Gazing into the crystal ball? prospective mri score to predict negative eular-response in patients with rheumatoid arthritis (RA) before therapy-escalation to a biological therapy

Objectives: To define and validate a new MRI score for the prediction of negative response in patients with rheumatoid arthritis (RA) before therapy-escalation to a biological therapy. Methods: Remission plus was a multicentre (n=11) German “real-life” MRI study with the aim to implement MRI in diagnostic and therapy-control in patients with RA. MRI was performed before therapy-escalation (T0) and after 12 month (T1) and were analysed by using the OMERACT rheumatoid arthritis MRI score (RAMRIS). In addition clinical and laboratory parameters (DAS-28 and CRP) were collected for each visit. The new MRI score was developed using a logistic regression and is a combination of the patients' age and the RAMRIS-T0 score. In the logistic regression we adjusted for possible confounders. Performance of the score is assessed in terms of area-under-the-curve (AUC). To validate the new score, a bootstrap analysis is performed with 5000 resamplings. Results: Overall 257 patients were investigated with MRI, while 29 (20 female, median age 57 (IQR 46 - 65) years, 95% anti-TNF-alpha therapy) of them were escalated to a biological therapy. Poor responders (n=5) and non-poor responders (n=24) had a mean RAMRIS-T0 score of 14.4 and 52.0, respectively (Wilcoxon test p<0.01). High RAMRIS-score seemed protective against non-response. The odds ratio for an increase of RAMRIS by one point was 0.90 (95% confidence interval 0.79-1.03, p=12%). The strength of the association was stable after adjusting for age, CCP, CRP, rheumatoid-factor (RF) and DAS28 at baseline. The median AUC in the bootstrap analysis is 88.9% with IQR 84.0-92.8%. Conclusions: RAMRIS-T0 possibly has the potential to predict poor EULARresponse in patients with RA before initiation of a biological therapy. By using the new score the predictive value was (AUC up to 95%) surprisingly high. Due to its high predictive value this score could be a possible tool in personalized medicine concepts after performing follow-up studies in clinical trials.