Gut microbiome, dietary habits, and prostate cancer: a two-step Mendelian randomization revealing the causal associations

Abstract

Background

Recent studies suggest that diet fizzy drinks may contribute to prostate cancer (PCa) development. However, the causal effects between diet fizzy drinks and PCa and whether gut microbiota (GM) act as a mediator remain unclear.

Methods

We conducted two-sample Mendelian Randomization (MR) analyses utilizing large-scale genome-wide association studies (GWAS) data from the UK Biobank, the MiBioGen consortium, and PCa-related datasets. The inverse-variance weighted (IVW) method was used to evaluate the causal effects of GM and dietary preferences on PCa risk. A mediation analysis was performed to investigate whether GM mediates the relationship between dietary factors and PCa risk.

Results

Diet fizzy drink consumption was causally associated with reduced PCa risk (OR = 0.83, 95% CI: 0.70–0.99, P = 0.041) and decreased abundance of PCa-risk-related GM taxa (Negativicutes and Selenomonadales). Mediation analysis did not reveal a statistically significant mediation effect, with a mediation proportion of 16% (95% CI: − 0.06–0.37, P = 0.13).

Conclusion

Consumption of diet fizzy drinks may reduce the risk of PCa, potentially through modulation of the GM; however, further studies are required to confirm these findings and clarify underlying mechanisms.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-02172-4.

Introduction

Prostate cancer (PCa), the second most frequently diagnosed cancer among men, significantly contributes to cancer-related mortality worldwide [1]. Common diagnostic methods for PCa include prostate biopsy, prostate-specific antigen (PSA) testing, digital rectal examination, magnetic resonance imaging (MRI), and routine health screenings [2]. Identified risk factors for PCa encompass family history, ethnicity, age, obesity, and environmental influences [2]. PCa is a heterogeneous disease with variations in its epidemiology and genetic characteristics. The complex interactions between genetic predispositions, environmental factors, and social determinants contribute to disparities in PCa survival rates across racial groups, which further leads to differences in PCa epidemiology between countries [3].

Gut microbiota (GM) refers to the collection of microorganisms residing in the gut, including bacteria, viruses, parasites, and fungi [4, 5]. Advances of the next-generation sequencing technologies development enables the deeper exploration of GM composition and their associations with cancer. Currently, GM has been identified as one of the factors affecting the incidence, development and response of prostate cancer [6–8]. Huang et al. reported that the decreased alpha-diversity of GM has been detected in patients with prostate cancer compared with the control group [4]. In terms of GM species, the ratio of Proteobacteria, Bacteroidia, Clostridia, Bacteroidales, Clostridiales, Prevotellaceae, Lachnospiraceae, Prevotella, Escherichia-Shigella, Faecalibacterium, and Bacteroides was higher in patients with prostate cancer [4], while the abundance ratio of Actinobacteria, Bacteroidetes, Firmicutes, Selenomonadales, Veillonella, and Megasphaera was higher in the control group. Diet participates in the determination of gut microbiome integrity [9], and high fat diet was reported to induce the imbalance in intestinal microbes [10]. In animal models, nutrients like fat, protein, carbohydrates, vitamins, and polyphenols have shown the potential of influencing the pathogenesis and progression of PCa [11]. Given the correlation between diet, GM, and PCa could be established preliminarily, the current study suggests that intervention of dietary patterns may contribute to the prevention of PCa via regulating GM [11], and the concept of “gut-prostate cancer axis” was proposed accordingly [12].

Mendelian randomization (MR) is an emerging approach to investigate causal relationships between exposure factors and disease outcomes [13]. While numerous studies have examined associations between GM and PCa, their causal relationship remains unclear, and large cohort studies on how dietary patterns influence PCa development are challenging to conduct. In this study, we apply MR to explore the causal relationship between GM, dietary preferences, and the risk of developing PCa.

Methods and materials

Data sources

Data on dietary habits were obtained from the study by May-Wilson et al., which utilized UK Biobank data to examine genetic influences on food preferences. In this study, dietary habits were assessed through questionnaires completed by participants of European ancestry, focusing on 144 diet-related traits. Genotyping was conducted using UK Biobank arrays, followed by GWAS to identify genetic variants associated with these food preference traits. Hierarchical factor analysis was applied to group related food items and uncover latent factors. Additionally, genetic correlations between food preference traits and other complex traits were analyzed, and gene prioritization was conducted to identify genes likely contributing to each dietary trait [14].

