Dietary Flavonoid Intake and Smoking-Related Cancer Risk: A Meta-Analysis

Purpose To systematically investigate the effects of dietary flavonoids and flavonoid subclasses on the risk of smoking-related cancer in observational studies. Methods Summary estimates and corresponding standard errors were calculated using the multivariate-adjusted odds ratio (OR) or relative risk (RR) and 95% CI of selected studies and weighted by the inverse variance. Results A total of 35 studies, including 19 case-controls (9,525 cases and 15,835 controls) and 15 cohort studies (988,082 subjects and 8,161 cases), were retrieved for the meta-analysis. Total dietary flavonoids and most of the flavonoid subclasses were inversely associated with smoking-related cancer risk (OR: 0.82, 95% CI: 0.72-0.93). In subgroup analyses by cancer site, significant associations were observed in aerodigestive tract and lung cancers. Total dietary flavonoid intake was significantly associated with aerodigestive tract cancer risk (OR: 0.67, 95% CI: 0.54-0.83) marginally associated with lung cancer risk (OR: 0.84, 95% CI: 0.71-1.00). Subgroup analyses by smoking status showed significantly different results. The intake of total flavonoids, flavonols, flavones, and flavanones, as well as the flavonols quercetin and kaempferol was significantly associated with decreased risk of smoking-related cancer in smokers, whereas no association was observed in non-smokers, except for flavanones. In meta-analysis for the effect of subclasses of dietary flavonoids by cancer type, aerodigestive tract cancer was inversely associated with most flavonoid subclasses. Conclusion The protective effects of flavonoids on smoking-related cancer risk varied across studies, but the overall results indicated that intake of dietary flavonoids, especially flavonols, was inversely associated with smoking-related cancer risk. The protective effects of flavonoids on smoking-related cancer risk were more prominent in smokers.


Introduction
Flavonoids are polyphenolic compounds that are abundant in fruits and vegetables. High intake of fruits and vegetables is associated with beneficial health effects, and these effects have been attributed in part to their high content of flavonoids. The World Cancer Research Fund (WCRF) and the American Institute of Cancer Research (AICR) reported summary estimates of the effects of fruit and vegetable consumption on cancer risk [1]. The consumption of fruits and vegetables most likely protects against cancers of the mouth, pharynx, larynx, esophagus, and stomach; the risk of lung cancer was only associated with fruit consumption. These cancers are smokingrelated cancers according to the International Agency for Research on Cancer (IARC) Monograph on tobacco smoking.
Smoking is classified as the cause of cancers of the lung, oral cavity, nasal and paranasal sinuses, pharynx, larynx, esophagus, kidney, liver, uterine cervix, stomach, bladder, pancreas, as well as myeloid leukemia [2]. Adverse effects of flavonoids in human health are rare, but the potential for detrimental health effects is based on in vitro studies in which pro-oxidant activities have been observed [3,4]. However, flavonoids are known to have powerful antioxidant, antiinflammatory, and anti-tumor activities against carcinogens [5,6,7]. Tobacco contains various carcinogens that can induce free radicals and cause gene mutations and the formation of DNA [8]. Thus, dietary flavonoids may play a role in protecting against smoking-related cancers.
The effects of dietary flavonoids on stomach and colorectal cancer were investigated using published studies [9] because flavonoids have been consistently found to be inversely associated with stomach and colorectal cancers in in-vitro studies [10,11,12]. However, based on a previous metaanalysis, no clear evidence supports that dietary flavonoids are associated with a reduced risk of stomach and colorectal cancers. The health effects of flavonoids and their subclasses might differ by cancer site. The cancer site that is affected by carcinogens such as tobacco smoking might be closely related to intakes of dietary flavonoids. Thus, a meta-analysis of published case-control and cohort studies was performed to calculate summary estimates of the effects of dietary flavonoid and their subclasses on smoking-related cancer risk.

