Non-agreement between 2 serologic techniques for detecting antibody to Brucella ovis in naturally infected sheep

Abstract

Brucella ovis is the causative agent of ovine brucellosis, which is an important infectious disease in sheep farming worldwide and is responsible for economic losses because of its negative effect on the reproductive system of rams and ewes. Serologic tests are the main tools for detection of infection; however, these tests commonly yield a high frequency of false-negative results. We compared 2 serologic tests, agar gel immunodiffusion (AGID) and ELISA, for the detection of anti–B. ovis antibodies in naturally infected sheep. Of the 728 serum samples analyzed, 0.3% were positive by AGID and 9.2% by ELISA. Positive results were obtained for different animals and flocks. There was no statistical difference between the detection frequency of the 2 methods (p = 0.674), and the kappa test indicated low concordance (κ = 0.005). The lack of agreement between results obtained using AGID and ELISA, associated with the absence of clinical signs, makes it difficult to detect ovine brucellosis efficiently, and demonstrates the need for effective tests for the definitive detection of B. ovis infection.

Ovine brucellosis, caused by Brucella ovis infection, is one of the most important infectious diseases affecting sheep farming worldwide. Therefore, diagnosis, control, and prevention of this disease are important. ³ B. ovis is a bacterial, non-zoonotic species, ¹⁰ responsible for clinical or subclinical chronic disease, ¹³ which may result in economic losses through decreased fertility in rams and lower conception rates in ewes. ¹² B. ovis is transmitted sexually, and B. ovis may be excreted by rams in semen and by ewes in vaginal discharge and milk. ¹ The most common clinical finding of B. ovis infection in rams is epididymitis; however, <50% of rams develop clinical disease. ¹²

Serologic tests are the most useful epidemiologic tools for the detection of B. ovis exposure and infection, ⁵ and include agar gel immunodiffusion (AGID), complement fixation, and ELISA. ⁸ The World Organisation for Animal Health (OIE) ¹³ recommends AGID for the detection of anti–B. ovis antibodies. However, serologic tests for anti–B. ovis antibodies commonly yield highly variable results with a considerable frequency of false-negatives. ⁹

We compared AGID and ELISA for the detection of anti–B. ovis antibodies in naturally infected sheep to assess their performance for use in detection of infection in sheep flocks, which may guide veterinarians and sheep farmers in deciding between assay choices. The experiments were approved by the Universidade Pitágoras Unopar (Londrina, Brazil) Ethics Committee on the use of animals (CEUA 006/16).

We collected 728 ovine serum samples by jugular venipuncture from males and females without clinical signs of brucellosis. The animals were from 15 flocks in the State of Parana, Brazil, and were sampled in March 2017 (Table 1). Although believed to be free of brucellosis, the sheep had never been tested for B. ovis, and could have been naturally infected with B. ovis. Sera were separated by centrifugation and stored in sterile microtubes at −80°C until analyzed.

Table 1.

Composition of the animals tested in sheep flocks historically free from Brucella ovis for surveillance by agar gel immunodiffusion and ELISA.

Farm	Total	Male	Female	Average age (mo)
1	48	24	24	21
2	48	24	24	22
3	48	24	24	22
4	48	24	24	32
5	48	24	24	33
6	48	24	24	33
7	48	24	24	35
8	48	24	24	37
9	48	24	24	40
10	49	24	25	41
11	49	24	25	43
12	49	24	25	43
13	49	24	25	46
14	50	25	25	49
15	50	25	25	49
Total	728	362	366	37

The AGID technique was performed as described previously ⁸ using a B. ovis antibody test (Parana Technology Institute) for the detection of IgG. The antigen used consisted of soluble proteins and lipopolysaccharides extracted from B. ovis (sample Reo 198); test sensitivity and specificity were reported to be 96.4% and 100%, respectively. Agar preparation was performed according to the manufacturer’s instructions. Four samples were evaluated in each petri plate (55 × 15 mm), in which positive control serum was applied to the center well. Reactions were read 72 h after sample application, and samples were considered to be positive when a clear line of precipitation was observed between the sample and the positive control.

The ELISA was performed using a commercial kit (Brucella ovis antibody test; Idexx) according to the manufacturer’s instructions; the results were interpreted with a surveillance cutoff value (S/P ratio = 45%) because the flocks were historically free from B. ovis. Diagnostic sensitivity and specificity were reported to be 91.7% and 95.2%, respectively, for detection of IgG, and the antigen used was extracted from the bacterium B. ovis, sample Reo 198.

