NHANES is a nationally representative cross-sectional survey targeting the non-institutionalized civilian population of the United States, which is designed as a national program aimed at assessing the health and nutritional status of the American people (21). The NHANES protocol was approved by the Institutional Review Board (IRB) of the National Center for Health Statistics (NCHS), and informed consent was obtained from all participants. The project is conducted biennially and boasts a sample size of over 8,000 individuals. To investigate the relationship between DII and BPH, this study utilized data from five cycles of the NHANES spanning from 1999 to 2008. The selected cycles were chosen due to their comprehensive coverage of variables required for dietary and BPH data, with all data meticulously processed in strict accordance with standardized protocols. Our analysis adhered strictly to pre-established inclusion and exclusion criteria, which included female participants, individuals under the age of 40, and those lacking complete dietary, prostate, or covariate information. Initially, our participant pool consisted of 51,623 individuals. However, after applying these stringent exclusion criteria, our study ultimately encompassed 3,517 participants (Figure 1).

The dietary inflammatory potential of habitual dietary intake was assessed using the DII, a literature-derived scoring system that integrates the pro- and anti-inflammatory effects of individual dietary components (20, 22). The index is constructed from evidence linking specific foods and nutrients to circulating inflammatory biomarkers, including interleukins, tumor necrosis factor-α, and C-reactive protein, thereby enabling a standardized evaluation of diet-related inflammatory burden across populations (20). In large-scale epidemiological studies such as NHANES, DII scores are calculated using dietary data obtained from 24-h dietary recall interviews, which capture detailed information on daily food and beverage consumption (23). Individual nutrient intakes are standardized against a global reference database, converted into centered percentile scores to minimize skewness, and subsequently weighted by their respective inflammatory effect scores derived from the literature (20). The weighted scores of all available dietary components are then summed to generate an overall DII score, with higher values indicating a more pro-inflammatory dietary pattern (24). Notably, the DII has demonstrated robust validity across diverse populations and remains reliably computable even when fewer than the full set of food parameters are available, supporting its applicability in large observational datasets and chronic disease research (25, 26). Ultimately, the sum of the results obtained from all dietary components yields the total Dietary Intake Index value for the study subjects. Detailed information regarding the dietary parameters used in the study is provided in the Supplementary Table S1.

Regarding the diagnosis of BPH, male participants were queried regarding their prostate health with the following questions: “Have you ever been told by a doctor or health professional that you have any disease of the prostate? This includes an enlarged prostate.” If the response was affirmative, an additional question was posed: “Have you been told you have prostate enlargement? Is it benign enlargement? ”

Only participants who answered affirmatively to both questions were categorized as having BPH. Participants with missing data or those diagnosed with malignant hyperplasia were excluded from the study (27–29). These questions are represented by the NHANES questionnaire codes KIQ490, KIQ121, and KIQ141. Participants with other prostate conditions, non-benign enlargement, or missing data were excluded.

Covariates were selected a priori based on biological plausibility and previous literature regarding BPH and metabolic health. Demographic and lifestyle variables included age, race/ethnicity, sex, smoking status, alcohol consumption, educational attainment, household income, and poverty-to-income ratio. Clinical and metabolic factors included body mass index (BMI), diabetes status, and Metabolic syndrome (MetS). MetS was defined according to established criteria and incorporated as a composite metabolic risk indicator to account for its recognized association with BPH (30). All covariates were derived from standardized NHANES questionnaires, physical examinations, and laboratory assessments.

We identified genetic tools for BPH from the latest R12 GWAS summary data in the Finngen database, which encompasses a total of 194,710 European individuals, including 41,137 patients with BPH. The Finngen database is an integrated project that combines the digital health records from the Finnish Health Registry with the genetic data from the Finnish Biobank (https://www.finngen.fi/en). Diagnoses of BPH patients were made according to the ICD-10 diagnostic criteria.

The GWAS data related to inflammatory dietary phenotypes were sourced from the GWAS Catalog. To specifically investigate the impact of inflammatory-related dietary preferences on the risk of BPH, we have selectively filtered and identified nine dietary preference GWAS. These datasets encompass information from up to 421,155 participants registered in the UK Biobank, and detailed information regarding the GWAS summary statistics used in this study can be found in the Supplementary Table S2.

