OWE-16 Development and clinical validation of a genetic risk score for coeliac disease

Introduction Specific HLA-DQ genes predispose to coeliac disease (CD) and HLA typing is occasionally used as a rule-out test in clinic. However, CD is polygenic and genome wide association studies (GWAS) have implicated ∼40 additional genetic variants. Using single nucleotide polymorphisms (SNPs) we aimed to combine all associated variants into a genetic risk score and assess its utility as a clinical tool. Methods We used imputation to identify SNPs strongly correlated (r2>0.95) with 4 key HLA-DQ haplotypes (DQ2.5/DQ2.2/DQ7.5/DQ8) in UK Biobank. We derived HLA-DQ odds ratios from 12,041 cases and 12,228 controls (Wellcome Trust). We combined this with additional SNPs from recent GWAS to generate a coeliac genetic risk score (C-GRS). We validated the C-GRS in a population based cohort (UK Biobank) with 1237 cases identified by hospital admission codes. We genotyped the C-GRS in 161 samples from a paediatric clinic where patients had been assessed using anti-tissue transglutaminase antibodies, biopsy and HLA typing. Results The C-GRS consisted of 42 SNPs and was highly discriminative of CD in UKBiobank. The C-GRS was more discriminative than HLA stratification alone (ROC-AUC=0.88 [95%CIs:0.87–0.89] v 0.81, p Conclusions A C-GRS can aid in identifying incident cases of CD and is more effective than HLA typing alone. Given the low costs of SNP genotyping relative to HLA typing a C-GRS could improve the availability and utility of coeliac genetic testing in CD diagnosis and in recruitment to research studies.