Adherence to the Diabetes Risk Reduction Diet and Bladder Cancer Risk in the Prostate, Lung, Colorectal, Ovarian (PLCO) Cohort

Bladder cancer predominantly originates in the urothelial epithelium. With continuous advancements in surgical techniques and corresponding pharmacologic developments, the therapeutic options for bladder cancer have grown increasingly diverse. However, it continues to present a significant global health challenge, with over 550,000 new cases and more than 200,000 fatalities reported worldwide each year (1). Bladder cancer ranks tenth in incidence among all cancers, with a substantially higher incidence in men than in women, making it the sixth most common malignancy among males. Smoking remains the primary risk factor for its development (2, 3).

Recent studies have investigated the impact of dietary factors on the risk of developing bladder cancer. Various studies have identified associations between bladder cancer risk and factors such as alcohol and coffee consumption, vegetable and fruit intake, and consumption of processed meat and animal protein (4–8). Hence, making rational dietary choices, particularly adherence to long-term healthy dietary patterns, can contribute to the prevention of malignancies. The current literature has demonstrated an inverse association between adherence to a Mediterranean diet and the incidence of bladder cancer (9). Furthermore, exposure to a higher dietary inflammatory index (DII) is associated with an increased risk of bladder cancer (10, 11).

One such exemplar of a healthy dietary pattern is the diabetes risk reduction diet (DRRD), which consists of nine essential nutritional factors: nuts, whole fruits, cereal fiber, coffee, ratio of polyunsaturated to saturated fat, glycemic index (GI), sugar-sweetened beverages/fruit juices, trans fat, and red and processed meats (12). The DRRD has been linked to decreased risk of liver cancer (13), breast malignancy (14), and pulmonary carcinoma (15). However, the association between the DRRD and bladder cancer risk remains unexplored, signifying a potential gap in the current research. Remarkably, large cohort studies have demonstrated a significant increase in the risk of bladder cancer risk among individuals with type 2 diabetes (16, 17). Given that the DRRD encompasses diverse dietary elements that are inversely correlated with cancer onset, and it independently mitigates the risk of diabetes, we postulate a potential negative association between DRRD and the incidence of bladder cancer. To address this research gap, the primary objective of this study was to evaluate the association between DRRD adherence and the incidence of bladder cancer using data derived from Prostate, Lung, Colorectal, and Ovarian (PLCO) cohort. Through this comprehensive examination, we aim to provide valuable insights into the current understanding of dietary factors and bladder cancer risk and inform potential preventative strategies.

This study used data acquired from the PLCO Cancer Screening Trial, a comprehensive randomized clinical trial funded by the United States NCI. The primary objective of this trial was to examine the efficacy of various screening methods in reducing mortality rates associated with PLCO cancers among individuals aged 55 to 74 years. From 1993 to 2001, approximately 154,000 eligible individuals were enrolled in the research project and randomly assigned to either an intervention group (subjected to a specific screening test) or a control group (given routine care) across 10 centers in the United States. During the course of the study, participants completed self-administered lifestyle surveys and were monitored for cancer incidence until 2009. These surveys comprised of a baseline questionnaire (BQ) and a diet history questionnaire (DHQ). The BQ was employed to collect baseline information and record cancer diagnoses at the time of enrollment. The DHQ, derived from a food frequency questionnaire containing 124 food items along with inquiries about portion size and frequency, takes approximately one hour to complete and was first implemented in the trial in December 1998, 5 years after the baseline was established in November 1993. Unlike the dietary questionnaire, which was exclusively administered to the intervention arm during baseline assessment, the DHQ was administered to both the intervention and control arms of the trial. Approximately 77% of all the participants (equivalent to approximately 113, 000 individuals) completed the DHQ. The effectiveness of the DHQ as a tool for estimating nutrient intake has been evaluated in various studies, demonstrating its reliability (18, 19).

In line with the study's objective, participants were excluded on the basis of the following criteria: (i) failure to complete the DHQ, (ii) history of any cancer prior to the DHQ, (iii) instances where bladder cancer was not the first diagnosed cancer during the trial, and (iv) failure to complete the BQ. The United States NCI granted approval for this study (CDAS project “PLCO-1138”). All participants provided written informed consent to participate in the study, and the Institutional Review Board of the United States NCI approved the study's methodology (https://cdas.cancer.gov/approved-projects/?study=plco).

