Background:

Prostate cancer and urinary bladder cancer are frequently occurring cancers with few risk factors identified. We examined the relation of Mediterranean diet (MD) adherence with risks of prostate and bladder cancer in the Netherlands Cohort Study (NLCS).

Methods:

Data were available for 58,279 men and 62,573 women, who completed a baseline questionnaire on diet and other cancer risk factors. Multiple MD scores, including the alternate Mediterranean diet score without alcohol (aMEDr), were calculated to assess MD adherence. After 20.3 years of follow-up, 3,868 prostate cancer cases (advanced: 1,256) and 1,884 bladder cancer cases could be included in multivariable Cox proportional hazards analyses.

Results:

aMEDr was not associated with advanced prostate cancer risk [hazard ratio (HR)per 2-point increment (95% confidence interval, 95% CI) = 1.06 (0.96–1.17)]. In contrast, higher aMEDr values were associated with a significantly increased risk of nonadvanced prostate cancer (Ptrend = 0.04). For bladder cancer risk, no association was observed with aMEDr [HRper 2-point increment (95% CI) = 1.00 (0.92–1.09)]. Absolute scores based on the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) dietary recommendations were not associated with prostate or bladder cancer risk.

Conclusions:

MD adherence, measured by aMEDr or other MD scores, was not associated with decreased risks of advanced prostate cancer and bladder cancer in the NLCS. Higher levels of care-seeking behavior, screening attendance, and prostate cancer awareness in higher educated men with healthier lifestyles could potentially explain the positive associations observed for nonadvanced prostate cancer risk.

Impact:

MD adherence does not seem to reduce the risk of (advanced) prostate cancer or bladder cancer.

Worldwide, cancers of the prostate and urinary bladder were estimated to be the second and tenth most commonly diagnosed cancer types in 2018 (1). Together, these cancer types were responsible for over half a million deaths in this year (1). So far, only advancing age, African-American race, family history of prostate cancer, and genetic predisposition have been identified as established risk factors for prostate cancer (2, 3). Tobacco smoking is the most important risk factor for bladder cancer (4). Other bladder cancer risk factors include Schistosoma haematobium infection, environmental and occupational exposure to chemicals, and exposure to arsenic in drinking water (5). The high incidences, slow disease development and progression (prostate cancer), and high recurrence rates (bladder cancer), make prostate and bladder cancer suitable targets for preventive approaches (6–8).

The traditional Mediterranean dietary pattern (MD) is mainly based on plant foods. Intakes of vegetables, legumes, fruits, nuts, whole grains, fish, and monounsaturated fatty acids (MUFA, from olive oil) are high in the MD, whereas animal foods (e.g., meats and dairy products) are consumed in limited amounts. Typically, alcohol is consumed in moderation and usually in the form of wine during meals (9–11).

Prostate cancer is a disease with a heterogeneous nature. Advanced and more aggressive prostate tumors may etiologically differ from early, screening-detected forms that otherwise might never have become clinically relevant (2). Risk factors for prostate cancer subtypes (defined by grade, stage, or survival) may differ as they may exert their effect via different biological pathways (12). Therefore, effects of potential risk factors on advanced prostate cancer risk are of primary interest. Up until now, two prospective cohort studies from the United States have evaluated the relation of a priori defined MD adherence with advanced prostate cancer risk and did not observe an association (13, 14). In contrast to the results for advanced prostate cancer, the prospective evidence suggests that MD adherence might be associated with a reduced risk of (invasive) bladder cancer/urothelial cell carcinoma (UCC; refs. 15, 16). However, the inverse associations were not statistically significant.

In this analysis of the Netherlands Cohort Study (NLCS), we examined associations between a priori defined MD adherence and risks of prostate and urinary bladder cancer. Associations were compared for subtypes of the investigated cancer sites classified by stage at diagnosis (prostate cancer) or malignancy grade (bladder cancer). In addition, the effect of exclusion of alcohol from the MD scores was evaluated and performances of the relative MD scores were compared with those of absolute scores based on the dietary recommendations to prevent cancer issued by the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR).

Study population and cancer follow-up

The prospective NLCS includes 58,279 men and 62,573 women, ages 55 to 69 years (17–20). Study participants consented to participate by completing a self-administered baseline questionnaire on diet and other cancer risk factors in September 1986. A case-cohort approach was applied to process and analyze the data efficiently (17, 20, 21). A subcohort (N = 5,000) was randomly drawn just after baseline to estimate accumulated person-years at risk, whereas cases originated from the entire cohort. Subcohort members were followed-up biennially for vital status. The NLCS was approved by the institutional review boards from Maastricht University and the Netherlands Organization for Applied Scientific Research, and was conducted in accordance with the Declaration of Helsinki.

Incident cancer cases were detected annually for 20.3 years of follow-up (baseline until 31 December 2006) through record linkage with the Netherlands Cancer Registry and the nationwide Dutch Pathology Registry (PALGA; ref. 18). In total, 3,978 prostate cancer cases (ICD-O-3 code C61), 2,049 bladder cancer cases (ICD-O-3 code C67), and 4,084 subcohort members (men: 2,057, women: 2,027) were eligible for inclusion in the present analyses (Supplementary Figs. S1 and S2). Prostate cancer cases were classified as nonadvanced (N = 2,397, stages T1/T2, N0, and M0) or advanced (N = 1,294, stages T3/T4 or N+ or M1) at diagnosis, whereas bladder cancer cases were categorized in noninvasive (N = 1,053, malignancy grade 2) and invasive (N = 996, malignancy grade 3) subtypes. All included cases were microscopically confirmed. Furthermore, eligible cases and subcohort members did not have prevalent cancer at baseline (except skin cancer), and had complete and consistent data available on diet, alcohol, and MD adherence.

Exposure assessment

At baseline, participants were asked about their usual dietary intake during the previous year via a validated, 150-item, semiquantitative food frequency questionnaire (FFQ; refs. 19, 22). Nutrient intakes were derived from the FFQ data utilizing the 1986 Dutch food composition (NEVO) table (23).

Mediterranean diet adherence

The alternate and modified Mediterranean diet scores (aMED and mMED, respectively) were calculated to estimate the relative level of MD adherence (24–26). These scores are adaptations of the traditional Mediterranean diet score (tMED) created by Trichopoulou and colleagues (27, 28) and are each composed of 9 dietary components. For aMED (24, 26), 1 point (and 0 otherwise) is assigned to mean daily intakes at or above the sex-specific median of vegetables (excluding potatoes), legumes, fruits, nuts, whole grains, fish, and the ratio of MUFA to saturated fatty acids (SFA). Inverse scoring is applied to red and processed meats. Finally, 1 point can be obtained for a moderate alcohol consumption, defined as 5 to 25 g/day (24, 26). mMED (25) differs from aMED as follows: fruits and nuts are combined, total intakes of cereal and meat are considered, dairy intake is included (1 point if below sex-specific median), and the ratio of unsaturated fatty acids (MUFA + polyunsaturated fatty acids) to SFA replaces the MUFA:SFA ratio. Furthermore, other cut-offs are used to define moderate alcohol consumption (men: 10–50 g/day, women: 5–25 g/day; ref. 25). Before calculation of the MD scores, food intakes were standardized to daily energy intakes of 2,000 kcal (women) and 2,500 kcal (men; refs. 26, 27). aMED and mMED range from 0 to 9 points (lowest to highest MD adherence). Because alcohol consumption has been associated with increased risks of several types of cancer (29), we also created aMED and mMED variants without alcohol (aMEDr and mMEDr, respectively) ranging from 0 to 8 points.

Statistical analyses

We evaluated relations between MD adherence and risks of prostate and bladder cancer (subtypes) using Cox proportional hazards models with duration of follow-up as time scale to estimate hazard ratios (HR) and 95% confidence intervals (95% CI). Person-years at risk in the subcohort were calculated from baseline until prostate or bladder cancer diagnosis, death, emigration, loss to follow-up or end of follow-up, whichever came first. Sampling from the cohort introduces additional variance. Therefore, standard errors were calculated using the Huber–White sandwich estimator (30). The proportional hazards assumption was checked by scaled Schoenfeld residuals tests and visual inspection of –ln(–ln) survival plots (31).

We tested associations of MD adherence with risks of prostate and bladder cancer (subtypes) in age- (and sex-)adjusted and fully adjusted analyses, in which MD scores were modelled both categorically [low: 0–3, middle: 4–5, high: 6–8(9)] and continuously (per 2-point increment; refs. 25, 26). Men and women were combined in the models for bladder cancer, because there was no statistically significant interaction by sex. The fully adjusted models concerning prostate cancer risk were adjusted for the following predefined confounders: age at baseline, body mass index (BMI), alcohol consumption (except for models containing MD scores including alcohol), total daily energy intake, highest level of education, and family history of prostate cancer. For bladder cancer risk, sex and cigarette smoking behavior (status, frequency, and duration) were also listed as predefined confounders. Additionally, these analyses were adjusted for family history of bladder cancer instead of prostate cancer. Other confounders considered, but not included (removal resulted in <10% change in the effect estimate of the MD score), were cigarette smoking status (prostate cancer only), non-occupational physical activity, history of diabetes, height (prostate cancer only), tea consumption, and coffee consumption (bladder cancer only). Sex-specific median MD score values in the subcohort were appointed to each adherence category and fitted continuously in Cox regression models to perform trend tests. A competing risks procedure was applied to test for heterogeneity across the prostate and bladder cancer subtypes (32). Standard errors for the observed differences were estimated using a bootstrapping method specifically designed for the case-cohort approach (33).

