Genetically predicted serum urate and the risk of all-cause and site-specific cancer: A Mendelian randomization study

doi:10.21203/rs.3.rs-5007746/v1

Background

Positive associations between urate levels and gout and the risk of some cancer types (urogenital, prostate, gastrointestinal and lung) have been reported in several observational studies; however, whether the relationship is causal remains uncertain.

Objectives

The study aim was to evaluate associations between genetically predicted levels of serum urate (SU) and cancer risk (overall and major cancer types) in individuals of European ancestry using Mendelian randomization (MR) analysis.

Methods

A set of 26 SU-related variants was used as proxy instrument to perform a range of one and two-sample MR analyses in individual-level and publicly available GWAS (genome-wide association study data), respectively. The causal relationship was assessed between genetically determined SU and 13 site-specific (bladder, breast, colorectal, gastric, hepatic, lung, pancreatic, prostate, renal, skin, lymphatic and hematopoietic cancers, gynecological cancers, and brain tumor) and all-cause cancer. We also performed epidemiological association analyses in individual-level data to determine a SU-cancer relationship.

Results

There was some suggestive evidence for an association between higher levels of genetically predicted SU and lower risk of brain (p = 0.04 in 1-sample MR) and colorectal (p = 0.02 in 2-sample MR) cancers, although not consistent in the two analyses. There were no indications for associations between genetically predicted SU and any of the other cancers (all p > 0.05).

Conclusions

Our MR study, using a series of causal inference approaches, provides suggestive but inconsistent evidence of an effect of genetically predicted SU on brain and colorectal cancers in individuals of European ancestry.

Urate is the end-product of purine degradation in humans and several genes are known to play a role in controlling its metabolism. The serum urate (SU) level in humans is about ten times higher than in other mammals [1]. This can be attributed to the evolutionary silencing of the gene that encodes for the enzyme ‘uricase’ [2, 3] that is responsible for degrading urate to more soluble products in other mammals [4]. Urate is a vital antioxidant that accounts for about 60% contribution to the respiratory and circulatory antioxidant activity in humans [5–7], which possibly could influence the risk for cancer [8, 9]. Despite the beneficial role, unusually high urate concentrations (hyperuricemia) could also have adverse health effects. Hyperuricemia is often defined based on the distribution in populations by blood urate concentrations of > 430 µmol/L in men and > 360 µmol/L in women [10–13]. A longstanding hyperuricemia can cause urate super-saturation in the blood and its crystallization in joints and soft tissues at serum concentrations above 405 µmol/L at 37°C – a process that can subsequently lead to the development of gout [14]. The prevalence of hyperuricemia among the individuals of the Western world is recorded to be 10 to 20% [11, 15–17]. A more recent data from the Swedish Cardiopulmonary Bioimage Study (SCAPIS), including individuals between 50 and 64 years, demonstrated hyperuricemia to be more common in Swedish men (18%) than in women (2%) [18]. At least 28 genetic loci conferring risk to develop hyperuricemia have been identified to date [19].

Several diseases are associated with gout and hyperuricemia, including renal and cardiovascular diseases [20], type 2 diabetes (T2D) [21] and malignancies [22, 23], although causality remains to be established. Associations between urate levels and the risk of cancer have been reported in a number of observational studies [24], and postulated to be the result of the intracellular oxidative action of urate [25]. While few studies have reported a protective role of higher urate levels against a number of malignancies [8, 9, 26], other reports have contradicted this claim reporting higher urate to be associated with increased risk of cancer [9, 27]. Such inconsistent results, despite the presence of many well-designed prospective studies [23, 28–30], can be attributed to the biases that frequently arised in these conventional observational studies due to unaccounted/unknown confounding. It is therefore not clear if there exists a causal relationship between urate and risk of cancer.

Mendelian randomization (MR) is an analytical technique that takes into account the random assignment of alleles at the time of conception to disentangle the cause-effect relationship. The technique uses genetic variants known to be associated with exposure of interest as a proxy to detect the causal association with the outcome accounting for any confounding and reverse causality [31, 32]. Several MR studies, using genetic variants in the urate transporter genes (SCL2A9 and ABCG2) or combined genetic risk score from a list of urate-related variants as genetic instrument, have demonstrated a causal relationship between urate and multiple health outcomes [33]. These studies provide evidence of both causal (e.g., body mass index [34], T2D [35]) and non-causal (e.g., chronic kidney disease [36] cardiovascular outcomes [37, 38]) relationships between genetically determined urate levels and defined outcomes. Nevertheless, the available data regarding the causal relationship between urate and cancer risk remains limited. This scarcity calls for more comprehensive analyses and an effort to address the contradictory findings prevalent in the current literature.

Here, we designed a study to evaluate a causal effect of genetically predicted SU concentrations on cancer risks, overall and major cancer types, using a range of MR analytic approaches. The study also used observational data along with MR analysis to determine if the genetic data are consistent with the urate-cancer relationship detected in the observational association analysis. To keep the observational and MR association analyses in line, and owing to the availability of the data, the study was conducted only in individuals of European ancestry.

2.1 Study design

Figure 1 illustrates an overview of the study design, which comprised of several components including observational and causal association (or MR) analyses to comprehensively assess the relationship between urate and cancer in European individuals. We first performed an observational association analysis combining the data from two population-based Swedish cohorts from southernmost part of Sweden, the Malmö Diet and Cancer Study (MDCS) [39] and the Malmö Preventive Project (MPP) [40]. The combined dataset is referred to as MDCS-MPP cohort in this study.

To ensure the robustness of causal associations, we implemented a range of, both primary and sensitivity, unidirectional MR analyses to determine the causal effect of urate on the risk of cancer. For this purpose, both one-sample and two-sample MR settings were employed using the individual-level data from MDCS-MPP and summary-statistics data from the Global Urate Genetic Consortium (GUGC) [19] and UK-Biobank cohort [41] respectively.

2.2 Study participants and data sources

The observational association between urate and cancer risk was investigated using the data of 17,597 individuals in MDCS-MPP cohort (Table S1). Both MDCS and MPP are population-based cohorts of 30,447 and 33,346 European (mostly born in Sweden) participants, respectively. At baseline, individuals in MDCS underwent a health examination from 1991 to 1996, while those in MPP had a health examination from 1974 to 1992. Individuals from both of these cohorts included in the study (n = 17,597) were middle-aged (mean age = 46.5 years, Table S2). The follow-up data descriptions and other details for MDCS and MPP have been published elsewhere [40, 42]. For the purpose of this study, data for SU measurements (mean SU = 4.96 mg/dL) (Table S2) from the health surveys was used, and cancer endpoints (mean follow-up = 21.2 years) (Table S1) were obtained using the national cancer register in Sweden, which has been found to have a high validity [43-45]. In total, data for 13 common site-specific cancers [bladder, breast, colorectal, gastric, hepatic, lung, pancreatic, prostate, renal, skin, lymphatic and hematopoietic cancers, gynecological (included ovarian, cervical and uterine cancers) cancers and brain tumor] were used, and the variable all-cause cancer was created by combining the cases from all 13 cancer types (n = 5,659) (Table S1). Cancer diagnosis was defined using International Classification of Diseases or ICD-9 and 10 codes (Table S3). To increase comparability, all cancer cases were excluded from the control group (n = 11,938).

