Bidirectional mendelian randomisation of kidney function and heart failure in african-ancestry populations

Background

Chronic kidney disease and heart failure frequently coexist, and observational studies suggest a correlation between CKD and heart failure risk. However, these associations are susceptible to confounding and reverse causation. Mendelian randomisation (MR) offers an approach to strengthen causal inference, yet evidence from African ancestry populations remains scarce even though their greater genetic diversity and shorter linkage disequilibrium structure which may enhance instrument resolution and causal inference.

Purpose

To investigate the bidirectional causal relationship between kidney function and heart failure using African ancestry–specific genome-wide association study (GWAS) data.

Methods

We performed a bidirectional two-sample MR analysis restricted to African ancestry GWAS. Genetic instruments for estimated glomerular filtration rate (eGFR) were obtained from the African ancestry subset of the Kidney multiome-based genetic scorecard study1 (N=67,943). Heart failure summary statistics were derived from the African ancestry subset of the Global Biobank Meta-analysis Initiative2. Forward (eGFR → heart failure) and reverse (heart failure → eGFR) analyses were conducted using ancestry-matched linkage disequilibrium clumping. Instrument strength was assessed using F-statistics and variance explained (R²). Primary causal estimates were obtained using inverse-variance weighted (IVW) random-effects models, with weighted median and weighted mode estimators as sensitivity analyses. Horizontal pleiotropy, heterogeneity, and directionality were formally evaluated.

To contextualise precision, we calculated the minimum detectable effect (MDE) at 80% power and α=0.05 using the observed R² and effective outcome sample size.

Results

eGFR instruments were strong (minimum F=30.5; median 35.5) but explained 0.37% of variance (R²=0.0037). Median effective heart failure sample size was 4,142.

Forward analysis (eGFR → heart failure) showed no association (IVW OR 0.83, 95% CI 0.17–4.20; p=0.83; 5 SNPs), with consistent sensitivity analyses and no evidence of directional pleiotropy. Power calculations indicated detection of only large effects (minimum detectable OR ≈1.9–2.0).

Reverse analysis (heart failure → eGFR) showed no significant association (IVW β −0.018, SE 0.011; p=0.10; 12 SNPs). Steiger filtering supported directionality.

Conclusion

In this African ancestry–specific bidirectional MR study, we observed no statistically significant evidence of causal association between kidney function and heart failure in either direction. However, despite strong instruments, limited variance explained, and a modest effective sample size, restricted statistical power to detect only large effects. These findings demonstrate the feasibility of ancestry-matched MR in African populations while highlighting the need for larger and more comprehensive African ancestry GWAS to robustly evaluate cardio-renal causal relationships.Figure 1.Mendelian randomisation frameworkFigure 2.Forest plot of MR estimates