Objectives: To evaluate the agreement and performance of two tests for aPLs with regard to association with manifestations of the APS in patients with SLE.
Methods: We investigated 712 SLE patients and 280 population controls. Cardiolipin and β(2) glycoprotein-I antibodies were measured with routine ELISA and a new automated method. Three positivity cut-offs (99%, 90% of controls and recommended cut-off by manufacturers) were used. Associations with previous thrombotic events, thrombocytopenia and, in a subgroup of patients, obstetric morbidity (n = 296) were evaluated. Results were compared with the LA test, performed in 380 patients.
Results: Inter-test agreement was moderate (demonstrated by κ-values 0.16-0.71). Performance of the two tests was similar: at the 99th percentile cut-off, sensitivity for any thrombotic event ranged from 3.7% to 24.8%, while specificity was 84.7-97.7%. Regardless of assay, IgG isotypes were associated with venous thrombosis and ischaemic cerebrovascular disease, whereas aPLs of IgM isotype were weakly associated with ischaemic heart disease. Associations were greatly affected by aPL level. LA performed better than the specific aPL tests. LA was associated with any thrombotic event, odds ratio 5.4 (95% CI 3.1, 9.4), while the specific aPL tests ranged from non-significant to an odds ratio of 1.9 (95% CI 1.03, 3.4) using criteria cut-off. LA was also convincingly associated with other APS manifestations.
Conclusion: In relation to thrombotic manifestations, there was moderate agreement but no clear advantages when comparing a routine aPL ELISA with an automated method. APL isotype and titre as well as LA positivity are important for risk assessment in SLE patients.