| |
Onions Reduce Pancreatic Cancer Risk: Flavonols and Pancreatic Cancer Risk
|
| |
| |
The Multiethnic Cohort Study
Ute Nothlings1,2, Suzanne P. Murphy1, Lynne R. Wilkens1, Brian E. Henderson3 and Laurence N. Kolonel1
1 Cancer Research Center of Hawaii, University of Hawaii, Honolulu, HI
2 Department of Epidemiology, German Institute of Human Nutrition Potsdam-Rehbruecke, Nuthetal, Germany
3 Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA
"....pancreatic cancer is one of the most fatal cancers.. Because of a 5-year survival rate of less than 5 percent ......In conclusion, our study provides evidence for a pancreatic cancer preventive effect of flavonols, especially for current smokers..... We observed the largest effect of flavonols among smokers. Smokers have about a twofold increased pancreatic cancer risk....smoking increases oxidative stress, and smoke contains a large quantity of carcinogens.........Major sources of flavonols are onions, curly kale, leeks, broccoli, apples, and blueberries..... (in this study) Intake of onions (p trend = 0.057) and black tea (p trend = 0.070), but not apples (p trend = 0.936) or green, herbal, or other tea (p trend = 0.807), showed modest inverse associations with pancreatic cancer (data not shown..........However, because of the reduced sample size (435 pancreatic cancer cases), none of the associations was statistically significant........ .... A family history of pancreatic cancer, a diagnosis of diabetes mellitus, and obesity have also been associated with the disease..... Flavonols are ubiquitous in plants, but they are present in the highest concentrations in onions, tea, apples, berries, kale, and broccoli (2, 4, 5, 38). Quercetin is most abundant in onions and apples....the highest correlation for quercetin was with apples. Other studies have found correlations of the same magnitude between intakes of flavonols and intakes of fruits or vegetables..."
ABSTRACT
Only a few prospective studies have investigated flavonols as risk factors for cancer, none of which has included pancreatic cancer. The latter is usually fatal, rendering knowledge about prevention particularly important. The authors estimated intakes of three flavonols-quercetin, kaempferol, and myricetin-for 183,518 participants in the Multiethnic Cohort Study and examined associations with incidence of pancreatic cancer.
Baseline data were collected in Hawaii and California in 1993-1996. Diet was assessed by using a quantitative food frequency questionnaire. During 8 years of follow-up, 529 cases of exocrine pancreatic cancer occurred.
Multivariate Cox regression models were calculated to estimate relative risks. Intake of total flavonols was associated with a reduced pancreatic cancer risk (relative risk for the highest vs. lowest quintile = 0.77, 95% confidence interval: 0.58, 1.03; p trend = 0.046).
Of the three individual flavonols, kaempferol was associated with the largest risk reduction (relative risk = 0.78, 95% confidence interval: 0.58, 1.05; p trend = 0.017).
Total flavonols, quercetin, kaempferol, and myricetin were all associated with a significant inverse trend among current smokers (relative risks for the highest vs. lowest quartile = 0.41, 0.55, 0.27, 0.55, respectively) but not never or former smokers. This study provides evidence for a preventive effect of flavonols on pancreatic cancer, particularly for current smokers.
INTRODUCTION
Flavonols are a class of flavonoids, polyphenols, which are ubiquitous in plant foods and may exert cancer preventive effects (1, 2). (3). Major sources of flavonols are onions, curly kale, leeks, broccoli, apples, and blueberries (3). Other classes of flavonoids include flavanols, anthocyanins, flavanones, and isoflavones (2). Anticarcinogenic effects in animals and in vitro studies have been attributed to the ability of these constituents to inhibit the cell cycle, cell proliferation, and oxidative stress and to induce detoxification enzymes and apoptosis (4-6). However, evidence from human population studies is scarce (2) and has been hampered primarily by a lack of coverage of flavonols, as well as flavonoids in general, in food composition databases (7).
Recent reviews of epidemiologic studies concluded that the evidence was limited for an inverse association between polyphenol intake and cancer risk (8, 9). The most consistent inverse association was found between flavonols, especially quercetin, and lung cancer. None of the studies with flavonols included pancreatic cancer. Because of a 5-year survival rate of less than 5 percent, pancreatic cancer is one of the most fatal cancers, and research into disease prevention is greatly needed (10). Smoking is the only established risk factor for pancreatic cancer so far. A family history of pancreatic cancer, a diagnosis of diabetes mellitus, and obesity have also been associated with the disease. Other risk factors include age, male sex, and Native Hawaiian or African-American ethnicity (11-15).