In this study, we collected data from the BQ, gathering information on factors such as follow-up years [days from trial entry to cancer diagnosis or death or end of follow-up (December 31, 2009)], age, sex, body mass index (BMI), marital status, race, smoking behavior, family history of bladder cancer and any cancer, and history of diabetes. In addition, the DHQ was used to collect data on alcohol intake, daily energy intake, and dietary food or nutrient intake, which were subsequently used to compute the DRRD score.

On the basis of previous knowledge of the relationship between various food items and diabetes, the DRRD score was initially established with eight food components, which later progressed to include to nine components (12, 14). The previous version of the DRRD score, comprising nine food components, has been applied in previous studies leveraging the PLCO database for cancer research. For our current study, we used an identical version to that used in our previous study (15). In this scoring model, a value of 1 represents a diet associated with the highest risk of type 2 diabetes, while a value of 5 represents the lowest type 2 diabetes risk. The nine dietary factors included cereal fiber, nuts, coffee (caffeinated and decaffeinated), whole fruits, and the ratio of polyunsaturated to saturated fat in ascending order, and GI, trans fat, sugar-sweetened ages/fruit juices, and red and processed meats in descending order. The DRRD score (range = 9–45) is calculated by summing the quintile values of these nine dietary factors. In this investigation, participants were divided into quartiles according to their DRRD scores, using combined quintile cut points for both sexes; Q1 to Q4 represented the levels of adherence to DRRD, ranging from low to high compliance. Baseline characteristics of the participants are presented according to their quartile rankings for the DRRD score, ranging from quartiles 1 to 4.

Throughout the study period, participants were continually monitored until bladder cancer diagnosis, death, or conclusion of the follow-up period (December 31, 2009). Annual questionnaires were dispatched to participants to identify potential bladder cancer cases. Cancer diagnoses were subsequently ascertained by reviewing medical records. Deaths were identified using annually mailed questionnaires and periodic links to the National Death Index. The primary outcome under investigation was the incidence of bladder cancer, a primary malignancy originating in the urinary bladder (International Classification of Diseases for Oncology, Second Edition, codes C67.0–C67.9). The conclusive diagnosis of this condition relies on evaluating the pathologic diagnosis recorded in the medical records of each center (20).

Continuous variables were expressed as mean (standard deviation), while categorical variables were expressed as numbers (percentages). The Kruskal–Wallis test was employed to determine two-sided P values for continuous variables, and the χ² test was used for categorical variables. We used the Cox proportional hazards regression model to compute HRs and 95% confidence intervals (CI) to evaluate the relationship between the DRRD score and bladder cancer incidence for all participants. The proportional hazards assumption was verified using the Schoenfeld residual test. For the calculation of the P_trend, absent covariate values were handled as dummy variables. This trend analysis was based on a continuous variable derived from median values nestled within the DRRD score quartiles. The selection of covariates included in the multivariate regression models was carried out through meticulous literature assessment and expert clinical discernment.

In this study, the following variables were treated as covariates: age at baseline (continuous), sex (male or female), BMI at baseline (continuous), family history of any cancer (yes or no), marital status (married or other), cigarette smoking status (never, former smoker, current smoker), race (non-Hispanic white or others), alcohol intake status (never, former, current, or unknown), daily energy intake (continuous), and history of diabetes (yes or no). We used a restricted cubic spline model with three knots set at the 10th, 50th, and 90th percentiles of the DRRD distribution to explore the potential nonlinear association between the DRRD score and bladder cancer risk after comprehensive adjustment (21). The reference group for comparison was the lowest DRRD score quartile. To ensure the robustness of our findings, sensitivity analyses were conducted, excluding individuals with: (i) extreme daily energy intake (>4,000 kcal/day or <500 kcal/day), (ii) extreme BMI (top and bottom 1% of each sex), (iii) diabetes, and (iv) a follow-up duration of less than 2 years. Considering the differential distribution of DRRD across sexes, we separated the data into male and female groups and carried out a new analysis using sex-specific quintile cut-off points. An exploratory analysis was conducted to investigate the relationship between each dietary component's quintile-based score within the DRRD and the risk of bladder cancer. This analysis used the quintile scores attributed to individual components within the DRRD (representing five intake levels for each component) and their associations with bladder cancer risk were evaluated using multivariate Cox regression. Furthermore, we carried out prespecified subgroup analyses to ascertain if the observed association of DRRD score with bladder cancer incidence was modified by sex (male vs. female), age (>65 vs. ≤65 years old), BMI (>25 vs. ≤25 kg/m²), smoking status (nonsmokers vs. current smokers vs. former smokers), and history of diabetes (yes vs. no). We also conducted a stage-specific analysis to evaluate the associations between DRRD scores and bladder cancer incidence by stage—non–muscle-invasive and muscle-invasive. Statistical analyses were performed using SPSS 25.0 (RRID:SCR_002865) and R 4.2.2.

