Validation using sera from a previously study to characterize the in-house Bb-cELISA1
Using the first set of sera, previously established to characterize the in-house competitive ELISA Bb-cELISA1, the BIO K 466 showed a diagnostic sensitivity of 97.9% (95% CI: 91.9% – 99.6%) and diagnostic specificity of 99.5% (95% CI: 96.9% – 100%) relative to reference consisting of four previously published tests, i.e. two immunoblots using B. besnoiti tachyzoite or bradyzoite antigens [15], IFAT based on tachyzoite antigen [15] and an in-house BbAPureELISA [10] were similar to those of the in-house Bb-cELISA1 [11] (Additional file 1: Dataset S1).
A further validation, using 200 immunoblot-confirmed positive bovine sera from France and 628 negative sera from other European countries, where cases of bovine besnoitiosis have not reported yet (Additional file 2: Dataset S2), and revealed a diagnostic sensitivity of 97.5% (95% CI: 93.9%–99.1%) and specificity 99.5% (95% CI: 98.5%–99.9%).
So far, only indirect ELISA tests based on total tachyzoite cell lysates, lyophilized tachyzoites or semi-purified but nevertheless complex tachyzoite antigens (i.e. antigens enriched for diagnostically relevant antigens) have been established. This also includes a number of commercialized tests, and no competitive ELISA was available to diagnose besnoitiosis yet [2]. For these indirect ELISA tests, median diagnostic sensitivities of 92.2% (minimum 75.5%, maximum 100%) and median diagnostic specificities of 97.3% (minimum 93%, maximum 100%) were reported, as retrieved from previously reviewed publications [2]. Sets of sera used to validate test characteristics of various indirect ELISA were often not the same; in addition, criteria to define positivity vary between studies. Thus, estimates on diagnostic characteristics provided by different groups for these ELISAs are not directly comparable. However, in the present study, and independent of the set of sera used, levels for estimates for diagnostic sensitivity and specificity reached values exceeding those reported previously. This is an indication that the diagnostic characteristics for the BIO K 466 are probably better than at least for some of the indirect in-house and commercial B. besnoiti ELISAs based on crude or semi-purified antigens. However, because authors of this publication have competing interests, an independent validation by others using additional sets of reference sera, optimally testing some of other indirect ELISAs in parallel, is needed.
One of the five tests used to characterize B. besnoiti positivity in the first set of sera (i.e. four reference tests and the Bb-cELISA1) used in this study was the IFAT. IFAT titres directly reflect the concentration of antibodies in a serum, specific for the surface of B. besnoiti tachyzoites. When selecting sera for this study we specifically focussed on sera which showed only low IFAT titres (1:100, 1:200) and added fewer sera with high or very high IFAT titres to the first panel. In the comparison of IFAT titres vs. the inhibition values in BIO K 466, all sera with IFAT titres of 1:200 and most (17/19) sera with IFAT titres of 1:100 showed inhibition in the competitive ELISA exceeding the cut-off of 0.50 (Fig. 1A). This finding is in accord with a previous study [15], which showed that at an IFAT titre of 1:100 the diagnostic specificity of this test started to fade and IFAT tended to produce false-positive results, at least under the test conditions in our laboratory [15].
The median values for the inhibition in BIO K 466 for B. besnoiti-positive bovine sera increased with the number of positive results which had been gained for individual sera in reference tests (Fig. 1B). The only two false-negative sera in BIO K 466 were positive in two or three of the four reference tests only (Fig. 1B). In addition, the results in Fig. 1B show that, even in cases where not all reference tests are positive, BIO K 466 was sensitive enough to test these sera positive.
Relation of inhibition in the Monoscreen AbELISA Besnoitia besnoiti to those obtained with the in-house Bb-cELISA1
The results of the BIO K 466 with the results in the in-house Bb-cELISA1 were well correlated, as characterized by a relatively high R2 value of 0.86 (Fig. 2). In addition, the cut-offs applied in both tests seem to be very similar and the positive-negative test result of both tests largely agreed (Fig. 2) with only five sera of a total number of 305 showing divergent results. These five sera consisted of two which were reference-positive and tested false-negative in BIO K 466; one reference positive serum tested correct-positive in BIO K 466 but false-negative in the in-house Bb-cELISA1, a reference-negative serum tested false positive in the BIO K 466 and the remaining serum was reference-negative but with an immunofluorescence titer of 1:100 which tested correct-negative in BIO K 466 but false-positive in the in-house Bb-cELISA1 (Fig. 2).
Overall, a direct comparison of the results revealed an almost perfect agreement between the results of the in-house Bb-cELISA1 and the commercial version (kappa 0.98; 95% CI: 0.95–1). Thus, it can be concluded that differences in the test design between both the commercial and the in-house version of the competitive ELISA, i.e. using crude B. besnoiti tachyzoite antigens instead of semi-purified antigen and the monoclonal antibody directly conjugated to the reporter enzyme instead of using a secondary anti-mouse conjugate [11], respectively, had only a marginal effect on the inhibition values observed for tested sera in the ELISA, the commercial and the in-house Bb-ELISA1 (Fig. 2).