A multicenter study to assess the reproducibility of antiphospholipid antibody results produced by an automated system

Background Inter-assay variability is a well-known problem in antiphospholipid antibody testing, because of the lack of standardization. Inter-laboratory reproducibility for the same assay is similarly important. Objectives Testing repeatability and reproducibility of HemosIL((R)) AcuStar for anticardiolipin (aCL) and anti2-glycoprotein I antibodies (a2GPI) IgG and IgM. Patients/Methods In this observational study, out of 420 samples from the thrombophilia centers of Ghent and Geneva, 100 samples were randomly selected and successively analyzed in three centers: Ghent (C1, in duplicate for repeatability evaluation), Geneva (C2) and Frankfurt (C3). Results Results from 99 samples were available, including 25 from patients with antiphospholipid syndrome (APS) and 74 from non-APS patients. The intra-center repeatability expressed as intra-class correlation coefficient (ICC) was higher than 0.99 for each parameter. Differences between two measurements rarely exceeded 1 U mL(-1) for values below 100 U mL(-1), except for a2GPI IgG, where differences varied from -4 to 4 U mL(-1). The inter-center ICCs were higher than 0.99, except for aCL IgM (ICC = 0.961). These ICCs remained high even when considering values below 100 U mL(-1) (0.943, 0.964 and 0.977 for aCL IgG, aCL gM and a2GPI IgM, respectively), except for a2GPI IgG (ICC = 0.652). Qualitative comparison showed less than 5% discordant classification between centers, with somewhat more discordant results for a2GPI IgG. Conclusions In terms of discriminating properties, the HemosIL((R)) AcuStar has excellent intra-center repeatability and a good inter-center reproducibility for aCL IgG, aCL IgM and a2GPI IgM. Some concern may arise for a2GPI IgG.