For one-sample MR analysis, the data for SU measurement and that for 14 cancer outcomes, were obtained from the MDCS-MPP cohort for all 17,597 participants. For two-sample MR, we retrieved publicly available GWAS summary statistics data for variant-urate association for 110,347 European individuals from GUGC [19]. The variant-cancer outcome association data for 13 site-specific and all-cause cancer were obtained from 367,570 European individuals (case = 36,815, control = 330,755) in the UK-Biobank. The details of the genotyping information in MDCS-MPP and UK-Biobank are provided in supplementary material (Text S1). The definitions for several cancer outcomes were similar but not identical between the MDCS-MPP and the UK-Biobank cohorts using ICD-9 and 10 codes (Table S3).

2.3 Statistical analyses

2.3.1 Instrument selection

We used a set of 26 variants (single nucleotide polymorphisms or SNPs), associated with SU at a genome-wide significance (p < 5 x 10^-08) in the GUGC consortium [19], as the exposure instrument to run several MR analyses in this study (Table S4).

2.3.2 Observational association analysis

For observational association analysis, multivariable cox-proportional hazard regression models, adjusted for age and sex (except for breast, prostate, and gynecological cancers, where the adjustment was done only for age), were used to evaluate the association of SU, separately, with the risk of each cancer type.

2.3.3 One-sample MR

We tested the association of all 26 SNPs with SU in the individual-level data from the MDCS-MPP cohort, using the linear regression models in controls only (Table S6 and S7). The F-statistic was also calculated in this regression analysis to assess the strength of the SU instrument. An F-statistic of > 10 is regarded as having a strong potential to predict causality without weak instrument bias [46]. SNPs to cancer endpoint association estimates were obtained using logistic regression models (Table S8 to S21). All analyses were adjusted for age and sex, except for breast, prostate, and gynecological cancers, where the adjustment was done only for age.

To indicate the robustness of causal associations, a range of MR analyses were performed to run one-sample MR using the summary-statistics from the above variant-urate and variant-cancer association analysis (wherever applicable) in the MDCS-MPP. These analyses primarily implemented MR using two-stage least square (2SLS) and inverse variance weighted (IVW) methods. Cochran’s Q test was used to assess the heterogeneity across the estimates for 26 SNPs. We further applied MR analysis in MDCS-MPP data using weighted genetic risk score (GRS), where the GRS were calculated based on the number of risk alleles for all SNPs and their effect sizes on the SU levels in the MDCS-MPP cohort.

2.3.4 Two-sample MR

To run two-sample MR, the summary statistics data for SNP/variant-urate associations was obtained from the GUGC consortium (Table S4) while for variant-cancer endpoints the data was obtained from the UK-Biobank (Table S22 to S35) for all 26 SNPs. The individual estimates for each SNP were generated using the Wald ratio estimator, while the standard error (SE) was calculated using the Delta method [47, 48]. The estimator provides the ratios for SNP-outcome estimate over SNP-exposure estimate. The individual estimates were then pooled for each cancer type in the IVW ratio method as the primary MR analysis using the random-effect model. All effects were adjusted for age and sex (wherever applicable) before running the MR analysis. Cochran’s Q test was used to assess the heterogeneity across the estimates for 26 SNPs. As IVW method is vulnerable to the horizontal pleiotropy through the confounding pathways independent of the exposure (SU in this case), a range of sensitivity analyses were carried out. Each of these sensitivity analyses assess the causal effect based on different assumptions that are designed to be less stringent to address the possible pleiotropy. The methods applied are weighted median and MR-Egger methods. The detailed description for these methods is provided in the supplementary material (Text S2). In addition, the Mendelian Randomization Pleiotropy Residual Sum and Outlier (MR-PRESSO) method was performed [49]. The MR-PRESSO removes the potential outliers that are determined by the square of residual errors from the SNP-outcome against SNP-exposure regression to calculate an outlier-free effect estimate. This test is more sensitive than Egger and has the power to detect any outlier SNP in the MR analysis that could introduce biasness in the results. We also performed leave-one-out analyses to detect the significant effect that an individual SNP could render upon the MR estimates. Power calculation were done using mRnD power calculator [50].

To deal with any possible horizontal pleiotropy, an additional sensitivity analysis was performed. The 26 SNPs used as urate instrument were examined for their association with the traits other than SU levels using PhenoScanner (http://www.phenoscanner.medschl.cam.ac.uk/). The PhenoScanner is an online platform that provides public access to a vast range of GWAS summary results. We identified seven SNPs (Table S5) that were exclusively associated with SU levels (and or gout) and repeated all MR analyses using seven SNPs as urate instrument.

We applied Bonferroni correction to account for multiple testing for the 14 cancer outcomes. The associations with p-values < 3.5E-03 (where p = 0.05/14) were deemed as strong evidence for causal associations, while associations with p-values < 0.05 but > 3.5E-03 were arbitrated as suggestive evidence of associations. All analyses were done using R software (v4.3.2) and Mendelian Randomization [51] and MR-PRESSO software [49] packages in R.

2.4. Ethics approval and consent to participate

The Regional Ethics Committee at Lund University (Dnr LU 90-51, 85/2004 and 2009/633) provided ethical approval for this study in MDCS-MPP cohort. The UK Biobank has approval from the North West Multi-centre Research Ethics Committee (MREC) as a Research Tissue Bank (RTB) approval, renewed in 2021 (REC reference: 21/NW/0157).

3.1 Observational association analysis in MDCS-MPP cohort

All hazard ratios (HRs) are presented as a change in the cancer risk per 1 SD increase in SU concentration (1 SD = 1.15 mgdL^-1). We found a significant association between SU and an increased risk of colorectal cancer [HR (95%CI); 1.18 (1.09 to 1.28), p = 5.9E-05]. The urate was also associated with an increased risk of prostate cancer [HR (95%CI); 1.08 (1.02 to 1.15), p = 0.005] and gynecological cancers [HR (95%CI); 1.15 (1.01 to 1.32), p = 0.03]. While SU indicated susceptible association with most of the cancer risks (except gastric and lung cancer where the relationship was inverse), the results were not significant statistically. However, SU was associated with an increased risk of all-cause cancer in both unadjusted and adjusted analyses [HR (95%CI); 1.06 (1.03 to 1.09), p = 1.18E-04]. The complete results for observational association analysis are provided in Table 1.

3.2 One-sample MR analysis in MDCS-MPP cohort

All 26 SNPs indicated an accumulative variance of 5.9% (R² = 0.059) in SU concentrations in the MDCS-MPP cohort, which is comparable to 5.8% (R² = 0.058) variance explained by these 26 SNPs in the GUGC cohort [19]. The urate instrument also indicated an F-statistics of 31.38, suggesting that the instrument was unlikely to introduce marked weak instrument bias in the results. All odds ratios (ORs) in MR analyses are presented as a change in the cancer risk per 1 mgdL^-1 increase in SU concentration. There was a consistent protective causal effect of urate on the risk of brain cancer in MDCS-MPP individuals in both 2SLS [OR (95%CI); 0.48 (0.24 to 0.98), p = 0.05] (Figure 2A) and GRS [OR (95%CI); 0.48 (0.24 to 0.98), p = 0.04] MR analyses, while the IVW estimates were not significant and indicated heterogeneity [OR (95%CI); 0.48 (0.17 to 1.41), p = 0.18, p_het = 0.04] in 26-SNP MR analysis (Table 2).

