Causal Association Between Gut Microbiota,Plasma Metabolites,and Prostate Cancer: Two-Step Mendelian Randomization Study

Abstract

Objectives

The specific role of metabolites in mediating the relationship between gut microbiota (GM) and prostate cancer (PCa) remains unclear. In this study, we utilized Mendelian randomization (MR) analyses to investigate the causal relationship between 430 GM traits, 1,400 plasma metabolites, and PCa.

Methods

Inverse variance weighted (IVW) method is the primary analytical approach, with supplementary validation using MR-Egger, weighted median, and weighted mode methods. Mediation analysis was performed to assess the contribution of metabolites to the GM-PCa association.

Results

IVW analysis identified 18 GM traits and 66 plasma metabolites showing a causal association with PCa. For instance, P_Firmicutes exhibited a causally protective effect against PCa (IVW: OR=0.937, P=0.008), while Parasutterella showed a causally risk-increasing effect (IVW: OR=1.029, P=0.039). Notably, four metabolites mediated the causal effects of GM on PCa: Subdoligranulum via unidentified metabolite X-25422 (20.785% mediation proportion, P=0.022), Bacteroides via N-acetylarginine (6.706%, P=0.020), and Parasutterella and G_Roseburia via N-acetyl-L-glutamine (-9.382%, P=0.031 and -9.094%, P=0.036, respectively). No significant heterogeneity or horizontal pleiotropy was detected.

Conclusion

These findings offer new insights into the metabolite-mediated pathways linking GM to PCa, which may serve as novel potential biomarkers and therapeutic targets for precision therapies.

Graphical Abstract

Keywords

gut microbiota prostate cancer plasma metabolites mendelian randomization mediation analysis

1. Introduction

In 2022, prostate cancer (PCa) accounted for an estimated 1.467 million new cases and 397,000 deaths globally, ranking as the second most common malignancy and the fifth leading cause of cancer-related mortality in men.¹ In the United States, PCa represents the most common non-cutaneous cancer among men, with an estimated 299,010 new cases and 35,250 deaths projected for 2024². Cancer remains one of the most lethal diseases in contemporary society, with modernization-driven factors, including lifestyle transformations and environmental degradation, have substantially amplified carcinogenic risks.³ To address this escalating challenge, innovative pharmaceutical strategies have emerged at the forefront of oncological research. Notably, novel multidentate ligand metal complexes, particularly L-glutamic acid-based Cu (II) and Ru (III) coordination compounds, demonstrate potent cytotoxicity against diverse cancer cell lines.⁴ Concurrently, thalidomide and its derivatives have experienced a remarkable renaissance following historical setbacks, now serving as critical therapeutic agents for metastatic PCa and other malignancies.⁵ The development of intelligent nanocarriers has substantially enhanced drug delivery efficiency and therapeutic precision.⁶ Furthermore, the biological applications of metal complexes extend beyond oncology, with pyrazoline-based Cu (II) and Ni (II) complexes demonstrating promising antifungal efficacy.⁷

Despite substantial progress in imaging modalities, prostate-specific antigen screening, nanotechnology-enabled drug development, and multimodal therapeutic strategies, the prognoses for advanced PCa remain suboptimal.^6,8 Most metastatic cases inevitably progress to castration-resistant PCa, an aggressive phenotype with limited therapeutic options and a 5-year survival of 30%.⁹ Although several novel agents have demonstrated promising efficacy in in vitro studies of cervical cancer, hepatocellular carcinoma and other malignancies, the rising prevalence of malignancies has imposed a substantial socioeconomic burden globally.^3-5 This underscores the need to elucidate novel pathogenic mechanisms, risk factors, and therapeutic targets.

The gut microbiota (GM), a dynamic and diverse microbial community in the gastrointestinal tract, has emerged as a key regulator of host metabolism and immunity.^10,11 Dysbiosis has been linked to PCa initiation and progression,¹² with taxa such as Streptococcus¹³ and Akkermansia¹⁴ enriched in patients. These microbes may promote tumorigenesis by impairing intestinal barrier function,¹⁵ modulating inflammatory mediators,¹⁶ and reshaping the immune microenvironment, thereby influencing immunotherapy responses.¹⁷

Metabolic reprogramming is a hallmark of cancer, and plasma metabolites reflect alterations in pathways including glycolysis, fatty acid oxidation, and amino acid turnover.¹⁸ Changes in circulating myo-inositol, phosphate, and palmitic acid in lung cancer highlight disruptions in glucose and lipid metabolism.¹⁹ In PCa, branched-chain amino acids,²⁰ sphingosine-1-phosphate,²¹ and urinary glutamate and pseudouridine²² have been proposed as biomarkers for diagnosis and prognosis. Deciphering the GM-metabolite-PCa axis could inform early detection strategies and precision therapeutics.

