Validating treat-to-target endpoints in childhood lupus: data-driven sensitivity analyses from the UK JSLE cohort study.

Abstract

To conduct data-driven sensitivity analyses to evaluate whether refined definitions of childhood-onset systemic lupus erythematosus (cSLE) treat-to-target goals provide better protection against moderate-severe flares and new damage, compared with original consensus-derived targets. The UK JSLE Cohort Study was utilized. Childhood-SLE target attainment was determined at each visit. Removal or transformation of cSLE target criteria ('variations') were investigated, for Childhood Lupus Low Disease Activity State (cLLDAS), cSLE Clinical Remission on Steroids (cCR) and cSLE Clinical Remission off Steroids (cCR-0). The impact of such variations on the hazards of subsequent moderate-severe flare and new damage was assessed, using Prentice-Williams-Peterson (PWP) models. Two-sided t-tests compared the hazard ratios (HRs) obtained from the PWP gap-time models for the original and varied cSLE target definitions. Two variations of cLLDAS demonstrated significantly better protection against moderate-severe flare, including transformation of SLEDAI-2K cut-off to ≤3 (HR 0.13 [0.09, 0.19], P < 0.001); and transformation of PGA cut-off to ≤0.25 (HR 0.14 [0.10, 0.20], P < 0.001). These variations in cLLDAS did not impact on the hazards of new damage. No variations of cCR and cCR-0 led to a significant improvement in hazards of moderate-severe flare/new damage (all P > 0.05). A modified version of cLLDAS, combining these two transformations was also assessed, demonstrating further improvement in protection against moderate-severe flare (HR 0.12 [0.08, 0.17], P < 0.001). Refining the cLLDAS definition by lowering the SLEDAI-2K cut-off to ≤3 and PGA to ≤0.25 may enhance protection against moderate-severe flare, but not new damage. No variations of cCR or cCR-0 showed significant improvement.