Cs

Ma

b

c

d

ARRAA

KAE(MDA

1

tavgt(tecta(1ut2

Df

0h

Journal of Virological Methods 187 (2013) 103– 109

Contents lists available at SciVerse ScienceDirect

Journal of Virological Methods

jou rn al h om epage: www.elsev ier .com/ locate / jv i romet

omparison of the performance of five different immunoassays to detectpecific antibodies against emerging atypical bovine pestivirus

agdalena Larskaa,b,∗, Mirosław P. Polakb, Lihong Liuc, Stefan Aleniusa, Åse Uttenthald

Department of Clinical Sciences, Swedish University of Agricultural Sciences (SLU), Box 7054, SE-750 07 Uppsala, SwedenNational Veterinary Research Institute, Virology Department, Al. Partyzantów 57, 24-100 Puławy, PolandDepartment of Virology, Immunology and Parasitology (VIP), National Veterinary Institute (SVA), Ulls väg 2B, SE-751 89 Uppsala, SwedenNational Veterinary Institute, Technical University of Denmark, Lindholm, 4771 Kalvehave, Denmark

rticle history:eceived 9 May 2012eceived in revised form 5 September 2012ccepted 10 September 2012vailable online 17 September 2012

eywords:

a b s t r a c t

Bovine pestiviruses represent a considerably variable group. In addition to the two accepted speciesBVDV-1 and BVDV-2, a number of atypical bovine pestiviruses have been detected both in foetal calf seraand in field samples. The sera collected during the initial six weeks of experimental infection of calves withatypical pestivirus, BVDV-1 and a combination of both viruses have been examined by routine and newdiagnostic tests to validate their robustness and sensitivity. As expected, virus neutralization tests usinghomologous virus were able to differentiate the two groups infected by BVDV-1 or atypical pestivirus,

typical bovine pestivirusnzyme-linked immunosorbent assayELISA)

icrosphere immunoassay (MIA)iagnosisntibody detection

whereas the animals inoculated with a mixture of these two viruses had a reaction pattern very similar tothe homologous virus alone. It was found that immunoassays using whole virus and polyclonal antibodiesare the most robust, but all tests examined were able to detect antibodies also from cattle infected withatypical pestivirus a few weeks after infection. The detection, however, was at a lower level and slightlydelayed. Statistical validation of the threshold suggested by the manufacturer showed that in some casesthe reduction of the cut-off values would improve the test sensitivity.

. Introduction

Atypical bovine pestiviruses are a group of viruses whichogether with two bovine viral diarrhoea virus species (BVDV-1nd BVDV-2), border disease virus (BDV) and classical swine feverirus (CSFV) form the Pestivirus genus of Flaviviridae family. Theroup was referred to previously as Hobi-like pestiviruses fromhe precursor strain isolated from Brazilian foetal bovine serumFBS) (Schirrmeier et al., 2004). Phylogenetic and antigenic rela-ionship of atypical bovine pestiviruses with BVDV and increasingvidence of further intercontinental spread of the viruses amongattle (Decaro et al., 2011; Xia et al., 2011) raised a discussion overhe emergence of a new BVDV species (Liu et al., 2009). Detection oftypical pestiviruses requires introduction of new testing protocolsLiu et al., 2008), since RT-PCR assays used commonly (Vilcek et al.,994) fail to detect them (Stahl et al., 2009). Information on the

se of serological tests designed to detect BVDV-1 and BVDV-2 forhe detection of those atypical pestiviruses is limited (Decaro et al.,012; Kampa et al., 2007; Larska et al., 2012; Schirrmeier et al.,

∗ Corresponding author at: National Veterinary Research Institute, Virologyepartment, Al. Partyzantów 57, 24-100 Puławy, Poland. Tel.: +48 818893068;

ax: +48 818862595.E-mail address: m.larska@piwet.pulawy.pl (M. Larska).

166-0934/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.jviromet.2012.09.010

© 2012 Elsevier B.V. All rights reserved.

2004). Virus neutralization test and enzyme-linked immunosor-bent assay (ELISA) (Schrijver and Kramps, 1998) are used mostcommonly for routine serological diagnosis of BVDV. The pres-ence of specific antibodies is considered a good indicator of naturalinfection in tested herds, both on the individual-level (serum ormilk sample) and herd-level (bulk milk), or for estimation of theprotective immunity after vaccination. The envelope glycoproteinE2 is considered the main neutralizing antigen which is relativelyconserved provides cross-neutralization between all pestiviruses(Schirrmeier et al., 2004). Other viral proteins, such as: the enve-lope RNase, Erns and non-structural protein NS3 (previously knownas p80/p125) are also used as targets for pestivirus specific anti-body responses. Virus neutralization allows quantitation of theantibodies and is considered as the ‘gold standard’ method, how-ever it requires well-equipped facilities, contamination control andskilled personnel. Commercial ELISAs provide a good alternativefor virus neutralization test, and are chosen more frequently asthey allow fast and specific screening of large numbers of samples.Modern technologies, such as microsphere immunoassay (MIA), areintroduced in BVDV diagnosis in response to a demand for rapid andefficient tests (Xia et al., 2010).