However, observational studies are often confounded by environmental and clinical variables. Mendelian randomization (MR), which uses genetic variants as instrumental variables (IVs), leverages their random allocation to minimize confounding and strengthen causal inference.²³ For example, Zhang et al used two-sample MR to link fatty acid profiles causally to dilated cardiomyopathy risk.²⁴ The causal interplay and mediating mechanisms among GM, plasma metabolites, and PCa remain largely unexplored. To address this gap, we integrate large-scale genome-wide association study (GWAS) datasets^25-27 and represent the first use of mediation MR to quantify the mediating role of metabolites in the GM-PCa causal pathway. This offers novel causal evidence for microbiota-metabolite-tumor interactions in PCa.

2. Materials and Methods

2.1. Data Sources

Genetic variants for 430 GM traits were obtained from a GWAS dataset of 8,956 German individuals, with accession numbers ranging from GCST90011301 to GCST90011730(25). Of these 430 GM traits, 232 were defined by abundance, and 198 by presence, further divided into eight categories: phyla, classes, orders, families, genera, OTU99, OTU97, and TestASV. We extracted 1,400 plasma metabolites data from the GWAS catalog with accession numbers ranging from GCST90199621 to GCST90201020(26). This dataset included 309 metabolite ratios and 1,091 metabolite levels, covering a broad range of metabolic pathways, including amino acid metabolism, lipid metabolism, and carbohydrate metabolism, making them crucial for understanding the biological mechanisms underlying cancer risks. PCa data were obtained from 122,188 cases and 604,640 controls of European ancestry, registered under accession number GCST90274714 in the GWAS catalog.²⁷ Detailed information of the three GWAS datasets is presented in Table 1. All datasets were collected with informed consent and appropriate ethical approval in the original studies. Thus, no additional ethical approval was required for the present analysis.

Table 1.

Detailed Information of the Three Datasets

Phenotype	Sample size	Ancestry	Journal	Year	Author	PMID	Data source
Gut microbiota	8,956	European	Nature genetics.	2021	Rühlemann et al.	33462482	https://www-nature-com-s.web.bisu.edu.cn/articles/s41588-020-00747-1
Plasma Metabolites	8,299	European	Nature genetics.	2023	Chen Y et al.	36635386	https://www-nature-com-s.web.bisu.edu.cn/articles/s41588-022-01270-1
Prostate cancer	726,828	European	Nature genetics.	2023	Wang A et al.	37945903	https://www-nature-com-s.web.bisu.edu.cn/articles/s41588-023-01534-4

2.2. Selecting Genetic Instruments

We selected single nucleotide polymorphisms (SNPs) as IVs based on the three core assumptions²⁸: 1 strong association with the exposure, 2 no association with potential confounders, and 3 no direct effect on the outcome independent of the exposure. According to previous studies,²⁹ SNPs were selected using a significance threshold of p < 1×10^-5. Linkage disequilibrium clumping was performed using the 1000 Genomes Project Phase 3 as the reference panel, with a clumping window of 10,000 kb and a linkage disequilibrium r² threshold of 0.001. Data harmonization was implemented via the TwoSampleMR (v0.6.20) and ieugwasr (v1.1.0) packages. Strand alignment was inferred from allele frequency information using default settings, and palindromic SNPs with a minor allele frequency close to 0.5 were excluded to avoid alignment ambiguity. F-statistics were calculated using the formula F = Beta²/SE², and only SNPs with F > 10 were retained to ensure instrument strength.

2.3. MR Analysis

2.3.1. Exploring the Causal Relationships Between GM, Plasma Metabolites, and PCa

Initially, we employed two-sample MR to assess the causal relationships between GM, plasma metabolites, and PCa, utilizing methods including inverse variance weighted (IVW), MR Egger, weighted median, and weighted mode approaches. IVW is considered the most precise and robust method.²⁹ The MR Egger method incorporates an intercept term to detect potential pleiotropy.³⁰ The weighted median method, as a supplementary approach, provides consistent estimates when at least 50% of the IVs are valid.³¹ Our analysis was conducted using R version 4.3.1 with the “TwoSampleMR” and “ieugwasr” packages, with statistical significance set at p < 0.05. We applied the Benjamini-Hochberg method for false discovery rate (FDR) correction. Consistent with previous studies,³² with an FDR-adjusted P-value (q value) < 0.2 deemed suggestive of a potential causal association and the q value < 0.05 considered indicative of a significant causal relationship. The primary evaluation metrics included odds ratios (ORs) and their corresponding 95% confidence intervals (CIs).

All reported β coefficients and ORs for GM and plasma metabolites correspond to the effect per 1-standard deviation (SD) increase in the exposure. MR effect estimates for microbial abundance traits reflect the change in the outcome per 1-SD increment in microbial abundance. For microbial prevalence traits, MR estimates denote the change in the outcome per unit increase in genetic liability to microbial presence. MR effect estimates for metabolites represent the change in the outcome per 1-SD increment in log-transformed metabolite levels.

