Evaluation of an indirect ELISA for detection of antibodies in bulk milk against bluetongue virus infections in the Netherlands

Veterinary Microbiology 146 (2010) 209–214

Contents lists available at ScienceDirect

Veterinary Microbiology journal homepage: www.elsevier.com/locate/vetmic

Evaluation of an indirect ELISA for detection of antibodies in bulk milk against bluetongue virus infections in the Netherlands M.H. Mars a,*, C. van Maanen a, P. Vellema a, J.A. Kramps b, P.A. van Rijn b a b

Animal Health Service Deventer, The Netherlands Central Veterinary Institute of Wageningen University—(CVI-WUR), Lelystad, The Netherlands

A R T I C L E I N F O

A B S T R A C T

Article history: Received 5 January 2010 Received in revised form 28 April 2010 Accepted 3 May 2010

After the introduction of bluetongue virus serotype 8 (BTV-8) in western Europe in 2006, an indirect ELISA for detection of serogroup-specific antibodies against BTV in serum samples was validated for individual milk samples by the Central Veterinary Institute and the Animal Health Service in the Netherlands (Kramps et al., 2008). In order to develop a cost-effective monitoring tool, we now have evaluated this ELISA also for use in bulk milk. Therefore, bulk milk samples and individual milk samples were collected from 92 herds in the affected southern region in the Netherlands in 2007, before the start of the vaccination campaign. In addition, bulk milk samples collected from 88 herds before the bluetongue introduction in 2006 (‘‘historically negative’’ samples) have been tested. With these results ROC analyses were performed and herd specificity and herd sensitivity of the bulk milk ELISA were estimated. All ‘‘historically negative’’ bulk milk samples were negative in the ELISA, with a mean S/P ratio of 10  0.8%. The herd sensitivity and herd specificity of the ELISA in bulk milk samples depend on the cut-off that is chosen. In order to detect a withinherd-prevalence of 1%, the optimal cut-off S/P ratio 13% was found. A few herds with one or two milk-positive animals would then be missed. The specificity will be 100%. A within-herdprevalence of 10% can be detected with 100% sensitivity at a cut-off S/P ratio of 96%. In conclusion, the indirect ELISA in bulk milk samples is a very specific and sensitive test which can be implemented in monitoring and surveillance systems in unvaccinated populations. ß 2010 Elsevier B.V. All rights reserved.

Keywords: Bluetongue virus ELISA Bulk milk Sensitivity Specificity Validation

1. Introduction Since August 2006, bluetongue virus serotype 8 (BTV-8) has spread widely in North-western Europe. Clinical signs have been reported extensively in sheep and cattle, and on small scale in goats (Dercksen et al., 2007; Van Schaik et al., 2008; Elbers et al., 2008). In the Netherlands, monitoring of this vector-borne disease of ruminants can be performed most efficiently in the cattle population because serum and

* Corresponding author at: Diagnostics, Research and Epidemiology, Arnsbergstraat 7, P.O. Box 9, 7400 AA Deventer, The Netherlands. Tel.: +31 570 660614; fax: +31 570 660176. E-mail address: [email protected] (M.H. Mars). 0378-1135/$ – see front matter ß 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.vetmic.2010.05.009

milk samples are easily available. Previously, we have validated an indirect antibody detecting ELISA for individual milk samples (Kramps et al., 2008). In addition, bulk milk samples from 10 high prevalent herds, tested strongpositive in this ELISA. These preliminary results showed that testing bulk milk samples could be feasible to detect infections with BTV in herds and regions. In the Netherlands, many ruminant diseases are monitored using bulk milk tests. For OIE-listed diseases like bovine leucosis and brucellosis, which are not present in the Netherlands, bulk milk tests are used for surveillance. For endemic infections such as neosporosis, salmonellosis, bovine herpesvirus infections and bovine virus diarrhoea virus infections, bulk milk tests are used for certification programmes and monitoring. Extensive studies have been performed in the

