545 Validation of the cutaneous features in the classification of dermatomyositis

Recognizing the need to improve the classification of patients with dermatomyositis (DM), international experts in the field of rheumatologic dermatology aimed to create criteria that are valid and more reliable to classify patients with DM. The DM Delphi Project produced 23 provisional clinical and laboratory items after an extensive literature search, three rounds of consensus exercises, and nominal group discussions. Local Institutional Review Board approval and data transfer agreements were secured by participating sites. Consent was obtained from participants prior to the assessment. DM cases included newly diagnosed, treatment-naïve patients, as well as those with mild DM-specific cutaneous features. Controls were lupus, eczemas, psoriasis, and other inflammatory skin diseases. Descriptive statistics including counts and percentages of each criterion overall and by diagnosis were calculated and compared using the chi-square test. A total of 235 subjects were enrolled from 13 international sites in North America, Europe, and Asia. The investigators classified 147/235 subjects as DM (62%), whereas 105/235 subjects were DM (45%) using the EULAR/ACR calculator. 15 of the 23 Delphi items were found to have a high distinguishing ability in classifying DM (p<0.05). Logistic regression of the 15 items created a model that is robust in classifying DM (AUC=0.91) with papules and macular erythema on joints, location on elbow, knee and upper chest, as well as presence of myositis-specific antibody having statistically significant odds ratio (p<0.05). Forward stepwise analysis will be done to generate several candidate models. A clinically relevant final model will be converted into a nomogram or points-based scoring system for implementation in clinical settings. The results will be used as stand-alone criteria in skin-predominant DM and as source of items in a larger validation effort to improve the current myositis criteria.