2.3.2. Excluding Potential Reverse Causality Between GM and PCa

We performed a reverse MR analysis by treating PCa as the exposure and GM as the outcome. This approach ensured that the identified causal pathways followed the hypothesized direction from GM to plasma metabolites and subsequently to PCa.

2.3.3. Mediation Analysis

Additionally, we analyzed the causal relationships between candidate GM traits and plasma metabolites to investigate the potential mediating role of plasma metabolites in the GM-PCa causal pathway. The total effect was decomposed into a direct effect (not mediated) and an indirect effect (mediated via plasma metabolites). The indirect effect was then divided into the effect of GM on plasma metabolites (Beta1) and plasma metabolites on PCa (Beta2). The indirect effect was calculated using the formula: indirect effect = Beta1 × Beta³³, with the ratio of indirect effect to the total effect defined as the mediation proportion.

2.3.4. Sensitivity Analysis

Heterogeneity was assessed using Cochran’s Q statistic under both the IVW and MR Egger methods, with P > 0.05 indicating no significant heterogeneity.³⁴ Horizontal pleiotropy was evaluated using the MR Egger intercept test, where the intercept reflects the average directional pleiotropic effect of genetic variants, with a P-value less than 0.05 suggests the presence of horizontal pleiotropy.³⁵ Additionally, symmetry in funnel plots was visually inspected as a qualitative indicator of potential pleiotropy. Leave-one-out analysis was performed to detect and correct influential outliers. Data download and MR analysis were completed in August 2025.

3. Results

3.1. Study Design

Our study was conducted in three steps, as illustrated in Figure 1. First, a bidirectional two-sample MR analysis was performed between GM and PCa to identify GM traits strongly associated with PCa risk and exclude potential reverse causality, thereby estimating the total effect. Second, plasma metabolites significantly associated with PCa risk were identified, obtaining Beta 2. In the third step, we assessed the causal relationships between the identified GM traits and candidate plasma metabolites, estimating Beta 1 and calculating the proportion of the mediation effect. Detailed information of the three GWAS datasets is presented in Table 1.

Figure 1.

A two-step MR study of GM on PCa mediated by plasma metabolites. MR, Mendelian randomization; GM, gut microbiota; PCa, prostate cancer

3.2. Bidirectional MR Analysis Between GM and PCa

Our rigorous MR analysis using the IVW approach identified 18 distinct GM traits with causal relationships to PCa (Figure 2 and Table S1). For example, P_Firmicutes (IVW: OR=0.937, 95% CI [0.893, 0.983], P=0.008), G_Roseburia (IVW: OR=0.945, 95% CI [0.899, 0.992], P=0.024), and Subdoligranulum (IVW: OR=0.946, 95% CI [0.903, 0.992], P=0.022) were found to be protective factors. Notably, the FDR-adjusted P-value (q value) for Bacteroides was 0.056, indicating that the corresponding causal inference is more credible than other findings. In contrast, Parasutterella (IVW: OR=1.029, 95% CI [1.001, 1.057], P=0.039), G_Sutterella (IVW: OR=1.034, 95% CI [1.003, 1.066], P=0.029), and Clostridiales (IVW: OR=1.047, 95% CI [1.006, 1.09], P=0.025) exhibited a positive correlation with PCa risk. Reverse MR revealed that G_Sutterella and Desulfovibrio had reverse causal relationships with PCa, leading to their exclusion from further analysis (Table S2).

Figure 2.

The causal relationships between 18 GM and PCa. GM, gut microbiota; PCa, prostate cancer

3.3. MR Analysis Between Plasma Metabolites and PCa

We obtained the causality between 66 types of plasma metabolite levels and ratios and PCa using IVW method (Table S3). Notably, six metabolite levels and two metabolite ratios were associated with PCa risk across four MR methods, with concordant effect directions (Figure 3). Specifically, higher levels of glucose-to-mannose ratio (IVW: OR=0.955, 95% CI [0.930, 0.981], P=0.001), N-acetylarginine (OR=0.970, 95% CI [0.954, 0.986], P<0.001), N-acetyl-L-glutamine (OR=0.970, 95% CI [0.955, 0.986], P<0.001), 5-methyluridine (ribothymidine; OR=0.978, 95% CI [0.958, 0.999], P=0.043), N-acetylhistidine (OR=0.981, 95% CI [0.964, 0.998], P=0.030), and spermidine-to-ergothioneine ratio (OR=0.991, 95% CI [0.983, 0.999], P=0.028) were associated with a reduced risk of PCa. In contrast, elevated levels of N6-methyllysine (OR=1.021, 95% CI [1.006, 1.037], P=0.006) and unidentified metabolite X-23680 (OR=1.030, 95% CI [1.005, 1.056], P=0.018) were linked to an increased risk.

