Genetic analysis of retinal traits

Abstract

Retina is a unique site in the human body where the microcirculation can be imaged directly and non-invasively, allowing us to study in vivo the structure and pathology of the human microcirculation. Retinal images can be quantitatively assessed with computerized imaging techniques, enabling us to measure several different quantitative traits derived from the retinal vasculature. Arterial and venular calibres are the most extensively studied traits of the retinal microvasculature and numerous epidemiological studies demonstrated promising associations with systemic and ocular diseases as well as with disease markers. However, there has been a lack of research into pathophysiological processes leading to retinal vascular signs, and how they link retinal microcirculation with coronary and cerebral microvasculature change. Information about genetic determinants underlying retinal vascular structure is therefore important for understanding the processes leading to microvascular pathophysiology. Two genome wide association studies have been published so far revealing four loci associated with retinal venular calibre and one locus with arteriolar calibre. Here the results from the genome-wide association analysis of 10 different retinal vessel traits in two population based cohorts are presented. Retinal images were measured in non-mydriatic fundus images from 808 subjects in the Orkney Complex Disease Study (ORCADES) and 390 subjects from the Croatian island of Korcula, using the semi-automated retinal vasculature measurement programme SIVA and VAMPIRE. Using pairwise estimates of kinship based on genomic sharing, heritability was calculated for each trait. Estimates of tortuosity measure and fractal dimensions present first published reports of heritability estimates for those traits. In addition correlation analysis with systemic risk factor was also completed, confirming already published results as well as revealing some new associations. A genome wide association analysis of retinal arteriolar width revealed a genome wide significant hit (1.8x10-7) in a region of chromosome 2q32 (within TTN gene). Replication was sought in a further independent Scottish population (LBC) and additional 400 retinal images were graded. The result did not replicate, however the direction of the effect was consistent and a larger sample size is required. Analysis of the remaining traits did not yield genome wide significant result,s and will also require larger sample sizes. Genetic analysis of a binary retinal trait was also explored in a case control study of retinal detachment, which is an important cause of vision loss. A two-stage genetic association discovery phase followed by a replication phase in a combined total of 2,833 RRD cases and 7,871 controls was carried out. None of the SNPs tested in the discovery phase reached the threshold for association. Further testing was carried out in independent case-control series from London (846 cases) and Croatia (120 cases). The combined meta-analysis identified one association reaching genome-wide significance for rs267738 (OR=1.29, p=2.11x10-8), a missense coding SNP and eQTL for CERS2 encoding the protein ceramide synthase 2. Additional genetic risk score, pathway analysis and genetic liability analysis were also carried out.