Colorectal cancer is the third most common cancer in the developed world(1) with incidence in Scotland being among the highest in Europe(2). Critical assessment of potential risk factors suggests that diets rich in plant-based foods, such as fruit and vegetables, may reduce the risk of developing colorectal cancer(3, Reference Duijnhoven, De Mesquita and Ferrari4). Though the mechanism by which these foods exert a protective effect is unclear, one hypothesis is the presence of high levels of potentially anti-carcinogenic phytochemicals(Reference Smith-Warner, Genkinger, Giovannucci, Herber, Blackburn and Go5). Flavonoids are a large and diverse group of phytochemicals and research into their anti-carcinogenic potential with animal and cellular model systems supports a protective role(Reference Nichenametla, Taruscio and Barney6, Reference Lambert, Hong and Yang7). Structurally distinct subclasses of flavonoids have varying capacities to modulate the progression of colorectal cancer, acting as antioxidants(Reference Surh, Kundu and Na8, Reference Duthie, Johnson and Dobson9), anti-inflammatory agents(Reference Sánchez de Medina, Vera and Gálvez10–Reference Peng, Dixon and Muga13), anti-proliferative agents(Reference Daskiewicz, Depeint and Viornery14–Reference Gosse, Guyot and Roussi16) or as regulators of signal transduction pathways(Reference Williams, Spencer and Rice-Evans17, Reference Van Dross, Xue and Knudson18). Of all the tissues in the human body, the large intestine may be exposed to higher flavonoid concentrations than other tissues(Reference Halliwell, Zhao and Whiteman19).
Epidemiological assessment of the relationship between dietary flavonoid intake and colorectal cancer is limited, with different cohort studies investigating different combinations of flavonoids(Reference Lin, Zhang and Wu20–Reference Arts, Jacobs and Gross25). With the exception of the Iowa Women's Health study(Reference Arts, Jacobs and Gross25), where a significant inverse association between catechin intake and incidence of rectal cancer was found (P < 0·01), no significant relationships between flavonoid intake and colorectal cancer incidence have been reported. Recently, two case–control studies utilised new and relatively comprehensive flavonoid food composition databases(Reference Kyle, Duthie, Andersen and Markham26, Reference Hertog, Hollman and Katan27) examining potential beneficial links between flavonoid intake and development of colorectal cancer. An Italian case–control study(Reference Rossi, Negri and Talamini28) found reduced risk of developing colorectal cancer with increasing isoflavone, anthocyanin, flavone and flavonol intakes, but not with catechin or flavanone intake. A large Scottish case–control study(Reference Theodoratou, Kyle and Cetnarskyj29) also observed reduced colorectal cancer risk with higher intakes of flavonols, catechins and procyanidins type B1–B4, though not with flavanones or phyto-oestrogens.
Black tea is an important source of flavonoids. It is likely that the distribution of flavonoid subclasses differs in different populations depending on the relative intakes of fruit, vegetables and tea. The typical Scottish diet is known to be low in fruit and vegetables and high in processed foods, with black tea commonly consumed(30, Reference McKie, MacInnes and Hendry31). Dietary flavonoid intakes in a North East of Scotland population were found to be comparable with other European countries including Italy(Reference Kyle, Duthie, Andersen and Markham26). Unlike the Scottish diet, high fruit and vegetable intake and low tea consumption characterise the Italian diet(Reference Turrini, Saba and Perrone32). Consequently, the observation of comparable levels of flavonoid intake may be misleading, with black tea being the major source of flavonoids in the Scottish population(Reference Kyle, Duthie, Andersen and Markham26). Additionally, when assessing associations between flavonoid intake and disease risk in populations where high levels of tea are commonly consumed, misclassification of risk may occur if non-tea dietary sources are neglected(Reference Arts, Hollman and de Mesquita24).
The present study investigated the associations between dietary flavonoid intake and the risk of developing colorectal cancer in the North East of Scotland. The aim was to assess whether this relationship varied when considering total dietary and non-tea intake of four flavonoid subclasses (flavonols, catechins (flavon-3-ols), procyanidins and flavanones).