Figure 3.

The causal relationships between 8 plasma metabolites and PCa with four methods. PCa, prostate cancer

3.4. The Mediating Role of Plasma Metabolites

The MR analyses between 16 GM and 66 plasma metabolites identified 41 potential causal associations using IVW method (Figure 4 and Table S4). We then performed mediation analyses to calculate the mediation P-values, effect sizes (β), and proportions of mediation. Four potential mediation pathways were revealed (Table 2). Specifically, the causal effect of Subdoligranulum on PCa was mediated by unidentified metabolite X-25422 levels (20.785% mediation, P = 0.022); the effect of Bacteroides was mediated by N-acetylarginine levels (6.706% mediation, P = 0.020); the influence of Parasutterella on PCa was negatively mediated by N-acetyl-L-glutamine levels (-9.382% mediation, P = 0.031); and the effect of G_Roseburia was also negatively mediated by N-acetyl-L-glutamine levels (-9.094% mediation, P = 0.036).

Figure 4.

The causal relationships between GM and plasma metabolites. GM, gut microbiota

Table 2.

Mediation Effect of GM on PCa via Plasma Metabolites

Traits of GM	Plasma metabolites	Total effect β (95% CI)	Direct effect A β (95% CI)	Direct effect B β (95% CI)	Mediation effect β (95% CI)	Mediation Proportion (%)	Mediation P_value
OTU99_13 (Subdoligranulum) abundance	X-25422 levels	-0.0551 (-0.1021 - -0.0080)	-0.1765 (-0.2995 - -0.0534)	0.0649 (0.0330 - 0.0967)	-0.0114 (-0.0212 - -0.0017)	20.785	0.022
OTU99_197 (Bacteroides) abundance	N-acetylarginine levels	-0.0239 (-0.0361 - -0.0116)	0.0518 (0.0182 - 0.0854)	-0.0309 (-0.0475 - -0.0143)	-0.0016 (-0.0030 - -0.0003)	6.706	0.020
OTU99_30 (Parasutterella) prevalence	N-acetyl-L-glutamine levels	0.0282 (0.0014 - 0.0550)	0.0877 (0.0223 - 0.1532)	-0.0301 (-0.0458 - -0.0145)	-0.0026 (-0.0050 - -0.0002)	-9.382	0.031
G_Roseburia abundance	N-acetyl-L-glutamine levels	-0.0570 (-0.1065 - -0.0075)	-0.1721 (-0.3054 - -0.0387)	-0.0301 (-0.0458 - -0.0145)	0.0052 (0.0004 - 0.0100)	-9.094	0.036

GM, gut microbiota; PCa, prostate cancer.

Positive mediation proportions indicate that the mediator amplifies the exposure-outcome association, while negative proportions indicate that the mediator attenuates this association. The negative mediation proportions of Parasutterella and G_Roseburia suggest that N-acetyl-L-glutamine acts as an anti-mediator, reducing the overall effect of these bacteria on PCa risk. Specifically, G_Roseburia was negatively associated with N-acetyl-L-glutamine, which in turn exhibited a protective effect against PCa. Consequently, the reduction in this protective metabolite partially counteracted the direct beneficial effect of G_Roseburia, leading to a negative mediation proportion. Parasutterella was positively associated with the protective metabolite N-acetyl-L-glutamine, which in turn reduced PCa risk and partially counteracted the pro-tumorigenic effect of Parasutterella.

3.5. Sensitivity Analysis and Visualization

In all our analyses, the p-values derived from Cochran’s Q statistic using both IVW and MR Egger methods were greater than 0.05, indicating no evidence of significant heterogeneity (Tables S5–S7). Similarly, the p-values from the MR Egger intercept tests exceeded 0.05, suggesting the absence of horizontal pleiotropy (Tables S5–S7). Corresponding forest plots, leave-one-out analyses, scatter plots, and funnel plots of the four significant mediation results are presented in supplementary figures S1–S4.

4. Discussion

Dysbiosis induced by a high-fat diet can compromise intestinal barrier integrity, allowing lipopolysaccharides to translocate into the circulatory system, thereby activating the Interleukin-6/STAT3 signaling pathway and promoting the progression of PCa.³⁶ Moreover, certain gut bacteria capable of synthesizing testosterone³⁷ or converting androgen precursors into bioactive androgens contribute to castration resistance in PCa.³⁸ These findings collectively support the existence of the “gut-prostate” axis, sparking continuous interest in the functional mechanisms of GM in PCa. Recent MR studies have further demonstrated a causal relationship between GM and PCa from a genetic perspective.³⁹ Building on a systematic evaluation of the causal relationships among GM, plasma metabolites, and PCa, our study is the first to introduce plasma metabolites as mediators to assess their bridging role in the GM-PCa causal pathway. We identified 18 GM traits significantly associated with PCa and, notably, revealed that four plasma metabolites act as key mediators in this causal chain.

