- Email: [email protected]
Evaluation of three competitive ELISAs and a fluorescence polarisation assay for the diagnosis of bovine brucellosis
The Veterinary Journal 216 (2016) 38–44
Contents lists available at ScienceDirect
The Veterinary Journal j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / t v j l
Evaluation of three competitive ELISAs and a fluorescence polarisation assay for the diagnosis of bovine brucellosis A. Praud a,*, M. Durán-Ferrer b,1, D. Fretin c, M. Jaÿ d, M. O’Connor e, A. Stournara f, M. Tittarelli g, I. Travassos Dias h, B. Garin-Bastuji d,2 a Ecole Nationale Vétérinaire d’Alfort, Epidemiology of Animal Infectious Diseases Unit, French Agency for Food, Occupational Health & Safety (Anses), Université Paris-Est, 94700 Maisons-Alfort, France b Laboratorio Central de Sanidad Animal (LCSA), Brucellosis National Reference Laboratory, 18320 Santa Fe, Granada, Spain c Veterinary and Agrochemical Research Centre (CERVA-CODA), Brucellosis National Reference Laboratory, 1180 Brussels, Belgium d Paris-Est University/Anses, Animal Health Laboratory, EU/OIE/FAO Brucellosis Reference Laboratory, 94706 Maisons-Alfort, France e Agri-Food and Biosciences Institute (AFBI), Veterinary Sciences Division, Immunodiagnostic Branch, Brucellosis National Reference Laboratory, Belfast BT4 3SD, Northern-Ireland, UK f Veterinary Laboratory of Larisa, Brucellosis National Reference Laboratory, 41110 Larisa, Greece g Istituto Zooprofilattico Sperimentale dell’Abruzzo e del Molise, Brucellosis National Reference Laboratory, 64100 Teramo, Italy h Instituto Nacional de Investigação Agrária e Veterinária (INIAV), Animal Production Health Unit, Brucellosis National Reference Laboratory, 1500-311 Lisbon, Portugal
A R T I C L E
I N F O
Article history: Accepted 29 June 2016 Keywords: Bovine brucellosis Competitive ELISA Fluorescence polarisation assay Screening tests
A B S T R A C T
Bovine brucellosis is an infectious disease of worldwide public health and economic importance. The usual tests for the diagnosis of this disease include the Rose-Bengal test (RBT), complement fixation test (CFT), serum agglutination test (SAT) and indirect ELISA. New tests such as competitive ELISAs (CELISA) and fluorescence polarisation assay (FPA) have been developed. However, C-ELISA may correspond to different protocols and a wide variation may exist in their diagnostic performance. The aim of this study was to evaluate three commercially available C-ELISA kits (C-ELISA1–3) and FPA for the diagnosis of bovine brucellosis and compare test performance with RBT, CFT, indirect ELISA and FPA. Sera submitted to EU laboratories in 2011 from 5111 adult cattle were tested. Individual test sensitivities (Se) and specificities (Sp) were estimated. Threshold assessment using the receiver operating characteristic method was also performed. The most sensitive tests were FPA (99.0%; 95% confidence interval [CI], 97.9-100%), C-ELISA1 (98.4%; 95% CI, 97.0-99.8%) and RBT (97.7%; 95% CI, 95.9-99.3%). The most specific tests were CFT (99.98%; 95% CI, 99.93-100%), SAT (99.98%; 95% CI, 99.93-100%) and RBT (99.89%; 95% CI, 99.79-99.99%). Among the new tests, none of the three C-ELISA kits studied could be recommended as a single screening test because of their low specificity, especially when used in a herd. C-ELISA3 could not be recommended as confirmatory test on individual animals to determine whether false positive serological test results had occurred. © 2016 Elsevier Ltd. All rights reserved.
Introduction Bovine brucellosis is a major zoonosis of worldwide public health and economic importance. This disease is frequently due to Brucella abortus and sometimes due to Brucella melitensis, in areas where cattle are kept close to sheep or goats. The detection of Brucella antibodies in serum is the routine method for the presumptive diagnosis of bovine brucellosis in eradication programmes in
* Corresponding author. E-mail address: [email protected] (A. Praud). 1 Present address: Central Veterinary Laboratory-Animal Health, Carretera M-106, km 1,4. 28110 Algete-Madrid, Spain. 2 Present address: Paris-Est University/Anses, European and International Affairs Directorate, 94706 Maisons-Alfort, France. http://dx.doi.org/10.1016/j.tvjl.2016.06.014 1090-0233/© 2016 Elsevier Ltd. All rights reserved.
endemic countries and for monitoring herds in officially brucellosis free (OBF) countries, or for movement or trade. The most popular tests include the Rose-Bengal test (RBT), complement fixation test (CFT) and indirect ELISA (I-ELISA), these tests being well standardised at international level1. While the serum agglutination test (SAT) is still considered a screening test despite its low diagnostic accuracy, in several countries new tests such as competitive ELISA (CELISA) and fluorescence polarisation assay (FPA) have been developed and are claimed to have better specificity (Sp) or sensitivity (Se; Gall and Nielsen, 2004). However, C-ELISA is performed using different
1 See: OIE (World Organisation for Animal Health), 2016. Chapter 2.1.4 Bovine brucellosis. In Manual of Diagnostic Tests and Vaccines for Terrestrial Animals. OIE, Paris, France. http://www.oie.int/fileadmin/Home/eng/Health_standards/tahm/2.01.04 _BRUCELLOSIS.pdf (accessed 23 June 2016).
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
39
protocols and different antigens, competitive antibodies and conjugates, and consequently, there could be wide variations in their diagnostic performance. RBT, I-ELISA and FPA are considered suitable screening tests for the detection of infected herds or to confirm the absence of infection in brucellosis-free herds. However, at an individual level, the status of seropositive animals may be difficult to interpret without a confirmatory and/or complementary strategy including either other serological tests (usually CFT or C-ELISA), brucellin skin test or direct tests such as culture and/or PCR on target organs sampled at the abattoir. According to the recommendations made by the European Food Safety Authority (EFSA), by request from the European Commission for brucellosis diagnostic methods for bovines, sheep and goats2, the C-ELISA ‘should remain in the EU legislation on intra-community trade as a complementary test, pending the conduct of further studies conducted in accordance with the World Organisation for Animal Health (OIE) procedure’. This study aimed to address this recommendation and to evaluate three C-ELISAs and FPA for the diagnosis of bovine brucellosis and compare their performance with RBT, CFT, indirect ELISA and FPA.
ity and affinity to the S-LPS of Brucella. For all tests, the same batch of the antigen or kit was used with all samples tested. Samples were stored frozen. Once thawed, they were stored at 4 °C and tested within 48 h in the EURL by the same two experienced technicians. All tests were performed according to OIE standards for RBT, SAT, CFT and FPA3 and according to the manufacturers’ instructions for the ELISAs. For CFT, cut-offs were those recommended by Annex C of the 64/432 EC directive for CFT. For SAT, cut-offs were those recommended by the test manufacturers (Table 1).
Materials and methods
Epidemiological indicators and statistical methods
Sera
The specificities of C-ELISAs and FPA were estimated in cattle from OBF herds. The sensitivities of these tests were estimated in animals that were positive for Brucella spp. by culture and had also obtained at least two positive results among the usual tests (RBT, CFT, SAT and I-ELISA). The sensitivities of the usual tests were assessed in animals that were positive for Brucella spp. by culture and had also obtained at least two positive results among three usual tests. The defined significance level of statistical tests (McNemar’s test for paired data) was 5%. The second aim of this study was to evaluate the C-ELISA and FPA thresholds to optimise their accuracy. Receiver operating curves were constructed using R (The R foundation) equipped with DiagnosisMed and ROCR packages. The area under curve (AUC) and the threshold maximising the Youden indices (J = Se + Sp − 1) were calculated.
