= Emerging. More than 5 years before clinical availability. (24.37%, 2023)
= Expected to be clinically available in 1 to 4 years. (39.50%, 2023)
= Clinically available now. (36.13%, 2023)
MSACL 2023 : Wang

MSACL 2023 Abstract

Self-Classified Topic Area(s): Metabolomics

Podium Presentation in Steinbeck 2 on Thursday at 14:40 (Chair: Kelly Hines / Helen Jordan)

Using Multiple Serum Sample Cohorts with Chemical Isotope Labeling LC-MS to Discover Biomarkers for Rheumatoid Arthritis Disease

Xiaohang Wang (1), Wei Han (1), Walter P. Maksymowych (2), Liang Li (1, 3)
(1) The Metabolomics Innovation Centre (TMIC), University of Alberta, Edmonton, Canada; (2) Canadian Research and Education (CaRE) Arthritis, Edmonton, Canada; (3) Department of Chemistry, University of Alberta, Edmonton, Canada

Xiaohang Wang, Master (Presenter)
The Metabolomics Innovation Centre

Presenter Bio: Xiaohang Wang graduated from Department of Chemisty, Univeristy of Alberta in 2018 with a master degree. He is currently working in The Metabolomics Innovation Centre (TMIC) as a research assistant since 2021.


INTRODUCTION: Rheumatoid arthritis (RA) is a common and chronic symmetric polyarticular arthritis and autoimmune disease. Early diagnosis of RA is hampered by suboptimal accuracy of currently available serological biomarkers. Metabolomics is a powerful tool to characterize the complex biochemical phenotypes. In this work, we applied a high-performance chemical isotope labeling (CIL) LC-MS technique for in-depth profiling of the amine/phenol-submetabolome in serum samples to obtain high-confidence biomarker candidates.

METHODS: To avoid false positives, four batches of serum samples were analyzed, including two separate discovery cohorts (Cohort 1 and 2), one verification cohort (Cohort 3) and one validation cohort (Cohort 4). The first three cohorts comprised 50, 49, and 131 RA patients, respectively. Within each cohort, there were sex/age-matched healthy controls: 50 in Cohort 1, 50 in Cohort 2, and 100 in Cohort 3. Among these 446 subjects, 75% were females and the average age was 52.5. The patient and group information for the Cohort 4 was totally blinded. After protein precipitation, amine/phenol-containing metabolites in each individual sample were labeled by 12C-dansyl chloride to improve the LC-MS detection. For each cohort, a pooled sample was generated by combining aliquots from all the individual samples, and labeled by 13C-dansyl chloride to serve as the reference sample for relative quantification. The total dansyl-labeled metabolite amount of each sample was quantified and normalized using LC-UV. After mixing individual 12C-labeled serum with 13C-labeled pool in equal amount, the mixtures were analyzed by an LC-QTOF-MS platform to output the intensity ratios of 12C/13C peak pairs. Peak pairs or metabolites were finally processed, identified and statistically analyzed.

RESULTS: 1267 amine/phenol-containing peak pairs or metabolites were commonly detected across the four sample cohorts. Among them, 352 metabolites were high-confidence identified. For the first three cohorts, visualized by the partial least squares discriminant analysis (PLS-DA), the overall amine/phenol-submetabolome demonstrated clear and consistent differences between RA and healthy controls, with cross-validation Q2 = 0.751, 0.796, 0.767, respectively. The selection of biomarker candidates was first conducted according to the fold change and false-discovery-rate-adjusted Welch's t-test in cohort 1. It demonstrated 118 metabolites having higher concentrations in the RA samples than the controls, and 80 metabolites with lowered concentrations, and of those, 69 metabolites were identified with high confidence. Uni-variate receiver operating characteristic (ROC) curves of these 69 metabolites were generated and 18 metabolites with the highest univariate classification performance were selected as biomarker candidates to form the initial biomarker panel. This panel was then tested by cohort 2 and six metabolites were decided to be kept for the panel. The six-biomarker panel was further tested and verified by cohort 3. We then implemented the logistic regression to build ROC curves to measure the performance of the classifier in three cohorts. The area-under-the-curve (AUC) values (95% confidence interval) were 0.948 (0.848-1.000), 0.889 (0.779-0.980) and 0.942 (0.936-0.988) for cohort 1, 2 and 3, respectively. Confounders, including gender, age, etc., did not show strong influences on the biomarker panel. The results revealed the importance of examining multiple sample sets and even in the worst case (Cohort 2), our biomarker candidates could differentiate RA at 76% sensitivity and 98% specificity. In addition, in Cohort 3, there were 30 RA patients negative for anti-cyclic citrullinated peptide and rheumatoid factor, and the six-metabolite panel demonstrated consistently high performance for differentiating these specific subjects from healthy controls. The diagnostic model is being further validated using the cohort 4 with blinded sample information.

CONCLUSION: Serum metabolic changes can well and consistently differentiate rheumatoid arthritis from healthy controls in two separate discovery cohorts and a verification cohort. Several significantly changed metabolites have been identified and verified in the first three cohorts with good discrimination power and consistency, which can serve as potential biomarkers.

Financial Disclosure

Board Memberno
IP Royaltyno

Planning to mention or discuss specific products or technology of the company(ies) listed above: