Detailed performance characteristics of patient profiles and clinical trial data for the top 5 candidate response criteria definitions presented at the consensus conference*
| Profiles (n=270)† | RIM trial (n=147) | |||||||
|---|---|---|---|---|---|---|---|---|
| Candidate definitions for response criteria, improvement category, core set measure | Sensitivity, % | Specificity, % | Threshold AUC | Total AUC | Candidate definition, improved physician's rating‡ | Candidate definition, not improved physician's rating‡ | p Value | Rank |
| Conjoint analysis absolute % change (model 3)§ | 1 | |||||||
| Minimal (improvement score ≥20) | 85 | 92 | 0.89 | 0.96 | 2.0 | 4.0 | <0.001 | |
| Moderate (improvement score ≥40) | 90 | 96 | 0.93 | 0.99 | 2.0 | 3.0 | <0.001 | |
| Major (total improvement score ≥60) | 92 | 98 | 0.95 | 1.00 | 2.0 | 3.0 | <0.001 | |
| Conjoint analysis relative % change (model 1)¶ | 2 | |||||||
| Minimal (improvement score ≥33) | 94 | 90 | 0.92 | 0.98 | 2.0 | 4.0 | <0.001 | |
| Moderate (improvement score ≥55) | 93 | 93 | 0.93 | 0.99 | 2.0 | 3.0 | <0.001 | |
| Major (improvement score ≥70) | 100 | 95 | 0.97 | 0.99 | 2.0 | 3.0 | <0.001 | |
| Conjoint analysis relative % change (model 2)¶ | 3 | |||||||
| Minimal (improvement score ≥30) | 94 | 92 | 0.93 | 0.98 | 2.0 | 4.0 | <0.001 | |
| Moderate (total improvement score ≥45) | 94 | 88 | 0.91 | 0.98 | 2.0 | 3.0 | <0.001 | |
| Major (improvement score ≥65) | 100 | 98 | 0.99 | 1.00 | 2.0 | 3.0 | <0.001 | |
| Weighted core set measure relative % change** | 4 | |||||||
| Minimal (improvement score ≥100) | 92 | 91 | 0.91 | 0.97 | 2.0 | 3.0 | <0.001 | |
| Moderate (improvement score ≥250) | 94 | 91 | 0.93 | 0.98 | 2.0 | 3.0 | <0.001 | |
| Major (improvement score ≥400) | 100 | 94 | 0.97 | 1.00 | 2.0 | 3.0 | <0.001 | |
| Logistic regression relative % change†† | 5 | |||||||
| Minimal (improvement score ≥75) | 89 | 93 | 0.91 | 0.97 | 2.0 | 3.0 | <0.001 | |
| Moderate (improvement score ≥150) | 94 | 88 | 0.91 | 0.98 | 2.0 | 3.0 | <0.001 | |
| Major (improvement score ≥300) | 100 | 96 | 0.98 | 1.00 | 2.0 | 3.0 | <0.001 | |
*Online supplementary table S2 (available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40064/abstract) shows definitions 6–18 from the consensus conference ratings. The threshold AUC was calculated as the AUC from the ROC curve for the total improvement score and the threshold for minimal, moderate, and major improvement. The total AUC was calculated as the AUC from the ROC curve, using the total improvement score and the threshold cutoffs for minimal, moderate, and major improvement, and applies only to continuous definitions.
†The reference standard for sensitivity and specificity was myositis expert consensus rating of improvement.
‡Physician's rating is the treating physician's rating on a Likert scale of 1–7, where lower scores represent a greater degree of improvement, at week 24 of the RIM trial.3 A 1-point difference in the physician's rating of improvement from no improvement to minimal improvement was considered not only statistically significant but also clinically significant.
§Conjoint analysis–based continuous candidate response criteria using absolute per cent change in core set measures (absolute per cent change model) is shown in table 3. These criteria are also the top response criteria for juvenile DM, but with different thresholds in the total improvement score for minimal, moderate and major improvement.20
¶Conjoint analysis–based continuous candidate response criteria using relative per cent change in core set measures are shown in online supplementary table S3 (available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40064/abstract). These criteria are also the second- and third-choice criteria for juvenile DM, but with different thresholds in the total improvement score for minimal, moderate, and major improvement.20
**The total improvement score is calculated as 2×(MD global % change)+(patient global % change)+3×(MMT % change)+1.5×(HAQ % change)+1.5×(extramusc % change)+(enzyme % change).
††The total improvement score is calculated as (MD global % change)+(patient global % change)+(MMT % change)+(HAQ % change)+(extramusc % change)+enzyme % change).
AUC, area under the curve; DM, dermatomyositis; extramusc, extramuscular; enzyme, most abnormal serum muscle enzyme value among aldolase, alanine aminotransferase, aspartate aminotransferase, lactate hydrogenase, and creatine kinase; HAQ, Health Assessment Questionnaire; MD global, physician global activity; MMT, manual muscle testing; patient global, patient global activity; RIM, Rituximab in Myositis; ROC, receiver operating characteristic.