Although renal cell carcinoma (RCC) accounts for only 3 % of adult malignancies in the USA, its incidence has been increasing in the USA for the last 30 years, with annual increments of 1·6 % in white men and 1·7 % in white women. Thirty years ago, rates of renal cancer were 12 per 100 000 white men and 5 per 100 000 white women. Recent rates are reported as 18 per 100 000 white men and 9 per 100 000 white women(Reference Devesa, Silverman, McLaughlin, Brown, Connelly and Fraumeni1, Reference Moore, Wilson and Campleman2). The increase cannot be fully explained by early detection of pre-symptomatic tumours. The reported ongoing epidemic of obesity in the USA(Reference Flegal3) and/or the increase in hypertension(Reference Fields, Burt, Cutler, Hughes, Roccella and Sorlie4) and diabetes(Reference Lindblad, Chow, Chan, Bergström, Wolk, Gridley, Mclaughlin, Nyrén and Adami5) may explain part of this increase, which occurred despite a drop in smoking rates(6). Although obesity(Reference Bergstrom, Hsieh, Lindblad, Lu, Cook and Wolk7), hypertension(Reference Grossman, Messerli, Boyko and Goldbourt8) and diabetes(Reference Zucchetto, Dal Maso and Tavani9) have consistently been associated with RCC risk, few studies have tried to disentangle the effects of obesity from increased dietary intake and lack of physical activity(Reference Yuan, Gago-Dominguez, Castelao, Hankin, Ross and Yu10, Reference Hu, Mao and White11). An increase in lipid peroxidation may partially explain some of the reasons for RCC risk(Reference Gago-Dominguez, Castelao, Yuan, Ross and Yu12, Reference Gago-Dominguez and Castelao13). To evaluate the association between an ‘energy-dense’ diet and the risk of RCC and to understand the interrelationship between dietary intake of fatty foods and its correlates, we analysed RCC dietary data, along with other established and potential risk factors collected as part of a large population-based case–control study.
Material and methods
Study sample
A population-based case–control study of RCC and five other cancers was conducted in Iowa between 1986 and 1989. Detailed methods are reported elsewhere(Reference Cantor, Lynch and Johnson14, Reference Parker, Cerhan, Janney, Lynch and Cantor15). Briefly, eligible cases were residents of the state of Iowa, aged 40–85 years, newly diagnosed with histologically confirmed RCC (ICD-O code 189.0) in July 1985 to December 1987, and without previous diagnosis of a malignant neoplasm. Cases were identified by the State Health Registry of Iowa(Reference Ries, Harkins and Krapcho16). An introductory letter was followed by a telephone call in which potential participants were invited to complete a mailed questionnaire, designed either for direct respondents or their proxies, sent per request during the telephone contact. Of the 463 eligible RCC cases, the questionnaires were completed for 406 of them (87·7 % response rate). Among these, 287 subjects completed the questionnaire designed for direct respondents and 119 completed a proxy questionnaire. The early version of the direct respondent questionnaire, which did not include a question about possible proxy status, was completed by 81 of the 287 ‘direct questionnaire’ respondents. In the present analysis, these respondents were assumed to be the study subjects since almost all of the 206 respondents, who completed the later version of the direct respondent's questionnaire that asked about possible proxy status, were study subjects.
Controls were frequency matched to all cases in the overall study by sex and 5-year age group. Controls, like cases, had to be without previous diagnosis of a malignant neoplasm. Controls under 65 years of age were selected randomly from computerised state of Iowa driver's license records(Reference Lynch, Logsden-Sackett, Edwards and Cantor17), whereas controls aged 65 years and older were selected randomly from the lists of Iowa residents provided by the US Health Care Financing Administration (now the Centers for Medicare and Medicaid Services). Both sampling frames have been shown to achieve greater than 95 % coverage of the intended population(Reference Hartge, Cahill, West, Hauck, Austin, Silverman and Hoover18). Of the 999 eligible controls under the age of 65 years, 817 (82 %) participated by returning a completed questionnaire, and 1617 of 2036 eligible controls aged 65 years or older participated (79 %). Among the 2432 sent direct reminder questionnaires, control subjects 2064 were completed by the subject, 241 by a proxy and 127 by an undetermined respondent (assumed to be a direct respondent, as described earlier). Proxy questionnaires were sent to two control subjects.
