Vegetable and fruit consumption and risk of renal cell carcinoma: Results from the Netherlands cohort study
Article first published online: 31 MAY 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 117, Issue 4, pages 648–654, 20 November 2005
How to Cite
van Dijk, B. A.C., Schouten, L. J., Kiemeney, L. A.L.M., Goldbohm, R. A. and van den Brandt, P. A. (2005), Vegetable and fruit consumption and risk of renal cell carcinoma: Results from the Netherlands cohort study. Int. J. Cancer, 117: 648–654. doi: 10.1002/ijc.21203
- Issue published online: 14 SEP 2005
- Article first published online: 31 MAY 2005
- Manuscript Accepted: 18 MAR 2005
- Manuscript Received: 15 NOV 2004
- Dutch Kidney Foundation. Grant Number: C99.1863
- Dutch Cancer Society
- renal cell carcinoma;
- cohort study
Vegetable and fruit consumption is generally inversely associated with various cancer types, including renal cell carcinoma (RCC). The Netherlands cohort study on diet and cancer (NLCS) consists of 120,852 men and women, aged 55–69 years, who filled out a self-administered questionnaire that includes 150-item food-frequency questions and additional questions on lifestyle factors, at baseline in 1986. A case-cohort approach was used. After 9.3 years of follow-up, 275 microscopically confirmed incident cases were identified. Subjects with incomplete or inconsistent dietary data were excluded, leaving 260 RCC cases for analyses on fruit consumption and 249 RCC cases for analyses on vegetable consumption. Incidence rate ratios (RR) and corresponding 95% confidence intervals (CI) were estimated using Cox proportional hazard models. RRs for exposure variables are expressed per increment of 25 g/day and are adjusted for age, sex, smoking, body mass index and history of hypertension at baseline. The RRs for vegetable consumption were further adjusted for fruit consumption and vice versa. Total vegetable and fruit consumption (RR: 1.00; 95% CI 0.97–1.02), vegetable (RR: 1.00, 95% CI 0.96–1.06) and fruit consumption (RR: 1.00; 95% CI 0.97–1.03) were not associated with RCC risk. Also, no association existed for botanical subgroups of vegetables and fruit. For 30 individual vegetables and fruits, we observed one that significantly increased RR (mandarin consumption, RR: 1.76; 95% CI 1.28–2.42), which must be regarded cautiously because of multiple testing. These results suggest the absence of an association between vegetable and/or fruit consumption and RCC risk. © 2005 Wiley-Liss, Inc.
In general, vegetable and fruit consumption is assumed to reduce risk of various cancer types. Most reviews on renal cell carcinoma (RCC) also conclude that vegetable and fruit consumption may reduce RCC risk.1, 2, 3, 4, 5 Statistically significant reduced risks for RCC have been reported for the highest tertile of vegetable and fruit consumption,6 for vegetable and fruit consumption but restricted to men,7 for cruciferous/dark green vegetables,8 for vegetable and vegetable juice consumption,9 for carrot consumption,10 for root vegetable and banana consumption11 and for fruit consumption.12, 13 Furthermore, 2 studies showed estimates pointing in the direction of a protective effect, but associations were not statistically significantly different from one.14, 15 Finally, 3 studies observed null associations.16, 17, 18
Antioxidant vitamins, fibres or enzyme inducers present in vegetable and fruits might be responsible for this preventive effect.5 Plants rich in nitrosation inhibitors, antioxidants or enzyme inducers, e.g., ascorbate and polyphenols or carotenoid-rich vegetables, garlic and cruciferous vegetables, may thus be most effective in preventing cancer.5 The biologic plausibility that vegetable and fruit consumption reduces cancer risk is present, but reported results do not support this notion unequivocally.6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18
We investigated total vegetable and fruit consumption, as well as the consumption of botanical groups of vegetables and fruits and individual vegetables and fruits and RCC risk in a large cohort. For some important risk factors stratified analyses will be carried out, e.g., for smoking because smoking increases oxidative stress and a more pronounced protective effect has been reported for never smokers.12, 15
Material and methods
Netherlands Cohort Study
The Netherlands Cohort Study on diet and cancer is a large cohort study that started in September 1986. The study design has been reported in detail elsewhere.19 Briefly, the cohort included 120,852 men and women aged 55–69 years in 1986. An elderly cohort was selected because dietary habits (and their contrasts) are stabilized, and such a cohort will yield sufficient cases for meaningful analyses within a reasonable time period.19 The case-cohort design was used, which means that a subcohort of 5,000 men and women was randomly sampled from the cohort after baseline exposure measurement to estimate the number of person-years for the entire cohort, whereas cases were enumerated for the entire cohort.20
Follow-up for incident cancers and vital status
The entire cohort was followed for incident cancer by computerized record linkage with the Netherlands Cancer Registry and PALGA, a national database of pathology reports. The method of record linkage to obtain information on cancer incidence has been described previously.21 The completeness of cancer follow-up was estimated to be more than 96%.22 After 9.3 years of follow-up, 275 incident RCC cases have been identified.
