|Home | About | Journals | Submit | Contact Us | Français|
Fish, vitamin D, flavonoids, and flavonoid-containing foods may have cardiovascular benefits and therefore may also reduce the risk of renal cell cancer. Risk was prospectively assessed in the Alpha-Tocopherol Beta-Carotene Cancer Prevention Study (1985–2002) cohort (N=27,111; 15.2 mean person-years of follow-up). At enrollment, demographic, health, and dietary history information was recorded. Individuals who smoked less than 5 cigarettes/day, with chronic renal insufficiency or prior cancer, were excluded. Hazard ratios and 95% confidence intervals from Cox regression were used to compare upper quartiles (quartiles 2–4) with the lowest quartile (quartile 1) of dietary intake. Among 228 cases, risk (quartile 4 vs. quartile 1) was associated with consumption of the flavonoid quercetin (hazard ratio=0.6, 95% confidence interval: 0.4, 0.9; Ptrend=0.015) and Baltic herring (hazard ratio=2.0, 95% confidence interval: 1.4, 3.0; Ptrend<0.001), with adjustment for age, body mass index, smoking, blood pressure, alcohol use, physical activity, urban residence, and education. In geographically stratified models, the risks associated with herring and total fish intake appeared to be highest in the urban coast region, although the interaction was not statistically significant. These results suggest that the flavonoid quercetin may prevent renal cell cancer among male smokers. The possible risk associated with fish intake warrants further investigation before conclusions may be drawn.
Renal cell cancer is the predominant form of cancer of the kidney (>80%) and one of the most rapidly increasing cancers worldwide. The incidence has risen for both early and late stage cancers, suggesting that incidental detection, through the increased adoption of clinical imaging technology, does not fully account for this increase (1, 2). Internationally, some of the highest incidence rates of renal cell cancer are known to occur in Scandinavia and Eastern Europe. In some countries, the incidence has increased approximately 2-fold in the last 3 decades (3–5).
Renal cell cancer shares known risk factors with cardiovascular disease, including hypertension, tobacco use, and obesity, that account for approximately 50% of all renal cell cancer diagnoses (2). A number of dietary factors, including vitamin D and flavonoids, have demonstrated possible cardiovascular benefits (6–8). Fish are an important source of vitamin D, essential fatty acids, and other micronutrients associated with cardiovascular benefits, including a reduced risk of hypertension (9). Consistent with a hypothesized protective effect of vitamin D, higher incidence has been reported among latitudes located farther away from the Earth's equator, possibly related to the total yearly exposure to ultraviolet B radiation (4, 10, 11).
Therefore, this study examines the association between intakes of fish, vitamin D, and flavonoids and renal cell cancer risk in a cohort of smokers in Finland.
Renal cell cancer cases (International Classification of Diseases, Ninth Revision, code 189.0) were identified through the Alpha-Tocopherol Beta-Carotene Cancer Prevention (ATBC) Study, a randomized, double-blind, placebo-controlled primary cancer prevention trial designed to determine the efficacy of alpha-tocopherol, beta-carotene, or both in preventing lung cancer among male smokers. As described elsewhere, the ATBC Study included men aged 50–69 years who smoked 5 or more cigarettes per day and were recruited from 14 study centers in southwestern Finland between 1985 and 1988 (12). Participants were excluded at the time of study entry if they had a history of cancer (other than nonmelanoma skin cancer or carcinoma in situ), severe angina upon exertion, chronic renal insufficiency, cirrhosis of the liver, alcoholism, anticoagulant use, vitamin E (>20 mg/day), vitamin A (>20,000 IU/day), beta-carotene (>6 mg/day), or other medical problems perceived to limit participation over the 6-year period of the intervention. Incidental renal cell cancer cases, diagnosed by April 2002, were identified through the Finnish Cancer Registry that provides nearly 100% case ascertainment (13). For cases diagnosed through April 1999, the medical records were reviewed centrally by 1 or 2 study physician(s) for diagnostic confirmation. Cases diagnosed after April 1999 had only the Finnish Cancer Registry data for site, histology, and date of diagnosis.