Written informed consent was obtained from all participants. The study was approved by the Institutional Review Boards at the US National Cancer Institute and at the University of Iowa.
Data collection
Data were collected by means of a self-administered mailed questionnaire, supplemented by a telephone interview where necessary. The questionnaire included information on demographics, anthropometric measures (weight history and usual adult height), usual non-occupational physical activity, smoking history, occupational history, past medical history (including self-report of physician-diagnosed hypertension and history of bladder/kidney infection), history of cancer among first-degree relatives and other factors. BMI was calculated as (weight (in kilograms))/(height (in meters))2 and subjects were classified as normal (BMI < 25 kg/m2), overweight (BMI 25–29·9 kg/m2) or obese (BMI ≥ 30 kg/m2) when they were in their 20s, 40s and 60s.
Of the 2434 controls, 548 did not have sufficient dietary data for analysis. We did not exclude any subject for ‘extreme’ values on any dietary variable. Sixty-six controls were missing information on BMI and/or a history of hypertension. Of the 406 RCC cases, seventy three did not have sufficient dietary information and ten did not have BMI and/or hypertension information. These subjects were excluded, leaving 323 cases and 1820 controls for the dietary analysis. Most of the 548 controls and seventy three cases, who were excluded due to insufficient dietary information, had responded to a truncated telephone questionnaire that did not include diet.
Dietary analysis
Information on usual adult dietary intake was gathered with a food frequency questionnaire that asked about the number of times per d, week, month or year (or rarely/never) of consumption for each of fifty-five food items, excluding dietary changes in the previous couple of years. Using these data, we calculated the intake per common time period for each item. We then summed these data to derive the frequency of intake within each food group. Estimates of usual intake were derived for individual food items by multiplying the frequency of consumption of each item by an average serving size for males and females, separately, obtained from the National Health and Nutrition Examination Survey II (NHANES II)(Reference Dixon, Zimmerman, Kahle and Subar19, Reference Dresser20). Nutrients were then estimated by multiplying the intake of these foods by nutrient density estimates derived from the US Department of Agriculture (USDA) food composition tables(Reference Dresser20) and a USDA-National Cancer Institute (NCI) food composition database(Reference Dixon, Zimmerman, Kahle and Subar19). An adjustment for total food intake was carried out by the nutrient density method(Reference Hu, Stampfer, Rimm, Ascherio, Rosner, Spiegelman and Willett21). Each nutrient was individually divided by the subject's total energy intake before the quartiles of intake were calculated. Quartiles of dietary intake by food group or nutrient were calculated based on the distribution among controls. When nutrients were analysed, total energy consumption in kJ (continuous variable) was entered into a logistic regression model along with the other potential confounders. Two statistical packages were used: Statistical Package for the Social Sciences (version 11) and EPICURE(Reference Preston, Lubin, Pierce and McConney22).
Multiple logistic regression analysis was used to adjust for confounding by age (continuous), sex, smoking (in eight pack-year categories), proxy status of respondents (direct or proxy respondent), history of high blood pressure (yes, no), BMI at age 40, alcohol intake (as per recent cohort studies(Reference Greving, Lee, Wolk, Lukkein, Lindblad and Bergström23, Reference Lee, Hunter and Spiegelman24)) and vegetable consumption (Tables 1 and 2). Physical activity, fruit intake, education, family history of kidney cancer, coffee, tea consumption and history of kidney infection were found not to be risk factors and when added to the models of the present analysis had no added effect on RCC risk; thus they were not included as confounders in further analyses. The maximum-likelihood estimate of the OR, with 95 % CI, was used as the measure of association between either dietary fat-related food group variables or nutrients and RCC(Reference Breslow and Day25). Tests for trend across quartiles were performed by assigning the mean value of each respective quartile to the score variable, and then testing linear trend using a likelihood ratio test(Reference Breslow and Day25). To evaluate possible interaction on the association of risk with dietary fat by other established risk factors, we examined stratified models and also tested multiplicative interaction by the log-likelihood ratio test. For example, interaction (ORinteraction) between blood pressure and fatty spread consumption (continuous) was tested by the log-likelihood ratio test in a logistic model, with main effects adjusting for sex, age, proxy status, smoking, energy, BMI at age 40, alcohol intake and vegetable intake(Reference Breslow and Day25). We previously reported the joint effect of obesity and hypertension on RCC risk in these data(Reference Brock, Gridley, Lynch, Ershow and Cantor26).