In line with previous research, our study confirmed that Bacteroides and Ruminococcaceae exert a protective effect against PCa,^40,41 while Butyrivibrio correlates with elevated PCa risk.⁴² Many gut bacteria, primarily within the Firmicutes and Bacteroidetes phyla, are capable of fermenting dietary fibers and other indigestible carbohydrates into short-chain fatty acids (SCFAs), such as acetate, propionate, and butyrate.⁴³ For instance, Ruminococcaceae degrades resistant starch into SCFAs,⁴⁴ while Prevotella metabolizes hexose and pentose sugars into propionate via the succinate pathway.⁴⁵ Through the inhibition of histone deacetylases, SCFAs suppress tumor cell proliferation and migration, and to induce apoptosis.⁴⁶ Conversely, Matsushita et al suggested that SCFAs may promote PCa progression by activating the IGF1-MAPK/PI3K signaling pathway,⁴⁷ indicating a potentially dual role. However, these studies did not distinguish the specific biological effects of individual SCFAs. Our study found that SCFAs-producing bacteria, including Firmicutes and Ruminococcaceae, were negatively associated with PCa risk, whereas Clostridiales were positively associated, highlighting the complexity and context-dependence of SCFAs function. Moreover, Clostridiales are known to catalyze the side-chain cleavage of glucocorticoids into androgens, facilitating PCa progression,⁴⁸ which is consistent with our findings.

By disrupting glycolysis, decreasing mitochondrial membrane potential, increasing reactive oxygen species levels, and reducing ATP production, ultimately inducing mitochondrial apoptosis, mannose can inhibit the growth and proliferation of PCa cells.⁴⁹ Additionally, mannose can suppress the Warburg effect in tumor cells, thereby slowing their proliferation and enhancing sensitivity to chemotherapeutic agents.⁴⁹ Although vitro and animal experiments have preliminarily demonstrated its antitumor effects,⁵⁰ further investigation is required to validate its clinical efficacy. On the other hand, spermidine has been reported to promote ferroptosis in PCa cells by modulating proteasome activity.⁵¹ We further identified that glucose-to-mannose ratio and the spermidine-to-ergothioneine ratio were protective factors against PCa and underscored the critical role of metabolite ratios in cancer interventions.

5-methyluridine, a key metabolite in the folate metabolic pathway, exerts antitumor effects by regulating the cell cycle and inhibiting cell proliferation.⁵² Clinical studies have demonstrated that 5-methyluridine levels are significantly lower in patients with PCa compared to those with benign prostatic hyperplasia,⁵³ suggesting its potential as a biomarker for early diagnosis and prognosis. Our study further provides robust evidence supporting a negative association between 5-methyluridine and PCa.

In our mediation analysis, the unknown metabolite X-25422 levels accounted for 20.785% mediation proportion of the causal pathway between Subdoligranulum and PCa. Overall, Subdoligranulum exhibited a protective effect against PCa, whereas X-25422 levels were positively associated with PCa risk. X-25422 levels are downregulated by Subdoligranulum, thereby exerting its mediating role. Notably, Subdoligranulum is a major butyrate-producing bacterium in the gut,⁵⁴ and its metabolite butyrate—an essential energy source for colonic epithelial cells—plays a critical role in preserving intestinal mucosal integrity and barrier function.⁵⁵ Moreover, it mitigates inflammation by downregulating pro-inflammatory cytokines such as Interleukin-17 and upregulating anti-inflammatory cytokines like Interleukin-22, thereby contributing to a reduced risk of gastrointestinal tumorigenesis.⁵⁶

Our study identified N-acetylarginine and N-acetyl-L-glutamine as potentially critical protective mediators against PCa. Notably, N-acetylarginine is a key N-acetylated amino acid metabolite and serves as a central node in the arginine metabolic pathway. Serum metabolomic profiling demonstrated that the level of N-acetylarginine was statistically significantly lower in men who were subsequently diagnosed with PCa.⁵⁷ Furthermore, prospective serum metabolomic analysis validated a statistically significant reduction in N-acetylarginine levels in patients with T4 stage PCa.⁵⁸ In contrast, N-acetyl-L-glutamine, a core derivative of glutamine metabolism, has remained relatively understudied in the context of PCa. Further metabolomics-based studies are thus warranted to clarify the metabolic pathways and functional mechanisms of these identified metabolites in tumors.