All genetic instruments (SNPs) associated with inflammatory dietary intake were selected at the genome-wide significance level (P < 5 × 10⁻⁸). Using the 1,000 Genomes Project reference panel based on European ancestry, we performed linkage disequilibrium (LD) clumping to identify independent SNPs for each trait. The LD threshold was set at r² < 0.001 within a clumping window of 10 kb. To ensure valid causal inference in the MR analyses, the effects of SNPs on exposure and outcome were harmonized so that they corresponded to the same effect allele. We conducted this harmonization using the “TwoSampleMR” R package and excluded palindromic SNPs with ambiguous strand orientation. To minimize potential confounding, SNPs that were strongly associated with the outcome (P < 5 × 10⁻⁵) were removed prior to analysis. The Steiger directionality test was performed for each instrumental SNP to confirm that the causal direction was from exposure to outcome rather than the reverse. In addition, we calculated the F-statistic for each instrument to assess its strength, with values below 10 indicating a weak instrument (Supplementary Table S3).

The random-effects Inverse-variance weighted (IVW) method was used as the primary analytical approach for all MR analyses. The IVW model provides a weighted regression of SNP-specific causal estimates and yields robust causal inferences, even in the presence of balanced horizontal pleiotropy (31).

To validate the core assumptions of the univariable MR analysis, we performed several sensitivity analyses, including the weighted median, MR-Egger regression, Bayesian weighted Mendelian randomization (BWMR), and MR Pleiotropy RESidual Sum and Outlier (MR-PRESSO) methods. The weighted median estimator computes the weighted median of individual SNP-specific causal effects, providing a consistent estimate when more than 50% of the total weight comes from valid instrumental variables (32). The MR-Egger regression allows for non-zero average horizontal pleiotropic effects but at the cost of reduced statistical power. The BWMR approach, based on a variational expectation-maximization (VEM) algorithm, applies Bayesian weighting to mitigate violations of instrumental variable assumptions caused by pleiotropy, thereby enabling causal inference even under pleiotropic conditions (33). The MR-PRESSO distortion test detects significant differences between causal estimates before and after the removal of outlier SNPs (34). We further applied the Egger intercept test to evaluate potential directional pleiotropy and performed the Steiger directionality test to confirm the causal direction between exposure and outcome. Cochran's Q test was used to assess heterogeneity among instrumental SNPs and to evaluate the consistency of MR assumptions across analyses. Based on these MR estimations and sensitivity analyses, we considered a causal inference to be robust and credible when the following criteria were met: the causal estimates from the four MR approaches and sensitivity analyses showed consistent directions of effect; the MR-Egger intercept indicated no evidence of pleiotropy.

All the above processes were carried out in R 4.1.0 (https://www.R-project.org/). R package “TwoSampleMR” [The MR-Base platform supports systematic causal inference across the human phenome]. And “MRPRESSO” [The MR-Base platform supports systematic causal inference across the human phenome] were used to perform MR and sensitivity analyses.

Twenty-four specific pathogen-free (SPF) male Sprague-Dawley (SD) rats (6 weeks old, body weight 200 ± 20 g) were procured from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). All animals were housed at the Experimental Animal Center of West China Hospital, Sichuan University (Chengdu, Sichuan, China), four rats per cage, with ad libitum access to food and water. The housing environment was maintained under controlled conditions (temperature: 23 °C ± 2 °C; humidity: 55% ± 5%; 12-h light/12-h dark cycle). Rats were maintained under sterilized conditions, with cages subjected to ultraviolet or high-pressure steam sterilization. Upon arrival, animals were randomly assigned to three dietary groups (n = 8 per group). The control group received a standard pellet diet (3.02 kcal/g; 24% protein, 62% carbohydrate, 13% fat), which served as a nutritionally balanced and inflammation-neutral reference diet. To model a pro-inflammatory dietary pattern, rats were fed a high-fat, high-sucrose diet (4.87 kcal/g; 16.4% protein, 41.1% carbohydrate, 42.5% fat), a composition that reflects a positive DII profile and has been widely used to induce chronic low-grade systemic inflammation in rodent models (20, 22, 35). The anti-inflammatory diet was formulated by modifying the AIN-93 purified diet to enrich ω-3 polyunsaturated fatty acids and dietary fiber (3.22 kcal/g; 24.8% protein, 58.4% carbohydrate, 16.8% fat), thereby representing a negative DII profile (36). This design was based on evidence that marine-derived ω-3 polyunsaturated fatty acids suppress pro-inflammatory mediator production and modulate immune signaling pathways, while high dietary fiber enhances short-chain fatty acid production and reduces pro-inflammatory cytokine expression in animal models (37–39). This graded dietary intervention strategy, grounded in the DII framework, was designed to enable a translational evaluation of the impact of dietary inflammatory load on prostatic hyperplasia. All animals were maintained on their respective diets for 12 weeks. At the end of the experimental period, rats were anesthetized by isoflurane inhalation (induction at 3%−4% and maintenance at 1.5%−2.0%). Adequate depth of anesthesia was confirmed prior to sample collection. Blood and prostate tissues were harvested under deep anesthesia to minimize pain and distress, and animals were subsequently euthanized by cervical dislocation in accordance with established guidelines for laboratory animal welfare. Serum samples were centrifuged and stored at −80 °C, while prostate tissues were weighed and either fixed in 4% paraformaldehyde or snap-frozen in liquid nitrogen for further analysis. All animal experimental protocols were reviewed and approved by the Animal Ethics Committee of West China Hospital, Sichuan University, to ensure adherence to the principles of Replacement, Reduction, and Refinement (3Rs) and to promote animal welfare and scientific rigor in experimental design. Approval was granted on 3 September 2024 (approval no. 20240903006). All animal procedures were conducted in strict accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals.

