Polygenic scores for estimated glomerular filtration rate in a population of general adults and elderly - comparative results from the KORA and AugUR study.

Background: Polygenic scores (PGSs) combining genetic variants found to be associated with creatinine-based estimated glomerular filtration rate (eGFR) have been applied in various study populations with different age ranges. This has shown that PGS explain less eGFR variance in the elderly. Our aim was to understand how differences in eGFR variance and the percentage explained by PGS varies between population of general adults and elderly.

Results: We derived a PGS for cystatin-based eGFR (eGFR) from published genome-wide association studies. We used the 634 variants known for eGFR and the 204 variants identified for eGFR to calculate the PGS in two comparable studies capturing a general adult and an elderly population, KORA S4 (n = 2,900; age 24-69 years) and AugUR (n = 2,272, age ≥ 70 years). To identify potential factors determining age-dependent differences on the PGS-explained variance, we evaluated the PGS variance, the eGFR variance, and the beta estimates of PGS association on eGFR. Specifically, we compared frequencies of eGFR-lowering alleles between general adult and elderly individuals and analyzed the influence of comorbidities and medication intake. The PGS for eGFR explained almost twice as much (R = 9.6%) of age-/sex adjusted eGFR variance in the general adults compared to the elderly (4.6%). This difference was less pronounced for the PGS for eGFR (4.7% or 3.6%, respectively). The beta-estimate of the PGS on eGFR was higher in the general adults compared to the elderly, but similar for the PGS on eGFR. The eGFR variance in the elderly was reduced by accounting for comorbidities and medication intake, but this did not explain the difference in R-values. Allele frequencies between general adult and elderly individuals showed no significant differences except for one variant near APOE (rs429358). We found no enrichment of eGFR-protective alleles in the elderly compared to general adults.

Conclusions: We concluded that the difference in explained variance by PGS was due to the higher age- and sex-adjusted eGFR variance in the elderly and, for eGFR, also by a lower PGS association beta-estimate. Our results provide little evidence for survival or selection bias.