Distribution of potential confounding factors (%) and their correlations by dietary fat consumption in controls of Iowa case–control study of renal cell carcinoma and diet
* Correlation was adjusted for energy.
Demographic and lifestyle risk factors: Iowa case–control study of renal cell carcinoma
* Adjusted for age, sex, proxy status, smoking, BMI at age 40, blood pressure, alcohol and vegetable consumption, where relevant.
Results
Table 1 presents the distribution between dietary fat consumption and covariates in the control population. This analysis was done both by percentage distribution by quartile of fat intake and also Pearson's correlation coefficients with dietary fat as a continuous variable, where relevant. The major associations in the controls with dietary fat were with age, proxy status, smoking, hypertension, alcohol consumption, coffee, physical activity, protein and carbohydrate intake and vegetable consumption. It should be noted that energy, as expected, was highly correlated with dietary fat consumption. We therefore present the correlations of the variables adjusted for energy.
In Table 2, smoking, increased BMI at age 40, age, a history of hypertension, low alcohol intake and low vegetable consumption were significant risk factors for RCC in our data. Compared with the controls, the cases were more likely to be current smokers (OR = 1·6 (95 % CI 1·1, 2·2)), obese at age 40 (OR = 1·9 (95 % CI 1·3, 2·9)), to report a history of hypertension (OR = 1·8 (95 % CI 1·4, 2·4)), not drink alcohol and consume vegetables at a low level. The cases were somewhat younger than the controls; thus age was included as a confounder (continuous) in subsequent analyses. Among the direct respondents, OR for smoking, obesity and hypertension and low alcohol and vegetable consumption followed patterns similar to those shown in Table 2 (data not shown). Thus, age, smoking, proxy status, obesity, hypertension and alcohol and vegetable consumption were included as confounders in subsequent analyses. In our data, neither physical activity, coffee/tea consumption nor fruit consumption remained as risk factors after adjustment for these confounders, and thus were not included as covariates in any models.
We compared energy and percentage of contribution of fat, protein and carbohydrate, by sex and case–control status, in our data with those in the NHANES II nutritional survey conducted contemporaneously(Reference Dresser20). This was done as no validation studies were available from 1986, and we wanted some indication of the generalisability of our data to the general US population at the time. The dietary composition of total energy and distribution of macronutrients among both male and female controls from the Iowa study was remarkably similar to those of the NHANES II study sample (i.e. men consumed approximately 8000 kJ/d, of which fat comprised almost 40 % and women consumed approximately 5550 kJ/d, of which fat comprised about 35 %; Appendix 1).
Table 3 presents analyses of fat-related food groups. High-fat spreads (e.g. mayonnaise, margarine, butter), red meat (bacon, breakfast sausage, beefsteaks, roasts, hamburgers, meat loaf, beef stew, pot pie, hot dogs, lunch meats, bratwurst, ham, pork, meat in pasta dishes), dairy foods (ice cream, cheese, milk) and cured meats (e.g. bacon, hot dogs) were found to be associated with a higher risk of RCC, with significant trends for high-fat spreads (P trend = 0·001), red meat (P trend = 0·01), dairy foods (P trend = 0·02) and cured meats (P trend = 0·02). Total meat consumption (data not shown) did not show this significant positive association with RCC risk. Subjects in the highest quartile (compared with the lowest quartile) of consumption of high-fat spreads, red meat, dairy foods and cured meat had significantly increased risks: OR = 2·0 (95 % CI 1·4, 3·0), OR = 1·7 (95 % CI 1·0, 2·2), OR = 1·6 (95 % CI 1·1, 2·3) and OR = 1·8 (95 % CI 1·2, 2·7), respectively. Similar risks were seen when analyses were limited to self-respondents (Table 3).
