Background/Purpose:   Preliminary definitions of improvement (DOIs) for juvenile dermatomyositis (JDM) combining IMACS or PRINTO core set activity measures (CSMs) have been used as a responder index in clinical trials.  However, these DOIs defined only minimal clinical improvement and are in need of further validation.

Methods:   Using natural history study data, JDM experts rated patient profiles containing IMACS or PRINTO CSMs and achieved consensus ( ≥ 70% agreement) in 98% (247 minimal, 174 moderate and 84 major improvement).  Conjoint analysis was performed on forced-choice surveys (by using 1000Minds software) administered to myositis experts. Candidate DOIs based on changes in core set measures (CSMs) were generated as follows: A) 23 published DOIs were retested; B) 436 DOIs developed using expert survey and variations of published DOIs; C) 56 DOIs derived from logistic regression analysis; D) 6 DOIs derived from a conjoint analysis survey that yielded scores with different levels of improvement in different CSMs; E) 8 DOIs drafted by combining changes in each CSM with respective conjoint analysis weights; and F) 194 DOIs drafted by applying weights from conjoint analysis to the base DOIs. Relative and absolute % change DOIs were tested.  The consensus patient profiles were then used to test DOIs for their validity, including sensitivity, specificity, kappa, OR and area under the curve (AUC). High performing DOIs were externally validated using Rituximab in Myositis (RIM) trial data (N=48) and PRINTO JDM (n=139) trials. 

Results: The 14 highest performing candidate DOIs with AUC ≥ 0.90 in profile data for minimal, moderate and major improvement, and comparable results in the RIM trial, were discussed among 12 JDM experts for their performance characteristics and clinical face validity at a consensus conference using Nominal Group Technique (NGT).  Many DOIs had significant discriminant validity in the PRINTO trial.  A final ranking of the pediatric experts’ top DOIs yielded 92% consensus for a conjoint analysis hybrid model DOI using absolute % change in CSMs with different cut points for minimal, moderate and major improvement (Table 1).  NGT discussion with the pediatric and adult working groups yielded consensus (91% agreement) in use of this hybrid DOI for adult DM/PM and JDM.

Conclusion:   A conjoint analysis-driven hybrid definition with a continuous score of improvement based on absolute % change in CSMs with different cut points for minimal, moderate and major improvement was selected by a data- and consensus-driven process as a final DOI to be used for JDM and adult DM/PM clinical trials.  ACR and EULAR funding support was received for this project, and their approval will be sought for these as new criteria for clinical response.

                      Table 1