The aim of this study was to evaluate the ability of currentdiagnostic methods to detect antibodies against atypical bovinepestiviruses. A panel of sera collected from naive calves infectedexperimentally with an atypical bovine pestivirus, BVDV-1 and the

1 logica

mre

2

2

smbsao5c2dfc

2

assecpcoi

2

2

S(suatstfsvwt

2

STiEuacpC

04 M. Larska et al. / Journal of Viro

ixture of BVDV-1 and atypical pestivirus were used. Increasingisk of the spread of atypical BVDVs prompts the search for thefficient tools supporting testing and control programs.

. Materials and methods

.1. Experimental design

Fifteen healthy calves, negative for the presence of BVDV andpecific antibodies were divided into three groups of five ani-als each, and were inoculated with the same dose of atypical

ovine pestivirus strain Th/04 KhonKaen (Group A); BVDV-1atrain Horton 916 (Ho916) (Group B); and a mixture of Ho916nd Th/04 KhonKaen (Group C). The procedure has been previ-usly described in detail (Larska et al., 2012). A fourth group of

calves (control group) was mock-inoculated. Blood samples wereollected from all animals before the inoculation and at 7, 14, 21,8, 35 and 42 day post-inoculation (dpi). The animals were han-led according to the protocol approved by the Ethics Committeeor Animal Experiments by the Ministry of Science and Higher Edu-ation in Poland (No. 79/2010).

.2. Virus neutralization test

To detect and quantify neutralizing antibodies against homo-nd heterologous BVDV strains, all sera were tested against BVDV-1train DK258 (used as the reference strain in Danish BVDV controlcheme) in Madin–Darby bovine kidney (MDBK) cells (Uttenthalt al., 2005) and Th/04 KhonKaen strain in bovine turbinate (BT)ells. The test was performed according to the protocol describedreviously (Larska et al., 2012). The neutralizing titres were cal-ulated using the Spearman–Kärber method. The inhibitory effectf serum dilution above 1:10 was regarded as positive and as anndicator of seroconversion.

.3. Indirect enzyme-linked immunosorbent assays (ELISAs)

.3.1. i-ELISA-1The Svanovir BVDV-Ab ELISA kit (Svanova Biotech, Uppsala,

weden) is designed for detection of BVDV specific antibodiesIgG1) in bovine serum or milk samples. The sensitivity andpecificity relative to virus neutralization test provided by the man-facturer is 100% and 98.2%, respectively. The assay was performedccording to the manufacturer’s protocol with 1-h incubation ofhe serum under test in the coated plate. Prior to the test, serumamples were diluted 1:25. The results were calculated first by sub-racting optical density (OD) value of the reference control samplerom OD value of the tested sample (corrected OD; COD) and pre-ented as percentage positivity (PP) by dividing the sample CODalue by positive reference sample COD value. The cut-off valueas set at 15 PP as recommended by the manufacturer of the

est.

.3.2. i-ELISA-2The HerdChek BVDV Antibody Test Kit (IDEXX, Liebefeld-Bern,

witzerland) was the second indirect ELISA-based assay evaluated.he assay is characterized by high specificity and high sensitiv-ty for detecting anti-BVDV-1 and anti-BVDV-2 antibodies. The

LISA was performed according to the manufacturer’s protocolsing 1:5 serum dilution and 1.5 h-incubation time of serum withntigen-coated plate. The results were expressed as corrected opti-al density (COD) values calculated by subtracting COD of the testositive control from the COD value of the sample tested. TheOD ≥ 0.3 was interpreted as positive.

l Methods 187 (2013) 103– 109

2.4. Blocking ELISAs

2.4.1. b-ELISA-1The Svanovir BVDV p80-Ab ELISA (Svanova Biotech, Uppsala,

Sweden) is a multispecies (bovine, ovine and caprine) blockingassay which consists of microplates coated with BVD NS3 (p80) pro-tein and a peroxidase labelled specific to NS3 monoclonal antibody.According to the manufacturer the test detects antibodies againstBVDV p80 (NS3) protein of BVDV-1, BVDV-2 and atypical pestivirusin bovine, caprine and ovine samples. The manufacturer’s protocolwas followed. The serum samples were diluted 1:10 in phosphatebuffered saline (PBS) and incubated for 1 h in the antigen-coatedplate. The results are shown as percent inhibition (PI) and the 45 PIwas implemented as the cut-off value as recommended by the kitmanufacturer.

2.4.2. b-ELISA-2A liquid phase blocking ELISA for BVDV was performed

(Rønsholt et al., 1997). The test is based on: (1) swine anti-BVDVcapture IgG produced by infection of pigs initially with BVDV-1 andinoculated subsequently with CSFV Brescia; (2) rabbit-anti-BVDVproduced by immunization of rabbits with a BVDV antigen adaptedto primary rabbit cells; (3) crude BVDV-1 antigen produced in cellcultures. Briefly, the plates were coated by swine anti-BVDV cap-ture IgG and blocked. The bovine serum sample was added shortlybefore the addition of an equal volume of BVDV antigen and incu-bated overnight. The next day plates were washed and bound BVDVantigen was detected by rabbit anti-BVDV and swine anti-rabbitantibodies. Negative and positive serum control wells were usedto calculate the percentage inhibition (PI) and values below 50 PIwere considered as positive. The sensitivity of the test was 96–97%,whereas the specificity was 94–98% compared to the virus neutral-ization test, according to the manufacturer.