The subcohort has been followed up for vital status information biennially by mail. The vital status of subcohort members who did not respond was completed by contacting municipal population registries. Only 2 male subcohort members were lost to follow-up after 9.3 years and were censored. Subcohort members with prevalent cancer at baseline (other than skin cancer) were excluded from analyses, leaving 4,779 subcohort members.
At baseline, all cohort members completed a mailed, self-administered questionnaire on dietary habits, lifestyle, smoking, personal and family history of cancer and demographic data. The dietary section was a 150-item semiquantitative food-frequency questionnaire, which was validated against 3-day diaries completed at 3 time points during a calendar year.23 The questionnaire concentrated on the habitual consumption of food and beverages during the year preceding the start of our study. With regard to vegetable consumption, participants were asked to report their frequency of consumption of a number of vegetables, both in summer and in winter. They could choose 1 of 6 categories, ranging from “never or less than once a month” to “3 to 7 times per week.” Usual serving sizes were asked for string beans and cooked endive only, the mean of which served as an indicator for serving sizes of all cooked vegetables. This procedure was chosen because in a pilot study (based on an extensive dietary history with food models and photos used to estimate individual portion sizes) it was shown that serving sizes of different types of cooked vegetables were correlated within subjects.24 To derive an individual serving size for each type of vegetable, the indicator serving size was multiplied with a type-specific factor calculated from the same pilot study data as the ratio of the means of the specific to the indicator serving sizes.24 For tomatoes and sweet peppers, consumption was asked in pieces per week and month, respectively, during summer and winter. With regard to fruit consumption, frequencies varying from “never or less than once a month” to “6 or 7 days per week” and amounts consumed could be reported for mandarins, oranges, grapefruits, grapes, bananas, apples/pears and strawberries. Using standard portion sizes, these frequencies and amounts have been converted to consumption in grams per day. The choice of items for inclusion in the questionnaire was such that it covered almost all vegetables and fruits eaten regularly, with the exceptions of chicory, red cabbage and cucumber. Broccoli was a rarely available vegetable in 1986 and therefore not included. However, an open-ended question on other foods eaten on a regular basis was included. Participants could write down how many times per week they ate such a food and how much they were used to eating on each occasion.
According to criteria published previously,23 subjects with incomplete or inconsistent dietary data were excluded; 260 RCC cases and 4,441 subcohort members remained for analyses on fruit consumption. In addition, we computed an error index based on the consistency of responses on vegetable questions. Questions on vegetable consumption appeared early in the questionnaire, which led some subjects to make mistakes on these particular items, whereas items appearing further along in the questionnaire were filled out without problems. When the vegetable error index exceeded a certain value, i.e., more than 3 errors, subjects were excluded from analyses on vegetable consumption. Therefore, data analysis regarding vegetable consumption was based on 249 RCC cases and 4,201 subcohort members.
