Two-sample Mendelian randomization study reveals no causal relationship between inflammatory bowel disease and urological cancers

Background: The relationship between inflammatory bowel disease (IBD) and urological cancers has been identified in epidemiological and observational studies, while the causality remains uncertain. We examined whether IBD is causally associated with urological cancers in a Mendelian randomization (MR) study. Methods: The causal relationship between IBD, its main subtypes, and urological cancers was investigated using genome-wide association study data. To obtain more reliable conclusions, all outcomes were divided into training and validation sets. Eligible single-nucleotide polymorphisms were selected as instrumental variables based on MR analysis assumptions. The inverse variance-weighted (IVW) method was employed as the main method along with four other complementary methods. Results: In this two-sample MR study, no genetic evidence for the causal effect of IBD on urological cancers was found in either the training or validation sets using the IVW method. Similarly, we did not observe any significant association between Crohn’s disease or ulcerative colitis and urological cancers. The results of the other methods are in accordance with those obtained using the IVW method. Conclusion: In this study, we confirmed that IBD is not a causal genetic risk factor for urological cancer in a European population.


Introduction
Inflammatory bowel disease (IBD), which includes Crohn's disease (CD) and ulcerative colitis (UC), is a chronic inflammatory disease of the intestine that accumulates mainly in the ileum, rectum, and colon (Kappelman et al., 2007).Epidemiological studies have shown that IBD has become a global condition, and its incidence has increased, particularly in newly industrialized areas (Taleban et al., 2016;Ng et al., 2017;Mak et al., 2020).IBD is an autoimmune disease of unknown etiology characterized by chronic gastrointestinal inflammation, with an increased risk of intestinal cancer in a subset of patients (Canavan et al., 2006;Lutgens et al., 2013).Previous metaanalyses and population-based studies have established that IBD can be accompanied by extraintestinal cancers (Pedersen et al., 2010;Katsanos et al., 2011;Parisian et al., 2013;Lo et al., 2021), which are assumed to be the result of an underlying inflammatory state and immunosuppressive treatment (Axelrad et al., 2016).
Despite limited data, recent studies have shown that patients with IBD are at an increased risk of urological cancer.Population-based cohort studies have suggested that IBD, especially CD, increases the risk of kidney cancer (So et al., 2017;Feng et al., 2021a).The risk of bladder cancer is only mildly elevated in patients with CD (So et al., 2017).However, some researchers have identified potential key genes involved in CD with bladder cancer (Zheng et al., 2023).IBD is a risk factor for prostate cancer (Burns et al., 2019;Ge et al., 2020;Meyers et al., 2020), and preneoplastic changes have been observed in a mouse model of chemically induced intestinal inflammation (Desai et al., 2022).While evidence suggests a link between IBD and urological cancers, the mechanism underlying this relationship remains unclear.In addition, the association between IBD and urological cancers identified in observational studies may be biased due to confounding factors.Previous studies have not illustrated the existence of a causal relationship between IBD and urological cancers; hence, our Mendelian randomization (MR) study was conducted from a genetic perspective.
MR, an epidemiological method of analysis, employs singlenucleotide polymorphisms (SNPs) as instrumental variables (IVs) to investigate the causal relationship between exposures and outcomes, with the advantage of reducing bias due to confounding factors and reverse causality (Sekula et al., 2016;Emdin et al., 2017;Davies et al., 2018).In this two-sample MR study, we attempted to examine whether IBD is causally associated with urological cancers.

Study design description
Figure 1 shows our two-sample MR design comparing IBD and its main subtypes (CD and UC) with urological cancers.We split the data into training and validation sets using summary statistics from a genome-wide association study (GWAS) to investigate the causal relationship between IBD and urological cancer.Fourteen MR analyses were performed, considering IBD and its main subtypes as exposures and urological cancers as outcomes.All the participants included in this study were European.No ethical approval was required, as the data for our study were sourced from publicly available summary statistics.

Filtering of instrumental variables for MR analyses
The following steps were used to select instrumental variables to ensure the accuracy and authenticity of the study: first, SNPs significantly associated with urological cancers and IBD were selected at a genome-wide significance threshold (p < 5 × 10 -8 ).Second, we conducted linkage disequilibrium (LD) analyses (r 2 = 0.001 and with 10000 kb window) among the included instrumental variables.Third, SNPs associated with the outcome (p < 5 × 10 -8 ) were excluded.The strength of the instrument was evaluated using the formula F = (n-k-1/k) × R 2 × (1-R 2 ), where n denotes the sample size, k denotes the number of selected instrumental variables, and R 2 is the exposure variance explained by the included SNPs.F < 10 indicates a weak association between instrumental variables and exposures (Burgess et al., 2011).

Selection of data sources and instrumental variables for IBD
We considered IBD, including overall IBD, CD, and UC, as exposure.Statistics on IBD obtained from European cases were downloaded from the International IBD Genetics Consortium (IIBDGC) (Liu et al., 2015), with 12,882 cases, 21,770 controls, and 12,716,084 SNPs in IBD, 5,956 cases, 14,927 controls, and 12,276,506 SNPs in CD, and 6,968 cases, 20,464 controls, and 12,255,197 SNPs in UC (Supplementary Material; Table 1).

