Table 4.
Explained variance and genetic risk score analyses.
| (a) | |||||
|---|---|---|---|---|---|
| Study | N | Study design | Number of variants | sd of age-/sex-adjusted log eGFRcrea in the respective study | R2 |
| UKB | 436,581 | Population-based | 634 | 0.15 | 9.3% |
| HUNT | 69,389 | Population-based | 625 | 0.15 | 6.7% |
| MGI | 47,219 | Hospital-based | 620 | 0.28 | 3.7% |
| MVP | 300,680 | Hospital-based | 620 | 0.28 | 4.1% |
| Second meta | 417,288 | Meta-analysis | 632 | 0.13a | 9.8% |
| 0.28b | 2.0% |
| (b) | ||||||
|---|---|---|---|---|---|---|
| Study | N | GRS | bGRS per sdGRS | Mean eGFRcrea difference for 95th vs 5th percentile of GRS | PGRS | R2 |
| HUNT | 26,254 | Unweighted | −2.62 ml/min/1.73 m2 | −8.6 ml/min/1.73 m2 | 1.5E-282 | 4.8% |
| Weighted | −2.88 ml/min/1.73 m2 | −9.5 ml/min/1.73 m2 | 6.3E-344 | 5.8% | ||
| AugUR | 1105 | Unweighted | −2.49 ml/min/1.73 m2 | −8.2 ml/min/1.73 m2 | 1.2E-08 | 2.9% |
| Weighted | −2.98 ml/min/1.73 m2 | −9.8 ml/min/1.73 m2 | 8.8E-12 | 4.2% |
aFrom population-based ARIC (as in Wuttke et al.7).
bFrom hospital-based MVP.
Shown are results from the explained variance and genetic risk score (GRS) analysis based on the 634 identified signal index variants. (a) Summary of explained variance analyses based on summary statistics from population- and hospital-based studies or from the second meta-analysis. UKB was part of the primary identifying meta-analysis. HUNT, MVP and MGI were independent studies, which were meta-analysed as second meta-analysis (n = 417,288). The variance explained by the 634 signal lead variants (R2) was computed based on genetic effects, genotype and phenotype variance from the respective study. Since phenotype variance was not available for the second meta-analysis, we here assumed phenotype variances taken from the population-based study ARIC or from the hospital-based study MVP. (b) Summary of GRS analyses. For the GRS, two example studies of different age range were analysed, each independent from the identifying GWAS meta-analysis: HUNT (n = 26,254, population-based, age 19–99 years, sd of age-/sex-adjusted eGFRcrea = 11.9 ml/min/1.73 m2) and AugUR (n = 1105, population-based of mobile elderly, age 70–95 years, sd of age-/sex-adjusted eGFRcrea = 14.6 ml/min/1.73 m2). The unweighted and weighted GRS (i.e., using effect sizes from eGFRcrea GWAS) were computed and the association of the GRS on eGFRcrea (not log-transformed) and the variance explained (R2) were derived via linear regression with GRS as covariate and eGFRcrea as outcome (adjusted for age, sex and principal components, “Methods“; GRS and eGFRcrea descriptives in Supplementary Table 1).