OA07 Patient-reported outcomes as early warning signals of flare following drug cessation in rheumatoid arthritis

Withdrawal of disease-modifying antirheumatic drugs (DMARDs) in patients with rheumatoid arthritis (RA) in remission offers an opportunity to reduce medication burden, but carries a risk of arthritis flare which requires rapid clinical intervention. Detecting flare using traditional measures such as DAS28 requires frequent face-to-face assessments, increasing both the complexity and cost of patient monitoring. In contrast, patient-reported outcomes (PROs) can measure changes in symptoms and disease impact remotely. In this study, we explored the potential of PROs for remote flare detection in patients with RA in sustained remission after conventional synthetic DMARD (csDMARD) cessation.

The BIO-FLARE study is a multicentre, prospective, open-label experimental study of csDMARD cessation amongst RA patients in remission. Four PROs (FLARE-RA, EQ5D, RAPID-3 and RA-FQ) were captured at baseline and at subsequent visits until time-of-flare (defined by DAS28-CRP ≥3.2, or DAS28-CRP ≥2.4 on two occasions within 14 days) or end of 6 months’ follow-up. Longitudinal changes were visualised using LOESS plots with K-fold cross validation. Cox regression models incorporating time as a covariate were fitted to evaluate association between PRO changes and the likelihood of subsequent flare. Receiver operator characteristic (ROC) curve analysis then enabled discriminatory changes in each PRO from baseline to be determined and compared as means of either predicting or confirming subsequent flare events (area under ROC; diagnostic test evaluation).

58 (47.9%) of the 121 participants (70.1% females, mean age 64.8 years) experienced a flare. There were no significant differences in baseline PRO scores between groups except for EQ5D visual analogue scale which was higher in the flare group (p = 0.015). Every 1-point change in FLARE-RA (HR 1.29 [95% CI 1.12 - 1.48]; p < 0.001), RA-FQ (HR 1.04 [95% CI 1.01 - 1.07]; p < 0.02), RAPID-3 (HR 1.11 [95% CI 1.05 - 1.17]; p < 0.001) and EQ5D Index (HR 0.01 [95% CI 0.00 - 0.17]; p = 0.001) scores was significantly predictive of subsequent flare in the multivariate model after adjustment for age and sex. Changes in PROs from baseline could not discriminate flare from remission patients at any point prior to time-of-flare (AUCs for all PROs ≤0.6). As expected, PRO increments at time-of-flare robustly distinguished flare from remission patients. This was most evident for the RA-FQ measure (AUC 0.87; 95% CI 0.80 - 0.94; p = 0.001), for which an increment of ≥ 5.5 confirmed objective flares with positive and negative predictive values of 81.4% (95% CI 71.7 - 88.2) and 87.3% (95% CI 77.3 - 93.3), respectively.

Changes in subjective elements of RA disease activity measured by PROs hold promise as remote monitoring tools for the early identification of flare amongst RA patients in remission following csDMARD cessation. These data inform strategies for the potential of remote monitoring of individuals in drug-free remission.

L. Gumber: None. F. Rayner: None. T. Bigirumurame: None. B. Dyke: None. A.R. Melville: None. S. Kerrigan: None. A. McGucken: None. N. Naamane: None. J. Pritchard: None. C.D. Buckley: None. A. Filer: None. I. McInnes: None. K. Raza: None. S. Siebert: None. J. Wason: None. W. Fai-Ng: None. A.E. Anderson: None. J.D. Isaacs: None. K.F. Baker: None. A.G. Pratt: None.