At baseline, study subjects were measured for height, weight, and resting blood pressure, and a 12-hour fasting blood sample was collected. Participants completed a questionnaire containing information on supplement use, medical history, leisure time physical activity, and pack-years of smoking. Serum levels of alpha-tocopherol, beta-carotene, and retinol were analyzed by high-performance liquid chromatography (14). Serum cholesterol (high density lipoprotein cholesterol and total cholesterol) was determined enzymatically by using the cholesterol oxidase/p-aminophenazone method (Boehringer Mannheim, Mannheim, Germany).
At baseline, food consumption over the previous 12 months was assessed with a validated, self-administered food frequency questionnaire, including both portion size and frequency of consumption for 203 food items and 73 mixed dishes, that was designed specifically for the ATBC Study (15, 16). Validation studies include vitamin D and flavonoid intake, in addition to several other macro- and micronutrients. A color-picture booklet was used to aid in portion size estimation. Fish intake questions included frozen fish, rainbow trout, Baltic herring, other fresh fish, and canned/salted fish. Total fish consumption was calculated as the sum of these fish intake subgroups. Food intakes of vitamin D (μg/day), flavonoids (μg/day), and other micronutrients and macronutrients were estimated according to food composition data from the National Public Health Institute of Finland (17).
Cumulative lifetime tobacco smoking exposure was calculated as the number of pack-years of smoking (pack-years=total cigarettes per day/20 × total number of years’ regularly smoking). Because the major outcome of the ATBC Study was lung cancer, participants were asked to report a history (ever/never) of employment in occupations associated with lung cancer, including mining, colliery, quarrying, stonemasonry, stonecutting, foundry work, asbestos work, lead refining, nickel refining, copper smelting, steel production, oil refining, gas manufacture, chromium pigments, and arsenic production. We included this information in our study as some of these occupations have been associated with renal cell cancer and possibly confound dietary results (2).
Place of residence was defined according to the district of residence at the time of recruitment into the study. Coastal residence (yes/no) was defined as residence in a city, town, or municipality the boundary of which extended to the Baltic coast. Urban residence (yes/no) was defined as residence in a city with a population greater than 50,000 according to the Finnish census at the time of recruitment into the study (Statistics Finland; http://www.stat.fi).
All dietary variables were energy adjusted according to the residual method of Willett et al., as previously described (15), and grouped into quartiles based on the intake distribution in the entire cohort. Because dietary supplements and tea, nuts, and wine were used by a relatively small proportion of the cohort, their intakes could not be divided into quartiles (i.e., median intake=0), and intake levels were not adjusted by kcal but were divided into tertiles among only the individuals reporting intake. Similarly, for dietary items with greater than 25% of the cohort reporting no intake, tertiles of the kcal-adjusted values among only those reporting intake were created and assessed with respect to the risk. Pearson's correlation coefficient was used to estimate the correlation between kcal-adjusted intakes of dietary vitamin D and vitamin D-containing foods and between total flavonoids and flavonoid-containing foods. Milk was not included in the correlation analysis, because milk was not fortified with vitamin D in Finland during the time period of this study.
Cox proportional hazards regression was used to assess the risk of renal cell cancer associated with fish, vitamin D, and flavonoids (18). Each variable was first assessed in models adjusting for age and randomization group (alpha tocopherol, beta carotene, both, or placebo). Dietary variables included intake of vitamin D (total and supplement), total flavonoids, individual flavonoids (catechin, epicatechin, kaemferol, luteolin, quercetin, and myricetin), foods known to contain vitamin D, and foods known to contain flavonoids. Because production of the active form of vitamin D within the kidney is related to dietary calcium availability, total dietary calcium (diet plus supplements) was also included in the analysis.
