Laboratory decision-making during the classical swine fever epidemic of 1997–1998 in The Netherlands

Laboratory decision-making during the classical swine fever epidemic of 1997–1998 in The Netherlands

Preventive Veterinary Medicine 42 (1999) 185±199 Laboratory decision-making during the classical swine fever epidemic of 1997±1998 in The Netherlands...

299KB Sizes 0 Downloads 20 Views

Preventive Veterinary Medicine 42 (1999) 185±199

Laboratory decision-making during the classical swine fever epidemic of 1997±1998 in The Netherlands A.J. de Smit*, P.L. EbleÂ, E.P. de Kluijver, M. Bloemraad, A. Bouma Department of Mammalian Virology, Institute for Animal Science and Health (ID-DLO), PO Box 65, NL-8200 AB, Lelystad, The Netherlands Accepted 18 June 1999

Abstract The National Reference Laboratory for classical swine fever (CSF) virus in the Netherlands examined more than two million samples for CSF virus or serum antibody during the CSF epizootic of 1997±1998. The immense amount of samples and the prevalence of border disease (BD) virus and bovine viral diarrhoea (BVD) virus infections in Dutch pig herds necessitated the diagnostic efforts of the laboratory to be focused on generating CSF specific test results throughout the eradication campaign. Detection of 82% of the 429 outbreaks was achieved through the combined use of a direct immunofluorescence and peroxidase assay (FAT/IPA) with samples (tonsils) collected from clinically-suspected pigs. This suggests that in the majority of the outbreaks, the pigs had clinical signs that were recognised by the farmer and/or veterinarians, indicating the presence of CSF virus in a pig herd. A positive diagnosis of 74% of all the tissue samples (tonsils) collected at infected pig holdings was established by FAT. More than 140,000 heparinised blood samples were examined by virus isolation, resulting in the detection of 4.5% of the infected herds. CSF virus was isolated in approximately 29% of all the blood samples collected from pigs at infected or suspected farms. Several serological surveys Ð each done within a different framework Ð led to the detection of 13.5% of the total number of outbreaks. The detection of CSF virus antibody in serum was carried out by semi-automated blocking ELISA. Approximately 28.5% of the sera which reacted in the ELISA were classified as CSF virus-neutralising antibody positive and 26.5% as positive for other pestiviruses following the virus neutralisation test (VNT). We concluded that two of the CSF laboratory diagnostic methods described were determinative in the eradication campaign: first, the FAT for the screening of diseased pigs; and second, the ELISA and VNT when millions of predominantly healthy pigs needed to be screened for the *

Corresponding author. Tel.: ‡31-320-238698; fax: ‡31-320-238668 E-mail address: [email protected] (A.J. de Smit) 0167-5877/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 5 8 7 7 ( 9 9 ) 0 0 0 7 5 - 6

186

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

presence of CSF serum antibody. Decision-making on the basis of results generated by either method can, however, be seriously hindered when samples are examined from pig herds with a high prevalence of non-CSF pestiviruses. # 1999 Elsevier Science B.V. All rights reserved. Keywords: Classical swine fever virus; Laboratory diagnosis; Samples

1. Introduction Clinical signs and pathological findings after infection of the pig with classical swine fever (CSF) virus are diverse and not pathognomonic (Van Oirschot, 1999; Depner et al., 1996). Suspicion of CSF in pig herds should therefore be confirmed by examining relevant samples with CSF-specific laboratory methods. Following the incidence of CSF virus in an area of high pig density in the Netherlands, millions of samples were submitted to the National Reference Laboratory for CSF. The majority of these samples were collected to monitor the infection status of pig herds located in restricted areas free from CSF infection for at least 30 days (Anonymous, 1980). The disease was eradicated from the pig population in the Netherlands by the so-called `stamping out' strategy used in the European Union (EU), within the framework of a non-vaccination policy for the control of a notifiable disease like CSF. A total of 429 pig holdings were declared CSF infected by the veterinary authorities. More than 12 million pigs were killed, the majority of which took place within the framework of market-support/welfarerelief activities. The organisations involved in the eradication were: the regional animal health services (AH); the Inspection Services for Livestock and Meat (RVV); the National Reference Laboratory for CSF (ID-DLO); and the Ministry of Agriculture (LNV). The measures prescribed in the CSF Council Directive 80/217/EEC formed the basis of the eradication campaign. Success of this campaign is dependent on the rapid implementation and control of several restrictive measures. Rapid enforcement of a complete standstill of the transportation of pigs, supplemented with the implementation of appropriate zoo-sanitary measures on all pig holdings, should limit further introduction of CSF virus into naive pig herds. Additionally, the sensitivity and specificity of the available laboratory diagnostic tools and their suitability for examining the pig population in the enclosed area determine the duration and extent of a CSF epizootic. If CSF virus is introduced into a region with a high pig density, all these aspects must be tuned perfectly to control the spread of virus within a limited period. The origin and type of the first Dutch CSF virus isolate found in 1997 was determined in collaboration with the European Reference Laboratory for CSF (Hannover, Germany). After sequencing parts of the genome, the Dutch CSF isolate was found to be identical (Widjojoatmodjo et al., 1999) to a CSF isolate from pigs originating from the Paderborn region in Germany in 1996 which belonged to subgroup 2.1 (Greiser-Wilke et al., 1998). It is well known that two other members of the pestiviruses genus, namely, bovine viral diarrhoea (BVD) virus and border disease (BD) virus, can infect the pig as well as their more usual hosts and, by doing so, may interfere with the clinical (Terpstra and Wensvoort, 1988, 1997) and laboratory (Wensvoort et al., 1994) diagnosis of CSF.

