Plant-Based Diets and Disease Progression in Men With Prostate Cancer

Key Points Question What is the association between postdiagnostic plant-based dietary patterns and risk of prostate cancer progression? Findings In a cohort study of 2062 men diagnosed with nonmetastatic prostate cancer, individuals with the highest intake of plant foods in the overall plant-based diet index had lower risk of prostate cancer progression compared with those with the lowest intake. Meaning These findings suggest that consuming a primarily plant-based diet may be associated with better prostate cancer–specific health outcomes among men with prostate cancer.


Introduction
Prostate cancer is the second most common cancer among men in the US.3][4][5][6] Current dietary recommendations for patients with cancer and the general population emphasize a plant-based diet high in fruits, vegetables, and whole grains. 7t, little is known about plant-based dietary patterns and prostate cancer-specific clinical outcomes after diagnosis.Many studies have reported that greater intake of individual plant-based foods (eg, cruciferous vegetables, cooked tomatoes, vegetable fats) is associated with lower risk of prostate cancer recurrence or mortality, [8][9][10][11][12][13][14] but single dietary factors in isolation may not accurately capture the health effects of whole diets. 15Given the increasing interest in plant-based food at the population level, examining whether plant-based dietary patterns are associated with disease outcomes has important implications for public health.
Therefore, we evaluated postdiagnosis intake of plant-based foods in relation to clinical outcomes among patients with prostate cancer.We focused on 2 plant-based diet indices: the overall plant-based diet index (PDI) and healthful plant-based index (hPDI).These indices were developed in 2016 in 3 large cohort studies and have been associated with risk of diabetes, coronary heart disease, and total mortality. 2,3In addition, in the Health Professionals Follow-Up Study, they were associated with a lower risk of fatal prostate cancer and better scores for quality of life among men diagnosed with prostate cancer. 16,17We hypothesized that greater consumption of plant foods in both indices would be associated with lower risk of prostate cancer progression and prostate cancerspecific mortality.

Study Design
This cohort study was conducted in accordance with the Belmont Report and the US Common Rule under local institutional review board approval.All participants provided written informed consent.
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.We used data from the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE), a longitudinal observational study of 15 310 men with biopsy-proven prostate cancer.Participants were enrolled concurrently from 43 urology practices across the US from 1999 to 2018. 18Participating urologists collected data on clinical and pathological factors, treatments, and recurrence.

Study Population
A subset of individuals from the CaPSURE study were invited to participate in the CaPSURE Diet and Lifestyle substudy, consisting of a comprehensive diet and lifestyle questionnaire with a validated food frequency questionnaire (FFQ).Invitations to participate in the substudy were sent to all active participants at 3 time points between 2004 and 2016.If more than 1 survey was completed, we used the first completed FFQ to standardize exposure assessment.
Participants with a last clinical follow-up or documented progression (defined as recurrence, secondary treatment, bone metastases, or prostate cancer-specific mortality) prior to completion of a diet and lifestyle questionnaire were excluded.To reduce measurement error in usual diet, 19,20 individuals with an extreme or unknown caloric intake (<800 kcal/d or >4200 kcal/d) or missing 70 or more FFQ items were excluded.Lastly, individuals with unknown clinical T-stage or T-stage T3a or higher were excluded.For the prostate cancer-specific mortality analyses (secondary outcome), we included individuals who had documented progression prior to completing the FFQ, since these people were still at risk for prostate cancer-specific mortality.

JAMA Network Open | Urology
Plant-Based Diets and Disease Progression in Men With Prostate Cancer

Dietary Assessment
Dietary data were collected with a validated semiquantitative FFQ based on the one used to develop the diet indices. 21Participants were asked on average, how often (ranging from never or <1 serving/mo to Ն6 servings/d) they consumed a standard portion size of approximately 140 distinct foods and beverages in the past year.
To compute the plant-based diet indices, 18 food groups were created based on nutrients and culinary similarities, then classified into 3 larger categories of 7 healthful plant foods (whole grains, fruits, vegetables, nuts, legumes, vegetable oils, and tea and coffee), 5 unhealthful plant foods (fruit juices, sugar-sweetened beverages, refined grains, potatoes, and sweets or desserts), and 6 animal foods (animal fats, dairy, eggs, fish and seafood, meat, and miscellaneous animal-based foods). 2 Intakes of the 18 food groups (servings per day) were ranked into quintiles (Qs).For PDI, greater amounts of both the healthful and unhealthful plant groups were given higher scores (ie, Q1 indicates a score 1; Q2, 2; Q3, 3; Q4, 4; Q5, 5), whereas animal food groups were given lower scores (ie, Q5 indicates a score of 1; Q4, 2; Q3, 3; Q2, 4; Q1, 5).For hPDI, the healthful plant food group was given increasing scores, while unhealthful plant food and animal food groups were given decreasing scores.
Scores for the 18 groups were summed, ranging from 18 (lowest plant intake) to 90 (highest plant intake).

