristics of the included studies are shown in Table 1.
Flavonoid intake and overall digestive tract cancer risk. We pooled the study-specific ORs using a fixed-effect model. No significant associations were detected between the highest compared with the lowest category of flavonoid intake and digestive tract cancers (overall OR = 0.96, 95% CI = 0.89-1.05, I 2 = 36.1%) (Fig. 2).
Flavonoid intake and oesophageal cancer risk. Seven studies examined the association between flavonoid intake and oesophageal cancer risk. The pooled OR of oesophageal cancer for the highest versus lowest categories of flavonoid intake was 0.91 (95% CI = 0.75-1.10, I 2 = 0.0%), suggesting that flavonoid intake was not significantly associated with the risk of oesophageal cancer (Fig. 2).
As shown in Table 2, a subgroup analysis was conducted by geographic location, study design, and histological type. However, non-significant associations of dietary flavonoid intake with oesophageal cancer were detected among all strata for the between-study subgroup analyses.
Flavonoid intake and colorectal cancer risk. A total of eight studies assessed the association between dietary flavonoid intake and colorectal cancer risk. The pooled OR of colorectal cancer risk for the highest versus the lowest categories of flavonoid intake was 1.02 (95% CI = 0.92-1.14, I 2 = 36.2%), indicating that flavonoid intake was not significantly associated with colorectal cancer risk.
The subgroup analysis indicated that dietary flavonoid intake has no significant effect on colorectal cancer risk in the US or European population. For colorectal cancer subtypes, no significant association was found between flavonoid intake and colon cancer (pooled OR = 0.92, 95% CI = 0.80-1.05) or rectal cancer (pooled OR = 0.90, 95% CI = 0.75-1.08), as shown in Table 2.
Flavonoid intake and gastric cancer risk. Six studies investigated flavonoid intake and gastric cancer risk. As shown in Fig. 2, the meta-analyses demonstrated no significant association between gastric cancer risk and flavonoid intake (OR = 0.88; 95% CI = 0.74-1.04, I 2 = 63.6%) for a comparison of the highest to the lowest category of intake.
In the subgroup analysis, we found that flavonoid intake was significantly associated with gastric cancer risk in Europe (pooled OR = 0.78, 95% CI = 0.62-0.97), but not in the United States or Asia (Table 2). With respect to study design, we did not find that flavonoid intake was statistically associated with gastric cancer in either the cohort (pooled OR = 0.86, 95% CI = 0.67-1.09) or case-control studies (pooled OR = 0.90, 95% CI = 0.72-1.13).
Heterogeneity analysis. For most of the outcomes of digestive tract cancer, the I 2 values of heterogeneity were lower than 50%. Only the levels of heterogeneity for gastric cancer were intermediate (I 2 = 63.6%). To explore the sources of heterogeneity, we performed subgroup analyses with stratification by geographic location and study design. We found that flavonoid intake was significantly associated with gastric cancer risk in Europeans (pooled OR = 0.78, 95% CI = 0.62-0.97) but not in Americans or Asians. Therefore, geographic location may partially account for the appreciable heterogeneity. Meta-regression showed that study design, geographic location, source of controls, and publication year had no significant impact on heterogeneity. Moreover, the leave-one-out analysis showed that the key contributor to heterogeneity was the study conducted by Zamora-Ros 16 . After excluding this study, the heterogeneity was reduced to I 2 = 23.4%, and the summary OR for oesophageal cancer was 1.304 (95% CI = 0.807-1.325), which was similar to the main finding.