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

187

Reports regarding the incidence of BVD and BD virus among the pig population in the Netherlands (Terpstra and Wensvoort, 1991, Wensvoort et al., 1994) indicate the presence of both diseases in pig herds in the Netherlands. Similar results have been found in other pig-populated regions (Liess et al., 1977; Holm Jensen, 1985; Afshar et al., 1989). Laboratory diagnosis of CSF is followed by measures which have severe consequences for the individual pig holder (culling of the farm) and the pig industry (export losses). Specific, sensitive and rapid laboratory results are therefore necessary for confirmation of a suspicion of CSF in the field. This paper describes the support to decision-making of different CSF laboratory diagnostic methods during an outbreak of CSF in one of the most pig-dense areas of the world. 2. Material and methods 2.1. Detection of viral antigen The direct immunofluorescence assay (FAT) was used for the detection of CSF viral antigen (Ressang, 1973) in organs (tonsil, spleen, kidney, ileum). The organs were collected from selected diseased pigs at CSF-suspected pig holdings or from diseased pigs during clinical inspections, for instance during the compulsory (Anonymous, 1980) clinical screening surveys in the protection zones. The direct immunoperoxidase assay (IPA) employing CSF-specific monoclonal antibodies was used to confirm if samples found positive by the FAT contained CSF or BVD/BD viral antigen (Wensvoort et al., 1986) (fulfilling the requirement of Annex I Council Directive 80/217/EEC). The FAT and IPA were carried out according to standard protocol (Fig. 1) within a certified qualitycontrol system. Duplicate sections cut (cryostat) from each tissue sample were examined. The veterinary authorities were notified of those samples found to be positive for CSF viral antigen, together with other relevant data of the sampled pigs and the pig holding. The following announcement was added to the test report send to the submitter of the samples if the test results were negative or inconclusive: ``If circumstances persist which suggest the presence of the disease please submit new samples''. Unclotted blood samples for virus isolations were collected randomly and/or on the basis of fever. Before heparinised or EDTA-blood samples were examined, a 4 ml aliquot of each sample was stored in pre-labelled containers (Sanbio). Virus was isolated from tissues and unclotted, freeze-thawed whole blood using a method of De Smit et al. (1994). Briefly, a volume of 300 ml of the undiluted sample (a 10% (w/v) tissue suspension or freeze-thawed whole blood) was incubated on a 70% confluent monolayer of SK6 cells in a 24 wells plate (Greiner) for 1 h at 378C. After removing the inoculum, the monolayer was washed twice with 800 ml of medium and then 800 ml medium was added per well. Subsequently, the monolayers were incubated at 378C in an atmosphere with 5% CO2 for 4 days, and then examined to determine the presence of virus by the immunoperoxidasemonolayer-assay (IPMA) using CSF virus-specific monoclonal antibodies (Wensvoort et al., 1986). Blood samples or tissue suspensions which gave toxic reactions in the virus isolation assay were tested again and, if similar reactions were found, results were issued as inconclusive.

188

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

Fig. 1. Test procedure for the detection of CSF virus antigen by FAT and IPA in tissue samples (tonsil, spleen, kidney, ileum).

2.2. Detection of antibody The sera from blood samples submitted for testing was collected in a 96-tube (Micronic1) storage-block system. This was done manually or semi-automatically by robots (AH) the latter using barcode-labelled collection tubes. Serum samples were screened for CSF virus-specific antibody (Ab) with a blocking ELISA (Colijn et al., 1997) in a semi-automated (Bloemraad et al., in preparation) and/or manual test set-up.