Outcome Ascertainment
Our primary outcome was time to prostate cancer progression, a composite outcome comprised of biochemical recurrence, secondary treatment, bone metastases, or death attributed to prostate cancer.If participants had multiple progression events, the first reported date was used.Biochemical recurrence was defined as either 2 consecutive prostate-specific antigen (PSA) levels at least 0.2 ng/mL (to convert to micrograms per liter, multiply by 1) after radical prostatectomy or 2 consecutive PSA levels at least 2.0 ng/mL greater than the postradiation nadir. 22Date of recurrence was recorded as the date of the second elevated PSA.Secondary treatment was defined as any treatment that started at least 6 months after primary treatment completion.Bone metastases were attributed to prostate cancer if a urologist reported prostate cancer progression to bone or advancement to stage M1b, the patient had a positive bone scan, or the patient underwent radiation to treat bone metastases.Cause of death was determined by the registry data coordinating center and confirmed by state death certificate or the National Center for Health Statistics National Death Index.For analyses focused on clinical progression, participants were administratively censored at their last known clinical follow-up date up until January 31, 2019.
Prostate cancer-specific mortality was our secondary outcome, given the small number of prostate cancer-specific mortality events in this cohort.For these analyses, participants with a last known clinical follow-up date beyond December 30, 2020 (last National Death Index search), were administratively censored on December 30, 2020.

Statistical Analysis
Medians and IQRs were calculated for continuous patient and clinical characteristics, and number and percentage was calculated for categorical characteristics, overall and by quintile, of the index scores.Median and IQR consumption of the 18 individual food groups were also computed in servings per day.
Pearson correlation coefficient was used to describe the correlation between the PDI and hPDI.
We used Cox proportional hazards models and cause-specific models to evaluate the associations between the PDIs and the risk of prostate cancer progression and prostate cancer-specific mortality, respectively.All models were clustered by CaPSURE clinical site, with robust standard errors used to calculate 95% CIs.Simple models were adjusted for days from diagnosis to FFQ, age at diagnosis (years), year of diagnosis, and total energy intake (kcal).In the full multivariable models, we additionally adjusted for clinical T-stage (T1, T2, T3a), Gleason score (<7, 7, >7), and PSA (Յ6, >6 to 10, >10 ng/mL) at diagnosis, primary treatment (radical prostatectomy, radiation, hormonal therapy,

JAMA Network Open | Urology
Plant-Based Diets and Disease Progression in Men With Prostate Cancer watchful waiting or active surveillance, other); self-reported race and ethnicity; smoking status (current, former, never); walking pace (<2, 2 to <3, 3 to <4, >4 mph, unable), and body mass index.
Race and ethnicity were self reported and categorized as African American, Asian or Pacific Islander, Latino, multiple, Native American, White, and unknown; race and ethnicity were used in adjustment as White or other.Race and ethnicity were included in analysis because they are strongly associated with prostate cancer outcomes.Additional covariates, including diabetes, family history of prostate cancer, household income, education level, height, alcohol use, multivitamin use, calcium supplement use, and selenium supplement use, were considered but did not meaningfully change results, so they were not included in final models.Log-minus-log plots and Schoenfeld tests were used to test the proportional hazards assumption, and Martingale residuals and smoothing were used to assess the linearity of predictors assumption.Contrast analyses were used to assess for linear trends.
In secondary analyses, we examined each of the 3 food groups comprising the indices (healthful, unhealthful, animal) in association with prostate cancer progression.We also explored potential modification by walking pace (<3 vs Ն3 mph), age (<65 years vs Ն65 years), stage (T1, T2, or T3a), PSA (<6, 6-10, or >10 ng/mL), and Gleason grade at diagnosis (<7, Ն7).To evaluate the significance of interactions between the PDI or hPDI and these variables, we used separate multivariable models including cross product terms between the index and effect modifier of interest.We then used likelihood ratio tests to compare models with and without these interaction terms.For the covariates that demonstrated statistically significant interactions, stratified subgroup analyses were performed.
All analyses were performed in Stata software version 17 (StataCorp) using a 2-sided α = .05to assess statistical significance.Data were analyzed from August 2022 to April 2023.

