Enter Note Done


Decrease font size Increase font size

PgmNr 11: Genetics of 38 blood and urine biomarkers in the UK Biobank.

N. Sinnott-Armstrong 1; Y. Tanigawa 2; S. Naqvi 1,3; N.J. Mars 4; D. Amar 2; H.M. Ollila 4,5,6; M. Aguirre 2; G.R. Venkataraman 2; M. Wainberg 7; J.P. Pirruccello 8,9; J. Qian 10; A. Shcherbina 2,11; F. Rodriguez 11; T.L. Assimes 11,12; V. Agarwala 11; R. Tibshirani 10; T. Hastie 10; S. Ripatti 3,9,13; M.J. Daly 3,9,15; J.K. Pritchard 1,4,14; M.A. Rivas 2; FinnGen

View Session  Add to Schedule

1) Genetics, Stanford Univ, Stanford, California.; 2) Biomedical Data Science, Stanford University, Stanford, California; 3) HHMI, Stanford, California; 4) Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Helsinki, Finland; 5) Stanford University, Department of Psychiatry and Behavioral Sciences, Palo Alto, CA, USA; 6) Center for Genomic Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA; 7) Department of Computer Science, Stanford University, Stanford, CA, USA; 8) Massachusetts General Hospital Division of Cardiology, Boston, MA, USA; 9) Program in Medical and Population Genetics and Stanley Center for Psychiatric Research, Broad Institute of Harvard and MIT, Cambridge, MA, USA; 10) Department of Statistics, Stanford University, Stanford, CA, USA; 11) Department of Medicine, School of Medicine, Stanford University, Stanford, CA, USA; 12) VA Palo Alto Health Care System, Palo Alto, CA, USA; 13) Department of Public Health, Clinicum, University of Helsinki, Helsinki, Finland; 14) Department of Biology, Stanford University, Stanford, CA, USA; 15) Analytic and Translational Genetics Unit, Massachusetts General Hospital, Boston, MA., USA

Biomarkers are well suited to testing how and when variation in the genome changes phenotype, as they are often understood on a molecular level. Here, we systematically evaluated the genetic basis of 38 blood and urine laboratory tests measured in 358,072 participants in the UK Biobank.

We identified 1,857 independent loci associated with at least one laboratory test, including 488 large-effect protein truncating, missense, and copy-number variants. These loci included membrane transporter SLC2A9 for urate; the chaperone IGFBP3 for IGF-1; and the activating enzyme SRD5A2 for testosterone, which were all key members of the corresponding gene pathways. More generally, up to 80% of genes in the relevant pathways contained common variation within 50 Kb that significantly altered biomarker levels. Moreover, rare variants also revealed novel coding associations with a number of genes with therapeutic potential.

Our findings suggest that biomarkers are driven by tissue-specific polygenic backgrounds and a few core genes with large effect. To this end, we found tissue- and cell-type specific polygenic signal in kidney tissue for urate (~35-fold); UACR and SHBG in podocytes and LDL in hepatocytes; and creatinine, alkaline phosphatase, and eGFR in proximal tubules. The polygenic architecture of biomarkers echos that of common diseases with the major exception that we have an a priori, molecularly-driven sense of core genes.

Finally, we built combined polygenic risk score (PRS) models using all 38 biomarker PRSs simultaneously. We found substantially improved prediction of incidence in FinnGen (n = 135,500) with the multi-PRS for renal failure and alcoholic cirrhosis (hazard ratio = 1.1 vs no association with trait PRS alone).

Our results reveal that biomarkers are an ideal model system to understand the genetic architecture of complex phenotypes. By combining disease associations with measurements from a number of relevant biomarkers, we can improve the utility, interpretability, and portability of genetic associations.