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

189

Fig. 2. Test procedure for the detection of CSF virus neutralising antibody (Ab) by ELISA and virus neutralisation test.

At times, approximately 15,000±20,000 sera samples were tested per day in the semiautomated set-up. Serum samples which were positive (30% inhibition) in the ELISA (tested in manually or semi-automated set-up) were always re-examined in the ELISA (manually). Confirmation of the presence of CSF virus-neutralising Ab in sera found positive by ELISA (fulfilling the requirements of Annex I EU Directive 80/217) was done by means of a virus-neutralisation test: the neutralisation-peroxidase-linked assay (Terpstra et al., 1984). The ELISA and VNT were performed according to standard protocol (Fig. 2) within a certified-quality system. Internal control sera were used to monitor the performance (sensitivity and specificity) of different batches of ELISA kits/ bulk ingredients. A VNT with BD virus (Vilcek and Belak, 1996), Strain F, was used to optimise discrimination of CSF virus-neutralising Ab from cross-reacting BD virus Ab (Wensvoort et al., 1994). The BD virus strain had been isolated from piglets in 1994. The use of this isolate in the VNT allowed us to discriminate numerous CSF-virus Ab-positive (ELISA)

190

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

cross-reacting sera in the period 1994±1996 which were hence classified as CSF falsepositive due to BD virus neutralising Ab. If VNT results were inconclusive, a third VNT using BVD virus strain Oregon was performed and the CSF and BD virus VNT was repeated. Sera (duplicate) were titrated to endpoints of 1 : 800 or 1 : 3200. Neutralising Ab titres were calculated as the highest reciprocal serum dilution neutralising 100 50% tissue culture infective doses (100 TCID50) of CSF virus in 50% of the cultures. End points of serum titration were calculated by the method of Reed and Muench (1938) Serum samples found toxic or heavily contaminated in duplicate wells in both neutralisation assays were in general not tested again and results were marked as inconclusive. Field sera collected from sows located at pig holdings free of CSF were screened for the presence of pestivirus antibodies in an ELISA described by Paton et al. (1991) in order to estimate the sero-prevalence of non-CSF pestivirus. The appropriate authorities were notified of results from the VNT and the ELISA leading to the conclusion that sera contained CSF -neutralising Ab, together with the relevant data concerning the sampled pig and the pig holding. 2.3. Samples The handling of a variety of types and amounts of samples for the different test methods was complex and an attempt is made to describe a few aspects of this process. Each individual sample was identified by the laboratory management computer system (LMS) and given a special code, making it easier for the laboratory experts to find specific data (e.g. sample list containing information of the location and identification of each sampled pig). Each test result (positive or negative) was recorded first in writing and then fed into the computer system. Blood samples for serological or virological examination were collected by veterinarians of varying status (practitioners, students, AH, RVV). After collection of the blood samples (unclotted whole blood samples or serum blood samples) on the pig holding the samples were gathered at the local AH centre, the crisis centre or the IDDLO. The majority of the serum samples were prepared for examination at an AH centre using robots for collecting samples (96) in Micronic blocks. The tissue samples for the FAT were gathered as follows: farmers and/or veterinarians diagnosed pigs clinically suspected of CSF and a number of the pigs were killed on the farm. Transport of the dead pigs to the local AH was than organised by the AH service. Pathologists at the AH performed a post-mortem and collected samples from the pig's tonsil, spleen, kidney, ileum for further examination (FAT/IPA). The samples were packaged in a sealed plastic container and necessary identification of animals, samples and containers were double checked and additional information was written on the accompanying forms before transport. All of this was stored at 48C in an insulated box. The majority of the samples during the outbreak were transported to the laboratory by courier and the approximate time of transport from the gathering point (AH, RVV crisis centre) to the laboratory was 2±3 h. The sample acceptance policy of the ID-DLO demands clear identification of samples, this means that individual samples must be marked and correlate to an accompanying list with all the relevant data concerning the pig holding, barns, pens, etc. and the