Model fits of the various MD scores considered (aMEDr and mMEDr, with and without alcohol) were compared using Akaike's Information Criterion (AIC; ref. 34). Because of the equal or better performance of aMEDr compared with mMEDr in both the current and previous NLCS analyses (35–37), the Results section of this article mainly focuses on associations with aMEDr and subgroup analyses were only performed using this score. We preferred the aMED variant without alcohol (aMEDr), because alcohol consumption is a risk factor for several types of cancer (29).

Potential effect modification by sex (bladder cancer only), cigarette smoking status (bladder cancer only), alcohol consumption, BMI, educational level, and family history of prostate/bladder cancer was explored by testing the statistical significance of interaction terms between these factors and aMEDr. In addition, HRs were estimated for strata of the potential effect modifying factors. For prostate cancer risk, we estimated associations with the MD scores within time periods before (1986–1994) and after (1995–2006) the introduction of PSA testing in clinical practice in the Netherlands. Furthermore, the effect of excluding the first 2 years of follow-up was evaluated.

Because of the use of cohort-specific cut-offs, the MD scores used measure relative levels of MD adherence. Therefore, we also evaluated associations of prostate and bladder cancer (subtypes) with an absolute score based on the dietary part of the 2007 cancer prevention recommendations published by the WCRF/AICR (38). The WCRF/AICR score used in this study includes the recommendations concerning intakes of foods and drinks that promote weight gain, plant foods, red and processed meats, alcohol, and salt. When possible, recommendations were operationalized as in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort (39, 40). For a detailed description of the scoring method and the absolute cut-offs used we refer to a previous NLCS article (35). The resulting sum score ranged from 0 to 5 points (lowest to highest adherence). In addition, we created a variant of the WCRF/AICR score without the alcohol recommendation that ranged from 0 to 4 points. Fully adjusted HRs and 95% CIs were estimated per SD increase in WCRF/AICR score. For comparison, we also estimated HRs and 95% CIs per SD-increment for the aMED variables (with and without alcohol). Finally, AIC was used to compare model fits of the WCRF/AICR score and aMED variables. All analyses were performed using Stata version 15. Two-sided P values <0.05 were considered statistically significant.

Baseline characteristics of NLCS of subcohort and cases of prostate and bladder cancer

Table 1 compares baseline characteristics of subcohort members with those of prostate and bladder cancer cases. The mean level of MD adherence was similar for cases and subcohort members, except for the slightly lower mean aMEDr value in female noninvasive bladder cancer cases. Compared with subcohort members, prostate cancer cases were less often current smokers, higher educated (particularly nonadvanced prostate cancer), more physically active (particularly nonadvanced prostate cancer), and more likely to have a family history of prostate cancer. When comparing nonadvanced with advanced cases at diagnosis, nonadvanced cases more frequently had a high level of education and were slightly more active. Bladder cancer cases were more often current smokers, consumed higher amounts of alcohol, and reported more commonly a family history of bladder cancer than subcohort members. Patterns were mostly comparable for noninvasive and invasive bladder cancer cases.

Table 1.

Baseline characteristics of subcohort members, and cases of prostate and bladder cancer in the Netherlands Cohort Study

MenWomen
Prostate cancer casesBladder cancer casesBladder cancer cases
SubcohortNonadvancedAdvancedNoninvasiveInvasiveSubcohortNoninvasiveInvasive
N = 2,057N = 2,397N = 1,294N = 913N = 831N = 2,027N = 140N = 165
aMEDr 3.9 (1.6) 4.1 (1.6) 4.0 (1.6) 3.9 (1.6) 3.8 (1.6) 4.0 (1.6) 3.7 (1.5) 4.0 (1.6) 
mMEDr 4.0 (1.5) 4.1 (1.5) 4.0 (1.5) 4.0 (1.5) 3.9 (1.5) 4.0 (1.5) 4.0 (1.3) 4.0 (1.4) 
Age (years)a 61 (7) 62 (7) 62 (7) 62 (7) 62 (6) 61 (7) 61 (7) 62 (7) 
Current cigarette smokers (%) 35.1 30.1 30.8 41.1 45.0 21.3 35.7 28.5 
Cigarette smoking frequency (cig/day)a,b 15 (10) 15 (10) 15 (10) 15 (10) 20 (15) 10 (13) 12 (15) 12 (14) 
Cigarette smoking duration (years)a,b 36 (17) 34 (19) 34 (18) 39 (16) 40 (15) 30 (20) 35 (19) 35 (19) 
Higher vocational education or university (%) 19.3 24.4 20.9 20.1 19.3 9.5 7.1 8.5 
Alcohol consumption (g/day)a 9.7 (20.9) 10.3 (21.3) 10.5 (20.1) 12.1 (21.7) 12.6 (23.9) 1.6 (7.8) 2.5 (8.7) 2.1 (11.1) 
Daily energy intake (kcal) 2162 (501) 2162 (483) 2164 (492) 2154 (475) 2161 (482) 1687 (392) 1621 (357) 1662 (366) 
Body mass index (kg/m224.9 (2.6) 24.9 (2.5) 25.0 (2.5) 25.1 (2.6) 24.9 (2.5) 25.0 (3.5) 25.1 (3.6) 25.1 (3.8) 
Non-occupational physical activity (min/day)a 62.1 (67.1) 66.4 (62.1) 64.3 (65.7) 62.5 (64.3) 63.6 (67.1) 54.3 (52.9) 53.6 (60.0) 55.7 (60.0) 
Family history of prostate cancer (%) 2.4 3.6 3.4 NA NA NA NA NA 
Family history of bladder cancer (%) 0.7 NA NA 0.9 1.0 1.4 2.9 1.8 
MenWomen
Prostate cancer casesBladder cancer casesBladder cancer cases
SubcohortNonadvancedAdvancedNoninvasiveInvasiveSubcohortNoninvasiveInvasive
N = 2,057N = 2,397N = 1,294N = 913N = 831N = 2,027N = 140N = 165
aMEDr 3.9 (1.6) 4.1 (1.6) 4.0 (1.6) 3.9 (1.6) 3.8 (1.6) 4.0 (1.6) 3.7 (1.5) 4.0 (1.6) 
mMEDr 4.0 (1.5) 4.1 (1.5) 4.0 (1.5) 4.0 (1.5) 3.9 (1.5) 4.0 (1.5) 4.0 (1.3) 4.0 (1.4) 
Age (years)a 61 (7) 62 (7) 62 (7) 62 (7) 62 (6) 61 (7) 61 (7) 62 (7) 
Current cigarette smokers (%) 35.1 30.1 30.8 41.1 45.0 21.3 35.7 28.5 
Cigarette smoking frequency (cig/day)a,b 15 (10) 15 (10) 15 (10) 15 (10) 20 (15) 10 (13) 12 (15) 12 (14) 
Cigarette smoking duration (years)a,b 36 (17) 34 (19) 34 (18) 39 (16) 40 (15) 30 (20) 35 (19) 35 (19) 
Higher vocational education or university (%) 19.3 24.4 20.9 20.1 19.3 9.5 7.1 8.5 
Alcohol consumption (g/day)a 9.7 (20.9) 10.3 (21.3) 10.5 (20.1) 12.1 (21.7) 12.6 (23.9) 1.6 (7.8) 2.5 (8.7) 2.1 (11.1) 
Daily energy intake (kcal) 2162 (501) 2162 (483) 2164 (492) 2154 (475) 2161 (482) 1687 (392) 1621 (357) 1662 (366) 
Body mass index (kg/m224.9 (2.6) 24.9 (2.5) 25.0 (2.5) 25.1 (2.6) 24.9 (2.5) 25.0 (3.5) 25.1 (3.6) 25.1 (3.8) 
Non-occupational physical activity (min/day)a 62.1 (67.1) 66.4 (62.1) 64.3 (65.7) 62.5 (64.3) 63.6 (67.1) 54.3 (52.9) 53.6 (60.0) 55.7 (60.0) 
Family history of prostate cancer (%) 2.4 3.6 3.4 NA NA NA NA NA 
Family history of bladder cancer (%) 0.7 NA NA 0.9 1.0 1.4 2.9 1.8 

NOTE: The % missing values in the total eligible population was <5% for all variables included in this table, with the exception of cigarette smoking frequency in men. Mean (SD) values are reported unless otherwise specified.

aMedian (IQR) values are reported.

bMedian (IQR) values for frequency and duration of smoking were based on former and current smokers.