The datasets presented in this article are not readily available because the data supporting this study's findings are available from the NIH (RRID:SCR_011417) PLCO study group. Restrictions apply to the availability of these data, which were used under license for this study. Requests to access the datasets should be directed to https://biometry.nci.nih.gov/cdas/datasets/plco/.

After applying the exclusion criteria, the study included 99,001 participants, with 761 bladder cancer cases (736 transitional cell carcinoma) identified after a median follow-up of 11.7 years. The DRRD scores ranged from 10 to 45, and the participants were divided into quartiles (25,427 in Q1; 29,633 in Q2; 20,502 in Q3; and 23,289 in Q4). The mean (SD) DRRD score for all participants was 26.88 (5.30), varying from 20.60 (2.29) to 33.58 (2.39) between the lowest and highest quartiles. Compared with individuals in the bottom quartile, those in the top quartile demonstrated reduced daily energy consumption and BMI, were primarily older and female, and had a higher probability of having a family history of cancer, particularly bladder cancer. In addition, this group were less likely to be non-Hispanic white, married, current alcohol consumers, or smokers. Furthermore, they ingested greater quantities of fruits, nuts, coffee, cereal fiber, and polyunsaturated fat compared with saturated fat, while adhering less to high GI diets and consuming fewer trans fats, sugar-sweetened beverages/fruit juices, and red and processed meats. The top quartile also reported a higher percentage of individuals with a definitive family history of any cancer. Table 1 provides comprehensive baseline details.

On the basis of the Schoenfeld residuals test, all P values for the covariates were above the significance level of 0.05, indicating that the data met the proportional hazards assumption, and validating our use of the Cox proportional hazards model. The calculated HRs exhibited a consistent pattern where higher quartiles were correlated with a diminished risk of bladder cancer, both before (HR, 0.65; 95% CI, 0.53–0.82; P_trend < 0.001) and after adjustment for covariates (HR, 0.79; 95% CI, 0.64–0.98; P_trend = 0.007) among all participants in the study. These findings should be interpreted with caution due to the limited sample size. Examining transitional cell carcinoma cases, both crude and adjusted HRs indicated a reduced risk associated with elevated quartiles (unadjusted HR, 0.65; 95% CI, 0.53–0.81; P_trend < 0.001; adjusted HR, 0.79; 95% CI, 0.64–0.99; P_trend = 0.007). In the sex-specific quintile groups, amongst men, we found a significant inverse relationship between DRRD scores and bladder cancer risk (HR, 0.74; 95% CI, 0.58–0.93; P_trend = 0.005). For women, however, no significant association was observed (HR, 0.99; 95% CI, 0.78–1.15; P_trend > 0.05; Table 2), no significant interaction by sex was observed in the analysis (P = 0.185). A restricted cubic spline model was employed to explore the dose–response relationship between the DRRD score and bladder cancer risk, revealing a linear association between the DRRD score and bladder cancer risk (P_nonlinearity > 0.05; Fig. 1). Results showed that bladder cancer risk decreased as DRRD scores increased.

A notable negative association between DRRD scores and bladder cancer risk was more evident in specific subgroups (males and nonobese individuals). However, the P_interaction was not statistically significant (Fig. 2). Sensitivity analyses indicated that HRs remained largely unchanged even after excluding participants with extreme BMI values (upper 1% and lower 1%), excessive energy intake (>4,000 kcal/day or <500 kcal/day), a history of diabetes, or a follow-up duration of <2 years. These findings reinforce the robustness of the strong association between DRRD scores and bladder cancer development (Table 3).