2.5. Microsphere immunoassay (MIA)

A microsphere immunoassay developed recently using recom-binant Erns protein of Th/04 KhonKaen strain as the antigen wasused for detection of antibodies (Vijayaraghavan et al., 2012).Briefly, five-fold diluted serum samples were incubated with theErns antigen-coupled-microspheres at 37 ◦C for 30 min on a plateshaker. After washing, a biotinylated anti-bovine IgG (JacksonImmunoResearch, West Grove, USA) was added and incubatedfor another 30 min. After subsequent washing, a streptavidin-R-phycoerythrin conjugate (ProZyme, Hayward, Canada) was addedand incubated at 37 ◦C for 30 min on a plate shaker. After washingtwice, the microspheres were resuspended in 100 �l of PBS with 1%BSA and analyzed by a Luminex 200 analyzer. The median fluores-cence intensity (MFI) was calculated based on the measurement of100 beads per sample. The cut-off value of the test was calculatedusing Receiver operating curve analysis (ROC) with the Medcalcsoftware (MedCalc Software, Mariakerke, Belgium).

2.6. Statistical analysis

All statistical analyses were performed using STATA softwareversion 11 (StataCorp, College Station, Texas, USA). Statistical sig-nificance between animal groups within each immunoassay resultsets was analyzed using ANOVA followed by Tukey’s HSD (Hon-estly Significant Difference) and Dunnett’s test. The results fromindividual sampling days were analyzed as repeated measures.

Tukey’s test was used to calculate a critical value to evaluatethe significance of the difference between two pairs of means,which was then compared to the Tukey critical value, whileDunnett’s test compared inoculated groups with the control

M. Larska et al. / Journal of Virological Methods 187 (2013) 103– 109 105

Fig. 1. Time-dependent kinetics of antibody response from pre-inoculation day (0 dpi) to 42 dpi according to mean neutralizing antibody titres (NT) using Th/04 KhonKaenstrain (A) and BVDV-1 strain (B); mean percent positivity (PP) in i-ELISA-1 (C): mean corrected optical densities in i-ELISA-2 (D); mean percent inhibition in two blocking tests:b-ELISA-1 (E); b-ELISA-2 (the y scale was reversed in order to compare graphs) (F); and mean of median fluorescence intensity (MFI) (G) in the calves from: Group A inoculatedwith atypical bovine pestivirus Th/04 KhonKaen strain (�); Group B inoculated with BVDV-1 strain Ho916 (�); Group C inoculated with a mixture of Th/04 KhonKaen and

Ho916 strains (�); and control group – uninfected animals ( ). The cut-off values are marked as dashed lines and the vertical bars represent standard error of mean (SEM).Statistically significant variation (p < 0.05) between Group A (a), Group B (b) and Group C (c) in respect to control group were analyzed using two-sided Dunnett test, whilethe differences between all the groups of inoculated calves (d = A versus B; e = A versus C: f = B versus C) were studied using Tukey (HSD) test. Error bars indicate standarddeviation (SD) of the results.

1 logical Methods 187 (2013) 103– 109

gp

tpata(shaawar

aar

3

3

socctgovgtawwFori

ottM(vsaGwiicBcsitEtp

Fig. 2. Comparison of the performance of ELISAs versus homologous virus neutral-ization test. (A) Se – sensitivity = proportion (%) of the true positive samples whichare identified correctly by the test; Sp – specificity = proportion (%) of the true nega-tive samples which are identified correctly by the test; (B) AUC – area under receiveroperating curve (ROC) analysis for perfect discrimination between positive and neg-ative animals. ROC analysis is a combined measure of sensitivity and specificity forassessing inherent validity of a diagnostic test. The maximum AUC of 1.00 meansthat the diagnostic test is perfect in differentiating true positive from true negativeserum samples, while 0.5 is the minimum value. The results were considered ‘truepositive’ when positive in virus neutralization test using atypical bovine pestivirusTh/04 KhonKaen strain in Group A (inoculated with Th/04 KhonKaen); and BVDV-1strain for Groups B (inoculated with BVDV-1 strain Ho916) and Group C (inocu-lated with the mixture of Th/04 KhonKaen and BVDV-1 strain Ho916). The results

06 M. Larska et al. / Journal of Viro

roup. Differences were considered statistically significant at < 0.05.

In order to establish the sensitivity and specificity of the assays,he results were classified into positive and negative using appro-riate cut-off values. The sensitivity and specificity were expresseds the percentage of the samples found positive and negative in allrue (detected by virus neutralization test) positive and true neg-tive samples, respectively. Areas under receiver operating curveAUC) with a range of 0.5–1.0 were calculated for each immunoas-ay as additional accuracy measure. The neutralization results usingomologous virus were considered as the ‘gold standard’ results inll analyses. BVDV-1 was the reference for Group C, and when allnimals were considered. The optimal cut-off values for each testere determined by ROC analysis using continuous values of ELISA

nd MIA readings in regard to the binomial (positive/negative)esults of the ‘gold standard’.

