Background In rheumatoid arthritis (RA) it is of major importance to distinguish non-responders to biological treatment to prevent a delay in effective treatment, potential side-effects and unnecessary healthcare costs. Prediction of response is a hot topic for years, yet predictive values remain modest. An explanation could be, that the measured response is not per definition a biochemically relevant change in disease activity. Recently, normal fluctuations in disease activity score (DAS28) were observed in RA patients that were stable for over 2 years on the same treatment (1). Based on these values, a cut-off of for DAS28 of ≥1.8 was considered significantly different. This implies that both moderate and good-responders according to EULAR response criteria (2), can still be explained by normal fluctuations. We investigated the predictive ability of proteomic biomarkers in addition to clinical parameters for prediction of response to biological treatment, comparing the different response criteria.
Objectives To investigate which proteomic biomarkers in addition to clinical parameters are predictive for response on biological treatment according to EULAR or DAS28-cd response criteria.
Methods In a dataset of 251 RA patients eligible for any biological treatment, serum was collected at baseline (before treatment) and analyzed on 57 inflammatory proteins using xMAP technology. EULAR response and DAS28-cd response were calculated after three months of treatment. After univariable analysis of both clinical markers and proteins, all with p<0,2 were used for logistic regression with backward selection procedure. All models were tested for predictive probabilities using C-statistics.
Results Clinical variables alone, were slightly predictive for EULAR response with an area under the curve (AUC) of 0,71 and reasonable for DAS28-cd response with an AUC of 0,78. For the more homogeneous group of naive TNF-alpha inhibitor starting patients (n=109), these values were 0,71 and 0,80 respectively. When proteomics were added in logistic regression and backward selection was performed again, the models almost remained as predictive. However for DAS28-cd the AUC was now increased from 0,80 to 0,86, which is highly predictive. For the AUC of 0,86, a proposed cut-off of resulted in a PPV of 85% and NPV of 78%.
Conclusions We showed that xMAP technology analysis is able to predict response to biologicals in RA patients. Predictive values are stronger in patients who are naïve for TNF-alpha inhibitor treatment. Proteomics in combination with clinical variables correlate better with the DAS28-cd score, possibly due to stricter classification of responders and non-responders.
Behrens F., Tony H., Altern R, Kleinert S., et al. Development and Validation of a New Disease Activity Score in 28 Joints–Based Treatment Response Criterion for Rheumatoid Arthritis. Arthritis Care & Research Vol. 65, No. 10, October 2013, pp 1608–1616.
van Gestel AM, Prevoo ML, van 't Hof MA, et al. Development and validation of the European League Against Rheumatism response criteria for rheumatoid arthritis.Comparison with the preliminary American College of Rheumatology and the World Health Organization/International League Against Rheumatism Criteria. Arthritis Rheum 1996; 39: 34 – 40.
Disclosure of Interest None declared