The study included sera from 5111 adult cattle, submitted to laboratories in 2011. Three French regional veterinary diagnostic laboratories were asked to provide the EU Brucellosis Reference Laboratory (EURL, Anses, Maisons-Alfort, France) with bovine sera from French OBF herds. Samples (n = 4330) were received in sufficient volume for the study (2–5 mL) from Saône-et-Loire (n = 1409), Normandy (n = 1483) and Brittany (n = 1538). French sera were a convenience sample of specimens submitted for routine serological diagnosis of bovine diseases in adult beef cattle (>24 months); there were no duplicates. Selection was independent of brucellosis test results. National Reference Laboratories from EU member states where bovine brucellosis had yet not been eradicated were requested to provide the EURL with sera from unvaccinated adult cattle confirmed to be infected with brucellosis (isolation of Brucella spp. by culture). Sera (n = 681) were received in sufficient volume from the following countries: Belgium (n = 83), Greece (n = 264), Italy (n = 37), Northern Ireland (n = 138), Portugal (n = 124) and Spain (n = 35). Animals from each country were from several herds. Screening tests Eight serological tests were performed. Five of them were approved by the EU (EU, 2008): RBT, CFT, SAT, I-ELISA and FPA. Antigens and kits (IDEXX Brucellosis Serum Ab Test) were supplied by IDEXX and all met the standardisation criteria in the annex C of the 64/432 EC directive (EU, 2008). SAT was performed in tubes. FPA antigen test kits (B. abortus antibody test kit) conformed to the standardisation criteria defined in the annex C of the 64/432 EC directive and were supplied by Diachemix. For the batch used in this study, results with the OIE ELISA standard serum (OIEELISASPSS) were as follows: positive up to 1/4, suspect/inconclusive at 1/8 and negative at 1/16 dilutions (made in a negative pool of bovine sera). Results with the French National secondary standard for ELISA were as follows: positive up to 1/8, suspect/ inconclusive at 1/16 and negative at 1/32 dilutions (made in a negative pool of bovine sera). The three C-ELISA kits available on the EU market (C-ELISA1, Ingezim Brucella Compac, Ingenasa; C-ELISA2, Compelisa, AHVLA Scientific; C-ELISA3, SVANOVIR Brucella-Ab c-ELISA, Svanova) were assessed. These three C-ELISAs were based on different biological principles. For C-ELISA1, an anti-Brucella S-LPS monoclonal antibody conjugated to the peroxidase was added after test sera were incubated for 1 h with the S-LPS coated plates (two-step competition). The reaction was then revealed with the addition of the chromogen substrate. For C-ELISA2, test sera were incubated for 30 min with S-LPS coated plates, with an anti-Brucella S-LPS monoclonal antibody conjugated to the peroxidase (one-step competition). The reaction was then revealed with the addition of the chromogen substrate. For C-ELISA3, test sera were incubated for 30 min with S-LPS coated plates with an anti-Brucella S-LPS monoclonal antibody (one-step competition). The reaction was revealed with the addition first of an anti-mouse Ig conjugated to the peroxidase, then with the addition of the chromogen substrate. The three monoclonal antibodies used in the three respective tests were from different sources; therefore they differed in their activ-
2 See: Opinion of the Scientific Panel on Animal Health and Welfare (AHAW) on a request from the Commission concerning Brucellosis Diagnostic Methods for Bovines, Sheep, and Goats http://www.efsa.europa.eu/en/efsajournal/pub/432 (accessed 23 June 2016).
Table 1 Cut-offs recommended by the manufacturers of the indirect ELISA (I-ELISA), fluorescence polarisation assay (FPA) and three competitive ELISAs (C-ELISAs1–3). Test
Cut-offs and interpretation of the result
I-ELISA
Negative, <110% Inconclusive, 110–120% Positive, >120% Negative, <10 Inconclusive, 10–20 Positive, >20 Positive, >40% Positive, ≤60% Positive, ≥30%
FPA
C-ELISA1 C-ELISA2 C-ELISA3
Results The results of the eight tests in the OBF population and in infected herds are displayed in Tables 2–4. Comparison with results from the OBF regions showed false positive serological reactions (FPSR) in 137 animals. Most FPSR occurred in Saône-et-Loire (85/ 1469; 5.8%) and Brittany (47/1538; 3.1%) and a few occurred in Normandy. Most FPSR occurred in only one test (124/137; 90.5%); 9.5% of samples obtained FPSR to at least two tests (13/137; 9.5%). These FPSR mainly occurred with the C-ELISAs, especially C-ELISA3. Only a few animals had inconclusive results with I-ELISA (2/4430 in OBF herds; 4/681 in infected herds), but inconclusive results were more frequent with FPA, especially in the infected population (29/ 4430 animals from OBF herds; 29/681 animals in infected herds). The estimated sensitivities and specificities are presented in Tables 5 and 6. The most specific tests were CFT (99.98%; 95% confidence interval, CI, 99.93–100%), SAT (99.98%; 95% CI, 99.93– 100%) and RBT (99.89%; 95% CI, 99.79–99.99%). C-ELISA1 (99.37%; 95% CI, 99.13–99.61%) and C-ELISA2 (99.35%; 95% CI, 99.11–99.59%) were less specific than RBT (P = 6 × 10−5 and 8 × 10−6, respectively), CFT (P = 1 × 10−5 and 8 × 10−7, respectively) and SAT (P = 2 × 10−6 and 3 × 10−7, respectively). C-ELISA3 (Sp, 98.28%; 95% CI, 97.91–98.67%) was the least specific test. C-ELISA1 (Se, 98.4%; 95% CI, 97.0– 99.8%) was significantly more sensitive than SAT (Se, 81.3%; 95% CI, 77.0–85.7%; P = 4 × 10−12), C-ELISA2 (Se, 90.0%; 95% CI, 86.6–93.3%; P = 1 × 10−5) and C-ELISA3 (Se, 94.2%; 95% CI, 91.6–96.8%; P = 2 × 10−4). When inconclusive results were interpreted as positive results, FPA (Se, 99.0%; 95% CI, 97.9–100%) was more sensitive than CFT (Se, 95.0%; 95% CI, 92.6–97.5%; P = 3 × 10−3), C-ELISA2 (P = 5 × 10−7) and C-ELISA3 (P = 3 × 10−4). Most likely due to the small sample size from
3
See footnote 1.
40
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
Table 2 Results of cattle from officially brucellosis free farms according to the geographic region. Brittany (n = 1538)
Test
RBT CFT SAT I-ELISA C-ELISA1 C-ELISA2 C-ELISA3 FPA
Normandy (n = 1483)
Saône et Loire (n = 1409)
Total (n = 4430)
P
I
N
P
I
N
P
I
N
P
I
N
0 0 1 (0.07) 3 (0.20) 14 (0.91) 25 (1.63) 49 (3.19) 6 (0.39)
– – – 1 (0.07) – – – 12 (0.78)
1538 1538 1537 1534 1524 1513 1489 1520
0 1 (0.07) 0 0 0 3 (0.20) 17 (1.15) 1 (0.07)
– – – 0 – – – 7 (0.47)
1483 1482 1483 1483 1483 1480 1466 1475
5 (0.35) 0 0 5 (0.35) 14 (0.99) 1 (0.07) 10 (0.71) 2 (0.14)
– – – 1 (0.07) – – – 10 (0.71)
1404 1409 1409 1403 1395 1408 1399 1397
5 (0.11) 1 (0.02) 1 (0.02) 8 (0.18) 28 (0.63) 29 (0.65) 76 (1.72) 9 (0.20)
– – – 2 (0.05) – – – 29 (0.65)
4425 4429 4429 4420 4402 4401 4354 4392
P, positive (% tested); I, inconclusive (% tested); N, negative; RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay.
Table 3 Test results in cattle from officially brucellosis free farms, according to the geographic region. RBT
CFT
SAT
I-ELISA
C-ELISA 1
2
N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N P N N N N N P N N N N P N N N N N P N N N N N P P N N N I N N N N N P N N N N N P P N N N N P P N N N N P P P N N P N P P N P N N N N P N N N N N P N N N P N P N N P P N Total Number of animals with false positive results to either test (% tested animals)
FPA
Brittany
Normandy
Saône and Loire
Total
N I P N I N N N N N N N N I P N N N N P
1436 11 6 0 1 21 2 10 2 1 1 3 0 0 0 1 0 0 0 0 1538 47 (3.1)
1454 7 1 0 0 3 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1483 5 (0.3)
1375 8 0 69 0 0 0 7 0 0 1 0 1 2 1 0 0 2 2 1 1409 85 (5.8)
4265 26 7 69 1 24 2 17 2 1 2 3 1 2 1 1 1 2 2 1 4430 137 (3.1)
3 N N N P P N P N P N N N N N P P N N N N
RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay; N, negative; I, inconclusive; P, positive.
most countries, no significant differences could be demonstrated at country level except for Greece, where the same overall trends were observed. These results suggested that the cut-offs for C-ELISAs and FPA could be changed to improve their sensitivities and/or specificities. ROC curves are displayed in Figs. 1A-D. The optimised cutoffs for C-ELISA1 (40%) and C-ELISA3 (30%) agreed with the manufacturers’ recommendations (Figs. 1A, C). The cut-off optimising both Se and Sp for C-ELISA2 was 70%, whereas the manufacturer recommended 60% (Fig. 1B). The optimised cut-off for FPA was 10, with no inconclusive range (Fig. 1D). Discussion The aims of this study were to estimate the Se and the Sp of C-ELISAs and FPA and to calculate the thresholds of C-ELISAs and FPA to optimise their accuracy. In our study, none of the usual tests (SAT, CFT, RBT and I-ELISA) recommended by the European legislation were able to confirm all cases. The most sensitive tests were C-ELISA1, FPA and RBT. The least sensitive test was SAT. The most specific tests were CFT, SAT and RBT, while the least specific test was C-ELISA3. Increasing the cut-off for C-ELISA2 from 60% to 70% could improve its accuracy when a better compromise between Se and Sp is needed, while in OBF regions the usual cut-off should be
used in order to optimise the Sp. A single cut-off could be chosen for FPA in order to eliminate the inconclusive range and make the interpretation of results easier. The opinion of the EFSA on a request from the European Commission concerning brucellosis diagnostic methods for bovines4 considered RBT, CFT and I-ELISA to be standard tests. It also suggested the inclusion of FPA in the EU legislation, since its Se and Sp were comparable to standard tests. It was therefore logical to recommend that SAT should not remain in the EU legislation on trade because of its low Se, and that C-ELISA should be used as a complementary test, until further investigations were performed. The results of a meta-analysis presented in this opinion are summarised in Table 7. The sensitivities estimated in our study were not significantly different from these results, whereas the specificities estimated in our study were significantly higher. Abernethy et al. (2012) estimated individual sensitivities of six tests for bovine brucellosis diagnosis and reported that RBT (78.1%) and I-ELISA (67.9%) had the highest sensitivities relative to culture. I-ELISA, RBT and CFT (100%) had the highest specificities. C-ELISA3 was also evaluated in that study and its Se and Sp were the lowest of the tests investigated, which agrees with our results. Other studies
4
See footnote 2.
