THU0104 HIGH-PERFORMANCE CHEMICAL ISOTOPE LABELING LIQUID CHROMATOGRAPHY MASS SPECTROMETRY FOR DISCOVERY OF METABOLITE BIOMARKERS OF RHEUMATOID ARTHRITIS

Background: Early diagnosis of rheumatoid arthritis (RA) is hampered by suboptimal accuracy of currently available serological biomarkers. Recent advancements in metabolomic profiling include liquid chromatography mass spectrometry (LC-MS) and in-depth profiling of amine/phenol and carboxyl submetabolomes, resulting in 1000-fold increase in detection sensitivity, and universal metabolome-standard methodology to facilitate metabolome comparisons among different data sets. Objectives: We aimed to identify a metabolite signature with consistently high accuracy for RA. Methods: Sera from 2 RA cohorts were analyzed: Cohort A samples were from 50 RA patients, 39 female (mean age 49.9), 11 male (mean age 47.8), symptom duration 3.7, naive to b-DMARD, and 50 age and sex-matched healthy controls. Cohort B samples were from 50 RA patients, 40 female (mean age 53.4), 10 male (mean age 57.2), symptom duration Results: A total of 3415 amine/phenol and 2114 carboxyl metabolites were commonly detected in more than 80% of the samples. For amine/phenol submetabolome profiling, partial least squares discriminant analysis (PLSDA) showed a clear separation of the groups for each cohort (R2=0.98;Q2=0.92 for cohort A and R2=0.93;Q2=0.79 for cohort B). Similarly, for carboxyl submetabolome profiling, a clear separation between RA and controls was demonstrable (R2=0.93;Q2=0.80 for cohort A and R2=0.84;Q2=0.55 for cohort B). 13 positively identified amine/phenol-containing metabolites, including o-phosphoethanolamine and glycyl-valine, were both significant in cohort A and cohort B, and each of these metabolites showed similar fold changes between RA and controls for both cohorts. 5 carboxyl-containing metabolites, with 1 positively identified, azelaic acid, and 4 unidentified, were both significant in 2 cohorts. The ROC AUC (95%CI) of the 13 amine/phenol-containing metabolite panel were 0.99 (0.94-1.00), 0.98 (0.92-1.00) and 0.98 (0.95-1.00) for cohort A and cohort B (pre- and post-treatment), respectively, with sensitivity/specificity of 94%/96%, 94%/95%, 94%/94%. The ROC AUC (95%CI) of the 5 carboxyl metabolite panel were 0.92 (0.86-0.97), 0.96 (0.91-0.99), 0.89 (0.82-0.95), respectively, with sensitivity/specificity of 86%/86%, 90%/91%, 80%/80%. The combined panel of 18 metabolites demonstrated ROC AUC>0.99 with sensitivity and specificity>95% for each cohort. None of these biomarker metabolites correlated with age, gender, or symptom duration. Conclusion: Consistent discrimination in metabolite profiles between discovery and verification cohorts generated high priority candidates for further biomarker validation in RA. Disclosure of Interests: Xiaohang Wang: None declared, Joel Paschke: None declared, Rana Dadashova: None declared, Edna Hutchings: None declared, Liang Li: None declared, Walter P Maksymowych Grant/research support from: AbbVie, Pfizer, Janssen, Novartis, Consultant for: AbbVie, Eli Lilly, Boehringer, Galapagos, Janssen, Novartis, Pfizer and UCB Pharma; Chief Medical Officer for Canadian Research and Education Arthritis