Rate ratios (RRs) were calculated for total vegetable and fruit consumption, total vegetable consumption, total fruit consumption, cooked vegetable consumption and raw vegetable consumption. RRs were also calculated for botanical groups of vegetables and fruits (composition of botanical groups is shown in Appendix I), with the exception of groups based on one main constituent (carrots, beets, tomatoes, grapes, bananas and strawberries). These were analysed in the individual vegetable and fruit analysis only. RRs were calculated per increment of 25 g/day. Also, subjects were classified into quintiles and tertiles of vegetable or fruit consumption, based on the distribution in the subcohort. Analyses for total vegetable and fruit consumption, total vegetable, and total fruit consumption have been repeated excluding the first 2 years of follow-up to evaluate whether preclinical RCC influenced results.
Based on the literature and previous analyses, considered confounders were age (continuous), sex, cigarette smoking (current smoker yes or no, number of smoking years and number of cigarettes per day), alcohol intake, body mass index (BMI), history of hypertension, physical activity, energy intake and social economic status (SES) based on education. We did not adjust for family history of RCC (present or not in a first-degree relative) because only 49 participants (4 cases and 45 subcohort members) reported a first-degree relative with RCC. Age and sex were included in all analyses. Factors that statistically significantly contributed to the model were entered in the multivariable model, leaving BMI and a history of hypertension. Smoking was also entered because some of the smoking variables to describe the smoking status satisfied this criterion as well. Moreover, these factors are well-known risk factors of RCC and may be associated with a “healthy” lifestyle, which may also be associated with vegetable and fruit consumption. For all analyses on vegetable consumption, fruit consumption was also included as a confounder and vice versa. We investigated possible interaction by sex by entering an interaction term in the model and assessing the significance of this term using the Wald test. Since no interaction on RCC risk between sex and dietary intakes was observed, results are shown for men and women combined. Furthermore, total vegetable and fruit consumption, vegetable consumption, and fruit consumption are presented stratified by smoking status (never, ex- or current smoker), BMI (<25 kg/m2 or ≥25 kg/m2) and a history of hypertension (yes or no).
RRs and corresponding 95% confidence intervals (CI) for RCC were estimated using Cox proportional hazard models processed with the STATA statistical software package (STATA statistical software, Release 7, 2001; STATA Corporation, College Station, TX) after testing the proportional hazards assumption using scaled Schoenfeld residuals.25 The proportional hazards assumption was not rejected. Standard errors were estimated using the robust Huber-White sandwich estimator to account for additional variance introduced by sampling from the cohort.26 To obtain p-values for dose-response trends, ordinal exposure variables were fitted as continuous terms. Two sided p-values are reported throughout this article.
The mean age (standard deviation) was 62.1 (3.9) for cases and 61.4 (4.2) for subcohort members. Sixty-five percent of cases were male compared to 49% of subcohort members. The mean BMI was somewhat higher for cases than for subcohort members (25.5 kg/m2 compared to 25.1 kg/m2). Almost a third (30%) of the cases and approximately a quarter (26%) of subcohort members reported a history of hypertension.
Also, cases were more often current smokers (38% compared to 29%) or ex-smokers (38% compared to 35%) and smoked more and longer than subcohort members (within the strata of ex- and current smokers).
No statistically significant interaction was shown for sex and total vegetable and fruit consumption (p-value = 0.40), for sex and vegetable consumption (p-value = 0.12) or for sex and fruit consumption (p-value = 0.99), so RRs for RCC were calculated for men and women combined.
Differences in vegetable consumption between cases and subcohort members were small (Table I). However, mean fruit consumption was somewhat lower for cases than for subcohort members, and this difference was present in all fruit groups (Table I). Table II shows means of vegetable and fruit consumption for never, ex- and current smokers, for a BMI <25 and ≥25 and for a positive history of hypertension or not (Table II). Similar means were observed for vegetable consumption in different smoking groups and for strata of BMI and a history of hypertension. Mean fruit consumption was higher in never smokers. In the stratum of current smokers, cases ate more vegetables and less fruit than subcohort members. Among never smokers, cases consumed less fruit than subcohort members (Table II). Fruit consumption was somewhat lower for cases with a BMI <25 (Table II).