Selection of data sources and instrumental variables for urological cancers
We selected kidney, bladder, and prostate cancers as outcomes, and each outcome was divided into training and validation sets.Two validation sets were used for prostate cancer.All related statistics of the training sets are available in the UK Biobank (https://www.ukbiobank.ac.uk/), in which SNPs associated with kidney cancer were obtained from 701 European cases and 360,493 controls, SNPs associated with bladder cancer were obtained from 1,554 cases and 359,640 controls, and SNPs associated with prostate cancer were obtained from 7,872 cases and 160,876 controls.The datasets for the validation sets were obtained from FinnGen (Kurki et al., 2023), with 1,830 cases and 259,583 controls in kidney cancer, 2,380 cases and 259,583 controls in bladder cancer, and 79,148 cases and 61,106 controls in prostate cancer.The data for the other validation set in prostate cancer were obtained from the Prostate Cancer Association Group to Investigate Cancer-Associated Alterations in the Genome (PRACTICAL) consortium (11,590 cases and 110,189 controls) (Schumacher et al., 2018).Detailed data are shown in Table 1.

Statistical analysis
Five different methods were used, namely, inverse varianceweighted (IVW), MR-Egger, weighted median, simple mode, and weighted mode.IVW was the main analysis used for assessing whether there was a causal effect of IBD on urological cancers.IVW can evaluate the causal effects of genetically predicted exposure on outcomes using a weighted regression of SNPspecific Wald ratios when no horizontal pleiotropy exists.We performed MR-Egger regression analysis for assessing whether there was potential horizontal pleiotropy in the included SNPs.Potential outliers reflecting likely pleiotropic biases were detected by performing Mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO), and any outlying SNP was excluded to correct for horizontal pleiotropic effects.Cochran's Q statistic was used to examine the heterogeneity among the selected SNPs.We also applied a leave-one-out sensitivity analysis to identify whether there was a potentially strong influence of SNPs.The results of the MR analyses are expressed as odds ratios (ORs) and 95% confidence intervals (CIs) to estimate the degree of the causal effect.All analyses were conducted using the TwoSampleMR package (version 0.5.6) in R software (version 4.2.3, https://www.r-project.org/).

Selected genetic instrumental variables
We screened IVs according to the criteria described above and selected 63, 52, and 39 independent SNPs as IVs for IBD, CD, and UC, respectively.The F-statistics revealed no weak instrumental bias between the screened IVs and exposure (all F > 10).Detailed information on the selected SNPs and F analyses is presented in Supplementary Tables S1-S3.The IVs of exposure used in 14 MR analyses (IBD and urological cancers) shared the same filtered SNPs.

Causal effect of IBD on kidney cancer
IVW results showed that overall IBD, CD, and UC were not causally related to kidney cancer in either the training set (overall IBD: OR = 1.000002, 95%CI = 0.9998-1.0002,p = 0.986; CD: OR = 1.00007, 95%CI = 0.9999-1.0002,p = 0.468; and UC: OR = 1.0001, 95%CI = 0.9999-1.0004,p = 0.343) or the validation set (overall IBD: OR = 0.987, 95%CI = 0.918-1.061,p = 0.714; CD: OR = 0.982, 95% CI = 0.927-1.041,p = 0.546; and UC: OR = 1.018, 95%CI = 0.932-1.111,p = 0.699).The overall effect of the training and validation sets also indicated no significant association between IBD and kidney cancer, as shown in the forest plot drawn based on the IVW method (Figure 2).We also drew a forest plot based on the MR-Egger method, and the results were consistent (Supplementary Figure S1).Scatter and forest plots derived from the IVW method are shown in Supplementary Figures S2, S3.No significant associations were observed in the other four models as well.Detailed information on MR analyses is presented in Supplementary Tables S4-S6.Therefore, we believe that IBD and its main subtypes are not causal risk factors for kidney cancer in European populations.

Sensitivity analysis
Pleiotropy and heterogeneity analyses were performed to validate the reliability of MR analyses (Table 2).No incidence of potential pleiotropy was identified in the MR-Egger regression analyses or MR-PRESSO tests.Outlier SNPs were excluded using the MR-PRESSO method.Additionally, we applied Cochran's Q p-value (Table 2) and funnel plots (Supplementary Figure S4) to IVW and MR-Egger methods, and heterogeneity was found in some analyses; such heterogeneity was acceptable in the MR study.The results of the leave-one-sensitivity analyses demonstrated that the estimates of the causal effects of the genetic prediction of IBD on urological cancers were reliable (Supplementary Figure S5).