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
41
Table 4 Test results for 681 sera from infected herds. RBT
N N N N N N N N N N N N N N N N N N N N N N N N N P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P Total
CFT
N N N N N N N N N N N N N N N N N N N N P P P P P N N N N N N N N N N N N N N N N P P P P P P P P P P P P P P P P
SAT
N N N N N N N N N N N N N N N N N N N P N N N N N N N N N N N N N N N P P P P P P N N N N N N N N P P P P P P P P
I-ELISA
N N N N N N N I I I P P P P P P P P P P N P P P P N N N N N I P P P P N N N N N P P P P P P P P P N N N N P P P P
C-ELISA 1
2
3
N N N N P P P N P P N N N P P P P P P N N N P P P N P P P P P N P P P P P P P P P N P P P P P P P P P P P N P P P
N N N N N N N N N N N N N N N N N P P N N P N N P N N N N P N N N N P N N N P P P P N N N P P P P N N P P P N P P
N N P P N N P N N N N N N N N N P N P N N P N P N N N N P N N N P P P N N P P P P P N P P N N P P N P P P P P N P
FPA
B
Sp
Gr
N-I
It
Pt
Total
N I N I N I N N N I N I P N I P I P P N N P P P P N N I I I I P I P P N I I I P P P P I P N P D P P P D P P P P P
56 – 3 1 2 – 1 – – – – – – – – 1 – – 1 – – 1 – – – 2 1 – – – – – – – – – 1 – – 1 – 1 1 – – – – – 1 – – 1 – 1 – – 7 83
12 1 – – – – – – – – 1 – – – 2 – – – – – – – 1 – – 1 – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – 17 35
17 – – – – – – – – – 1 2 – – 1 3 1 – – 1 – – – 2 1 – 2 – 1 – 1 1 2 1 1 – – – – – 1 – 2 1 7 1 3 1 23 1 – – 1 – 5 2 178 264
118 – 2 – 1 1 – – – 1 – – – – – – – – – – – – 1 – – – 2 1 – 1 – – – – – 1 – 1 1 3 – – – – – – – – – – 1 – 1 – – – 2 138
11 – – – – – – – – – – – – – 1 – 1 – – – 1 – – – – 2 – – – – – – – – – – – – – – – – – – – – – – 2 – – – – – – – 19 37
102 2 1 – 1 – – 1 1 – 2 – 1 1 1 – – 1 – – – – – – – 1 – – – – – – – – – – – – – – – – 1 – – – – – 1 – – – – – – – 7 124
316 3 6 1 4 1 1 1 1 1 4 2 1 1 5 4 2 1 1 1 1 1 2 2 1 6 5 1 1 1 1 1 2 1 1 1 1 1 1 4 1 1 4 1 7 1 3 1 27 1 1 1 2 1 5 2 230 681
RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay ; B, Belgium; Sp, Spain; Gr, Greece ; N-I, Northern Ireland; It, Italy; Pt, Portugal; N, negative; I, inconclusive; P, positive.
published in the last decade have aimed to assess the sensitivities and specificities of tests for the diagnosis of bovine brucellosis (Fosgate et al., 2006; Muma et al., 2007; Matope et al., 2011; Rahman et al., 2013; Sanogo et al., 2013; Gorsich et al., 2015), but they were performed in different epidemiological contexts in different countries (Bangladesh, Ivory Coast, South Africa, Trinidad, Zambia, Zimbabwe), and in different species (cattle, African buffalos, water buffaloes, sheep, goats). Further, they were performed on smaller sample sets than our study and specificities were often estimated
in brucellosis free herds from infected regions. Test methods and thresholds varied greatly from one study to another. According to Muma et al. (2007), the Se of C-ELISA3 was 97% (95% credibility interval, CrI, 93–100%) and its Sp was 54% (95% CrI, 34–75%); FPA was less sensitive (Se, 72%; 95% CrI, 63–80%) than C-ELISA3 but more specific (FPA: Sp, 93%; 95% CrI, 85–99%). Matope et al. (2011) also reported that C-ELISA3 was highly sensitive (Se, 99%; 95% CrI, 94.8–100%) and quite specific (Sp, 95.4%; 95% CrI, 93.7–96.7%). The FPA performed on serum was more specific (Sp, 99%; 95% CrI,
42
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
Table 5 Estimation of the specificities of the eight tests (n = 4430, officially brucellosis free animals). Test
RBT CFT SAT I-ELISA C-ELISA1 C-ELISA2 C-ELISA3 FPA
Brucellosis-free population P
I
N
5 1 1 8
– – – 2
4425 4429 4429 4420
28 29 76 9
– – – 29
4402 4401 4354 4392
Estimated specificity mean (95% CI) 99.89% (99.79–99.99%) 99.98% (99.93–100%) 99.98% (99.93–100%) I = N: 99.82% (99.70–99.95%) I = P: 99.77% (99.64–99.92%) 99.37% (99.13–99.61%) 99.35% (99.11–99.59%) 98.28% (97.91–98.67%) I = N: 99.80% (99.67–99.93%) I = P: 99.14% (98.88–99.42%)
CI, confidence interval; RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay ; P, positive; I, inconclusive; N, negative; I = N, inconclusive results interpreted as negative results; I = P, inconclusive results interpreted as positive results.
98.0–99.6%) but less sensitive (FPA: Se, 87%; 95% CrI, 78.4–93.4%) than C-ELISA3. However, these results are unlikely to be generalisable to European cattle or to our results, since the contexts, sampling schemes and statistical methods were different. The estimation of the diagnostic Se and Sp of a test should ideally be derived from testing an adequate sampling of animals of known history and infection status. The selection of an appropriate reference standard infected population (culture-positive animals) is complicated and expensive. Furthermore, Brucella-culture should be performed in conditions that optimise Se, which could lead to an overestimation of the diagnostic Se of the tests under study. In our study, the infection status of herds was confirmed by culture, but the individual infection status of animals was unknown. In the absence of single reference standard, we used a composite reference standard (Rutjes et al., 2007), based on the infection status of the herd and the individual results obtained with the other tests, to perform a direct assessment. Our results were then conditional on a positive reference and should be interpreted as relative Se and Sp. This method is quite easy to implement and is widely used for many diagnostic tests (Alonzo and Pepe, 1998; Rutjes et al., 2007). This kind of reference standard is more discriminant than each of its components in isolation (Alonzo and Pepe, 1999; Pepe, 2003; Rutjes et al., 2007) and even if the reference is imperfect, the infection status of such animals is almost certainly correct. Moreover,
Table 6 Sensitivity estimations for the eight tests (n = 309, infected animalsa). Test
RBT CFT SAT I-ELISA C-ELISA1 C-ELISA2 C-ELISA3 FPA
Infected animals P
I
N
289 288 244 289 304 278 291 298
– – –
7 15 56 13 5 31 18 3
– – – 8
Estimated sensitivity mean (95% CI) 97.7% (95.9–99.3%) 95.0% (92.6–97.5%) 81.3% (77.0–85.7%) 95.7% (93.4–98.0%) 98.4% (97.0–99.8%) 90.0% (86.6–93.3%) 94.2% (91.6–96.8%) I = N: 96.4% (94.4–98.5%) I = P: 99.0% (97.9–100%)
CI, confidence interval; RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay; P, positive; I, inconclusive; N, negative; I = N, inconclusive results interpreted as negative; I = P, inconclusive results interpreted as positive. a Animals from an infected herd and with positive results to at least two of any of the four usual tests (RBT, CFT, SAT and I- ELISA); sensitivities of the usual tests were performed on animals that had positive results to at least two of the three other usual tests and were from an infected herd.