The study by Kurilshikov et al. provides GWAS data on the gut microbiome. The MiBioGen consortium curated and analyzed genome-wide genotypes alongside 16S rRNA fecal microbiome data from 18,340 individuals. Microbial composition exhibited high variability across cohorts, with only 95 out of 410 genera detected in more than 9% of samples. This GWAS of host genetic variation in microbial taxa identified 31 loci associated with the microbiome, each meeting the genome-wide significance threshold (P < 5 × 10⁻⁸) [15].

In this study, the outcome variable is PCa. To enhance the robustness of our analysis, we used two distinct prostate cancer datasets from the IEU database, labelled ieu-b-85 and ebi-a-GCST006085, and conducted separate analyses on each. These datasets were derived from a study by Schumacher et al., published in 2018. This meta-analysis included genotype data from 46,939 PCa cases and 27,910 controls of European ancestry obtained using a custom high-density array, along with previously genotyped data from 32,255 PCa cases and 33,202 European ancestry controls. The study ultimately identified 62 novel loci significantly associated with PCa (P < 5 × 10⁻⁸), as well as one locus significantly associated with early-onset PCa [16].

Study design

This study aims to explore the causal relationship between diet, GM, and PCa and to determine whether GM mediates the association between diet and PCa. First, we performed two-sample MR analyses to separately assess the causal effects of GM and dietary preferences on PCa. Next, we conducted two-sample MR analyses on the significant findings from these assessments to identify any causal relationships between PCa-related dietary factors and PCa-related GM taxa. Finally, we conducted a mediation MR analysis with diet as the exposure, GM as the mediator, and PCa as the outcome. Each of the first two steps was repeated using two distinct PCa datasets with unique IDs, and the intersecting results were used in the final analysis steps. The detailed study design is shown in Fig. 1.

Fig. 1.

Study design. Our study is divided into four steps. The first step is to identify which gut microbiome are associated with prostate cancer. The second step is to determine which dietary liking may be related to prostate cancer. In the third step, we perform Mendelian randomization on the positive results from the first and second steps to identify the associations between gut microbiota and dietary habits linked to prostate cancer. The final step is to conduct a mediation analysis involving the relevant dietary liking, microbiome, and prostate cancer

Mendelian randomization

For the two-sample MR analysis, we strictly adhered to the three core assumptions required for MR. The first is the relevance assumption, which requires that the selected SNPs are strongly associated with the exposures, specifically gut microbiota and dietary preferences. To ensure a sufficient number of SNPs for analysis, we applied a significance threshold of P < 5 × 10⁻⁶ for gut microbiota and P < 5 × 10⁻⁸ for dietary preferences, both of which are widely accepted in MR studies. The second assumption is independence, indicating that the SNPs used in this study are not associated with any confounding factors. The third assumption is the exclusion restriction, which stipulates that the SNPs influence PCa only through the exposures (GM or dietary preferences) and not through any other pathways. In Fig. 1, the second and third assumptions are represented by dashed lines, while the first assumption is represented by a solid line. The mediation MR analysis is based on the traditional two-step method proposed by Carter et al. To maintain consistency throughout the study, we applied the same P-value thresholds as in the previous two-sample MR analyses: P < 5 × 10⁻⁸ for the association between the exposure and the mediator, P < 5 × 10⁻⁶ for the mediator and the outcome, and P < 5 × 10⁻⁸ for the exposure and the outcome [17].

Instrumental variables (IVs) were further processed, the clump was set to remove the IVs for linkage disequilibrium analysis with parameters R²= 0.001 and kb = 10,000, as well as the weak IVs were omitted based on the F-statistic for each SNP, the F values was conculcated as the following equation: F = [R²/(1-R²)] × (N-K-1)/K, and F < 10 was considered as weak IVs [18].

Statistics analysis

The primary results of the MR analyses were obtained using the inverse-variance weighted (IVW) method. To support these findings, we conducted additional analyses using MR Egger, weighted median, and weighted mode methods. For significant associations, we further performed heterogeneity and pleiotropy tests to evaluate the robustness of the results. In cases where the number of instrumental variables was limited, sensitivity or heterogeneity tests could not be conducted. Statistical significance for the IVW analysis was defined as a P-value < 0.05. For the heterogeneity test, a P-value > 0.05 indicated the absence of heterogeneity. In addition to the standard pleiotropy test, the MR-PRESSO test was applied in analyses with more than three SNPs to detect potential horizontal pleiotropy. Results were considered free from horizontal pleiotropy if all P-values exceeded 0.05. The software used in this study included R Studio with R version 4.3.0, and relevant R packages such as “TwoSampleMR” and “MRPRESSO.”