Mediterranean diet adherence and risks of prostate and bladder cancer

Fully adjusted associations of MD adherence with prostate and bladder cancer incidence are presented in Tables 2 and 3, respectively. For age- (and sex-)adjusted associations we refer to Supplementary Tables S1 (prostate) and S2 (bladder). The number of subjects included in the Cox models is slightly smaller than the number of eligible participants due to missing values in covariates.

Table 2.

Fully adjusted associations of aMED and mMED (including and excluding alcohol) with prostate cancer risk in the Netherlands Cohort Study

Prostate cancer
AllNonadvancedAdvanced
PYsubcohortCasesHR (95% CI)aAICCasesHR (95% CI)aAICCasesHR (95% CI)aAIC
aMEDr 
 0–3 12,637 1,473 1.00  854 1.00  497 1.00  
 4–5 13,363 1,713 1.09 (0.96–1.24)  1,048 1.14 (0.99–1.31)  545 1.03 (0.88–1.21)  
 6–8 4,817 682 1.14 (0.96–1.36) 55,789 427 1.22 (1.01–1.48) 33,513 214 1.08 (0.87–1.35) 18,198 
Ptrend   0.139   0.037   0.483  
 Continuous, per 2 pts 30,816 3,868 1.09 (1.01–1.17) 55,780 2,329 1.12 (1.04–1.22) 33,505 1,256 1.06 (0.96–1.17) 18,195 
aMEDb 
 0–3 9,833 1,156 1.00  666 1.00  392 1.00  
 4–5 13,722 1,661 1.03 (0.89–1.18)  1,008 1.07 (0.92–1.25)  528 0.97 (0.81–1.15)  
 6–9 7,261 1,051 1.19 (1.02–1.40) 55,778 655 1.28 (1.08–1.53) 33,505 336 1.14 (0.93–1.39) 18,192 
Ptrend   0.065   0.012   0.376  
 Continuous, per 2 pts 30,816 3,868 1.09 (1.01–1.17) 55,777 2,329 1.12 (1.04–1.21) 33,503 1,256 1.07 (0.98–1.17) 18,192 
mMEDr 
 0–3 11,536 1,329 1.00  790 1.00  437 1.00  
 4–5 14,176 1,857 1.10 (0.97–1.26)  1,111 1.11 (0.96–1.28)  614 1.12 (0.95–1.32)  
 6–8 5,104 682 1.10 (0.92–1.31) 55,790 428 1.17 (0.97–1.41) 33,519 205 1.01 (0.81–1.26) 18,196 
Ptrend   0.285   0.111   0.889  
 Continuous, per 2 pts 30,816 3,868 1.07 (0.99–1.16) 55,787 2,329 1.11 (1.02–1.21) 33,511 1,256 1.02 (0.92–1.13) 18,197 
mMEDb 
 0–3 8,128 955 1.00  571 1.00  310 1.00  
 4–5 14,828 1,817 1.00 (0.87–1.16)  1,069 0.99 (0.84–1.16)  612 1.05 (0.87–1.26)  
 6–9 7,860 1,096 1.12 (0.95–1.32) 55,788 689 1.18 (0.99–1.41) 33,512 334 1.07 (0.87–1.31) 18,197 
Ptrend   0.122   0.034   0.577  
 Continuous, per 2 pts 30,816 3,868 1.07 (1.00–1.16) 55,784 2,329 1.11 (1.02–1.20) 33,509 1,256 1.04 (0.95–1.14) 18,195 
Prostate cancer
AllNonadvancedAdvanced
PYsubcohortCasesHR (95% CI)aAICCasesHR (95% CI)aAICCasesHR (95% CI)aAIC
aMEDr 
 0–3 12,637 1,473 1.00  854 1.00  497 1.00  
 4–5 13,363 1,713 1.09 (0.96–1.24)  1,048 1.14 (0.99–1.31)  545 1.03 (0.88–1.21)  
 6–8 4,817 682 1.14 (0.96–1.36) 55,789 427 1.22 (1.01–1.48) 33,513 214 1.08 (0.87–1.35) 18,198 
Ptrend   0.139   0.037   0.483  
 Continuous, per 2 pts 30,816 3,868 1.09 (1.01–1.17) 55,780 2,329 1.12 (1.04–1.22) 33,505 1,256 1.06 (0.96–1.17) 18,195 
aMEDb 
 0–3 9,833 1,156 1.00  666 1.00  392 1.00  
 4–5 13,722 1,661 1.03 (0.89–1.18)  1,008 1.07 (0.92–1.25)  528 0.97 (0.81–1.15)  
 6–9 7,261 1,051 1.19 (1.02–1.40) 55,778 655 1.28 (1.08–1.53) 33,505 336 1.14 (0.93–1.39) 18,192 
Ptrend   0.065   0.012   0.376  
 Continuous, per 2 pts 30,816 3,868 1.09 (1.01–1.17) 55,777 2,329 1.12 (1.04–1.21) 33,503 1,256 1.07 (0.98–1.17) 18,192 
mMEDr 
 0–3 11,536 1,329 1.00  790 1.00  437 1.00  
 4–5 14,176 1,857 1.10 (0.97–1.26)  1,111 1.11 (0.96–1.28)  614 1.12 (0.95–1.32)  
 6–8 5,104 682 1.10 (0.92–1.31) 55,790 428 1.17 (0.97–1.41) 33,519 205 1.01 (0.81–1.26) 18,196 
Ptrend   0.285   0.111   0.889  
 Continuous, per 2 pts 30,816 3,868 1.07 (0.99–1.16) 55,787 2,329 1.11 (1.02–1.21) 33,511 1,256 1.02 (0.92–1.13) 18,197 
mMEDb 
 0–3 8,128 955 1.00  571 1.00  310 1.00  
 4–5 14,828 1,817 1.00 (0.87–1.16)  1,069 0.99 (0.84–1.16)  612 1.05 (0.87–1.26)  
 6–9 7,860 1,096 1.12 (0.95–1.32) 55,788 689 1.18 (0.99–1.41) 33,512 334 1.07 (0.87–1.31) 18,197 
Ptrend   0.122   0.034   0.577  
 Continuous, per 2 pts 30,816 3,868 1.07 (1.00–1.16) 55,784 2,329 1.11 (1.02–1.20) 33,509 1,256 1.04 (0.95–1.14) 18,195 

Abbreviation: PYsubcohort, person-years in the subcohort.

aAdjusted for age at baseline (years), body mass index (kg/m2), alcohol consumption (g/day), daily energy intake (kcal), highest level of education (primary school or lower vocational, secondary school or medium vocational, higher vocational or university), and family history of prostate cancer (no, yes).

bNot adjusted for alcohol consumption.

Table 3.

Fully adjusted associations of aMED and mMED (including and excluding alcohol) with bladder cancer risk in the Netherlands Cohort Study

