Evaluation of American College of Rheumatology Provisional Composite Response Index in Systemic Sclerosis (CRISS) in the FaSScinate Trial D Khanna 1, V Berrocal 1, C Denton 2, A Jahreis 3, H Spotswood 4, C Lin 3, J Siegel 3 and D Furst 5, 1 University of Michigan, Ann Arbor, MI, 2 University College London, Royal Free Hospital, London, United Kingdom, 43 Genentech, South San Francisco, CA, 4 Roche Products Ltd., Welwyn Garden City, United Kingdom, 5 University of California, Los Angeles, David Geffen School of Medicine, Los Angeles, CA
Disclosures Dr Furst has consultancy relationships with Actelion,Genentech/Roche,Amgen, Corbus,Cytori,Pfizer.Investigator funding from Amgen,Corbus, NIH.
Why a combined response index for Systemic Sclerosis (SSc)? SSc has validated outcomes for individual manifestations SSc is a multisystem disorder Trials have largely been negative or modestly positive Inappropriate pharmacologic targets or poor outcome measures? Combined indices successful in other complex systemic rheumatic diseases: RCTs FDA/EMA approval of new therapies RA-ACR/DAS: joint count, HAQ, APRs, all separate measures SLE-SELENA-SLEDAI: SLEDAI, PhyGA, medication change Vasculitis: BVAS versions are single measures of overall disease We seek approaches to treat the underlying disease; thus, we want measures of overall disease
Consensus and data-driven methodology Delphi exercise and nominal group consensus meeting to select core items Assessment of psychometric properties of core items using 2 longitudinal dcssc cohorts and consensus meeting to select items for patient profiles Development and rating of patient profiles by experts Development of candidate definitions for response and assessment for performance Selection of top indices based on statistical performance and ranking by experts using OMERACT attributes Test in prospective trial
2-Step Process Cardio-pulmonary-renal Worsening? Yes No CRISS Score=0.00 PROCEED to Step 2
STEP 1: Expert consensus on definition of a patient who is not-improved during a trial Patient is considered not improved* if he/she develops ANY ONE OF THE FOLLOWING DURING THE TRIAL** RENAL: New scleroderma renal crisis ILD: Decline in forced vital capacity (FVC)% predicted 15% (relative) in established ILD and FVC% predicted below 80% predicted* HEART: New onset of left ventricular failure (defined as left ventricular ejection fraction 45%) requiring treatment* CARDIOPULMONARY: New onset of pulmonary arterial hypertension (PAH) on right heart catheterization requiring treatment*. *Irrespective of improvement in other core items ** Attributable to SSc
STEP 2 Step 2 involves computing the predicted probability of improving for each patient using a probability equation that includes change in modified Rodnan skin score, FVC%, HAQ-DI, patient global assessment, and physician global assessment. All changes are absolute change (Time 2 Time baseline ).
CRISS score Hypothetical example CRISS calculates the probability of improvement on 0.00 to 1.00 scale for each individual patient where 0.00 is no improvement and 1.00 is marked improvement Probability=0.0-0.2 Placebo or ineffective drug Probability> 0.40 Effective drug
48 week double-blind, placebo-controlled trial of early dcssc 5 years disease duration 15 and 40 mrss units Elevated acute-phase reactants 1:1 randomization Khanna D, et al. Lancet. 2016;387(10038):2630-2640.
Statistical analysis Step 1 for CRISS was captured using review of serious adverse event data. Step 2 calculated the CRISS index using the probability equation. Non-parametric Wilcoxon test and proportion z-tests were used to assess whether significant differences exist between the CRISS score in both its continuous and binary form in TCZ and PBO. Analyses utilized all subjects, with missing data imputed using the best-fitting linear mixed model determined through Chi-square significance test on the deviance. The linear mixed models selected included a subject-specific random intercept, an indicator for treatment and a function of time.
Change From Baseline in mrss at Week 24 (Primary End Point) & Week 48 (Secondary End Point) 1.00 0.00 PBO 162 mg QW SC TCZ 162 mg QW SC Change (95% CI) From Baseline in mrss 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0 3 8 16 24 32 40 48 Weeks Week 24 Week 48 PBO 162 mg QW SC 1.2 ΔLSM 2.8 ΔLSM (TCZ-PBO): 2.70 (TCZ-PBO): 3.6 TCZ 162 mg QW SC 3.9 p = 0.09 6.3 p = 0.06 CI, confidence interval; LSM, least square means; ΔLSM (TCZ-PBO), difference in LSM between treatment arms. Negative change indicates improvement. Means and 95% CI are from the repeated-measures model.