210

M.H. Mars et al. / Veterinary Microbiology 146 (2010) 209–214

Netherlands to validate tests for this purpose. For some diseases, bulk milk tests appeared to be very sensitive and useful (Kramps et al., 1999; Wellenberg et al., 1998; Veling et al., 2001, 2002; Emmerzaal et al., 2002; Mars and van Maanen, 2005; Bartels et al., 2005; Brinkhof et al., 2009). For other purposes such as screening for paratuberculosis and bovine tuberculosis, the use of bulk milk tests might be restricted (Van Weering et al., 2007; Fischer et al., 2005). The feasibility of bulk milk tests depends on the characteristics of the infection as well as on the characteristics of the test. Bulk milk testing is in general a simple and costefficient method for surveillance. For an OIE-listed disease such as bluetongue, monitoring and surveillance tools should also detect low-prevalent herds. On the other hand, due to the serious consequences of antibody detection for such a notifiable disease, specificity should be very high. Therefore we have performed a validation study in order to investigate the possibilities of using the indirect ELISA for BTV antibodies in bulk milk samples for monitoring and surveillance programmes. 2. Materials and methods 2.1. Samples In the framework of a sentinel study performed in the Netherlands in 2007, individual milk samples were collected from 300 dairy herds from all over the country. All herds consisted of more than 50 adult dairy cows. At that time in spring 2007, only the southern part of the Netherlands had been infected, and BTV vaccines were not yet available. From 92 herds in the infected area, one bulk milk sample and individual samples from all dairy cows were collected. The bulk milk samples were collected in May and the individual milk samples were collected in the period May–July 2007. These samples are referred to as panel A. In addition, bulk milk samples were available from 26 dairy herds in which clinical signs of bluetongue were observed, and from 29 herds in which no clinical signs were observed. These bulk milk samples were tested for the presence of BTV antibodies. These samples are referred to as panel B. Historically negative bulk milk samples from the Netherlands, sampled before August 2006 (n = 88) were also tested. These samples are being referred to as panel C. 3. Methods An indirect ELISA based on the recombinant VP7 protein of BTV (ID Screen1 Blue Tongue Indirect Milk, ID.VET, Montpellier, France) was used to test both the individual and the bulk milk samples. Antibodies present in the sample will bind to this VP7 and will be detected by adding anti-ruminant-IgG-conjugate and substrate. The intensity of staining is correlated with the amount of antibodies in the sample. This ELISA was developed for use in individual milk samples. The manufacturer advised a cut-off of S/P ratio 50% so all samples with an optical density of more than 50% of the optical density of the positive control

sample should be interpreted as positive. Previously, we have validated this ELISA for use in individual samples and found it useful (Kramps et al., 2008). The sensitivity and specificity were 99% and 96.5%, respectively as compared to a competitive ELISA in serum. For routine use however, we have adapted the cut-off from S/P ratio 50% to S/P 100% in order to increase the specificity to 99%. With this cut-off, the sensitivity is 98% as compared to the competitive ELISA in serum. For use in bulk milk no cut-off was proposed in the manual. Receiver operating characteristic (ROC) curves (Kraemer, 1992) of the ELISA were constructed by plotting the sensitivity and specificity on the ordinate as a function of different cut-off values. The area under the ROC curve is a quantitative measure of the test’s performance. 4. Results The results of the samples of 92 herds from the southern part of the Netherlands (panel A) are presented in Tables 1a and 1b). The within-herd-prevalence is the percentage of the individual cows that were tested positive in the individual milk samples of each herd. All bulk milk samples from herds with a prevalence of 0% (n = 23) had S/P ratios 13%. All bulk milk samples from herds with more than 10% milk-positive cows (n = 69) had S/P ratios >100%. Based on the results mentioned in Table 1, bulk milk test specificity would be 100% with a cut-off S/P ratio = 13%, however, in that case 8 herds with one or two positive cows would have been falsely classified as antibodynegative. The sensitivity to detect every single infection in a herd would be 87% (confidence interval at 87% = 79–95%). Fig. 1 presents the relation between S/P ratio in bulk milk samples and the percentage positive individual milk Table 1a Results (S/P ratio) of ELISA in bulk milk and individual milk samples in herds (n = 23) in which all individual milk samples were tested negative. Bulk milk S/P%

