Background/Purpose: The CRISS has been proposed as a composite outcome measure for trials in systemic sclerosis¹. CRISS is a 2-step process that assigns a probability of improvement for each individual patient on 0.00 [no improvement] to 1.00. Step 1 assesses clinically meaningful decline in cardio-pulmonary-renal involvement (based on consensus expert opinion) and assigns a probability of 0.00. For remaining subjects who do not experience a clinically meaningful decline, 5 variables are used over 12 months to calculate probability of improvement. These are: FVC%, mRSS, patient and physician global assessments (MD GA), and HAQ-DI. Scleroderma Lung Study-I compared oral daily cyclophosphamide (CYC) vs. placebo (PLA) in subjects with SSc-ILD ², which showed modest efficacy favoring CYC at 12-month but greater improvements in patient-reported outcomes. Our objective was to assess if CRISS index can discriminate CYC from PLA in the SLS-I over 12 months.

Methods: We used patient-level data for this analysis. For Step 2, MD GA was not assessed in the SLS-I. Therefore, linear regression model was developed to predict MD GA from variables which were similar between the original CRISS cohort and SLS-I. Stepwise regression (using maximum R² method) was used to identify the best prediction model for MD GA and applied to the SLS-1 data. We compared the CYC vs. PLA for the whole population and for the diffuse cutaneous SSc using the Wilcoxon rank-sum test for CRISS and Student’t t-test for individual variables. In addition, we explored the proportion of subjects who had a probability score ≥ 0.60¹ and compared the 2 groups using Chi-square.

Results: The mean (SD) baseline disease duration was 3.2 (2.1) years; 59% had diffuse cutaneous SSc. The prediction model for MD GA included the FVC (at baseline and 12 month change), mRSS (at baseline and 12 month change), gender, disease duration, and joint contractures ( R² of 0.44).

For Step 1, the definition of worsened cardio-pulmonary-renal involvement was met by 9 in CYC vs. 13 in PLA and were given a CRISS score of 0.0. For the remaining subjects, [60 in CYC and 52 in PLA], we applied Step 2. Using the CRISS as a continuous measure, CYC was statistically superior to PLA for the whole group (p=0.007) and diffuse SSc (p=0.0.02; Table). Lack of administration of MD GA in the trial is a limitation of this analysis.

Conclusion: In this post-hoc analysis using individual data from SLS-I, CRISS is able to discriminate CYC from PLA, supporting its validity in an independent cohort.

¹Khanna D. Arthritis Rheum. 2016

²Tashkin et al. N Engl J Med 2008

Table: Performance of individual components of Step 2 in CRISS and overall CRISS in SLS-1

*includes Step 1 and Step 2 probability scores

Negative score is improvement in MRSS, Patient Global, HAQ-DI, and Physician Global assessments and worsening in FVC% predicted

« Back to 2017 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/performance-of-the-american-college-of-rheumatology-provisional-composite-response-index-in-systemic-sclerosis-criss-in-the-scleroderma-lung-study-i/