Study selection
A systematic search for relevant studies published through October 31, 2012 was conducted with PubMed using the terms (cancer risk) and (flavonoid or flavonol or quercetin or kaempferol or myricetin or isorhamnetin or flavone or luteolin or apigenin or flavanone or eriodictyol or hesperetin or naringenin or flavan-3-ol or catechin or epicatechin or theaflavin or anthocyanidin or cyanidin or delphinidin or malvidin or pelargonidin or peonidin or petunidin or isoflavones). The inclusion criteria were as follows: (1) the original article described a case-control or cohort design; (2) the article reported the intake of either dietary flavonoids or subclasses of flavonoids; (3) the article reported the risk of smoking-related cancers that were defined according to the IARC monograph (cancers included those of the oral cavity, paranasal sinuses, nasal cavity, larynx, pharynx, lung, esophagus, stomach, liver, pancreas, kidney, bladder, and uterine cervix, as well as myeloid leukemia); and (4) the article reported 95% confidence intervals (CI) with adjusted odds ratios (OR) or relative risks (RR) for smoking-related cancer risk in subjects with the highest dietary flavonoid intake compared with those with the lowest dietary flavonoid intake.

Data collection
Data on the authors, publication year, cancer site, country in which the study was performed, study design, study period, dietary assessment method, reported flavonoid types, included subclasses for calculation of total flavonoid intake, controlled confounders, and multivariate-adjusted OR/RR and 95% CI were collected for the meta-analysis. If the study reported only the subgroup analyses by sex, viral infection status, or ethnicity, then the summary estimate of the subgroup was used for the meta-analysis to include an example from each article. However, the articles that reported individual risk estimates by cancer sites but not an all-cancer risk were used only in the subgroup meta-analysis that was conducted by cancer site. If multiple articles reported the risk of the same cancer site from the same cohort data, then the most recently published data were selected.

Statistical analysis
All statistical analyses were performed using the Stata software package (version 10, College Station, TX). Summary estimates and the corresponding standard errors were calculated using the multivariate-adjusted OR/RR and 95% CI of selected studies and weighted by the inverse variance. The heterogeneity was tested using the I 2 test and Q-test based on the χ 2 statistic, considering significant statistical heterogeneity as p < 0.1. Subgroup analyses were conducted by study design, sex, and smoking status. Publication bias was examined using Begg's tests. A subgroup analysis by smoking status was performed when the studies reported the risk in both smokers and non-smokers because the effect of smoking on cancer risk can vary by cancer site. Sensitivity analyses were performed to test the robustness of the results of the combined effects. Based on the heterogeneity of the included studies, fixed or random effects models were selected to calculate the pooled effect measures.

Results
A total of 1401 studies were examined, and the 174 studies that remained after excluding articles based on the titles and abstracts were further reviewed; 139 studies were excluded for the following reasons: 106 studies included a cancer type that was not smoking-related; 20 studies were not relevant for dietary flavonoid intake or any of the flavonoid subclasses; 6 studies were review articles; 4 studies did not report the cancer risk according to categories of flavonoid intake; 2 studies were updated with more recent studies from the same cohort; and 2 studies were not related to cancer risk ( Figure 1). Finally, 19 case-control studies [13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31] and 15 cohort studies [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46] were selected for the meta-analysis (Tables 1, 2). Begg's tests for publication bias showed non-significant results. Table 3 shows the meta-analysis of the risk of smokingrelated cancer in people with the highest intake of total dietary flavonoids compared with those with the lowest intake of total dietary flavonoids. Total dietary flavonoids were inversely associated with smoking-related cancer risk. However, in the subgroup analysis conducted by study design, this association was observed only in the case-control studies. Intake of total dietary flavonoids was inversely associated with smokingrelated cancer risk among smokers, but the associations were not significant for the non-smokers. The summary estimate calculated for the association between total dietary flavonoids and aerodigestive tract cancer risk was significant (OR: 0.67, 95% CI: 0.54-0.83). The association with lung cancer risk was only marginally significant (OR: 0.84, 95% CI: 0.71-1.00).
The summary estimates calculated for the association between the subclasses of dietary flavonoids and smokingrelated cancer risk are presented in Table 4. Dietary flavonols, flavones, flavanones, flavan-3-ols, and isoflavones significantly lowered the risk of smoking-related cancers, and the significant associations were maintained among the flavones, flavanones, and isoflavones in case-control studies. Flavonols were strongly associated with lower smoking-related cancer risk. The associations with flavonols were significant in only cohort studies, and all of the flavonol subgroups were significantly associated with smoking-related cancer risk if the fixed effects model was used. The subgroup analysis conducted by smoking status showed significantly different results. The dietary intake of flavonols, flavones, and flavanones was significantly associated with a lower risk of smoking-related cancer among smokers, but no association was observed among nonsmokers, except for flavanones. Quercetin and kaempferol, the components of flavonols, showed significant associations, but flavonol myricetin showed no association. However, the associations with flavonol quercetin, kaempferol, and myricetin differed by smoking status. The components of flavones (luteolin and apigenin), flavanones (naringenin), and flavan-3ols (catechin, epicatechin) were not significantly associated with smoking-related cancer risk, but hesperetin, a component of flavanones, was positively associated with smoking-related cancer risk. Table 5 shows summary estimates for the effect of subclasses of dietary flavonoids on cancer risk by cancer site. In aerodigestive tract cancer including oral, larynx, pharynx, and esophageal cancer, most of flavonoids subclasses were significantly associated with reduced risk. In lung cancer, flavonol quercetin and kaempferol were inversely associated with cancer risk, but these associations were not observed in total flavonol intake and any other flavonoid subclasses.