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

191

identification of the sampled pigs. The ID-DLO was confronted with numerous violations of this policy but could not refuse incorrectly identified samples. On several occasions it was not possible to identify the samples correctly and these samples were given a sequential number. The normal capacity of 4±5 members of staff working at the National Reference Laboratory for the diagnosis of CSF virus was expanded to a work force of approximately 30±40 during the first weeks of the outbreak. Several teams were formed to deal with samples for separate test methods in separate laboratories. All CSF virus-positive test results were discussed and judged by at least two experienced persons. In total, six laboratory rooms (each covering an area of approximately 50 m2) were necessary to process and examine the samples. Also, more than 25,000 forms with relevant information about the samples were processed and had to be stored at the laboratory. After testing, all the samples were stored at ÿ208C or ÿ708C (except samples found negative by the FAT). Thus in total, more than two million samples were stored at the laboratory for reasons of quality-control regulations and judicial and research interests. 3. Results During the period from February 1997 to June 1998, 2.3 million samples were examined for CSF viral antigen, CSF virus or CSF antibody. These samples were collected from pigs located at approximately 50% of the total number of pig holdings (22,000) in the Netherlands in 1997. The processing of samples and accompanying information within the laboratory facilities is shown in a schematic representation in Figs. 3 and 4. 3.1. Detection of CSF viral antigen Tonsils collected from pigs suspected of CSF and examined by FAT/IPA detected 82% of the 429 CSF infected pig herds. Of the tonsils (2100) collected from CSF-infected pig herds, 74% were found positive by FAT. After FAT positive findings in the tonsils of pigs from four other herds BVD or BD virus was isolated. This concurred with findings in the IPA. In all four cases samples originated from young piglets with clinical signs of CSF. On another four occasions tonsils found to be positive in the FAT did not reveal the presence of pestivirus after subsequent IPA or virus isolation. The deteriorated condition of three percent of the samples submitted (18,000) were regarded as unsuitable for further testing. 3.2. Detection of CSF virus Approximately 144,000 heparinised-blood samples were tested for the presence of CSF virus by virus isolation in cell culture and led to the detection of 4.5% of the infected pig herds. The majority of the heparinised blood samples (81%) were collected for marketsupport/welfare reasons. The remaining 19% were collected on pig holdings visited during tracing activities and infected pig holdings just before culling. CSF virus was isolated by cell culture in approximately 29% of the samples collected from infected pig herds. Five percent of the samples examined were toxic in the assay.

192

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

Fig. 3. The number of samples examined during outbreaks of classical swine fever in the Netherlands in the period 1984±1998.

Two infected pig herds were detected after isolation of CSF virus from the tonsils. The tonsils were collected from pig herds (320) which had received semen from a CSF virus infected artificial insemination station were examined by virus isolation only. Of the 3891 tonsil samples examined, 10% were found to be toxic in cell culture. 3.3. Detection of CSF virus neutralising antibody The test results obtained from the different serological surveys led to the detection of 13.5% of the 429 outbreaks. Of the 2.15 million sera which were examined in the ELISA reactions were observed in approximately 0.7% sera collected from about 3,100 pig herds. Of these sera, 28.5% were classified as positive for CSF virus-neutralising antibodies and originated from 286 of the 429 infected pig herds. Of the remaining sera 26.5% were classified as positive for BD or BVD virus neutralising antibodies, 42% as negative and 3% as toxic. The CSF virus Ab ELISA showed consistently high reactions of sera with CSF virus neutralising titres  1 : 200 (Fig. 5). The majority (59%) of the CSF VNT positive sera had an inhibition score of over 90% in this ELISA whereas approximately 18% of the CSF positive sera had an inhibition score of less than 70%. The majority of the sera classified as positive for BD/BVD neutralising antibodies scored between 30% and 80% inhibition in the CSFV-Ab ELISA, independent of the titre of the neutralising antibody (Fig. 5). Of the sera classified as negative in the VNT for antibody to CSF/BD/BVD 87% scored less than 70% inhibition (Fig. 6) in the CSFV-Ab ELISA.

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

193

Fig. 4. Schematic representation of the processing of incoming samples with accompanying information, the internal transport of samples and test results and the outgoing flow of test results.

3.4. Detection of pestivirus serum antibody A BD/BVD virus serum antibody prevalence of 11% was found among 12,000 sows from 215 different pig herds. Sera found positive were collected from 181 pig herds which were free of CSF. A survey among 300 slaughtered sows and boars demonstrated on the basis of neutralising antibodies titres that a total of 20% of the pigs had developed antibodies against BD virus (61%) or BVD virus (39%). On a mixed farm where BVD virus had been isolated from piglets, the seroprevalence of BVD virus-neutralising antibodies among 125 sows was 59%. 4. Discussion Rapid and unambiguous laboratory results are important to substantiate presence or absence of disease in pig herds, especially when a stamping out strategy is used to eradicate CSF from an area of high pig density. The prevalence of diseases which lower the specificity of the laboratory methods needed to detect a notifiable disease can frustrate laboratory experts, veterinary authorities and the pig industry. The presence of