Materials and methods
Subjects and study design
Participants were recruited as part of a population-based case–control study of colorectal cancer investigating colorectal cancer and genetic polymorphisms in xenobiotic metabolising enzymes in the North East of Scotland(Reference Little, Sharp and Masson33). Patients (cases) presenting with their first primary cancer, diagnosed between September 1998 and February 2000, and resident in the Grampian health board area were approached after histological confirmation of incident invasive tumours of the colon or rectum. Population-based controls, frequency matched to cases, were selected from the Grampian Community Health Index (a list of all those registered with a general practitioner in the National Health Service). Controls who declined to participate were replaced. The present study was conducted according to the guidelines laid down in the declaration of Helsinki and all procedures involving human subjects and patients were approved by the Joint Ethical Committee of the Grampian Health Board and the University of Aberdeen. Written consent was obtained from all participants.
Dietary assessment and analysis
With the permission of their general practitioner, subjects were contacted by mail and asked to complete the Scottish Collaborative Group FFQ version 6.31 (www.foodfrequency.org.uk) and a questionnaire which included questions on a range of sociodemographic and lifestyle factors relevant to colorectal cancer aetiology. Of those invited to participate, 62 % of cases and 61 % of controls completed and returned our questionnaires (264 cases and 408 controls). Of these, seven subjects (three cases and four controls) were excluded as their FFQ were returned incomplete(Reference Little, Sharp and Masson33).
Dietary data were converted into estimates of nutrient intakes using the UK McCance and Widdowson's food composition tables(34). Twelve subjects (six cases and six controls) were excluded on the basis of implausible total energy intakes ( ± 3 sd of mean intake)(Reference Kyle, Masson and Duthie35). Total dietary and non-tea flavonol, flavon-3-ol, procyanidin and flavanone intakes were computed using a flavonoid food composition database compiled by systematic review of the available composition literature representing foods consumed in the UK(Reference Kyle, Duthie, Andersen and Markham26). In this database, as numerous conjugated forms of flavonoids are commonly present in foods, all food content and intake values are expressed in their free (aglycone) form. Additionally, each subclass is a summary of individual flavonoid compounds as outlined: flavonols – quercetin, kaempferol and myricetin; flavones – lutein and apigenin; flavon-3-ols – catechin, epicatechin, epigallocatechin, epigallocatechin gallate, epicatechin gallate and gallocatechin; procyanidins – types BI-IV; flavanones – hesperidin and naringenin. The 150-item semi-quantitative FFQ was previously tested in the local population and compared with 4 d weighed intake records for macro- and micronutrients as well as each flavonoid subclass(Reference Kyle, Masson and Duthie35, Reference Masson, McNeill and Tomany36). Relatively good Spearman rank correlation and agreements were found for flavonols, procyanidins and flavon-3-ols (correlations 0·70–0·94), but not for flavanones and flavones, 0·33 and 0·18 respectively. As a result, flavone data were not used. The FFQ included questions on frequency of consumption of each food item in the last year as ‘rarely or never’, ‘monthly’ or 1, 2, 3, 4, 5, 6, 7 d per week and amount (‘measures’) of consumption of a range of food items. The measure for each food was designed to be a small portion or unit measure such as a tablespoon; hence a single standard portion of a particular food might be two measures(Reference Masson, McNeill and Tomany36).
Statistical analysis
SPSS (version 13; SPSS, Inc., Chicago, IL, USA) and Stata 8 (StataCorp LP, College Station, TX, USA) were used for statistical analysis. Case and control demographic and lifestyle factors were summarised. Independent t tests and χ2 tests assessed the significance of any differences. The mean values and standard deviations of measures of tea, fruit and vegetables were calculated. The measure for tea was a cup, while fruit and vegetable measures were the total number of single portions per d as reported on the FFQ.
