P898 Uncovering of tissue-associated risk factors that lead to different genetic risk profiles across Inflammatory Bowel Disease patients

Gaite, A.(1)*;Sánchez-Mayor, M.(1);Bujanda, L.(2);M. Marigorta, U.(1);

(1)CIC bioGUNE, Integrative Genomics, Derio, Spain;(2)Biodonostia, Gastrointestinal Disease Group, San Sebastián, Spain;

Background

Polygenic risk scores (PRSs) are emerging as a tool of choice to summarize the risk of disease in each individual. The calculation of PRSs is based on genetic variants and effect sizes obtained from large studies using many different cohorts of patients. In inflammatory bowel disease (IBD), GWAS meta-analysis based on over 75.000 IBD patients from different multi-ethnic cohorts has discovered around 300 independent associated variants. This information can be used to estimate genetic risk and detect individuals with several higher odds of developing the disease during their lifetime. However, PRSs usually do not incorporate functional information about causal variants, and the potential for heterogeneity in effect sizes among subgroups of individuals is disregarded entirely. Hence, PRSs might be ill-suited to capture subcomponents of genetic risk that may account for key differences among specific etiological profiles of patients. Fine mapping of GWAS causal variants show an enrichment in variants associated with changes in gene expression in specific tissues, suggesting that risk of complex diseases is led by an accumulation of variants that change the transcriptome towards a pathogenic profile. To explore the biological nature of IBD risk we propose a tissue transcriptional risk (t-TRS) score

Methods

We perform Transcription Wide Association studies (TWAS) based on gene expression prediction models in nine IBD relevant tissues and a meta-analysis of European IBD GWAS. We next estimated the t-TRS of people from UK biobank and compared its prediction performance against PRS measuring the area under the curve (AUC)

Results

We performed nine TWAS studies and obtained 636 genes that are genome wide significant in at least one tissue. Although eQTL weights are shared across tissues, we observe a high number of tissue specific signals, specifically in terminal ileum and lymphocytes. Pathway enrichment analysis on the 600-gene set showed enrichment of immune functions like “Signaling by interleukins” and “Antigen presentation”. We also show the predicting value of t-TRS is better than the PRS in all tissues. In addition, clustering of individuals according with t-TRS yield 3 subgroups of patients

Conclusion

Functionally informed genetic predictors allow not only to better stratify patients, but also to discover which are the biological pathways that accumulate risk variants in subgroups of IBD patients