Purpose : Keratoconus is a relatively common, progressive disease of young adults with a strong genetic component. We aim to identify genetic risk factors.

Methods : A genome-wide association study was conducted using the Illumina HumanCoreExome SNP array. Australian keratoconus cases (n=522) were compared to Australian controls (n= 2036) drawn from the Blue Mountains Eye Study and patients with age-related macular degeneration previously genotyped (Fritsche et al 2016). Following data cleaning, genotypes were imputed against the 1000 Genomes Project reference panel (Phase III, version 5) using Minimac3 v2.0.1 and association analysis was completed in Plink V1.9. SNPs with p<1x10-6 were evaluated by imputation in a previous keratoconus GWAS data set (Li et al 2012) from the USA and lead SNPs were genotyped in a second Australian case-control cohort. All top ranked SNPs are undergoing genotyping in a third replication cohort consisting of cases from Australia and Northern Ireland.

Results : The GWAS identified 40 variants at 8 loci with p<1x10-6. Previously reported loci near RXRA-COL5A1 and FOXO1 (to which this cohort contributed) were confirmed, reaching genome-wide significance of p<5x10-8. FNDC3B was also confirmed, and all three loci were replicated in both additional cohorts examined to date. SNP rs7593141 (P-discovery = 3.83x10-7) in a gene desert on chr2 was replicated in the second Australian cohort (p=0.003) as was the novel locus on chr16 at rs980304 (p-discovery = 9.89x10-7, P-replication = 5.12x10-6). Both SNPs showed a trend in the same direction in the US cohort but did not reach nominal significance. SNPs on chr18 upstream of CDH7 (e.g. rs9950788, P-discovery= 2.37x10-7) showed replication in the Australian cohort (p= 8.23x10-5), but the US sample showed the opposite direction of association across the whole region. Analysis of the third replication cohort and meta-analyses of all cohorts are ongoing.

Conclusions : At least two new loci for keratoconus risk have been identified, namely a gene desert on chr2, and on chr16 in the region of FBXO31, MAP1LC3B and ZCCHC14 genes. Due to differences in the direction of association between the Australian and US cohorts, additional data are required to draw conclusions on the locus upstream of CDH7 on chr18. Meta-analysis of all available cohorts will also determine if any novel loci reach genome-wide significance in this study, the largest keratoconus GWAS to date.