|Exposure variables||Cases Mean (SD)||Subcohort Mean (SD)|
|Total vegetables and fruit1||350.5 (141.3)||362.1 (151.9)|
|Total vegetables1||190.4 (76.7)||188.4 (75.6)|
|Cooked vegetables1||152.4 (65.3)||149.3 (61.3)|
|Raw vegetables1||38.0 (23.7)||39.1 (29.6)|
|Legumes1||36.1 (25.0)||32.4 (22.1)|
|Brassicas1||32.9 (20.2)||31.7 (19.8)|
|Leafy vegetables, cooked1||21.9 (15.5)||21.3 (15.8)|
|Leafy vegetables, raw1||9.1 (7.9)||10.0 (9.0)|
|Allium vegetables, cooked1||29.6 (24.8)||29.1 (24.3)|
|Total fruit2||161.9 (112.2)||175.3 (119.4)|
|Citrus fruit2||73.0 (71.4)||77.0 (74.3)|
|Apples, pears2||81.4 (74.6)||87.3 (82.4)|
|Vegetable consumption||Fruit consumption|
|Cases (n)||Cases Mean (SD)||Subcohort members (n)||Subcohort Mean (SD)||Cases (n)||Cases Mean (SD)||Subcohort members (n)||Subcohort Mean (SD)|
|Never smoker||61||183.9 (71.4)||1,496||187.4 (74.6)||63||188.2 (113.6)||1,588||202.5 (123.3)|
|Ex-smoker||92||186.5 (73.4)||1,525||192.9 (74.5)||100||172.2 (106.3)||1,594||173.4 (117.1)|
|Current smoker||96||198.4 (82.8)||1,180||183.8 (78.2)||97||134.3 (112.5)||1,259||143.3 (108.7)|
|BMI < 25||119||187.3 (72.7)||2,221||188.3 (76.4)||124||151.4 (99.3)||2,344||175.9 (119.5)|
|BMI ≥ 25||130||193.4 (80.4)||1,980||188.4 (74.8)||136||171.5 (122.4)||2,097||174.5 (119.4)|
|No history of hypertension||172||187.7 (79.5)||3,118||186.9 (74.7)||180||161.0 (115.0)||3,290||173.2 (119.7)|
|History of hypertension||77||196.6 (70.0)||1,083||192.6 (78.3)||80||164.0 (106.5)||1,151||181.3 (118.4)|
Multivariable rate ratios of RCC for total vegetable and fruit consumption as well as for botanical groups of vegetable and fruit consumption are shown in Table III. We also analysed age- and sex-adjusted rates, but these were not very different (data not shown). “All vegetables and fruits,” “all vegetables” and “all fruits” were not (inversely) associated with RCC risk (all RRs equalled 1). RRs of RCC for these groups hardly changed after exclusion of the first 2 years of follow-up (data not shown).
|Vegetable/fruit group||Quintile of consumption||p-value trend||Continuous/25g per day|
|1 (low)1||2||3||4||5 (high)|
|All vegetables and fruits|
|Median consumption (g/day)||189||275||343||418||556|
|Cases of RCC||50||34||48||51||39||222|
|Multivariable-adjusted rate ratio (RR)2||1||0.69||0.94||1.01||0.78||0.79||1.00|
|Median consumption (g/day)||104||145||178||217||287|
|Cases of RCC||48||44||38||50||42||222|
|Median consumption (g/day)||79||114||142||173||230|
|Cases of RCC||48||40||42||42||50||222|
|Median consumption (g/day)||8||22||34||48||76|
|Cases of RCC||33||47||50||52||40||222|
|Median consumption (g/day)||10||19||28||38||60|
|Cases of RCC||44||39||41||40||58||222|
|Median consumption (g/day)||10||20||28||38||57|
|Cases of RCC||43||34||51||52||42||222|
|Leafy vegetables, cooked4|
|Median consumption (g/day)||4||12||19||27||41|
|Cases of RCC||41||48||46||41||46||222|
|Leafy vegetables, raw4|
|Median consumption (g/day)||2||4||7||12||22|
|Cases of RCC||54||45||42||43||38||222|
|Allium vegetables, cooked4|
|Median consumption (g/day)||5||16||24||37||60|
|Cases of RCC||65||31||45||38||43||222|
|Median consumption (g/day)||44||107||155||215||326|
|Cases of RCC||47||45||51||39||50||232|
|Median consumption (g/day)||3||26||60||98||176|
|Cases of RCC||43||40||63||35||51||232|
|Apples and pears5|
|Median consumption (g/day)||6||40||80||116||181|
|Cases of RCC||45||42||59||42||44||232|
Legume consumption was associated with an increased risk for every 25 grams of legume consumed more per day (multivariable adjusted RR: 1.14; 95% CI 0.99–1.33) (Table III). This increased risk was restricted to the highest quintile of legume consumption (RR: 1.31; 95% CI 0.83–2.07). An increment of 25 grams per day of raw, leafy vegetables was associated with a reduced RCC risk (RR: 0.68; 95% CI 0.45–1.04), with an indication of a decreasing risk with increasing consumption (p-value for trend: 0.11) (Table III).