Person-years were counted from the date of randomization until diagnosis of renal cell cancer, death, or end of the study (April 30, 2002). Person-years continued to accrue for participants whether or not they had dropped out of the primary ATBC Study intervention. Tests for trend by quartile or tertile were calculated by regressing the median value of each quartile or tertile as a continuous variable in multivariate Cox models. For trends in categorical variables, including physical activity and education, the ordinal rank of the category was used, if applicable.
Multivariate Cox proportional hazards regression models were constructed by stepwise selection and backward variable elimination techniques (SAS, version 9.1, software; SAS Institute, Inc., Cary, North Carolina). Selection models included age, body mass index, blood pressure, smoking (pack-years), leisure time physical activity, urban residence, educational level, and dietary variables that demonstrated an association with Ptrend ≤0.15 in age- and randomization group-adjusted models. In order to avoid the possible influence of extreme values of any one variable in the model selection process, 2 separate sets of models were run that included the following: 1) each quartile of dietary intake and 2) the median value of each quartile. The consistencies of variables remaining from backward and stepwise selection were compared. Variables selected for the final model included all variables demonstrating a statistically significant association (P<0.05) in both stepwise selection and backward variable elimination models. Age and pack-years of smoking were retained in the final model because of their biologic significance related to renal cell cancer. Education and urban residence were also included in the final model in order to adjust for possible residual effects of socioeconomic status related to other dietary risk factors.
Multiplicative interactions in the final model were tested by multiplication of the 2 variables in question and then including the main effect and interaction terms in the final model. In order to investigate regional differences in risk, we stratified the final model according to the region of residence at the time of enrollment into the study: urban inland (urban/noncoastal), rural inland (nonurban/noncoastal), urban Baltic coast (urban/coastal), and rural Baltic coast (nonurban/coastal), using the geographic definitions previously described. For all hazard ratios, 2-sided P values and 95% confidence intervals are reported. In order to protect confidentiality, tabular counts of less than 5 are suppressed.
There were 228 renal cell cancer cases diagnosed among 27,111 men with dietary information available, comprising 93% of the entire cohort of 29,133 participants. The mean and median follow-up times among the 27,111 men were both equal to 15.2 years. The response rate to enrollment in the primary intervention study was 79.3% (42,947/54,171) among smokers in the target population of southern Finland, among whom 67.8% (29,133/42,947) met the inclusion criteria. Drop-out rates (including from deaths) within the randomized groups did not differ significantly at the end of the intervention period (30.1%–31.3% range) (data not shown).
Vitamin D intake values were correlated with total fish (r=0.45), Baltic herring (r=0.28), rainbow trout (r = 0.22), margarine (r=0.22), and egg dishes (r=0.12) (data not shown). Flavonoid intake values were correlated with tea (r=0.93), apples and pears (r=0.29), and berries (r = 0.21), with lower correlations for peanuts (r=0.07) and wine (r = 0.08). All correlations were statistically significant (P<0.05).
There was no effect of the alpha-tocopherol or beta-carotene intervention on the incidence of renal cell cancer, and serum levels of alpha-tocopherol and beta-carotene were not associated with risk (data not shown). Body mass index, pack-years of smoking, systolic blood pressure, alcohol intake, leisure time physical activity, urban residence, and dietary intake of quercetin, vitamin D (dietary and total), and Baltic herring were significantly associated with renal cell cancer risk in models adjusting for age and randomization group (Table 1). There was no statistically significant association with total flavonoids, other flavonoids, other vitamin D-containing foods, total fish intake, or other types of fish (rainbow trout, frozen, fresh, and salted/canned). Ptrend values were statistically significant for body mass index, systolic blood pressure, alcohol intake, and intakes of quercetin, vitamin D (dietary and total), and Baltic herring. In addition to the statistically significant variables above, variables with Ptrend<0.15 that were included in the stepwise selection and backward variable elimination models included catechin, total calcium, frozen fish, and liver dishes.
