Diagnostic accuracy study of a factor VIII ELISA for detection of factor VIII antibodies in congenital and acquired haemophilia A.

Antibody formation to factor VIII (FVIII) remains the greatest clinical and diagnostic challenge to the haemophilia-treating physician. Current guidance for testing for inhibitory FVIII antibodies (inhibitors) recommends the functional Nijmegen-Bethesda assay (NBA). A FVIII ELISA offers a complementary, immunological approach for FVIII antibody testing. It was the aim of this study to retrospectively evaluate the performance of a FVIII ELISA (index) for detection of FVIII antibodies, compared with the NBA (reference). All samples sent for routine FVIII antibody testing at two haemophilia Comprehensive Care Centres, were tested in parallel using the NBA and a solid-phase, indirect FVIII ELISA kit (Immucor). A total of 497 samples from 239 patients (severe haemophilia A=140, non-severe haemophilia A=85, acquired haemophilia A=14) were available for analysis. Sixty-three samples tested positive by the NBA (prevalence 12.7%, 95% confidence interval [CI], 9.9-15.9 %), with a median inhibitor titre of 1.2 BU/ml (range 0.7-978.0). The FVIII ELISA demonstrated a specificity of 94.0% (95%CI, 91.3-96.0), sensitivity of 77.8% (95%CI, 65.5-87.3), negative predictive value of 96.7% (95%CI, 94.5-98.2), positive predictive value 65.3% (95%CI, 53.5-76.0), negative likelihood ratio 0.2 (95%CI, 0.1-0.4), positive likelihood ratio 13.0 (95%CI, 8.7-19.3) and a diagnostic odds ratio of 54.9 (95%CI, 27.0-112.0). Strong positive correlation (r=0.77, p<0.001) was seen between the results of the NBA (log adjusted) and FVIII ELISA optical density. In conclusion, FVIII ELISA offers a simple, specific, surveillance method enabling batch testing of non-urgent samples for the presence of FVIII antibodies.