Detection of diphtheria antitoxin by four different methods.

Abstract

To investigate the reliability of the different methods used in Norway and Russia for detection of diphtheria antitoxin. One hundred and twenty-two sera were selected among Russian serum samples previously collected for seroepidemiologic studies of diphtheria antitoxin. The sera were selected to cover the total antitoxin range and were analyzed by four different antidiphtheria toxin assays: an in vitro toxin neutralization test using Vero cells (in vitro NT), an in vivo neutralization test using rabbit skin inoculation (in vivo NT), an indirect enzyme immunoassay (EIA) and a passive hemagglutination assay (PHA). The results were expressed according to the international standard as: not protected (<0.01 IU/mL), relatively protected (0.01-0.1 IU/mL) or protected (≥0.1 IU/mL). The sensitivity, specificity and inter-rater agreement (K or Kw) of each method were related to the in vitro NT selected as the reference method. The in vivo NT test corresponded very well with the in vitro NT in its ability to differentiate between protection/relative protection and no protection (sensitivity 97%, specificity 87% and K=0.84). The EIA test showed a high sensitivity (96%), but since many sera were categorized as protected rather than not protected, the specificity (30%) and inter-rater agreement (K=0.29) were low. The PHA test had a very high specificity (100%) but a low sensitivity (86%). The agreement between the two neutralization tests was high. If none of the neutralization assays is routinely available, the PHA test can be used to predict the need for vaccination on an individual basis but should not be used for seroepidemiologic studies, since the protection rate for diphtheria would be falsely too low, due to the lower sensitivity. The indirect EIA test used in this study should not be used routinely.