Evaluation of antiphospholipid antibody assays using latent class analysis to address the lack of a reference standard

Background: Method evaluation of new assays for the detection of antiphospholipid antibodies (aPL) such as anti--cardiolipin (aCL) or anti-beta(2)-glycoprotein I (a beta(2)-GPI) is challenging, as no internationally accepted reference material is available yet. Besides a lack of standardization, unacceptable inter-laboratory comparability of established tests is regularly observed. Owing to the absence of a commonly accepted reference standard, the evaluation of two research surface plasmon resonance (SPR) biosensor assays was performed using statistical methods from latent class analysis (LCA). Methods: aCL and a beta(2)-GPI IgG and IgM were measured in sera from 63 antiphospholipid syndrome patients, fulfilling the Sydney criteria, and in 34 healthy controls with four commercial assays. LCA was performed on the results and sera were assigned to the antibody-positive or antibody- negative group. Sera were subsequently evaluated in the SPR assays for aCL and a beta(2)-GPI. Optimal cutoffs and diagnostic performances of the research systems were established employing the LCA-derived gold standard. Results: With area under the curve results of 0.96 and 0.89 for the detection of aCL and a beta(2)-GPI, the research SPR assays discriminated well between antibody-positive and antibody-negative sera. Their sensitivities and specificities were comparable to the investigated commercial immunoassays. Conclusions: SPR assays are a suitable tool for the detection of aCL and a beta(2)-GPI with diagnostic performances not different from currently available commercial tests. LCA enabled the calculation of sensitivities and specificities for aPL assays in absence of a reference standard.