The distributions of flavonoid variables were checked. Log transformation normalised the non-tea flavonoid intake, but not total flavanone intake. Median intake and interquartile range of flavonoid subclasses and individual compounds were computed for cases and controls before investigating any significant differences by applying the non-parametric Wilcoxon rank-sum test. The correlation between flavonoid intakes and other dietary and non-dietary factors (for example, BMI) were calculated using the Spearman rank correlation coefficients.
Flavonoid, tea, fruit and vegetable intakes were adjusted for total energy using the nutrient residual method(Reference Willett and Willett37) before division of subjects into quartiles of intake (lowest to highest). The odds of developing colorectal cancers with increasing intake of flavonoids were then computed. Multivariate adjusted OR were adjusted for sex, age, correlated dietary variables and potential confounders; factors making a significant contribution (likelihood ratio test P < 0·1) were retained in the regression model, including variables with significant correlations with flavonoid intake. The likelihood ratio test was used to assess trends across the quartiles of flavonoid intake (P trend). The primary analysis was for all colorectal cancers combined; secondary analyses were carried out for colon and rectal cancers separately.
Results
A total of 261 colorectal cancer patients (cases) (186 and seventy-five with colon and rectal cancer, respectively) and 404 population-based control subjects were included in the full analysis of the present investigation (Table 1). Cases (age range 39–92 years) tended to be older than controls (age range 32–88 years) and a higher proportion of cases were male. Although fewer cases were current smokers (P < 0·01), more were ex-smokers (P < 0·01) than controls. A higher percentage of cases reported a family history of colorectal cancer compared with controls, while non-steroidal anti-inflammatory drugs were more frequently taken by controls.
Table 1 Demographic characteristics of cases and controls
NSAID, non-steroidal anti-inflammatory drugs.
Value was significantly different from that of the controls: * P < 0·05, ** P < 0·01, *** P < 0·001.
When stratified by sex, energy intake of cases and controls did not differ significantly, though energy intake in men (10·5 (sd 3·8) MJ/d) was significantly higher than in women (9·1 (sd 3·9) MJ/d) for both cases and controls (P < 0·05). Significantly higher total dietary intake of flavonols, procyanidins and flavanones was recorded for cases than controls. The converse was observed for non-tea flavonol, procyanidin and flavon-3-ol intake, with controls reporting higher intakes (Table 2).
Table 2 Dietary flavonoid intakes of cases and controls
(Medians and interquartile ranges (IQR))
Median was significantly different from that of the controls: * P < 0·05, ** P < 0·01, *** P < 0·001.
Tea intake was highly correlated with flavonol, procyandin and flavon-3-ol intakes (r>0·90; P < 0·001). Exclusion of tea as a potential source of flavonoids yielded markedly lower estimates of intake (Table 2), with black tea consumption accounting for between 71 and 93 % of total flavonol, procyanidin and flavon-3-ol intakes. After exclusion of tea flavonoids, gallated catechin esters (epigallocatechin gallate, epicatechin gallate and epigallocatechin) were replaced with catechin and epicatechin as the main contributors to dietary flavon-3-ol intake, while quercetin remained the main flavonol consumed before and after adjustment for tea consumption.
Colorectal cancer patients consumed more tea than control subjects (P < 0·01), drinking an average of 3·7 (sd 0·1) and 3·3 (sd 0·1) cups per d, respectively. There were no significant differences between reported fruit and vegetable intake between cases and controls (1·9 (sd 1·8) and 1·9 (sd 1·9) measures of fruit per d and 3·1 (sd 4·2) and 3·6 (sd 3·9) measures of vegetables per d for cases and controls, respectively). Controls under the age of 55 years consumed less tea and more vegetables than controls over 55 years (P < 0·01), but this effect was not seen in cases. This had an impact on intake of both total and non-tea flavonoids, with controls reporting a lower intake of flavonols (P < 0·05) and a higher intake of procyanidins (P < 0·05) and flavon-3-ols (P < 0·01) than cases.
