We identified 14 gut microbiota species significantly associated with the risk of RPC (Fig. 2 and Table S1). The associated 158 SNPs are listed in Table S2.

Mendelian randomization results for the causal effects between the gut microbiota and cancer

Among these species, 10 were positively correlated with increased RPC risk, including Actinobacteria, Bacillaceae A, CAG-390 sp003523225, CAG-452, Faecalicatena sp002397985, Lawsonibacter sp000492175, Succiniclasticum, UBA1777 sp900319835, UBA8621, and Victivallis sp002998355. In contrast, 4 species were negatively correlated with RPC risk, including Collinsella, Dokdonella, Holdemania, and UCG-010 sp003150215. The specific OR, 95% CI, and P-values for these associations are detailed in Fig. 2 and Table S1.

Figure 3A illustrates the causal relationships between several inflammatory factors and RPC. Specifically, Macrophage Colony-Stimulating Factor 1 (CSF-1) (OR = 2.003, P = 0.0072), C-X-C Motif Chemokine 11 (CXCL11) (OR = 1.791, P = 0.172), CXCL5 (OR = 1.381, P = 0.489), CXCL6 (OR = 1.329, P = 0.0338), Interleukin-12 Subunit Beta (IL-12RA) (OR = 1.380, P = 0.0252), Interleukin-1 Alpha (IL-1A) (OR = 1.641, P = 0.491), IL-5 (OR = 2.022, P = 0.469), IL-6 (OR = 2.178, P = 0.329), and Tumor Necrosis Factor Receptor Superfamily Member 9 (TNFRSF9) (OR = 1.795, P = 0.0057) were identified as risk factors for RPC. In contrast, Fibroblast Growth Factor 21 (FGF-21) (OR = 0.593, P = 0.257), Interleukin-18 Receptor 1 (IL-18R1) (OR = 0.784, P = 0.0292), and Stem Cell Factor (SCF) (OR = 0.707, P = 0.322) were protective factors.

Results of Mendelian randomization. A. Inflammatory factors and cancer. B.Blood cells and cancer

Side fluorescence median of RBC 2 in response to TMAO perturbation measured by reticulocyte dye (RET_TMAO_RBC2_Med_SFL), Side fluorescence median of neutrophil 4 at baseline measured by WDF dye (WDF_Baseline_NE4_Med_SFL), Forward scatter standard deviation of eosinophil 2 in response to KCl perturbation measured by WDF dye (WDF_KCl_EO2_SD_FSC), Neutrophil 2/neutrophil 4 ratio in response to KCl perturbation measured by WDF dye (WDF_KCl_NE2_NE4_Med_SFL), Forward scatter standard deviation of eosinophil 2 in response to KCl NE2_NE4_ratio) and Neutrophil 2 percentage in response to water perturbation measured by WDF dye (WDF_Water_NE2_%) were positively correlated with RPC, with ORs of 2.003, 1.791, 1.381, 1.3291, 1.381, 1.381, and 1.391, respectively, 1.381, 1.329 and 1.380, and the P-values were 0.0069, 0.0349, 0.0228, 0.0164 and 0.0092, respectively.On the other hand, Forward scatter coefficient of variation of neutrophil 1 in response to Pam3CSK4 perturbation measured by WDF dye (WDF_Pam3CSK4_NE1_CV_FSC) and Side fluorescence standard deviation of platelet in response to isobutyric acid perturbation measured by WNR dye (WNR_Isobutyric_A._PLT_SD_SFL) were negatively correlated with RPC, with ORs of 0.593 and 0.784, and P-values of 0.0493 and 0.0193, respectively.

A total of 18 species from the gut microbiota were found to be significantly associated with UC (Fig. 2 and Table S1), involving 169 SNPs(Table S2).

Ten species were found to be negatively correlated with UC risk, including An181, Bifidobacterium kashiwanohense, CAG − 274 sp000432155, CAG − 345, CAG − 485 sp002404675, Faecalicatena sp002161355, Lawsonibacter sp900066645, Parabacteroides, Prevotellamassilia sp000437675, and Rhodanobacter. Eight species, including Enterococcus, Lawsonibacter sp000492175, Paenibacillus J, Prevotella sp002437285, Prevotella sp900318625, RUG147 sp900315495, RUG472, and Syntrophorhabdaceae, were positively correlated with UC risk(Fig. 2 and Table S1).

Figure 3A shows that IL-4 (OR = 0.366, P = 0.0242) and SCF (OR = 0.584, P = 0.0066) were protective factors for UC, while TNFRSF9 (OR = 1.660, P = 0.0434) was a risk factor for UC.

Blood cell measures also played a role in UC risk. Specifically, the side fluorescence coefficient of variation of RBC in response to H₂O₂ perturbation (PLT-F_H₂O₂_RBC_CV_SFL) and the side scatter standard deviation of platelets in response to captopril perturbation (WNR_Captopril_PLT_SD_SSC) were negatively correlated with UC (ORs of 0.701 and 0.932, P-values of 0.0141 and 0.0207). On the other hand, the side fluorescence coefficient of variation of monocyte 2 in response to captopril perturbation (WDF_Captopril_MO2_CV_SFL) and the forward scatter coefficient of variation of monocyte in response to Pam3CSK4 perturbation (WDF_Pam3CSK4_MO_CV_FSC) were positively correlated with UC (ORs of 1.261 and 1.502, P-values of 0.0011 and 0.0063) (Fig. 3B).

Eleven gut microbiota species were significantly associated with BC risk (Fig. 2 and Table S1), involving 111 SNPs (Table S2).