We examined data from the large Multiethnic Cohort to investigate flavonol intake as a risk factor for pancreatic cancer during 8 years of follow-up. For this analysis, we used a food composition table developed specifically for the multiethnic study population. Many of the flavonol values in the food composition table were obtained by analyses of local foods (16).
RESULTS
Intakes of total flavonols, quercetin, kaempferol, and myricetin in the Multiethnic Cohort Study are shown in table 1. Intakes were higher in women than in men. Quercetin contributed most to total flavonol intake (about 70 percent), followed by kaempferol (about 25 percent). African Americans had the lowest flavonol intake and Latinos the highest.
Because flavonols are ubiquitous in plant foods, we used correlation coefficients between flavonol intake and selected foods and food groups to better understand food sources (table 2). Overall, correlation coefficients were moderate in size and differed across foods for the individual flavonols.
In the lowest quintile of flavonol intake, 62 percent of study participants were men, compared with only 29 percent in the highest quintile of intake (table 3). The percentage of current and former smokers was also larger in the lowest quintile of flavonol intake than in the higher quintiles. When we compared ethnicities, the largest group in the fifth quintile of flavonol intake were Japanese Americans; they had the highest intake of flavonols relative to their energy intakes. Intake of processed meat and red meat, to a lesser extent, was inversely associated with flavonol intake.
In a multivariate model, intake of total flavonols was associated with a reduced pancreatic cancer risk (relative risk comparing extreme quintiles = 0.77), with a statistically significant trend (p = 0.046) (table 4). Regarding intake of the individual flavonols, kaempferol was associated with the largest risk reduction, with a relative risk comparing extreme quintiles of 0.78 and a statistically significant trend (p = 0.017). Stratification by study area confirmed inverse associations for Hawaii and Los Angeles, suggesting that the food analysis of flavonols applies to both regions. Intake of onions (p trend = 0.057) and black tea (p trend = 0.070), but not apples (p trend = 0.936) or green, herbal, or other tea (p trend = 0.807), showed modest inverse associations with pancreatic cancer (data not shown). When we excluded cases of pancreatic cancer that occurred in the first 2 years of follow-up to eliminate possible cases with preclinical disease at baseline, relative risks for flavonols generally showed the same patterns (data not shown). However, because of the reduced sample size (435 pancreatic cancer cases), none of the associations was statistically significant.
We also stratified our analyses by sex, ethnicity, and smoking status (table 5). Relative risks for total and single flavonol intakes were lower for women (0.72-0.79) than for men (0.85-0.97) when we compared extreme quartiles, although differences between the sexes were not statistically significant. The associations with kaempferol and myricetin showed statistically significant decreasing trends among women (p for trend = 0.016 and p for trend = 0.025, respectively).
Risks were examined across three ethnicities (African Americans, Japanese Americans, and Caucasians). The numbers of cases among Latinos (n = 85) and Native Hawaiians (n = 43) were too small for meaningful analyses. Interactions between flavonol intake and ethnicity were not statistically significant, indicating that the risk reduction was present for all groups. Indeed, a reduced risk with increasing intake of flavonol was observed in each of the groups except for total flavonols in Caucasians. The only significant trend was seen with kaempferol in African Americans (table 5), with a reduction in pancreatic cancer risk of more than 50 percent in the highest compared with the lowest quartile of intake. Tests for trend for total flavonols, quercetin, and kaempferol were <0.10 among Japanese Americans and for myricetin among Caucasians.
