POS0295\u2005NO CAUSAL EFFECTS OF GENETICALLY DETERMINED SERUM URATE LEVELS ON THE RISK OF ALL-CAUSE AND SITE-SPECIFIC CANCER: A MENDELIAN RANDOMIZATION STUDY

Abstract

Positive associations between urate levels and gout and the risk of some cancer types (urogenital, prostate, gastrointestinal and lung) have been reported in a number of observational studies; however, whether the relationship is causal remains uncertain.The study aim was to evaluate a causal effect of genetically determined serum urate (SU) concentrations on cancer risks (overall and major cancer types) in individuals with European ancestry using Mendelian randomization (MR) analyses design.We used the individual-level data from two population-based Swedish cohorts including middle-aged subjects (mean follow-up = 21.2 years), Malmö Diet Cancer and Malmö Preventive Project (MDC/MPP), for one-sample MR setting. Data from a total of 17,597 individuals (n = 17,597 for SU at baseline, diagnoses during follow-up: 5659 for all-cause, 516 for bladder, 545 for lung, 791 for bowel, 1521 for prostate and 729 for breast cancer) was included. For two-sample MR, summary-statistic data for SU was obtained from Global Urate Genetic Consortium (GUGC: n = 110,347), while UK-Biobank data was employed for several major cancer outcomes (n = 36,815 for all-cause, 2,245 for bladder, 2,590 for lung, 4,488 for bowel, 6,474 for prostate and 10,274 for breast cancer). The definitions for cancer endpoints were matched for ICD9 and 10 codes between MDC/MPP and UK-Biobank cohorts. For both MR settings, a set of 26 urate-associated single nucleotide variants was selected to build-up the SU instrument (SU-instr) to test for a causal effect of SU on cancer outcomes. Statistical analysis, adjusted for age and sex, was done using multiple conventional MR methodologies and MR package in R (v4.0.2). A p < 0.05 was designated as statistically significant.We found no causal effect of our SU-instr on neither all-cause nor site-specific cancer across all MR analyses (all p > 0.05). In MDC/MPP, SU-instr did not show a causal effect on the risk of all-cause [OR = 1.06, p = 0.32], bladder [OR = 0.96, p = 0.84], lung [OR = 1.26, p = 0.17], bowel [OR = 0.96, p = 0.81], prostate [OR = 1.05, p = 0.62], and breast [OR = 0.99, p = 0.98] cancer. Similar findings were made in the two-sample settings. Detailed results are provided in Table 1.Our MR study, using a series of causal inference approaches, does not support a causal effect of genetically determined SU for major cancer outcomes. There is no evidence to support changing SU levels by lifestyle or pharmacological intervention to attenuate the risk of major cancer types.Table 1.Results from a range of MR analyses for causal effect of SU on cancer risk in MDC/MPP (one-sample MR) and GUGC and UK-Biobank (two-sample MR) cohortsOne-sample MRCancer typeIVW2SLSGRSOR(95% CI)p-causalp-HetOR(95% CI)p-causalOR(95% CI)p-causalBladder0.96(0.68; 1.36)0.840.550.99(0.70; 1.42)0.990.94(0.67; 1.34)0.76Lung1.26(0.90; 1.77)0.170.891.29(0.91; 1.82)0.141.26(0.90; 1.77)0.17Bowel0.96(0.72; 1.28)0.810.440.95(0.72; 1.27)0.770.97(0.73; 1.28)0.83Prostate1.05(0.84; 1.31)0.620.841.04(0.83; 1.30)0.721.04(0.83; 1.30)0.69Breast0.99(0.74; 1.33)0.980.110.98(0.73; 1.32)0.920.98(0.73; 1.32)0.94All-cause1.06(0.93; 1.21)0.320.61.07(0.94; 1.21)0.281.07(0.94; 1.21)0.29Two-sample MRCancer typeIVWWeighted medianMR-EggerOR(95% CI)p-causalp-HetOR(95% CI)p-causalOR(95% CI)p-causalBladder1.02(0.88; 2.16)0.770.361.03(0.86; 2.10)0.71.03(0.83; 2.13)0.75Lung0.86(0.73; 1.08)0.070.040.86(0.73; 1.09)0.090.84(0.66; 1.16)0.15Bowel0.82(0.69; 1.02)0.03< 0.00010.89(0.78; 1.06)0.060.92(0.72;1.67)0.52Prostate1.00(0.91; 2.64)0.970.720.98(0.88; 2.17)0.770.94(0.83; 1.45)0.37Breast0.97(0.88; 1.95)0.670.00021.00(0.91; 2.70)0.990.95(0.81; 1.59)0.46All-cause0.96(0.90; 1.20)0.18< 0.00010.97(0.92; 1.23)0.210.93(0.86; 1.12)0.11OR; odds ratio, 95% CI; 95% confidence interval, IVW; Inverse variance weighted method, 2SLS; two-stage least square, GRS; genetic risk score, p-Het; p-value for heterogeneity.None declared