|Home | About | Journals | Submit | Contact Us | Français|
Cruciferous vegetables, tomato sauce, and legumes have been associated with reduced risk of incident advanced prostate cancer. In vitro and animal studies suggest these foods may inhibit progression of prostate cancer, but there are limited data in men. Therefore, we prospectively examined whether intake of total vegetables, and specifically cruciferous vegetables, tomato sauce, and legumes, after diagnosis reduce risk of prostate cancer progression among 1,560 men diagnosed with non-metastatic prostate cancer and participating in the Cancer of the Prostate Strategic Urologic Research Endeavor, a United States prostate cancer registry. As a secondary analysis, we also examined other vegetable sub-groups, total fruit, and subgroups of fruits. The participants were diagnosed primarily at community-based clinics and followed from 2004–2009. We assessed vegetable and fruit intake via a semi-quantitative food frequency questionnaire, and ascertained prostate cancer outcomes via urologist report and medical records. We observed 134 events of progression (53 biochemical recurrences, 71 secondary treatments likely due to recurrence, six bone metastases, four prostate cancer deaths) during 3,171 person-yrs. Men in the fourth quartile of post-diagnostic cruciferous vegetable intake had a statistically significant 59% decreased risk of prostate cancer progression compared to men in the lowest quartile (hazard ratio (HR): 0.41; 95% confidence interval (CI): 0.22, 0.76; p-trend: 0.003). No other vegetable or fruit group was statistically significantly associated with risk of prostate cancer progression. In conclusion, cruciferous vegetable intake after diagnosis may reduce risk of prostate cancer progression.
More than 2.2 million men currently live with prostate cancer in the United States (US).1 Cruciferous vegetables, tomato sauce, and legumes have been linked to a lower risk of incident prostate cancer and, in some cases, reduced risk of advanced or aggressive disease at diagnosis.2–6 However, there are limited data on the chemotherapeutic effects of diet after diagnosis of prostate cancer. Among 1,202 men with non-metastatic prostate cancer in the Health Professionals’ Follow-up Study (HPFS), post-diagnostic intake of tomato sauce and fish were associated with reduced risk of prostate cancer progression.7 In the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE™), intakes of poultry with skin and eggs after diagnosis were associated with elevated risk of prostate cancer progression.8 Saturated fat intake has also been reported to increase risk of prostate cancer progression,9, 10 and small clinical trials conducted in men with prostate cancer and examining biomarker outcomes support a role of diet in prostate cancer progression.11–14
Thus, we prospectively examined post-diagnostic intake of vegetables and fruits in relation to risk of prostate cancer progression among men with non-metastatic prostate cancer at diagnosis. We hypothesized that total vegetables, and specifically cruciferous vegetables, tomato sauce, and legumes, would be inversely associated with risk of prostate cancer progression.
CaPSURE is a US prostate cancer registry study initiated in 1995.15, 16 Forty sites (34 community-based clinics, three academic institutions, three Veterans Administration hospitals) have enrolled men with biopsy-verified prostate cancer. CaPSURE participants complete surveys at baseline and every 6 mos. thereafter, and urologists provide clinical data at baseline and subsequent clinic visits. The base population for this study included 2,134 participants in CaPSURE who completed a semi-quantitative food frequency questionnaire (FFQ) during 2004–2005. The Institutional Review Boards of the University of California, San Francisco and collaborating institutions approved this study.
Vegetable and fruit groups of interest for this analysis are listed in Table 1. Men were asked how often on average they consumed a specific portion of 127 foods and beverages during the past year, with nine frequency options ranging from <1/mo. to ≥6/d. We also asked participants whether they were eating more, less, or the same amount of each item compared to before their diagnosis. Our FFQ was based on a FFQ that has been used to study diet and chronic disease relations in a variety of populations.17, 18 In a validation study, the median correlation between two 1-wk diet records and the original FFQ for vegetable and fruit items was 0.58 (range:0.19 for garlic to 0.95 for bananas).19
We abstracted data on treatment, biopsy Gleason sum, stage, prostate specific antigen (PSA), and metastases from medical records or urologists’ reports. The National Death Index and Bureau of Vital Statistics were checked for mortality data, and death certificates were used to verify the date and cause of death.