(Odds ratios and 95% confidence intervals)
* OR adjusted for age, sex, proxy status, smoking, BMI at age 40, blood pressure, alcohol and vegetable consumption in total population.
† OR adjusted for age, sex, proxy status, smoking, BMI at age 40, blood pressure, alcohol and vegetable consumption in direct respondents.
‡ High-fat spreads: butter/margarine and mayonnaise; red meat: bacon, breakfast sausage, beef (steaks, roasts, hamburgers, meat loaf), beef stew, pot pie, hot dogs, lunch meats, bratwurst, ham, pork, meat in pasta dishes; dairy: ice cream, cheese, cheese spread, cheese or cream in pasta dishes, whole and skimmed milk; cured meat: bacon, breakfast sausage, hot dogs, bratwurst, lunch meats.
Table 4 presents energy density nutrient values for the macronutrients. Total energy was not significantly associated with the risk of RCC (P trend = 0·31; top quartile v. bottom quartile of intake: OR = 1·3, 95 % CI 0·8, 2·0). There was no association between increased protein intake and RCC (OR = 1·2, 95 % CI 0·7, 1·6, high v. low quartile). The significantly reduced risks and trends for carbohydrate consumption disappeared when adjusted for fat intake (OR = 1·1, 95 % CI 0·6, 2·0, high v. low quartile; protein: OR = 0·7, 95 % CI 0·5, 1·2, high v. low quartile). However, increased fat intake was associated with significant risk of RCC (OR = 2·0 (95 % CI 1·3, 3·0), P trend = 0·001), even after adjustment for protein or carbohydrate intake. As was found for the food group associations, results from the analyses limited to direct respondents were similar to those from the total study sample.
(Odds ratios and 95% confidence intervals)
* OR adjusted for age, sex, proxy status, smoking, BMI at age 40, blood pressure, energy, alcohol and vegetable consumption.
† OR adjusted for age, sex, proxy status, smoking, BMI at age 40, blood pressure, energy, alcohol and vegetable consumption, and the model also includes percentage of energy from fat and carbohydrates.
‡ OR adjusted for age, sex, proxy status, smoking, BMI at age 40, blood pressure, energy, alcohol and vegetable consumption, and the model also includes percentage of energy from protein and carbohydrates.
§ OR adjusted for age, sex, proxy status, smoking, BMI at age 40, blood pressure, energy, alcohol and vegetable consumption, and the model also includes percentage of energy from fat and protein.
Table 5 presents results of the analyses for the types of fat nutrients, using energy density estimates for these fat nutrient variables. For saturated fat, animal fat, oleic acid and cholesterol, there were significant dose–response increases in the risk for RCC, with the risk increasing as the intake of each type of fat increased. Those in the highest quartile of each type of fat nutrient had a significant twofold risk: OR = 2·6 (95 % CI 1·6, 4·0, P trend < 0·001), OR = 1·9 (95 % CI 1·3, 2·9, P trend < 0·001), OR = 1·9 (95 % CI 1·2, 2·9, P trend = 0·01) and OR = 1·9 (95 % CI 1·3, 2·8, P trend = 0·006), respectively. Increasing intake of vegetable fat and polyunsaturated fat (linoleic acid) showed little association with RCC risk. Results from the analyses limited to direct respondents were similar to those from the total study sample. As the nutrients related to the types of fat were highly correlated with one another (Appendix 2), the individual fat-related nutrients were not further adjusted for each other, nor for protein or carbohydrate.
(Odds ratios and 95% confidence intervals)
* OR adjusted for sex, age, proxy status, smoking, BMI at age 40, hypertension, alcohol, vegetable consumption and energy.
† OR direct respondent analysis adjusted for sex, age, smoking, BMI at age 40, hypertension, alcohol, vegetable consumption and energy.
In an attempt to disentangle food group findings from energy-adjusted nutrient findings, we adjusted dairy intake for cholesterol (and vice versa). The risks for both dairy and cholesterol remained unchanged and significant. By contrast, the risk for cured meat was reduced by adjusting for cholesterol (but not vice versa; ORcured meat = 1·4 (95 % CI 0·9, 2·3) and ORcholesterol = 2·2 (95 % CI 1·3, 3·7), for the highest compared with the lowest intake quartile).