Results

Causal associations between gut microbiome and prostate cancer

After a rigorous selection process, a total of 196 gut microbiota taxa had sufficient data for MR analysis with PCa. Harmonized data are provided in Tables S2 and S4, while all results are summarized in Tables S1 and S3. Across two datasets with distinct IDs, 13 gut microbiota taxa (ID: ieu-b-85) and 14 gut microbiota taxa (ID: ebi-a-GCST006085) demonstrated statistically significant associations with PCa, with 10 taxa showing consistent positive associations in both datasets. These taxa include class Alphaproteobacteria (OR = 0.81, P = 0.001), genus unknowngenus (ID: 2755) (OR = 0.87, P = 0.020), family Rhodospirillaceae (OR = 0.89, P = 0.049), genus Holdemania (OR = 0.89, P = 0.014), genus Adlercreutzia (OR = 0.89, P = 0.033), genus Anaerofilum (OR = 1.09, P = 0.033), genus unknowngenus (ID: 2041) (OR = 1.13, P = 0.033), class Negativicutes (OR = 1.15, P = 0.011), order Selenomonadales (OR = 1.15, P = 0.011), and family Porphyromonadaceae (OR = 1.21, P = 0.011). All results showed no evidence of heterogeneity or horizontal pleiotropy. The statistically significant results are presented in Table 1 and visualized with forest plots in Fig. 2.

Table 1.

Positive results of Mendelian randomized analysis for gut microbiome and prostate cancer

Outcome: prostate cancer, ID: ieu-b-85			P-value
Exposures	SNP number	OR (95%CI)	IVW outcome	Heterogeneity test	Pleiotropy test	MR Presso test
class.Alphaproteobacteria	3	0.81 (0.71–0.92)	0.001	0.422	0.659	–
genus.unknowngenus(id.2755)	6	0.87 (0.77–0.98)	0.020	0.118	0.676	0.20
family.Rhodospirillaceae	7	0.89 (0.79–0.99)	0.049	0.062	0.103	0.07
genus.Holdemania	7	0.89 (0.80–0.98)	0.014	0.265	0.638	0.31
genus.Adlercreutzia	7	0.89 (0.80–0.99)	0.033	0.136	0.678	0.17
genus.Allisonella	3	0.89 (0.83–0.96)	0.002	0.981	0.935	–
genus.Anaerofilum	6	1.09 (1.01–1.19)	0.033	0.268	0.873	0.41
genus.Prevotella7	4	1.10 (1.01–1.20)	0.026	0.221	0.566	0.34
genus.unknowngenus(id.2041)	6	1.13 (1.01–1.27)	0.033	0.068	0.615	0.13
family.Veillonellaceae	7	1.14 (1.00–1.29)	0.044	0.205	0.336	0.27
class.Negativicutes	8	1.15 (1.03–1.29)	0.011	0.644	0.128	0.61
order.Selenomonadales	8	1.15 (1.03–1.29)	0.011	0.644	0.128	0.64
family.Porphyromonadaceae	5	1.21 (1.04–1.40)	0.011	0.456	0.606	0.47

Outcome: prostate cancer, ID: ebi-a-GCST006085			P-value
Exposures	SNP number	OR (95%CI)	IVW outcome	Heterogeneity test	Pleiotropy test	MR Presso test
class.Alphaproteobacteria	3	0.81 (0.71–0.92)	0.001	0.422	0.659	–
family.Porphyromonadaceae	5	1.21 (1.04–1.40)	0.011	0.456	0.606	0.43
class.Negativicutes	8	1.15 (1.03–1.29)	0.011	0.644	0.128	0.60
order.Selenomonadales	8	1.15 (1.03–1.29)	0.011	0.644	0.128	0.71
genus.Anaerofilum	7	1.09 (1.02–1.17)	0.012	0.378	0.880	0.38
genus.unknowngenus(id.2755)	6	0.87 (0.77–0.98)	0.020	0.118	0.676	0.17
family.Acidaminococcaceae	5	0.87 (0.78–0.98)	0.023	0.903	0.699	0.90
genus.Enterorhabdus	4	1.13 (1.01–1.26)	0.026	0.230	0.708	0.04
genus.Holdemania	8	0.90 (0.82–0.99)	0.029	0.239	0.817	0.19
genus.DefluviitaleaceaeUCG011	8	0.92 (0.85–0.99)	0.030	0.638	0.556	0.60
genus.Adlercreutzia	7	0.89 (0.80–0.99)	0.033	0.136	0.678	0.14
genus.unknowngenus(id.2041)	6	1.13 (1.01–1.27)	0.033	0.068	0.615	0.14
family.Defluviitaleaceae	8	0.92 (0.85–0.99)	0.048	0.615	0.495	0.59
family.Rhodospirillaceae	7	0.89 (0.79–0.99)	0.049	0.624	0.103	0.10