Table IV shows RRs of RCC for individual vegetables and fruits. No statistically significant reduced or increased risks were observed for individual vegetables. A significantly increased risk of RCC was observed for mandarin consumption (RR: 1.76; 95% CI 1.28–2.42). Other citrus fruits, however, were not associated with either an increased or decreased risk of RCC (Table IV). Banana consumption may be associated with a reduced RCC risk (RR: 0.85; 96% CI 0.72–1.01).
|Variable1||Median consumption (grams/day)||Age, and sex-adjusted rate ratio (RR) per 25 g increment||95% CI||Multivariable-adjusted RR2 per 25 g increment||95% CI|
|String and sliced beans||17||1.16||0.94–1.43||1.17||0.94–1.46|
Table V shows multivariable-adjusted RRs for vegetable and fruit consumption, stratified by smoking status, BMI and history of hypertension. The estimated RRs did not differ largely between strata. None of the interaction terms of smoking, BMI or history of hypertension with tertiles of total vegetable consumption or with total fruit consumption were statistically significant.
|No. of cases person-years in subcohort||Tertile of vegetable and fruit consumption||p-value trend||Continuous per 25 g increment|
|1 (low)||2||3 (high)|
|Never smoker1||59/12,970||1 (reference)||0.92 (0.47–1.77)||1.11 (0.57–2.18)||0.77||0.99 (0.91–1.09)|
|Ex-smoker2||82/12,221||1 (reference)||0.91 (0.52–1.59)||0.84 (0.48–1.49)||0.56||0.96 (0.87–1.05)|
|Current smoker2||81/8,921||1 (reference)||0.95 (0.53–1.71)||1.24 (0.72–2.15)||0.44||1.05 (0.98–1.13)|
|BMI <253||112/18,673||1 (reference)||0.84 (0.52–1.36)||1.08 (0.67–1.73)||0.77||1.00 (0.94–1.07)|
|BMI ≥253||117/16,564||1 (reference)||1.00 (0.63–1.59)||1.04 (0.65–1.66)||0.87||1.01 (0.94–1.08)|
|No history of hypertension4||151/25,241||1 (reference)||0.86 (0.57–1.31)||1.01 (0.67–1.51)||0.98||1.00 (0.94–1.06)|
|History of hypertension4||71/8,870||1 (reference)||1.02 (0.54–1.92)||1.07 (0.56–2.04)||0.83||1.01 (0.93–1.08)|
|Never smoker1||61/13,755||1 (reference)||0.76 (0.40–1.41)||0.71 (0.38–1.33)||0.30||0.98 (0.93–1.04)|
|Ex-smoker2||90/12,734||1 (reference)||0.94 (0.54–1.63)||1.21 (0.71–2.06)||0.46||1.01 (0.97–1.05)|
|Current smoker2||81/9,532||1 (reference)||1.17 (0.69–1.99)||0.79 (0.43–1.46)||0.54||0.99 (0.93–1.05)|
|BMI <253||117/19,698||1 (reference)||0.97 (0.61–1.53)||0.81 (0.50–1.31)||0.39||0.97 (0.93–1.01)|
|BMI ≥253||122/17,627||1 (reference)||0.85 (0.53–1.34)||0.95 (0.60–1.49)||0.82||1.01 (0.97–1.06)|
|No history of hypertension4||159/26,597||1 (reference)||1.04 (0.70–1.55)||0.96 (0.63–1.44)||0.83||1.01 (0.97–1.04)|
|History of hypertension4||73/9,423||1 (reference)||0.81 (0.44–1.46)||0.88 (0.49–1.57)||0.67||0.98 (0.93–1.03)|
Neither total vegetable and/or fruit consumption, non-consumption of a botanical group of vegetables or fruits or individual vegetable or fruit consumption was associated with a decreased RCC risk in this cohort study. The only statistically significant result observed in 30 analyses of specific vegetables and fruits was an increased risk for mandarin consumption, which must be regarded cautiously because of multiple testing. Also, no modifying effect for smoking, BMI or a history of hypertension could be shown.
