SAT0170 COMPOSITE OF RELEVANT ENDPOINTS FOR SJÖGREN’S SYNDROME (CRESS)

Abstract

Background:Defining a primary study endpoint that is able to discriminate between active treatment and placebo is crucial for clinical trials in primary Sjögren syndrome (pSS). Recent trials used the validated ESSDAI as primary endpoint, but found large ‘response rates’ in the placebo group too. Since pSS is a very heterogenous disease, a composite endpoint including multiple aspects (i.e., systemic, patient-reported, functional and biological) may be more appropriate to demonstrate clinical efficacy.Objectives:To develop a composite endpoint for pSS based on expert opinion and analysis of trial data.Methods:Based on expert opinion, 5 items were found to be most relevant to assess the effect of treatment in pSS patients: ESSDAI, ESSPRI, OSS, SWS and RF/IgG (Figure 1). These items were tested using data at week 24 of the randomized, double blind, placebo-controlled ASAP-III trial.1ROC analysis was used to assess the discrimination of effect between the abatacept (n=40) and placebo (n=39) treatment groups. The optimal cut-off point per item was defined by the highest sum of sensitivity and specificity. The percentage of patients responding to the individual items (Figure 1) and the composite endpoint (named CRESS) was calculated.Results:For ESSDAI, ROC analysis showed that both absolute and relative change in ESSDAI were not able to discriminate between treatment groups (AUC 0.536 and 0.559) and no optimal cut-off point could be identified. According to an in SLE developed endpoint and based on expert opinion, it was decided to aim for the validated definition of low disease activity (ESSDAI<5)2.For ESSPRI, ROC analysis (AUC 0.629) showed an optimal cut-off point of -13.8%. Therefore, the validated definition of ESSPRI response (≥-15% or 1 point)2was used. For OSS and SWS, ROC analysis (AUC 0.555 for OSS>3 at baseline and AUC 0.556 for SWS>0 at baseline) could not identify an optimal cut-off point, so the definitions based on expert opinion were kept (Figure 1).For serological items, ROC analysis (AUC 0.861 for RF>0 at baseline and 0.615 for IgG) showed optimal cut-off points of -23% and -2.2%, respectively. It was decided to round these numbers to ≥25% decrease in RF or ≥5% decrease in IgG. Responding to ≥3 of the 5 items discriminated best between the abatacept and placebo groups. The final response rate to our composite endpoint (CRESS responders) was 55% vs. 13% in the abatacept and placebo groups, respectively (P<0.001). Further analysis of how many patients who met the composite endpoint also met the single endpoints and vice versa demonstrated that all individual items contributed to the overall response rate.Conclusion:This concept of the new ‘Composite of Relevant Endpoints for Sjögren’s Syndrome’ (CRESS) is developed. With this composite endpoint, it is possible to discriminate between abatacept and placebo response in pSS patients. Additional validation analyses in independent, global, multi-center, placebo-controlled trials of biological DMARDs in pSS and NECESSITY will be performed.