Fig. 2.

Positive Mendelian randomization results for prostate cancer and gut microbiota. A showed the results with ID ieu-b-85, and B shows the results with ID ebi-a-GCST006085

Causal associations between dietary habits and prostate cancer

A total of 87 dietary liking had sufficient IVs available for MR analysis. The data for these instrumental variables are presented in Table S6 and S8, and the results of dietary preferences in relation to prostate cancer are shown in Table S5 (outcome ID: ieu-b-85) and Table S7 (outcome ID: ebi-a-GCST006085). All significant results are summarized in Table 2. Across two repeated analyses, the shared positive results include F-coffee/alcohol (OR = 1.08, P = 0.028), Diet fizzy drinks (OR = 0.83, P = 0.041), Extra virgin olive oil (OR = 1.38, P = 0.029), Lemon (OR = 1.23, P = 0.043), Melon (OR = 1.48, P = 0.022), Orange juice (OR = 1.32, P = 0.007), Potatoes (OR = 2.13, P = 0.028), Salty pretzels (OR = 1.35, P = 0.001), F-sodas (OR = 1.32, P < 0.001), and Tomatoes (OR = 1.31, P = 0.018). All results showed no heterogeneity or horizontal pleiotropy, the positive results are presented in Table 2 and visualized using forest plots in Fig. 3.

Table 2.

Positive results of Mendelian randomized analysis for diet liking and prostate cancer

Outcome: prostate cancer, ID: ieu-b-85			P-value
Exposures	SNP number	OR (95%CI)	IVW outcome	Heterogeneity test	Pleiotropy test	MR presso test
Biscuits	9	1.14 (1.01–1.29)	0.029	0.755	0.637	0.77
Brussel sprout	4	1.18 (1.00–1.38)	0.045	0.203	0.880	0.31
Adding butter to bread	2	1.30 (1.10–1.53)	0.002	0.336	–	–
F-chocolate/coffee	11	1.06 (1.01–1.11)	0.023	0.227	0.397	0.27
F-coffee/alcohol	12	1.08 (1.01–1.17)	0.028	0.048	0.750	0.08
Diet fizzy drinks	4	0.83 (0.70–0.99)	0.041	0.245	0.693	0.35
Extra virgin olive oil	3	1.38 (1.03–1.85)	0.029	0.270	0.444	–
Fizzy drinks	2	1.41 (1.03–1.94)	0.031	0.104	–	–
Lemon	6	1.23 (1.01–1.50)	0.043	0.223	0.676	0.29
Melon	2	1.48 (1.06–2.07)	0.022	0.353	–	–
Orange juice	2	1.32 (1.08–1.61)	0.007	0.549	–	–
Potatoes	2	2.13 (1.08–4.19)	0.028	0.080	–	–
Red meat	4	1.30 (1.11–1.53)	0.001	0.430	0.288	0.46
Salty pretzels	3	1.35 (1.13–1.61)	0.001	0.484	0.918	–
F-sodas	2	1.32 (1.16–1.51)	< 0.001	0.526	–	–
Tomatoes	4	1.31 (1.05–1.64)	0.018	0.792	0.418	0.83