Number of ELISA positive individual milk samples

Total number of individual milk samples

13 7 6 7 9 7 7 8 7 7 7 6 6 6 6 7 7 7 7 6 6 6 6

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

37 52 89 42 45 34 51 55 60 42 54 51 49 58 42 48 60 58 43 53 38 36 53

M.H. Mars et al. / Veterinary Microbiology 146 (2010) 209–214 Table 1b Results (S/P ratio) of ELISA in bulk milk and individual milk samples in herds in which one or more individual milk samples were tested positive (n = 69). Results of 31 herds with more than 10% positive individual milk samples were summarized in the last part of the table. Bulk milk S/P%

Number of ELISA positive individual milk samples

Total number of individual milk samples

Within-herdprevalence

10 17 23 7 7 7 7 28 9 30 7 51 35 27 26 34 27 40 6 56 80 85 95 28 54 20 62 77 122 59 70 155 51 118 22 123 153 117 107–240a

2 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 3 2 2 2 2 2 3 1 3 2 3 3 2 5 3 3 3 3 5 4 6 5 3–37

208 81 80 70 62 59 56 56 55 52 51 50 97 47 44 39 100 62 53 51 47 47 67 22 62 41 59 56 34 84 48 42 41 39 57 42 59 49 13–64

1% 1% 1% 1% 2% 2% 2% 2% 2% 2% 2% 2% 2% 2% 2% 3% 3% 3% 4% 4% 4% 4% 4% 5% 5% 5% 5% 5% 6% 6% 6% 7% 7% 8% 9% 10% 10% 10% 11–85%

a 11 herds had with a within-herd-prevalence of 11–19%, 9 herds had a within-herd-prevalence of 20–49% and 11 herds had a within-herdprevalence of 50–85%.

211

samples for each herd in panel A. Above a within-herdprevalence of 30%, S/P ratios reached a maximum signal in this ELISA. The optimal cut-off for the ELISA in bulk milk depends on the aim of testing. In this study, several ROC analyses were performed. For example, in order to detect a withinherd-prevalence of at least 1%, the optimal cut-off would be an S/P ratio of 13% (max Youden). With that cut-off the sensitivity to detect a herd with at least 1% milk-positive cows will be 88% whereas the specificity would be 100%. In order to detect all herds with only 1% milk-positive cows by testing one bulk milk sample, the cut-off in bulk milk samples should be set on an S/P ratio of 6% (see Table 1b). With such a low threshold, many not-infected herds also would be tested false-positive (Table 1a). In Fig. 2, three ROC curves are presented: bulk milk ELISA test characteristics with several cut-off values in order to detect a within-herd-prevalence of 2%, 5% and 10% respectively. In Table 2, the results of the ROC analyses are summarized: the minimum within-herd-prevalence which has to be detected, the optimal cut-off, the area under the curve (AUC), the sensitivity in order to detect the specified prevalence and the accompanying specificity. It should be noted that the within-herd-prevalence that needs to be detected, the optimal cut-off and the sensitivity and specificity are related to each other. For example, when only herds with a within-herd-prevalence of at least 5% should be detected, and the optimal cut-off (S/P ratio 56) is used, 95.3% of these herds with 5% prevalence will be detected. However, some of the bulk milk-positive herds will have a within-herd-prevalence lower than 5% but none of the positive herds will be free of BTV antibodies. The bulk milk samples of panel B, originating from 26 dairy herds with clinical signs of bluetongue and from 29 herds with no clinical signs of bluetongue, were tested without confirmation of the actual infection status of these herds. The S/P ratio of the bulk milk ELISA in the first group was always higher than 56%, and most S/P ratios were higher than 100%. The S/P ratio in the bulk milk samples from the latter group was always lower than 25% (Fig. 3). The results of the ELISA in bulk milk samples of panel C, originating from 88 dairy herds that had been sampled before the outbreak of bluetongue, were all negative, with a mean S/P ratio of 10.2  0.8%. Based on the mean S/P percentage +3SD of these historically negative bulk milk samples, the cut-off of the ELISA in bulk milk should be an S/P ratio of at least 12.6%. 5. Discussion

Fig. 1. Relation between S/P ratio in the bulk milk sample of a herd with the within-herd-prevalence (% positive individual milk samples) for each herd.