Bladder cancer
AllNoninvasiveInvasive
PYsubcohortCasesHR (95% CI)aAICCasesHR (95% CI)aAICCasesHR (95% CI)aAIC
aMEDr 
 0–3 24,302 790 1.00  398 1.00  392 1.00  
 4–5 28,005 796 0.95 (0.82–1.09)  418 0.98 (0.83–1.17)  378 0.91 (0.76–1.08)  
 6–8 11,055 298 1.00 (0.83–1.21) 28,540 156 1.02 (0.81–1.28) 14,711 142 0.98 (0.77–1.24) 13,840 
Ptrend   0.931   0.898   0.783  
 Continuous, per 2 pts 63,362 1,884 1.00 (0.92–1.09) 28,540 972 1.01 (0.92–1.12) 14,709 912 0.99 (0.89–1.09) 13,840 
aMEDb 
 0–3 20,107 646 1.00  316 1.00  330 1.00  
 4–5 28,038 795 0.90 (0.78–1.05)  422 0.98 (0.81–1.17)  373 0.84 (0.70–1.00)  
 6–9 15,217 443 0.99 (0.83–1.18) 28,544 234 1.05 (0.85–1.29) 14,712 209 0.93 (0.75–1.16) 13,840 
Ptrend   0.591   0.850   0.279  
 Continuous, per 2 pts 63,362 1,884 1.00 (0.93–1.08) 28,546 972 1.03 (0.94–1.14) 14,710 912 0.97 (0.88–1.07) 13,843 
mMEDr 
 0–3 22,721 725 1.00  360 1.00  365 1.00  
 4–5 30,384 882 0.98 (0.85–1.13)  465 1.04 (0.87–1.23)  417 0.92 (0.77–1.10)  
 6–8 10,256 277 0.90 (0.74–1.09) 28,539 147 0.94 (0.74–1.19) 14,710 130 0.85 (0.67–1.09) 13,839 
Ptrend   0.273   0.645   0.182  
 Continuous, per 2 pts 63,362 1,884 0.97 (0.89–1.06) 28,539 972 1.00 (0.90–1.11) 14,709 912 0.95 (0.85–1.06) 13,839 
mMEDb 
 0–3 17,618 531 1.00  265 1.00  266 1.00  
 4–5 30,953 888 0.91 (0.78–1.06)  462 0.95 (0.78–1.14)  426 0.87 (0.72–1.05)  
 6–9 14,791 465 0.95 (0.80–1.14) 28,545 245 0.99 (0.80–1.23) 14,712 220 0.91 (0.73–1.13) 13,842 
Ptrend   0.730   0.954   0.533  
 Continuous, per 2 pts 63,362 1,884 0.99 (0.91–1.08) 28,546 972 1.03 (0.93–1.13) 14,710 912 0.96 (0.87–1.07) 13,842 
Bladder cancer
AllNoninvasiveInvasive
PYsubcohortCasesHR (95% CI)aAICCasesHR (95% CI)aAICCasesHR (95% CI)aAIC
aMEDr 
 0–3 24,302 790 1.00  398 1.00  392 1.00  
 4–5 28,005 796 0.95 (0.82–1.09)  418 0.98 (0.83–1.17)  378 0.91 (0.76–1.08)  
 6–8 11,055 298 1.00 (0.83–1.21) 28,540 156 1.02 (0.81–1.28) 14,711 142 0.98 (0.77–1.24) 13,840 
Ptrend   0.931   0.898   0.783  
 Continuous, per 2 pts 63,362 1,884 1.00 (0.92–1.09) 28,540 972 1.01 (0.92–1.12) 14,709 912 0.99 (0.89–1.09) 13,840 
aMEDb 
 0–3 20,107 646 1.00  316 1.00  330 1.00  
 4–5 28,038 795 0.90 (0.78–1.05)  422 0.98 (0.81–1.17)  373 0.84 (0.70–1.00)  
 6–9 15,217 443 0.99 (0.83–1.18) 28,544 234 1.05 (0.85–1.29) 14,712 209 0.93 (0.75–1.16) 13,840 
Ptrend   0.591   0.850   0.279  
 Continuous, per 2 pts 63,362 1,884 1.00 (0.93–1.08) 28,546 972 1.03 (0.94–1.14) 14,710 912 0.97 (0.88–1.07) 13,843 
mMEDr 
 0–3 22,721 725 1.00  360 1.00  365 1.00  
 4–5 30,384 882 0.98 (0.85–1.13)  465 1.04 (0.87–1.23)  417 0.92 (0.77–1.10)  
 6–8 10,256 277 0.90 (0.74–1.09) 28,539 147 0.94 (0.74–1.19) 14,710 130 0.85 (0.67–1.09) 13,839 
Ptrend   0.273   0.645   0.182  
 Continuous, per 2 pts 63,362 1,884 0.97 (0.89–1.06) 28,539 972 1.00 (0.90–1.11) 14,709 912 0.95 (0.85–1.06) 13,839 
mMEDb 
 0–3 17,618 531 1.00  265 1.00  266 1.00  
 4–5 30,953 888 0.91 (0.78–1.06)  462 0.95 (0.78–1.14)  426 0.87 (0.72–1.05)  
 6–9 14,791 465 0.95 (0.80–1.14) 28,545 245 0.99 (0.80–1.23) 14,712 220 0.91 (0.73–1.13) 13,842 
Ptrend   0.730   0.954   0.533  
 Continuous, per 2 pts 63,362 1,884 0.99 (0.91–1.08) 28,546 972 1.03 (0.93–1.13) 14,710 912 0.96 (0.87–1.07) 13,842 

Abbreviation: PYsubcohort, person-years in the subcohort.

aAdjusted for age at baseline (years), sex (men, women), cigarette smoking status (never, former, current), cigarette smoking frequency (cigarettes smoked per day, centered), cigarette smoking duration (years, centered), body mass index (kg/m2), alcohol consumption (g/day), daily energy intake (kcal), highest level of education (primary school or lower vocational, secondary school or medium vocational, higher vocational or university), and family history of bladder cancer (no, yes).

bNot adjusted for alcohol consumption.

Prostate cancer.

Higher MD adherence, measured by aMEDr, was associated with an increased risk of prostate cancer [HRper 2-point increment (95% CI): 1.09 (1.01–1.17); Table 2], although this positive association was mainly present in nonadvanced cases. For nonadvanced prostate cancer risk, the HR (95% CI) comparing high to low aMEDr values was 1.22 (1.01–1.48) with a significant test for trend (P = 0.04). However, aMEDr was not significantly associated with the risk of advanced prostate cancer [HRper 2-point increment (95% CI): 1.06 (0.96–1.17)]. Despite this difference between the prostate cancer subtypes, tests for heterogeneity were not statistically significant. Associations of similar directions were observed when MD adherence was assessed using mMEDr and inclusion of alcohol in the MD scores did not notably change the results. Comparison of model performances showed equal or better (nonadvanced prostate cancer) fits for aMEDr compared with mMEDr (Table 2). In addition, model fits were generally better for scores with alcohol than scores without alcohol.

Associations of aMEDr with risks of nonadvanced and advanced prostate cancer did not significantly differ across strata of potential effect modifiers (Table 4). Nevertheless, increasing aMEDr was associated with a significantly increased risk of both prostate cancer subtypes among men in the highest education category, whereas there was no clear evidence of an association in the other education categories. Results were comparable after exclusion of the first 2 years of follow-up [HRper 2-point increment in aMEDr: 1.09 (total), 1.13 (nonadvanced), 1.06 (advanced)]. In addition, the strength of the associations did not significantly differ before (1986–1994) and after (1995–2006) the introduction of PSA testing in clinical practice in the Netherlands (Table 4). However, the positive association between aMEDr and nonadvanced prostate cancer risk was only statistically significant in the late period (1995–2006).

Table 4.

Fully adjusted associations of aMEDr (per two-point increment) with prostate cancer risk for various subgroups in the Netherlands Cohort Study

Prostate cancer
AllNonadvancedAdvanced
CasesHR (95% CI)a,bCasesHR (95% CI)a,bCasesHR (95% CI)a,b
Overall 3,868 1.09 (1.01–1.17) 2,329 1.12 (1.04–1.22) 1,256 1.06 (0.96–1.17) 
Alcohol consumptionc 
 0 g/day 488 1.11 (0.91–1.35) 291 1.10 (0.89–1.37) 154 1.22 (0.93–1.61) 
 >0–<15.0 g/day 1,891 1.15 (1.03–1.28) 1,135 1.20 (1.06–1.36) 628 1.08 (0.94–1.24) 
 ≥15.0 g/day 1,489 1.02 (0.90–1.16) 903 1.07 (0.93–1.22) 474 0.99 (0.85–1.15) 
Pinteractiond  0.484  0.506  0.387 
Body mass indexe 
 ≥18.5–<25.0 kg/m2 2,054 1.08 (0.97–1.20) 1,256 1.12 (1.00–1.25) 646 1.07 (0.94–1.22) 
 ≥25.0 kg/m2 1,804 1.10 (0.98–1.23) 1,067 1.14 (1.01–1.29) 608 1.05 (0.92–1.21) 
Pinteractiond  0.699  0.623  0.897 
Highest level of educationf 
 Primary school or lower vocational 1,572 1.03 (0.92–1.17) 929 1.07 (0.93–1.22) 525 0.97 (0.84–1.13) 
 Secondary school or medium vocational 1,405 1.06 (0.95–1.20) 827 1.12 (0.98–1.27) 470 1.05 (0.91–1.22) 
 Higher vocational or university 891 1.24 (1.04–1.47) 573 1.26 (1.04–1.52) 261 1.27 (1.01–1.60) 
Pinteractiond  0.235  0.362  0.188 
Family history of prostate cancerg 
 No 3,731 1.10 (1.02–1.18) 2,244 1.14 (1.05–1.24) 1,214 1.07 (0.97–1.18) 
 Yes 137 0.85 (0.53–1.38) 85 0.84 (0.50–1.41) 42 0.85 (0.47–1.54) 
Pinteractiond  0.191  0.164  0.363 
Follow-up period 
 1986–1994 985 1.06 (0.96–1.18) 455 1.08 (0.95–1.24) 397 1.09 (0.95–1.25) 
 1995–2006 2,883 1.09 (1.01–1.19) 1,874 1.13 (1.04–1.24) 859 1.05 (0.94–1.17) 
Pinteractiond  0.611  0.456  0.440 
Prostate cancer
AllNonadvancedAdvanced
CasesHR (95% CI)a,bCasesHR (95% CI)a,bCasesHR (95% CI)a,b
Overall 3,868 1.09 (1.01–1.17) 2,329 1.12 (1.04–1.22) 1,256 1.06 (0.96–1.17) 
Alcohol consumptionc 
 0 g/day 488 1.11 (0.91–1.35) 291 1.10 (0.89–1.37) 154 1.22 (0.93–1.61) 
 >0–<15.0 g/day 1,891 1.15 (1.03–1.28) 1,135 1.20 (1.06–1.36) 628 1.08 (0.94–1.24) 
 ≥15.0 g/day 1,489 1.02 (0.90–1.16) 903 1.07 (0.93–1.22) 474 0.99 (0.85–1.15) 
Pinteractiond  0.484  0.506  0.387 
Body mass indexe 
 ≥18.5–<25.0 kg/m2 2,054 1.08 (0.97–1.20) 1,256 1.12 (1.00–1.25) 646 1.07 (0.94–1.22) 
 ≥25.0 kg/m2 1,804 1.10 (0.98–1.23) 1,067 1.14 (1.01–1.29) 608 1.05 (0.92–1.21) 
Pinteractiond  0.699  0.623  0.897 
Highest level of educationf 
 Primary school or lower vocational 1,572 1.03 (0.92–1.17) 929 1.07 (0.93–1.22) 525 0.97 (0.84–1.13) 
 Secondary school or medium vocational 1,405 1.06 (0.95–1.20) 827 1.12 (0.98–1.27) 470 1.05 (0.91–1.22) 
 Higher vocational or university 891 1.24 (1.04–1.47) 573 1.26 (1.04–1.52) 261 1.27 (1.01–1.60) 
Pinteractiond  0.235  0.362  0.188 
Family history of prostate cancerg 
 No 3,731 1.10 (1.02–1.18) 2,244 1.14 (1.05–1.24) 1,214 1.07 (0.97–1.18) 
 Yes 137 0.85 (0.53–1.38) 85 0.84 (0.50–1.41) 42 0.85 (0.47–1.54) 
Pinteractiond  0.191  0.164  0.363 
Follow-up period 
 1986–1994 985 1.06 (0.96–1.18) 455 1.08 (0.95–1.24) 397 1.09 (0.95–1.25) 
 1995–2006 2,883 1.09 (1.01–1.19) 1,874 1.13 (1.04–1.24) 859 1.05 (0.94–1.17) 
Pinteractiond  0.611  0.456  0.440 