Agreement and correlation between ELISA and MIA readingsnd neutralizing titres against homologous pestivirus strain weressessed by linear regression analysis and by Spearman’s rank cor-elation. The relationship was considered significant at p < 0.05.

. Results

.1. Kinetics of seroconversion

All samples from the control group were negative in allerological tests throughout the experiment. As described previ-usly (Larska et al., 2012), virus neutralization (Fig. 1A and B)onfirmed the infection of all animals inoculated without anyross-contamination between the groups. The animals from allhe inoculated groups began to seroconvert against the homolo-ous viruses at 14 dpi (4 animals out of 5 from Group A and 2/5f Group B). The mean antibody titres against Th/04 KhonKaenirus (Fig. 1A) in the sera of calves inoculated with the homolo-ous strain increased from 54 at 14 dpi to 1763 at 42 dpi, whilehe sera of calves from Group B neutralized atypical pestivirust the significant level (p < 0.05) only after 28 dpi. The resultsere reversed in virus neutralization using BVDV-1 (Fig. 1B) inhich the titres were significantly higher in Groups B and C.

our out of five animals from Group A cross-neutralized heterol-gous pestivirus at the mean titre of 17 at 42 dpi while one calfemained negative for the presence of BVDV-1 specific antibod-es.

Similarly to the virus neutralization test results, the increasef antibodies detected by other methods was correlated withhe post-inoculation time. The earliest detection of antibodies inhe Group A was observed at 14 dpi in the b-ELISA-2 (Fig. 1F),

IA (Fig. 1G), and indirect i-ELISA-1 (in 2 out of 5 animals)Fig. 1C), which was comparable to the results of Th/04 KhonKaenirus neutralization test. Using i-ELISA-2 (Fig. 1D) seroconver-ion was detected first in two animals from Group A at 21 dpind in all animals at 28 dpi. Detection of specific antibodies inroup A was the most delayed by the end of the experimenthen b-ELISA-2 test was used (Fig. 1E), despite the fact that

t detected seroconversion in two remaining inoculated groupsn the third week post-inoculation. All assays showed high effi-iency, comparable to the ‘gold standard’ in early detection ofVDV-1 specific antibodies in Groups B and C. Most ELISAs indi-ated that the calves, which were inoculated with both viruses,howed kinetics of immune response comparable to the calvesnfected with BVDV-1. Only the b-ELISA-2 detected antibodies

o BVDV-1 and atypical pestivirus equally well. Immunoassays i-LISA-a, i-ELISA-2 and b-ELISA-1 detected preferentially antibodieso BVDV-1, whereas the MIA detected preferentially the atypicalestivirus.

presented for the serological tests in last column (All groups) included inoculatedanimals (Groups A–C) and control group in respect to virus neutralization test usingBVDV-1 strain as reference.

3.2. Evaluation of test sensitivity and specificity

The specificity, the sensitivity (Fig. 2A) and the values of areasunder ROC curve (AUC) (Fig. 2B) were generated for all the ELISAsand MIA based on the neutralization test results with homolo-gous pestivirus strains (Th/04 KhonKaen virus in Group A andBVDV-1 strain in Groups B and C used as the reference results).One hundred percent sensitivity for detecting Th/04 KhonKaenvirus specific antibodies in Group A was obtained by b-ELISA-2and MIA, notwithstanding lower specificities of 81.8% and 91.0%,respectively. Other assays were characterized by 100% specificity,however the sensitivity varied from only 25% for b-ELISA-2 and

BVDV-1 neutralization test to the highest value of 83.3% for i-ELISA-1. Most assays detected 100% of seropositive animals fromGroup B, which were categorized as positive in BVDV-1 neutraliza-tion test, except for b-ELISA-1 which had only 72% sensitivity. The

M. Larska et al. / Journal of Virological Methods 187 (2013) 103– 109 107

Table 1Sensitivity (Se), specificity (Sp) and area under ROC curve (AUC) calculated using ROC optimized cut-off values (Opt C/O) for each test using sera from animals infected withatypical pestivirus (Group A) including control group (n = 60), and sera from all uninfected and infected animals (n = 120).

Serologic test Manufacturer’s cut-off valuea Atypical bovine pestivirus infected calves All groups

Opt C/O Se/Sp (%) AUC Opt C/O Se/Sp (%) AUC

i-ELISA-1 15.0 13.0 91.6/100.0 0.93 21.0 98.1/80.7 0.91i-ELISA-2 0.3 0.13 95.8/91.0 0.98 = 90.4/88.6 0.90b-ELISA-1 45.0 19.0 83.3/81.8 0.83 30.0 90.4/91.0 0.91b-ELISA-2 50.0 37 100.0/81.8 0.91 6.7 100.0/84.1 0.89MIA 2800 1446–8244.5 100.0/91.0 0.95 = 100.0/75.0 0.88

= the manufacturer.LISA-2).

spmeBar

3

uaafwaDEsTae

3E

bBbaifgttlacbttvtwbBvtb

Table 2Spearman’s rank correlations coefficients (�) between the titres of the antibodiesagainst Th/04 KhonKaen (in sera from Groups A, C and control; n = 105) and BVDV-1(in sera from Groups B, C and control; n = 105) and ELISA results. All the correlationswere significant (p < 0.0001).