Prostate cancer progression was defined as: prostate cancer death, bone metastases from prostate cancer, biochemical recurrence, or initiation of secondary treatment. A death was attributed to prostate cancer if prostate cancer was listed as the primary, secondary, or tertiary cause of death and no other malignancy was listed as a higher order cause. An outcome of bone metastases was defined as urologist report of: (1) prostate cancer progression to bone, (2) positive bone scan, (3) radiation for metastasis at a bone site, or (4) M1b stage. Biochemical recurrence was defined as two consecutive PSA values ≥0.2ng/ml ≥8 wks after radical prostatectomy or a PSA ≥2ng/ml above post-radiation nadir.20 Secondary treatment was defined as any treatment initiated ≥6 mos. after primary treatment.21, 22 The date of prostate cancer progression was the first of the following: prostate cancer death, diagnosis of bone metastases, second PSA ≥0.2ng/ml for radical prostatectomy patients, first PSA ≥2ng/ml above nadir for radiation patients, or initiation of secondary treatment.
We excluded men with extra-prostatic disease at diagnosis (T-stage≥T3b), men missing treatment information, and men who reported an energy intake outside 800–4200 kcal/d (n=241). To maintain the prospective nature of our analysis and reduce the potential for recall bias, we excluded men whose prostate cancer progressed prior to the FFQ (n=333), resulting in 1,560 men for analysis.
We examined associations between post-diagnostic vegetable and fruit intake and prostate cancer progression using Cox proportional hazards regression. Person-time was calculated from date of the FFQ until prostate cancer progression, non-prostate cancer death, last contact, or August 21, 2009, whichever occurred first. We modeled quartiles of vegetables and fruits with indicator variables and tested for linear trends using the median of each quartile as a continuous term.
Model 1 was adjusted for age at diagnosis (continuous), energy intake (continuous), and days from diagnosis to FFQ (continuous). Model 2 included the variables in Model 1 plus prognostic risk at diagnosis (low, intermediate, high), primary treatment (radical prostatectomy, radiation, other/active surveillance, androgen deprivation therapy), body mass index (BMI; <25, 25–29.9, ≥30 kg/m2), and walking metabolic equivalent task (MET)-h/wk (quartile rank).23 We classified participants’ prognostic risk using modified D’Amico definitions as follows: [High: PSA>20ng/ml or Gleason sum=8–10 or T-Stage≥T3a; else Intermediate: PSA=10.1–20ng/ml or Gleason sum=7 or secondary 4–5 pattern or T-Stage=T2b/T2c (2002) or T2b (1997); else Low: PSA≤10ng/ml and Gleason sum=2–6 and T-Stage=≤T2a].24, 25 Model 3 was additionally adjusted for quartile ranks of eggs, poultry with skin, fruits, and vegetables other than the exposure of interest. Adjustment for education, income, race, prostate cancer family history, smoking, and intakes of sweets, grains, or dairy did not change the results; therefore these variables were omitted from the final models.
We examined whether biopsy Gleason sum (<7 v. ≥7), age at diagnosis (<60 v. ≥60 y), smoking (ever v. never), BMI (<25 v. ≥ 25 kg/m2), or walking (< 7.5 v. ≥7.5 MET-h/wk) modified any of the relations using likelihood ratio tests. The cut-points for age at diagnosis and walking were chosen based on their distribution in the study population.
We performed a sensitivity analysis excluding events defined by secondary treatment that lacked evidence of a preceding PSA rise. In addition, we were concerned men with higher prognostic risk may increase their tomato intake more than men with lower prognostic risk; therefore we examined whether self-reported change in tomato items was associated with prognostic risk at diagnosis using chi-square tests.
All statistical tests were two-sided and considered significant at p<0.05. All analyses were conducted using SAS v. 9.1.3.
We observed 134 events of progression (53 biochemical recurrences, 71 secondary treatments, six bone metastases, four prostate cancer deaths) among 1,560 men during 3,171 person-yrs. The median year of diagnosis was 2002 [interquartile range (IQR): 2000–2003]. The median follow-up after the FFQ was 23 mos. (IQR: 10–32 mos.). Approximately 14% (n=213) of the men who completed the FFQ did not participate in CaPSURE follow-up after the FFQ; these men did not differ from the remaining men in terms of their biopsy Gleason sum, clinical T-stage, primary treatment, or intake of total vegetables, cruciferous vegetables, tomato sauce, or legumes. However, they were younger (mean=63y) compared to the remaining men (mean=65y) (p-value=0.001).
