Definition of low disease activity state based on the SLE-DAS: derivation and validation in a multicentre real-life cohort

Objectives. To derive and validate a definition of low disease activity (LDA) for SLE based on the SLE Disease Activity Score (SLE-DAS), in a real-life multicentre cohort of SLE patients. Methods. Derivation was conducted using data from a monocentric cohort of SLE (Portugal), and validation was performed in a multicentre cohort (Italy, France and Spain). The Lupus Low Disease Activity State (LLDAS) was used as comparator. We applied receiver operating characteristics curve analysis against the LLDAS to determine the cut-off of SLE-DAS for LDA using bootstrap methodology. In a second step, we tested a definition of SLE-DAS LDA that included: (i) the statistically derived SLE-DAS upper threshold for LDA and (ii) prednisone dose <= 7.5 mg/day. In the multicentre validation cohort, we assessed the classification performance of this SLE-DAS LDA definition. Results. We included 774 patients, 300 in the derivation and 474 in the validation cohort. In the derivation cohort, the optimal cut-off to identify patients in LLDAS was SLE-DAS <= 2.48, presenting an area under the curve of 0.965 (95% CI 0.935, 0.994). When applied to the multicentre validation cohort, the SLE-DAS LDA definition showed a sensitivity of 97.1% and a specificity of 97.7% for LLDAS and an almost perfect agreement (Cohen's Kappa =0.933; P < 0.001). McNemar's test found no significant differences between the two definitions (P = 0 .092). Conclusion. The SLE-DAS LDA is a validated, accurate and easy-to-use definition for classifying SLE patients in LDA state.