194

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

Fig. 5. Field sera which were found positive (CSF-Ab-ELISA and VNT) for CSF virus neutralising (~, n ˆ 3387) or BD/BVD virus neutralising (*, n ˆ 3100) antibodies were grouped according to the VNT titre of each individual serum. Of each group, the mean (SD) percentage of inhibition is shown.

such diseases in the population of interest should, in this respect, be recognised as unwanted. The prevalence (Terpstra and Wensvoort, 1991; Wensvoort et al., 1994) of BVD virus and BD virus in Dutch pig herds hindered laboratory decision-making during the CSF epizootic of 1997±1998 in the Netherlands. The fact that 82% of the CSF outbreaks were detected by examining samples in the FAT/IPA suggests that the farmer and/or veterinarians were often capable to recognise clinical signs of CSF virus-infected pigs. This implies that they can play a key role in the early detection of CSF-infected pig herds. According to the information on the registration forms which accompanied the submitted tissues the predominant signs observed were fever, anorexia and apathy. Such signs can develop after an incubation period of three days, and the multiple occurrence of theses signs often led to reporting of suspected CSF. The results of the FAT were available within several hours, generally leading to rapid implementation of control measures in the field. The FAT and IPA are both qualitative test methods and interpretation of the assays is subjective. Positive FAT and IPA sections were therefore preferably judged by three persons and the FAT was always repeated (Fig. 1). Non-specific staining or a low level of infected epithelial cells can hinder the interpretation of the IPA. Here, isolation of CSF virus from the tissue samples by cell culture is required for confirmation (Fig. 1). This can be accompanied by a request for additional sampling of the herd in question. An alternative method for confirming positive findings in the FAT which would still support rapid decision-making could be a CSF specific RT-PCR (McGoldrick et al., 1998). Unfortunately, the killing of suspected pigs and the restrictive regulations implemented by the veterinary authorities after sampling of the pigs, can discourage farmers and veterinarians from reporting

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

195

diseased pigs. This will postpone the detection of infected herds and increase the risk of further spread of CSF. The killing of pigs may, however, be omitted by taking biopsies from the tonsils or skin (Pan et al., 1993, personal observations) of diseased pigs. Furthermore, adaptation of the regulations following the reporting of suspected CSF may support a stamping-out strategy more efficiently. After further analysis of the available field samples, a comparison can be made between the effectiveness of examining tonsil samples by the FAT and the isolation of CSF virus from whole-blood samples or from tonsil suspensions to detect CSF-infected pigs in the field with clinical sign of the disease. Virus isolation from blood or tissue samples played an important role in the excessive workload the laboratory faced but did not detect many (4.5%) CSF infected herds. The method we have described for virus isolation minimises background staining. Problems concerned with the interpretation of this assay were rarely encountered. Isolation of CSF virus from the blood of experimentally infected pigs can be achieved in the early phase of infection (Kaden et al., 1999). Viremia is, however, generally a transient phenomenon and the duration and height of viremia are variable, both factors depending, for instance, on the age of the pig (Depner et al., 1996; Koenen et al., 1998). Consequently, the virus isolation assay when used for screening clinically healthy pig herds by random sampling was thought to have little predictive value. As expected, few infected pig herds, two, were detected through the sampling of clinically healthy pigs within the framework of market support/welfare reasons. Experts explained to the authorities that in this case a method based on random sampling for CSF virus antibody was more appropriate, since antibodies last for years and the number of seropositive pigs increases with time if virus circulates within the herd. Three months later these statements were acknowledged and random serological sampling resulted in the detection of 21 infected pig herds. Several commercial antigen capture ELISAs are available for the detection of pestivirus antigen in a range of different samples (leukocytes, plasma, serum, whole blood, tissue suspensions). The advantages of these assays are that they facilitate processing of large numbers of samples in a limited period. However, factors (duration and height of viremia) mentioned before and the sensitivity and specificity of these assays have discouraged us from using these assays, particularly when screening clinically healthy pigs by random sampling (Kaden et al., 1999). Laboratory results which create uncertainty because of failure to confirm positive results, may demand frequent resampling of the herds in question. False-negative results provided by these assays can also create a false sense of security, both of these aspects are unwanted and introduce extra difficulties during the eradication of an outbreak of CSF in an area of high-pigdensity. The positive predictive value of these assays providing that they are CSFspecific, could be increased if samples are collected from pigs suspected of CSF on the basis of for instance fever (Kaden et al., 1999, personal observations). A total of 13.5% of the infected pig herds were detected after examination of sera collected during the different serological surveys. On a number of occasions (138 times), serological test results (ELISA plus VNT) resulted in the re-sampling of pig herds before decisions were made regarding the presence of CSF virus on the farms in question. In the end, 39% of these pig holdings were declared as CSF infected by the veterinary authorities. In eight cases, a diagnostic follow up was impossible because the holdings were already emptied under market support/welfare regulations. The prevalence of CSF