Identification of novel GM-metabolite axes and metabolic ratio signatures linked to PCa pathogenesis enabled us to establish a framework for advancing PCa risk stratification and therapeutic optimization. The protective effects of Bacteroides, Ruminococcaceae, and Subdoligulum support probiotic supplementation or fecal microbiota transplantation as potential primary prevention strategies, while Butyrivibrio-based risk stratification tools could facilitate early screening of high-risk populations. As integrated biomarkers that better reflect metabolic network dynamics than individual metabolites, these metabolic ratios require validation of their supplementary diagnostic value in prostate-specific antigen screening-negative cohorts via prospective studies. For translational research, we propose a “microbiota-metabolite-host” trinity multi-omics platform, and recommend prospective nested case-control studies to verify causal associations and assess the safety and efficacy of metabolic interventions combined with standard therapies (e.g., androgen deprivation therapy or novel endocrine therapies) for PCa.

Our study provides compelling evidence supporting the pivotal role of GM and metabolites in PCa. Nonetheless, this study has certain limitations. In our present MR analysis, most associations did not meet the strict FDR threshold of < 0.05, likely due to limited statistical power, mild genetic heterogeneity, and minor measurement errors in exposure or outcome phenotypes. Notably, our findings are consistent with previous MR studies where marginally significant signals were later validated in larger datasets. Future multi-ancestry GWAS with expanded sample sizes will be critical to enhance statistical power and permit more robust identification of modest causal effects.

5. Conclusion

Our two-sample MR analysis identified 18 GM traits and 66 plasma metabolites with potential causal effects on PCa risk, among which 4 metabolites acted as mediators. These findings provide novel mechanistic insights into microbiota-driven PCa pathogenesis, highlighting the intricate crosstalk between the gut ecosystem and host metabolism in modulating PCa susceptibility. We further recognized protective microbial taxa (e.g., Bacteroides, Ruminococcaceae, Subdoligranulum) and metabolites (e.g., mannose, spermidine, N-acetylarginine), which represent promising targets for mechanism-guided preventive strategies. Future translational studies are warranted to validate these microbial and metabolic biomarkers in prospective cohorts and explore their utility as personalized intervention targets for PCa management.

Supplemental Material

Supplemental Material - Causal Association Between Gut Microbiota, Plasma Metabolites, and Prostate Cancer: Two-Step Mendelian Randomization Study

Supplemental Material for Causal Association Between Gut Microbiota, Plasma Metabolites, and Prostate Cancer: Two-Step Mendelian Randomization Study by Junxiong Li, Jian Hou, Haolin Liu, Shicheng Fan, Yumin Wang, Pinyao Liang, Jingbo Qin, Peng Gu, Xiaodong Liu and Xiangyang Wen in Dose-Response.

Footnotes

Acknowledgements

We acknowledge the use of summary-level data from the NHGRI-EBI GWAS Catalog. We thank all investigators and participants involved in these studies.

ORCID iD

Jian Hou

Ethical Considerations

This study leveraged publicly available GWAS datasets, which had already acquired ethical approval in the primary investigations.

Consent to Participate

Participant consent was obtained in the primary investigations.

Author Contributions

J.X. Li, J. Hou, and H.L. Liu conceived and designed the study. J.X. Li and S.C. Fan analyzed the data. H.L. Liu, Y.M. Wang, and P. Gu prepared the supplementary materials. P.Y. Liang, J.B. Qin, and X.D. Liu generated the figures and tables. X.Y. Wen supervised the project. J.X. Li drafted the manuscript. All authors contributed to manuscript revision and approved the final version.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data utilized in this study were provided in the table and supplementary materials.*

Supplemental Material

Supplemental material for this article is available online.

References

Bray

Laversanne

Sung

, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: a cancer journal for clinicians. 2024;74:229-263.

Raychaudhuri

Lin

Montgomery

. Prostate Cancer: A Review. Jama. 2025;333:1433-1446.

Ali

Saleem

Aboul-Enein

Rather

Imran

. Social Aspects of Cancer Genesis. Cancer Therapy. 2011;8:6-14.

Ali

Wani

Saleem

Wesselinova

. Syntheses, DNA binding and anticancer profiles of L-glutamic acid ligand and its copper(II) and ruthenium(III) complexes. Medicinal chemistry (Shariqah (United Arab Emirates)). 2013;9:11-21.

Imran

Waseem

Kishwar

Ashanul

. Thalidomide: A Banned Drug Resurged into Future Anticancer Drug. Current Drug Therapy. 2012;7:13-23.

Ali

Nadeem Lone

Suhail

Danish Mukhtar

Asnin

, Advances in Nanocarriers for Anticancer Drugs Delivery. Curr Med Chem. 2016;23:2159-2187.

Ali

Wani

Khan

, et al. Synthesis and synergistic antifungal activities of a pyrazoline based ligand and its copper(II) and nickel(II) complexes with conventional antifungals. Microbial pathogenesis. 2012;53:66-73.