Cumulative Distribution by Absolute Change in % pfvc From Baseline to Week 48 Worsening [decrease in % pfvc ] No Decline [increase in % pfvc ] 100 80 PBO 162 mg QW SC (n = 31) TCZ 162 mg QW SC (n = 30) p = 0. 0373 ( Van Elteren) 84% Patients, % 60 40 57% 20 23%* 0 10%* 40 30 20 10 0 10 20 % pfvc Change From Baseline pfvc, predicted forced vital capacity. *Percentage of patients with 10% worsening in % pfvc. 1
2-Step Process Cardio-pulmonary-renal Worsening? Yes: 4 in placebo* 0 in TCZ No CRISS Score=0.00 PROCEED to Step 2 * 1 each developed scleroderma renal crisis, heart failure, worsening ILD and PAH
Imputed data in fasscinate study (week 24) TCZ group Mean (SD) N=43 PBO group Mean (SD) N=40 P-value* mrss, 0-51* -4.20 (6.9) -2.65 (6.0) 0.28 FVC% predicted -0.66 (4.6) -4.20 (2.7) <0.00005 24 Weeks HAQ-DI, 0-3* 0.17 (0.60) 0.18 (0.44) 0.93 Patient global assessment, 0-10* -0.36 (1.78) -0.06 (2.31) 0.51 Physician global assessment, 0-10* -1.92 (2.62) -1.79 (2.47) 0.82 *Negative is improvement
Imputed data in fasscinate study (week 48) TCZ group Mean (SD) N=43 PBO group Mean (SD) N=40 P-value* mrss, 0-51* -5.26 (7.18) -3.0 (5.76) 0.12 FVC% predicted -2.21 (2.33) -6.50 (3.48) <0.00001 48 Weeks HAQ-DI, 0-3* 0.15 (0.69) 0.23 (0.50) 0.53 Patient global assessment, 0-10* -0.85 (2.22) -0.36 (2.26) 0.33 Physician global assessment, 0-10* -3.18 (2.62) -1.88 (2.74) 0.03 *Negative is improvement
CRISS scores at 24 and 48 weeks CRISS, median [IQR] at 24 wks, Imputed data CRISS, median [IQR] at 48 wks, Imputed data CRISS, median [IQR] at 24 wks, Completers only TCZ PBO** P value 0.19 [0.006; 0.92], N=43 0.32 [0.01; 0.93], N=43 0.23 [0.003; 0.95], N=33 0.0006 [0.0001; 0.13], N=40 0.001 [0.0002; 0.16], N=40 0.008 [0.00; 0.10], N=36 0.01* 0.002* 0.04* CRISS, median [IQR] at 48 wks, Completers only 0.31 [0.008; 0.99], N=27 0.001 [0.00; 0.18], N=35 0.01* *Using Wilcoxon test as CRISS data is not normally distributed ** there are 4 subjects in the PBO who met step 1 and were given a score of 0.0. One patient each with heart failure, Worsening ILD, SRC, and PAH.
CRISS scores vs. individual components at week 48 0-3 Individual Measures mrss FVC Pt. Glob MD Glob -0.85-0.36-2.21-1.88-3 -3.18 0.4 0.3 p=0.01 0.31 CRISS p=0.002 0.32 TCZ Placebo -6-9 -5.26 TCZ Placebo -6.5 P=0.12 P=0.0001 P=0.33 P=0.0.03 HAQ-DI:TCZ (0.15) vs. Placebo (0.23) p=0.53 0.2 0.1 0 CRISS completers 0.001 0.001 CRISS imputed Conclusion: CRISS works best, although FVC and mrss also work
CRISS score in TCZ at 48 week Placebo TCZ
Correlation coefficients between the CRISS score and individual variables in CRISS 24-weeks 48-weeks Change in mrss Change in FVC% TCZ PBO TCZ PBO -0.76* -0.84* -0.75* -0.73* 0.06 0.11 0.16 0.16 Change in patient global -0.22-0.17-0.41* -0.005 assessment Change in physician global -0.45* -0.24-0.26-0.35* assessment Change in HAQ-DI -0.49* -0.26-0.58* -0.43* *p<0.05 using Pearson correlation coefficient
Conclusion In this post-hoc analysis from fasscinate trial, CRISS discriminated TCZ from PBO, supporting its validity in an independent cohort from its development cohort. Although mrss had the highest correlation with CRISS score, pt global assessment, physician global assessment, and HAQ-DI had significant correlations. Lack of correlation with FVC% may be related to nonenrichment of ILD in this trial. In the SLS-I (abstract#726), the correlation between CRISS and FVC% ranged between 0.41-0.51 (<0.05). Current work is ongoing to define clinically important differences for CRISS.
Funded by NIH/NIAID and NIH/NIAMS