Restricting the urate instrument criteria to 7-SNPs showed a suggestive protective effect of urate on brain cancer [2SLS; OR (95%CI); 0.36 (0.15 to 0.82), p = 0.01 (Figure 2A) and GRS; OR (95%CI); 0.38 (0.17 to 0.86), p = 0.02, IVW; OR (95%CI); 0.36 (0.15 to 0.87), p = 0.02] with non-significant heterogeneity (p_het = 0.07) (Table 2). SU also indicated a consistent trend towards a significant causal effect on an increased risk of gynecological cancers in restricted (7-SNPs) analyses. Complete results for one-sample MR are presented in Figure 2A and Table 2.

3.3 Two-sample MR analysis in publicly available GWAS

All 26 SNPs had F-statistic values larger than 10 (with lowest being 30.1) (Table S4), that indicated that the chance of introducing weak instrument bias in two-sample MR analyses was unlikely. The individual MR estimates for each SNP are illustrated in Figure S1 while the detailed results for two-sample MR are presented in Figure 2B and Table 3, where all ORs are presented as a change in the cancer risk per 1 mgdL^-1 increase in SU concentration. The main MR analysis revealed a suggested causal effect of higher urate on the decreased risk of colorectal cancer with high heterogeneity [OR (95%CI); 0.82 (0.69 to 0.98), p = 0.02, p_het < 0.0001] (Figure 2B). Restricting the instrument criteria to 7-SNP set-up still indicated a significant effect of urate for a decreased risk of colorectal cancer, with no heterogeneity [OR (95%CI); 0.89 (0.79 to 1.00), p = 0.05, p_het = 0.74] (Figure 2B). The outlier removal via MR-PRESSO still resulted in a suggestive effect [OR (95%CI); 0.82 (0.68 to 0.97), p = 0.04] (Table 3). The only other cancer for which urate indicated a trend of association was lung cancer with significant heterogeneity [OR (95%CI); 0.86 (0.73 to 1.02), p = 0.07, p_het = 0.04] (Figure 2B). There was no association between urate and any other cancer risk and results were consistent throughout the range of MR and sensitivity analyses that were implied in two-sample setting. None of the leave-one-out analyses showed a particular SNP to be an outlier and having a significant effect on MR estimates (Figure S2). Results from power calculations for both one- and two-sample MR are detailed in Table S36.

In this population-based study, we explored a potential causal role of genetically predicted urate levels for the development of 13 site-specific (bladder, breast, colorectal, gastric, hepatic, lung, pancreatic, prostate, renal, skin, lymphatic and hematopoietic cancers, gynecological cancers and brain tumor) and all-cause cancers using MR analysis. To account for possible pleiotropy, several complementary and sensitivity MR analyses designs were applied. The study also included observational association analysis in the MDCS-MPP cohort to provide the urate-cancer risk associations without controlling for other confounders apart from age and sex (where applicable). The results from the observational association analysis indicated an association between higher SU levels and increased risk of colorectal, prostate and all-cause cancers. Despite these and other conventional studies showing strong association between urate and cancer risk [52], the MR analysis in our study did not provide strong evidence to support a causal role of urate levels in cancer incidence. While there was evidence supporting a potential causal effect of higher urate on the decreased risk of colorectal cancer and brain tumor, these findings were only rendered as being suggestive as the significance diminished after correction for multiple testing. To the best of our knowledge, there are only a few MR studies that investigated the urate to cancer causal relationship till date. One of the studies that was conducted in Copenhagen General Population Study demonstrated higher plasma urate to be causally associated with cancer incidence (OR = 1.22) and all-cause mortality (OR = 1.49) using only rs7442295 in SLC2A9 gene as an instrument for urate [53], while one showed causal association of higher urate with higher risk of prostate cancer in Japanese population using 34-SNP urate instrument [54]. Contrary to these, other studies did not report any causality of urate levels on neither colorectal [55] nor eight cancer outcomes [56] using 26-SNP urate instrument and several publicly available cancer datasets. Similarly, no evidence of a causal relationship between urate and lung cancer was reported using UK-Biobank data in one-sample MR setting [57]. None of these studies, however, offered details regarding the selection of cancer patients for one cancer type as controls for another cancer type in their analysis. Additionally, they did not adjust for potential confounding covariates while calculating variant-urate and variant-cancer risk associations for several MR analyses.

Several reports suggest that urate might potentially offer protection against cancer in humans due to its remarkable antioxidant properties [4, 5, 25, 58]. However, persistently higher levels of urate have also been associated with inflammation and development of gout [7, 59], and thereby set a question mark for its protective effect against cancer. In a study of middle-aged men in Sweden (Malmö), cancer mortality was positively associated with SU levels only in the first 2.5 years after urate screening [60]. Another study that included 493,281 adults (followed-up for 25 years) from the Swedish Apolipoprotein-related MOrtality RISk study (AMORIS) identified a correlation between elevated serum uric acid levels and an increased risk of colorectal, hepatobiliary, kidney, and non-melanoma skin cancers in men, and head and neck cancers in women [61]. The same study also observed an inverse relationship between serum uric acid levels and the risk of pulmonary cancers in men, along with lymphatic, hematological, and breast cancers in women. Moreover, a positive association between urate levels and prostate cancer incidence was observed in Japanese men in Hawaii, but it attenuated over time and became non-significant >10 years after the urate measurements [30]. A recent prospective population-based study in EPIC-Heidelberg cohort reported higher levels of SU to be associated with lower risk of breast cancer. The study also reported no evidence of any association between urate levels and risk of other common cancers i.e., prostate, colorectal and lung cancers [62]. The findings from these studies support the paradoxical role of serum urate as both an anti- and prooxidant. These findings also align with the observational association analysis results in our study, where we observed both protective and susceptible relationship between higher SU levels and risk of a few cancer types, while for others the analyses showed no association at all even after adjustments for age and sex as primary confounders (Table 1).

We did observe suggestive, but inconsistent, evidence for a causal role of SU and risk of a few cancer types in several MR settings in our study. In MDCS-MPP cohort, using one-sample MR approach, higher SU indicated an association with reduced risk of brain cancer in the Swedish population, especially when seven variants exclusively associated with SU were considered as exposure instrument. Having a lower antioxidant capacity, brain is considered to be particularly vulnerable to oxidative stress [63, 64], and urate has been suggested to play a role as a possible neuroprotector [65, 66]. Wang et al. (2015) demonstrated a significant delay in tumor growth after the administration of uric acid as an adjuvant to a dendritic cell-based vaccine [67]. Urate is also known to be associated with decreased risks of developing Alzheimer’s and Parkinson’s disease [68-70], however, there are no previous MR studies analyzing the causal relation with brain cancer. In addition, a meta-analysis of three prospective cohort studies have not been able to demonstrate any association between brain cancer risk and gout [71].

