Background/Purpose: The existing 1980 classification criteria for systemic sclerosis (SSc) are suboptimal for patients with early SSc and some patients with limited cutaneous SSc (lcSSc) where an important subset do not meet those criteria. Therefore, a joint EULAR and ACR committee was established to develop new classification criteria for SSc. These results are preliminary and not yet approved by the organizational sponsors.

Methods: Delphi exercises and a nominal group technique were used to create a set of potential items for classification of SSc.  The validity of the 23 items was tested in existing databases of SSc cases and controls. Next, twenty cases were prospectively collected to represent the spectrum of SSc (low probability to high probability) which were ranked by SSc experts and conjoint analysis (1000 Minds®) was used to assign weights to the items. This resulted in 17 items clustered in 13 domains with weights for the importance of each item. For the classification algorithm, experts agreed that all patients with sclerodactyly and scleroderma skin involvement proximal to the MCPs were considered SSc; patients with skin involvement due another scleroderma-like disorder (e.g. scleromyxedema) were excluded. A provisional threshold was established to classify definite SSc based on the sum of weights of 17 items.  To test the provisional algorithm, data on the items were prospectively collected in cases and controls in North America and Europe. To refine the threshold, a subset of cases (n=25) within the range of borderline probability of SSc was selected from the collected data. Experts were then asked to determine whether each case had ‘definite SSc’ or not, leading to a new threshold. During a face-to-face meeting of the Steering Committee, the 17 items were reduced to 9 while maintaining adequate sensitivity and specificity, tested in a random sample of cases and controls of 100 from North America and 100 from Europe (derivation sample). Cases of SSc and mimickers were collected at several sites in NA and Europe where approximately half of SSc had early disease. Weights were simplified by dividing each weight by 5 and rounding to the nearest whole number. The data were then re-analyzed in the remaining cases and controls (validation set; n=405). 

Results: The table provides the final items with the proposed weights. Only the maximum score in each domain is counted. A cut-off of >= 9 (out of 19) had a sensitivity of 91% and specificity of 92% in the validation cohort (n=405). The sensitivity and specificity of the former 1980 ARA criteria in this database were 75% and 72%, respectively. 

Conclusion: The preliminary results of the classification criteria for SSc performed better than 1980 Preliminary ARA Criteria for SSc. These criteria can be endorsed for epidemiological studies and clinical trials after approval by ACR and EULAR.

Table Preliminary Classification Criteria for Systemic Sclerosis

Disclosure: This work was sponsored by ACR and EULAR but has not been endorsed and is preliminary.