Serological results obtained with the three different kits showed that the blocking design kits performed better than the indirect ones as had been concluded for the use of similar kits for BoHV-1 , and identified that an alphaherpesvirus serologically related to BoHV-1 is present in semi-domesticated reindeer in Finnmark. The blocking kits were found to work efficiently without any changes to the manufacturers' protocols or pre-defined cut-off values unlike Kit A which could not be used without adaptations.
Data obtained in the virus neutralization strongly indicates that CvHV-2 is most likely the virus present in this reindeer population.
The percentage of seropositive reindeer ranged from 40–42% between kits. The low variation between the kits verified the consistency of the results. In this study, reindeer samples were tested in serological kits designed for bovine sera and it was therefore necessary to verify if the pre-established cut off values could be used for reindeer sera. From the data obtained in the kits with blocking design (B and C) even considerable changes in the cut off values (10% up or downwards) would not significantly change the results.
The Spearman coefficient is based on the ranks, reducing sensitivity to outliers that could affect the Pearson correlation coefficient. The value of Spearman's ρ calculated for each sub-populations of positive or negative results, showed that there was an association between ranks when the two blocking kits were compared as could have been expected given they were based on the same blocking ELISA design. Samples tended to score similar percentages of competition for Kits B and C even when we analyzed only those samples flanking the cut-off lines. The clustering in two populations above and below the cut-off line with similar quantitative and qualitative results was shown to be concurrent with the VNT results with the exception of two samples.
Despite using a different method, Kit A showed qualitative results (animal classified as positive or negative) very similar to the other two kits. Some association within positive results between Kits A and B further showed that the tested ELISA kits correctly classified samples even when using different methods.
Regarding the samples that scored negative in Kit A while positive or doubtful in Kits B and C, one could also consider that the difference may be due to a non specific inhibitory character in the sera or a possible difference in available epitopes for reaction between the two ELISA methods.
The analysis of variability serves as an important tool to study repeatability, and the differences between samples tested in duplicate in the same plate is a good evaluator. A mean variability in OD of 0.06 (5–95% [0.001; 0.123]) for Kit C and of 0.03 (5–95% [0.001; 0.160]) for Kit B are good evidences that gB blocking kits had a better repeatability compared to the indirect ELISA (Kit A), which had a mean variability in OD of 0.37 (5–95% [0.018; 1.188]). It is however important to remember that a direct comparison is difficult since the protocol of Kit A had to be adapted to test reindeer sera. In Kits B and C variability was obtained from the absolute difference between the observed OD for a given sample (|ODS1 - ODS2|), where S1 and S2 represent the duplicates of a given test sample. In Kit A however, there was an intermediate step for the calculation of the same value (|(ODIBR1 - ODCONTROL1) – (ODIBR2 - ODCONTROL2)|), whereCONTROL represents the control wells,IBRthe well containing the antigen and 1 and 2 the duplicates. This additional step in Kit A might also have contributed to the higher variability in Kit A versus Kits B and C.
If we consider that analytical sensitivity is the largest dilution of a high-level positive serum in which antibody is no longer detected, we observed a similar pattern for all kits, in which sample 24 remained positive at 1:256 for Kits B and C and at 1:3200 for Kit A. Sample FA16, which was another strong positive (though not as strong as number 24), became negative at 1:200, 1:64, 1:32 for Kits A, B and C respectively.
The abnormal curve observed in Kit A (Figure 3A) was repeated and confirmed and could possibly be explained by unspecific factors in the sera which interfered with the binding of the antibodies.
Given the reduced number of samples tested it is difficult to present a final conclusion for sensitivity, but we might conclude for Kits B and C that they have a good sensitivity as positive samples are still detectable 3 to 4 dilution steps below their testing dilution. Further, it is possible to conclude from the three serial dilution curves, that the blocking design kits presented a more stable curve with a moderate decrease in competition percentage when compared to the indirect ELISA kit where OD values changed abruptly and oscillated even though sensitivity also seemed to be high considering how the positive samples scored.
When comparing the ELISA designs used in this study, it is demonstrated from the serology but also from the variance and serial dilution analysis that the gB blocking design kits should be preferred to the indirect ELISA kit. This was also the situation when testing cattle, where the BoHV-1 gB kits was found more suitable as compared to kits with an indirect ELISA design [14, 24–26]. The lower performance of Kit A in this trial may have derived from the adaptations introduced and the conclusions drawn are therefore only valid regarding its adaptation to test reindeer sera as required by the aim of this study.
When comparing the two blocking ELISA kits little differences can be found, though Kit C gave less doubtful results and a slightly better repeatability. The positive control serum of Kit C performed however worse in the dilution analysis compared to the positive control of Kit B, becoming negative at dilutions of 1:8 and 1:128 respectively.
Regarding the VNT, Kramps et al.  clarified that VNT did not present sufficient advantages to be the method of choice for cattle. They showed that the ELISA kits had a higher sensitivity and specificity and that they were time and cost saving when large numbers of samples were to be tested.
Even though the ELISA kits compared in this study were designed for cattle, the genetic similarity between BoHV-1 and CvHV-2 was sufficient for all kits to detect reindeer antibodies against CvHV-2. The VNT confirmed this by showing an unequivocal higher neutralization against CvHV-2 with an average difference of three dilution steps to BoHV-1. Neutralization against other alphaherpesviruses was not performed in this study given their unlikely presence in Norway.
The VNT further showed that the cut-off values of the ELISA kits were placed at a correct percentage of competition for Kits B and C and correct OD value for Kit A. Samples classified as doubtful (Kits B and C) where negative in the VNT and only one low positive in Kit B (doubtful in Kit C) and one negative sample in Kit A might have been misclassified by the ELISA kits, if one wishes to consider the VNT as a potential gold standard test for this type of wildlife screening.
The present coefficient of antigenic similarity of 8.8 is in line with previous calculations by Lyaku et al. and Rimstad et al. who calculated it to be 9 and 8.8 respectively, even though the titres against CvHV-2 were lower in this study (1:256 maximum) than in previous ones, where reindeer sera neutralized CvHV-2 up to 1: 1024 and 1:512 respectively [18, 20].
It is important to clarify that the VNT was used mostly to confirm the presence of another alphaherpesvirus than BoHV-1, as would be expected given the BoHV-1 free status in cattle in Norway, and not specifically to compare the performance of ELISA versus VNT despite the agreement found between the two types of tests.
Kits B and C used as antigen the gB glycoprotein which is strongly immunogenic and induces a humoral response that appears in an early stage of infection . This response persists two to three years after infection in cattle . Because of the persistence of anti-gB antibodies, as well as the fact that the gB antigen is genetically conserved between alphaherpesviruses of ruminants, gB can be regarded as an ideal antigen for serology in wild animals for which the time of infection is unknown and no validated serological tests are commercially available.