Men who consumed more vegetables were more educated, had higher household incomes, and expended more energy walking than men who consumed the least vegetables (Table 2). We observed a non-significant inverse trend for total vegetables and risk of progression (Table 3), which appeared to be driven by cruciferous vegetables (e.g. broccoli; cabbage, coleslaw; cauliflower; Brussels sprouts; kale, mustard, chard greens). Men in the fourth quartile of post-diagnostic intake of cruciferous vegetables had a 59% reduced risk of prostate cancer progression compared to men in the lowest quartile (hazard ratio (HR): 0.41, 95% confidence interval (CI): 0.22, 0.76; p-trend: 0.003). The remaining vegetable groups were not associated with risk of prostate cancer progression. Most of the individual cruciferous items were inversely related to risk of prostate cancer progression, but none were statistically significant on its own, likely due to the low consumption of these foods (Table 4).
Total fruit and fruit groups, with the exception of berries, were also not associated with risk of prostate cancer progression (Table 5). For berries, there was an inverse association in the age- and calorie-adjusted model (HR comparing extreme quartiles: 0.60; 95%CI: 0.37, 0.97), which was somewhat attenuated and not statistically significant after multivariate adjustment (HR comparing extreme quartiles: 0.68; 95%CI: 0.40, 1.15).
There was no evidence of effect modification by biopsy Gleason sum, BMI, age at diagnosis, or smoking. However, there was an interaction between walking and total vegetable intake (p-interaction = 0.02). Among the 732 men who walked ≥7.5 MET-h/wk after diagnosis (approximately 150 min/wk), total vegetable intake after diagnosis was inversely associated with risk of prostate cancer progression (HR comparing extreme quartiles: 0.35; 95%CI: 0.15, 0.79). There was no association among the 729 men who walked <7.5 MET-h/wk (HR comparing extreme quartiles: 0.91; 95%CI: 0.36, 2.31).
Our results remained unchanged when excluding events defined by secondary treatment without evidence of a preceding PSA rise. Additionally, few men reported any change in tomato intake compared to before diagnosis (≤15%) and there was no association between self-reported change in any tomato item and prognostic risk at diagnosis (data not shown).
In this novel analysis of post-diagnostic vegetable and fruit intake and clinical outcomes among men with prostate cancer, we observed a strong inverse association between cruciferous vegetable intake after diagnosis and prostate cancer progression. No other vegetable or fruit group after diagnosis was statistically significantly associated with risk of prostate cancer progression.
This is the first study to examine cruciferous vegetable intake after diagnosis in relation to clinical outcomes among men with prostate cancer. However, two recent prospective studies reported inverse associations between cruciferous vegetables or glucosinolate, a metabolite of cruciferous vegetables, and risk of incident prostate cancer. Kirsh et al. reported a 40% reduced risk of incident extra-prostatic prostate cancer comparing men with high and low cruciferous vegetable intake (HR: 0.60, 95%CI: 0.36, 0.98; p-trend: 0.02).26 In the EPIC-Heidelberg cohort, high glucosinolate consumption was associated with a 32% decreased risk of incident prostate cancer (HR Q4 versus Q1: 0.68; 95%CI: 0.48, 0.97; p-trend: 0.03).27
Glucosinolates are hydrolyzed to form isothiocyanates and indoles, which have anti-carcinogenic effects in vitro and in vivo.28–31 The isothiocyanate, sulforaphane, promotes apoptosis and cell cycle arrest in prostate cancer cells.32–34 Phenethyl isothiocyanate inhibits prostate cancer cell growth and migration, reduces androgen receptor levels, impairs mRNA translation, and promotes transcription of p21.35–38 Additionally, indole-3-carbinol promotes cell cycle arrest, growth inhibition, and apoptosis, and has been shown to inhibit components of oncogenic cell signaling pathways.39–41 In humans, consumption of broccoli sprouts inhibited histone deacetylase in blood42 and a broccoli-rich diet altered global gene expression in the prostate.43
Furthermore, Joseph et al. observed a stronger inverse relation between broccoli consumption and prostate cancer risk among men with the glutathione S-transferase mu 1 (GSTM1)-present genotype compared to men with null deletions in this gene.44 Glutathione S-transferase enzymes are induced by metabolites of cruciferous vegetables and may reduce risk of prostate cancer progression through detoxification of carcinogens and elimination of reactive oxidative species.45
We observed evidence of an interaction between total vegetable intake after diagnosis and walking, similar to results from the control arm of the Women’s Healthy Eating and Living trial among women with breast cancer.46 In that study, women who consumed ≥5 servings/d of vegetables and fruits and engaged in physical activity equivalent to walking 30 min/d 6 d/wk had a 44% reduced risk of mortality compared to women who consumed <5 servings/d of vegetables and fruits and engaged in <30 min/d 6 d/wk of activity (HR: 0.56; 95% CI: 0.31, 0.98). Future studies should consider the possible synergy between plant-based diets and physical activity in reducing risk of prostate cancer progression.