aAll HRs were estimated per two-point increment in aMEDr.

bAdjusted for age at baseline (years), body mass index (kg/m2), alcohol consumption (g/day), daily energy intake (kcal), highest level of education (primary school or lower vocational, secondary school or medium vocational, higher vocational or university), and family history of prostate cancer (no, yes).

cNot adjusted for alcohol consumption.

dP-values for interaction were obtained by testing the statistical significance of interaction terms between aMEDr and the stratifying covariates in fully adjusted models.

eNot adjusted for body mass index.

fNot adjusted for highest level of education.

gNot adjusted for family history of prostate cancer.

Bladder cancer.

MD adherence was not significantly associated with bladder cancer risk, regardless of the MD score used (Table 3). HRs (95% CIs) per 2-point increase in aMEDr were 1.00 (0.92–1.09), 1.01 (0.92–1.12), and 0.99 (0.89–1.09) for total, noninvasive, and invasive bladder cancer, respectively. In contrast to the fully adjusted analyses, inverse trends (not always significant) seemed to be present between MD adherence and risks of total and invasive bladder cancer in age- and sex-adjusted analyses (P = 0.04 for mMED and invasive bladder cancer; Supplementary Table S2). Comparable performances were observed for models containing aMEDr and mMEDr (Table 3). Furthermore, MD scores without alcohol fitted equally or better than their equivalents including alcohol.

There was no evidence of effect modification by sex, cigarette smoking status, alcohol consumption, educational level, and family history of bladder cancer (Table 5). Higher aMEDr values seemed to be associated with a nonsignificantly decreased risk of bladder cancer in subjects with a normal BMI, but not in overweight or obese subjects [Pinteraction = 0.026 (total), 0.037 (noninvasive), 0.128 (invasive)]. Associations did not essentially change after exclusion of the first 2 years of follow-up [HRper 2-point increment in aMEDr: 1.01 (total), 1.02 (noninvasive), 1.00 (invasive)].

Table 5.

Fully adjusted associations of aMEDr (per two-point increment) with bladder cancer risk for various subgroups in the Netherlands Cohort Study

Bladder cancer
AllNoninvasiveInvasive
CasesHR (95% CI)a,bCasesHR (95% CI)a,bCasesHR (95% CI)a,b
Overall 1,884 1.00 (0.92–1.09) 972 1.01 (0.92–1.12) 912 0.99 (0.89–1.09) 
Sexc 
 Men 1,593 1.01 (0.92–1.11) 840 1.04 (0.93–1.16) 753 0.98 (0.87–1.10) 
 Women 291 0.94 (0.79–1.10) 132 0.86 (0.69–1.08) 159 1.00 (0.81–1.25) 
Pinteractiond  0.472  0.091  0.622 
Cigarette smoking statuse 
 Never 271 0.99 (0.82–1.18) 137 0.93 (0.73–1.18) 134 1.05 (0.83–1.33) 
 Former 855 1.09 (0.97–1.22) 457 1.10 (0.96–1.27) 398 1.07 (0.92–1.24) 
 Current 758 0.93 (0.80–1.08) 378 0.98 (0.82–1.16) 380 0.89 (0.74–1.07) 
Pinteractiond  0.206  0.457  0.151 
Alcohol consumptionf 
 0 g/day 234 1.00 (0.82–1.22) 111 0.91 (0.69–1.20) 123 1.09 (0.86–1.38) 
 >0–<15.0 g/day 888 1.01 (0.90–1.13) 475 1.03 (0.89–1.18) 413 0.98 (0.85–1.14) 
 ≥15.0 g/day 762 0.96 (0.84–1.11) 386 1.00 (0.84–1.18) 376 0.93 (0.78–1.11) 
Pinteractiond  0.988  0.807  0.661 
Body mass indexg 
 ≥18.5–<25.0 kg/m2 1,019 0.92 (0.82–1.03) 513 0.91 (0.79–1.05) 506 0.92 (0.80–1.06) 
 ≥25.0 kg/m2 857 1.10 (0.97–1.24) 455 1.12 (0.97–1.30) 402 1.07 (0.92–1.26) 
Pinteractiond  0.026  0.037  0.128 
Highest level of educationh 
 Primary school or lower vocational 816 0.91 (0.80–1.04) 410 0.94 (0.80–1.11) 406 0.87 (0.74–1.03) 
 Secondary school or medium vocational 728 1.03 (0.90–1.16) 384 1.01 (0.87–1.18) 344 1.04 (0.89–1.21) 
 Higher vocational or university 340 1.20 (0.97–1.48) 178 1.20 (0.92–1.57) 162 1.20 (0.92–1.56) 
Pinteractiond  0.056  0.303  0.055 
Family history of bladder canceri,j 
 No 1,862 1.01 (0.93–1.09)     
 Yes 22 0.34 (0.05–2.34)     
Pinteractiond  0.094     
Bladder cancer
AllNoninvasiveInvasive
CasesHR (95% CI)a,bCasesHR (95% CI)a,bCasesHR (95% CI)a,b
Overall 1,884 1.00 (0.92–1.09) 972 1.01 (0.92–1.12) 912 0.99 (0.89–1.09) 
Sexc 
 Men 1,593 1.01 (0.92–1.11) 840 1.04 (0.93–1.16) 753 0.98 (0.87–1.10) 
 Women 291 0.94 (0.79–1.10) 132 0.86 (0.69–1.08) 159 1.00 (0.81–1.25) 
Pinteractiond  0.472  0.091  0.622 
Cigarette smoking statuse 
 Never 271 0.99 (0.82–1.18) 137 0.93 (0.73–1.18) 134 1.05 (0.83–1.33) 
 Former 855 1.09 (0.97–1.22) 457 1.10 (0.96–1.27) 398 1.07 (0.92–1.24) 
 Current 758 0.93 (0.80–1.08) 378 0.98 (0.82–1.16) 380 0.89 (0.74–1.07) 
Pinteractiond  0.206  0.457  0.151 
Alcohol consumptionf 
 0 g/day 234 1.00 (0.82–1.22) 111 0.91 (0.69–1.20) 123 1.09 (0.86–1.38) 
 >0–<15.0 g/day 888 1.01 (0.90–1.13) 475 1.03 (0.89–1.18) 413 0.98 (0.85–1.14) 
 ≥15.0 g/day 762 0.96 (0.84–1.11) 386 1.00 (0.84–1.18) 376 0.93 (0.78–1.11) 
Pinteractiond  0.988  0.807  0.661 
Body mass indexg 
 ≥18.5–<25.0 kg/m2 1,019 0.92 (0.82–1.03) 513 0.91 (0.79–1.05) 506 0.92 (0.80–1.06) 
 ≥25.0 kg/m2 857 1.10 (0.97–1.24) 455 1.12 (0.97–1.30) 402 1.07 (0.92–1.26) 
Pinteractiond  0.026  0.037  0.128 
Highest level of educationh 
 Primary school or lower vocational 816 0.91 (0.80–1.04) 410 0.94 (0.80–1.11) 406 0.87 (0.74–1.03) 
 Secondary school or medium vocational 728 1.03 (0.90–1.16) 384 1.01 (0.87–1.18) 344 1.04 (0.89–1.21) 
 Higher vocational or university 340 1.20 (0.97–1.48) 178 1.20 (0.92–1.57) 162 1.20 (0.92–1.56) 
Pinteractiond  0.056  0.303  0.055 
Family history of bladder canceri,j 
 No 1,862 1.01 (0.93–1.09)     
 Yes 22 0.34 (0.05–2.34)     
Pinteractiond  0.094     

aAll HRs were estimated per two-point increment in aMEDr.

bAdjusted for age at baseline (years), sex (men, women), cigarette smoking status (never, former, current), cigarette smoking frequency (cigarettes smoked per day, centered), cigarette smoking duration (years, centered), body mass index (kg/m2), alcohol consumption (g/day), daily energy intake (kcal), highest level of education (primary school or lower vocational, secondary school or medium vocational, higher vocational or university), and family history of bladder cancer (no, yes).

cNot adjusted for sex.

dP-values for interaction were obtained by testing the statistical significance of interaction terms between aMEDr and the stratifying covariates in fully adjusted models.

eNot adjusted for cigarette smoking status.

fNot adjusted for alcohol consumption.

gNot adjusted for body mass index.

hNot adjusted for highest level of education.

iThe analyses stratified by family history of bladder cancer were only performed for total bladder cancer risk because of the low number of cases with a positive family history.

jNot adjusted for family history of bladder cancer.