Our study also demonstrated suggestive evidence for a protective role of SU against colorectal cancer in publicly available larger cohorts (GUGC and the UK-Biobank in two-sample MR settings), albeit with high heterogeneity. However, the association sustained when corrected for an outlier as well as when variants exclusively associated with urate were used as urate-instrument in more stringent analysis with no significant heterogeneity. We, however, did not observe a causal effect of urate in changing the risk of colorectal cancer in Swedish population, and hence, the association was deemed inconsistent. The previous MR analyses were not able to report a causal effect of urate on colorectal cancer using the GUGC,UK-Biobank and other publicly available data in several European cohorts [55, 56]. One possible explanation for these divergent results could be our exclusion of patients developing any cancer types from the control group that was used for all comparisons and calculating variant-urate associations in MR analyses. We also accounted for adjustments for age and sex as potential confounders for urate while calculating urate-cancer associations in the UK-Biobank cohort. In their study, Jiang et al. (2021) [56] also reported one variant, rs12498742, to be significantly affecting the MR estimates in leave-one-out analysis, which is line with our results (Figure S2). However, the association sustained in our analysis when corrected for this and other outlier variants via MR-PRESSO (OR = 0.82, p = 0.04, Table 3). It is important to take in consideration here that rs12498742 (in SLC2A9 gene) is so far the strongest urate-associated variant reported in European GWAS [19]. Therefore, it would be undesirable to not select this variant as a proxy to reflect serum urate levels in humans in any MR analysis. In addition, we did find a similar protective association of urate for colorectal cancer risk when using variants strictly associated with urate and gout as urate instrument only in two-sample MR (OR = 0.89, p = 0.04, Table 3). Published data in most observational studies have reported a possible association of high urate levels and increased risk of colorectal cancer [72-74]. Our observational analysis also indicating an association between higher urate levels and an increased risk of developing colorectal cancer (OR = 1.18, p = 5.9E-05, after adjustment for age and sex), which is in line with these studies but could include major effects of residual confounding, a problem which is largely avoided using genetically predicted urate levels as the exposure. In contrast, a study that reported 10-year prevalence of colorectal cancer in gout verses osteoarthritis patients, using the computerized patient record system of the VA New York Harbor Health Care System, reported that individuals with gout had significantly lower (0.8%) prevalence of colorectal cancer than those with osteoarthritis (3.7%) (p = 0.0008) [75]. Furthermore, Mi et al. (2022) reported a U-shaped relationship between serum uric acid and the risks of colon and rectal cancers i.e., they found higher risks for both cancer types in the lowest (≤3.5 mgdL^-1) and highest (>8.4 mgdL^-1) serum uric acid groups compared to the reference group (>3.5-5.4 mgdL^-1) [74]. This suggests that the association between SU and colorectal cancer is more complex than a simple linear relationship, and given the known common genetic variants explain only a minor proportion of the variance (23.9%) in SU levels [76], the inconsistent MR estimates through a range of MR analyses in our study merits further exploration.

The study has several strengths. First, it investigates the observational association between SU levels and cancer risk using a comprehensive list of cancer outcomes in observational analysis in the Swedish (MDCS-MPP) cohort. Second, it applied one-sample and two-sample MR analyses in three independent European cohorts to avoid overlapping, hence improving the reliability of the results. Third, the study prevented competing risk via excluding the patients for any cancer type from the control group altogether. Fourth, we adjusted for common population covariates, age and sex, in all MR analyses to account for their possible association with urate levels. To the best of our knowledge, the study presents the largest MR-analyses performed so far in individuals of any ancestry to investigate SU to cancer risk causality with most of cancer outcomes included in the analysis. Moreover, the study used a range of robust MR-analyses methods to account for possible violations of the MR assumptions, such as horizontal pleiotropy.

The first limitation of this study is lower statistical power for several MR analyses, where a consistent lower case to control proportions for several cancer types may have largely contributed. While the data in two-sample MR setting indicated to have enough power (82%) to detect a causal effect of urate on colorectal cancer risk, it is hard to immaculately claim causal effect of urate on risk of other cancer types, without having a lack of accuracy, as the analyses consistently suffered from lower statistical power. This is, unfortunately, an inherent limitation and cannot be addressed unless cases to control proportion is increased. In addition, the GWAS data for SU and cancer outcomes were obtained from heterogenous populations and it is possible that population differences for potentially important features, other than age and sex that we adjusted for, may have contributed to masking the causal effect of urate on cancer. Though it would be premature to draw definitive conclusions, such heterogeneity could potentially have contributed to the somewhat different results in the MR analyses between one-sample and two-sample MR settings. Moreover, our investigation (observational and MR analyses) was confined to the European population, thereby limiting the generalizability of these findings to other population settings.

This study did not find statistically significant strong evidence to support a causal role of genetically predicted SU levels for cancer risk. The study, however, does report suggestive, but inconsistent, evidence for causality between higher urate levels and lower risk of colorectal and brain cancer. The inconsistence in MR estimates, particularly where suggestive evidence was found in one or the other MR setting, warrants further investigation of potential causal role of SU levels in relation to cancer risk in larger cohorts as well as in other population settings. Moreover, MR investigation(s) in data particularly well stratified for SU levels for SNP-urate and SNP-cancer endpoint associations may be helpful in delineating a potential non-linear relationship between SU and cancer risk.

SE=serum urate, MR= Mendelian randomization, SCAPIS= the Swedish Cardiopulmonary Bioimage Study; T2D= type 2 diabetes, GWAS= genome-wide association study; , MDCS= the Malmö Diet and Cancer Study, MPP= the Malmö Preventive Project, GUGC=the Global Urate Genetic Consortium, UK-Biobank cohort= the United Kingdom Biobank cohort, ICD=International Classification of Diseases Codes; SNP=single nucleotide polymorphisms, 2SLS= two-stage least square, IVW= inverse variance weighted, SE=the standard error, GSR=genetic risk score, HR= Hazard Ratio, OR= odds ratio, MR-PRESO= the Mendelian Randomization Pleiotropy Residual Sum and Outlier, mRnD=Mendelian randomization power calculator, AMORIS=the Swedish Apolipoprotein-related MOrtality RISk study

Acknowledgements

We thank the participants of the Malmö Preventive Project (MPP), Malmö Diet and Cancer Study (MDCS) and the UK Biobank.

Funding information

MPP was supported by the City of Malmö, the Swedish Medical Research Council, and The Swedish Heart and Lung Foundation. MDCS received financial support from the Swedish Cancer Foundation, the Swedish Medical Research Council, the European Commission, the City of Malmö, the Swedish Dairy Association and the Albert Pâhlsson Foundation.

Contributions

TF, LTHJ and MCK design the study, oversaw its execution, interpreted the data/results, and wrote the manuscript; TF performed the formal analyses, interpreted the data and wrote the first draft of the manuscript; SB and AMM provided the UK-Biobank data, contributed and helped in the analyses and interpreted the data; MD, PMN, and OM helped to interpret the results. All authors contributed equally to revise the manuscript for important intellectual content and read and approved the final version of the manuscript.

Data availability statement

The data for MDCS and MKC cohorts can be obtained via reasonable request, while the data used for two-sample MR analysis are publicly available.

Conflict of interest statement

TF, MD, PMN, OM, LTHJ and MCK report no conflict of interest. SB is supported by the Wellcome Trust (225790/Z/22/Z) and the United Kingdom Research and Innovation Medical Research Council (MC_UU_00002/7, MC_UU_00040/01). AMM is supported by core funding from the: British Heart Foundation (RG/18/13/33946), NIHR Cambridge Biomedical Research Centre (BRC-1215-20014; NIHR203312) [*], Cambridge BHF Centre of Research Excellence (RE/18/1/34212), BHF Chair Award (CH/12/2/29428) and by Health Data Research UK, which is funded by the UK Medical Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Department of Health and Social Care (England), Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Health and Social Care Research and Development Division (Welsh Government), Public Health Agency (Northern Ireland), British Heart Foundation and Wellcome. *The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.

