POS0338 DEVELOPMENT AND VALIDATION OF RANKED COMPOSITE IMPORTANT DIFFERENCE (RCID) SCORE IN DIFFUSE CUTANEOUS SYSTEMIC SCLEROSIS

• Published in: Annals of the rheumatic diseases Vol. 82; no. Suppl 1; p. 416
• Main Authors: Del Galdo, F., Bissell, L.A., Huang, S., Hansen, P., Johnson, S., Furst, D., Khanna, D.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier B.V 01.06.2023; Elsevier Limited
• Subjects: Clinical trials; Conjoint analysis; Patient-led research; Patients; Placebos; Randomized control trial; Scleroderma; Statistical analysis; Surveys; Systemic sclerosis; Patient-led research; Randomized control trial; Systemic sclerosis
• ISSN: 0003-4967; 1468-2060
• DOI: 10.1136/annrheumdis-2023-eular.2638

The revised Combined Response Index in diffuse cutaneous Systemic Sclerosis (dcSSc) (rCRISS) is a composite outcome measure relying on 6 core set variables including Organ Failure, Forced Vital Capacity (FVC), HAQ-DI and modified Rodnan skin score (mRSS), together with patient and clinician global assessments of disease activity [1]. rCRISS is calculated in patients not experiencing any organ failure and reported as proportion of patients who improve a given % (20,30,40 etc) in at least 3 domains. rCRISS discriminated treatment vs placebo in 3 randomised, placebo-controlled trials on early diffuse SSc [1]. Notwithstanding, it is not clear what is the minimal rCRISS score that captures an overall improvement in disease. Further, being developed as a response index, patients that do not experience a predefined organ failure but do worsen in their disease cannot be scored or discriminated from patients that simply do not improve. To improve the clinical interpretation of the rCRISS by creating a continuous ranked score of clinician and patient meaningful changes in its individual measures. Following OmerACT guidelines, 100 physicians with experience in dcSSc and 100 patients with dcSSc who have participated in a dcSSc trial were identified in 5 continents. An adaptive survey was developed, based on 1000Minds conjoint analysis software. Patients and doctors were asked to choose, on successive pairs of two hypothetical patients, who had better or worse outcome according to presence of organ failure and Minimally Clinical Important Differences (MCID) in two domains among FVC, HAQ-DI and mRSS. Pairwise comparisons of choices were analysed utilising PAPRIKA methodology [2] to rank and weight the MCIDs against each other. The resulting score was tested in the same cohort utilised for rCRISS development. Briefly, 354 SSc patients data from three randomized, placebo-controlled trials (Asset, Foccussced and Fasscinate) were randomised in 10 development (2/3 pts) and validation (1/3 pts) sets, stratified by study and treatment group. Mean scores of the replicate datasets were calculated for active and placebo treatments. Bootstrapping was employed to determine 95% Confidence Intervals. Kappa statistics were used to analyse concordance. 160 of the 200 participants completed the survey. Discrete choices conjoint analysis defined ranks and relative weights of the 4 domains in the improved and worsened outcomes groups. A continuous composite ranked score reflecting the relative weighting of the individual outcome measures (Ranked Composite Important Difference, RCID) was developed. The score ranges from -1 (worst possible outcome) to 1 (best possible outcome), with patients who experience no organ failure and do not meet any MCID in any of the 3 domains scoring 0. Positive RCID scores showed a strong correlation with ACR CRISS >0.6 with kappa statistics in development and validation sets of 0.71 (CI 0.64,0.73) and 0.71 (0.66,0.84), respectively. 73.1% of subjects with positive RCIDs (74% in the validation set) met revised CRISS10, with 63.7 and 51.3 meeting rCRISS 20% and 30%, respectively. rCRISS scores between 20 and 30 were associated with only positive RCIDs. Additionally, 32% (C.I 29.7,34.3) of subjects on active treatment showed a positive RCID (>0) vs 25.8 (CI: 23.3,26.6) in the placebo group. Data were confirmed in the validation set (31.7 vs 27.1). Conversely, 12.3% (10.7,14) of subjects on treatment arms showed negative RCID (Any worsening, <0) vs 21.1 (19.2,23.3) in the placebo groups. Our process adopted a robust methodology with a large patient representation in outcome measure development, to develop a patient and physician derived anchor (RCID) to other composite scores, such as ACR CRISS and revised CRISS. RCID provides a clinician and patient meaningful score that can be used as a dichotomous variable with a positive (improved) or negative (worsened) outcome. [1]Khanna D, Huang S, Lin CJF, et al. Ann Rheum Dis 2020-219100 [2]Hansen P et al Multi-Criteria Decis. Anal 2008) NIL. Francesco Del Galdo Speakers bureau: AstraZeneca, Boehringer Ingelheim,Janssen, Consultant of: AstraZeneca, Boehringer Ingelheim, Capella, Chemomab, Ergomed, Janssen, Mitsubishi Tanabe, Grant/research support from: Abbvie, AstraZeneca, Boehringer Ingelheim, Capella, Chemomab, Janssen, Mitsubishi Tanabe, Lesley-Anne Bissell Consultant of: Ucb, Abbvie, Galabagos, Suiyuan Huang: None declared, Paul Hansen: None declared, Sindhu Johnson: None declared, Daniel Furst: None declared, Dinesh Khanna Shareholder of: Eicos Sciences, Speakers bureau: Bayer, Bristol-Myers Squibb, Boehringer Ingelheim, Cytori, CSL Behring, EMD Merck-Serono, Genentech/Roche, GlaxoSmithKline, Genkyotex, Sanofi-Aventis, UCB, Actelion, and Gilead;, Consultant of: Bayer, Bristol-Myers Squibb, Boehringer Ingelheim, Cytori, CSL Behring, EMD Merck-Serono, Genentech/Roche, GlaxoSmithKline, Genkyotex, Sanofi-Aventis, UCB, Actelion, and Gilead; has stock options in.