Classification performance of the 2023 ACR/EULAR antiphospholipid syndrome criteria: a systematic review and meta-analysis
Leyre Riancho-Zarrabeitia, José A RianchoAbstract
Objectives
To systematically evaluate the classification performance of the 2023 American College of Rheumatology/European Alliance of Associations for Rheumatology (ACR/EULAR) classification criteria for antiphospholipid syndrome (APS).
Methods
We conducted a systematic review of literature from August 2023 through May 2026, including studies that evaluated the ability of the 2023 ACR/EULAR APS criteria to classify patients previously classified according to expert clinical diagnosis and/or the 2006 revised Sapporo (Sydney) criteria. Data on true and false positives and negatives were extracted. Sensitivity and specificity were pooled using a bivariate random-effects model when both parameters were available; sensitivity was additionally pooled using a univariate random-effects model.
Results
Twenty four studies comprising 25 data sets were included, of which 13 incorporated non-APS control groups. In the bivariate meta-analysis, pooled sensitivity was 0.74 (95%CI 0.67-0.81) and pooled specificity was 0.99 (95% CI 0.98–1.00). Sensitivity demonstrated substantial heterogeneity across studies and varied markedly by APS phenotype. Sensitivity was low in obstetric APS (pooled sensitivity 0.20 in studies restricted to obstetric APS and 0.31 when obstetric subgroup data from additional studies were included), whereas it was higher among cases of thrombotic APS (0.82). Across cohorts, non-classification was mainly driven by isolated IgM antiphospholipid antibody positivity and obstetric APS phenotypes.
Conclusion
The 2023 ACR/EULAR APS classification criteria exhibit consistently excellent specificity but reduced and phenotype-dependent sensitivity. These findings reinforce the intended role of the criteria for classification rather than clinical diagnosis and highlight important limitations in specific APS subgroups.