Associations of Genetically Predicted CKD With Urinary Tract Cancer and Lung Cancer: A Mendelian Randomization Analysis

Rationale & Objective: Chronic kidney disease (CKD) is reported to be associated with cancer, especially for urinary tract and lung cancer. However, whether this suggests causality has not been resolved. This study aimed to investigate the causal relation of CKD to overall, urinary tract, and lung cancer. Study Design: Mendelian randomization (MR) analysis. Setting & Participants: Single-nucleotide polymorphism (SNP)-CKD data were obtained from genome-wide association studies of a meta-analysis of CKDGen and UK Biobank. SNP-cancer data from genome-wide association studies of several community-based and cancer-specific consortia were extracted and then meta-analyzed. Exposures: Impaired kidney function (creatinine-based estimated glomerular filtration rate [eGFRcr] <60 mL/min/1.73 m²), increased albuminuria (urinary albumin-creatinine ratio [UACR] >30 mg/g), and on a continuous scale, eGFRcr, cystatin C-based eGFR (eGFRcys), and UACR. Outcomes: Incidences of overall cancer, urinary tract cancer, and lung cancer. Analytical Approach: Pooled MR estimates of single-SNP Wald ratios were obtained using an inverse variance-weighted method. Results: In inverse variance-weighted MR analyses, a higher genetic liability to impaired kidney function was not significantly associated with higher overall cancer risk (OR, 1.00; 95% CI, 1.00-1.01; P_OR = 0.06). No associations of other kidney phenotypes including increased albuminuria, eGFRcr, eGFRcys, and UACR with overall cancer risk were found at the Bonferroni-corrected significance level (all P_OR > 0.0025). Similarly, risk estimates of generally null were detected in the associations of all these CKD phenotypes with specifically urinary tract and lung cancer (all P_OR > 0.0025). The results of our sensitivity analyses, including pleiotropy-robust MR, reverse MR, and multivariable MR analyses, corroborated the results of our main analyses. Limitations: Uncertain extrapolation to other ethnicities. Conclusions: Genetically predicted CKD is not associated with the risk of incident overall, urinary tract, and lung cancer. Our findings thus provide no genetic evidence for a causal relationship between CKD and cancer. Plain-Language Summary: Despite growing evidence for an association between chronic kidney disease (CKD) and cancer, causality remains unresolved because of the confounding bias inherent to observational studies. To test causality, we applied Mendelian randomization, which uses genetic variants to minimize confounding. We genetically predicted 5 CKD traits using genetic variants as instruments and examined their associations with several sets of large-scale cancer genome-wide data. Our Mendelian randomization analyses showed robust null associations of genetically predicted CKD traits and cancer outcomes. Although we found no genetic evidence for causality linking CKD to cancer, our study does not exclude the potential role of CKD for predicting cancer incidence and emphasizes the need for alternative study approaches to investigate causality in the future.