Abstract
Genome-wide association study (GWAS) meta-analyses of complex traits have identified thousands of loci, whose biological characterization is underway. In meta-analyses, genetic variant effect sizes vary across contributing studies due to environmental or genetic differences. However, how much study-specific features may allow individual studies to contribute to the elucidation of underlying mechanisms is unclear. As a case study, we compared a GWAS of estimated glomerular filtration rate (eGFR) in a population-based study from the Alps, where thyroid disease is common, versus a meta-analysis on >1 million individuals that previously identified 147 kidney function relevant loci.
We performed a GWAS of age- and sex-adjusted ln(eGFR) on 10,146 participants in the Cooperative Health Research in South Tyrol (CHRIS) study, accounting for genomic kinship (genomic inflation λ=1.02). We assessed direction-consistent replication at any variant strongly correlated with the locus lead variant (r2≥0.8), accounting for multiple testing. Replicated variants underwent phenome-wide mediation analysis across 72 health biomarkers.
We replicated 10 loci. The variants’ effect magnitudes were 1.3-to-5.8 times larger than in the meta-analysis, at similar minor allele frequencies (MAF). Free triiodothyronine and thyroxine, measured in individuals with low or high thyroid stimulating hormone levels, resulted as potential mediators for most loci. We observed potential interactions with hyperthyroidism at SHROOM3 (p=0.03) and with hypothyroidism at PIP5K1B (p=0.03) and GAB2 (p=0.04).
Similar MAF, same ancestry and loci independency suggest the larger effects on eGFR in the CHRIS study being probably of environmental origin. The potential modifier role of thyroid disease warrants independent replication.
H2020-MSCA-ITN-2019 ID:860977 (TrainCKDis)