Phospholipid autoantibodies and the antiphospholipid antibody syndrome: diagnostic accuracy of 23 methods studied by variation in ROC curves with number of clinical manifestations.

BACKGROUND We analyzed the diagnostic accuracies for the diagnosis of antiphospholipid syndrome (APS) of 23 antiphospholipid antibody (APL-Ab) assays targeted at different antigen preparations and immunoglobulin isotypes. METHODS In 144 patients with suspected APS, anti-cardiolipin (aCL) and anti-beta2-glycoprotein I (abeta2GPI) antibodies were measured with 23 different ELISAs from three manufacturers. Data were analyzed by ROC curves. In the absence of an accepted criterion standard, the endpoint "diagnosis of APS" was varied according to the number (two through five) of signs and symptoms of APS. RESULTS Although the presence of lupus anticoagulant was associated significantly with APL-Ab in 10 of 23 assays (P = 0.01-10(-4)) and recurrent arterial or venous occlusions were significantly associated with APL-Ab of IgM isotype in 5 of 6 assays (P = 0.02-10(-4)), sensitivity for detection of APS did not exceed 67%. With the exception of IgA APL-Ab, the diagnostic accuracy of the assays improved when the diagnosis of APS was based on an increasing number of simultaneous features of APS. For most methods, areas under the ROC curves were >0.8 irrespective of the method's subclass specificity and antigen preparation (aCL or abeta2GPI), if the clinical diagnosis of APS was based on four or more signs and symptoms of APS. CONCLUSION Despite considerable heterogeneity in the individual test results, a single test of IgG or IgM isotype targeted at either aCL or abeta2GPI antibodies has excellent diagnostic accuracy when the criterion for diagnosis requires four or more typical manifestations of APS.