The prospective nature of a cohort study together with the completeness of follow-up, as has been achieved in our study, reduced the potential for selection bias to a minimum. Information bias, i.e., a change in (report of) dietary habits of RCC cases due to the disease, is also largely avoided in a prospective study because dietary habits were reported before the disease was diagnosed. A change in dietary habits of subjects with preclinical RCC at the time of completing the baseline questionnaire remains possible. An indicator of advanced disease such as weight loss is estimated to be associated with RCC in approximately 35% of cases.27 Weight loss may be induced by substances such as cytokines, insulin, inflammatory mediators, etc., produced in pathologic amounts by the tumour.27 However, results excluding the first 2 years of follow-up did not differ from presented results (including the first 2 years of follow-up), indicating that our results were not influenced by the possible presence of preclinical RCC cases. Also, more than 50% of RCCs are now detected incidentally because of the more pervasive use of noninvasive imaging for the evaluation of a variety of nonspecific symptom complexes.27
Residual confounding by risk factors for RCC such as smoking or BMI is a realistic threat. Clustering of low vegetable and fruit consumption with smoking has been reported for the Dutch population,28 and insufficient control of 1 factor will then confound the association between the other factor and RCC. We tried to model cigarette smoking habits such that they best explained RCC risk, resulting in a model including the habitual number of cigarettes smoked per day (smoking amount) and the number of years smoked (smoking duration), both as continuous variables. Furthermore, estimated RRs were not different for smoking strata (never, ex-, current).
Correction for BMI may have influenced results because high vegetable and fruit consumption might be associated with a lower BMI (as a result of the dietary pattern). However, the correlation of BMI with total vegetable and fruit consumption is not large (r = 0.025; p-value = 0.0781). Also, mean BMI did not differ for quintiles of total vegetable and fruit consumption, vegetable consumption, and fruit consumption (mean BMI range: 24.8–25.2 kg/m2). Moreover, the RR and corresponding 95% CI for vegetable and fruit consumption did not change after removing BMI from the model.
Although we measured vegetable and fruit consumption extensively, a potential limitation remains misclassification of exposure. Usual vegetable consumption is not easy to assess in food-frequency questionnaires (or in other methods of dietary assessment), particularly if portion sizes have to be estimated. In the NLCS validation study, the FFQ was tested against 3-day diaries completed at 3 time points during a calendar year among 212 randomly selected subcohort members. The correlation coefficient for total vegetable consumption was 0.4,23 which is moderate but comparable to the figure reported for many other prospective studies.29, 30 One of the reasons for the low correlation may be the relative lack of true contrast in the frequency of vegetable consumption in a population such as the Dutch, because people are accustomed to a diet including only 1 hot meal per day, which almost always includes vegetables. This relative lack of contrast and thus a relatively large measurement error may result in attenuation of the estimated RR for the association of total vegetable consumption and RCC. Due to individual preferences, contrast in consumption frequency of specific vegetables is much higher, which means that attenuation is probably less important for RRs estimated for specific vegetables. It was not possible to assess validity for specific vegetables in the NLCS validation study because 9 days of dietary record (3-day diaries completed at 3 time points during a calendar year) are not sufficient to estimate consumption frequency of specific vegetables. To minimize the amount of uninformative data in addition to the general dietary exclusion criteria, we excluded subjects who appeared not to have understood how to fill out the questions on vegetable consumption, which occurred in the first part of the food-frequency questionnaire; an extreme score on the vegetable error index defined those subjects. For the same cohort, statistically significant inverse associations have been described between most categories of vegetable or fruit consumption and lung cancer24 as well as between brassica vegetables and cooked leafy vegetables and colon cancer,31 indicating that measurement error or too little contrast most likely do not mask a possible association.