Discussion
Meta-analyses performed in this study including 19 casecontrols (9,525 cases and 15,835 controls) and 15 cohort studies (988,082 subjects and 8,161 cases) showed various results according to flavonoid type, smoking status, and cancer site. However, overall results revealed that dietary flavonoids were inversely associated with the risk of smoking-related cancers, and the associations were more prominent among smokers.  Garcia [14] 1999 Stefani [16] 1999 The summary estimates suggested that total dietary flavonoids and several flavonoid subclasses were associated with reduced risks of smoking-related cancers. Total dietary flavonoids were inversely associated with smoking-related cancer risk in both the case-control and cohort studies, although a non-significant association was observed in the   [9]. Significant associations were observed in the summary estimates of only the case-control studies, which are subject to more recall bias than cohort studies. However, the non-significant results might be caused by low statistical power due to the small numbers of cohort studies in the present study. In the summary estimates of cohort studies, dietary flavonols were significantly associated with a reduced risk of smoking-related cancer, and total flavonoids were also significantly associated with the reduced risk when the fixed effects model was applied. Thus, in contrast to previous study of colorectal and stomach cancers [9], dietary flavonoids may play a role in reducing smoking-related cancer risk. The preventative effects of dietary flavonoids may differ by cancer site among the smoking-related cancers. Lung cancer, which is  mostly caused by tobacco smoke, was the most commonly studied cancer site, but total dietary flavonoids was only marginally associated with lung cancer in the present study. Significant association between flavonoids and lung cancer risk was observed in the previous meta-analysis [47]. In their study, summary estimate of flavonoids were combined all studies reported total flavonoid intake and their subclasses altogether. However, the results of each flavonoid subclasses were similar. Kaempferol alone showed significant association with reduced lung cancer risk in the previous meta-analysis, and quercetin, although it is not statistically significant, showed most reduced risk of lung cancer. In our study, quercetin and kaempferol were significantly associated with reduced risk of lung cancer. Intake of total dietary flavonoids and most of their subclasses was significantly associated with aerodigestive tract cancer risk. Dietary flavonoids were not significantly associated with risk of cancer at any other site. The limited number of studies that were eligible for the subgroup analyses conducted by cancer site rendered the results inconclusive. However, the lower risk of cancers of the aerodigestive tract may be closely related to the protective effects of dietary flavonoids. The protective effects of flavonoids on cancer risk have been explained by several mechanisms. Flavonoids are polyphenolic compounds that are known to have antioxidant properties, so the free radical scavenging properties of flavonoids are closely related to the beneficial effects on cancer risk. The strong antioxidant properties of flavonoids effectively reduced various types of oxidants [48,49]. However, direct antioxidant activities of dietary flavonoids may not be the only explanation of the protective effects on cancer risk [50,51]. Dietary flavonoids are found at very low concentrations in human plasma due to their low bioavailability. The bioavailability of dietary flavonoids can vary according to their chemical structure and food sources. The bioavailability of isoflavones and quercetin was relatively high, whereas that of tea catechins and anthocyanins is very low [50,52]. The other explanation for the beneficial effects of flavonoids on cancer is that flavonoids play a role in regulating enzymatic pathways. Carcinogens that enter the body are first metabolized to more