Smoking is the most well established risk factor for pancreatic cancer. Intakes of total and all three single flavonols were inversely associated with pancreatic cancer risk among current smokers, and all of the trends were statistically significant (table 5). The largest risk reduction was seen with higher intake of kaempferol (relative risk = 0.27, p for trend < 0.0001). The relative risk for total flavonol intake was 0.41 when we compared extreme quartiles, with a monotonic inverse trend (p = 0.002). No statistically significant associations were seen among never or former smokers. The interaction between smoking and kaempferol intake was statistically significant at p = 0.007, whereas interactions with total flavonol and quercetin intakes were of borderline significance (p = 0.067 and p = 0.068, respectively). Although their risk was still somewhat elevated, current smokers with high flavonol intakes had relative risks similar to those for never smokers. The relative risks, compared with those for never smokers with low flavonol intake (at median or below), were 0.87 (95 percent confidence interval: 0.66, 1.14) for never smokers with high intake (>median), 1.99 (95 percent confidence interval: 1.46, 2.70) for current smokers with low intake, and 1.15 (95 percent confidence interval: 0.76, 1.72) for current smokers with high intake.
DISCUSSION
In this analysis of the Multiethnic Cohort Study, intake of flavonols was associated with a reduced pancreatic cancer risk. Kaempferol in particular showed slightly stronger inverse associations than quercetin or myricetin. The associations were strongest in high-risk groups, that is, current smokers and African Americans. Among current smokers, as opposed to never and former smokers, intakes of total and each of the single flavonols showed statistically significant inverse trends for pancreatic cancer risk, with a relative risk below 0.5 for total flavonols in the highest quartile. We previously reported inverse associations between dark green vegetables and pancreatic cancer in current smokers and African Americans (22), which might be explained by the flavonol contents of these vegetables. The moderate size of the correlation coefficients between flavonols and particular foods and food groups suggests that these results do not simply reflect an association with common vegetable groupings. Indeed, the results for vegetables were much weaker (22): the relative risks for the highest quintile of intake compared with the lowest were 0.86 (p for trend = 0.134) for total vegetables, 0.80 (p for trend = 0.081) for dark green vegetables, and 0.83 (p for trend = 0.156) for cruciferous vegetables. When we evaluated models including both total flavonols and total vegetables, dark green vegetables, or cruciferous vegetable intakes, we found that associations with vegetable intakes remained nonsignificant (relative risks for the highest vs. the lowest tertiles = 0.91 (p for trend = 0.50), 0.96 (p for trend = 0.51), and 0.98 (p for trend = 0.81)), while associations with total flavonol intake were only slightly changed (relative risks for the highest vs. the lowest tertile = 0.82 (p for trend = 0.08), 0.81 (p for trend = 0.07), and 0.80 (p for trend = 0.04) adjusted for total, dark green, and cruciferous vegetables, respectively). Because of correlations among these dietary components (table 2), an evaluation of their independent effects is difficult. However, the associations with flavonol intake appear to be stronger.
To our knowledge, no large epidemiologic study has yet analyzed the association between intakes of flavonols and pancreatic cancer risk. A prospective study that included only 29 pancreatic cancer cases found no association between intake of combined flavonoids, of which 95 percent was quercetin, and pancreatic cancer (23). The mean flavonoid intake in that study was 4.0 mg/day, considerably lower than values reported in other studies (24). Few and inconsistent data are available on associations between flavonol intake and cancers at other sites. Of three prospective (25-27) and five case-control (28-32) studies, three reported inverse associations for lung cancer, and one each for total, stomach, and prostate cancer. Lagiou et al. (32) reported a positive association between intake of flavonols and lung cancer risk among women in Greece. Other cancers investigated included colorectum, urinary tract, and breast; none of these studies reported significant inverse associations. Discrepancies between epidemiologic studies investigating health effects of flavonoids may be due to differences between subclasses of flavonoids. Thus, it may be important to examine intakes of flavonoid subclasses individually, rather than all flavonoids combined, as we did in this investigation of flavonols (33).
Some evidence for the potential of flavonols to decrease pancreatic cancer risk is available from animal studies and in vitro systems. For example, in an animal study, quercetin decreased primary pancreatic tumor growth, increased survival, and prevented metastasis (34). The authors suggested that the beneficial effects may be due to the enhancement of apoptosis, which was confirmed in in vitro studies. Lee et al. (35) showed that the blockade of epidermal growth factor receptor tyrosine kinase activity by quercetin led to growth inhibition and apoptosis of pancreatic tumor cells. However, addition of quercetin to the diet has also been shown to lead to severe dysplastic lesions, carcinoma in situ, and microcarcinoma in a model of rat pancreatic carcinogenesis (36).
