SAT0651 GOOD DISCRIMINATIVE ABILITY OF THE SIMPLE EROSION NARROWING SCORE COMPARED TO THE SHARP/VAN DER HEIJDE SCORE

Abstract

Background The Simple Erosion Narrowing Score (SENS), a simplification of the Sharp/van der Heijde score (SHS), has been recommended for use in clinical practice because it combines a simpler, less time-consuming scoring method with measurement properties comparable to the SHS [1]. The SENS assesses both joint space narrowing (JSN) and erosions in the same joints as the SHS, but the SENS sums the number of joints with erosions and the number of joints with JSN without grading per joint. However, some studies have suggested that the SENS is less sensitive to radiographic progression than the SHS. Further erosion of already eroded joints added to the discriminative ability of the SHS in the COBRA trial, suggesting some discrimination might be lost when using the SENS, especially in more advanced RA populations [2]. This is an important issue, as time savings using SENS are substantial (7 instead of 25 minutes [1,3]), but contrast between interventions is in general lower than in the early years of RA research as the standard of care improves and trials are more and more characterized by active controls and disease activity guided treatment strategies that tend to converge. Objectives To assess whether the SENS can discriminate between treatment groups when the treatment contrast is low, using data from the DRESS (Dose REduction Strategy of Subcutaneous TNF inhibitors) trial Methods The DRESS study (Dutch trial register, NTR 3216, CMO region Arnhem-Nijmegen, NL37704.091.11) is an open label non-inferiority randomised controlled trial in which RA patients with low disease activity on a stable adalimumab or etanercept dose were randomised 2:1 to disease activity guided tapering or usual care [4]. Radiographs of hand and feet at baseline and the end of follow up (18 months) were assessed independently in chronological order by 2 blinded trained readers using the SHS. The SENS was determined based on the individual joint scores of the SHS. For both scores, the mean of the two readers was used to minimise measurement error. In order to compare the discriminative ability of both scores, we repeated the analyses from the DRESS study using the SENS and compared the results. In addition, we performed bootstrapping (1000 replications) to investigate the consistency of the results. Results Table 1 shows baseline characteristics of the study population. A significant but clinically unimportant mean increase (95% CI) in radiographic progression of 0.60 (0.16-1.0) points SHS in the dose reduction group was found in the original study. Using the SENS score, this analysis shows a difference of 0.33 (0.11-0.54) points. In relation to the maximum score in each method (SHS: 448; SENS: 86), these changes are comparably low (SHS: 0.13% (0.04%-0.23%); SENS: 0.38% (0.13%-0.63%)). In bootstrapping replications of the DRESS study, using a T-test, the SHS detected a significant difference in progression in 86,0% of replications, the SENS did so in 91,3%. Conclusion The SENS and SHS were equally able to detect a difference in radiographic progression, even with a very low treatment contrast between the two groups and in this relatively advanced RA population.