Table 6 shows the interaction between hypertension and high-fat spreads for the risk of RCC. As this interaction was interesting but only marginally significant for the high-fat spreads food group (P = 0·06), we also investigated the interaction between hypertension and the other fat-related food groups and nutrients. We found similar interactions that were only marginally significant for saturated fat and oleic acid (data not shown). When we investigated other potential interactions, none were found with age, sex, tobacco, BMI, alcohol intake or vegetable intake for the association between fat intake and RCC risk.
(Odds ratios and 95% confidence intervals)
* Interaction between blood pressure and high-fat spread consumption was tested by the likelihood ratio test in a logistic model adjusted for age (continuous), smoking, proxy status, sex, blood pressure, BMI at age 40, alcohol and vegetable consumption (ORinteraction). ORinteraction = 1·2 (95% CI 0·99, 1·6), P interaction = 0·06.
†OR adjusted for sex, age, proxy status, BMI at age 40, smoking, alcohol and vegetable intake.
Discussion
Results from this population-based case–control study provide evidence for a link between high dietary saturated fat, animal fat, oleic acid and cholesterol intake and an excess risk of RCC. In initial macronutrient analysis, once the effect of fat was taken into account, neither protein, carbohydrate nor total energy intake was significantly associated with RCC. Increased risks were associated with high-fat spreads, red and cured meats and dairy products. In both the fat-related food groups and nutrients, there was a significant dose–response with increased intake. Our data also indicated that the association of RCC with fatty foods may be stronger among individuals with hypertension.
Our findings of a significant effect of animal and saturated fat intake, cholesterol, high-fat spreads, dairy products and red and cured meat are consistent with indications from very early ecological observations noted both in the USA(Reference Muscat, Hoffmann and Wynder27) and internationally(Reference Armstrong and Doll28), where average national intake of animal products was significantly correlated with national RCC mortality in thirty-two countries (r 2 = 0·8). Our data showing an increased risk for selected fats in the diet are similar to those of a US case–control study, collected in a similar time period, which reported an OR of 2·2 (95 % CI 1·2, 3·9) for saturated fat, an OR = 1·8 (95 % CI 1·0, 3·1) for animal fat and yet little association with animal protein (OR = 1·3 (95 % CI 0·8, 2·3))(Reference Maclure and Willett29). An Italian case–control study (with hospital controls) reported a significant twofold association, similar to ours, between margarine and oils and RCC risk(Reference Talamini, Barón, Barra, Bidoli, La Vecchia, Negri, Serraino and Franceschi30). Out of the eleven case–control(Reference Muscat, Hoffmann and Wynder27, Reference Maclure and Willett29–Reference Lindblad, Wolk, Bergstrom and Adami38) and six cohort studies(Reference Wolk, Larsson, Johansson and Ekman39–Reference Washio, Mori and Sakauchi44), four medium-sized case–control studies (all but one with population controls)(Reference Maclure and Willett29, Reference Talamini, Barón, Barra, Bidoli, La Vecchia, Negri, Serraino and Franceschi30, Reference Mellemgaard, McLaughlin, Overvad and Olsen32, Reference Boeing, Schlehofer and Wahrendorf33) and one cohort study(Reference Rashidkhani, Akesson, Lindblad and Wolk40) reported a specific association of intake of some form of fat with RCC risk.