in diagnostic test accuracy studies, the animals sampled should also be representative of the region where the test is to be used5. Our sample set included bovine sera from several European countries. Some of them were collected in France, which has been an OBF country since 2005. In France, screening tests are performed annually in farms. When a brucellosis outbreak occurs, all animals of the infected herd are slaughtered and an epidemiological survey is launched to find the sources of contamination and determine the potential consequences of the outbreak in epidemiologically related herds. The epidemiological situation is thus strictly controlled. For this reason, the brucellosis free status of cattle from OBF herds is reliable and OBF samples can be considered representative of OBF European regions. Nevertheless, in one of the OBF regions (Saone-et-Loire), FPSR due to cross-reactions with environmental contaminants occur quite frequently (Godfroid et al., 2002). This suggests that highly specific tests should be chosen in such regions (Pouillot et al., 1998). The sera from infected herds were collected in several countries where bovine brucellosis outbreaks occurred. Infection was confirmed by the isolation of Brucella abortus or Brucella melitensis. Infected samples could thus be considered representative of animals from infected herds. Since the infected sera originated from several herds, a cluster effect cannot be excluded, but given the large sample size and the number of herds selected, this effect should be limited. Furthermore, little information was available about the animals (e.g. age, breed, gestational status, herd production type, or species of Brucella isolated in infected herds). It would have been interesting to take these parameters into account to study the accuracy of tests in various situations and disease conditions, e.g. according to Abernethy et al. (2012), the tests are likely to be more sensitive in cows in late gestation than those in early gestation. Conclusions This study was based on a large sample of cattle, which were representative of European cattle. CFT and RBT were the most accurate screening tests for surveillance in OBF regions. Among the new tests, FPA and C-ELISA1 were the most sensitive. C-ELISAs did not have higher specificities than the usual tests, and their sensitivities varied according to the kit used. C-ELISA2 was less sensitive than the usual tests, but was quite specific. Therefore, C-ELISA2 should be used as a screening test only in OBF regions; it should not be used as a confirmatory test. C-ELISA3 was the least specific test investigated and it was not very sensitive. It should not be used as a screening test in OBF regions, or as confirmatory test on single animals in a FPSR context. FPA could be used as a screening test, with inconclusive results interpreted as negative in OBF regions and as positive in infected regions. Given the imperfect sensitivities and specificities of tests studied, a combination of several tests is required to improve the accuracy of diagnosis: (1) in series in a low epidemiological risk context and (2) in parallel in outbreaks to hasten the eradication of the disease. These recommendations should nevertheless be confirmed by other studies because of the limitations of our study protocol. A larger sample set should be used and the concordances between the tests should be studied to assess the characteristics of screening schemes that use these tests in series or in parallel. Conflict of interest statement All authors certify that they have no affiliations with or involvement in any organisation or entity with any financial interest or
5
See footnote 1.
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
43
0.6 0.4 0.0
0.2
True positive rate
0.8
1.0
(A)
0.0
0.2
0.4
0.6
0.8
1.0
False positive rate
0.6 0.4 0.0
0.2
True positive rate
0.8
1.0
(B)
0.0
0.2
0.4
0.6
0.8
1.0
0.8
1.0
0.8
1.0
False positive rate
0.6 0.4 0.0
0.2
True positive rate
0.8
1.0
(C)
0.0
0.2
0.4
0.6 False positive rate
0.6 0.4 0.0
0.2
True positive rate
0.8
1.0
(D)
0.0
0.2
0.4
0.6 False positive rate
Fig. 1. Receiver operating curve (ROC) for competitive ELISAs (C-ELISAs) and fluorescence polarisation assay (FPA). ROC for C-ELISA1, C-ELISA2, C-ELISA3, and FPA are presented in (A), (B), (C), and (D), respectively. Cut-off optimising both the sensitivity (Se) and the specificity (Sp) of C-ELISA1, C-ELISA2, C-ELISA3, and FPA was 40% (Se, 98.7%; Sp, 99.3%; area under curve, AUC, 99.1%), 70% (Se, 92.2%; Sp, 97.0%; AUC, 97.6%), 30% (Se, 94.5%; Sp, 97.6%; AUC, 98.5%), and 10% (Se, 99.4%; Sp, 99.1%; AUC, 99.8%), respectively. For C-ELISA2, the result index decreases with the antibody titre; the curve was adapted to make reading easier.
44
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
Table 7 Predictions of the sensitivities and specificities of tests evaluated for cattle, based on logistic meta-regression (European Food Safety Authority, 2006). Test
Sensitivity (95% credibility interval)
Specificity (95% credibility interval)
C-ELISA (various) C-ELISA3 CFT FPA I-ELISA RBT SAT
0.983 (0.977–0.988) 0.954 (0.917–0.981) 0.960 (0.949–9.970) 0.987 (0.981–0.992) 0.976 (0.941–0.992) 0.981 (0.968–0.991) 0.817 (0.749–0.872)
0.998 (0.997–0.998) 0.992 (0.987–0.995) 0.998 (0.997–0.998) 0.998 (0.998–0.999) 0.975 (0.972–0.979) 0.998 (0.997–0.998) 0.987 (0.986–0.987)
C-ELISA, competitive ELISA; CFT, complement fixation test; FPA, fluorescence polarisation assay; I-ELISA, indirect ELISA; RBT, Rose-Bengal test; SAT, serum agglutination test.
non-financial interest in the subject matter or materials discussed in the manuscript. Acknowledgements The authors thank the French Departmental Veterinary Laboratories of Ille-et-Vilaine, Manche and Saône-et-Loire, Drs Eric LE DREAN, Michaël TREILLES and Pascal VERY, particularly for the collection of the sera from French bovine brucellosis officially free herds. Antoine DRAPEAU and Yannick CORDE are also duly acknowledged for their technical assistance. All serological analyses could be performed thanks to the European Commission’s support to the EU reference laboratory for Brucellosis. References Abernethy, D.A., Menzies, F.D., McCullough, S.J., McDowell, S.W.J., Burns, K.E., Watt, R., Gordon, A.W., Greiner, M., Pfeiffer, D.U., 2012. Field trial of six serological tests for bovine brucellosis. The Veterinary Journal 191, 364–370. Alonzo, T.A., Pepe, M.S., 1998. Assessing the accuracy of a new diagnostic test when a gold standard does not exist. UW Biostatistics Working Paper Series 156.
Alonzo, T.A., Pepe, M.S., 1999. Using a combination of reference tests to assess the accuracy of a new diagnostic test. Statistics in Medicine 18, 2987–3003. European Food Safety Authority, 2006. Scientific opinion on “Performance of Brucellosis Diagnostic Methods for Bovines, Sheep, and Goats”. EFSA Journal 432, 1–44. EU, 2008. Commission Decision of 10 December 2008 amending Annex C to Council Directive 64/432/EEC and Decision 2004/226/EC as regards diagnostic tests for bovine brucellosis (2008/984/EC). Official Journal of the European Union L. 352/38–L. 352/45. Fosgate, G.T., Adesiyun, A.A., Hird, D.W., Hietala, S.K., 2006. Likelihood ratio estimation without a gold standard: A case study evaluating a brucellosis C-ELISA in cattle and water buffalo of Trinidad. Preventive Veterinary Medicine 75, 189–205. Gall, D., Nielsen, K., 2004. Serological diagnosis of bovine brucellosis: A review of test performance and cost comparison. Revue Scientifique et Technique de l’OIE 23, 989–1002. Godfroid, J., Saergerman, C., Walravens, K., Letesson, J.J., Tibor, A., Macmillan, A., Spencer, S., Sanaa, M., Bakker, D., Pouillot, R., et al., 2002. How to substantiate eradication of bovine brucellosis when aspecific serological reactions occur in the course of brucellosis testing. Veterinary Microbiology 90, 461–677. Gorsich, E.E., Bengis, R.G., Ezenwa, V.O., Jolles, A.E., 2015. Evaluation of the sensitivity and specificity of an enzyme-linked immunosorbent assay for diagnosing brucellosis in African Buffalo (Syncerus Caffer). Journal of Wildlife Diseases 51, 9–18. Matope, G., Muma, J.B., Toft, N., Gori, E., Lund, A., Nielsen, K., Skjerve, E., 2011. Evaluation of sensitivity and specificity of RBT, C-ELISA and fluorescence polarization assay for diagnosis of brucellosis in cattle using latent class analysis. Veterinary Immunology and Immunopathology 141, 58–63. Muma, J.B., Toft, N., Oloya, J., Lund, A., Nielsen, K., Samui, K., Skjerve, E., 2007. Evaluation of three serological tests for brucellosis in naturally infected cattle using a latent class analysis. Veterinary Microbiology 125, 187–192. Pepe, M.S., 2003. The Statistical Evaluation of Medical Tests for Classification and Prediction. Oxford University Press, Oxford, p. 320. Pouillot, R., Lescoat, P., Garin-Bastuji, B., Repiquet, D., Terrier, P., Gerbier, G., Bénet, J.J., Sanaa, M., 1998. Risk factors for false-positive serological reactions for bovine brucellosis in Saône-et-Loire (France). Preventive Veterinary Medicine 35, 165–179. Rahman, A.K.M.A., Saegerman, C., Berkvens, D., Fretin, D., Gani, M.O., Ershaduzzaman, M., Ahmed, M.U., Abatih, E., 2013. Bayesian estimation of the true prevalence, sensitivity and specificity of indirect ELISA, Rose Bengal Test and Slow Agglutination Test for the diagnosis of brucellosis in sheep and goats in Bangladesh. Preventive Veterinary Medicine 110, 242–252. Rutjes, A.W.S., Reitsma, J.B., Coomarasamy, A., Khan, K.S., Bossuyt, P.M.M., 2007. Evaluation of diagnostic tests when there is no gold standard. A review of methods. Health Technology Assessment 11, 72. Sanogo, M., Thys, E., Achi, Y.L., Fretin, D., Michel, P., Abatih, E., Berkvens, D., Saegerman, C., 2013. Bayesian estimation of the true prevalence, sensitivity and specificity of the Rose Bengal and indirect ELISA tests in the diagnosis of bovine brucellosis. The Veterinary Journal 195, 114–120.