Results
A total of 2891 participants completed at least 1 survey, and a total of 2062 participants (median [IQR] age at diagnosis, 65.0 [59.0-70.0]years) met the inclusion criteria; 61 (3%) identified as African American, 3 (<1%) identified as American Indian or Alaska Native, 9 (<1%) identified as Asian or Pacific Islander, 15 (1%) identified as Latino, and 1959 (95%) identified as White.Participant characteristics, overall and by quintile of the PDI and hPDI at baseline, are displayed in Table 1.
Compared with participants in the lowest PDI and hPDI quintile, participants in the highest quintile of PDI and hPDI had a faster walking pace, lower body mass index, and lower diagnostic PSA and were less likely to smoke (Table 1).Participants in the highest PDI quintile consumed more calories than those in the lowest PDI quintile, whereas individuals in hPDI Q5 consumed fewer calories and were younger than those in Q1.Characteristics for participants in the prostate cancer-specific mortality analyses were nearly identical (eTable 1 in Supplement 1).PDI and hPDI scores were moderately positively correlated (r = 0.34; P < .001).PDI scores ranged from 27 to 76, and hPDI scores ranged from 29 to 84.
Servings per day of individual dietary score components by lowest and highest quintile of PDI and hPDI are shown in Table 2. Participants in the highest vs lowest quintile, consumed a mean of approximately 1.9 additional servings of vegetables, 1.6 additional servings of fruit, 0.9 more servings of whole grains, 1.0 less serving of dairy, 0.4 less servings of animal fat, slightly less egg, and marginally less meat (Table 2).
The sample for our secondary analysis of prostate cancer-specific mortality included 2274 participants, with 61 prostate cancer-specific deaths and 302 other deaths.While there were no statistically significant associations between either dietary index and risk of prostate cancer-specific mortality (Table 3), CIs were too wide to draw meaningful conclusions from point estimates (Q4 vs Q1: HR, 0.33; 95% CI, 0.14-0.78;Q5 vs Q1: HR, 0.53; 95% CI, 0.17-1.66;P for trend = .16).
For both indices, we found no evidence of interactions between age, PSA, stage at diagnosis, or walking pace.For hPDI, Gleason grade at diagnosis was associated with modifying the association of hPDI with prostate cancer progression (P for interaction = .03).Among participants with Gleason grade 7 or higher, participants in the highest quintile had a 55% lower risk of progression compared with the lowest quintile (HR, 0.45; 95% CI, 0.25-0.81;P for trend = .01)(eTable 3 in Supplement 1).
There was no statistically significant association in individuals with Gleason grade less than 7.We did not detect association modification by any factors for the associations of the indices with prostate cancer-specific mortality.