Increasing energy-adjusted black tea consumption demonstrated a weak increased risk of developing colorectal cancer (OR 1·5; 95 % CI 1·0, 2·4; highest v. lowest quartile; P trend = 0·08), but this was attenuated after adjustment for confounding variables. Energy-adjusted vegetable intake was significantly associated with colon cancer (highest quartile of energy-adjusted intake OR 0·6; 95 % CI 0·3, 0·9; P trend = 0·03). Again this association was not seen after multivariate adjustment (highest v. lowest quartile of energy-adjusted intake OR 1·0; 95 % CI 0·6, 1·8; P trend = 0·52). No effect was seen with rectal cancer. There was also no evidence of a relationship between fruit consumption and colorectal cancers (data not shown).
No association between total dietary flavonol, procyanidin or flavon-3-ol intake and risk of developing colorectal cancer was observed (Table 3). There was a weak (P trend = 0·04) trend towards increased risk of colorectal cancer with higher levels of flavanone intake. Stratification by cancer site (Table 4) strengthened this observation, with a significant trend apparent for colon cancer (multivariate OR 1·3; 95 % CI 0·7, 2·4; highest v. lowest quartile; P trend < 0·01).
Table 3 Total dietary and non-tea flavonoid intakes by quartile of intake and risk of colorectal cancer (261 cases and 404 controls)
(Odds ratios and 95 % confidence intervals)
* P < 0·05.
† Energy adjusted.
‡ Adjusted for energy, age at diagnosis, family history, non-steroidal anti-inflammatory drugs, aspirin, Mn, riboflavin, vitamin C, folate.
Table 4 Association between total dietary and non-tea flavonoid intakes by quartile of intake and colon and rectal cancers†
(Odds ratios and 95 % confidence intervals)
* P < 0·05.
† Multivariate adjusted for energy, age at diagnosis, family history, non-steroidal anti-inflammatory drugs, aspirin, Mn, riboflavin, vitamin C, folate.
Analysis of non-tea flavonoid intake indicated a significant inverse association between non-tea flavonol intake and risk of colorectal cancer (P < 0·05), but not for non-tea procyanidin or flavon-3-ol intake (Table 3). Separate analyses of colon and rectal cancer cases demonstrated that non-tea flavonol intake was significantly associated with a reduced risk of developing colon (OR 0·5; 95 % CI 0·3, 0·8; highest v. lowest quartile; P trend < 0·01), but not rectal cancer in the adjusted model (Table 4). Further assessment of the relationship with intake of individual non-tea flavonol compounds highlighted an association between quercetin (highest v. lowest quartile multivariate adjusted OR 0·4; 95 % CI 0·2, 0·8; P trend < 0·01) and colon cancer (Table 5).
Table 5 Association between non-tea flavonol intake by quartile of intake and colorectal cancers†
(Odds ratios and 95 % confidence intervals)
* P < 0·05, ** P < 0·01.
† Adjusted for energy, age, education, family history, non-steroidal anti-inflammatory drugs, aspirin, vitamin C, folate, fruit and vegetables.
Discussion
Associations between dietary intake of four different flavonoid subclasses and colorectal cancer risk were assessed in the present case–control study amongst men and women from the North East of Scotland. Total dietary flavonol, flavan-3-ol or procyanidin intakes were not associated with colorectal cancer risk. However, there was a weak negative trend with flavanone intake and colon cancer risk. Black tea and its flavonoids did not appear to be related to colorectal cancer risk in this population. An inverse association between colorectal cancers, more specifically colon cancer, and non-tea quercetin intake was observed.
While the retrospective nature of the case–control design of the present study makes it potentially susceptible to recall bias, the fact that the study has a population-based study design helps to minimise selection biases. A similar participation rate was achieved as in a previous UK postal contact study(Reference Macfarlane, Gray and Kincey38). In addition, response rates and the quality of completion of the dietary questionnaires in the cases and controls were comparable. Moreover, it is worth noting that assessment of known risk factors resulted in associations consistent with previous evidence(Reference Little, Sharp and Masson33), while reported energy intake was comparable with previous observations for this population. Multiple testing may also be a potential source of bias, though due to the relatively small sample size, the present study's aim was hypothesis generating rather than hypothesis testing.