Contents lists available at ScienceDirect
The Veterinary Journal j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / t v j l
Evaluation of three competitive ELISAs and a fluorescence polarisation assay for the diagnosis of bovine brucellosis A. Praud a,*, M. Durán-Ferrer b,1, D. Fretin c, M. Jaÿ d, M. O’Connor e, A. Stournara f, M. Tittarelli g, I. Travassos Dias h, B. Garin-Bastuji d,2 a Ecole Nationale Vétérinaire d’Alfort, Epidemiology of Animal Infectious Diseases Unit, French Agency for Food, Occupational Health & Safety (Anses), Université Paris-Est, 94700 Maisons-Alfort, France b Laboratorio Central de Sanidad Animal (LCSA), Brucellosis National Reference Laboratory, 18320 Santa Fe, Granada, Spain c Veterinary and Agrochemical Research Centre (CERVA-CODA), Brucellosis National Reference Laboratory, 1180 Brussels, Belgium d Paris-Est University/Anses, Animal Health Laboratory, EU/OIE/FAO Brucellosis Reference Laboratory, 94706 Maisons-Alfort, France e Agri-Food and Biosciences Institute (AFBI), Veterinary Sciences Division, Immunodiagnostic Branch, Brucellosis National Reference Laboratory, Belfast BT4 3SD, Northern-Ireland, UK f Veterinary Laboratory of Larisa, Brucellosis National Reference Laboratory, 41110 Larisa, Greece g Istituto Zooprofilattico Sperimentale dell’Abruzzo e del Molise, Brucellosis National Reference Laboratory, 64100 Teramo, Italy h Instituto Nacional de Investigação Agrária e Veterinária (INIAV), Animal Production Health Unit, Brucellosis National Reference Laboratory, 1500-311 Lisbon, Portugal
A R T I C L E
I N F O
Article history: Accepted 29 June 2016 Keywords: Bovine brucellosis Competitive ELISA Fluorescence polarisation assay Screening tests
A B S T R A C T
Bovine brucellosis is an infectious disease of worldwide public health and economic importance. The usual tests for the diagnosis of this disease include the Rose-Bengal test (RBT), complement fixation test (CFT), serum agglutination test (SAT) and indirect ELISA. New tests such as competitive ELISAs (CELISA) and fluorescence polarisation assay (FPA) have been developed. However, C-ELISA may correspond to different protocols and a wide variation may exist in their diagnostic performance. The aim of this study was to evaluate three commercially available C-ELISA kits (C-ELISA1–3) and FPA for the diagnosis of bovine brucellosis and compare test performance with RBT, CFT, indirect ELISA and FPA. Sera submitted to EU laboratories in 2011 from 5111 adult cattle were tested. Individual test sensitivities (Se) and specificities (Sp) were estimated. Threshold assessment using the receiver operating characteristic method was also performed. The most sensitive tests were FPA (99.0%; 95% confidence interval [CI], 97.9-100%), C-ELISA1 (98.4%; 95% CI, 97.0-99.8%) and RBT (97.7%; 95% CI, 95.9-99.3%). The most specific tests were CFT (99.98%; 95% CI, 99.93-100%), SAT (99.98%; 95% CI, 99.93-100%) and RBT (99.89%; 95% CI, 99.79-99.99%). Among the new tests, none of the three C-ELISA kits studied could be recommended as a single screening test because of their low specificity, especially when used in a herd. C-ELISA3 could not be recommended as confirmatory test on individual animals to determine whether false positive serological test results had occurred. © 2016 Elsevier Ltd. All rights reserved.
Introduction Bovine brucellosis is a major zoonosis of worldwide public health and economic importance. This disease is frequently due to Brucella abortus and sometimes due to Brucella melitensis, in areas where cattle are kept close to sheep or goats. The detection of Brucella antibodies in serum is the routine method for the presumptive diagnosis of bovine brucellosis in eradication programmes in
* Corresponding author. E-mail address: [email protected] (A. Praud). 1 Present address: Central Veterinary Laboratory-Animal Health, Carretera M-106, km 1,4. 28110 Algete-Madrid, Spain. 2 Present address: Paris-Est University/Anses, European and International Affairs Directorate, 94706 Maisons-Alfort, France. http://dx.doi.org/10.1016/j.tvjl.2016.06.014 1090-0233/© 2016 Elsevier Ltd. All rights reserved.
endemic countries and for monitoring herds in officially brucellosis free (OBF) countries, or for movement or trade. The most popular tests include the Rose-Bengal test (RBT), complement fixation test (CFT) and indirect ELISA (I-ELISA), these tests being well standardised at international level1. While the serum agglutination test (SAT) is still considered a screening test despite its low diagnostic accuracy, in several countries new tests such as competitive ELISA (CELISA) and fluorescence polarisation assay (FPA) have been developed and are claimed to have better specificity (Sp) or sensitivity (Se; Gall and Nielsen, 2004). However, C-ELISA is performed using different
1 See: OIE (World Organisation for Animal Health), 2016. Chapter 2.1.4 Bovine brucellosis. In Manual of Diagnostic Tests and Vaccines for Terrestrial Animals. OIE, Paris, France. http://www.oie.int/fileadmin/Home/eng/Health_standards/tahm/2.01.04 _BRUCELLOSIS.pdf (accessed 23 June 2016).
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
39
protocols and different antigens, competitive antibodies and conjugates, and consequently, there could be wide variations in their diagnostic performance. RBT, I-ELISA and FPA are considered suitable screening tests for the detection of infected herds or to confirm the absence of infection in brucellosis-free herds. However, at an individual level, the status of seropositive animals may be difficult to interpret without a confirmatory and/or complementary strategy including either other serological tests (usually CFT or C-ELISA), brucellin skin test or direct tests such as culture and/or PCR on target organs sampled at the abattoir. According to the recommendations made by the European Food Safety Authority (EFSA), by request from the European Commission for brucellosis diagnostic methods for bovines, sheep and goats2, the C-ELISA ‘should remain in the EU legislation on intra-community trade as a complementary test, pending the conduct of further studies conducted in accordance with the World Organisation for Animal Health (OIE) procedure’. This study aimed to address this recommendation and to evaluate three C-ELISAs and FPA for the diagnosis of bovine brucellosis and compare their performance with RBT, CFT, indirect ELISA and FPA.
ity and affinity to the S-LPS of Brucella. For all tests, the same batch of the antigen or kit was used with all samples tested. Samples were stored frozen. Once thawed, they were stored at 4 °C and tested within 48 h in the EURL by the same two experienced technicians. All tests were performed according to OIE standards for RBT, SAT, CFT and FPA3 and according to the manufacturers’ instructions for the ELISAs. For CFT, cut-offs were those recommended by Annex C of the 64/432 EC directive for CFT. For SAT, cut-offs were those recommended by the test manufacturers (Table 1).
Materials and methods
Epidemiological indicators and statistical methods
Sera
The specificities of C-ELISAs and FPA were estimated in cattle from OBF herds. The sensitivities of these tests were estimated in animals that were positive for Brucella spp. by culture and had also obtained at least two positive results among the usual tests (RBT, CFT, SAT and I-ELISA). The sensitivities of the usual tests were assessed in animals that were positive for Brucella spp. by culture and had also obtained at least two positive results among three usual tests. The defined significance level of statistical tests (McNemar’s test for paired data) was 5%. The second aim of this study was to evaluate the C-ELISA and FPA thresholds to optimise their accuracy. Receiver operating curves were constructed using R (The R foundation) equipped with DiagnosisMed and ROCR packages. The area under curve (AUC) and the threshold maximising the Youden indices (J = Se + Sp − 1) were calculated.