Validation studies for the use of BTV ELISAs in bulk milk have not been published yet, but promising results were presented at the EPIZONE Bluetongue Symposium in 2007 (Kirkland, 2007) However, Chaignat et al described some experiences with the BTV bulk milk ELISA in which specificity could be optimised (Chaignat et al., 2009). The indirect ELISA that has been validated in our study detects BTV-VP7 specific antibodies in milk samples. However, there was neither a reference test for BTV-ELISA in milk samples nor a proficiency test for BTV antibody

M.H. Mars et al. / Veterinary Microbiology 146 (2010) 209–214

212

Fig. 2. (a) ROC analysis of bulk milk ELISA sensitivity (solid line) and specificity (dotted line) with several cut-off values in order to detect a within-herdprevalence of 2%. (b) ROC analysis of bulk milk ELISA sensitivity (solid line) and specificity (dotted line) with several cut-off values in order to detect a within-herd-prevalence of 5%. (c) ROC analysis of bulk milk ELISA sensitivity (solid line) and specificity (dotted line) with several cut-off values in order to detect a within-herd-prevalence of 10%.

detection in milk samples available. Therefore we previously validated this indirect ELISA in individual milk samples using the ID.VET competition ELISA in serum samples as the reference test (Kramps et al., 2008). The ID.VET competition ELISA in serum showed a good performance in proficiency testing (Workshop on bluetongue diagnostics and epidemiology, Brussels, Belgium, November 28–29, 2006; Report Proficiency Testing 2009, VAR Brussels, Batten et al., 2009). The indirect ELISA in individual milk samples showed good agreement to results of competition ELISA in serum samples (Kramps et al., 2008). Therefore in the present study we validated the use of this indirect ELISA in bulk milk samples, using the ELISA

in individual milk samples as reference test. ELISA results of bulk milk samples have been evaluated in comparison to the ELISA results of individual milk samples of all milking cows in the herds. Consequently, the infection status of non-milking cows and young stock of the herd is not taken into account. This could cause an underestimation of the number of infected animals in the herds and/or the number of infected herds. However, we assume that the lactating cows in most cases will be representative for the herd. This might be inaccurate when, for example, the milking cows were housed and the other animals were on pasture, assuming that risk of exposure to midges might be higher on pasture (Santman-Berends et al., 2010).

Table 2 Relation between detection of within-herd-prevalence, optimal cut-off (max Youden) and test characteristics. Detection of herds with within-herd-prevalence of

Optimal cut-off value for the specified within-herd-prevalence

AUCa

Sensitivity at the specified within-herd-prevalence

Specificity at the specified within-herd-prevalence

1% 2% 3% 5% 10%

S/P S/P S/P S/P S/P

0.959 0.978 0.970 0.984 0.992

88.4% 98.2% 90.5% 95.3% 100%

100% 91.1% 97.4% 93.8% 93.2%

a

Area under the curve.

ratio ratio ratio ratio ratio

13 17 35 56 96

M.H. Mars et al. / Veterinary Microbiology 146 (2010) 209–214

Fig. 3. Bulk milk ELISA S/P ratios of herds with or without bluetongue clinical signs.