Serum concentrations of TNF-α, IL-1β, and IL-6 were quantified using commercial ELISA kits (Bioswamp, Wuhan, China) in strict accordance with the provided instructions. Absorbance was measured at 450 nm with a microplate reader. Standard curves, generated from the known concentrations of the provided standards, were used to interpolate the cytokine concentrations in the rat samples. Details of the raw measurement data for ELISA experiments, including absorbance OD values at 450 nm and standard curves, are provided in Supplementary Table S4.

Tissue samples were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at a thickness of 4 μm. The sections were subjected to hematoxylin and eosin (HE) staining for histopathological evaluation, and Masson's trichrome staining to visualize collagen deposition. All stained sections were examined under an optical microscope (Olympus, Tokyo, Japan). The collagen-positive areas in Masson's trichrome-stained sections and the Picro-Sirius Red-positive areas in Picro-Sirius Red stain were quantified using ImageJ software.

For IHC analysis, sections were deparaffinized, rehydrated, and incubated overnight at 4 °C with specific primary antibodies. After treatment with appropriate secondary antibodies, antigen visualization was achieved using diaminobenzidine (DAB), followed by counterstaining with Mayer's hematoxylin. Immunoreactivity was evaluated with the IHC Profiler plugin in ImageJ: National Institutes of Health, Bethesda, MD, United States (40), which integrates staining intensity (average gray value) and the percentage of positive area to assign a composite score: high positive (4), positive (3), low positive (2), or negative (1). Details regarding primary antibodies, including sources and dilution ratios, are provided in Supplementary Table S5.

Results are expressed as mean ± standard deviation (SD) from a minimum of three independent replicates. Group comparisons were conducted using Student's t-test (for two groups) or one-way ANOVA (for multiple groups). All statistical analyses and graph generation were carried out with GraphPad Prism version 9.0: GraphPad Software, San Diego, CA, United States. A P-value of less than 0.05 was deemed statistically significant, with the following markers indicating degree of significance: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, and ^****P < 0.0001.

A total of 3,517 participants were included in the analysis, of whom 531 were classified as having BPH and 2,986 served as controls. The mean age was 69 years in the BPH group and 54 years in the non-BPH group. Survey-weighted baseline characteristics stratified by BPH status are presented in Table 1. Compared with participants without BPH, individuals with BPH were more likely to be older, non-Hispanic White, and to have higher educational attainment and Poverty-to-income ratio (PIR). In addition, the BPH group exhibited a higher prevalence of cardiometabolic comorbidities, including hypertension, coronary heart disease, and MetS. Lifestyle-related differences were also observed, with a greater proportion of current or former smokers among participants with BPH. Collectively, these findings indicate that BPH was associated with both adverse metabolic profiles and distinct sociodemographic characteristics in this population.

Our analyses demonstrated a consistent association between the DII and BPH across multiple regression models (Table 2). In survey-weighted logistic regression analyses, higher DII scores were associated with an increased risk of BPH in both unadjusted and adjusted models. In the univariate model, each one-unit increase in DII was associated with a higher odds of BPH. This association remained statistically significant after sequential adjustment for demographic factors, including age, race/ethnicity, and body mass index, as well as additional sociodemographic and clinical covariates. Notably, in the fully adjusted model, higher DII scores continued to show a robust positive association with BPH risk, indicating that the observed relationship was independent of multiple potential confounders.

