POS0474 FACTORS ASSOCIATED WITH EROSIVE RHEUMATOID ARTHRITIS, A MULTIMARKER PRINCIPAL COMPONENT ANALYSIS (PCA) AND PRINCIPAL COMPONENT REGRESSION (PCR) ANALYSIS

Background

Various clinical (disease activity, seropositive RA etc.) and metabolic risk factors (Dkk1 etc.) have been associated with erosive rheumatoid arthritis (RA). However, such risk factors might be intertwined, and multicollinearity might reduce our ability to discern the individual contribution to erosive score. Principal component analysis (PCA) is statistical technique for reducing dataset's dimension and principal component regression (PCR) is a regression analysis based on PCA. PCR overcomes the multicollinearity problem.

Objectives

To investigate the clinical and bone metabolic risk factors associated with erosive RA using PCA and PCR.

Methods

We conducted a cross-sectional analysis on seropositive RA patients not responding to first-level disease-modifying antirheumatic drug, candidate to bDMARD treatment. Clinical, radiographic (both hands and feet x-ray), laboratoristic and densitometric (BMD parameters were collected. Sharp van der Heijde Score (SvdHS) was calculated by two independent readers. Serum samples were collected and assayed for C-terminal telopeptide of type I collagen (CTX), Procollagen I Intact N-Terminal Peptide (P1NP), Dkk1, Sclerostin (SOST), 25-OH-Vitamin D (VitD), and PTH. PCA was applied to reduce dimensionality of the dataset and find clusters of variables recording largely redundant information. PCs were selected based on eigenvalues explaining >75% of total variance. PCR was used to predict the SvdHS. Results were analyzed using PCA package on GraphPad Prism version 9.5.0 for Windows, GraphPad Software, San Diego, California USA.

Results

62 RA patients aged 57.2 years (SD 12.1) were consecutively enrolled. Mean DAS28-CRP was 4.17 (SD 1.27) and median SvdHS was 24 (IQR 12-53). The loadings plot (Figure 1) shows the clusters of correlated variables in the dataset (vectors). In Table 1 are presented the results of the PCR with SvdHS as outcome. We found that age, GC treatment, ACPA titer, RF titer, CRP levels, ESR, CTX serum levels and Dkk1 serum levels were significantly positively correlated with SvdHS, whereas P1nP serum levels and PGA were negatively correlated with SvdHS.

Conclusion

We found that age, seropositivity, and inflammation were the independent clinical risk factors associated with erosive RA. CTX and Dkk1 serum levels were the metabolic factors independently associated with erosive disease whereas P1nP serum levels were associated with less erosions.

REFERENCES:

NIL.

Acknowledgements:

NIL.

Disclosure of Interests

Giovanni Adami Speakers bureau: EliLilly, Theramex, Amgen, UCB, Galapagos, Fresenius Kabi, Giovanni Orsolini: None declared, Angelo Fassio: None declared, Ombretta Viapiana: None declared, Elena Sorio: None declared, Camilla Benini: None declared, Davide Gatti: None declared, Davide Bertelle: None declared, Maurizio Rossini: None declared.