Our estimates of intake of flavonols and catechins are comparable with other tea-drinking populations and higher than non-tea-drinking countries such as Italy and America(Reference Lin, Zhang and Wu20–Reference Arts, Jacobs and Gross25), while flavanone intake was comparable with a Finnish study(Reference Knekt, Kumpulainen and Järvinen23), but lower than an Italian population study(Reference Rossi, Negri and Talamini28). Intakes of flavonols, procyanidin and flavon-3-ols were higher among controls, but not cases, when compared with another recently published Scottish colorectal cancer study(Reference Theodoratou, Kyle and Cetnarskyj29). One reason for this may be a difference in the age distribution of subjects between the studies; in the present study both cases and controls were on average 6 years older than in the other study. Though a relatively small difference, in the UK tea consumption by the 64- to 74-year age group is 25 % higher than by the 50- to 64-year-olds(39).
The results of our regional North East of Scotland study are only partially consistent with those of the larger Scottish study of Theodoratou et al. (Reference Theodoratou, Kyle and Cetnarskyj29). In their study, increasing total dietary flavonol (OR 0·73; P trend < 0·02), quercetin (OR 0·68; P trend < 0·001), catechin (OR 0·68; P trend < 0·001), epicatechin (OR 0·74; P trend < 0·05) and procyanidins (OR 0·78; P trend < 0·05) intake significantly reduced the risk of developing colorectal cancer, while no effect was found for total flavanones or flavon-3-ols. The latter result is consistent with our findings; however, in contrast to Theodoratou et al. (Reference Theodoratou, Kyle and Cetnarskyj29) we did not observe a trend in risk with total flavonols, catechin or epicatechin intakes and colorectal cancer. One explanation may be that our findings may reflect the smaller sample size or the different age distribution of subjects. Alternatively it may be that, due to regional variations in diet within Scotland, our population drank more tea and consumed less fruit and vegetables, providing relatively lower non-tea sources of flavonoids. Our findings with regard to non-tea flavonoid intake reflect those of a large Italian case–control study(Reference Rossi, Negri and Talamini28). Researchers in the Italian study identified a positive trend against colorectal cancers with increasing intake of flavonols (OR 0·64; P trend < 0·001), but not flavon-3-ols or flavanones in a population with high fruit and vegetable intake and infrequent tea consumption. Each of the now three case–control studies have utilised recently compiled and relatively comprehensive flavonoid composition databases, one developed in the UK reflecting the flavonoid content of foods commonly consumed in Europe(Reference Kyle, Duthie, Andersen and Markham26) and the USA(40) unlike the previously published cohort studies, in which an earlier more limited Dutch database with regional additions were employed(Reference Hertog, Hollman and Katan27, Reference Hertog, Hollman and van de Putte41–Reference Arts, van De Putte and Hollman43). Of earlier studies, only one reported significant associations between flavonoid intake and risk of colorectal cancer(Reference Knekt, Kumpulainen and Järvinen23). The Iowa Women's Health study, a cohort of 34 651 postmenopausal women, observed a significant inverse association between total flavon-3-ol intake and rectal cancer. The present study did not replicate these findings even when the analysis was limited to females only, but there were only twenty-eight women with rectal cancer in the present study. Black tea is not commonly consumed in America, and comparison of the US intake with the Scottish non-tea flavon-3-ol intake indicates that the Scottish non-tea intake (>8·5 mg/d) may be too low to detect a protective effect.
The lack of an association between black tea consumption and colorectal cancer risk is in agreement with previous findings(3). As total dietary flavonol, procyanidin and flavon-3-ol intakes were highly correlated with black tea consumption, a case has been made to study total and non-tea flavonoid intake separately(Reference Arts, Hollman and de Mesquita24). Assessment of these independent associations with risk of developing colorectal cancer suggests that flavonoids from dietary sources other than tea may modulate colorectal cancer risk.