To our knowledge, only 3 cohort studies11, 14, 16 have evaluated the relationship between vegetable and/or fruit consumption and RCC. The Iowa Women's Health Study with data on 124 women reported no association for “total fruit and vegetables” and for “cruciferous or green leafy vegetables”.16 In an earlier report from this study on 62 cases, a null association was reported for consumption of “fruit + vegetables,” “fruit” and “vegetables” and RCC risk.32 These results were confirmed in our study.
The Swedish Mammography Cohort included 122 Swedish women and observed a nonsignificant inverse association between the combined consumption of total fruits and total vegetables and RCC risk. Individual vegetables were associated with nonsignificantly decreased risks, except for root vegetables, which were statistically significantly inversely associated. For individual fruits, a nonsignificant inverse association with apples, a null association with citrus fruit, a statistically significant inverse association with banana and a nonsignificant increased risk with fruit juice consumption were observed.11 The only agreement with the current study was the protective effect for banana consumption. A direct relationship between total phenolic content and total antioxidant activity in phytochemical extracts of different fruits has been observed; bananas ranked 6th of 11 fruits tested.33 Bananas had the highest bound-W phenolics content, but the significance of bound phytochemicals in fruits to human health is not clear.33 The other cohort study was conducted amongst Californian Seventh-Day Adventists with 8 male and 6 female incident cases.14 Participants in this study seldom smoked (only 3% current smokers and 20% ex-smokers) and were mostly vegetarian. This study only reported on green salad and fruit consumption for which a statistically nonsignificant inverse association was found.
Ten case-control studies6, 7, 8, 9, 10, 12, 13, 15, 17, 18 have evaluated the relationship between vegetable and/or fruit consumption and RCC risk. Three studies are based on a large number of cases and population-based controls. The first was conducted in Los Angeles and reported on dark green vegetables, yellow-orange vegetables, tomatoes or tomato products, citrus fruits and citrus fruit juices in relation to RCC, based on 1,204 RCC cases and an equal number of neighbourhood controls matched by sex, date of birth (within 5 years) and ethnicity.8 Only dark green vegetables showed a significant inverse association with RCC risk. Results were not different for smokers or never smokers, for persons with a BMI greater or less than 24.4 kg/m2 or for persons with or without a history of hypertension, which was confirmed in this study. The second study was performed in Canada among 1,279 incident cases and 5,370 controls. A significant inverse association with RCC was observed with increasing total consumption of vegetables and vegetable juices for males and females combined. Statistically significant inverse associations between total consumption of dark green vegetables (broccoli and spinach) and cruciferous vegetables (broccoli, cabbage, cauliflower and Brussels sprouts) were observed for females only. Smoking did not modify the association between vegetable and fruit consumption and RCC.9 The other large case-control study was a multicenter study from Australia, Denmark, Sweden and the United States (International renal-cell cancer study)15 and investigated total vegetables, orange/dark green vegetables, cruciferous vegetables, allium vegetables, total fruit, citrus, and apples and pears among 1,185 cases and 1,526 controls. A significantly decreased risk in quartile 4 for orange/dark green vegetables was observed, with most estimates less than, but close to one and a more pronounced statistically significant inverse association with RCC in never smokers. Four of the case-control studies, referred to earlier, were part of the International Renal Cell Cancer Study.12, 13, 17, 18 These studies reported a statistically significant inverse association with RCC for fruit consumption12, 13 and null associations.16, 17, 18 However, results from one study were not included in the aggregated article because other methods were used for dietary assessment,13 whereas another article included a larger case group,18 and a third article also reported on food intake 20 years before the interview (with similar results).12 Data from case-control studies from Italy, Germany and China mostly showed decreased risks with vegetable and fruit consumption.6, 7, 10, 13 Statistically significant decreased associations were only observed for carrot consumption,10 the highest tertile of green vegetable consumption,6 total vegetable and total fruit consumption but confined to men7 and the highest tertile of fruit consumption.6, 13 None of the earlier mentioned studies reported on mandarin consumption, which suggests this was either not investigated or that no statistically significant association was observed.
