Interval-specific likelihood ratios improve interpretation of antiphospholipid antibody testing across different analytical platforms.

OBJECTIVES

Antiphospholipid antibodies (aPL) are essential biomarkers for classification, diagnosis, and risk stratification of antiphospholipid syndrome (APS). Although automated analytical platforms, such as chemiluminescent immunoassay (CLIA), and particle-based multi-analyte technology (PMAT), are increasingly implemented in routine laboratories, uncertainty remains regarding their diagnostic performance and interpretation of quantitative antibody results. We compared ELISA, CLIA, and PMAT for aPL detection and evaluated clinical utility of interval-specific likelihood ratios (LRs).

METHODS

In this retrospective single-center diagnostic accuracy study, 351 patients with APS (238 thrombotic and 113 obstetric APS) and 197 controls were included. IgG and IgM anticardiolipin (aCL) and anti-β2-glycoprotein I (aβ2GPI) antibodies were measured using ELISA, CLIA, and PMAT. Analytical agreement was assessed by correlation analysis, and diagnostic performance by ROC analysis. Interval-specific LRs were calculated using predefined specificity thresholds (90 %, 95 %, and ≥99 %).

RESULTS

Strong correlations were observed among all analytical platforms (p<0.001), with the highest agreement found between CLIA and PMAT. Automated platforms showed higher positivity rates than ELISA, particularly for IgM isotypes. PMAT achieved the highest diagnostic accuracy in thrombotic APS for IgG aCL (0.810), IgG aβ2GPI (0.816), and IgM aβ2GPI (0.848). Increasing antibody concentrations were associated with progressively higher interval-specific LRs, especially at specificity thresholds ≥99 %. Diagnostic performance was lower in obstetric APS.

CONCLUSIONS

Automated platforms, including PMAT represent analytically robust and clinically reliable methodologies for aPL testing, and represent valid alternatives to conventional ELISA assays. Interval-specific LR interpretation provides clinically relevant probabilistic information beyond dichotomous cut-offs, and may improve laboratory-based risk stratification in APS across analytical platforms.