In stepwise selection and backward variable elimination models, statistically significant variables (P<0.05) included body mass index, systolic blood pressure, physical activity, and intakes of alcohol, quercetin, and Baltic herring.
In the final model (Table 2), there was a 40% decreased risk among individuals consuming the highest amounts of quercetin (for individuals consuming >9.1 mg/day compared with individuals consuming ≤4.8 mg/day: hazard ratio (HR)=0.6, 95% confidence interval (CI): 0.4, 0.9), and the trend was statistically significant (Ptrend=0.015). These results were similar among individuals above (>26 kg/m2) and below (<26 kg/m2) the median body mass index (Pinteraction=0.613), although the trend was statistically significant only among those with values above the median body mass index (Ptrend=0.045).
Conversely, there was a 2-fold increased risk of renal cell cancer among individuals consuming, on average, 12.2 g or more per day of Baltic herring (HR=2.0, 95% CI: 1.4, 3.0), compared with individuals consuming 2.1 g or less, and the trend was statistically significant (Ptrend<0.001). Among individuals whose body mass index was below the median (<26 kg/m2), intakes at the third and fourth quartiles were associated with a 2.0- and a 2.7-fold elevated risk (Ptrend<0.001), whereas the same intake levels among individuals whose body mass index was above the median (>26 kg/m2) were associated with a 1.7- and a 1.6-fold elevated risk, respectively (Ptrend=0.090). The interaction was not statistically significant (Pinteraction=0.110).
In other models classifying those individuals with both lowest body mass index (≤23.7 kg/m2) and lowest herring intake (≤2.1 g/day) as the reference group, the risk for renal cell cancer ranged from 2.3- to 4.3-fold higher among individuals with both higher body mass index (quartiles 2–4) and highest Baltic herring intake (>12.2 g/day) (Figure 1). (Statistically significant hazard ratios are indicated by solid shapes in the figure.)
There was no association between vitamin D intake (total or dietary) and renal cell cancer with adjustment for the effects of the other variables in the final model (for the highest vs. the lowest quartile of total vitamin D intake: HR=1.0, 95% CI: 0.7, 1.4) (data not shown).
There were no statistically significant interactions between place of residence (urban, coastal, or combinations thereof) and fish intake (data not shown), but there was a suggestion of increased risk with higher consumption of Baltic herring, rainbow trout, and total fish intake, especially in the urban Baltic coast (Table 3).
This study observed a decreased risk of renal cell cancer associated with increased dietary intake of the flavonoid quercetin. Elevated renal cell cancer risk was observed with greater intake of Baltic herring in the urban Baltic coast region.
We found a suggestion that fish consumption was associated with an increased risk of renal cell cancer, mainly due to the consumption of Baltic herring which made up one fifth of the total fish consumption (17, 19, 20). We also found that this association appeared to be more pronounced among individuals with both high herring intake and higher body mass index, compared with individuals with both low herring intake and low body mass index. However, there was not statistical evidence of an interaction between fish intake and body mass index. Other studies have not reported an interaction with body mass index. A study of women in Sweden by Wolk et al. (21) reports a reduced risk of renal cell cancer with higher fatty fish consumption (i.e., salmon, herring, sardines, and mackerel), although the risks associated with specific fish types are not reported. Another study by Bravi et al. (22), including both men and women, reports no association with fish intake. If fish consumption is a significant factor in reducing or increasing renal cell cancer risk, it is possible that the effects are sex specific or that there are other aspects of fish consumption that are important to elucidate.
Fish consumption is positively correlated with blood levels of polychlorinated biphenyls, dioxin, cadmium, lead, and mercury (23–35). Cadmium, mercury, and lead are known nephrotoxicants that induce oxidative stress and damage to the proximal renal tubule, the location where nearly all renal cell cancers arise (36–39). Cadmium has a long half-life (10–30 years in kidney tissues) and has been associated with renal cell cancer in previous epidemiologic studies (40–44). Mercury induces renal tumors in male but not female mice, although it is not clear that this result translates directly to gender differences among humans (45, 46). Notably, it has been hypothesized that an increasing cadmium uptake in plankton and herring (~5.2 μg/g/year) in the aquatic ecosystem could be related to decreasing Baltic salinity and global climate change (47–49).