In our exploratory analysis, in which we investigated the relationship between bladder cancer risk and quintile scores of individual DRRD components, several significant associations were uncovered. Notably, the ‘Fruit’ score (adjusted HR, 0.95; 95% CI, 0.93–1.07; P = 0.998) and ‘GI' score (adjusted HR, 0.95; 95% CI, 0.89–0.99; P = 0.048) exhibited an inverse association with bladder cancer risk. Similarly, the ‘Red and Processed Meat' score also exhibited an inverse relationship (adjusted HR, 0.89; 95% CI, 0.83–0.95; P = 0.001). Conversely, the ‘Coffee' score was the only component that demonstrated a positive association (adjusted HR, 1.06; 95% CI, 1.01–1.13; P = 0.033; Table 4). In our stage-specific analysis, we found no apparent differences in associations for noninvasive and muscle-invasive cases. However, it's worth noting that our ability to draw robust conclusions was limited by the relatively small numbers of cases that could be included in these analyses—341 non–muscle-invasive cases and 70 muscle-invasive cases (Supplementary Table S1).

In this extensive prospective study conducted on a PLCO cohort, we discovered a potential association between dietary patterns, as assessed by the DRRD score, and the risk of developing bladder cancer. Our findings consistently indicated a negative association between DRRD scores and bladder cancer incidence, even after accounting for potential confounding variables, which was further validated in the sex-specific quintile group of males. Importantly, the robustness of these results was confirmed by sensitivity analyses, where we excluded patients with limited follow-up, extreme energy intake, and high BMI. Our analyses also revealed a linear dose–response relationship between DRRD tendencies and bladder cancer incidence using restricted cubic spline models, indicating that individuals with higher DRRD scores had a lower risk of developing bladder cancer. Subgroup analyses yielded consistent and encouraging results for males, those with lower BMI, and individuals without diabetes, providing additional evidence supporting the protective role of DRRD in bladder cancer. However, these characteristics did not modify the observed associations, as we found no significant interactions between DRRD scores and these subgroups (P > 0.05). In our exploratory analysis, we investigated the associations between individual components of DRRD and bladder cancer risk. Significant associations were observed among four components. The scores of the three components (high fruit intake, low GI intake, and red and processed meat) were inversely associated with bladder cancer risk. In contrast, high coffee intake was positively associated with the risk of bladder cancer.

DRRD is a dietary pattern comprising nine food components that have been shown decrease the risk of type 2 diabetes (12). Several studies have found a negative association between DRRD scores and tumor occurrence (13–15). The relationship between diabetes and bladder cancer may be attributed to various hormones such as insulin, insulin-like growth factor 1, immune inflammation, and metabolic characteristics like hyperglycemia (22, 23). Elevated sugar levels or insulin resistance may trigger multiple mechanisms that directly or indirectly promote cancer cell proliferation, migration, or invasion. Given the bladder's role as an excretory organ, diet likely plays a critical role in bladder cancer development (24–26).

In recent decades, observational nutritional research has primarily focused on the relationship between individual food items and disease risk. However, given that people consume a diverse mix of foods and nutrients, determining the effects of the interactions between these foods on disease risk is challenging, particularly when using a single food or nutrient-based approach. Consequently, researchers have adopted a more comprehensive approach to evaluate dietary patterns by characterizing the dietary intake of populations using food consumption patterns and examining the potential relationship between these patterns and disease risk. Dietary Approaches to Stop Hypertension, the Mediterranean diet, and the inflammatory potential of the diet (estimated using the DII) have been studied (9, 27, 28), but the results have been inconsistent. Therefore, DRRD, as a dietary pattern, can play a dual role in preventing bladder cancer, leveraging dietary factors to reduce the risk of diabetes, and inhibiting diabetes-related carcinogenesis.

Currently, no studies have explored the potential link between DRRD and the risk of bladder cancer. However, DRRD components, such as a higher intake of fruits (7) and whole grain fiber, (29) have been inversely associated with the incidence of bladder cancer in multiple studies. Furthermore, increased consumption of unsaturated fatty acids (30) and nuts (31) has been linked to a reduced risk of other cancers. Conversely, high GI and intake of red and processed meats, sugar-sweetened beverages, and fats have been associated with a greater risk of bladder cancer (32–35). In our exploratory analysis, we examined the specific association between each component score and the risk of bladder cancer. We found a positive association between coffee scores and bladder cancer, consistent with previous research (36). Interestingly, this presents a contrasting relationship between overall DRRD score and coffee, its individual components, and bladder cancer risk. While DRRD is a dietary pattern designed to reduce the risk of diabetes, it does not consider that high coffee intake may increase the risk of other diseases. Nevertheless, our results indicate that following the DRRD pattern may reduce the risk of bladder cancer. This seemingly contradictory relationship highlights the complexity of dietary patterns. Even if the relationship between individual components and disease risk may contrast, this does not affect the overall association between the dietary pattern and disease risk. This emphasizes the importance of examining dietary patterns holistically rather than focusing solely on individual nutrients or food items when studying nutrition-cancer links.