Discussion
The present meta-analysis first evaluated the association between dietary flavonoid intake and the risk of digestive tract cancers based on the highest versus the lowest categories. We found no significant association between the highest dietary flavonoid intake and oesophageal cancer, colorectal cancer, or gastric cancer. However, studies conducted in European populations indicated an association between higher flavonoid intake and a decreased risk of gastric cancer.
Previous epidemiological studies have suggested that flavonoid intake has no significant effect on breast cancer (RR = 0.98, 95% CI = 0.86-1.12) 31 or colorectal neoplasms (RR = 1.03, 95% CI = 0.88-1.20 ) 32 . The meta-analysis also suggested a significant association between the highest category of flavonoid intake and a reduced risk of lung cancer (RR = 0.76, 95% CI = 0.63-0.92) 33 . The association between dietary flavonoid intake and the risk  of digestive cancers identified in our meta-analysis adds new information regarding the relationship between flavonoid intake and cancer risk. The lack of a strong association between flavonoid intake and oesophageal or colorectal cancer is particularly surprising because numerous in vitro and animal studies have demonstrated an inverse association between flavonoid and oesophageal or colorectal cancer [34][35][36][37] . However, many flavonoids present in foods cannot be absorbed in their native form, including esters, glycosides, and polymers 38 . It has been demonstrated that the amount of flavonoids that are bioavailable is only a small proportion of the ingested amount, ranging from 0.2-0.9% for tea catechins to 20% for quercetin and isoflavones 39,40 . In in vitro and animal studies, the intake of flavonoids is much higher than that in humans, and it remains unclear whether the beneficial antiproliferative and antioxidative effects observed during vitro studies are also present in humans 41 . Other studies have also suggested that flavonoids have weaker actions in vivo than in vitro 42 .

Figure 2. The forest plot between highest versus lowest categories of flavonoids intake and digestive tract cancers risk.
In the subgroup analysis by geographic location, we found an inverse association between flavonoid intake and gastric cancer in Europe but not in America or Asia. The reasons for this may include the complexity due to the presence of flavonoids from various food sources, the occurrence of a large amount of flavonoids in nature, and the diversity of dietary culture 43 . The main dietary sources of flavonoids are fruits, vegetables, tea, and red wine 44 . However, the flavonoids in vegetables and fruits depend on the type of cultivation, crop variety and location, as well as the specific morphological part of the plant 45 . Moreover, we could not exclude cultural differences in the storage and preparation of foods, particularly vegetables, which may also lead to this result.
Medium heterogeneity was detected for the association between flavonoid intake and gastric cancer. A meta-regression was used to explore the sources of heterogeneity, which showed that the study design, geographic location, source of controls, and publication year had no significant impact on heterogeneity. Then, we performed subgroup analyses by study design and geographic location and, found that flavonoid intake was significantly associated with gastric cancer risk in Europe, but not in America or Asia, suggesting that the intermediate heterogeneity is partially explained by geographic location. Moreover, the leave-one-out analysis showed that the key contributor to heterogeneity was the study conducted by Zamora-Ros 16 . After excluding this study, the summary estimate was not materially altered, but the I 2 of heterogeneity was reduced from 63.6% to 23.4%.
The role of flavonoids in gastric carcinogenesis has been attributed to several mechanisms. Flavonoids have anticarcinogenic effects by way of their antioxidant properties, which are attributed to their ability to modulate antioxidant pathways 46 . Moreover, flavonoids can regulate cell proliferation and apoptosis, modulate phase I and II metabolic enzymes, and affect inflammatory pathways 47 . Another possible explanation for the protective effects of flavonoids m may be their potential anti-Helicobacter pylori effect 48 , including direct bactericidal activity, the neutralisation of VacA, the reduction of urease secretion, and interference with Toll-like receptor 4 signaling 49 , which is unique to gastric cancer. Therefore, high dietary flavonoids intakes can possibly reduce the risk of gastric but not esophageal or colorectal cancer.
Several potential limitations of our meta-analyses should be acknowledged. The relatively small number of studies analysed makes it difficult to evaluate heterogeneity. Furthermore, we only compared the risk of cancer for those in the highest category of flavonoid intake with that of those in the lowest category of flavonoid intake, and the definition of these intake categories varied between studies. In addition, only published studies were included in the meta-analysis, which may have biased the results.

Conclusion
In summary, this meta-analysis provides little evidence of an association between the highest category of dietary flavonoid intake and the risk of oesophageal cancer or colorectal cancer. However, a subgroup analysis