Outcome: prostate cancer, ID: ebi-a-GCST006085			P-value
Exposures	SNP number	OR (95%CI)	IVW outcome	Heterogeneity test	Pleiotropy test	MR presso test
F-coffee/alcohol	13	1.08 (1.00–1.15)	0.030	0.058	0.770	0.09
Diet fizzy drinks	4	0.83 (0.70–0.99)	0.041	0.245	0.693	0.34
Extra virgin olive oil	3	1.38 (1.03–1.85)	0.029	0.270	0.444	–
Honey	2	0.66 (0.52–0.85)	0.001	0.488	–	–
F-acquired taste	35	1.08 (1.02–1.51)	0.012	< 0.001	0.271	< 0.01
Lemon	6	1.23 (1.01–1.50)	0.043	0.223	0.676	0.30
F-lentils/beans	3	0.78 (0.62–0.99)	0.046	0.896	0.929	–
Melon	2	1.48 (1.06–2.07)	0.022	0.353	–	–
Orange juice	2	1.32 (1.08–1.61)	0.007	0.549	–	–
Potatoes	2	2.13 (1.08–4.19)	0.028	0.080	–	–
Salty pretzels	4	1.32 (1.13–1.55)	< 0.001	0.661	0.827	0.70
F-sodas	2	1.22 (1.07–1.39)	0.003	0.516	–	–
Tomatoes	4	1.31 (1.05–1.64)	0.018	0.792	0.418	0.83

Fig. 3.

Positive Mendelian randomization results for dietary habits and prostate cancer. A showed the results with ID ieu-b-85, and B shows the results with ID ebi-a-GCST006085

Causal associations between PCa associated gut microbiome and PCa associated dietary habits

We conducted MR analysis using the intersecting positive results, linking dietary associated with PCa and gut microbiome associated with PCa. The results indicated an inverse causal relationship between Diet fizzy drinks and class Negativicutes (OR = 0.81, P = 0.038) as well as order Selenomonadales (OR = 0.81, P = 0.038). These results were robust, showing no evidence of heterogeneity or horizontal pleiotropy (Table 3). Additional results and instrumental variable data are presented in Table S9 and Table S10.

Table 3.

Positive results of Mendelian randomized analysis for PCa associated gut microbiome and PCa associated dietary liking

Exposures	Outcome	SNP number	OR (95%CI)	P-value
				IVW outcome	Heterogeneity test	Pleiotropy test	MR presso test
Diet fizzy drinks	Class Negativicutes	4	0.81 (0.66–0.99)	0.038	0.715	0.906	0.73
	Order Selenomonadales	4	0.81 (0.66–0.99)	0.038	0.715	0.906	0.73

Two-step MR analysis of Diet fizzy drinks, order Selenomonadales and prostate cancer

Since order Selenomonadales belongs to class Negativicutes, and the MR analysis results for both were identical in relation to Diet fizzy drinks, we retained order Selenomonadales as a potential mediator. We further explored whether the inverse causal relationship between Diet fizzy drinks and PCa could be explained by a reduction in the abundance of order Selenomonadales, which is associated with increased PCa risk. The results indicated a partial mediation effect, but it was not statistically significant. The total effect was OR = 0.83, 95% CI: 0.70–0.99, the direct effect was OR = 0.86, 95% CI: 0.72–1.03, the mediation effect was OR = 0.97, 95% CI: 0.93–1.01, and the proportion of the mediation effect was 16% (95% CI: − 0.06–0.37), P = 0.13 (Table 4).

Table 4.

Two-step MR analysis

Exposure	Mediator	Outcome*	Intermediary effect		Direct effect		Total effect		Mediation effectd Proportion
			OR (95%CI)	Beta, SE	OR (95%CI)	Beta, SE	OR (95%CI)	Beta, SE	Proportion %(95%CI)	P-value
Diet fizzy drinks	Order Selenomonadales	ebi-a-GCST006085	0.97 (0.93–1.01)	− 0.03, 0.02	0.86 (0.72–1.03)	− 0.15, 0.09	0.83 (0.70–0.99)	− 0.18, 0.09	0.16 (− 0.06–0.37)	0.13
	Order Selenomonadales	ieu-b-85	0.97 (0.93–1.01)	− 0.03, 0.02	0.86 (0.72–1.03)	− 0.15, 0.09	0.83 (0.70–0.99)	− 0.18, 0.09	0.16 (− 0.06–0.37)	0.13

*The outcome is prostate cancer with two different ID datasets

Discussion

In this study, we employed advanced MR techniques to investigate the potential causal relationships between diet, GM, and PCa risk. Our findings indicate that the consumption of diet fizzy drinks is associated with a reduced risk of PCa. Additionally, two types of GM—class Negativicutes and order Selenomonadales—were found to significantly increase PCa risk, while their abundance can be suppressed by diet fizzy drinks. Thus, it is plausible to suggest that the intake of diet fizzy drinks may reduce PCa risk by lowering the abundance of specific GM, particularly class Negativicutes and order Selenomonadales.