The study included sera from 5111 adult cattle, submitted to laboratories in 2011. Three French regional veterinary diagnostic laboratories were asked to provide the EU Brucellosis Reference Laboratory (EURL, Anses, Maisons-Alfort, France) with bovine sera from French OBF herds. Samples (n = 4330) were received in sufficient volume for the study (2–5 mL) from Saône-et-Loire (n = 1409), Normandy (n = 1483) and Brittany (n = 1538). French sera were a convenience sample of specimens submitted for routine serological diagnosis of bovine diseases in adult beef cattle (>24 months); there were no duplicates. Selection was independent of brucellosis test results. National Reference Laboratories from EU member states where bovine brucellosis had yet not been eradicated were requested to provide the EURL with sera from unvaccinated adult cattle confirmed to be infected with brucellosis (isolation of Brucella spp. by culture). Sera (n = 681) were received in sufficient volume from the following countries: Belgium (n = 83), Greece (n = 264), Italy (n = 37), Northern Ireland (n = 138), Portugal (n = 124) and Spain (n = 35). Animals from each country were from several herds. Screening tests Eight serological tests were performed. Five of them were approved by the EU (EU, 2008): RBT, CFT, SAT, I-ELISA and FPA. Antigens and kits (IDEXX Brucellosis Serum Ab Test) were supplied by IDEXX and all met the standardisation criteria in the annex C of the 64/432 EC directive (EU, 2008). SAT was performed in tubes. FPA antigen test kits (B. abortus antibody test kit) conformed to the standardisation criteria defined in the annex C of the 64/432 EC directive and were supplied by Diachemix. For the batch used in this study, results with the OIE ELISA standard serum (OIEELISASPSS) were as follows: positive up to 1/4, suspect/inconclusive at 1/8 and negative at 1/16 dilutions (made in a negative pool of bovine sera). Results with the French National secondary standard for ELISA were as follows: positive up to 1/8, suspect/ inconclusive at 1/16 and negative at 1/32 dilutions (made in a negative pool of bovine sera). The three C-ELISA kits available on the EU market (C-ELISA1, Ingezim Brucella Compac, Ingenasa; C-ELISA2, Compelisa, AHVLA Scientific; C-ELISA3, SVANOVIR Brucella-Ab c-ELISA, Svanova) were assessed. These three C-ELISAs were based on different biological principles. For C-ELISA1, an anti-Brucella S-LPS monoclonal antibody conjugated to the peroxidase was added after test sera were incubated for 1 h with the S-LPS coated plates (two-step competition). The reaction was then revealed with the addition of the chromogen substrate. For C-ELISA2, test sera were incubated for 30 min with S-LPS coated plates, with an anti-Brucella S-LPS monoclonal antibody conjugated to the peroxidase (one-step competition). The reaction was then revealed with the addition of the chromogen substrate. For C-ELISA3, test sera were incubated for 30 min with S-LPS coated plates with an anti-Brucella S-LPS monoclonal antibody (one-step competition). The reaction was revealed with the addition first of an anti-mouse Ig conjugated to the peroxidase, then with the addition of the chromogen substrate. The three monoclonal antibodies used in the three respective tests were from different sources; therefore they differed in their activ-
2 See: Opinion of the Scientific Panel on Animal Health and Welfare (AHAW) on a request from the Commission concerning Brucellosis Diagnostic Methods for Bovines, Sheep, and Goats http://www.efsa.europa.eu/en/efsajournal/pub/432 (accessed 23 June 2016).
Table 1 Cut-offs recommended by the manufacturers of the indirect ELISA (I-ELISA), fluorescence polarisation assay (FPA) and three competitive ELISAs (C-ELISAs1–3). Test
Cut-offs and interpretation of the result
I-ELISA
Negative, <110% Inconclusive, 110–120% Positive, >120% Negative, <10 Inconclusive, 10–20 Positive, >20 Positive, >40% Positive, ≤60% Positive, ≥30%
FPA
C-ELISA1 C-ELISA2 C-ELISA3
Results The results of the eight tests in the OBF population and in infected herds are displayed in Tables 2–4. Comparison with results from the OBF regions showed false positive serological reactions (FPSR) in 137 animals. Most FPSR occurred in Saône-et-Loire (85/ 1469; 5.8%) and Brittany (47/1538; 3.1%) and a few occurred in Normandy. Most FPSR occurred in only one test (124/137; 90.5%); 9.5% of samples obtained FPSR to at least two tests (13/137; 9.5%). These FPSR mainly occurred with the C-ELISAs, especially C-ELISA3. Only a few animals had inconclusive results with I-ELISA (2/4430 in OBF herds; 4/681 in infected herds), but inconclusive results were more frequent with FPA, especially in the infected population (29/ 4430 animals from OBF herds; 29/681 animals in infected herds). The estimated sensitivities and specificities are presented in Tables 5 and 6. The most specific tests were CFT (99.98%; 95% confidence interval, CI, 99.93–100%), SAT (99.98%; 95% CI, 99.93– 100%) and RBT (99.89%; 95% CI, 99.79–99.99%). C-ELISA1 (99.37%; 95% CI, 99.13–99.61%) and C-ELISA2 (99.35%; 95% CI, 99.11–99.59%) were less specific than RBT (P = 6 × 10−5 and 8 × 10−6, respectively), CFT (P = 1 × 10−5 and 8 × 10−7, respectively) and SAT (P = 2 × 10−6 and 3 × 10−7, respectively). C-ELISA3 (Sp, 98.28%; 95% CI, 97.91–98.67%) was the least specific test. C-ELISA1 (Se, 98.4%; 95% CI, 97.0– 99.8%) was significantly more sensitive than SAT (Se, 81.3%; 95% CI, 77.0–85.7%; P = 4 × 10−12), C-ELISA2 (Se, 90.0%; 95% CI, 86.6–93.3%; P = 1 × 10−5) and C-ELISA3 (Se, 94.2%; 95% CI, 91.6–96.8%; P = 2 × 10−4). When inconclusive results were interpreted as positive results, FPA (Se, 99.0%; 95% CI, 97.9–100%) was more sensitive than CFT (Se, 95.0%; 95% CI, 92.6–97.5%; P = 3 × 10−3), C-ELISA2 (P = 5 × 10−7) and C-ELISA3 (P = 3 × 10−4). Most likely due to the small sample size from
3
See footnote 1.
40
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
Table 2 Results of cattle from officially brucellosis free farms according to the geographic region. Brittany (n = 1538)
Test
RBT CFT SAT I-ELISA C-ELISA1 C-ELISA2 C-ELISA3 FPA
Normandy (n = 1483)
Saône et Loire (n = 1409)
Total (n = 4430)
P
I
N
P
I
N
P
I
N
P
I
N
0 0 1 (0.07) 3 (0.20) 14 (0.91) 25 (1.63) 49 (3.19) 6 (0.39)
– – – 1 (0.07) – – – 12 (0.78)
1538 1538 1537 1534 1524 1513 1489 1520
0 1 (0.07) 0 0 0 3 (0.20) 17 (1.15) 1 (0.07)
– – – 0 – – – 7 (0.47)
1483 1482 1483 1483 1483 1480 1466 1475
5 (0.35) 0 0 5 (0.35) 14 (0.99) 1 (0.07) 10 (0.71) 2 (0.14)
– – – 1 (0.07) – – – 10 (0.71)
1404 1409 1409 1403 1395 1408 1399 1397
5 (0.11) 1 (0.02) 1 (0.02) 8 (0.18) 28 (0.63) 29 (0.65) 76 (1.72) 9 (0.20)
– – – 2 (0.05) – – – 29 (0.65)
4425 4429 4429 4420 4402 4401 4354 4392
P, positive (% tested); I, inconclusive (% tested); N, negative; RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay.
Table 3 Test results in cattle from officially brucellosis free farms, according to the geographic region. RBT
CFT
SAT
I-ELISA
C-ELISA 1
2
N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N P N N N N N P N N N N P N N N N N P N N N N N P P N N N I N N N N N P N N N N N P P N N N N P P N N N N P P P N N P N P P N P N N N N P N N N N N P N N N P N P N N P P N Total Number of animals with false positive results to either test (% tested animals)
FPA
Brittany
Normandy
Saône and Loire
Total
N I P N I N N N N N N N N I P N N N N P
1436 11 6 0 1 21 2 10 2 1 1 3 0 0 0 1 0 0 0 0 1538 47 (3.1)
1454 7 1 0 0 3 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1483 5 (0.3)
1375 8 0 69 0 0 0 7 0 0 1 0 1 2 1 0 0 2 2 1 1409 85 (5.8)
4265 26 7 69 1 24 2 17 2 1 2 3 1 2 1 1 1 2 2 1 4430 137 (3.1)
3 N N N P P N P N P N N N N N P P N N N N
RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay; N, negative; I, inconclusive; P, positive.
most countries, no significant differences could be demonstrated at country level except for Greece, where the same overall trends were observed. These results suggested that the cut-offs for C-ELISAs and FPA could be changed to improve their sensitivities and/or specificities. ROC curves are displayed in Figs. 1A-D. The optimised cutoffs for C-ELISA1 (40%) and C-ELISA3 (30%) agreed with the manufacturers’ recommendations (Figs. 1A, C). The cut-off optimising both Se and Sp for C-ELISA2 was 70%, whereas the manufacturer recommended 60% (Fig. 1B). The optimised cut-off for FPA was 10, with no inconclusive range (Fig. 1D). Discussion The aims of this study were to estimate the Se and the Sp of C-ELISAs and FPA and to calculate the thresholds of C-ELISAs and FPA to optimise their accuracy. In our study, none of the usual tests (SAT, CFT, RBT and I-ELISA) recommended by the European legislation were able to confirm all cases. The most sensitive tests were C-ELISA1, FPA and RBT. The least sensitive test was SAT. The most specific tests were CFT, SAT and RBT, while the least specific test was C-ELISA3. Increasing the cut-off for C-ELISA2 from 60% to 70% could improve its accuracy when a better compromise between Se and Sp is needed, while in OBF regions the usual cut-off should be
used in order to optimise the Sp. A single cut-off could be chosen for FPA in order to eliminate the inconclusive range and make the interpretation of results easier. The opinion of the EFSA on a request from the European Commission concerning brucellosis diagnostic methods for bovines4 considered RBT, CFT and I-ELISA to be standard tests. It also suggested the inclusion of FPA in the EU legislation, since its Se and Sp were comparable to standard tests. It was therefore logical to recommend that SAT should not remain in the EU legislation on trade because of its low Se, and that C-ELISA should be used as a complementary test, until further investigations were performed. The results of a meta-analysis presented in this opinion are summarised in Table 7. The sensitivities estimated in our study were not significantly different from these results, whereas the specificities estimated in our study were significantly higher. Abernethy et al. (2012) estimated individual sensitivities of six tests for bovine brucellosis diagnosis and reported that RBT (78.1%) and I-ELISA (67.9%) had the highest sensitivities relative to culture. I-ELISA, RBT and CFT (100%) had the highest specificities. C-ELISA3 was also evaluated in that study and its Se and Sp were the lowest of the tests investigated, which agrees with our results. Other studies
4
See footnote 2.