196

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

serum antibody in the sampled pig herds varied considerably (0±100%). Data gathered during outbreaks of CSF in Germany and Belgium (Koenen et al., 1996) in recent years have demonstrated a low prevalence of CSF serum antibody in CSF-infected pig herds. At the beginning of the epizootic the sample size during the serological surveys of the pig herds in restricted areas was as prescribed in the EU regulations, which is based on a sampling rate sufficient to detect 5% seroprevalence of antibody to CSF with 95% confidence. However, the low prevalence of pigs with CSF serum antibody led to a decision to increase the sample size to a level sufficient to detect 2±3% seroprevalence. The authorities believed that increasing the sample size would minimise the risk of overlooking CSF serum antibody positive `carrier sows'. The `carrier sow' syndrome can be of considerable epidemiological importance for the perpetuation of CSF virus in a pig population (Terpstra, 1987). The sensitivity of ELISAs used to detect of CSF seropositive pigs in herds with a low prevalence of CSF is important. It is generally agreed that current ELISAs are not as sensitive as the VNT (MuÈller et al., 1997), which on comparison vary between 85±98% (Wensvoort et al., 1988; Moser et al., 1996; Colijn et al., 1997; MuÈller et al., 1997). However, a small increase in the herd sample size can secure detection of infected pig holdings easily even at low seroprevalence (Martin et al., 1992). From this point of view, the lowered sensitivity of the ELISA when compared with the VNT is of minor importance if the sample size is adequate. Annexe I of the Council Directive 80/217/EEC prescribes how to interpret the results of the VNT. The use of this test is compulsory in the EU for confirmation of positive results obtained by CSF antibody ELISA. From the results shown in Figs. 5 and 6, it is clear that decision-making on the basis of ELISA results alone is unwise. The specificity of the ELISA can be improved by increasing the cut-off level (Jacobson, 1998) the subsequent decrease in sensitivity should however be carefully weighed. For instance, increasing the cut-off level from 30 to 70% inhibition would decrease the sensitivity with 18% if we use the results presented here. Similar findings have been reported by MuÈller et al. (1996). The VNT is regarded as the `golden standard' assay, it can discriminate sera that react falsely in the ELISA due to BD or BVD virus-neutralising antibodies from sera with CSF virus-neutralising antibodies only. The neutralising peroxidase-linked antibody assay described by Holm Jensen (1981) is the most-commonly used technique in the EU to detect CSF neutralising antibody. The VNT-neutralisation titres can, however, be influenced by the virus strains (CSF, BD and BVD virus strains) used in the assay (MuÈller et al., 1997; Wensvoort et al., 1994; Dekker et al., 1995) resulting in variations in specificity and sensitivity of the VNT. The VNT, being a laborious method, is not suitable for testing large numbers of sera in a short period. However, if applied as such Terpstra et al. (1984) suggest a cut-off titre for the VNT of 1 : 25, interpretation of VNT titres would then be optimised by reducing the number of sera with low cross-reactive titres caused by BD/BVD virus infections or nonspecific factors. The interference of BD or BVD neutralising antibodies with the laboratory diagnosis of CSF has been described by many authors (Liess et al., 1977; Holm Jensen, 1981; Dahle et al., 1987, 1993; Wensvoort et al., 1994) and it is agreed that this can compromise interpretation of ELISA and VNT test results considerably. The specificity of current CSF ELISAs has enabled us to screen large numbers of field sera for