In our previous study (Kramps et al., 2008), we have estimated a specificity of 99% for individual milk samples, consequently some of the single reactors in this study might have been false-positive results. In order to obtain more certainty about the infection status of these single reactors, blood samples of these cattle should have been tested with serum ELISA and RT-PCR. This was not performed in this study. Moreover, in some cases the individual cows were sampled several weeks after collection of the bulk milk. Therefore, in these herds antibodies might not yet have been detected in bulk milk samples whereas the bluetongue infections spread some weeks later inducing antibodies in individual cows. Therefore the sensitivity of the ELISA in bulk milk samples could be underestimated in this study. The choice of the cut-off in bulk milk depends on the goals to be achieved, and the consequences of the test specifications going with that cut-off. For example, in order to detect a herd with only 1% milk-positive cows by testing only a bulk milk sample, the cut-off in bulk milk samples should be set on an S/P ratio of 6%. With such a low threshold, many not-infected herds also would be tested positive in bulk milk. Based on the mean S/P percentage +3SD of historically negative bulk milk samples, the cut-off of the ELISA in bulk milk should be set at an S/P ratio of at least 13%. When a within-herd-prevalence of 10% should always be detected, than this ELISA in bulk milk would have a sensitivity of 100% using a cut-off at an S/P ratio of 96% (Table 2). The relatively low specificity at this cut-off (Table 2) may be misleading, since it only implies that herds with ‘‘false-positive’’ bulk milk samples have a within-herd-prevalence of lower than 10%, but higher than 0%. The good relation between within-herd-prevalence and S/P ratio of bulk milk samples (Figs. 1 and 3) indicates that the S/P ratio can be used to indicate the within-herdprevalence. Additional validation would be needed. To obtain a good differentiation in higher within-herdprevalences, titration of the bulk milk samples could be useful. Probably, when bulk milk samples would be tested in several dilutions, a better differentiation can be made in the herds with higher within-herd-prevalences. An indica-

213

tion of the within-herd-prevalence might be informative for farmers in order to decide whether or not to vaccinate their cattle. The use of bulk milk ELISAs is a convenient and cheap method for surveillance, because of the availability of these samples from other monitoring systems. However, we have not studied the responses of this ELISA in bulk milk after BTV vaccination, and most likely this ELISA based on the conservative VP7 protein will not differentiate between infected animals and animals vaccinated with the currently available vaccines. Testing bulk milk samples might be more sensitive than only reporting clinical signs. This can be illustrated by the fact that bulk milk samples from herds without any clinical symptoms sometimes seemed to be infected (S/P ratio >13%). In conclusion, the indirect ELISA in bulk milk samples is very useful for surveillance and monitoring of bluetongue infections in unvaccinated dairy herds. Acknowledgements The authors would like to thank Rianne Buter for technical assistance and the Dutch Ministry of Agriculture, Nature and Food Quality, the Product Board for lifestock, meat and eggs and the Dutch Dairy Board for funding. References Bartels, C.J., van Maanen, C., van der Meulen, A.M., Dijkstra, T., Wouda, W., 2005. Evaluation of three enzyme-linked immunosorbent assays for detecting of antibodies to Neospora caninum in bulk milk. Vet. Parasitol. 131 (3–4), 235–246. Batten, C.A., Sanders, A.J., Bachanek-Bankowska, K., Bin-Tarif, A., Oura, C.A.L., 2009. Bluetongue virus: European Community proficiency test (2007) to evaluate ELISA and RT-PCR detection methods with special reference to pooling of samples. Vet. Microbiol. 135 (3–4), 380–383. Brinkhof, J.M., Houwers, D.J., Moll, L., Dercksen, D., van Maanen, C., 2009. Diagnostic performance of ELISA and PCR in identifying SRLV-infected sheep and goats using serum, plasma and milk samples and in early detection of infection in dairy flocks through bulk milk testing. Vet. Microbiol. (epub). Chaignat, V., Nitzsche, S., Scharrer, S., Feyer, D., Schwermer, H. and Thur, B., 2009. Milk concentration improves Bluetongue antibody detection by use of an indirect ELISA. doi:10.1016 vetmic, 2009.11.036. Dercksen, D., Groot Nibbelink, N., Paauwe, R., Rijn van, P., Vellema, P., 2007. First outbreak of bluetongue in goats in The Netherlands. Tijdschr. Dierg. 132 (10), 786–790. Elbers, A.R., Backx, A., Meroc, E., Gerbier, G., Staubach, C., Hendrickx, G., van der Spek, A., Mintiens, K., 2008. Field observations during the bluetongue serotype 8 epidemic in 2006. I. Detection of first outbreaks and clinical signs in sheep and cattle in Belgium, France and the Netherlands. Prev. Vet. Med. 87 (1–2), 21–30. Fischer, E.A., van Roermund, H.J., Hemerik, L., van Asseldonk, M.A., De Jong, M.C., 2005. Evaluation of surveillance strategies for bovine tuberculosis (Mycobacterium bovis) using an individual based epidemiological model. Prev. Vet. Med. 15 (67 (4)), 283–301. Emmerzaal, A., de Wit, J.J., Dijkstra, T., Bakker, D., van Zijderveld, F.G., 2002. The Dutch Brucella abortus monitoring programme for cattle: the impact of false-positive serological reactions and comparison of serological tests. Vet. Q. 24 (1), 40–46. Kirkland, P.D., June 2007. In: Proceedings EPIZONE Bluetongue Satellite Symposium, Brescia, Italy, p. 26. Kraemer, H.C., 1992. Evaluating Medical Tests, Objective and Quantitative Guidelines. Sage Publications, London, pp. 63–95. Kramps, J.A., van Maanen, C., van de Wetering, G., Stienstra, G., Quak, S., Brinkhof, J., Ronsholt, L., Nylin, B., 1999. A simple, rapid, and reliable enzyme-linked immunosorbent assay for the detection of bovine virus diarrhoea virus (BVDV) specific antibodies in cattle serum, plasma and bulk milk. Vet. Microbiol. 64 (2–3), 135–144.