Seven species were associated with a decreased BC risk, including CAG − 177 sp002438685, CAG − 194 sp002441865, CAG − 274 sp000432155, Faecalicatena lactaris, Intestinimonas massiliensis, Tepidanaerobacteraceae, and UBA1066. Four species were positively correlated with increased BC risk, including Chloroflexales, Dorea, Ruminococcus C sp000437255, and UBA6960 (Fig. 2 and Table S1).

In terms of inflammatory factors, IL-20 (OR = 1.306, P = 0.0280), IL-24 (OR = 1.310, P = 0.0352), and IL-6 (OR = 1.223, P = 0.0397) were identified as risk factors for BC. In contrast, Leukemia Inhibitory Factor Receptor (LIF-R) was protective (OR = 0.891, P = 0.0431) (Fig. 3A).

Figure 3B demonstrates, side fluorescence standard deviation of neutrophil 1 in response to butyric acid perturbation measured by WDF dye (WDF_Butyric_A._NE1_SD_ SFL), side scatter coefficient of variation of neutrophil 3 in response to cholic acid perturbation measured by WDF dye (WDF_Cholic_A._NE3_CV_SSC), side fluorescence standard deviation of neutrophil 4 in response to LPS perturbation measured by WDF dye (WDF_LPS_NE4_SD_SFL) and forward scatter coefficient of variation of neutrophil 4 in response to Pam3CSK4 perturbation measured by WDF dye (WDF_Pam3CSK4_NE4_CV_FSC) were positively correlated with BC, with ORs of 1.034, 1.042, 1.062 and 1.114, with P-values of 0.0139, 0.0144, 0.0486 and 0.0187, respectively; while neutrophil 2 percentage in response to water perturbation measured by WDF dye (WDF_Water_ NE2%) and forward scatter coefficient of variation of unknown cell population 1 in response to nigericin perturbation measured by WNR dye (WNR_Nigericin_ UK1_CV_FSC), on the other hand, were negatively correlated with BC, with ORs of 0.910 and 0.987, and p-values of 0.0026 and 0.0226, respectively..

Cochran's Q test (Table S4) indicated significant heterogeneity in the causal relationship for BC, specifically for UBA6960 (P = 0.0156), but no significant heterogeneity was observed for other analyses (P > 0.05). MR-Egger intercept analysis revealed horizontal pleiotropy for CAG − 345 in UC (P = 0.0438), but no other SNPs showed horizontal pleiotropy (P > 0.05). "Leave-one-out" analysis suggested that the inclusion of all SNPs significantly influenced the causal conclusions (Fig.S1-3). The scatter plots further illustrated the overall impact of gut microbiota on urothelial carcinoma (Fig.S4-6), while the forest plots provided additional evidence of causal relationships (Fig.S7-9).

Table S5 illustrates a reverse association between RPC and Victivallis sp002998355 (OR = 0.926, P = 0.0223), and between BC and CAG − 177 sp002438685 (OR = 0.918, P = 0.0108). Table S6 reveals a negative correlation between BC and WDF_Cholic_A._NE3_CV_SSC (OR = 0.690, P = 0.0253). No other reverse causal relationships were detected.

After excluding heterogeneity, horizontal pleiotropy, and bidirectional causal effects, we performed mediation analysis to investigate the effects of gut microbiota, inflammatory factors, and blood cells in cancer development.

Figure 4 indicates that Bacillaceae A was positively correlated with IL-6 (OR = 1.926, P = 0.0358), and IL-6 was a risk factor for RPC. Bacillaceae A significantly increased RPC risk (OR = 62.129, P = 0.0244), with an indirect effect via IL-6 contributing to 12.35% of the risk (Table S8). CAG − 452 was negatively correlated with SCF (OR = 0.894, P = 0.0022) and elevated RPC risk (OR = 1.709, P = 0.0144), with a 7.25% mediating effect via SCF. Similarly, CAG − 390 sp003523225 positively correlated with TNFRSF9 (OR = 1.194, P = 0.0244), significantly increasing RPC risk (OR = 1.795, P = 0.0057), with an indirect effect of 8.47% (Table S8).

Mediator Mendelian randomization results between the gut microbiota, inflammatory factors, and cancer

Further analysis (Fig. 5) showed that Succiniclasticum was negatively correlated with WNR_Isobutyric_A._PLT_SD_SFL (OR = 0.144, P = 0.0223) but elevated RPC risk (OR = 2.325, P = 0.0393), with a 34.01% mediating effect. In contrast, UCG − 010 sp003150215 was positively correlated with WNR_Isobutyric_A._PLT_SD_SFL (OR = 7.775, P = 0.0051), reducing RPC risk (OR = 0.370, P = 0.0093), with a mediating effect of 30.51%(Table S8).

Mediator Mendelian randomization results between the gut microbiota, blood cells, and cancer

Figure 4 demonstrates that Syntrophorhabdaceae was negatively correlated with IL-4 (OR = 0.521, P = 0.0016), with IL-4 being a protective factor for UC. Syntrophorhabdaceae significantly increased UC risk (OR = 18.484, P = 0.0129), with an indirect effect of 22.46% via IL-4 (Table S8).

Figure 5 further indicated that Prevotella sp002437285 was negatively correlated with WNR_Captopril_PLT_SD_SSC (OR = 0.011, P = 0.0124) but increased UC risk (OR = 2.840, P = 0.0202), with a 30.58% mediating effect(Table S8).

Although causal relationships were observed between gut microbiota, inflammatory factors, and blood cells in BC (Table S7), no significant mediating effects were detected in this analysis. Thus, inflammatory factors and blood cells do not appear to mediate the causal relationship between gut microbiota and BC.