Development of the 2023 ACR/EULAR Antiphospholipid Syndrome Classification Criteria, Phase III-C Report: Assessment of Patient Scenarios (Derivation Cohort) and Refinement of Definitions

Objective: The 2023 American College of Rheumatology (ACR)/EULAR antiphospholipid syndrome (APS) classification criteria aim to identify patients with a high likelihood of APS for research. Phases I/II of our four-phase methodologic approach resulted in 27 candidate criteria organized in clinical and laboratory domains. Here, we summarize phase III efforts to reduce and refine criteria using patient scenarios. Methods: Using standardized definitions for candidate criteria, the Steering Committee collected antiphospholipid antibody (aPL)–positive cases referred for “suspected APS.” Treating physicians assessed APS case likelihood using a Likert scale. Poisson regression calculated risk ratios (RRs) and 95% confidence intervals (CIs) to quantify the direction and size of the association of candidate criteria with “highly likely” versus “equivocal or unlikely” APS, which guided Steering Committee candidate criteria refinement and organization. Results: We collected 314 suspected APS cases (137 [44%] highly likely and 177 [56%] equivocal/unlikely APS). Provoking venous thromboembolism (VTE) or arterial thrombosis (AT) risk factors reduced the size of the association with highly likely APS (RR 4.31 [95% CI 2.11–8.78] to RR 1.56 [95% CI 0.89–2.75] for VTE and RR 3.48 [95% CI 1.91–6.32] to RR 1.64 [95% CI 0.77–3.51] for AT). Persistent lupus anticoagulant, anticardiolipin IgG antibody ≥40 U, and anti–β₂-glycoprotein-I IgG antibody ≥40 U were positively associated with highly likely APS (all P < 0.05). Eventually, items within eight additive and independent clinical and laboratory domains were refined. Conclusion: Referred suspected APS cases provided insight into associations of individual candidate criteria with APS likelihood. RR analyses helped refine items and organize the draft classification system into eight additive and independent clinical and laboratory domains.