Mandal AK, Mount DB. The molecular physiology of uric acid homeostasis. Annu Rev Physiol. 2015;77:323–45.
Oda M, Satta Y, Takenaka O, Takahata N. Loss of urate oxidase activity in hominoids and its evolutionary implications. Mol Biol Evol. 2002;19:640–53.
Kratzer JT, Lanaspa MA, Murphy MN et al. Evolutionary history and metabolic insights of ancient mammalian uricases. Proceedings of the National Academy of Sciences. 2014;111:3763-8.
Briggs J, Levitt M, Abramson R. Renal excretion of allantoin in rats: a micropuncture and clearance study. Am J Physiology-Renal Physiol. 1977;233:F373–81.
Nieto FJ, Iribarren C, Gross MD, Comstock GW, Cutler RG. Uric acid and serum antioxidant capacity: a reaction to atherosclerosis? Atherosclerosis. 2000;148:131–9.
Carocho M, Ferreira IC. A review on antioxidants, prooxidants and related controversy: Natural and synthetic compounds, screening and analysis methodologies and future perspectives. Food Chem Toxicol. 2013;51:15–25.
Kang D-H, Ha S-K. Uric acid puzzle: dual role as anti-oxidantand pro-oxidant. Electrolytes Blood Pressure: E BP. 2014;12:1.
Manifar S, Rahimzamani A, Shirkhoda M, Ghamsari MN, Bakhshi M. Role of serum uric acid as a protective biomarker in patients with different histopathological grades of oral squamous cell carcinoma: a case-control study. BioMed Research International. 2020;2020.
Allegrini S, Garcia-Gil M, Pesi R, Camici M, Tozzi MG. The Good, the Bad and the New about Uric Acid in Cancer. Cancers. 2022;14:4959.
Smith E, March L, editors. Global prevalence of hyperuricemia: a systematic review of population-based epidemiological studies. Arthritis & Rheumatology. NJ USA: WILEY-BLACKWELL 111 RIVER ST, HOBOKEN 07030 – 5774; 2015.
Mikuls TR, Farrar J, Bilker W, Fernandes S, Schumacher H, Saag K. Gout epidemiology: results from the UK general practice research database, 1990–1999. Ann Rheum Dis. 2005;64:267–72.
Liu R, Han C, Wu D et al. Prevalence of hyperuricemia and gout in mainland China from 2000 to 2014: a systematic review and meta-analysis. BioMed research international. 2015;2015.
Lin K-C, Lin H-Y, Chou P. Community based epidemiological study on hyperuricemia and gout in Kin-Hu, Kinmen. J Rhuematol. 2000;27:1045–50.
Towiwat P, Chhana A, Dalbeth N. The anatomical pathology of gout: a systematic literature review. BMC Musculoskelet Disord. 2019;20:1–14.
Campion EW, Glynn RJ, Delabry LO. Asymptomatic hyperuricemia. Risks and consequences in the Normative Aging Study. Am J Med. 1987;82:421–6.
Zhu Y, Pandya BJ, Choi HK. Prevalence of gout and hyperuricemia in the US general population: the National Health and Nutrition Examination Survey 2007–2008. Arthr Rhuem. 2011;63:3136–41.
Annemans L, Spaepen E, Gaskin M, et al. Gout in the UK and Germany: prevalence, comorbidities and management in general practice 2000–2005. Ann Rheum Dis. 2008;67:960–6.
Drivelegka P, Forsblad-D’elia H, Bergström G, Fagman E, Jacobsson L, Dehlin M. POS0188 HYPERURICEMIA IS ASSOCIATED WITH CORONARY ARTERY ATHEROSCLEROSIS IN MEN: RESULTS FROM THE SCAPIS STUDY. BMJ Publishing Group Ltd; 2023.
Köttgen A, Albrecht E, Teumer A, et al. Genome-wide association analyses identify 18 new loci associated with serum urate concentrations. Nat Genet. 2013;45:145–54.
Culleton BF, Larson MG, Kannel WB, Levy D. Serum uric acid and risk for cardiovascular disease and death: the Framingham Heart Study. Ann Intern Med. 1999;131:7–13.
Bhole V, Choi JWJ, Kim SW, De Vera M, Choi H. Serum uric acid levels and the risk of type 2 diabetes: a prospective study. Am J Med. 2010;123:957–61.
Dovell F, Boffetta P. Serum uric acid and cancer mortality and incidence: a systematic review and meta-analysis. Eur J Cancer Prev. 2018;27:399–405.
Lai S-W, Liu C-S, Hwang B-F. Risk of Cancer in Middle-aged Patients With Gout: A Nationwide Population-based Study in Korea. J Rhuematol. 2021.
Fini MA, Elias A, Johnson RJ, Wright RM. Contribution of uric acid to cancer risk, recurrence, and mortality. Clin translational Med. 2012;1:1–15.
Ames BN, Cathcart R, Schwiers E, Hochstein P. Uric acid provides an antioxidant defense in humans against oxidant-and radical-caused aging and cancer: a hypothesis. Proceedings of the National Academy of Sciences. 1981;78:6858-62.
Peden DB, Hohman R, Brown ME et al. Uric acid is a major antioxidant in human nasal airway secretions. Proceedings of the National Academy of Sciences. 1990;87:7638-42.
Strasak A, Rapp K, Hilbe W, et al. The role of serum uric acid as an antioxidant protecting against cancer: prospective study in more than 28 000 older Austrian women. Ann Oncol. 2007;18:1893–7.
Levine W, Dyer AR, Shekelle RB, Schoenberger JA, Stamler J. Serum uric acid and 11.5-year mortality of middle-aged women: findings of the Chicago Heart Association Detection Project in Industry. J Clin Epidemiol. 1989;42:257–67.
Hiatt RA, Fireman BH. Serum uric acid unrelated to cancer incidence in humans. Cancer Res. 1988;48:2916–8.
Kolonel LN, Yoshizawa C, Nomura A, Stemmermann GN. Relationship of serum uric acid to cancer occurrence in a prospective male cohort. Cancer Epidemiol biomarkers prevention: publication Am Association Cancer Res cosponsored Am Soc Prev Oncol. 1994;3:225–8.
Lawlor DA, Harbord RM, Sterne JA, Timpson N, Davey Smith G. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med. 2008;27:1133–63.
Davey Smith G, Hemani G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet. 2014;23:R89–98.
Li X, Meng X, Timofeeva M et al. Serum uric acid levels and multiple health outcomes: umbrella review of evidence from observational studies, randomised controlled trials, and Mendelian randomisation studies. BMJ. 2017;357.
Palmer TM, Nordestgaard BG, Benn M et al. Association of plasma uric acid with ischaemic heart disease and blood pressure: mendelian randomisation analysis of two large cohorts. BMJ. 2013;347.
Sluijs I, Holmes MV, Van Der Schouw YT, et al. A Mendelian randomization study of circulating uric acid and type 2 diabetes. Diabetes. 2015;64:3028–36.
Jordan DM, Choi HK, Verbanck M, et al. No causal effects of serum urate levels on the risk of chronic kidney disease: a Mendelian randomization study. PLoS Med. 2019;16:e1002725.
Kleber ME, Delgado G, Grammer TB, et al. Uric acid and cardiovascular events: a Mendelian randomization study. J Am Soc Nephrology: JASN. 2015;26:2831.
Keenan T, Zhao W, Rasheed A, et al. Causal assessment of serum urate levels in cardiometabolic diseases through a Mendelian randomization study. J Am Coll Cardiol. 2016;67:407–16.
Manjer J, Carlsson S, Elmståhl1 S et al. The Malmö Diet and Cancer Study: representativity, cancer incidence and mortality in participants and non-participants. Eur J Cancer Prev. 2001489–99.
Berglund G, Nilsson P, Eriksson KF, et al. Long-term outcome of the Malmö Preventive Project: mortality and cardiovascular morbidity. J Intern Med. 2000;247:19–29.
Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12:e1001779.
Trell E. Community-based preventive medical department for individual risk factor assessment and intervention in an urban population. Prev Med. 1983;12:397–402.
Barlow L, Westergren K, Holmberg L, Talbäck M. The completeness of the Swedish Cancer Register–a sample survey for year 1998. Acta Oncol. 2009;48:27–33.
Linder G, Lindblad M, Djerf P, et al. Validation of data quality in the Swedish National Register for Oesophageal and Gastric Cancer. J Br Surg. 2016;103:1326–35.
Löfgren L, Eloranta S, Krawiec K, et al. Validation of data quality in the Swedish National Register for Breast Cancer. BMC Public Health. 2019;19:1–11.
Palmer TM, Lawlor DA, Harbord RM, et al. Using multiple genetic variants as instrumental variables for modifiable risk factors. Stat Methods Med Res. 2012;21:223–42.
Didelez V, Meng S, Sheehan NA. Assumptions of IV methods for observational epidemiology. 2010.
Thompson JR, Minelli C, Fabiola Del Greco M. Mendelian randomization using public data from genetic consortia. Int J biostatistics. 2016;12.
Verbanck M, Chen C-Y, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50:693–8.
Brion M-JA, Shakhbazov K, Visscher PM. Calculating statistical power in Mendelian randomization studies. Int J Epidemiol. 2013;42:1497–501.
Yavorska OO, Burgess S. MendelianRandomization: an R package for performing Mendelian randomization analyses using summarized data. Int J Epidemiol. 2017;46:1734–9.
Mi S, Gong L, Sui Z. Friend or foe? An unrecognized role of uric acid in cancer development and the potential anticancer effects of uric acid-lowering drugs. J Cancer. 2020;11:5236.
Kobylecki CJ, Afzal S, Nordestgaard BG. Plasma urate, cancer incidence, and all-cause mortality: a Mendelian randomization study. Clin Chem. 2017;63:1151–60.
Deng Y, Huang J, Wong MCS. Association between serum uric acid and prostate cancer risk in East Asian populations: a Mendelian randomization study. Eur J Nutr. 2023;62:1323–9.
Zhou J, Fu R, Zhang J et al. Association between serum uric acid and colorectal cancer risk in European population: a two-sample Mendelian randomization study. Front Oncol. 202414:1394320.
Jiang M, Ren L, Chen S, Li G. Serum uric acid levels and risk of eight site-specific cancers: a mendelian randomization study. Front Genet. 2021;12:608311.
Horsfall LJ, Hall IP, Nazareth I. Serum urate and lung cancer: a cohort study and Mendelian randomization using UK Biobank. Respir Res. 2021;22:1–11.
Glantzounis G, Tsimoyiannis E, Kappas A, Galaris D. Uric acid and oxidative stress. Curr Pharm Design. 2005;11:4145–51.
Fang P, Li X, Luo JJ, Wang H, Yang X. -f. A double-edged sword: uric acid and neurological disorders. Brain disorders therapy. 2013;2:109.
Petersson B, Trell E. Raised serum urate concentration as risk factor for premature mortality in middle aged men: relation to death from cancer. Br Med J (Clin Res Ed). 1983;287:7–9.
Yiu A, Van Hemelrijck M, Garmo H, et al. Circulating uric acid levels and subsequent development of cancer in 493,281 individuals: findings from the AMORIS Study. Oncotarget. 2017;8:42332.
Kühn T, Sookthai D, Graf ME, et al. Albumin, bilirubin, uric acid and cancer risk: results from a prospective population-based study. Br J Cancer. 2017;117:1572–9.
Salim S. Oxidative stress and the central nervous system. J Pharmacol Exp Ther. 2017;360:201–5.
Cobley JN, Fiorello ML, Bailey DM. 13 reasons why the brain is susceptible to oxidative stress. Redox Biol. 2018;15:490–503.
Proctor PH. Uric acid and neuroprotection. Stroke. 2008;39:e126–e.
Mijailovic NR, Vesic K, Borovcanin MM. The Influence of Serum Uric Acid on the Brain and Cognitive Dysfunction. Front Psychiatry. 2022;13.
Wang Y, Ma X, Su C, et al. Uric acid enhances the antitumor immunity of dendritic cell-based vaccine. Sci Rep. 2015;5:16427.
Andreadou E, Nikolaou C, Gournaras F, et al. Serum uric acid levels in patients with Parkinson's disease: their relationship to treatment and disease duration. Clin Neurol Neurosurg. 2009;111:724–8.
Auinger P, Kieburtz K, Mcdermott MP. The relationship between uric acid levels and Huntington's disease progression. Mov Disord. 2010;25:224–8.
Shi X, Zheng J, Ma J, et al. Low serum uric acid levels are associated with the nonmotor symptoms and brain gray matter volume in Parkinson’s disease. Neurol Sci. 2022;43:1747–54.
Wang W, Xu D, Wang B et al. Increased risk of cancer in relation to gout: a review of three prospective cohort studies with 50,358 subjects. Mediators of Inflammation. 2015;2015.
Yan J, Zhu C. Hyperuricemia is a adverse prognostic factor for colon cancer patients. Int J Gen Med. 2021;14:3001.
Li W, Liu T, Siyin ST, et al. The relationship between serum uric acid and colorectal cancer: a prospective cohort study. Sci Rep. 2022;12:1–9.
Mi N, Huang J, Huang C, et al. High serum uric acid may associate with the increased risk of colorectal cancer in females: A prospective cohort study. Int J Cancer. 2022;150:263–72.
Slobodnick A, Krasnokutsky S, Lehmann RA, et al. Colorectal Cancer among gout patients undergoing colonoscopy. JCR: J Clin Rheumatol. 2019;25:335–40.
Major TJ, Topless RK, Dalbeth N, Merriman TR. Evaluation of the diet wide contribution to serum urate levels: meta-analysis of population based cohorts. BMJ. 2018;363.