Similar to past case–control findings, our data show a stronger association with red meat than with total meat consumption. Eight of the case–control studies reported positive significant associations with high intakes of meat, some specifically with animal protein(Reference Maclure and Willett29, Reference Talamini, Barón, Barra, Bidoli, La Vecchia, Negri, Serraino and Franceschi30, Reference Handa and Kreiger34–Reference Chow, Gridley, McLaughlin, Mandel, Wacholder, Blot, Niwa and Fraumeni36), beef(Reference Hu, Mao and White11, Reference Maclure and Willett29, Reference Handa and Kreiger34), red meat(Reference De Stefani, Fierro, Mendilaharsu, Ronco, Larrinaga, Balbi, Alonso and Deneo-Pellegrini35, Reference Chow, Gridley, McLaughlin, Mandel, Wacholder, Blot, Niwa and Fraumeni36), fried meat(Reference De Stefani, Fierro, Mendilaharsu, Ronco, Larrinaga, Balbi, Alonso and Deneo-Pellegrini35, Reference Wolk, Gridley, Niwa, Lindblad, McCredie, Mellemgaard, Mandel, Wahrendorf, McLaughlin and Adami37), processed meat(Reference Hu, Mao and White11) and poultry(Reference Lindblad, Wolk, Bergstrom and Adami38). Our data also showed some association with cured/processed meat. In a pooled case–control study from four countries(Reference Wolk, Gridley, Niwa, Lindblad, McCredie, Mellemgaard, Mandel, Wahrendorf, McLaughlin and Adami37) and in a California study(Reference Yuan, Gago-Dominguez, Castelao, Hankin, Ross and Yu10), cured meat was not found to be a risk factor.
Our finding of selected types of dietary fat as the major nutrient associated with RCC risk is not in total accord with the few other studies that investigated the role of macronutrients, where either protein(Reference Talamini, Barón, Barra, Bidoli, La Vecchia, Negri, Serraino and Franceschi30, Reference Chow, Gridley, McLaughlin, Mandel, Wacholder, Blot, Niwa and Fraumeni36), fat(Reference Boeing, Schlehofer and Wahrendorf33) or total energy(Reference Mellemgaard, McLaughlin, Overvad and Olsen32) was determined to be a risk factor, after mutual adjustment. In an attempt to elucidate the macronutrient involved, case–control data from five countries were combined and a risk for RCC of 1·7 for the highest v. the lowest quartile of total energy intake was reported(Reference Wolk, Gridley, Niwa, Lindblad, McCredie, Mellemgaard, Mandel, Wahrendorf, McLaughlin and Adami37). In subsequent cohort studies, only one investigated macronutrients and a null effect was reported for total energy(Reference Rashidkhani, Akesson, Lindblad and Wolk40). Most cohort studies have shown little association of RCC risk with high-fat foods(Reference Wolk, Larsson, Johansson and Ekman39–Reference Nicodemus, Sweeney and Folsom43), except for a Japanese study in which ‘a fondness for fatty foods’ was associated with RCC risk(Reference Washio, Mori and Sakauchi44), although it must be noted that the numbers of RCC cases in these studies were small (n 14–122).
The disparity between case–control and cohort studies may have two origins. First, cohort studies to date have had very limited numbers of RCC cases as it is a rare cancer. Thus, they may not have had adequate power to detect an effect. Second, it is also possible that the effect we have observed could be due to recall bias. Only further investigations in large cohort studies or consortia will resolve this issue.
Several types of putative mechanisms may shed light on these findings. First, as diabetes may also be related to RCC risk, one explanation for an association with specific types of fat in the diet is that high insulin levels may increase the risk of RCC(Reference Lindblad, Chow, Chan, Bergström, Wolk, Gridley, Mclaughlin, Nyrén and Adami5, Reference Zucchetto, Dal Maso and Tavani9), as certain types of fat in the diet have previously been thought to be associated with high insulin levels and development of type 2 diabetes(Reference Zucchetto, Dal Maso and Tavani9), although these associations are now in question(Reference Lichtenstein and Schwab45). In animal models, insulin directly stimulates carcinogenesis and neoplastic differentiation by promoting DNA synthesis(Reference Lupulescu46). A second mechanism is hormonal, as animal studies indicate that the deposition of lipids in the kidney may be regulated by hormones and the kidney is rich in prolactin receptors(Reference Marshall, Bruni and Meites47). Thus, there is a possibility that fat intake, obesity, diabetes and hypertension could all be intermediate steps in a causal pathway to RCC. An overriding hypothesis that incorporates all these steps has recently been proposed as a ‘lipid peroxidation hypothesis’ to explain the associations of specific types of fats in the diet, obesity and hypertension with RCC(Reference Gago-Dominguez, Castelao, Yuan, Ross and Yu12, Reference Gago-Dominguez and Castelao13, Reference Greenland, Gago-Dominguez and Castelao48). This hypothesis is supported by observations in both experimental chemically induced models(Reference Greenland, Gago-Dominguez and Castelao48, Reference Okamoto, Asai, Saito, Okabe and Gomi49) and human renal cell tissue(Reference Zhang, Yamashita, Uetsuki and Kakehi50).