We were able to assess the independent association with specific vegetable and fruit groups and for individual vegetables and fruits by adjusting for total vegetable and/or fruit consumption. Only one cohort study11 and one hospital-based case-control study also adjusted for total vegetable/fruit consumption.6 All significantly reduced risks reported so far seem to be confined to a subgroup such as one specific type of vegetable or fruit or to a specific tertile quartile or quintile, which may suggest chance results due to multiple testing. In our study, statistically significant results for a quintile were also incidentally observed, but we did not observe this in multivariable-adjusted analyses based on the continuous vegetable or fruit variable or in the trend over the categories.
To our knowledge, this is the largest prospective study on vegetable and fruit consumption and RCC risk currently available. Our results suggest the absence of an association between vegetable and fruit consumption and RCC risk.
The authors thank the staffs of the Dutch regional cancer registries and the Netherlands national database for pathology (PALGA) for providing incidence data. They also thank Dr. E. Dorant and Ms. C.A. de Brouwer for their preparatory work for their study; Dr. A. Volovics and Dr. A. Kester for statistical advice; Ms. S. van de Crommert, Ms. H. Brants, Ms. J. Nelissen, Ms. C. de Zwart, Ms. M. Moll, Ms. W. van Dijk, Ms. M. Jansen and Ms. A. Pisters for data entry and processing; and Mr. H. van Montfort, Mr. T. van Moergastel, Ms. L. van den Bosch and Mr. R. Schmeitz for programming assistance.
- 5Fruit and vegetables IARC handbooks of cancer prevention, vol. 8. Lyon: IARC Press, 2003. 293., .
- 22Estimation of the coverage of Dutch municipalities by cancer registries and PALGA based on hospital discharge data. Tijdschr Soc Gezondheidsz 1994; 72: 80–4., , .
- 2775. Renal tumors. In: WalshPC, RetikAB, Darracott VaughanJrE, WeinAJ, KavoussiLR, NovickAC, PartinAW, PetersCA, eds. Campbell's urology, 8th ed., vol. 4. Philadelphia: Saunders, 2002. 2672–719., .
|Group name||Vegetables or fruit represented in this group|
|Cooked vegetables||Endive (prepared); cauliflower (prepared); kale (prepared); mushrooms; leek (prepared); spinach (prepared); Brussels sprouts (prepared); onion (prepared); carrots (prepared); sauerkraut; beets (prepared); broad beans (prepared); cabbage (prepared); sliced beans, string beans (prepared); sweet peppers; other vegetables prepared|
|Raw vegetables||Raw endive; lettuce; carrots (raw); tomatoes; other raw vegetables|
|Legumes||Broad beans (prepared); pulses; sliced beans, string beans (prepared)|
|Brassicas||Cauliflower (prepared), cabbage (prepared), kale (prepared); Brussels sprouts (prepared)|
|Leafy vegetables, cooked||Endive (prepared); spinach (prepared)|
|Leafy vegetables, raw||Raw endive; Lettuce; Cress1|
|Allium vegetables||Leek (prepared); onions (prepared); cocktail onions (sweet-sour),1Garlic2|
|Carrots||Carrots (prepared); canned carrots1|
|Beets||Beets (prepared); beet juice|
|Tomatoes||Tomatoes (raw); tomato juice|
|Other cooked vegetables||Other vegetables prepared1|
|Other raw vegetables||Other raw vegetables1|
|Citrus fruit||Lemons,1 fresh lemon juice,1 grapefruit, fresh grapefruit juice; mandarins; oranges, fresh orange juice|
|Grapes||Grapes (blue and white)|
|Apples/pears||Applesauce; apples, pears|
|Other fruit and fruit juices||Other fruits and fruit juices1|