Dietary WCRF/AICR recommendations and risks of prostate and bladder cancer

Because values of aMED indices are population-dependent, we compared these indices to absolute WCRF/AICR scores (Table 6). The WCRF/AICR scores were not significantly associated with prostate and bladder cancer risk, but as with the aMED indices, associations with prostate cancer risk were in the positive direction. For prostate cancer risk, WCRF/AICR scores had worse model fits compared with aMED indices, particularly when considering the nonadvanced subtype (Supplementary Table S3). Comparable model performances were observed for bladder cancer risk (Supplementary Table S3).

Table 6.

Fully adjusted associations of the absolute WCRF/AICR score and aMED (per SD-increment) with risks of prostate and bladder cancer in the Netherlands Cohort Study

Prostate cancerBladder cancer
AllNonadvancedAdvancedAllNoninvasiveInvasive
HRSD (95% CI)a,bHRSD (95% CI)a,bHRSD (95% CI)a,bHRSD (95% CI)a,cHRSD (95% CI)a,cHRSD (95% CI)a,c
PYsubcohort/casesd 30,049/3,763 30,049/2,272 30,049/1,212 61,976/1,836 61,976/945 61,976/891 
Excluding alcohol 
 WCRF/AICR scoree 1.03 (0.97–1.09) 1.03 (0.97–1.10) 1.05 (0.97–1.13) 0.98 (0.92–1.05) 0.99 (0.91–1.07) 0.98 (0.90–1.06) 
 aMEDr 1.07 (1.01–1.13) 1.09 (1.03–1.17) 1.05 (0.98–1.14) 0.99 (0.93–1.06) 1.00 (0.93–1.09) 0.97 (0.90–1.06) 
Including alcohol 
 WCRF/AICR scoree,f 1.04 (0.97–1.10) 1.04 (0.98–1.11) 1.04 (0.97–1.13) 0.95 (0.89–1.02) 0.97 (0.90–1.05) 0.94 (0.86–1.02) 
 aMEDf 1.07 (1.01–1.14) 1.10 (1.03–1.17) 1.07 (0.99–1.15) 0.99 (0.93–1.06) 1.02 (0.95–1.11) 0.96 (0.89–1.04) 
Prostate cancerBladder cancer
AllNonadvancedAdvancedAllNoninvasiveInvasive
HRSD (95% CI)a,bHRSD (95% CI)a,bHRSD (95% CI)a,bHRSD (95% CI)a,cHRSD (95% CI)a,cHRSD (95% CI)a,c
PYsubcohort/casesd 30,049/3,763 30,049/2,272 30,049/1,212 61,976/1,836 61,976/945 61,976/891 
Excluding alcohol 
 WCRF/AICR scoree 1.03 (0.97–1.09) 1.03 (0.97–1.10) 1.05 (0.97–1.13) 0.98 (0.92–1.05) 0.99 (0.91–1.07) 0.98 (0.90–1.06) 
 aMEDr 1.07 (1.01–1.13) 1.09 (1.03–1.17) 1.05 (0.98–1.14) 0.99 (0.93–1.06) 1.00 (0.93–1.09) 0.97 (0.90–1.06) 
Including alcohol 
 WCRF/AICR scoree,f 1.04 (0.97–1.10) 1.04 (0.98–1.11) 1.04 (0.97–1.13) 0.95 (0.89–1.02) 0.97 (0.90–1.05) 0.94 (0.86–1.02) 
 aMEDf 1.07 (1.01–1.14) 1.10 (1.03–1.17) 1.07 (0.99–1.15) 0.99 (0.93–1.06) 1.02 (0.95–1.11) 0.96 (0.89–1.04) 

Abbreviation: PYsubcohort, person-years in the subcohort.

aHRs were estimated per SD-increment in the scores.

bAdjusted for age at baseline (years), body mass index (kg/m2), alcohol consumption (g/day), daily energy intake (kcal), highest level of education (primary school or lower vocational, secondary school or medium vocational, higher vocational or university), and family history of prostate cancer (no, yes).

cAdjusted for age at baseline (years), sex (men, women), cigarette smoking status (never, former, current), cigarette smoking frequency (cigarettes smoked per day, centered), cigarette smoking duration (years, centered), body mass index (kg/m2), alcohol consumption (g/day), daily energy intake (kcal), highest level of education (primary school or lower vocational, secondary school or medium vocational, higher vocational or university), and family history of bladder cancer (no, yes).

dA lower number of subjects could be included in these analyses as a result of missing values for salt intake.

eScore based on WCRF/AICR dietary recommendations to prevent cancer.

fNot adjusted for alcohol consumption.

In the large prospective NLCS, a priori defined MD adherence (aMEDr) was associated with a significantly increased risk of nonadvanced prostate cancer. In contrast, no association was observed with advanced prostate cancer risk. MD adherence was not associated with risks of total, noninvasive, and invasive bladder cancer. Model fits were equal or better for aMEDr compared with mMEDr. In addition, inclusion of alcohol in the MD scores resulted in generally better model fits for prostate cancer risk, whereas the opposite was observed for bladder cancer risk. Finally, adherence to the dietary WCRF/AICR recommendations was not associated with risks of both prostate and bladder cancer.

Previously conducted cohort studies in the United States and Europe consistently found no association between a priori defined MD adherence and prostate cancer risk (13, 14, 41, 42). Similar results were obtained when focusing on advanced cases of prostate cancer specifically (13, 14). Case–control studies showed less consistent results. One study observed a significant inverse association between a priori defined MD adherence and prostate cancer risk, whereas no relation was present in another study (43, 44). The vulnerability of the case–control design to several types of bias, including recall and selection biases, could potentially explain this inconsistency. Furthermore, prospective cohort studies may also have some limitations. For example, reliance on a single assessment of dietary intake at baseline may lead to exposure misclassification and attenuated associations. Although exposure misclassification could have contributed to the null findings of the previously conducted cohort studies, some cohorts did have updated dietary information available during follow-up (14, 42).

Results of this study were partially in concordance with results of previous cohort studies. We found higher MD adherence to be significantly associated with an increased risk of nonadvanced prostate cancer, whereas there was no evidence of an association with advanced prostate cancer risk. Prostate cancer is a heterogeneous disease with potentially etiologically different subtypes that may differ in risk factors (2, 12, 45). The subgroup of nonadvanced cancers at diagnosis mainly encompasses relatively nonaggressive forms of prostate cancer that progress slowly and might never have become clinically relevant. Approximately half of the diagnosed prostate cancers were estimated to not have caused any harm if they had remained undiagnosed and untreated (45). The prevalence of undiagnosed prostate cancer in elderly men is high, in 47.3% of US White and European men above the age of 80 years incident prostate cancer was detected at autopsy (45). Because of the often indolent nature of nonadvanced prostate cancers, advanced prostate cancer is the subtype of our primary interest. In this study, MD adherence was not significantly associated with advanced prostate cancer risk.

The significant positive associations that we observed between MD adherence and nonadvanced prostate cancer risk could potentially be explained by differences in care-seeking behavior, screening attendance, and prostate cancer awareness related to education and lifestyle. Male NLCS subcohort members with higher MD adherence overall seemed to have a healthier lifestyle judged by lower levels of smoking and alcohol consumption, and higher levels of physical activity, and were higher educated (36). Higher educated men with a more health-conscious lifestyle may be more aware of prostate cancer and more prone to seek care or attend screenings, resulting in a larger number of nonadvanced prostate cancer diagnoses in this group, part of which never would have become clinically relevant and otherwise would have remained undiagnosed. Nonadvanced prostate cancer cases in our study were more physically active and higher educated than advanced cases and subcohort members, implying a healthier lifestyle. Furthermore, the positive association between MD adherence and nonadvanced prostate cancer risk was strongest and only statistically significant in the highest education category, which fits our hypothesis that nonadvanced prostate tumors are more likely to be detected in highly educated men. Finally, results of our stratified analyses by follow-up period showed that the positive association between MD adherence and nonadvanced prostate cancer risk only reached statistical significance in the period after the introduction of PSA testing in clinical practice in the Netherlands (1995–2006).

