PO:36:251 | Rheumatoid factor predicts vasculitis relapse in eosinophilic granulomatosis with polyangiitis: a machine learning approach

Luca Iorio¹, Federica Davanzo¹, Giulia Lorenzoni², Eleonora Fiorin¹, Marta Codirenzi¹, Roberta Prevedello¹, Debora Campaniello¹, Dario Gregori², Roberto Padoan¹, Andrea Doria¹. | ¹Rheumatology Unit, DIMED, University of Padova, Padova, Italy; ²Unit of Biostatistics, Epidemiology and Public Health, DCTV, University of Padova, Padova, Italy.

Background. The present study was conducted with the objective of evaluating the frequency and prognostic significance of rheumatoid factor (RF) in patients diagnosed with eosinophilic granulomatosis with polyangiitis (EGPA). A particular emphasis was placed on the potential role of RF in predicting vasculitis relapses.

Materials and Methods. A retrospective cohort study was conducted, encompassing patients diagnosed with EGPA who underwent RF testing (immunoturbidimetric assay/nephelometric assay/enzyme-linked immunosorbent assay) at the time of diagnosis. Clinical and laboratory data were prospectively collected starting from diagnosis and throughout the following 60 months. Relapse was defined as the new onset, recurrence, or worsening of systemic signs or symptoms (excluding asthma and nasal polyps exacerbations) associated with a BVASv3 > 0, requiring the modification of systemic therapy. Kaplan-Meier curves were used to assess relapse-free survival. A Random Forest machine learning model was developed to identify relapse predictors, with support from SHAP analysis.

Results. A total of 83 patients diagnosed with EGPA were included in the present study, 45.8% of whom (38/83) were found to have positive RF. During follow-up, 19.3% (16/83) of patients experienced at least one relapse. Patients with a positive RF test had a significantly higher relapse rate than those with a negative RF test (39.5% vs. 2.2%, p=0.003). The median time to relapse was 29 [11.8-71.8] months. The overall relapse rate at 12, 24 and 60 months were 4.8%, 9.1% and 26.2%, respectively. As illustrated in Figure, the Kaplan-Meier analysis demonstrated a significant lower 5-year relapse-free survival in RF-positive patients compared to RF-negative patients (log-rank test, p < 0.001). Furthermore, patients with a positive RF test had a higher frequency of mononeuritis multiplex (52.6% vs. 13.3%, p = 0.011) and elevated C-reactive protein levels (49.1 mg/L vs. 11.0 mg/L, p = 0.007) compared to RF-negative patients. The Random Forest model identified RF positivity as the strongest relapse predictor (VIMP 0.2031, 95%CI -0.012 to 0.307), outperforming established variables such as ANCA positivity and age at disease onset. Subsequent SHAP analysis confirmed that negative RF was associated with a greater relapse-free probability (phi +0.1469).

Conclusions. RF is a prevalent and clinically significant biomarker in EGPA. In our cohort, it is associated with increased systemic relapse risk. These findings suggest that RF may define a distinct immunopathogenic subset of EGPA, characterised by increased systemic inflammation and vasculitic phenotype. The integration of RF into future prognostic models has the potential to enhance patient stratification and guide personalized treatment approaches.