Assay Th/04 KhonKaen antibody titre BVDV-1 antibody titre

i-ELISA-1 0.81 0.87i-ELISA-2 0.82 0.92b-ELISA-1 0.75 0.85b-ELISA-2 0.84 0.82MIA 0.84 0.62

Indicates when the optimized cut-off values were equal to the cut-off point set bya The value corresponds to PP (i-ELISA-1); COD (i-ELISA-2); PI (b-ELISA-1 and b-E

pecificity of the tests was lower with respect to the ‘gold standard’,ossibly due to the lower sensitivity of the latter. The AUC for theethods detecting atypical pestivirus antibodies were the high-

st reaching 0.93 for MIA and the lowest (0.62) for b-ELISA-1 andVDV-1 neutralization test. These values in calves inoculated with

single BVDV-1 strain or mixed strains were significantly higheranging from 0.81 for b-ELISA-2 to 1.00 for i-ELISA-2.

.3. Evaluation of optimal cut-off values

In order to test if implementing new optimized cut-off val-es would improve the sensitivities, specificities of the assaysnd would shorten the time of detection of seroconversion innimals inoculated with atypical pestivirus, ROC analysis was per-ormed (Table 1). The results of all sera from the control groupere included in each analysis to improve the statistical power

nd to avoid false positive results with optimized cut-off values.ecreasing the cut-off values by 2 PP in i-ELISA-1; by 0.17 COD in i-LISA-2; and by 26 PI in b-ELISA-1 improved the sensitivity and thepecificity for detecting the antibodies against atypical pestivirus.wo tests: MIA and b-ELISA-2 showed more flexible cut-offs, whichllowed the values to fluctuate within a wider range without influ-ncing their performance.

.4. Evaluation of the correlation between neutralizing titres andLISA results

The linear regression model showed that the correlationsetween specific antibody titres against Th/04 KhonKaen andVDV-1, and ELISA and MIA results were significant (at proba-ility of p < 0.0001). The highest correlations between the specificntibody titre against BVDV-1 and atypical pestivirus and serolog-cal assays were found for the i-ELISA-2 (Pearson’s coefficients r2

rom 0.68) and b-ELISA-1 (r2 from 0.67) in each of the inoculatedroups and independently of the group. The correlation betweenhe tests declined gradually with increasing neutralizing antibodyitres (in connection to dpi). The values of all ELISAs were corre-ated significantly with the neutralizing titres against BVDV-1 andtypical pestivirus (p < 0.0001) when tested by the Spearman rankorrelation coefficients (�) (Table 2). Anti-Th/04 KhonKaen anti-ody titres showed comparable correlations with all ELISAs, withhe highest correlation to b-ELISA-2 (� = 0.84). The lowest correla-ion was observed between Th/04 KhonKaen antibody titres andirus neutralization using BVDV-1 strain (� = 0.65). The correla-ions between the BVDV-1 antibody titres and the ELISA resultsere also significant (p < 0.0001) with � values between 0.82 for

-ELISA-2 and 0.92 for i-ELISA-2. The BVDV-1 antibody titres in the

VDV-1 antisera showed higher correlation (� = 0.87) to the titres inirus neutralization using Th/04 KhonKaen strain, which suggestedhat Th/04 KhonKaen strain cross-neutralized BVDV-1 antibodiesetter than vice versa.

BVDV-1 NT 0.65 1.00Th/04 KhonKaen NT 1.00 0.87

4. Discussion

Atypical bovine pestiviruses were described first as contami-nants of fetal bovine serum originating from Brazil (Schirrmeieret al., 2004; Peletto et al., 2012) and later they were confirmed incattle and buffalo in the same country (Stalder et al., 2005; Cortezet al., 2006). Further studies showed that the spread of atypicalpestiviruses is much wider (Xia et al., 2011), corresponding to thefrequent incidence of BVDV-1 and to a lesser extend to BVDV-2in cattle population all over the word. Considering the intensityof the global trade of cattle, semen or bovine bio-products, fur-ther transmission of atypical pestiviruses was a matter of time,and dependent on the sensitivity of the available diagnostic meth-ods. Recently, the atypical pestivirus was detected in Europe incalves from a herd in which an outbreak of severe respiratory dis-ease occurred (Decaro et al., 2011). This fact raised the question ifsuitable detection assays are available. The sensitivity of the ‘goldstandard’ virus neutralization test is influenced significantly by thehomology of predominating field viruses and the reference strainused (Sandvik, 2005). The sensitivity of this method also dependslargely on qualifications of the laboratory staff who may face prob-lems regarding sample contamination, cell culture, and handlingof live virus. The best known alternative methods are the antibodyELISAs, however, these commercial assays are based on BVDV-1antigen which may influence detection of other bovine pestiviruses.Most ELISAs give comparable and compatible results with the ref-erence test (Beaudeau et al., 2001; Mahecha et al., 2011; Patonet al., 1991) and are capable of detecting seroconversion earlieror at a similar time point to virus neutralization test (Durhamand Hassard, 1990), which should be considered in quarantineprocedures. Detection of pestivirus antibodies in cattle by ELISA,however does not provide any information about the pestivirustype involved unless cross-neutralization tests are performed.