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

197

CSF serum antibody. Specificity is of great importance because, as described above, test results have to be confirmed in the VNT which may still provide ambiguous results. Local BD/BVD virus field isolates circulating in a pig herd can compromise the interpretation of VNT titres if these sera are detected by the ELISA (Wensvoort et al., 1994; Matschullat et al., 1994). This can hinder and delay laboratory decision-making considerably. MuÈller et al. (1997) suggest therefore, a cut-off limit of >1 : 40 for CSF virus neutralising titres as decisive for the presence of a CSF infection in a herd regardless of the titre against BD or BVD virus. In our opinion this conclusion should be carefully weighed because of earlier experience (Wensvoort et al., 1994) and data gathered during the recent outbreak. It is not unlikely that a large percentage of the sera classified as `negative' (Fig. 6) in the VNT contain in fact BD or BVD virus neutralising antibodies, titres remained however undetected because of antigenic differences between the causative field strains and the strains used in the assay (Dekker et al., 1995). If several sera from a single submission are encountered with high inhibition percentages in the ELISA, immediate re-sampling (virological) of the suspected herd is appropriate to speed the detection of a pig holding possibly infected with CSF. Acknowledgements We wish to thank all the technicians involved for their efforts and dedication during the CSF epidemic, and greatly appreciate the assistance of B. Wieman in gathering relevant data from the LMS database. References Anonymous 1980. Council Directive 80/217/EEC of 22 January 1980 introducing community measures for the control of classical swine fever. Afshar, A., Dulac, G.C., Bouffard, A., 1989. Application of peroxidase labelled antibody assays for detection of porcine IgG antibodies to hog cholera and bovine viral diarrhoea viruses. J. Virol. Meth. 23, 253±262. Colijn, E.O., Bloemraad, M., Wensvoort, G., 1997. An improved ELISA for the detection of serum antibodies directed against classical swine fever virus. Vet. Microbiol. 59, 15±25. Dahle, J., Liess, B., Frey, H.-J., 1987. Interspecies transmission of pestiviruses: experimental infections with bovine viral diarrhoea virus in pigs and hog cholera virus in cattle. In: Pestivirus infections of ruminants, Commission of the European Communities, EUR 10238 EN, pp. 195±211. Dahle, J., Schageman, G., Moennig, V., Liess, B., 1993. Clinical, virological and serological findings after intranasal inoculation of pigs with bovine viral diarrhoea virus and subsequent intranasal challenge with hog cholera virus. J. Vet. Med. B 40, 46±54. Dekker, A., Wensvoort, G., Terpstra, C., 1995. Six antigenic groups within the genus pestivirus as identified by cross neutralization assays. Vet. Microbiol. 47, 317±329. Depner, K.R., Rodriguez, , Pohlenz, J., Liess, B., 1996. Persistent classical swine fever virus infection in pigs infected after weaning with a virus isolated during the 1995 epidemic in Germany: clinical, virological, serological, serological and pathological findings. Eur. J. Vet. Pathol. 2(2), 61±66. De Smit, A.J., Terpstra, C., Wensvoort, G., 1994. Comparison of virus isolation methods from whole blood or blood components for early diagnosis of CSF. In: Report on Meeting of the EU National Swine Fever Laboratories, Brussels, 24±25 November 1994, pp. 21±22. Greiser-Wilke, I., Depner, K., Fritzemeier, J., Haas, L., Moennig, V., 1998. Application of a computer program for genetic typing of classical swine fever virus isolates from Germany. J. Virol. Meth. 75, 141±150.