Our results, showing elevated risks associated with fish intake within the urban Baltic coast region, suggest that place of residence may modify risks related to fish intake, although studies with a larger number of cases are needed to identify an interaction. Elevated risks of cancer of the stomach, breast, and skin, as well as of multiple myeloma, and cardiovascular mortality have been reported among individuals with high levels of Baltic fish intake (26, 50–53). It is still possible, however, that individuals consuming high amounts of Baltic fish may have other characteristics that place them at high risk of renal cell cancer. This study adjusted for the major risk factors known to account for approximately half of all renal cell cancer cases (hypertension, smoking, and body mass index), as well as education and place of residence, which reduces the likelihood somewhat that the observed association is due to an unknown confounding factor. Future epidemiologic studies incorporating biomarkers of exposure to contaminants present in fish will be important.
The active form of vitamin D (1,25-dihydroxyvitamin D3) has been shown to inhibit growth of renal cell cancer cell lines and to increase survival in animal models of renal cell cancer (9–11). There is also some epidemiologic evidence that intake of vitamin D may reduce the risk of renal cell cancer (54). We did not observe a decreased risk of renal cell cancer in association with dietary vitamin D intake nor with vitamin D supplement use (including vitamin D from multivitamins) among smokers. In contrast, our initial risk estimates suggested an increased risk associated with vitamin D intake. However, vitamin D intake was not statistically significant in the final model, suggesting that our initial results are possibly due to confounding by other factors such as Baltic herring consumption (r=0.28). A nested case-control study within the same (ATBC Study) cohort reports an increased risk of pancreatic cancer with higher serum vitamin D levels, although it is not known whether higher serum levels were directly related to Baltic herring consumption (55). Because there were few participants taking supplemental vitamin D, it is likely that this study did not have sufficient statistical power to fully assess this association.
Toxicity related to high-dose vitamin D includes kidney stone formation (nephrolithiasis), which has been associated with renal cell cancer in some case-control studies, although not consistently (56–58). The Women's Health Initiative reported a 17% increased incidence of nephrolithiasis among participants receiving the recommended daily dose of vitamin D (10 μg or 400 IU/day of vitamin D3) and calcium (1,000 mg/day), compared with participants receiving placebo (59–61), although in the ATBC Study cohort, kidney stone incidence was not associated with vitamin D intake (62). The results of this study do not support a preventative association with dietary vitamin D intake among men who smoke.
Higher quercetin intake was associated with a significantly reduced risk of renal cell cancer. We did not observe a significant association for all flavonoids combined, of which quercetin comprises about 75%, nor for other flavonoids including catechin, epicatechin, kaemferol, luteolin, or myricetin. In the validation of the ATBC Study instrument, flavonoid intakes correlated reasonably well with food records (correlation statistics ranging from 0.66 to 0.76), with the exception of luteolin (r=0.43) (16). It is possible that other flavonoids may have an association with renal cell cancer, but we were unable to identify an association because of misclassification error. For example, risk estimates for catechin were reduced but not statistically significant in the highest intake quartile in the model adjusting for age and randomization group alone, although the Ptrend value was close to 0.05. In addition, if the dietary intake of quercetin were randomly misclassified among study participants (i.e., nondifferential misclassification), this suggests that the reduced risk of renal cell cancer associated with quercetin may be more pronounced when using a more accurate and reliable estimate of intake.
