Fig. 2

Optimal cut-off and comparison of univariate ROC curve analysis. Abbreviations: AUC, area under the curve; aCL, anticardiolipin antibody; aβ2GPI, anti-β2GPI antibody; CIA, chemiluminescent immunoassay; ROC, receiver operating characteristic. A Receiver operating characteristics (ROC) analysis for anti-β2GPI/HLA-DR antibodies was performed to determine the most suitable cut-off value that maximized the area under the curve (AUC) between non-arterial thrombosis (n = 627) and arterial thrombosis (n = 77). The ROC analysis revealed an AUC of 0.601 with a sensitivity and specificity of 33.8% and 91.4%, respectively, at the optimal cut-off level of 172.359 U/mL, using Youden’s index. B Five ROC plots of the individual aPLs were compared to each parameter that contains aCL IgG [EIA], aCLβ2GPI IgG [EIA], aCL IgG [CIA], aβ2GPI [CIA], and the anti-β2GPI/HLA-DR antibody. Other differential aPLs are listed in Table 2. The ROC curve for the anti-β2GPI/HLA-DR antibody was the same blue curve seen in A. C, D Comparison between anti-β2GPI/HLA-DR antibody titers and aβ2GPI IgG antibody or aCL IgG antibody titers. The cut-off value of 172.359 U/ml for the anti-β2GPI/HLA-DR antibody was shown as the blue line. The 99th percentile cut-off value of 20 U/ml for aβ2GPI IgG antibody or aCL IgG antibody was shown as the green or yellow line, respectively. The number and percentage of patients with histories of arterial thrombosis among included patients in each quadrant were presented