Non-linear causal effects of kidney function on the risk of heart failure: a Mendelian randomization study

Abstract Background Kidney insufficiency is defined as the early stage of heart failure, but Mendelian randomization (MR) analyses based on several large heart failure cohorts suggest no linear causal effect between kidney function and heart failure (HF), inconsistent with observational studies evidence. However, whether there is a non-linear causal relationship between kidney function and HF has not been reported. Methods and Results Observational and non-linear MR analyses were performed using individual-level data from a UK Biobank cohort of white British ancestry subjects (N = 386,304). There were non-linear observational associations of between kidney function as assessed with estimated glomerular filtration rate (eGFR) and HF (P for nonlinear <0.001). Non-linear MR analyses showed a U-shaped association between genetically predicted eGFR values (both creatinine- and cystatin C-based eGFR) and the risk of HF (Quadratic P value < 0.05) and the risk of HF increased with both reduced and supranormal eGFR levels (β1 = -1.70E-05, P1-value = 0.004; β2 = 3.56E-06, P2-value = 0.004). Conclusion This study observed a non-linear U-shaped genetic relationship between eGFR and HF outcomes, indicating that both reduced and supranormal eGFR may be causally linked to a higher risk of HF.Central Illustration