The results of the present study demonstrated that all assaysexamined could detect specific antibodies against atypical pes-tivirus Th/04 KhonKaen strain, but many of the methods had a

delay in the detection of seroconversion to Th/04 KhonKaen. TheMIA had a good sensitivity for atypical pestivirus, but it shouldbe kept in mind that the antigen used in this study was based on

1 logica

Tiaiadtatdewtosudotvatehlr1vsoToisB2antwsvHA(eamDwhaEi2liqebd(U

ot

08 M. Larska et al. / Journal of Viro

h/0 KhonKaen whereas all the ELISAs were based on BVDV-1. Its important to search for diagnostic test, which would be suit-ble for early detection of antibodies against the virus. Since thenitial host immune response is based on the production of IgMntibodies, there is a growing need for assays that are indepen-ent of IgG detection. Therefore, virus neutralization is expectedo result in rapid detection of the virus, whereas, the tests whichre based on recognition of the proteins expressed in the laterime periods postinfection such as the NS3 protein, may delay theetection of the infection and seroconversion. Thus, if the presentxperiment had a longer duration, the appearance of IgG antibodiesith high avidity would supposedly change the resulting sensi-

ivity of some of the tests. The pattern of antibody developmentbserved in all tests in all inoculated groups was consistent witheroconversion described previously after experimental and nat-ral infections (Decaro et al., 2011; Kramps et al., 1999). A clearifferentiation was seen in virus neutralization test with the usef the homologous and heterologous virus. Neutralizing antibodyitres of sera from cattle infected with both BVDV-1 and atypicalirus showed that the homologous part of the virus inoculum wasble to induce the production of antibodies in calves in spite ofhe interference by the heterologous virus. Despite these differ-nces, stronger cross-reactivity against homologous in respect toeterologous strains was observed, which is consistent with ear-

ier studies (Couvreur et al., 2002; Schirrmeier et al., 2004). Theesults of the current study showed that the blocking b-ELISA-, based on detection of antibodies against the conservative NS3iral protein, had the lowest sensitivity and the latest detection oferoconversion. Similar observations were described previously byther research groups (Bauermann et al., 2012; Paton et al., 1991).his could be explained in part by mutations in the epitope siter the specific differences in virus affinity to monoclonal antibod-es (mAbs) (Schirrmeier et al., 2004). Atypical bovine pestiviruseseem to share more mAb domains and neutralizing activity withVDV-2 than with BVDV-1 (Bauermann et al., 2012; Kampa et al.,007; Schirrmeier et al., 2004). Low sensitivity for detecting anti-typical pestivirus antibodies was observed with BVDV-1 viruseutralization when compared to b-ELISA-1. This finding stresseshat virus neutralization could easily fail to detect animals infectedith another bovine pestivirus, so even the ‘gold standard’ method

hould be interpreted with caution. Interestingly, Th/04 KhonKaenirus cross-neutralized sera of calves inoculated with the BVDV-1o916 strain to a greater extent than BVDV-1 virus sera of Group

calves. It is consistent with the findings of Ståhl and co-workers2009), who showed that the atypical pestiviruses have the high-st cross-species neutralizing reactivity among bovine pestiviruses,nd may indicate falsely the presence of BVDV-1 strains, com-only used in diagnostic methods and for vaccine production.evelopment of an immunoassay using microspheres conjugatedith atypical pestivirus Erns protein (Xia et al., 2010) resulted inighly sensitive and specific assay, which, however, did not enable

reliable, specific discrimination of the pestivirus. However, therns protein-based ELISA was successfully used for distinguish-ng between BVDV and CSFV infection in pigs (Langedijk et al.,001). MIA format using Luminex platform allows multiplex sero-

ogical diagnosis (Anderson et al., 2011) and has a great potentialn future surveillance of infectious diseases appearing more fre-uently as disease complexes of multifactorial aetiology (Raaperit al., 2012). Additionally, pestivirus species specific NS3 and Erns-ased immunoassays are considered as the candidates tests forifferentiating between infected and vaccinated animals (DIVA)Álvarez et al., 2012; Langedijk et al., 2001; Raue et al., 2011;

ttenthal et al., 2010).

The sensitivity and specificity of serological tests are dependentn the selected cut-off values. While a high cut-off value eliminateshe false positive results, it will also decrease the sensitivity of the

l Methods 187 (2013) 103– 109

test leading to a risk of false negative results. On the other hand,low cut-off values can result in increased sensitivity, yet increas-ing the possibility of obtaining false positive results. The assaystested for the detection of antibodies against atypical pestiviruswere characterized by fairly high sensitivity and specificity (exceptfor b-ELISA-1). The performance of the tests could be improvedfurther by optimizing the cut-off values, however the results pre-sented here should be interpreted with caution due to the limitednumber of samples used. Lower specificity observed for b-ELISA-2 was due to some positive reactions obtained for sera collectedbefore the inoculation, despite the fact that the animals were neg-ative in neutralization test. The persistence of maternal antibodiesin calves which originated from herds without known BVDV statuscould provide a possible explanation. Good linear and rank corre-lations between the values generated by the immunoassays testedand virus neutralizing titres indicated a possible use of bulk milksamples for herd testing, as one of the ways to screen and controlpestivirus infections in cattle (Beaudeau et al., 2001; Kramps et al.,1999).