For the causal relationship between fizzy drink consumption (including F-sodas such as Fanta, Faygo, Fresca, and Fentimans) and the risk of PCa, our result aligns with previous studies on nutrient intake and PCa risk. Matsushita et al. demonstrated that high intake of monosaccharides and disaccharides may contribute to PCa progression via mechanisms involving insulin-like growth factor (IGF-1)–mediated inflammation and activation [11, 19], which promote the states of abundance and cell growth [20, 21].

The use of dietary manipulation in cancer therapy has been explored for over a century [22]. Low-carbohydrate and ketogenic diets are common strategies for inhibiting IGF-1–induced PCa tumorigenesis, as calorie restriction can reduce circulating insulin levels and decrease insulin resistance [23, 24]. Replacing sugary beverages with diet fizzy drinks also constitutes a dietary modification aimed at reducing sugar intake, potentially lowering PCa risk by decreasing circulating insulin and IGF-1 levels. Additionally, daily dietary components profoundly impact the composition of GM [25], which play essential roles in producing arginine and B vitamins, such as folate, biotin, and riboflavin [6]. Elevated levels of urea cycle metabolites—including aspartate, argininosuccinate, arginine, proline, and the oncometabolite fumarate—have been detected in PCa patients [26]. Interestingly, meta-analyses indicated that while dietary and total folate intake do not affect PCa risk, high blood folate levels are associated with an increased risk, suggesting that GM–synthesized folic acid may influence PCa occurrence [27]. Beyond metabolic changes, microorganism-associated molecular patterns (MAMPs) that mediate inflammatory responses are also considered potential mechanisms for tumorigenesis, as GM can drive inflammation through pro-inflammatory cytokines such as IL-17, IL-23, TNF-α, and IFNγ [28–30]. However, the mechanisms by which class Negativicutes and order Selenomonadales increase PCa risk remain unclear, and no studies have yet explored the effect of gut microbiota alterations on PCa risk factors such as androgens, warranting further investigation.

The health impacts of both diet and regular fizzy drinks remain relatively poorly understood. A study focusing on adolescents in Australia highlighted potential knowledge gaps regarding the health effects of diet versus regular sodas, as well as nutritional knowledge relevant to public health interventions [31]. Our study provides genetic evidence supporting the claim that ‘diet drinks are healthier’ and may have implications for beverage choices. However, this does not suggest that diet fizzy drinks can be consumed without restriction. These drinks are still linked to a higher prevalence of metabolic syndrome [32] and diabetes [33], making them unsuitable for long-term, high-quantity consumption.

Strengths and limitations

Our research is the first to identify a causal relationship between diet and PCa risk at the genetic level. Our study has several notable strengths. First, the large sample size and the comprehensive GWAS database we used to enhance the robustness and reliability of our findings. Additionally, our study establishes associations between diet, GM, and PCa, supporting the concept of the ‘gut-prostate cancer axis.’ These findings have potential implications for choices between diet and regular fizzy drinks.

However, several limitations remain. Firstly, while our MR analysis identified associations between diet fizzy drinks, GM, and PCa, GM cannot be confirmed as an ‘intermediary agent’ between diet fizzy drinks and PCa risk. Our results suggest associations rather than definitive causal relationships for two GM types with diet fizzy drinks and PCa. Secondly, we included only European populations in our analysis. Given that microbiota compositions vary among different populations, our findings may not be generalizable across all racial groups. Thirdly, we used a p-value threshold of 5 × 10⁻⁶ to select SNPs for GM due to insufficient SNPs at the standard 5 × 10⁻⁸ threshold in MR analyses. This could introduce potential bias into our results. Also, due to the limitations of the MR design, we could not determine the exact intake level of diet fizzy drinks associated with a reduced risk of PCa. Lastly, many other factors associated with PCa, such as life habits, kidney function, sexual function, may also be related to dietary habits and gut microbiota, though we did not investigate this further [34, 35].

Author contributions

Junhua Guo: Conceptualization, Methodology, Writing—Original Draft Preparation. Ting Huang: Data Curation, Formal Analysis. Heran Zhou: Supervision, Project Administration.

Funding

This research received financial support from the Construction Fund of Medical Key Disciplines of Hangzhou. (No. 2020SJZDXK004).

Data availability

Data available within the article or its supplementary materials.

Declarations

Ethics approval and consent to participate

The present study is based on summary-level data that are publicly available. In all original studies, ethical approval had been obtained.

Competing interests

The authors declare no competing interests.