The possible interaction between diets high in specific types of fat and hypertension is of interest. We previously reported the joint effect of obesity and hypertension on RCC risk in these data and speculated that the increase in RCC risk related to obesity may be rather mild unless blood pressure was poorly controlled(Reference Brock, Gridley, Lynch, Ershow and Cantor26). Unhealthy diets that are high in certain types of fat may be associated with poorly controlled blood pressure, which could partly explain these observations. Also in line with the ‘lipid peroxidation hypothesis’(Reference Gago-Dominguez, Castelao, Yuan, Ross and Yu12, Reference Gago-Dominguez and Castelao13), we speculated that diets high in specific types of fat may play a synergistic role with hypertension for RCC risk.
The strengths of the present study include the use of a well-established tumour registry to ascertain cases(Reference Moore, Wilson and Campleman2), a randomly selected control sample representative of the general population and reasonable participation rates among the cases and controls. Additional strengths of the present study were our ability to adjust for a wide variety of potential confounding factors and the high prevalence of fat intake among the present study subjects. A difficulty in sorting out the effects of specific high-energy nutrients lay in their high intercorrelation. Although we did not find total energy to be a significant confounder in the present study, we controlled for energy intake in the analysis of nutrients in order to adjust for potential general over- or under-reporting of all foods.
In addition to limitations inherent in case–control studies of past diet, other limitations of the present study deserve a mention. Height, weight at various ages and hypertension were self-reported. It is possible that the risk associated with our high-fat spread food group was higher in individuals with hypertension, but the present study had limited power to detect the interaction between consumption of this food group and hypertension. Larger studies are necessary to test this hypothesis. In addition, the dietary questionnaire was retrospective and limited to fifty-five items, was not validated and portion sizes were not asked. The questions about meat were limited and did not ascertain inner and outer doneness and various forms of meat preparation. The questionnaire asked about past diet and responses may be subject to recall bias. If differences in dietary recall occur non-differentially with respect to case–control status, estimates of risk are typically biased towards the null. If recall is differential, risk estimates could be biased in either direction. It is known that although diet has some consistency over time, reported food intakes may not accurately reflect past behaviour(Reference Dwyer, Gardner, Halvorsen, Krall, Cohen and Valadian51). Dietary changes due to hypertension or other conditions were not ascertained. Dietary changes may also have occurred in the food supply (marketplace) over the past 20 years. Survey data suggest that the amount and proportion of energy from total and saturated fat have steadily declined over the last 20 years in the USA(Reference Kant and Graubard52). Thus, the present results may not be as relevant in the society today, or may reflect a latency effect. Given that 99 % of the participants in the present study were white, the present results may have limited generalisability to other racial/ethnic groups. Some observed associations may have been due to chance.
While RCC is not common in the general population, it is increasing, both in the USA and worldwide, despite a drop in smoking rates. It would therefore be worthwhile to further evaluate these findings in larger prospective studies.
Acknowledgements
This research was supported by the Intramural Research Program of the National Institutes of Health (NIH), NCI, Division of Cancer Epidemiology and Genetics (DCEG), and Sydney University, NSW, Australia Sabbatical Program for K. E. B. In addition, we acknowledge the invaluable support of Mr David Check, research assistant, Biostatistics Branch, DCEG, NCI, NIH. Contributions of the co-authors were as follows: K. P. C. designed the study and had overall responsibility for the project; A. G. C. designed the collection of dietary information; C. F. L. was responsible for overseeing data collection; G. G., K. E. B. and B. C.-H. C. conducted the data analysis; K. E. B. drafted the paper and all authors contributed to the final completion of the manuscript. None of the authors have any conflicts of interest (personal, commercial, political, academic or financial).
Nutritional distributions: Iowa case–control study compared with National Health and Nutrition Examination Survey II (NHANES II)
Pearson correlation coefficients between fat-related nutrients in the control population