The results of this study have shown that all immunoassays eval-uated were suitable for detection of antibodies raised in calvesinoculated with atypical pestivirus, however, only some of theassays could be used for the early detection of the seroconver-sion. Two ELISAs (i-ELISA-1 and b-ELISA-2) and novel MIA wereconsidered to be superior (even in respect to the BVDV-1 virusneutralization test) for the routine detection of atypical bovinepestivirus antibodies in cattle. The analysis of the optimal cut-offvalues suggested that cautious result interpretation may diminishthe possibility of obtaining false results.

Acknowledgements

The study was supported by European FP6 Network of Excel-lence EPIZONE (Contract no FOOD-CT-2006-016236). We wouldlike to thank Heidi Elbrink from DTU-VET at Lindholm, Denmark;Maj Hjort and Balaje Vijayaraghavan at SLU in Uppsala, Sweden,and Małgorzata Głowacka at NVRI in Pulawy, Poland for the helpwith performing the assays.

References

Álvarez, M., Donate, J., Makoschey, B., 2012. Antibody responses against non-structural protein 3 of bovine viral diarrhoea virus in milk and serum samplesfrom animals immunised with an inactivated vaccine. Veterinary Journal 191,371–376.

Anderson, S., Wakeley, P., Wibberley, G., Webster, K., Sawyer, J., 2011. Develop-ment, evaluation of a Luminex multiplex serology assay to detect antibodies tobovine herpes virus 1, parainfluenza 3 virus, bovine viral diarrhoea virus andbovine respiratory syncytial virus, with comparison to existing ELISA detectionmethods. Journal of Immunological Methods 366, 79–88.

Bauermann, F.V., Flores, E.F., Ridpath, J.F., 2012. Antigenic relationships betweenbovine viral diarrhea virus 1, 2 and HoBi virus. Possible impacts on diagnosis,control. Journal of Veterinary Diagnostic Investigation 24, 253–261.

Cortez, A., Heinemann, M.B., Castro, A.M.M.G., Soares, M.S., Pinto, A.M.V., Alfieri,A.A., Flores, E.F., Leite, R.C., Richtzenhain, L.J., 2006. Genetic characterization ofBrazilian bovine viral diarrhea virus isolates by partial nucleotide sequencing ofthe 5′-UTR region. Pesquisa Veterinaria Brasileira 26, 211–216.

Couvreur, B., Letellier, C., Collard, A., Quenon, P., Dehan, P., Hamers, C., Pastoret, P.P.,Kerkhofs, P., 2002. Genetic, antigenic variability in bovine viral diarrhea virus(BVDV) isolates from Belgium. Virus Research 85, 17–28.

Decaro, N., Lucente, M.S., Mari, V., Cirone, F., Cordioli, P., Camero, M., Sciarretta,R., Losurdo, M., Lorusso, E., Buonavoglia, C., 2011. Atypical pestivirus andsevere respiratory disease in calves in Europe. Emerging Infectious Diseases 17,1549–1552.

Decaro, N., Mari, V., Lucente, M.S., Sciarretta, R., Moreno, A., Armenise, C., Losurdo, M.,Camero, M., Lorusso, E., Cordioli, P., Buonavoglia, C., 2012. Experimental infec-tion of cattle, sheep and pigs with ‘Hobi’-like pestivirus. Veterinary Microbiology155, 165–171.

Durham, P.J.K., Hassard, L.E., 1990. An enzyme-linked immunosorbent assay (ELISA)for antibodies to bovine viral diarrhea virus. Veterinary Microbiology 22, 1–10.

Kampa, J., Ståhl, K., Renström, L.H., Alenius, S., 2007. Evaluation of a commercial Erns-capture ELISA for detection of BVDV in routine diagnostic cattle serum samples.Acta Veterinaria Scandinavica 49, 1–7.

logica

K

L

L

L

L

M

P

P

R

R

R

S

M. Larska et al. / Journal of Viro

ramps, J.A., van Maanen, C., van de Wetering, G., Stienstra, G., Quak, S., Brinkhof, J.,Rønsholt, L., Nylin, B., 1999. A simple, rapid, reliable enzyme-linked immunosor-bent assay for the detection of bovine virus diarrhoea virus (BVDV) specificantibodies in cattle serum, plasma and bulk milk. Veterinary Microbiology 64,135–144.

angedijk, J.P.M., Middel, W.J.G., Meloen, R.H., Kramps, J.A., de Smit, J.A., 2001.Enzyme-linked immunosorbent assay using a virus type-specific peptide basedon a subdomain of envelope protein Erns for serologic diagnosis of pestivirusinfections in swine. Journal of Clinical Microbiology 39, 906–912.

arska, M., Polak, M.P., Riitho, V., Strong, R., Belák, S., Alenius, S., Uttenthal, Å., Liu,L., 2012. Kinetics of single and dual infection of calves with an Asian atyp-ical bovine pestivirus and a highly virulent strain of bovine viral diarrhoeavirus 1. Comparative Immunology, Microbiology and Infectious Diseases 35,381–390.