For agreement analysis of the 2 methods, the kappa coefficient and corresponding CI were calculated using R software (http://www.R-project.org/). The confidence level for all analyses was 95%.

Of the 728 serum samples analyzed, only 2 (0.3%) were positive by AGID (95% CI 0.03–0.99); 67 (9.2%) were positive by ELISA (95% CI 7.3–11.5). There was no statistical difference between the methods in detecting the frequency of positive animals (p = 0.674); there was low agreement between the assays based on the κ coefficient (κ = 0.005).

Although the low number of seropositive animals did not permit statistical comparisons of the sensitivity and specificity of the tests, the results clearly indicated the lack of agreement between AGID and ELISA because seropositive samples were from different animals of different flocks for both tests, and AGID-positive animals were not ELISA-positive, as has been observed in other studies.^7,9 We concluded that neither of the tests was able to avoid false-positive serologic reactions under the conditions in which the study was carried out, given that the 2 AGID-seropositive sheep were ELISA-negative, and all of the ELISA-seropositive sheep were AGID-negative.

In our study, the ELISA detected B. ovis exposure in a higher percentage of sheep than AGID in asymptomatic flocks never tested previously. This is not surprising, given that ELISA can detect lower anti–B. ovis antibody concentrations than AGID. ⁶ However, the ELISA may also have detected false-positives because the ELISA did not detect the 2 animals that were positive by AGID. The AGID may have not detected truly infected animals because all of the ELISA-positive samples were AGID-negative, and these samples may have contained lower antibody levels.

International standardization of B. ovis antibody detection is lacking, and numerous independent validation studies have shown that ELISA is more sensitive than AGID ¹³ ; conversely, however, ELISA has been reported to be a less specific method because it frequently yields false-positive results. ¹¹ Thus, it would appear to be appropriate for serologic screening tests to require confirmatory tests, which would minimize the possibility of cross-reaction(s), and false-negative and false-positive results.

Similar to our results, non-agreement between AGID and ELISA was reported in a study that analyzed 1,033 sheep samples, in which AGID did not detect animals that were positive by ELISA ⁹ ; those authors interpreted the ELISA including the suspect category, and we used the new ELISA cutoff, without the suspect category, as implemented by the manufacturer in 2016. Studying the use of a commercial antigen for detection of ovine brucellosis antibodies, researchers standardized an indirect ELISA and tested 448 samples, resulting in 11 ELISA-reactive samples, which were not positive by AGID. ⁷ Although the OIE recommends AGID for the detection of anti–B. ovis antibodies, using both AGID and ELISA may increase the sensitivity of detection. ¹³

In a study of 2,276 sheep sera in the United States, no significant difference was found in B. ovis seroprevalence between an indirect ELISA and the Idexx B. ovis ELISA, although there was also poor agreement between these ELISAs. ⁴ However, when AGID was used as the reference test, the Idexx ELISA, with a moderate cutoff value (S/P ratio = 45%), demonstrated the highest relative sensitivity (38.1%) and specificity (92.0%). ⁴

The frequent non-agreement of serologic tests to detect B. ovis infection in sheep flocks and the high occurrence of subclinical infections^2,4,7,9,11 suggests that PCR may be a more promising detection method. PCR can detect B. ovis DNA from semen or vaginal swabs; however, PCR can yield false-negative results because of intermittent bacterial shedding and would be impractical for most sheep producers because these samples are difficult to obtain.^2,4 A study reported no statistical difference and low concordance between a PCR assay for detection of B. ovis DNA in urine and a serum ELISA for detection of B. ovis antibody. ² Despite the low agreement between serology and urine PCR, the combination of these methods significantly increased the frequency of detection compared with either method individually (p < 0.05). Thus, PCR may be a suitable complementary tool for B. ovis detection because seronegative rams can excrete the organism whereas seropositive rams may not. ² In any case, detection of B. ovis remains a challenge.

The most important finding of our study was the lack of agreement between AGID and ELISA, as well as the inability of AGID to detect animals that were positive by ELISA. We conclude that neither test avoided false-positive and false-negative serologic reactions. The selection of a gold standard test is an important but difficult step in studies investigating laboratory tests; however, an easily executed and reliable test for B. ovis is not available and is a limitation of our study. Our study was also limited by the fact that all of the samples tested were from farms that had never been tested for B. ovis. This issue might be resolved through a controlled study with known infection status of animals confirmed through molecular tests, with independent calculation of the sensitivity and specificity of each assay.