The null association for tomato sauce was contrary to our hypothesis and the previously observed inverse association.7 We considered whether the lack of an inverse association for tomato sauce was due to reverse causation (e.g. men with higher prognostic risk at diagnosis increased their tomato sauce intake more than men with low prognostic risk), but there was no association between change in any tomato item and prognostic risk at diagnosis. Overall, the results of intervention studies on tomato sauce and/or lycopene supplementation after prostate cancer diagnosis in relation to intermediate endpoints have been inconsistent,11, 47 and further research on the role of tomatoes after prostate cancer diagnosis is needed.
No other study has examined post-diagnostic intake of legumes in relation to clinical outcomes in men with prostate cancer. However, our observation of no association between post-diagnostic legume intake and prostate cancer progression is consistent with many prospective studies on incident prostate cancer in Western populations, although two reported inverse associations.6, 48–50 In the Multi-Ethnic Cohort Study, greater legume consumption was associated with small to moderate reductions in risk of total and aggressive incident prostate cancer (HR: 0.89, 95% CI: 0.89, 0.99 for total; HR: 0.74, 95% CI: 0.61, 0.91 for aggressive).48 However, this association was only significant among Latinos, who had much higher legume consumption than any other ethnic group. A recent meta-analysis reported similar variation across ethnic groups, with evidence of a protective association in Asian populations, but no association in Western populations.50 Associations between dietary factors and risk of incident prostate cancer versus post-diagnostic intake and prostate cancer progression likely differ, and more research is needed before firm conclusions may be drawn. Yet, based on the available data, if there is an effect of legumes on prostate cancer progression, it appears to be modest and likely varies across populations depending on their level of legume intake.
Limitations of this study include our lack of pre-diagnostic diet, loss to follow-up, and few events. Our lack of pre-diagnostic diet prevents us from concluding the association we observed between post-diagnostic cruciferous vegetable intake was independent of what the men consumed prior to diagnosis. In addition, 14% of participants did not participate in CaPSURE follow-up after the FFQ, and thus were considered lost to follow-up immediately after completing the FFQ. Fortunately, these men did not differ from the remaining men in terms of their vegetable intake or clinical prognostic factors, and therefore it is unlikely that loss of these men biased our results. Lastly, we acknowledge that this is a small study and caution is warranted in interpreting the strong inverse relation we observed between post-diagnostic cruciferous vegetables and risk of prostate cancer progression. While suggestive, further study of cruciferous vegetables in men with prostate cancer is needed from randomized controlled trials before translating these results to clinical practice.
In conclusion, cruciferous vegetable consumption after diagnosis was strongly associated with reduced risk of prostate cancer progression among men initially diagnosed with non-metastatic prostate cancer. These data strengthen the rationale to investigate the phytochemicals of cruciferous vegetables in men with prostate cancer, and if confirmed, provide dietary guidance for men with prostate cancer.
We thank the participants and staff of CaPSURE for their invaluable contributions, and Asher Schranz for his review of the literature on isothiocyanates and indoles.
CaPSURE is supported by an unrestricted educational gift from Abbott Laboratories (Chicago, IL), by the National Institutes of Health/National Cancer Institute (5RC1CA146596), and by the Agency for Healthcare Research and Quality (1U01CA88160). Richman was supported by the National Institutes of Health training grant (R25 CA098566). June Chan was supported by the Department of Defense (W81XWH-04-1-0850) and Prostate Cancer Foundation.
None of the authors have conflicts of interest.