Pos1021 high throughput multiplex proteomics identifies biomarkers and networks associated with the rheumatoid arthritis-cardiovascular disease (ra-cvd) multimorbid axis

Background Rheumatoid arthritis (RA) patients have higher mortality rates compared to the general population, primarily due to excess cardiovascular (CV) disease (CVD). We have shown the earliest stages of RA are associated with CV abnormalities but blood biomarkers to identify CVD in early RA are lacking. Objectives In a treatment-naïve new-onset RA trial cohort, to identify (i) blood-based protein biomarkers that associate with cardiovascular magnetic resonance imaging (CMR) abnormalities, (ii) whether these biomarkers are sensitive to change; (iii) the predominant inflammatory and metabolic pathways implicated. Methods CADERA (Coronary Artery Disease in Early RA) [1] was an add-on study to a parent RCT of an early RA inception cohort where 81 patients underwent CMR at baseline and year 1. Proximity extension immunoassay (OIink) was used to measure normalised protein expression (NPX) across inflammation, CV II/III and cardiometabolic panels at time of CMR. Bayesian linear mixed effects regression was used to identify proteins associated with CMR parameters of myocardial oedema/fibrosis [MO/MF = native T1, myocardial extracellular volume (ECV) and late gadolinium enhancement (LGE)] and vascular stiffness [VS = aortic distensibility and stiffness index] at baseline, year 1 and those sensitive to change over time. An expanded protein interaction (PPI) network using baseline proteins was created using an induced network approach (String-DB) and clusters identified using k-means that were then subjected to Gene ontology (GO) and KEGG enrichment analysis. Results Of 340 proteins analysed using Olink, 108 proteins were associated with CMR markers of MO/MF (64 proteins; 28 positively, 36 negatively) and VS (44 proteins; 13 positively, 38negatively) at baseline. 46/108 proteins identified at baseline were sensitive to change over time and of these, 15 remained associated with CMR parameters at Year 1 (see Figure 1A & B). No overlapping proteins were identified that associated with focal (LGE) and diffuse fibrosis (ECV). Table 1 reports posterior estimates and 95% credible intervals for two top proteins. Figure 1C displays change in predicted NPX for 2 time-sensitive proteins associated longitudinally with CMR parameters. K-means clustering of the expanded baseline PPI network (enrichment p < 0.0001) identified 4 clusters (Figure 1D) with roles in IL27 receptor binding and NF-kB signalling pathway (red), vascular endothelial growth factor receptor binding and JAK-STAT signalling pathway (green); Macrophage CSF receptor binding and ErbB signalling pathway (blue); and IL10 receptor activity and complement and coagulation cascades (yellow). Conclusion This is the first study to identify sensitive protein biomarkers that may help with early diagnosis and monitoring of RA-CVD and inform on possible therapeutic targets. Treatment group differences and subgroup analyses with responders are ongoing. Further validation of these candidate biomarkers is needed in an independent cohort. Table 1. Top proteins sensitive to change and longitudinally associated with CMR detected subclinical myocardial oedema(MO)/fibrosis(MF) and vascular stiffness (VS). PE = posterior estimate, CrI = credible intervals. Proteins Role Baseline PE(95%CrI) Year 1 PE(95%CrI) MO/MF *Protein A Lipid metabolism -4.33(-8.07 to -0.6) -52.18 (-97.4 to -6.8) VS *Protein G Growth factor, heparin binding 0.84(0.12 to 1.58) 1.31(0.2 to 2.4) *Proteins, clusters subject to intellectual property discussions via Innovation Factory, University of Manchester. Figure 1. A(middle):Overlap of proteins associated with CMR parameters at baseline, year 1 and those sensitive to change. B(top right): Posterior estimate of 15 proteins from the intersect of figure 1A. C(bottom right): Predicted normalised protein expression of protein D(bottom) and H (top)at baseline and Year 1. D(left): Baseline PPI network analysis identifies 4 clusters. Reference [1] Plein S, et al Ann. Rheum. Dis. 2020;79:1414-1422. Acknowledgements: NIL. Disclosure of Interests None Declared.