Similar to our findings for bladder cancer risk, a priori defined MD adherence was not significantly associated with total UCC risk in the prospective EPIC and Melbourne Collaborative Cohort Study (MCCS) cohorts (15, 16). Although not statistically significant, HR estimates for risks of total, nonaggressive, and aggressive UCC in EPIC were suggestive of an inverse association with MD adherence (15). In addition, higher MD adherence seemed to be associated with a decreased risk of invasive UCC (P = 0.06), but not superficial UCC, in the Australian MCCS (16). HR estimates in our study were much closer to unity. Residual confounding by cigarette smoking, which is the most important bladder cancer risk factor (4), could (partially) explain the nonsignificant inverse associations between MD adherence and (invasive) UCC risk that were observed in the previously conducted prospective studies. The adjustment for cigarette smoking seemed to be less comprehensive in these studies compared with this study and the inverse associations were restricted to (former and) current smokers. Besides, we also observed (non-significant) inverse associations with invasive bladder cancer risk when we only adjusted our models for age and sex. Very recently, a pooled analysis of 13 prospective cohorts showed that MD adherence was associated with a significantly decreased risk of total, non-muscle-invasive, and muscle-invasive bladder cancer (46). However, this study did not sufficiently adjust for smoking behavior. Finally, an Italian hospital-based case-control study found higher MD adherence to be associated with a significantly decreased UCC risk (47).

In this study, BMI significantly modified the association between MD adherence and risks of total and noninvasive bladder cancer, with nonsignificant inverse associations being observed among subjects with a normal BMI, but not among overweight or obese subjects. An inverse association has been suggested between BMI and levels of urinary 8-hydroxydeoxyguanosine, a marker of oxidative DNA damage, particularly in smokers (48). Therefore, subjects with a normal BMI potentially benefit most from the high antioxidant content (e.g., vitamins and polyphenols) of the MD (49, 50). However, this interaction could also be a chance finding. The interaction with BMI was not detected in the EPIC cohort (15) and requires attention in future research. The association of MD adherence with bladder cancer risk did not significantly differ across strata of other potential effect modifiers including sex and smoking status.

A major strength of the NLCS is its prospective design and the nearly complete follow-up of 20.3 years. The large number of prostate and bladder cancer cases allowed subtype-specific analyses based on tumor stage/invasiveness at diagnosis, extensive adjustment for confounding, and stratified analyses for potential effect modifying factors. Despite the comprehensive adjustment for cigarette smoking habits in the analyses concerning bladder cancer risk, residual confounding by smoking (bladder cancer) or unmeasured factors (prostate and bladder cancer) could still have affected our results. For example, we had no information about PSA testing. Nevertheless, associations of MD adherence with prostate cancer risk did not statistically significantly differ in time periods before and after the introduction of PSA testing in the Netherlands, making a relevant effect on our results unlikely. Moreover, we were not able to adjust the analyses concerning bladder cancer risk for environmental and occupational exposures to chemicals or exposure to arsenic in drinking water. A final strength of our study includes the high quality of the dietary data. The single baseline measurement of the NLCS-FFQ was shown to perform adequately when compared with 9-day dietary records and dietary habits were reproducible for over at least 5 years (19, 22). However, changes in dietary habits and confounding factors after baseline as well as measurement errors may have attenuated associations. The population-dependent assignment of scores is a weakness of the MD scores that we used to assess MD adherence. Therefore, high MD score values may not necessarily represent a truly Mediterranean way of eating, especially in the Netherlands and other non-Mediterranean countries. Nonetheless, largely similar results were obtained when we used absolute scores based on the WCRF/AICR dietary recommendations to prevent cancer. In addition, aMED-containing models performed equally well or better than models containing WCRF/AICR scores in our study population.

In conclusion, high adherence to the MD was not associated with decreased risks of advanced prostate cancer as well as total, noninvasive, and invasive bladder cancer in the NLCS. The positive associations that we observed between MD adherence and nonadvanced prostate cancer risk potentially resulted from higher levels of care-seeking behavior, screening attendance, and prostate cancer awareness in higher educated men with healthier lifestyles, causing a larger number of nonadvanced prostate cancer diagnoses in this group. Therefore, we recommend that future studies on dietary factors and prostate cancer risk also report results specifically for advanced prostate cancer.

No potential conflicts of interest were disclosed.