198

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

Holm Jensen, M., 1981. Detection of antibodies against hog cholera virus and bovine viral diarrhea virus in porcine serum: a comparative examination using CF, PLA and NPLA assays. Acta Vet. Scand. 22, 85±98. Holm Jensen, M., 1985. Screening for neutralising antibodies against hog cholera virus and/or bovine viral diarrhea virus in Danish pigs. Acta Vet. Scand. 26, 72±80. Jacobson, R.H., 1998. Validation of serological assays for diagnosis of infectious diseases. Rev. Sci. tech. Off. Int. Epiz. 17(2), 469±486. Kaden, V., HuÈbert, P., Strebelow, G., Lange, E., Steyer, H., Steinhagen, P., 1999. Comparison of different diagnostic methods for the detection of the classical swine fever virus (CSFV) in the early infection period. Berl. Munch. Tierartzl. Wschr. 112, 52±57. Koenen, F., Van Caenegem, G., Vermeersch, J.P., Vandenheden, J., Deluyker, H., 1996. Epidemiological characteristics of an outbreak of classical swine fever in an area of high pig density. Vet. Rec. 12, 367±371. Koenen, F., Vanderhallen, H., Laevens, H., Deluyker, H., 1998. Serological and virological evolution of classical swine fever diagnosis after an experimental infection simulating field conditions. In: Proc. 15th Int. Pig Veterinary Society Congress. Birmingham, England, p. 191. Liess, B., Frey, H.-J., PraÈger, D., 1977. Antibody response of pigs following experimental infection with strains of hog cholera, bovine viral diarrhoea virus. In: Hog Cholera/Classical Swine Fever and African Swine Fever. Commission of the European Communities, EUR 5904 EN, pp. 200±213. Martin, S.W., Shoukri, M., Thorburn, A., 1992. Evaluating the health status of herds on tests applied to individuals. Prev. Vet. Med. 14, 33±43. Matschullat, G., Dahle, J., Roder, B., Moennig, V., Liess, B., 1994. Field infection with BVDV virus in pigs: epidemiology and diagnostic investigations. Dtsch. Tierartzl. Wschr. 101, 22±26. McGoldrick, A., Lowings, J.P., Ibata, G., Sands, J.J., Belak, S., Paton, D.J., 1998. A novel approach to the detection of classical swine fever virus by RT-PCR with a fluorogenic probe (TagMan). J. Virol. Meth. 72, 125±135. Moser, C., Ruggli, N., Tratschin, J.D., Hofmann, M.A., 1996. Detection of antibodies against classical swine fever virus in swine sera by indirect ELISA using recombinant envelope glycoprotein E2. Vet. Microbiol. 51, 41±53. MuÈller, A., Depner, K.R., Liess, B., 1996. Evaluation of a gp 55 (E2) recombinant-based ELISA for the detection of antibodies induced by classical swine fever virus. Dtsch. Tierartzl. Wschr. 103, 449±488. MuÈller, T., Depner, K.R., Burow, J., Ahl, R., Conraths, F.J., Moennig, V., 1997. Comparison of different BVDVstrains for their use in differential diagnosis of classical swine fever Ð an attempt to standardize the neutralization test. Dtsch. Tierartzl. Wschr. 104, 91±96. Pan, I.C., Huang, T.S., Pan, C.H., Chern, Shenq-Yi, Lee, Shu-Hwae, Lin, Y.L., Huang, B.Y., Lin, C.C., Li, N.J., Lin, J.P., Yang, Y.H., Chiu, S.Y., Chang, J.S., Hue, D.K., Lee, H.C., Chang, C.N., 1993. The skin, tongue, andbrain as favorable organs for hog cholera diagnosis by immunofluorescence. Arch. Virol. 131, 475±481. Paton, D.J., Ibata, G., Edwards, S., Wensvoort, G., 1991. An ELISA detecting antibody to conserved pestivirus epitopes. J. Virol. Meth. 31, 24±315. Reed, L.J., Muench, H., 1938. A simple method of estimating fifty percent end points. Am. J. Hyg. 27, 709±716. Ressang, A.A., 1973. Studies on the pathogenesis of hog cholera. I. Demonstration of hog cholera virus subsequent to oral exposure. Zbl. Vet. Med. B 20, 256±271. Terpstra, C., Bloemraad, M., Gielkens, A., 1984. The neutralizing peroxidase-linked assay for detection of antibody against swine fever virus. Vet. Microbiol. 9, 113±120. Terpstra, C., 1987. Epizootiology of swine fever. Vet. Quart. 9 (Suppl. 1), 50±60. Terpstra, C., Wensvoort, G., 1988. Natural infections of pigs with bovine viral diarrhoea virus associated with signs resembling swine fever. Res. Vet. Sci. 19, 137±142. Terpstra, C., Wensvoort, G., 1991. Bovine viral diarrhoea virus infections in pigs. Tijdschr. Diergeneeskd. 166, 943±948. Terpstra, C., Wensvoort, G., 1997. A congenital persistent infection of bovine virus diarhoea in pigs: clinical, virological, virological and immunological observations. Vet. Quart. 19, 97±101. Van Oirschot van, J.T., 1999. Hog Cholera. In: Straw et al. (Eds.), Diseases of Swine, 8th ed. Iowa State University Press, Ames, pp. 159±172. Vilcek, S., Belak, S., 1996. Genetic identification of pestivirus strain F, isolated from pigs, as a border disease virus. J. Virol. Meth. 69, 103±108.

A.J. de Smit et al. / Preventive Veterinary Medicine 42 (1999) 185±199

199

Wensvoort, G., Bloemraad, M., Terpstra, C., 1988. An enzyme immunoassay employing monoclonal antibodies and detecting specifically antibodies to classical swine fever virus. Vet. Microbiol. 17, 129±140. Wensvoort, G., de Smit, A.J., Terpstra, C., 1994. A non-CSFV pestivrus is currently circulating among a number of Dutch pig herds and causing false positive CSFV serology. In: Report on the Meeting of the National Swine Fever Laboratories, 24±25 November 1994, Brussels, pp. 22±24. Wensvoort, G., Terpstra, C., Boonstra, J., Bloemraad, M., Zaane, D. van, 1986. Production of monoclonal antibodies against swine fever virus and their use in laboratory diagnosis. Vet. Microbiol. 12, 8±101. Widjojoatmodjo, M.N., Van Gennip, H.G.P., de Smit, A.J., Moormann, R.J.M., 1999. Comparative sequence analysis of classical swine fever virus isolates from the epizootic in the Netherlands in 1997±1998. Vet. Microbiol. 66, 291±299.