Terlizzi

Bossi

. High-risk Locally Advanced Prostate Cancer: Multimodal Treatment Is the Key. European urology open science. 2022;38:14-16.

Kensler

Rebbeck

. Cancer Progress and Priorities: Prostate Cancer. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. 2020;29:267-277.

10.

Nobels

van Marcke

Jordan

Van Hul

Cani

. The gut microbiome and cancer: from tumorigenesis to therapy. Nature metabolism. 2025;7:895-917.

11.

Jyoti

Dey

. Mechanisms and implications of the gut microbial modulation of intestinal metabolic processes. npj Metabolic Health and Disease. 2025;3:24.

12.

Matsushita

Fujita

Hatano

, et al. Emerging Relationship between the Gut Microbiome and Prostate Cancer. The world journal of men's health. 2023;41:759-768.

13.

Liss

White

Goros

, et al. Metabolic Biosynthesis Pathways Identified from Fecal Microbiome Associated with Prostate Cancer. European urology. 2018;74:575-582.

14.

Daisley

Chanyi

Abdur-Rashid

, et al. Abiraterone acetate preferentially enriches for the gut commensal Akkermansia muciniphila in castrate-resistant prostate cancer patients. Nature communications. 2020;11:4822.

15.

Ivleva

Grivennikov

. Microbiota-driven mechanisms at different stages of cancer development. Neoplasia (New York, N.Y.). 2022;32:100829.

16.

Tilg

Zmora

Adolph

Elinav

. The intestinal microbiota fuelling metabolic inflammation. Nature reviews. Immunology. 2020;20:40-54.

17.

Zitvogel

Raoult

Kroemer

Gajewski

, The microbiome in cancer immunotherapy: Diagnostic tools and therapeutic strategies. Science (1979). 2018;359:1366-1370.

18.

Kumar

Mishra

Gautam

, et al. Metabolic Pathways, Enzymes, and Metabolites: Opportunities in Cancer Therapy. Cancers. 2022;14:5268.

19.

Bamji-Stocke

van Berkel

Miller

Frieboes

. A review of metabolism-associated biomarkers in lung cancer diagnosis and treatment. Metabolomics : Official journal of the Metabolomic Society. 2018;14:81.

20.

Bidgood

Philp

Rockstroh

, et al. Targeting valine catabolism to inhibit metabolic reprogramming in prostate cancer. Cell death & disease. 2024;15:513.

21.

Nunes

Naymark

Sauer

, et al. Circulating sphingosine-1-phosphate and erythrocyte sphingosine kinase-1 activity as novel biomarkers for early prostate cancer detection. British journal of cancer. 2012;106:909-915.

22.

Yang

Zhang

Cheng

Griebel

Neuhaus

. Novel Metabolic Signatures of Prostate Cancer Revealed by (1)H-NMR Metabolomics of Urine. Diagnostics (Basel, Switzerland). 2021;11:149.

23.

Davies

Holmes

Davey Smith

. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ (Clinical research ed.). 2018;362:k601.

24.

Zhang

Luo

Hou

, et al. Fatty acids and risk of dilated cardiomyopathy: A two-sample Mendelian randomization study. Frontiers in nutrition. 2023;10:1068050.

25.

Rühlemann

Hermes

Bang

, et al. Genome-wide association study in 8,956 German individuals identifies influence of ABO histo-blood groups on gut microbiome. Nature genetics. 2021;53:147-155.

26.

Chen

Pettersson-Kymmer

, et al. Genomic atlas of the plasma metabolome prioritizes metabolites implicated in human diseases. Nature genetics. 2023;55:44-53.

27.

Wang

Shen

Rodriguez

, et al. Characterizing prostate cancer risk through multi-ancestry genome-wide discovery of 187 novel risk variants. Nature genetics. 2023;55:2065-2074.

28.

Nie

Luo

Yan

Zhang

Wang

. Rheumatoid arthritis and coronary atherosclerosis: a two-sample Mendelian randomization study. Frontiers in cardiovascular medicine. 2023;10:1033644.

29.

Wang

Gao

Zhang

, et al. Causal relationship between the gut microbiota and insomnia: a two-sample Mendelian randomization study. Frontiers in cellular and infection microbiology. 2024;14:1279218.

30.

Burgess

Thompson

. Interpreting findings from Mendelian randomization using the MR-Egger method. European journal of epidemiology. 2017;32:377-389.

31.

Bowden

Davey Smith

Haycock

Burgess

. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genetic epidemiology. 2016;40:304-314.

32.

Zhang

Liu

. Causal role of immune cells in varicose veins: insights from a Mendelian randomization study. Scientific reports. 2025;15:19527.

33.

MacKinnon

Lockwood

Hoffman

West

Sheets

. A comparison of methods to test mediation and other intervening variable effects. Psychological methods. 2002;7:83-104.

34.