214

M.H. Mars et al. / Veterinary Microbiology 146 (2010) 209–214

Kramps, J.A., van Maanen, C., Mars, M.H., Popma, J.K., van Rijn, P.A., 2008. Validation of a commercial ELISA for the detection of bluetongue virus (BTV)-specific antibodies in individual milk samples in Dutch dairy cows. Vet. Microbiol. 130 (1–2), 80–87. Mars, M.H., van Maanen, C., 2005. Diagnostic assays applied in BVDV control in the Netherlands. Prev. Vet. Med. 72 (1–2), 43–48. Santman-Berends, I.M., Bartels, C.J., van Schaik, G., Stegeman, J.A., Vellema, P., 2010. The increase in seroprevalence of bluetongue virus (BTV) serotype 8 infections and associated risk factors in Dutch dairy herds, in 2007. Vet. Microbiol. 142 (3–4), 268–275. Van Schaik, G., Berends, I.M., van Langen, H., Elbers, A.R., Vellema, P., 2008. Seroprevalence of bluetongue serotype 8 in cattle in the Netherlands in spring 2007, and its consequences. Vet. Rec. 163 (15), 441–444. Veling, J., van Zijderveld, F.G., van Zijderveld-van Bemmel, A.M., Schukken, Y.H., Barkema, H.W., 2001. Evaluation of two enzyme-linked

immunosorbent assays for detecting Salmonella enterica subsp. Enterica Serovar Dublin antibodies in milk. Clin. Diagn. Lab. Immunol. 8 (6), 1049–1055. Veling, J., Barkema, H.W., van der Schans, J., van Zijderveld, F.G., Verhoeff, J., 2002. Herd-level diagnosis for Salmonella enterica subsp. Enterica Serovar Dublin infections in bovine dairy herds. Prev. Vet. Med. 53 (1–2), 31–42. Van Weering, H., van Schaik, G., van der Meulen, A., Waal, M., Franken, P., van Maanen, C., 2007. Diagnostic performance of the Pourquier ELISA for detection of antibodies against Mycobacterium avium subspecies paratuberculosis in individual milk and bulk milk samples of dairy herds. Vet. Microbiol. 125 (1–2), 49–58. Wellenberg, G.J., Verstraten, E.R., Mars, M.H., van Oirschot, J.T., 1998. ELISA detection of antibodies to glycoprotein E of bovine herpesvirus 1 in bulk milk samples. Vet. Rec. 28 (128 (9)), 219–220.