active forms by phase I enzymes such as cytochrome P450 (CYP), and the active forms can be detoxified by phase II enzymes such as UDP-glucuronyl transferase, glutathione S-transferase, and quinone reductase. Flavonoids are responsible for the inactivation of phase I enzymes as well as the activation of phase II enzymes [53]. Several studies examined for this meta-analysis performed a subgroup analysis by smoking status. Total flavonoid intake was associated with a reduced cancer risk among smokers but not among non-smokers. These different associations were observed for dietary flavonols, flavones, flavanones, and the flavonol components quercetin, kaempferol, and myricetin. Because smoking can increase oxidative stress, dietary flavonoids may function as antioxidants; therefore, the more prominent effects of flavonoids on the reduced cancer risk may be more likely to be observed among smokers. Smokers have lower levels of some antioxidants in their plasma than nonsmokers, suggesting the larger involvement of these antioxidants in smokers [54]. Similar results were observed in a previous meta-analysis of dietary flavonoids and the risk of lung cancer, and the results suggested that flavonoids function as antioxidants to reduce the lung cancer risk in smokers [47]. However, the antioxidant activities of dietary flavonoids might be limited due to their low bioavailability, as discussed earlier.
Tobacco has many types of carcinogens including PAH, benzo[a]pyrene (BaP), N-nitroso compound, and 4aminobiphenyl, and these carcinogens and their metabolites can cause gene mutations such as p53 mutations and the formation of DNA or protein adducts [8]. Dietary flavonoids protect human hepatoma cells from N-nitrosodimethylamine-, N-nitrosopyrrolidine-and BaP-induced DNA damage [55]. Associations between CYP1A1 and lung squamous cell carcinoma (SCC) and associations between CYP2E1 and lung adenocarcinoma (AC) were observed, suggesting a specificity of tobacco smoke PAHs for lung SCC and tobacco-specific nitrosamines for lung AC [56]. Quercetin inhibited the CYP450 enzymes that were activated by BaP [57], and the relationship between the consumption of onions rich in flavonoids and lung cancer risk was modified by CYP1A1 [17]. Dietary flavonoids may protect against carcinogenesis by modulating enzymatic pathways, suggesting that dietary flavonoids can function as anti-tumor agents among smokers.
Flavonoids may be associated with early carcinogenesis [58], suggesting that the cohort design is more suitable for association studies, but only a limited number of cohort studies have been published. Definition of high vs. low intake of flavonoids was slightly different across studies, which may result in high heterogeneity in the meta-analysis. Bioactive food components are highly correlated, and the bioavailability of dietary flavonoids differ considerably by food source, even among the same flavonoid types. Previous work has suggested that flavonoid food sources rather than specific flavonoids should be analyzed for associations with disease risk [59]. However, consistent associations between dietary flavonoids and their subclasses and smoking-related cancer risk were observed in the studies reviewed here. Furthermore, the subgroup analysis conducted by smoking status showed a consistently inverse association between dietary flavonoids and smoking-related cancer risk. Thus, despite these limitations, the present meta-analysis showed that dietary flavonoids are inversely associated with smoking-related cancer risk.
Although the protective effects of flavonoids varied according to cancer site and flavonoid type, the overall results indicated that dietary flavonoids might have preventative effects against smoking-related cancers. The beneficial effects of flavonoids on cancer risk were more prominent among smokers than nonsmokers. Total flavonols and flavonol quercetin may be the most powerful agents for smoking-induced carcinogenesis. Thus, we suggest that smokers might benefit from a flavonoidrich diet.