Appendix 76 Post-vaccinal serosurveillance for FMD: a European perspective on progress and problems D J Paton1, K de Clercq2, A Dekker3 1 FAO World Reference Laboratory for FMD, Institute for Animal Health, Pirbright, Surrey GU24 ONF, UK 2 CODA-CERVA-VAR, Department of Virology, Section Epizootic Diseases, Groeselenberg 99, B-1180 Ukkel, Belgium 3 Institute for Animal Science and Health, Department of Mammalian Virology, P.O. Box 65, NL 8200 AB
There has been much debate about the use of the so-called vaccinate-to-live policy for the control of FMD in Europe. According to this approach, spread of the FMD virus from future outbreaks could be controlled by a short period of “emergency” vaccination of surrounding herds, reducing the need for large-scale preemptive culling of at-risk animals. Since vaccinated ruminants may become subclinically and persistently infected with FMD virus following challenge exposure, it is necessary to either kill or slaughter under controlled conditions foreseen in the OIE Terrestrial Animal Health Code all vaccinates (vaccinate-to-kill) or to detect and kill or slaughter under controlled conditions all vaccinates that have become persistently infected (vaccinate-to-live), in order to rapidly regain the most favoured trading status of FMD-free without vaccination. The latter approach can be attempted by testing vaccinated animals for the presence of antibodies to certain non-structural proteins (NSP) of FMD virus, which are induced by FMD infection, but not by vaccination with purified vaccines. The numbers of herds and animals to be sampled and tested to be confident that infection has not been missed will depend upon the expected prevalence of subclinical infection amongst and within herds. This in turn will depend upon the manner in which infection is spread and on how vaccination is applied. The sensitivity of the tests used and the size of the herds will also influence the numbers of samples required to be collected and tested. The new Council Directive 2003/85/EC on FMD takes account of these factors in its provisions for vaccination and for the use of post-vaccination serosurveillance to detect subclinical infection (Anon, 2003). According to the Directive, blood samples should be collected and tested from vaccinated animals and herds within a vaccination zone and from the unvaccinated offspring of vaccinated animals. Either all animals within vaccinated herds must be sampled and tested (Article 56, 3 (b)) or else sufficient numbers must be sampled and tested to enable a 5% prevalence of subclinical infection to be detected with 95% confidence (Annex III, point 2.2). Such sampling is not to take place until at least 30 days after the completion of emergency vaccination. EC legislation, taking into account that the necessary tests are regarded as herd tests and are not suited to verify the status of an individual animal, requires that herds within which at least one confirmed persistently infected animals has been detetcted must be slaughtered. However, difficulties in predicting the likely prevalence of post-vaccinal, subclinical infection and uncertainty over the performance of NSP tests has cast doubt over the suitability of these proposed sampling regimes. There are now several commercially available NSP antibody ELISA tests. Under favourable sampling conditions, our estimates are that the sensitivity of these tests for detecting individual persistently infected vaccinated cattle can be as high as 90%, with a 99% specificity. This assumes that the vaccines used have been purified to remove traces of NSP and have been given in an emergency setting involving the application of a single vaccine dose. If all animals can be sampled, at least one animal should score positive to detect a herd. If a test with a 100% sensitivity is used in a herd of 100 animals a prevalence of 1% can be detected, but in a small herd of 10 animals 1 positive animal is equal to a prevalence of 10%. If the sensitivity is lower, e.g. 80%, the chance of missing positive animals is 0.2. To be sure we will detect a farm with 95% confidence the chance of missing animals (0.2n) should be lower than 0.05. This means the number of positive animals should be at least 2 (0.2log(0.05)=log(0.05)/log(0.2)=1.9). In this case on a large farm (100 animals) a prevalence of 2% can be detected (95% confidence), but on a small farm (10 animals) only a prevalence of 20%. This therefore sets a limit on the degree of certainty that can be achieved for detecting low levels of persistently infected animals. It has been difficult to obtain reliable information on carrier prevalence under field conditions following vaccine breakdowns and even where available may not be relevant to regimes of husbandry and vaccination intensity that would prevail under European conditions. Sutmoller and Gagero (1965) reported a 50% prevalence at four months after a vaccine trial breakdown in Brazil. 455
