Exploring the Causal Relationship Between Body Mass Index and Kidney Function: Tissue-Partitioned Mendelian Randomization

Background: Chronic kidney disease (CKD) represents a leading non-communicable disease, significantly contributing to global morbidity and mortality. Mendelian randomization studies (MR) have been integral in providing robust evidence that increased body mass index (BMI) has a causal impact on CKD. However, dissecting which specific mechanisms are primarily consistent with disease development remains challenging, and tissue-partitioned genetic analyses may provide exploratory insight into this underlying biological architecture. Objective: To investigate whether BMI-associated genetic variants prioritized according to colocalization with brain or subcutaneous adipose tissue gene expression exhibit differential associations with kidney function using tissue-partitioned MR. Methods: We employ a two-sample univariable and multivariable MR (MVMR) methodology that segregates genetic variants associated with BMI based on colocalization with gene expression in either brain or subcutaneous adipose tissue. We utilize sets of adipose and brain expression quantitative trait loci (eQTLs) that demonstrated colocalization with BMI (88 and 141 SNPs, respectively). We also use GWAS summary statistics of creatinine and cystatin C-based eGFR (eGFRcrea; n = 1,004,040 and eGFRcys; n = 460,826) blood urea nitrogen (BUN; n = 852,678), eGFR decline (n = 343,339) and CKD (defined as eGFRcrea < 60 mL/min/1.73m2; 41,395 cases and 439,303 controls) of European ancestry. Results: Univariable MR showed consistent positive associations between BMI and CKD (OR = 1.17, 95% CI: 1.1 to 1.23) and inverse associations with eGFRcys (beta = −0.051, 95% CI: −0.056 to −0.047). Both brain- and adipose-instrumented BMI showed similar effect sizes. However, MVMR provided no evidence that either tissue-partitioned BMI component exhibited a distinguishable association with CKD after mutual adjustment, (ORadipose = 1.2, 95% CIadipose = 0.87 to 1.7, ORbrain = 1.18, 95% CIbrain = 0.9 to 1.55) or other kidney function traits. Conclusions: While genetically predicted BMI was associated with kidney function, our tissue-partitioned MR analysis found no evidence that BMI genetic variants prioritized according to brain or subcutaneous adipose tissue gene expression exhibited distinguishable associations with kidney outcomes in MVMR. These findings may reflect shared genetic architecture, limited ability to distinguish highly correlated tissue-partitioned BMI components, or insufficient power to detect modest differences between them.