Although the P values for the interaction were not statistically significant in this study, subgroup analysis revealed an inverse relationship between bladder cancer risk and the DRRD score in the male subgroup, but not in the female subgroup. Similar inverse relationships were observed in the BMI ≤ 25 kg/m² subgroup. The lack of statistical significance in these interactions may be attributed to the relatively small number of cases in the subgroups. Due to the differences in DRRD distribution between males and females, we used sex-specific quintile cut-off points to explore the relationship more thoroughly. Ultimately, the Cox regression analysis results under sex-specific quintiles aligned with the subgroup analysis results. It is worth noting that Dianatinasab and colleagues conducted a study on the association between a Western dietary pattern and bladder cancer incidence in a bladder cancer epidemiology and nutritional determinants cohort and found that male participants had significantly higher compliance with the Western dietary pattern, leading to a higher incidence of bladder cancer in the male subgroup and significant heterogeneity between the male and female subgroups (37). Our study also observed a higher proportion of males in the low DRRD score group (Q1; 59.1%), which gradually decreased with increasing DRRD scores. In addition, as previously mentioned, we found that higher DRRD scores in males were associated with a lower risk of bladder cancer, which is consistent with the findings of Dianatinasab and colleagues, where a higher score for the Western dietary pattern reflects unhealthy eating habits, while a higher DRRD score reflects healthy eating habits (37). This indicates that maintaining good eating habits in male populations can reduce the risk of bladder cancer, whereas the opposite may increase the risk of bladder cancer. Notably, this sex-specific effect is more evident in males than in females, which may be influenced by sex hormones and genetic factors (38). Gender-specific genetic variations may contribute to similar environmental exposures having different effects on bladder cancer incidence in males and females (39). It has also been suggested that sex-specific differences in the metabolism of bladder carcinogens, influenced by sex hormones, may explain the sex differences in bladder cancer risk (40). Further investigation is required to elucidate the association between sex, dietary habits, and bladder cancer development.

Our study provides innovative insights into the association between DRRD and bladder cancer, offering significant contributions not previously explored. Our large-scale population-based study within the PLCO cohort offers a unique and robust design with a large sample size, sufficient follow-up duration, and completion rate, thus reducing potential selection bias. Detailed information on the various dietary factors provided by the DHQ and BQ enabled us to control for potential confounding factors. Furthermore, we employed sensitivity analyses to substantiate the stability of the inverse association between DRRD scores and bladder cancer incidence. As a comprehensive dietary pattern, the DRRD offers a more holistic representation of the relationship between diet and bladder cancer than individual nutrients or components.

Our study has inherent limitations, the most significant being the potential for measurement errors in dietary data due to the reliance on self-reported intake and one-time assessment using the DHQ. This could introduce recall bias and potentially weaken the observed associations between dietary factors and bladder cancer risk. Moreover, long-term changes in dietary habits that were not captured by a single assessment using the DHQ may introduce information bias. In addition, the small size of certain subgroups in our study could limit the statistical power of our analysis, highlighting the need for further research with larger sample sizes. As the study was conducted within a specific demographic group, our results may not be generalizable to other populations or age groups. Therefore, further research is needed to confirm the generalizability of our findings or assess potential demographic differences.

In a prospective study with a large population, we noted a substantial negative association between DRRD scores and the occurrence of bladder cancer. These findings suggest that adherence to a DRRD diet in daily life may serve as a preventative measure against bladder cancer.

No disclosures were reported.

The statements contained herein are solely those of the authors and do not represent or imply concurrence or endorsement by NCI.

Y. Chen: Conceptualization, data curation, software, formal analysis, methodology, writing–original draft. S. Zeng: Conceptualization, data curation, software, formal analysis, methodology, writing–original draft. B. Jiao: Methodology. H. Zhang: Conceptualization, methodology. G. Li: Validation, investigation, visualization. X. Zhang: Supervision, writing–review and editing. X. Hu: Conceptualization, supervision, writing–review and editing.

The authors thank the NCI for access to NCI's data collected by the PLCO Cancer Screening Trial.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.