We observed the largest effect of flavonols among smokers. Smokers have about a twofold increased pancreatic cancer risk, which we confirmed in the Multiethnic Cohort (22, 37). Smoking increases oxidative stress, and smoke contains a large quantity of carcinogens. Therefore, the actions of flavonols as antioxidants or as modulators of the expression of cytochrome P-450 enzymes involved in the activation of carcinogens may explain the larger risk reduction effect in smokers compared with nonsmokers (5). The relative risk of pancreatic cancer for current smokers with high flavonol intake remained higher than the relative risk for never smokers with low flavonol intake, although the difference was not statistically significant. At this point, this observation cannot be taken to suggest that the effects of smoking can be offset, since other characteristics in the group of smokers with high flavonol intake might account for the observed association. Additional evidence from other studies is needed.
Flavonols are ubiquitous in plants, but they are present in the highest concentrations in onions, tea, apples, berries, kale, and broccoli (2, 4, 5, 38). Quercetin is most abundant in onions and apples. Our correlations with flavonol intakes reflect this picture to some extent; for example, the highest correlation for quercetin was with apples. Other studies have found correlations of the same magnitude between intakes of flavonols and intakes of fruits or vegetables (32).
Few estimates of flavonol intake in populations are available, and reported intakes vary considerably between studies (3, 24). Flavonol intake by US adults has recently been estimated as 12.9 mg/day (39). Sampson et al. (40) estimated intake of flavonols and flavones to be 20-22 mg/day for men and women. In Finland, Knekt et al. (25) estimated mean intake of quercetin, kaempferol, and myricetin as 3.3 mg, 0.6 mg, and 0.12 mg, respectively, whereas researchers in the Netherlands estimated quercetin intake to be 16 mg/day (41). A recent review listed mean intakes of total flavonols for studies in the United States, United Kingdom, Korea, Japan, Germany, and the Netherlands at between 9.4 mg/day and 64 mg/day (3). The mean estimated intakes of quercetin (10 mg/day), kaempferol (3.9 mg/day), and myricetin (0.9 mg/day) in our cohort are within the range of previously reported intakes.
One reason for the large difference between estimates of flavonol intake might be the poor characterization of flavonols in existing tables of food composition. Few analytic data are available, and those published come predominantly from European laboratories. Different methods of analysis and variations in flavonol content by geography and plant cultivars may affect the accuracy of food composition tables (7, 16). However, differences across countries might also be real given diverse dietary patterns of flavonol consumption. The US Department of Agriculture has recently made available a literature-based food database of flavonoids, which will facilitate future studies of intakes (42). The flavonol values used in our study were based primarily on analyses of foods commonly consumed by cohort members and utilized up-to-date analytic methods (16).
A limitation to be considered when interpreting our study findings is that some degree of measurement error in the estimation of flavonol intake was certainly present in our data. However, a validation study of our food frequency questionnaire showed reasonably good performance (18). Furthermore, our use of flavonol densities rather than absolute values should have reduced the measurement error to some extent (43). However, to our knowledge, no validation study has been conducted to investigate a food frequency questionnaire's ability to reflect flavonol intake. Unlike many of the dietary components, flavonols are concentrated in specific foods rather than in broader food groups, for example, in apples rather than in all fruit. Thus, a food frequency questionnaire might be a better assessment instrument because shorter-term recalls and records would miss these single foods if they were not consumed on the interview day(s). Onions as a main source of flavonols are often consumed in mixed dishes and are not covered as a single item in the questionnaire. However, our quantitative food frequency questionnaire is likely to capture the major sources of flavonol-rich vegetables, such as onions, because we ask about consumption of many representative mixed dishes containing these vegetables, as well as about consumption of single vegetable items.
Our study has several strengths. Our food composition table included analyzed flavonol values, and the quantitative food frequency questionnaire has been especially designed for use in our multiethnic population. This combination is a major strength of the study since flavonol contents of foods can differ significantly by country (33). Furthermore, the prospective design of our study rules out recall bias, which is a major drawback in case-control studies, especially for pancreatic cancer where proxy interviews are usually necessary because of the rapid fatality of the disease.
In conclusion, our study provides evidence for a pancreatic cancer preventive effect of flavonols, especially for current smokers. Further epidemiologic studies in other populations and geographic regions are needed to confirm these findings.
|
|
| |
| |
|
|
|