Conception and design: P.A. van den Brandt

Development of methodology: P.A. van den Brandt

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.A. van den Brandt

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Schulpen

Writing, review, and/or revision of the manuscript: M. Schulpen

Study supervision: P.A. van den Brandt

This study was funded by the Wereld Kanker Onderzoek Fonds Nederland (WCRF-NL), as part of the World Cancer Research Fund International grant program (Grant No. 2015/1390 to P.A. van den Brandt). We are indebted to the participants of the Netherlands Cohort Study (NLCS) and further wish to thank the Netherlands Cancer Registry and the Dutch Pathology Registry. Additionally, NLCS staff members are acknowledged for their valuable assistance and advice.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Bray
F
,
Ferlay
J
,
Soerjomataram
I
,
Siegel
RL
,
Torre
LA
,
Jemal
A
. 
Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2018
;
68
:
394
424
.
2.
Hsing
AW
,
Chokkalingam
AP
. 
Prostate cancer epidemiology
.
Front Biosci
2006
;
11
:
1388
413
.
3.
Pernar
CH
,
Ebot
EM
,
Wilson
KM
,
Mucci
LA
. 
The epidemiology of prostate cancer
.
Cold Spring Harb Perspect Med
2018
;
8
.
doi: 10.1101/cshperspect.a030361
.
4.
Cumberbatch
MGK
,
Jubber
I
,
Black
PC
,
Esperto
F
,
Figueroa
JD
,
Kamat
AM
, et al
Epidemiology of bladder cancer: a systematic review and contemporary update of risk factors in 2018
.
Eur Urol
2018
;
74
:
784
95
.
5.
World Cancer Research Fund/American Institute for Cancer Research
. 
Continuous update project expert report 2018. Diet, nutrition, physical activity and bladder cancer
. 
2018
.
Available from:
https://www.wcrf.org/dietandcancer.
6.
Crawford
ED
. 
Understanding the epidemiology, natural history, and key pathways involved in prostate cancer
.
Urology
2009
;
73
(
5 Suppl
):
S4
10
.
7.
Clark
PE
,
Agarwal
N
,
Biagioli
MC
,
Eisenberger
MA
,
Greenberg
RE
,
Herr
HW
, et al
Bladder cancer
.
J Natl Compr Canc Netw
2013
;
11
:
446
75
.
8.
Al-Zalabani
AH
,
Stewart
KF
,
Wesselius
A
,
Schols
AM
,
Zeegers
MP
. 
Modifiable risk factors for the prevention of bladder cancer: a systematic review of meta-analyses
.
Eur J Epidemiol
2016
;
31
:
811
51
.
9.
Willett
WC
,
Sacks
F
,
Trichopoulou
A
,
Drescher
G
,
Ferro-Luzzi
A
,
Helsing
E
, et al
Mediterranean diet pyramid: a cultural model for healthy eating
.
Am J Clin Nutr
1995
;
61
(
6 Suppl
):
1402S
6S
.
10.
Trichopoulou
A
,
Lagiou
P
. 
Healthy traditional Mediterranean diet: an expression of culture, history, and lifestyle
.
Nutr Rev
1997
;
55
(
11 Pt 1
):
383
9
.
11.
Fung
TT
,
Rexrode
KM
,
Mantzoros
CS
,
Manson
JE
,
Willett
WC
,
Hu
FB
. 
Mediterranean diet and incidence of and mortality from coronary heart disease and stroke in women
.
Circulation
2009
;
119
:
1093
100
.
12.
Giovannucci
E
,
Liu
Y
,
Platz
EA
,
Stampfer
MJ
,
Willett
WC
. 
Risk factors for prostate cancer incidence and progression in the health professionals follow-up study
.
Int J Cancer
2007
;
121
:
1571
8
.
13.
Bosire
C
,
Stampfer
MJ
,
Subar
AF
,
Park
Y
,
Kirkpatrick
SI
,
Chiuve
SE
, et al
Index-based dietary patterns and the risk of prostate cancer in the NIH-AARP diet and health study
.
Am J Epidemiol
2013
;
177
:
504
13
.
14.
Kenfield
SA
,
DuPre
N
,
Richman
EL
,
Stampfer
MJ
,
Chan
JM
,
Giovannucci
EL
. 
Mediterranean diet and prostate cancer risk and mortality in the Health Professionals Follow-up Study
.
Eur Urol
2014
;
65
:
887
94
.
15.
Buckland
G
,
Ros
MM
,
Roswall
N
,
Bueno-de-Mesquita
HB
,
Travier
N
,
Tjonneland
A
, et al
Adherence to the Mediterranean diet and risk of bladder cancer in the EPIC cohort study
.
Int J Cancer
2014
;
134
:
2504
11
.
16.
Dugue
PA
,
Hodge
AM
,
Brinkman
MT
,
Bassett
JK
,
Shivappa
N
,
Hebert
JR
, et al
Association between selected dietary scores and the risk of urothelial cell carcinoma: a prospective cohort study
.
Int J Cancer
2016
;
139
:
1251
60
.
17.
van den Brandt
PA
,
Goldbohm
RA
,
van 't Veer
P
,
Volovics
A
,
Hermus
RJ
,
Sturmans
F
. 
A large-scale prospective cohort study on diet and cancer in the Netherlands
.
J Clin Epidemiol
1990
;
43
:
285
95
.
18.
van den Brandt
PA
,
Schouten
LJ
,
Goldbohm
RA
,
Dorant
E
,
Hunen
PM
. 
Development of a record linkage protocol for use in the Dutch Cancer Registry for Epidemiological Research
.
Int J Epidemiol
1990
;
19
:
553
8
.
19.
Goldbohm
RA
,
van den Brandt
PA
,
Brants
HA
,
van 't Veer
P
,
Al
M
,
Sturmans
F
, et al
Validation of a dietary questionnaire used in a large-scale prospective cohort study on diet and cancer
.
Eur J Clin Nutr
1994
;
48
:
253
65
.
20.
Volovics
A
,
van den Brandt
PA
. 
Methods for the analyses of case-cohort studies
.
Biometrical J
1997
;
39
:
195
214
.
21.
Prentice
RL
. 
A case-cohort design for epidemiologic cohort studies and disease prevention trials
.
Biometrika
1986
;
73
:
1
11
.
22.
Goldbohm
RA
,
van 't Veer
P
,
van den Brandt
PA
,
van 't Hof
MA
,
Brants
HA
,
Sturmans
F
, et al
Reproducibility of a food frequency questionnaire and stability of dietary habits determined from five annually repeated measurements
.
Eur J Clin Nutr
1995
;
49
:
420
9
.
23.
NEVO
.
Dutch food composition table 1986–1987
.
The Hague, the Netherlands
:
Voorlichtingsbureau voor de Voeding
; 
1986
.
24.
Fung
TT
,
McCullough
ML
,
Newby
PK
,
Manson
JE
,
Meigs
JB
,
Rifai
N
, et al
Diet-quality scores and plasma concentrations of markers of inflammation and endothelial dysfunction
.
Am J Clin Nutr
2005
;
82
:
163
73
.
25.
Trichopoulou
A
,
Orfanos
P
,
Norat
T
,
Bueno-de-Mesquita
B
,
Ocke
MC
,
Peeters
PH
, et al
Modified Mediterranean diet and survival: EPIC-elderly prospective cohort study
.
BMJ
2005
;
330
:
991
.
26.
Mitrou
PN
,
Kipnis
V
,
Thiebaut
AC
,
Reedy
J
,
Subar
AF
,
Wirfalt
E
, et al
Mediterranean dietary pattern and prediction of all-cause mortality in a US population: results from the NIH-AARP Diet and Health Study
.
Arch Intern Med
2007
;
167
:
2461
8
.
27.
Trichopoulou
A
,
Kouris-Blazos
A
,
Wahlqvist
ML
,
Gnardellis
C
,
Lagiou
P
,
Polychronopoulos
E
, et al
Diet and overall survival in elderly people
.
BMJ
1995
;
311
:
1457
60
.
28.
Trichopoulou
A
,
Costacou
T
,
Bamia
C
,
Trichopoulos
D
. 
Adherence to a Mediterranean diet and survival in a Greek population
.
N Engl J Med
2003
;
348
:
2599
608
.
29.
World Cancer Research Fund/American Institute for Cancer Research
. 
Diet, nutrition, physical activity and cancer: a global perspective. Continuous update project expert report 2018
. 
2018
. https://www.wcrf.org/dietandcancer.
30.
Lin
DY
,
Wei
LJ
. 
The robust inference for the Cox proportional hazards model
.
J Am Stat Assoc
1989
;
84
:
1074
8
.
31.
Grambsch
PM
,
Therneau
TM
. 
Proportional hazards tests and diagnostics based on weighted residuals
.
Biometrika
1994
;
81
:
515
26
.
32.
de Vogel
S
,
Bongaerts
BW
,
Wouters
KA
,
Kester
AD
,
Schouten
LJ
,
de Goeij
AF
, et al
Associations of dietary methyl donor intake with MLH1 promoter hypermethylation and related molecular phenotypes in sporadic colorectal cancer
.
Carcinogenesis
2008
;
29
:
1765
73
.
33.
Wacholder
S
,
Gail
MH
,
Pee
D
,
Brookmeyer
R
. 
Alternative variance and efficiency calculations for the case-cohort design
.
Biometrika
1989
;
76
:
117
23
.
34.
Akaike
H
. 
A new look at the statistical model identification
.
IEEE Trans Automat Contr
1974
;
AC-19
:
716
23
.
35.
van den Brandt
PA
,
Schulpen
M
. 
Mediterranean diet adherence and risk of postmenopausal breast cancer: results of a cohort study and meta-analysis
.
Int J Cancer
2017
;
140
:
2220
31
.
36.
Schulpen
M
,
van den Brandt
PA
. 
Adherence to the Mediterranean diet and risk of lung cancer in the Netherlands Cohort Study
.
Br J Nutr
2018
;
119
:
674
84
.
37.
Schulpen
M
,
Peeters
PH
,
van den Brandt
PA
. 
Mediterranean diet adherence and risk of esophageal and gastric cancer subtypes in the Netherlands Cohort Study
.
Gastric Cancer
2019
;
22
:
663
74
.
38.
World Cancer Research Fund/American Institute for Cancer Research
.
Food, nutrition, physical activity, and the prevention of cancer: a global perspective
.
Washington DC
:
American Institute for Cancer Research
; 
2007
.
39.
Romaguera
D
,
Vergnaud
AC
,
Peeters
PH
,
van Gils
CH
,
Chan
DS
,
Ferrari
P
, et al
Is concordance with World Cancer Research Fund/American Institute for Cancer Research guidelines for cancer prevention related to subsequent risk of cancer? Results from the EPIC study
.
Am J Clin Nutr
2012
;
96
:
150
63
.
40.
Vergnaud
AC
,
Romaguera
D
,
Peeters
PH
,
van Gils
CH
,
Chan
DS
,
Romieu
I
, et al
Adherence to the World Cancer Research Fund/American Institute for Cancer Research guidelines and risk of death in Europe: results from the European Prospective Investigation into Nutrition and Cancer cohort study
.
Am J Clin Nutr
2013
;
97
:
1107
20
.
41.
Ax
E
,
Garmo
H
,
Grundmark
B
,
Bill-Axelson
A
,
Holmberg
L
,
Becker
W
, et al
Dietary patterns and prostate cancer risk: report from the population based ULSAM cohort study of Swedish men
.
Nutr Cancer
2014
;
66
:
77
87
.
42.
Lavalette
C
,
Adjibade
M
,
Srour
B
,
Sellem
L
,
Fiolet
T
,
Hercberg
S
, et al
Cancer-specific and general nutritional scores and cancer risk: results from the prospective NutriNet-Sante Cohort
.
Cancer Res
2018
;
78
:
4427
35
.
43.
Moller
E
,
Galeone
C
,
Andersson
TM
,
Bellocco
R
,
Adami
HO
,
Andren
O
, et al
Mediterranean Diet Score and prostate cancer risk in a Swedish population-based case-control study
.
J Nutr Sci
2013
;
2
:
e15
.
44.
Russo
GI
,
Solinas
T
,
Urzi
D
,
Privitera
S
,
Campisi
D
,
Cocci
A
, et al
Adherence to Mediterranean diet and prostate cancer risk in Sicily: population-based case-control study
.
Int J Impot Res
2019
;
31
:
269
75
.
45.
Jahn
JL
,
Giovannucci
EL
,
Stampfer
MJ
. 
The high prevalence of undiagnosed prostate cancer at autopsy: implications for epidemiology and treatment of prostate cancer in the prostate-specific antigen-era
.
Int J Cancer
2015
;
137
:
2795
802
.
46.
Witlox
WJA
,
van Osch
FHM
,
Brinkman
M
,
Jochems
S
,
Goossens
ME
,
Weiderpass
E
, et al
An inverse association between the Mediterranean diet and bladder cancer risk: a pooled analysis of 13 cohort studies
.
Eur J Nutr
2019 Feb 8
[Epub ahead of print]
.
47.
Bravi
F
,
Spei
ME
,
Polesel
J
,
Di Maso
M
,
Montella
M
,
Ferraroni
M
, et al
Mediterranean diet and bladder cancer risk in Italy
.
Nutrients
2018
;
10
. doi: 10.3390/nu10081061.
48.
Mizoue
T
,
Kasai
H
,
Kubo
T
,
Tokunaga
S
. 
Leanness, smoking, and enhanced oxidative DNA damage
.
Cancer Epidemiol Biomarkers Prev
2006
;
15
:
582
5
.
49.
Brill
JB
. 
The Mediterranean diet and your health
.
Am J Lifestyle Med
2009
;
3
:
44
56
.
50.
Verberne
L
,
Bach-Faig
A
,
Buckland
G
,
Serra-Majem
L
. 
Association between the Mediterranean diet and cancer risk: a review of observational studies
.
Nutr Cancer
2010
;
62
:
860
70
.