PT  - JOURNAL ARTICLE
AU  - Nikita Telkar
AU  - Theresa Reiker
AU  - Robin G. Walters
AU  - Kuang Lin
AU  - Deepti Gurdasani
AU  - Arthur Gilly
AU  - Lorraine Southam
AU  - Emmanouil Tsafantakis
AU  - Maria Karaleftheri
AU  - Janet Seeley
AU  - Anatoli Kamali
AU  - Gershim Asiki
AU  - Iona Y. Millwood
AU  - Huaidong Du
AU  - Yu Guo
AU  - Group Understanding Society Scientific Group
AU  - Meena Kumari
AU  - George Dedoussis
AU  - Liming Li
AU  - Zhengming Chen
AU  - Manjinder S. Sandhu
AU  - Eleftheria Zeggini
AU  - Karoline Kuchenbaecker
TI  - The transferability of lipid-associated loci across African, Asian and European cohorts
AID  - 10.1101/525170
DP  - 2019 Jan 01
TA  - bioRxiv
PG  - 525170
4099  - http://biorxiv.org/content/early/2019/01/20/525170.short
4100  - http://biorxiv.org/content/early/2019/01/20/525170.full
AB  - The under-representation of non-European samples in genome-wide association studies could ultimately restrict who benefits from medical advances through genomic science. Our aim was therefore to address the fundamental question whether causal variants for blood lipids are shared across populations.A polygenic score based on established LDL-cholesterol-associated loci from European discovery samples had consistent effects on serum levels in samples from the UK, Uganda and Greek population isolates (correlation coefficient r=0.23 to 0.28 per LDL standard deviation, p<1.9×10−14). Trans-ethnic genetic correlations between European ancestry, Chinese and Japanese cohorts did not differ significantly from 1 for HDL, LDL and triglycerides. In each study, >60% of major lipid loci displayed evidence of replication with one exception. There was evidence for an effect on serum levels in the Ugandan samples for only 10% of major triglyceride loci. The PRS was only weakly associated in this group (r=0.06, SE=0.013). We establish trans-ethnic colocalization as a method to distinguish shared from population-specific trait loci.Our results provide evidence for high levels of consistency of genetic associations for cholesterol biomarkers across populations. However, we also demonstrate that the degree of shared causal genetic architecture can be population-, trait- and locus-specific. Efforts to implement genetic risk prediction in clinical settings should account for this.