Discussion
We appraised the genetic overlap and potential causal relationships between IBD and urological cancers using GWAS summary statistics; however, no genetic evidence for the causal effect of IBD on urological cancers was found in this two-sample MR study.The results of the validation set were consistent.We confirmed that IBD was not a causal genetic risk factor for urological cancers in the European population.
Although we did not obtain significant results, several hypotheses have been proposed regarding the associations observed in previous observational and epidemiological studies.First, the relationship between IBD and the increased risk of urological cancers remains controversial.Given scarce studies (Feng et al., 2021a;Feng et al., 2021b), speculations regarding the association between IBD and kidney cancer await confirmation.In the present meta-analysis, patients with IBD showed a trend toward an increased risk of bladder cancer in the CD subgroup (Geng and Geng, 2021).However, the association between IBD and bladder cancer was not observed to be statistically significant in another study (Feng et al., 2021b).IBD was not associated with an increased risk of prostate cancer in this population-based retrospective cohort study (Na et al., 2022).In another metaanalysis, the increased risk of prostate cancer compared with that in the general population was only little (Carli et al., 2020).Second, IBD and its main subtypes have been demonstrated to increase the risk of extraintestinal malignancies, assumed to be a result of an underlying inflammatory state and immunosuppressive therapies (Axelrad et al., 2016).Inflammation is assumed to play a role in the well-established association between IBD and the development of various solid tumors.Chronic kidney disease (CKD) is a risk factor for kidney cancer (Stengel, 2010).The potential mechanism connecting CKD and IBD involves immunoglobulin A (IgA) nephropathy, which is believed to be related to mucosal inflammation (Ambruzs et al., 2014).Patients with IBD probably have an increased risk of kidney cancer through the IBD-CKD pathway.Inflammation may lead to the activation of angiogenesis, which ultimately leads to bladder cancer progression in patients with IBD (Wigner et al., 2021).Chronic inflammation can cause DNA damage and promote carcinogenic epigenetic alterations, which may contribute to prostate tumorigenesis (Sfanos et al., 2018).In addition, a range of drugs used for the treatment of IBD, such as immunomodulators and biological agents, may increase the risk of extraintestinal cancer (Mason and Siegel, 2013).A population-based cohort study also concluded that patients with IBD who were chronically exposed to immunosuppressive treatment may be at a higher risk of being diagnosed with cancer (Van Den Heuvel et al., 2016).Patients with IBD receiving thiopurines are more likely to develop urinary tract cancers, including kidney and bladder cancer (Bourrier et al., 2016).Previous observational studies may not eliminate bias in the risk of cancer due to the use of immunosuppressive medications in patients with IBD.However, no direct evidence has been provided to accurately elucidate the correlation between immunosuppressive drugs and the risk of prostate cancer in patients with IBD.Moreover, patients with chronic inflammatory diseases commonly have more frequent healthcare utilization, which could be explained by a detection bias (Beckmann et al., 2019).Due to the nature of the disease, men with IBD have a higher prevalence of rectal examinations.Cancer can be detected during routine medical tests, with an increasing number of physician office visits over time.
The application of comprehensive GWAS data for MR analyses improves the precision of the estimated effects.The most significant advantage of this study is that causal estimates were obtained from MR, avoiding reverse causality and confounding bias.To the best of our knowledge, this is the first MR analysis on this subject.However, our MR study had some limitations.First, the mechanisms of cancers caused by IBD are complex, and this study only offers genetic evidence for the absence of a causal relationship between IBDs and urological cancers.Therefore, they cannot provide detailed temporal information on disease development.Integrating MR results with long-term clinical observations and patient data may contribute to a more comprehensive understanding.Second, some heterogeneity could not be avoided using Cochran's Q p-value in the previous MR analyses.In addition, MR studies are limited to specific races; thus, conclusions drawn from the analyses of populations of different races are invalid.Therefore, our results are limited to the European population.Finally, the MR method relies on certain assumptions, and there may be instances of improper IV selection or failure that account for potential confounding factors.
In conclusion, our MR study demonstrated that IBD has no causal effect on urological cancers in either the training or validation sets in European populations.Considering that patients with IBD are at an increased risk of developing cancer, further studies are needed to explore the relationship between IBD and cancer.

FIGURE 1
FIGURE 1Flowchart of the Mendelian randomization analysis of Crohn's disease and ulcerative colitis and the risk of development of kidney cancer, bladder cancer, and prostate cancer.IIBDGC, International Inflammatory Bowel Disease Genetics Consortium; IBD, inflammatory bowel disease; PRACTICAL, Prostate Cancer Association Group to Investigate Cancer-Associated Alterations in the Genome; LD, linkage disequilibrium; and SNPs, single-nucleotide polymorphisms.

FIGURE 2
FIGURE 2Forest plot: causal associations of inflammatory bowel disease and its main subtypes with urological cancers using the IVW method.(A) Effect of inflammatory bowel disease on urological cancers.(B) Effect of Crohn's disease on urological cancers.(C) Effect of ulcerative colitis on urological cancers.PRACTICAL, Prostate Cancer Association Group to Investigate Cancer-Associated Alterations in the Genome; IVW, inverse variance-weighted; OR, odds ratio; and CI, confidence interval.

TABLE 1
Characteristics of GWAS on the exposures and outcomes.
PRACTICAL, Prostate Cancer Association Group to Investigate Cancer-Associated Alterations in the Genome; SNPs, single-nucleotide polymorphisms.