To further characterize the nature of this association, we examined potential nonlinear relationships between DII and BPH risk (Table 3). The linear term of DII was significantly associated with BPH, whereas the quadratic term was not statistically significant. These findings suggest that the association between DII and BPH risk is predominantly linear, with no evidence supporting a nonlinear dose-response relationship. Collectively, these results indicate that increasing dietary inflammatory potential is linearly and consistently associated with a higher risk of BPH.

Subgroup analyses demonstrated that the positive association between the Dietary Inflammation Index (DII) and BPH risk was generally consistent across a wide range of demographic, socioeconomic, lifestyle, and clinical characteristics (Figure 2). The association appeared more pronounced among participants younger than 65 years compared with those aged 65 years or older. Comparable positive associations were observed across categories of educational attainment, marital status, and income level. With respect to lifestyle and clinical factors, higher DII scores were associated with increased BPH risk in both smokers and non-smokers, as well as in participants with and without hypertension or coronary heart disease. When stratified by metabolic status, the positive association between DII and BPH remained evident in both participants with and without MetS, with no substantial attenuation of effect size across strata. Similarly, the association persisted across categories of serum uric acid levels, with a slightly stronger association observed among individuals with hyperuricemia. Although the association was attenuated and did not reach statistical significance in participants with established diabetes, it remained significant in non-diabetic and pre-diabetic individuals. Overall, these subgroup analyses indicate that the association between dietary inflammatory potential and BPH risk is robust across most subgroups, with only modest variation in magnitude.

After 12 weeks of dietary intervention, rats receiving the pro-inflammatory diet exhibited a significant increase in body weight–adjusted prostate mass compared with both the control and anti-inflammatory diet groups (P < 0.01). The prostate index (PI) was highest in the pro-inflammatory group (mean ± SD: 0.810 ± 0.054 mg/g), moderate in the control group (0.679 ± 0.041 mg/g), and lowest in the anti-inflammatory diet group (0.585 ± 0.044 mg/g), indicating that high dietary inflammatory load promotes prostate enlargement (Table 4).

Hematoxylin–eosin (H&E) staining (Figure 4C) revealed typical hyperplastic features in the pro-inflammatory diet group, including multilayered epithelial proliferation, papillary infoldings into glandular lumina, and reduced luminal area. The control prostates showed regular acinar morphology with a single epithelial layer, whereas the anti-inflammatory diet group preserved normal architecture with minimal epithelial thickening.

Masson's trichrome staining (Figure 4D) demonstrated abundant collagen deposition and pronounced stromal fibrosis in the pro-inflammatory diet group, while the anti-inflammatory group showed sparse collagen fibers comparable to controls. Similarly, Picro-Sirius Red staining under polarized light (Figure 4E) confirmed increased total collagen content and thick, red-stained collagen bundles in the pro-inflammatory group, whereas thinner, green-stained fibers predominated in the anti-inflammatory group, indicating reduced fibrosis.

Immunohistochemical staining for Ki-67 (Figure 4F) demonstrated a marked elevation in proliferative activity within the epithelial cells of the pro-inflammatory group, consistent with hyperplastic histology. In contrast, Ki-67 positivity was sparse in the anti-inflammatory diet group, suggesting inhibition of epithelial proliferation. Furthermore, IHC for IL-1β (Figure 4G) showed intense cytoplasmic staining in both glandular epithelium and stromal regions of the pro-inflammatory group, while weak or minimal IL-1β expression was detected in the anti-inflammatory and control groups. These findings highlight the dietary modulation of local inflammatory signaling within the prostate.

Consistent with the histological observations, serum ELISA assays revealed significantly higher levels of IL-6, TNF-α, and IL-1β in the pro-inflammatory diet group compared with the other groups (P < 0.05). The anti-inflammatory diet group exhibited the lowest cytokine concentrations, approaching baseline values of the controls.

Collectively, these data demonstrate that chronic intake of a pro-inflammatory diet induces BPH–like changes characterized by epithelial proliferation, stromal fibrosis, and both local and systemic inflammation. Conversely, an anti-inflammatory diet attenuates these pathological alterations, preserving normal glandular morphology and suppressing inflammatory cytokine expression.

These findings provide experimental evidence supporting the causal association inferred from NHANES and MR analyses, indicating that inflammatory dietary patterns contribute to BPH pathogenesis through activation of systemic and local inflammatory pathways (Figures 4A–G).