Higgins

Thompson

Deeks

Altman

. Measuring inconsistency in meta-analyses. BMJ (Clinical research ed.). 2003;327:557-560.

35.

Bowden

Davey Smith

Burgess

. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. International journal of epidemiology. 2015;44:512-525.

36.

Zhong

Long

, et al. Gut dysbiosis promotes prostate cancer progression and docetaxel resistance via activating NF-κB-IL6-STAT3 axis. Microbiome. 2022;10:94.

37.

Cross

Kasahara

Rey

. Sexual dimorphism of cardiometabolic dysfunction: Gut microbiome in the play? Molecular metabolism. 2018;15:70-81.

38.

Pernigoni

Zagato

Calcinotto

, et al. Commensal bacteria promote endocrine resistance in prostate cancer through androgen biosynthesis. Science. 2021;374:216-224.

39.

Wang

Zheng

Yin

, et al. Causal relationship between gut microbiota and prostate cancer contributes to the gut-prostate axis: insights from a Mendelian randomization study. Discover oncology. 2024;15:58.

40.

Xie

. Effects of gut microbiota on prostatic cancer: a two-sample Mendelian randomization study. Frontiers in microbiology. 2023;14:1250369.

41.

Liu

Yang

Wang

, et al. Sex hormones, blood metabolites and proteins mediating the causal associations between gut microbiota and prostatic diseases: evidences from Mendelian randomization study. Prostate international. 2025;13:49-59.

42.

Yue

, et al. Causality investigation among gut microbiota, immune cells, and prostate diseases: a Mendelian randomization study. Frontiers in microbiology. 2024;15:1445304.

43.

Cao

Shen

, et al. Elucidating the specific mechanisms of the gut-brain axis: the short-chain fatty acids-microglia pathway. Journal of neuroinflammation. 2025;22:133.

44.

Duncan

Louis

Flint

. Ruminococcus bromii is a keystone species for the degradation of resistant starch in the human colon. The ISME journal. 2012;6:1535-1543.

45.

Zhang

Lingga

De Groot

Hackmann

. The oxidoreductase activity of Rnf balances redox cofactors during fermentation of glucose to propionate in Prevotella. Scientific reports. 2023;13:16429.

46.

Zhang

Shen

, et al. Short-Chain Fatty Acids and Their Association with Signalling Pathways in Inflammation, Glucose and Lipid Metabolism. International journal of molecular sciences. 2020;21:6356: 1-6356: 16.

47.

Matsushita

Fujita

Hayashi

, et al. Gut Microbiota-Derived Short-Chain Fatty Acids Promote Prostate Cancer Growth via IGF1 Signaling. Cancer research. 2021;81:4014-4026.

48.

Ridlon

Ikegawa

Alves

JMP

, et al. Clostridium scindens: a human gut microbe with a high potential to convert glucocorticoids into androgens. Journal of lipid research. 2013;54:2437-2449.

49.

Deng

Liu

Cai

, et al. Mannose inhibits the growth of prostate cancer through a mitochondrial mechanism. Asian journal of andrology. 2022;24:540-548.

50.

Jin

, et al. Mannose inhibits PKM2 lactylation to induce pyroptosis in bladder cancer and activate antitumor immune responses. Communications biology. 2025;8:689.

51.

Feng

Zhang

Niu

, et al. Spermidine inactivates proteasome activity and enhances ferroptosis in prostate cancer. Acta pharmaceutica Sinica. B. 2025;15:2095-2113.

52.

Zhou

Cao

Wang

. Methylation modifications in tRNA and associated disorders: Current research and potential therapeutic targets. Cell proliferation. 2024;57:e13692.

53.

Buzatto

de Oliveira Silva

Poppi

Simionato

AVC

. Assessment of nucleosides as putative tumor biomarkers in prostate cancer screening by CE-UV. Analytical and bioanalytical chemistry. 2017;409:3289-3297.

54.

Radjabzadeh

Bosch

Uitterlinden

, et al. Gut microbiome-wide association study of depressive symptoms. Nature communications. 2022;13:7128.

55.

Liu

Wang

, et al.

Butyrate: A Double-Edged Sword for Health?

Advances in nutrition (Bethesda, Md. 2018;9:21-29.

56.

Siddiqui

Cresci

GAM

. The Immunomodulatory Functions of Butyrate. Journal of inflammation research. 2021;14:6025-6041.

57.

Huang

Mondul

Weinstein

, et al. Serum metabolomic profiling of prostate cancer risk in the prostate, lung, colorectal, and ovarian cancer screening trial. British journal of cancer. 2016;115:1087-1095.

58.

Huang

Mondul

Weinstein

Karoly

Sampson

Albanes

. Prospective serum metabolomic profile of prostate cancer by size and extent of primary tumor. Oncotarget. 2017;8:45190-45199.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.16 MB