Table 1 Urate association with site specific and all-cause cancers in MDCS-MPP cohort

Cancer type	Unadjusted		Adjusted*
Cancer type	HR (95%CI)	p-value	HR (95%CI)	p-value
Bladder	0.97 (0.884 ; 1.080)	0.65	1.08 (0.978 ; 1.206)	0.12
Colorectal	0.98 (0.918 ; 1.059)	0.71	1.18 (1.091 ; 1.288)	5.9E-05
Brain	0.90 (0.747 ; 1.086)	0.27	1.16 (0.934 ; 1.439)	0.17
Gastric	0.82 (0.691 ; 0.991)	0.04	0.92 (0.760 ; 1.107)	0.36
Hepatic	1.04 (0.768 ; 1.401)	0.81	1.01 (0.765 ; 1.318)	0.97
Renal	0.89 (0.778 ; 1.034)	0.13	1.01 (0.857 ; 1.185)	0.92
Lung	0.85 (0.775 ; 0.932)	5.6E-04	0.94 (0.858 ; 1.035)	0.21
Lymphatic and hematopoietic	0.90 (0.822 ; 0.995)	0.04	1.02 (0.919 ; 1.131)	0.71
Pancreatic	0.97 (0.805 ; 1.191)	0.83	1.09 (0.888 ; 1.344)	0.40
Skin	0.85 (0.771 ; 0.954)	0.005	1.03 (0.911 ; 1.173)	0.60
Breast	1.13 ( 1.032 ; 1.244)	0.008	1.02 (0.929 ; 1.130)	0.62
Gynecological	1.16 (1.022 ; 1.324)	0.02	1.15 (1.014 ; 1.317)	0.03
Prostate	1.06 (1.002 ; 1.124)	0.04	1.08 (1.024 ; 1.148)	0.005
All-cause	0.86 (0.835 ; 0.883)	< 2E-16	1.06 (1.030 ; 1.094)	1.2E-04

All hazard ratios (HRs) are presented as a change in the cancer risk per 1 SD increase in SU concentration (1 SD = 1.15 mgdL^-1).