Previous dietary flavonoid intake assessment in the North East of Scotland revealed that the main dietary sources of flavonoids were tea (46 % of the intake), onions (14 %), apples (10 %) and processed foods and beverages (13 %) for flavonols(Reference Kyle44). This suggests that flavonol intake from a diverse combination of fruit- and vegetable-based foods may be responsible for the observed inverse association.
Individual flavonoid compounds and their dietary sources have differing relative bioavailability(Reference Hollman and Arts45). Consequently, when assessing several flavonoid subclasses in relation to disease, the sources and types of flavonoids could have different potential relationships with colorectal cancer. The structure–activity relationship of flavonoids provides an analytical example to support this observation. Ranking of the ability of flavonoids to scavenge different reactive oxygen and nitrogen species in both aqueous and lipophilic environments indicates that quercetin is a potent antioxidant in vitro (Reference Duthie, Duthie and Kyle46, Reference Heim, Tagliaferro and Bobilya47). As large amounts of flavonoids remain unabsorbed in the lower gastrointestinal tract, they may exert their antioxidant effects at that site(Reference Halliwell, Zhao and Whiteman19). An imbalance between cellular levels of reactive oxygen species and antioxidants can result in damage to DNA, leading to mutation and dysregulation of oncogenes or tumour-suppressor genes(Reference Duthie, Duthie and Kyle46). Quercetin, as one of the most potent antioxidants in vitro, may act to minimise mutations. Suppression of cellular proliferation may also be structure dependent. For example, quercetin, but not catechin, can suppress proliferation of colon cancer cell lines(Reference Agullo, Gamet and Besson48). More recently quercetin has been reported to modulate cell growth signalling pathways(Reference Gosse, Guyot and Roussi16, Reference Kim, Bang and Kim49–Reference Wenzel, Hertzog and Kuntz52). Finally, chronic inflammation is an important target for preventive measures against colon cancer, with non-steroidal anti-inflammatory drugs and aspirin, for example, being thought to reduce risk by up to 50 %(Reference Dubé, Rostom and Lewin53, Reference Vane and Botting54). This is principally achieved through modulation of arachidonic acid release and its subsequent metabolism by cyclo-oxygenase and lipoxygenase, mediators of the inflammatory response(Reference Vane and Botting54). Flavonoids are known to modulate expression of these enzymes with quercetin, in particular, inhibiting their activity(Reference Surh, Kundu and Na8, Reference Kim, Son and Chang12, Reference Peng, Dixon and Muga13).
The inverse association observed between colorectal cancer risk and non-tea flavonol intake but not tea flavonoids in the present study is interesting and highlights the importance of assessing their different dietary sources in relation to disease risk. It is only the non-tea dietary sources of flavonols that are significantly associated with a protective effect against colorectal cancer. This may circumstantially suggest that other dietary factors in fruit and vegetables in addition to flavonols have important roles in preventing the pathogenesis of the disease. Further investigation of the effect of flavonols from different fruit and vegetables and their processed products is required to determine whether the observed association is due to flavonols per se or to the other as yet unidentified components of fruit and vegetables which are co-associated with flavonoids. Intervention studies comparing the effects of individual flavonoids with flavonoid-rich diets may be required to elucidate whether the main protective effects are actually due to these phytochemicals.
Acknowledgements
The original case–control study was funded by a grant from the National Hospitals Trust. The flavonoid analysis was financially supported by the Scottish Executive Environment and Rural Affairs Department and the Food Standards Agency.
J. A. M. K. led the development of the flavonoid database and analysis, contributed to the interpretation of results and drafted the manuscript. L. S. and J. L. were the principal investigators of the case–control study, contributed to the statistical analysis and interpretation of the results, and contributed to the paper. G. G. D. and G. McN. supervised the flavonoid analysis, and contributed to the interpretation of the results and the paper. We thank Graeme McHardy, Diane Thom, Seonaidh Cotton, Nigel Brockton and David Grubb for assistance with development of study procedures, data collection and analysis.
None of the authors has any conflict of interest.