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
41
Table 4 Test results for 681 sera from infected herds. RBT
N N N N N N N N N N N N N N N N N N N N N N N N N P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P Total
CFT
N N N N N N N N N N N N N N N N N N N N P P P P P N N N N N N N N N N N N N N N N P P P P P P P P P P P P P P P P
SAT
N N N N N N N N N N N N N N N N N N N P N N N N N N N N N N N N N N N P P P P P P N N N N N N N N P P P P P P P P
I-ELISA
N N N N N N N I I I P P P P P P P P P P N P P P P N N N N N I P P P P N N N N N P P P P P P P P P N N N N P P P P
C-ELISA 1
2
3
N N N N P P P N P P N N N P P P P P P N N N P P P N P P P P P N P P P P P P P P P N P P P P P P P P P P P N P P P
N N N N N N N N N N N N N N N N N P P N N P N N P N N N N P N N N N P N N N P P P P N N N P P P P N N P P P N P P
N N P P N N P N N N N N N N N N P N P N N P N P N N N N P N N N P P P N N P P P P P N P P N N P P N P P P P P N P
FPA
B
Sp
Gr
N-I
It
Pt
Total
N I N I N I N N N I N I P N I P I P P N N P P P P N N I I I I P I P P N I I I P P P P I P N P D P P P D P P P P P
56 – 3 1 2 – 1 – – – – – – – – 1 – – 1 – – 1 – – – 2 1 – – – – – – – – – 1 – – 1 – 1 1 – – – – – 1 – – 1 – 1 – – 7 83
12 1 – – – – – – – – 1 – – – 2 – – – – – – – 1 – – 1 – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – 17 35
17 – – – – – – – – – 1 2 – – 1 3 1 – – 1 – – – 2 1 – 2 – 1 – 1 1 2 1 1 – – – – – 1 – 2 1 7 1 3 1 23 1 – – 1 – 5 2 178 264
118 – 2 – 1 1 – – – 1 – – – – – – – – – – – – 1 – – – 2 1 – 1 – – – – – 1 – 1 1 3 – – – – – – – – – – 1 – 1 – – – 2 138
11 – – – – – – – – – – – – – 1 – 1 – – – 1 – – – – 2 – – – – – – – – – – – – – – – – – – – – – – 2 – – – – – – – 19 37
102 2 1 – 1 – – 1 1 – 2 – 1 1 1 – – 1 – – – – – – – 1 – – – – – – – – – – – – – – – – 1 – – – – – 1 – – – – – – – 7 124
316 3 6 1 4 1 1 1 1 1 4 2 1 1 5 4 2 1 1 1 1 1 2 2 1 6 5 1 1 1 1 1 2 1 1 1 1 1 1 4 1 1 4 1 7 1 3 1 27 1 1 1 2 1 5 2 230 681
RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay ; B, Belgium; Sp, Spain; Gr, Greece ; N-I, Northern Ireland; It, Italy; Pt, Portugal; N, negative; I, inconclusive; P, positive.
published in the last decade have aimed to assess the sensitivities and specificities of tests for the diagnosis of bovine brucellosis (Fosgate et al., 2006; Muma et al., 2007; Matope et al., 2011; Rahman et al., 2013; Sanogo et al., 2013; Gorsich et al., 2015), but they were performed in different epidemiological contexts in different countries (Bangladesh, Ivory Coast, South Africa, Trinidad, Zambia, Zimbabwe), and in different species (cattle, African buffalos, water buffaloes, sheep, goats). Further, they were performed on smaller sample sets than our study and specificities were often estimated
in brucellosis free herds from infected regions. Test methods and thresholds varied greatly from one study to another. According to Muma et al. (2007), the Se of C-ELISA3 was 97% (95% credibility interval, CrI, 93–100%) and its Sp was 54% (95% CrI, 34–75%); FPA was less sensitive (Se, 72%; 95% CrI, 63–80%) than C-ELISA3 but more specific (FPA: Sp, 93%; 95% CrI, 85–99%). Matope et al. (2011) also reported that C-ELISA3 was highly sensitive (Se, 99%; 95% CrI, 94.8–100%) and quite specific (Sp, 95.4%; 95% CrI, 93.7–96.7%). The FPA performed on serum was more specific (Sp, 99%; 95% CrI,
42
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
Table 5 Estimation of the specificities of the eight tests (n = 4430, officially brucellosis free animals). Test
RBT CFT SAT I-ELISA C-ELISA1 C-ELISA2 C-ELISA3 FPA
Brucellosis-free population P
I
N
5 1 1 8
– – – 2
4425 4429 4429 4420
28 29 76 9
– – – 29
4402 4401 4354 4392
Estimated specificity mean (95% CI) 99.89% (99.79–99.99%) 99.98% (99.93–100%) 99.98% (99.93–100%) I = N: 99.82% (99.70–99.95%) I = P: 99.77% (99.64–99.92%) 99.37% (99.13–99.61%) 99.35% (99.11–99.59%) 98.28% (97.91–98.67%) I = N: 99.80% (99.67–99.93%) I = P: 99.14% (98.88–99.42%)
CI, confidence interval; RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay ; P, positive; I, inconclusive; N, negative; I = N, inconclusive results interpreted as negative results; I = P, inconclusive results interpreted as positive results.
98.0–99.6%) but less sensitive (FPA: Se, 87%; 95% CrI, 78.4–93.4%) than C-ELISA3. However, these results are unlikely to be generalisable to European cattle or to our results, since the contexts, sampling schemes and statistical methods were different. The estimation of the diagnostic Se and Sp of a test should ideally be derived from testing an adequate sampling of animals of known history and infection status. The selection of an appropriate reference standard infected population (culture-positive animals) is complicated and expensive. Furthermore, Brucella-culture should be performed in conditions that optimise Se, which could lead to an overestimation of the diagnostic Se of the tests under study. In our study, the infection status of herds was confirmed by culture, but the individual infection status of animals was unknown. In the absence of single reference standard, we used a composite reference standard (Rutjes et al., 2007), based on the infection status of the herd and the individual results obtained with the other tests, to perform a direct assessment. Our results were then conditional on a positive reference and should be interpreted as relative Se and Sp. This method is quite easy to implement and is widely used for many diagnostic tests (Alonzo and Pepe, 1998; Rutjes et al., 2007). This kind of reference standard is more discriminant than each of its components in isolation (Alonzo and Pepe, 1999; Pepe, 2003; Rutjes et al., 2007) and even if the reference is imperfect, the infection status of such animals is almost certainly correct. Moreover,
Table 6 Sensitivity estimations for the eight tests (n = 309, infected animalsa). Test
RBT CFT SAT I-ELISA C-ELISA1 C-ELISA2 C-ELISA3 FPA
Infected animals P
I
N
289 288 244 289 304 278 291 298
– – –
7 15 56 13 5 31 18 3
– – – 8
Estimated sensitivity mean (95% CI) 97.7% (95.9–99.3%) 95.0% (92.6–97.5%) 81.3% (77.0–85.7%) 95.7% (93.4–98.0%) 98.4% (97.0–99.8%) 90.0% (86.6–93.3%) 94.2% (91.6–96.8%) I = N: 96.4% (94.4–98.5%) I = P: 99.0% (97.9–100%)
CI, confidence interval; RBT, Rose-Bengal test; CFT, complement fixation test; SAT, serum agglutination test; I-ELISA, indirect ELISA; C-ELISA, competitive ELISA; FPA, fluorescence polarisation assay; P, positive; I, inconclusive; N, negative; I = N, inconclusive results interpreted as negative; I = P, inconclusive results interpreted as positive. a Animals from an infected herd and with positive results to at least two of any of the four usual tests (RBT, CFT, SAT and I- ELISA); sensitivities of the usual tests were performed on animals that had positive results to at least two of the three other usual tests and were from an infected herd.
