OP0015 THE ROLE OF MULTI-CRITERIA DECISION ANALYSIS IN THE DEVELOPMENT OF CANDIDATE CLASSIFICATION CRITERIA FOR ANTISYNTHETASE SYNDROME: ANALYSIS FROM THE CLASS PROJECT (2024)

Abstract

Background: Previously proposed antisynthetase syndrome (ASSD) criteria lack validation and broad consensus. The Classification Criteria of Anti-Synthetase Syndrome (CLASS) project working team aims to develop data- and expert consensus-driven candidate classification criteria for ASSD. This work will focus on the expert consensus-driven candidate criteria, which was based on multi-criteria decision analysis (MCDA), a technique that is based on the interpretation of the human decision-making process as an intuitive weighting of trade-offs between alternatives.

Objectives: Our main objective was to develop and evaluate the performance of MCDA-based candidate classification criteria for ASSD.

Methods: A list of variables associated with ASSD was developed using a systematic literature review and then refined into ASSD key domains and variables list by international myositis and interstitial lung disease (ILD) experts. This list was used to create preferences surveys where experts were presented with a series of pairwise comparisons of fictional clinical vignettes with only two different variables. Assuming all other aspects were the same, experts were asked to select the clinical vignette that was more likely to represent an ASSD case. Expert’s answers were analysed using the Potentially All Pairwise RanKings of All Possible Alternatives (PAPRIKA) method to determine the weights of the key variables to formulate the MCDA-based classification criteria. Clinical vignettes scored by the experts as consensus cases or controls were used to test the performance of candidate classification criteria using receiver operating characteristic curves. The best cut-off for the candidate criteria was defined as simultaneously having >80% sensitivity and specificity and the highest Youden index.

Results: The experts recommended that exclusion criteria should be applied before using the classification criteria (Table 1). Patients with a non-anti-synthetase myositis-specific autoantibody (MSA) or a myositis-associated autoantibody (MAA) confirmed by a gold standard method and a clinical picture compatible with the said antibody should not be considered for classification as ASSD. Given the sub-optimal performance of methods other than immunoprecipitation in detecting anti-ARS, especially for non-Jo1 anti-ARS, the experts considered the positivity for anti-Jo1 and non-anti-Jo1 anti-ARS as distinct domains in the candidate classification criteria original model. However, because anti-ARS antibodies are generally considered mutually exclusive, a modified model was also created by merging these two anti-ARS domains. The weights and rankings for each key variable and domain did not substantially differ between the original and modified models. The positivity for anti-synthetase antibodies (anti-ARS) had the highest weight for ASSD classification, followed by ILD, myositis, mechanic’s hands, arthritis, dermatomyositis-specific rashes, Raynaud’s phenomenon, fever, and pulmonary hypertension (Table 2). The best cut-off for the original model had a sensitivity of 89.42%, specificity of 81.97%, positive predictive value (PPV) of 89.42%, negative predictive value (NPV) of 81.97%, Youden’s index of 0.714 and an area under the curve (AUC) of 0.935 compared to the gold standard of expert consensus. The best cut-off for the modified model had a sensitivity of 89.42%, specificity of 81.97%, PPV of 89.42%, NPV of 81.97%, Youden’s index of 0.714 and an AUC of 0.931.

Conclusion: The MCDA-based candidate classification criteria for ASSD performed well against the gold standard of expert consensus. This expert consensus-driven candidate classification criteria will be simplified and externally validated on data from ASSD and control patients collected through the CLASS project.

Acknowledgements: The American College of Rheumatology and the European Alliance of Associations for Rheumatology (ACR/EULAR) funded the CLASS project. This research was supported in part by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences.

Disclosure of Interests: Giovanni Zanframundo: None declared, Eduardo Dourado: None declared, Iazsmin Bauer-Ventura: None declared, Sara fa*ghihi-Kashani: None declared, Akira Yoshida: None declared, Aravinthan Loganathan: None declared, Daphne Rivero Gallegos: None declared, Darosa Lim: None declared, Francisca Bozan: None declared, Gianluca Sambataro Boheringer Ingelheim, Sangmee Bae: None declared, Yasuhiko Yamano: None declared, Francesco Bonella: None declared, Tamera J. Corte: None declared, Tracy Doyle Dr. Doyle has received support from Bayer and has been part of a clinical trial funded by Genentech, all unrelated to this study., David Fiorentino: None declared, Miguel Ángel González-Gay: None declared, Marie Hudson: None declared, Masataka Kuwana Asahi Kasei Pharma, Boehringer-Ingelheim, Chugai, Eisai, Janssen, Kissei, Mochida, Nippon Shinyaku, and Ono Pharmaceuticals, AstraZeneca, Boehringer-Ingelheim, Chugai, GSK, Kissei, and Mochida, Research grants: Boehringer-Ingelheim, MBL, and Ono Pharmaceuticals, Ingrid E. Lundberg: None declared, Andrew Mammen: None declared, Neil McHugh: None declared, Fredrick Miller: None declared, Carlomaurizio Montecucco: None declared, Chester V Oddis: None declared, Jorge Rojas-Serrano: None declared, Jens Schmidt: None declared, Albert Selva-O’Callaghan: None declared, Victoria P. Werth: None declared, Paul Hansen: None declared, Davide Rozza: None declared, Carlo Alberto Scirè: None declared, Garifallia Sakellariou: None declared, Lorenzo Cavagna Speeking fee: Boehringer-Ingelheim, Rohit Aggarwal 1. Actigraph: Consultant, 2. Alexion: Consultant, 3. ANI Pharmaceuticals: Consultant, 4. Argenx: Consultant, 5. AstraZeneca: Consultant, 6. Boehringer-Ingelheim: Consultant, 7. Bristol Myers-Squibb: Consultant, 8. CabalettaBio: Consultant, 9. Capella Bioscience: Consultant, 10. Corbus: Consultant, 11. CSL Behring: Consultant, 12. EMD Serono: Consultant, 13. Galapagos: Consultant, 14. Horizon Therapeutics: Consultant, 15. I-Cell: Consultant, 16. Janssen: Consultant, 17. Kezar: Consultant, 18. Kyverna: Consultant, 19. Merck: Consultant, 20. Novartis: Consultant, 21. Nuvig Therapeutics: Consultant, 22. Octapharma: Consultant, 23. Pfizer: Consultant, 24. Regeneron: Consultant, 25. Roivant: Consultant, 26. Sanofi: Consultant, 27. Teva: Consultant, 28. Artsome: Consultant, 29. Capstanx: Consultant, 30. Manta: Consultant, 1. Boehringer Ingelheim (BI): 2. Bristol Myers-Squibb, 3. EMD Serono, 4. Janssen, 5. Mallinckrodt, 6. Pfizer, 7. Q32.