iu, L., Xia, H., Wahlberg, N., Belák, S., Baule, C., 2009. Phylogeny, classification andevolutionary insights into pestiviruses. Virology 385, 351–357.

iu, L., Xia, H., Belák, S., Baule, C., 2008. A TaqMan real-time RT-PCR assay for selec-tive detection of atypical bovine pestiviruses in clinical samples and biologicalproducts. Journal of Virological Methods 154, 82–85.

ahecha, F.O.L., Ogas Castells, M.L., Combessies, G., Lavoria, M.A., Wilda, M.,Mansilla, F.C., Seki, C., Grigera, P.R., Capozzo, A.V., 2011. Single dilution avidity-blocking ELISA as an alternative to the bovine viral diarrhea virus neutralizationtest. Journal of Virological Methods 175, 228–235.

aton, D.J., Ibata, G., Edwards, S., Wensvoort, G., 1991. An ELISA detecting anti-body to conserved pestivirus epitopes. Journal of Virological Methods 31,315–324.

eletto, S., Zuccon, F., Pitti, M., Gobi, E., De Marco, L., Caramelli, M., 2012. Detection,phylogenetic analysis of an atypical pestivirus, strain IZSPLV To. Research inVeterinary Science 92, 147–150.

aaperi, K., Bougeard, S., Aleksejev, A., Orro, T., Viltrop, A., 2012. Association ofherd BRSV and BHV-1 seroprevalence with respiratory disease, reproductiveperformance in adult dairy cattle. Acta Veterinaria Scandinavica 54, 4.

aue, R., Harmeyer, S.S., Nanjiani, I.A., 2011. Antibody responses to inactivated vac-cines, natural infection in cattle using bovine viral diarrhoea virus ELISA kits.Assessment of potential to differentiate infected and vaccinated animals. Vet-erinary Journal 187, 330–334.

ønsholt, L., Nylin, B., Bitsch, V., 1997. A BVDV antigen-, antibody blocking ELISA(DVIV) system used in a Danish voluntary eradication program. In: 3rd PestivirusSymposium, Lelystad, Netherlands, pp. 150–153.

andvik, T., 2005. Selection and use of laboratory diagnostic assays in BVD controlprogrammes. Preventive Veterinary Medicine 72, 3–16.

l Methods 187 (2013) 103– 109 109

Schirrmeier, H., Strebelow, G., Depner, K., Hoffmann, B., Beer, M., 2004. Genetic andantigenic characterization of an atypical pestivirus isolate, a putative memberof a novel pestivirus species. Journal of General Virology 85, 3647–3652.

Stalder, H.P., Meier, P., Pfaffen, G., Wageck-Canal, C., Rüfenacht, J., Schaller, P.,Bachofen, C., Marti, S., Vogt, H.R., Peterhans, E., 2005. Genetic heterogeneity ofpestiviruses of ruminants in Switzerland. Preventive Veterinary Medicine 72,37–41.

Schrijver, R.S., Kramps, J.A., 1998. Critical factors affecting the diagnostic reliability ofenzyme-linked immunosorbent assay formats. Revue Scientifique Et TechniqueDe L Office International Des Epizooties 17, 550–561.

Uttenthal, Å., Parida, S., Rasmussen, T.B., Paton, D.J., Haas, B., Dundon, W.G., 2010.Strategies for differentiating infection in vaccinated animals (DIVA) for foot-and-mouth disease, classical swine fever and avian influenza. Expert Review ofVaccines 9, 73–87.

Uttenthal, Å., Stadejek, T., Nylin, B., 2005. Genetic diversity of bovine viral diar-rhoea viruses (BVDV) in Denmark during a 10-year eradication period. Actapathologica, microbiologica, et immunologica Scandinavica 113, 536–541.

Vijayaraghavan, B., Xia, H., Harimoorthy, R., Liu, L., Belák, S., 2012. Evaluation ofenvelope glycoprotein Erns of an atypical bovine pestivirus as antigen in amicrosphere immunoassay for the detection of antibodies against bovine viraldiarrhea virus 1 and atypical bovine pestivirus. Journal of Virological Methods158, 193–198.

Vilcek, S., Herring, A.J., Herring, J.A., Nettleton, P.F., Lowings, J.P., Paton, D.J., 1994.Pestiviruses isolated from pigs, cattle, sheep can be allocated into at least threegenogroups using polymerase chain reaction, restriction endonuclease analysis.Archives of Virology 136, 309–323.

Xia, H., Liu, L., Nordengrahn, A., Kiss, I., Merza, Eriksson, R., Blomberg, J., Belak,S., 2010. A microsphere-based immunoassay for rapid, sensitive detection ofbovine viral diarrhea virus antibodies. Journal of Virological Methods 168,18–21.

Xia, H., Vijayaraghavan, B., Belák, S., Liu, L., 2011. Detection, identification of theatypical bovine pestiviruses in commercial foetal bovine serum batches. PLoSOne 6 (12), e28553 (Epub 2011 December 8).

Further reading

Ståhl, K., Kampa, J., Alenius, S., Persson Wadman, A., Baule, C., Aiumlamai, S., Belák,S., 2007. Natural infection of cattle with an atypical ‘HoBi’-like pestivirus-implications for BVD control and for the safety of biological products. VeterinaryResearch 38, 517–523.