in diagnostic test accuracy studies, the animals sampled should also be representative of the region where the test is to be used5. Our sample set included bovine sera from several European countries. Some of them were collected in France, which has been an OBF country since 2005. In France, screening tests are performed annually in farms. When a brucellosis outbreak occurs, all animals of the infected herd are slaughtered and an epidemiological survey is launched to find the sources of contamination and determine the potential consequences of the outbreak in epidemiologically related herds. The epidemiological situation is thus strictly controlled. For this reason, the brucellosis free status of cattle from OBF herds is reliable and OBF samples can be considered representative of OBF European regions. Nevertheless, in one of the OBF regions (Saone-et-Loire), FPSR due to cross-reactions with environmental contaminants occur quite frequently (Godfroid et al., 2002). This suggests that highly specific tests should be chosen in such regions (Pouillot et al., 1998). The sera from infected herds were collected in several countries where bovine brucellosis outbreaks occurred. Infection was confirmed by the isolation of Brucella abortus or Brucella melitensis. Infected samples could thus be considered representative of animals from infected herds. Since the infected sera originated from several herds, a cluster effect cannot be excluded, but given the large sample size and the number of herds selected, this effect should be limited. Furthermore, little information was available about the animals (e.g. age, breed, gestational status, herd production type, or species of Brucella isolated in infected herds). It would have been interesting to take these parameters into account to study the accuracy of tests in various situations and disease conditions, e.g. according to Abernethy et al. (2012), the tests are likely to be more sensitive in cows in late gestation than those in early gestation. Conclusions This study was based on a large sample of cattle, which were representative of European cattle. CFT and RBT were the most accurate screening tests for surveillance in OBF regions. Among the new tests, FPA and C-ELISA1 were the most sensitive. C-ELISAs did not have higher specificities than the usual tests, and their sensitivities varied according to the kit used. C-ELISA2 was less sensitive than the usual tests, but was quite specific. Therefore, C-ELISA2 should be used as a screening test only in OBF regions; it should not be used as a confirmatory test. C-ELISA3 was the least specific test investigated and it was not very sensitive. It should not be used as a screening test in OBF regions, or as confirmatory test on single animals in a FPSR context. FPA could be used as a screening test, with inconclusive results interpreted as negative in OBF regions and as positive in infected regions. Given the imperfect sensitivities and specificities of tests studied, a combination of several tests is required to improve the accuracy of diagnosis: (1) in series in a low epidemiological risk context and (2) in parallel in outbreaks to hasten the eradication of the disease. These recommendations should nevertheless be confirmed by other studies because of the limitations of our study protocol. A larger sample set should be used and the concordances between the tests should be studied to assess the characteristics of screening schemes that use these tests in series or in parallel. Conflict of interest statement All authors certify that they have no affiliations with or involvement in any organisation or entity with any financial interest or
5
See footnote 1.
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
43
0.6 0.4 0.0
0.2
True positive rate
0.8
1.0
(A)
0.0
0.2
0.4
0.6
0.8
1.0
False positive rate
0.6 0.4 0.0
0.2
True positive rate
0.8
1.0
(B)
0.0
0.2
0.4
0.6
0.8
1.0
0.8
1.0
0.8
1.0
False positive rate
0.6 0.4 0.0
0.2
True positive rate
0.8
1.0
(C)
0.0
0.2
0.4
0.6 False positive rate
0.6 0.4 0.0
0.2
True positive rate
0.8
1.0
(D)
0.0
0.2
0.4
0.6 False positive rate
Fig. 1. Receiver operating curve (ROC) for competitive ELISAs (C-ELISAs) and fluorescence polarisation assay (FPA). ROC for C-ELISA1, C-ELISA2, C-ELISA3, and FPA are presented in (A), (B), (C), and (D), respectively. Cut-off optimising both the sensitivity (Se) and the specificity (Sp) of C-ELISA1, C-ELISA2, C-ELISA3, and FPA was 40% (Se, 98.7%; Sp, 99.3%; area under curve, AUC, 99.1%), 70% (Se, 92.2%; Sp, 97.0%; AUC, 97.6%), 30% (Se, 94.5%; Sp, 97.6%; AUC, 98.5%), and 10% (Se, 99.4%; Sp, 99.1%; AUC, 99.8%), respectively. For C-ELISA2, the result index decreases with the antibody titre; the curve was adapted to make reading easier.
44
A. Praud et al. / The Veterinary Journal 216 (2016) 38–44
Table 7 Predictions of the sensitivities and specificities of tests evaluated for cattle, based on logistic meta-regression (European Food Safety Authority, 2006). Test
Sensitivity (95% credibility interval)
Specificity (95% credibility interval)
C-ELISA (various) C-ELISA3 CFT FPA I-ELISA RBT SAT
0.983 (0.977–0.988) 0.954 (0.917–0.981) 0.960 (0.949–9.970) 0.987 (0.981–0.992) 0.976 (0.941–0.992) 0.981 (0.968–0.991) 0.817 (0.749–0.872)
0.998 (0.997–0.998) 0.992 (0.987–0.995) 0.998 (0.997–0.998) 0.998 (0.998–0.999) 0.975 (0.972–0.979) 0.998 (0.997–0.998) 0.987 (0.986–0.987)
C-ELISA, competitive ELISA; CFT, complement fixation test; FPA, fluorescence polarisation assay; I-ELISA, indirect ELISA; RBT, Rose-Bengal test; SAT, serum agglutination test.
non-financial interest in the subject matter or materials discussed in the manuscript. Acknowledgements The authors thank the French Departmental Veterinary Laboratories of Ille-et-Vilaine, Manche and Saône-et-Loire, Drs Eric LE DREAN, Michaël TREILLES and Pascal VERY, particularly for the collection of the sera from French bovine brucellosis officially free herds. Antoine DRAPEAU and Yannick CORDE are also duly acknowledged for their technical assistance. All serological analyses could be performed thanks to the European Commission’s support to the EU reference laboratory for Brucellosis. References Abernethy, D.A., Menzies, F.D., McCullough, S.J., McDowell, S.W.J., Burns, K.E., Watt, R., Gordon, A.W., Greiner, M., Pfeiffer, D.U., 2012. Field trial of six serological tests for bovine brucellosis. The Veterinary Journal 191, 364–370. Alonzo, T.A., Pepe, M.S., 1998. Assessing the accuracy of a new diagnostic test when a gold standard does not exist. UW Biostatistics Working Paper Series 156.
Alonzo, T.A., Pepe, M.S., 1999. Using a combination of reference tests to assess the accuracy of a new diagnostic test. Statistics in Medicine 18, 2987–3003. European Food Safety Authority, 2006. Scientific opinion on “Performance of Brucellosis Diagnostic Methods for Bovines, Sheep, and Goats”. EFSA Journal 432, 1–44. EU, 2008. Commission Decision of 10 December 2008 amending Annex C to Council Directive 64/432/EEC and Decision 2004/226/EC as regards diagnostic tests for bovine brucellosis (2008/984/EC). Official Journal of the European Union L. 352/38–L. 352/45. Fosgate, G.T., Adesiyun, A.A., Hird, D.W., Hietala, S.K., 2006. Likelihood ratio estimation without a gold standard: A case study evaluating a brucellosis C-ELISA in cattle and water buffalo of Trinidad. Preventive Veterinary Medicine 75, 189–205. Gall, D., Nielsen, K., 2004. Serological diagnosis of bovine brucellosis: A review of test performance and cost comparison. Revue Scientifique et Technique de l’OIE 23, 989–1002. Godfroid, J., Saergerman, C., Walravens, K., Letesson, J.J., Tibor, A., Macmillan, A., Spencer, S., Sanaa, M., Bakker, D., Pouillot, R., et al., 2002. How to substantiate eradication of bovine brucellosis when aspecific serological reactions occur in the course of brucellosis testing. Veterinary Microbiology 90, 461–677. Gorsich, E.E., Bengis, R.G., Ezenwa, V.O., Jolles, A.E., 2015. Evaluation of the sensitivity and specificity of an enzyme-linked immunosorbent assay for diagnosing brucellosis in African Buffalo (Syncerus Caffer). Journal of Wildlife Diseases 51, 9–18. Matope, G., Muma, J.B., Toft, N., Gori, E., Lund, A., Nielsen, K., Skjerve, E., 2011. Evaluation of sensitivity and specificity of RBT, C-ELISA and fluorescence polarization assay for diagnosis of brucellosis in cattle using latent class analysis. Veterinary Immunology and Immunopathology 141, 58–63. Muma, J.B., Toft, N., Oloya, J., Lund, A., Nielsen, K., Samui, K., Skjerve, E., 2007. Evaluation of three serological tests for brucellosis in naturally infected cattle using a latent class analysis. Veterinary Microbiology 125, 187–192. Pepe, M.S., 2003. The Statistical Evaluation of Medical Tests for Classification and Prediction. Oxford University Press, Oxford, p. 320. Pouillot, R., Lescoat, P., Garin-Bastuji, B., Repiquet, D., Terrier, P., Gerbier, G., Bénet, J.J., Sanaa, M., 1998. Risk factors for false-positive serological reactions for bovine brucellosis in Saône-et-Loire (France). Preventive Veterinary Medicine 35, 165–179. Rahman, A.K.M.A., Saegerman, C., Berkvens, D., Fretin, D., Gani, M.O., Ershaduzzaman, M., Ahmed, M.U., Abatih, E., 2013. Bayesian estimation of the true prevalence, sensitivity and specificity of indirect ELISA, Rose Bengal Test and Slow Agglutination Test for the diagnosis of brucellosis in sheep and goats in Bangladesh. Preventive Veterinary Medicine 110, 242–252. Rutjes, A.W.S., Reitsma, J.B., Coomarasamy, A., Khan, K.S., Bossuyt, P.M.M., 2007. Evaluation of diagnostic tests when there is no gold standard. A review of methods. Health Technology Assessment 11, 72. Sanogo, M., Thys, E., Achi, Y.L., Fretin, D., Michel, P., Abatih, E., Berkvens, D., Saegerman, C., 2013. Bayesian estimation of the true prevalence, sensitivity and specificity of the Rose Bengal and indirect ELISA tests in the diagnosis of bovine brucellosis. The Veterinary Journal 195, 114–120.