SD: standard daviation, MDCS-MPP: Malmö Diet Cancer and Malmö Preventive Project

*Values are adjsuted for age and sex, except for for breast, prostate, and gynecological cancers where values are adjusted only for age.

Table 2 Results of one-sample MR (using MDCS-MPP cohort) for causal relationship between serum urate levels and cancer risk

Cancer type	MR instrument	Cochran's Q test		GRS		IVW
Cancer type	MR instrument	Q	p-Het	OR (95% CI)	p-value	OR (95% CI)	p-value
Bladder	26 SNPs	23.47	0.55	0.94 (0.671; 1.342)	0.76	1.00 (0.707 ; 1.420)	0.80
Bladder	7 SNPs	2.71	0.84	0.82 (0.555 ; 1.234)	0.35	0.85 (0.567 ; 1.294)	0.46
Colorectal	26 SNPs	25.35	0.44	0.97 (0.732 ; 1.287)	0.83	0.96 (0.724 ; 1.282)	0.80
Colorectal	7 SNPs	2.99	0.81	0.98 (0.712 ; 1.372)	0.94	0.97 (0.696 ; 1.372)	0.89
Brain	26 SNPs	38.75	0.04	0.48 (0.238 ; 0.983)	0.04	0.48 (0.169 ; 1.412)	0.18
Brain	7 SNPs	11.52	0.07	0.38 (0.170 ; 0.867)	0.02	0.36 (0.153 ; 0.869)	0.02
Gastric	26 SNPs	26.62	0.37	1.05 (0.586 ; 1.904)	0.86	1.08 (0.590 ; 1.999)	0.78
Gastric	7 SNPs	6.07	0.41	0.89 (0.455 ; 1.774)	0.74	0.92 (0.445 ; 1.908)	0.83
Hepatic	26 SNPs	29.43	0.25	2.57 (0.823 ; 8.264)	0.11	2.67 (0.810 ; 8.823)	0.11
Hepatic	7 SNPs	7.52	0.27	2.32 (0.621 ; 9.095)	0.22	2.83 (0.674 ; 11.90)	0.15
Renal	26 SNPs	15.92	0.92	0.82 (0.461 ; 1.466)	0.5	0.85 (0.482 ; 1.522)	0.59
Renal	7 SNPs	0.99	0.98	0.80 (0.412 ; 1.576)	0.52	0.83 (0.420 ; 1.632)	0.58
Lung	26 SNPs	16.19	0.91	1.26 (0.900 ; 1.778)	0.18	1.29 (0.926 ; 1.818)	0.13
Lung	7 SNPs	2.24	0.89	1.16 (0.788 ; 1.168)	0.43	1.17 (0.790 ; 1.752)	0.42
Lymphatic and hematopoietic	26 SNPs	19.98	0.74	1.18 (0.036 ; 1.182)	0.37	1.20 (0.827 ; 1.748)	0.33
Lymphatic and hematopoietic	7 SNPs	3.99	0.67	0.99 (0.651 ; 1.531)	0.98	0.99 (0.641 ; 1.559)	0.99
Pancreatic	26 SNPs	33.08	0.13	1.04 (0.511 ; 2.133)	0.91	1.15 (0.576 ; 2.314)	0.68
Pancreatic	7 SNPs	10.62	0.10	1.22 (0.538 ; 2.853)	0.63	1.45 (0.641 ; 3.296)	0.37
Skin	26 SNPs	22.01	0.63	1.00 (0.674 ; 1.491)	0.99	0.99 (0.672 ; 1.483)	0.99
Skin	7 SNPs	6.96	0.32	0.85 (0.542 ; 1.355)	0.5	0.84 (0.529 ; 1.347)	0.48
Breast	26 SNPs	33.58	0.12	0.98 (0.739 ; 1.323)	0.94	0.99 (0.737 ; 1.333)	0.95
Breast	7 SNPs	2.24	0.89	0.89 (0.641 ; 1.260)	0.53	0.89 (0.632 ; 1.276)	0.55
Gynecological	26 SNPs	25.48	0.43	1.18 (0.893 ; 1.583)	0.23	1.22 (0.911 ; 1.632)	0.18
Gynecological	7 SNPs	1.43	0.96	1.33 (0.953 ; 1.866)	0.09	1.36 (0.967 ; 1.935)	0.07
Prostate	26 SNPs	17.78	0.85	1.04 (0.838 ; 1.302)	0.69	1.06 (0.854 ; 1.325)	0.57
Prostate	7 SNPs	2.58	0.85	1.02 (0.799 ; 1.328)	0.82	1.05 (0.811 ; 1.364)	0.69
All-cause	26 SNPs	22.51	0.61	1.07 (0.943 ; 1.213)	0.29	1.07 (0.945 ; 1.215)	0.27
All-cause	7 SNPs	1.29	0.97	1.04 (0.902 ; 1.207)	0.56	1.04 (0.895 ; 1.206)	0.61

All odds ratios (ORs) are presented as a change in the cancer risk per 1 mgdL^-1 increase in SU concentration.

MR: Mendelian randomization, MDCS-MPP: Malmö Diet Cancer and Malmö Preventive Project, 95% CI: 95% confidence interval, 2SLS: two-stage least square, GRS: genetic risk score, IVW: Inverse variance weighted method, Q; Q value from Cochran’s Q-test where higher value indicates higher heterogeneity, p; p-value for the given analysis.

Table 3 Results from two-sample MR (using GUGC and the UK Biobank cohort) for causal relationship between serum urate levels and cancer risk

