BACKGROUND: Foot-and-mouth disease (FMD) vaccine potency testing involves hundreds of animals each year. Despite considerable efforts during the past decades, a challenge-free alternative vaccine potency test to replace the European protective dose 50% test (PD(50)) has not been implemented yet. The aim of the present study was to further characterize the properties of serological vaccine potency models.

METHODS: Logistic regression models were built for 5 serological assays from 3 different laboratories. The serum samples originated from 5 repeated PD(50) vaccine potency trials with a highly potent A/IRN/11/96 vaccine. Receiver Operating Characteristic analysis was used to determine a serological pass mark for predicting in vivo protected animals. Subsequently, an estimated PD(50) was calculated and the serotype dependency of the logistic models was investigated.

RESULTS: Although differences were observed between the laboratories and the serological assays used, the logistic models accurately predicted the in vivo protection status of the animals in 74-93% of the cases and the antibody pass levels corresponded to 84-97% of protection, depending on the serological assay used. For logistic models that combine different serotypes, the model fit can be increased by inclusion of a serotype factor in the logistic regression function.

CONCLUSIONS: The in vitro estimated PD(50) method may be at least as precise as the in vivo PD(50) test and may accurately predict the PD(50) content of a vaccine. However, the laboratory-effect and the serotype-dependency should be further investigated.