A previous study in the ATBC Study cohort reported an inverse relation with total flavonoid intake, although this result was not statistically significant, and individual flavonoid compounds were not investigated (63). At least one additional study reports a reduced risk associated with dietary flavones and flavonols (64). Toxicologically, high dose quercetin can produce chronic nephropathy and adenocarcinoma of the renal tubule in male F344 rats, although it is not known whether these results can be extrapolated to humans (65, 66). Quercetin is found in a variety of fruits and vegetables—including onions, apples, tea, and red grapes—and has been associated with reduced risk of cardiovascular disease in human studies (67). Case-control studies in humans have identified a reduced risk of renal cell cancer associated with dietary onion (68) and wine (69–71) consumption, although this has not been observed in all studies (72, 73). In clinical trials, flavonoid-rich foods have been associated with lower blood pressure (74), although one study reports that supplementation with quercetin did not demonstrate significant reductions in low density lipoprotein cholesterol, triglycerides, platelet aggregation, or blood pressure (75). Unlike with several other cancers, there is no clear evidence of an association between diets high in fruits and vegetables and a reduced risk of renal cell cancer (72, 76, 77).
The strengths of this study include the assessment of risk in a large, representative, longitudinal cohort study of smokers in whom exposure and biologic measures were assessed prior to cancer diagnosis, a high rate (approximately 100%) of cancer follow-up, and confirmation of cancer diagnoses through review of medical records by 1 or 2 study physician(s). This study used a comprehensive dietary assessment method that included a color picture booklet of the portion size.
This study is limited because the results may not be generalizable to nonsmoking populations or to women. In addition, although the response rate among smokers in southern Finland was nearly 80%, the ATBC Study is a controlled clinical trial, and therefore the study population is not a random sample of male smokers but comprised individuals willing and able to participate in a long-term lung cancer prevention trial. Further, each individual's exposure status was estimated at baseline, and changes over time were not considered. We did not have information on sunlight exposure, which can profoundly impact serum vitamin D levels (10). At the same time, dietary intake, through supplements and foods like fish, provides important sources of vitamin D, particularly in regions like Finland that have less ultraviolet B radiation year round. Compared with food records, the dietary instrument may underestimate vitamin D intake by approximately 12% (15), although the average vitamin D intake in this study is slightly higher than in other Finnish surveys conducted during a similar time period (20, 78). In addition, our study could not account for food preparation methods, including pickling, although pickled foods are not known to increase renal cell cancer risk (2). Finally, this study did not observe a significant trend with tobacco use, possibly because, unlike other studies, nonsmokers were not in the comparison group and possibly because of decreased tobacco use during the course of the study. At the time of enrollment, study participants were advised of the harmful effects of smoking, and more than 6,100 participants reported having stopped smoking during the intervention period (12, 79).
This study suggests that dietary intake of the flavonoid quercetin reduces the risk of renal cell cancer among men who smoke. Consumption of certain fish may be associated with increased renal cell cancer risk. Future studies are warranted, particularly those that investigate the potential synergy among fish intake, place of residence, and body mass index.
Author affiliations: Epidemiology Division, Department of Public Health Sciences, Pennsylvania State University College of Medicine, Hershey, Pennsylvania (Robin Taylor Wilson, Jiangyue Wang, Vernon Chinchilli, John P. Richie); Department of Chronic Disease Prevention, National Institute for Health and Welfare, Helsinki, Finland (Jarmo Virtamo); and Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, Maryland (Lee E. Moore, Demetrius Albanes).
This research was supported in part by the Intramural Research Program of the National Institutes of Health, by a postdoctoral research fellowship within the Division of Cancer Epidemiology and Genetics of the National Cancer Institute, and by salary support from the Pennsylvania State University. Additionally, this research was supported by US Public Health Service contracts N01-CN-45165, N01-RC-45035, and N01-RC-37004 from the National Cancer Institute, Department of Health and Human Services.
Special thanks to Dr. Stephanie Weinstein of the National Cancer Institute and Rusty Shields of Information Management Services, Inc., for resourceful technical and administrative support. The authors also thank Diane Pague, Kimberly Newman, and Kristen Brannen of the Pennsylvania State University for their assistance in the preparation of this manuscript.
Conflict of interest: none declared.