Cancer type	MR instrument	Cochran's Q test		Weighted median		MR-Egger		Egger-intercept		MR-PRESSO
		Cochran's Q test		Weighted median		MR-Egger		Egger-intercept		Raw		No. of outilers	Outlier corrected
		Q	p	OR (95% CI)	p	OR (95% CI)	p	Est.	p	OR (95% CI)	p	No. of outilers	OR (95% CI)	p
Bladder	26 SNPs	20.32	0.72	1.03 (0.864 ; 1.227)	0.73	0.97 (0.796 ; 1.190)	0.79	0.011	0.24	1.06 (0.921 ; 1.221)	0.37	0	NA	NA
Bladder	7 SNPs	4.57	0.59	1.02 (0.863 ; 1.225)	0.75	1.03 (0.824 ; 1.286)	0.79	0.002	0.89	1.04 (0.884 ; 1.225)	0.6	0	NA	NA
Colorectal	26 SNPs	75.77	< 0.0001	0.89 (0.785 ; 1.008)	0.06	0.92 (0.722 ; 1.178)	0.51	-0.014	0.20	0.82 (0.691 ; 0.980)	0.03	4	0.82 (0.689 ; 0.979)	0.04
Colorectal	7 SNPs	3.49	0.74	0.89 (0.788 ; 1.012)	0.07	0.89 (0.761 ; 1.042)	0.15	0.001	0.96	0.89 (0.796 ; 1.002)	0.04	0	NA	NA
Brain	26 SNPs	27.21	0.34	1.08 (0.769 ; 1.535)	0.63	1.08 (0.724 ; 1.635)	0.68	-0.016	0.38	0.95 (0.720 ; 1.271)	0.76	0	NA	NA
Brain	7 SNPs	8.41	0.21	1.10 (0.784 ; 1.544)	0.57	1.23 (0.740 ; 2.056)	0.42	-0.039	0.32	1.03 (0.712 ; 1.502)	0.86	0	NA	NA
Gastric	26 SNPs	25.93	0.41	1.18 (0.930 ; 1.517)	0.16	1.19 (0.905 ; 1.582)	0.21	-0.012	0.35	1.09 (0.898 ; 1.323)	0.39	0	NA	NA
Gastric	7 SNPs	5.17	0.52	1.19 (0.935 ; 1.528)	0.15	1.26 (0.931 ; 1.721)	0.13	-0.004	0.86	1.24 (0.996 ; 1.551)	0.08	0	NA	NA
Hepatic	26 SNPs	33.02	0.13	0.99 (0.683 ; 1.453)	0.98	1.22 (0.744 ; 1.999)	0.43	-0.02	0.37	1.04 (0.737 ; 1.411)	0.82	0	NA	NA
Hepatic	7 SNPs	6.76	0.34	1.12 (0.770 ; 1.637)	0.54	1.03 (0.607 ; 1.778)	0.88	0.022	0.59	1.14 (0.794 ; 1.658)	0.49	0	NA	NA
Renal	26 SNPs	18.09	0.83	0.90 (0.711 ; 1.145)	0.4	0.88 (0.674 ; 1.153)	0.36	0.004	0.75	0.91 (0.755 ; 1.091)	0.25	0	NA	NA
Renal	7 SNPs	2.64	0.85	0.89 (0.703 ; 1.134)	0.35	0.82 (0.611 ; 1.108)	0.19	0.017	0.44	0.89 (0.716 ; 1.106)	0.16	0	NA	NA
Lung	26 SNPs	38.15	0.04	0.86 (0.736 ; 1.024)	0.09	0.84 (0.666 ; 1.065)	0.15	0.003	0.76	0.86 (0.735 ; 1.016)	0.08	0	NA	NA
Lung	7 SNPs	4.67	0.58	0.87 (0.741 ; 1.032)	0.11	0.79 (0.642 ; 0.972)	0.02	0.026	0.1	0.88 (0.764 ; 1.034)	0.13	0	NA	NA
Lymphatic and hematopoietic	26 SNPs	54.28	0.0006	0.96 (0.886 ; 1.050)	0.41	0.93 (0.809 ; 1.078)	0.35	0.007	0.28	0.98 (0.893 ; 1.093)	0.82	1	0.99	0.99
Lymphatic and hematopoietic	7 SNPs	4.97	0.54	0.96 (0.886 ; 1.051)	0.41	0.95 (0.852 ; 1.059)	0.36	0.006	0.48	0.97 (0.901 ; 1.056)	0.53	0	NA	NA
Pancreatic	26 SNPs	32.44	0.14	1.07 (0.808 ; 1.436)	0.61	1.07 (0.732 ; 1.565)	0.72	0.005	0.76	1.11 (0.859 ; 1.449)	0.41	0	NA	NA
Pancreatic	7 SNPs	2.69	0.84	1.07 (0.810 ; 1.433)	0.61	1.04 (0.729 ; 1.506)	0.79	0.005	0.86	1.07 (0.821 ; 1.395)	0.48	0	NA	NA
Skin	26 SNPs	39.76	0.03	0.93 (0.783 ; 1.111)	0.44	0.91 (0.714 ; 1.175)	0.49	0.013	0.26	1.01 (0.850 ; 1.209)	0.87	0	NA	NA
Skin	7 SNPs	5.63	0.46	0.92 (0.781 ; 1.104)	0.4	0.91 (0.724 ; 1.146)	0.42	0.008	0.65	0.94 (0.804 ; 1.109)	0.5	0	NA	NA
Breast	26 SNPs	57.24	0.0002	1.00 (0.916 ; 1.089)	0.99	0.94 (0.817 ; 1.096)	0.46	0.004	0.54	0.97 (0.883 ; 1.082)	0.67	1	0.98 (0.884 ; 1.083)	0.81
Breast	7 SNPs	11.94	0.06	0.99 (0.903 ; 1.076)	0.76	0.93 (0.802 ; 1.093)	0.41	0.01	0.42	0.97 (0.866 ; 1.091)	0.7	0	NA	NA
Gynecological	26 SNPs	23.24	0.56	0.94 (0.804 ; 1.114)	0.5	0.92 (0.767 ; 1.111)	0.4	0.003	0.75	0.94 (0.828 ; 1.073)	0.37	0	NA	NA
Gynecological	7 SNPs	3.13	0.79	0.94 (0.800 ; 1.108)	0.47	0.91 (0.740 ; 1.115)	0.36	0.009	0.57	0.94 (0.814 ; 1.098)	0.35	0	NA	NA
Prostate	26 SNPs	20.51	0.71	0.98 (0.884 ; 1.096)	0.77	0.94 (0.837 ; 1.068)	0.37	0.007	0.19	1.00 (0.919 ; 1.090)	0.96	0	NA	NA
Prostate	7 SNPs	4.16	0.65	0.97(0.873 ; 1.081)	0.6	0.96 (0.840 ; 1.099)	0.56	0.004	0.68	0.97 (0.877 ; 1.081)	0.64	0	NA	NA
All-cause	26 SNPs	62.4	< 0.0001	0.97 (0.926 ; 1.017)	0.21	0.93 (0.861 ; 1.016)	0.11	0.004	0.35	0.96 (0.908 ; 1.019)	0.19	1	0.96 (0.910 ; 1.019)	0.21
All-cause	7 SNPs	9.85	0.13	0.97 (0.924 ; 1.018)	0.22	0.93 (0.866 ; 1.003)	0.06	0.006	0.26	0.95 (0.890 ; 1.013)	0.18	0	NA	NA

All odds ratios (ORs) are presented as a change in the cancer risk per 1 mgdL^-1 increase in SU concentration.

MR: Mendelian randomization, GUGC: Global Urate Genetic Consortium, MR-PRESSO; Mendelian Randomization Pleiotropy Residual Sum and Outlier, Est.; Estimate for Egger intercept, 95% CI: 95% confidence interval, Q; Q value from Cochran’s Q-test where higher value indicates higher heterogeneity, p; p-value for the given analysis.

Competing interest reported. TF, MD, PMN, OM, LTHJ and MCK report no conflict of interest. SB is supported by the Wellcome Trust (225790/Z/22/Z) and the United Kingdom Research and Innovation Medical Research Council (MC_UU_00002/7, MC_UU_00040/01). AMM is supported by core funding from the: British Heart Foundation (RG/18/13/33946), NIHR Cambridge Biomedical Research Centre (BRC-1215-20014; NIHR203312) [*], Cambridge BHF Centre of Research Excellence (RE/18/1/34212), BHF Chair Award (CH/12/2/29428) and by Health Data Research UK, which is funded by the UK Medical Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Department of Health and Social Care (England), Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Health and Social Care Research and Development Division (Welsh Government), Public Health Agency (Northern Ireland), British Heart Foundation and Wellcome. *The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.

FatimaetalARTSuppl.docx

Genetically predicted serum urate and the risk of all-cause and site-specific cancer: A Mendelian randomization study

Status:

Version 1

Abstract

Background

Objectives

Methods

Results

Conclusions

Figures

1 INTRODUCTION

2 METHODS

3 RESULTS

4 DISCUSSION

Conclusions

Abbreviations

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1