POS0762 CAN THE SLE-DAS SUBSTITUTE BILAG TO MEASURE LUPUS DISEASE ACTIVITY IN CLINICAL TRIALS? POST-HOC ANALYSIS OF THE BLISS-76 TRIAL

Abstract

Background:The primary endpoint for randomized clinical trials (RCT) in Systemic lupus erythematosus (SLE) is usually defined as proportion of responders in a composite index. The most widely used are British Isles Lupus Assessment Group (BILAG)-based Composite Lupus Assessment (BICLA) and Systemic Lupus Responder Index (SRI). Both comprise BILAG along with SLE Disease Activity Index (SLEDAI). BICLA and SRI are complex and time-consuming to assess.The SLE Disease Activity Score (SLE-DAS) is an easy to apply, validated, continuous disease activity measure, highly correlated with SLEDAI, with higher accuracy and sensitivity-to-change as compared to SLEDAI1.We hypothesize that SLE-DAS can also identify the SLE disease activity information from BILAG, thus dispensing the use of composite indexes for RCT.Objectives:To compare the ability of the SLE-DAS and the Safety of Estrogen in Lupus National Assessment (SELENA)-SLEDAI to discriminate between BILAG classification of mild vs. moderate vs. severe disease activity.Methods:Post-hoc analysis of all intention-to-treat patients in the BLISS-76 (NCT00410384) RCT at the baseline study visit. SELENA-SLEDAI and BILAG were assessed at time of the study visits and SLE-DAS was retrospectively scored from the study database. Patients’ disease activity was classified as: (i) mild (no BILAG B or A scores in any organ domain); (ii) moderate (1 BILAG B, no A scores); (iii) severe (≥2 BILAG B and/or ≥1 BILAG A). Ability of the SLE-DAS and SELENA-SLEDAI to differentiate between: (i) mild vs. moderate/severe disease activity; (ii) mild/moderate vs. severe disease activity (according to BILAG), were evaluated using receiver operating characteristic (ROC) analysis. The area under the ROC curves (AUCs) with 95% confidence intervals (95%CI) as a measure of discriminatory ability of the SLE-DAS and SELENA-SLEDAI were compared using Delong’s test for two correlated curves. Because AUC measurements might have restricted accuracy for imbalanced datasets, precision-recall (PR) curves and area under PR curves (AUC-PR) were also performed. Statistical significance was set at 0.05.Results:We included 819 patients, classified by BILAG as presenting mild (7.7%), moderate (28.8%) or severe (63.5%) disease activity. To differentiate mild vs. moderate/severe disease activity, the discriminatory ability of SLE-DAS was outstanding (AUC 0.948; 95%CI 0.923-0.973), while that of SELENA-SLEDAI was acceptable (AUC 0.729; 95%CI 0.657-0.801) (p<0.005) (figure 1A). To differentiate mild/moderate vs. severe disease activity, the discriminatory ability of SLE-DAS was excellent (AUC 0.873; 95%CI 9.846-0.899), while that of SELENA-SLEDAI was acceptable (AUC 0.707; 95%CI 0.670-0.744) (p<0.005) (figure 1B). The AUC-PR confirmed the higher performance of SLE-DAS over SELENA-SLEDAI to differentiate mild vs. moderate/severe disease activity (0.995 vs. 0.965, respectively) (figure 1C) and mild/moderate vs. severe disease activity (0.902 vs. 0.794, respectively) (figure 1D).Figure 1.Receiver operating characteristics (ROC) curves comparing the ability of the SLE-DAS and SELENA-SLEDAI to differentiate (A) mild vs. moderate/severe disease activity and (B) mild/moderate vs. severe disease activity, as assessed by BILAG; and Precision-recall (PR) curves comparing the performance of the SLE-DAS and SELENA-SLEDAI to differentiate (C) mild vs. moderate/severe disease activity and (D) mild/moderate vs. severe disease activity.Conclusion:The SLE-DAS presents excellent performance in assessing SLE disease activity categorized by BILAG scores, which is not the case for SELENA-SLEDAI. Further studies will aim to better define ability of SLE-DAS to substitute composite responder indices.