Discussion
This longitudinal cohort study investigated associations of plant-based dietary patterns after a diagnosis of localized prostate cancer with risk of prostate cancer progression.We did not evaluate the unhealthful plant-based diet index, as it would not be recommended for improving health outcomes.We observed an association whereby individuals who scored the highest on the overall PDI had lower risk of prostate cancer progression compared with those who scored the lowest.
Our findings align with previous reports that plant-based diets may improve prostate cancer outcomes.For example, in a study that evaluated PDI in association with risk of incident prostate cancer (47 243 men followed up for 28 years), Loeb et al 16 reported that a higher PDI score was associated with 19% lower risk of incident prostate cancer that went on to be fatal (HR, 0.81; 95% CI, 0.64-1.01;P for trend = .04).
Our results contribute to the evolving body of research indicating the positive associations of plant-based diets with health outcomes.The PDI and its subindices were originally developed by Satija et al 2,3 to evaluate the associations of PDI with type 2 diabetes and coronary heart disease.
Satija et al 2,3 found inverse associations between the overall PDI and hPDI for both outcomes.For PDI, other studies have observed a lower risk for diabetes, 5 cardiovascular disease risk, 23 cardiovascular mortality, 6,23 and total mortality. 6,23For hPDI, studies have reported lower risk for diabetes, 5 cardiovascular disease risk, 4 cardiovascular mortality, 6 and total mortality. 4,6These results are important in the context of localized prostate cancer, where men are more likely to die from these chronic diseases than their cancer.
We did not observe statistically significant associations for hPDI.Inconsistencies between the hPDI and overall PDI have been noted by others as well.In a 2022 study, Loeb et al 16 observed associations between PDI and risk of developing fatal prostate cancer, whereas associations for hPDI were only seen for risk of developing localized prostate cancer.A study by Kim et al 23 also reported statistically significant associations of PDI, but not hPDI, with risk of CVD. 23It may be that because the distribution of hPDI was relatively compressed compared with the distribution of PDI, the variance of the estimator increased corresponding with the association of hPDI.There were modest differences in servings per day between highest and lowest categories of fruit juice, refined grains, and sweets and desserts-all of which are categorized with equal weighting into the unhealthful food group and run in opposite directions for the PDI vs hPDI.Moreover, when looking at the healthful, unhealthful, and animal components of the subindices separately, there was no association with the unhealthful component with prostate cancer progression.Perhaps classifying some of these unhealthful plant foods as "bad actors" in the absence of an established detrimental association specifically with prostate cancer outcomes have attenuated the findings for hPDI.For PDI, we observed that people in Q5 (compared with Q1) consumed a mean of an additional 0.9 to 1.9 servings per day of healthful plant foods (particularly vegetables, fruits, and whole grains), while they consumed 0.3 to 1.0 fewer servings per day of animal products (particularly dairy, animal fat, and egg).5][26] While the similar consumption of fish and seafood, meat, and miscellaneous animal products among the extreme quintiles were unexpected, these results suggest that slightly reducing intake of animal products and placing more emphasis on more nutrient-dense plant-based foods may be advantageous.
Previous studies suggest several mechanisms through which plant-based diets may improve prostate cancer outcomes.Fruits and vegetables contain a variety of phytochemicals, including antioxidants and anti-inflammatory compounds, that have been shown to protect against prostate cancer. 24,26,27Plant foods are also a source of dietary fiber, which may promote satiety and regulate blood glucose levels. 280][31] High intake of red and processed meats has been associated with increased insulin resistance and insulin-like growth factor-  [32][33][34][35][36][37] Furthermore, milk and dairy (a primary source of insulin-like growth factor-1), have been associated with increased risk of prostate cancer [37][38][39][40][41] ; whole milk, in particular, has been associated with increased risk of prostate cancer recurrence. 38

Limitations
There are several limitations to the study.First, measurement error is a known limitation of selfreported data, including nutritional information.However, the FFQ used in this cohort has been validated, and the dietary data were collected prior to events of progression.Therefore, we expect measurement error in dietary intakes to be comparable in participants who experienced an event and those who did not.Second, even participants in the highest quintile of PDI consumed meat and dairy products; therefore we are unable to assess the associations of fully plant-based diets (eg, vegan, vegetarian).Third, the CaPSURE registry also is comprised of primarily of college-educated White men, which limits generalizability.Fourth, given that this was an observational study, we could not control for any unknown or unmeasured confounders.Other healthy behaviors and social determinants of health may be common causes of consuming more plant-based food and risk of prostate cancer progression.However, adjustment for income and education did not alter associations.Additionally, we were unable to adjust for prediagnostic diet in this cohort, so cannot conclude that the results are independent of prediagnostic exposure.Conversely, this study has several notable strengths, including a well-characterized cohort with extensive clinical follow-up and detailed diet data, as well as being the first to examine PDI and oncologic outcomes after prostate cancer diagnosis, to our knowledge.

Conclusions
The findings of this cohort study suggest that plant-based dietary patterns may be inversely associated with risk of prostate cancer progression, although future research and replication of our findings is needed.These data are consistent with prior research demonstrating the importance of dietary factors in overall health and well-being.

JAMA Network Open | Urology Plant
years after FFQ completion.In the fully adjusted models, participants in the highest quintile of PDI had a 47% lower risk of progression compared -Based Diets and Disease Progression in Men With Prostate Cancer

Table 1 .
Patient and Clinical Characteristics of Men With Localized Prostate Cancer, Overall and Stratified by Quintiles of PDI and hPDI Scores

Table 2 .
Median Servings per Day of Individual Food Components by Quintile of Each Index