Psoriatic arthritis (PsA) is distinctive amongst the inflammatory/autoimmune joint diseases in that its onset is commonly preceded by cutaneous psoriasis (PsC). This provides an unparalleled opportunity for the identification of predictive biomarkers to determine which of the approximately 25% of psoriasis vulgaris (PsV) patients will develop PsA. Over the past decade, we have expanded our genetic study of PsV to focus on PsA, resulting in the collection of 1,279 PsA patients at Michigan, 743 of whom have already been subjected to GWAS. Initiated in 2007, the International Psoriatic Arthritis Research Team (IPART) has accumulated 1,919 Canadian PsA patients, of whom 1,370 have already been subjected to GWAS. In 2015, we completed a meta- GWAS of PsC and PsA involving a discovery cohort of 1,430 PsA cases and 1,417 controls, with 9,293 additional PsV replication samples (3,061 PsA, 3,110 PsC) and 13,670 controls. We detected 10 associations for PsA and 11 for PsC, as well as a new association for PsV. Utilizing an innovative core exome array to genotype additional cases and controls, we carried out the largest meta-GWAS of PsV to date (~40,000 subjects) and found 16 more susceptibility regions, highlighting the roles of interferon signaling and the NF?B cascade, and demonstrating strong enrichment for psoriasis genetic signals in T-cell regulatory elements. Using machine learning to model ~200 genetic variants in our PsA vs. PsC GWAS, we achieved 82% area under receiver operator curve for distinguishing PsA vs PsC, with 98% accuracy among the top 10% of patients with the highest genetic load. We also carried out RNA-seq on mRNA and miRNA from 65 pairs of pre- and post-conversion samples from PsC patients who developed PsA. Suggestive of a shift from skin- focused to systemic autoimmunity, we found significant post-conversion enrichment for central memory CD4+ T-cell (CD4-Tcm) transcripts among the up-regulated genes. We also observed highly correlated pre- and post- conversion behaviors of 54 differentially-expressed miRNAs and their mRNA targets, as well as multiple serum miRNAs that are significantly differentially expressed in PsA vs. PsC. A metabolomic study of 50 paired converter sera revealed 293 biochemicals with significant alterations, 275 of which were increased. Finally, we identified noncoding eQTLs for IL23R that correlate strikingly with a region of selective PsA association. Based on these results, we hypothesize that PsA and PsC have pathogenetic mechanisms that are T-cell, osteoblast and osteoclast-driven and progress from skin to systemic during transition to PsA. We propose that this paradigm can be used to develop a useful test to predict PsA in PsC patients, while increasing our basic understanding of PsA. To test this hypothesis, we propose four aims: (1) to maintain and grow our longitudinal clinical resource; (2) To identify biomarkers for the development of PsA in PsC patients; (3) To integrate the biomarkers identified in Aim 2 into a clinically-useful tool for PsA identification using machine learning ; and (4) to explore the mechanisms by which the IL23R gene contributes to PsA pathogenesis.
Psoriatic Arthritis (PsA) is a major health problem in the United States and around the world. The mechanisms that predispose patients with psoriasis to develop PsA are unknown. The proposed research will utilize the power of genome-wide association studies, transcriptome analysis, metabolomics, and the largest longitudinal resource of PsA in the world to develop a clinically-relevant tool for PsA prediction and to address a major gap in our mechanistic understanding of the causes of PsA. The results of this research are also likely to be relevant to other autoimmune diseases.
Showing the most recent 10 out of 38 publications
