Why is Enferplex considered to be advantageous?
- Need for optimal specificity when screening for disease at the herd level (Surveillance Testing)
As with all herd screening for disease (as opposed to individual animal testing), the need to minimise the risk of false positive diagnoses is essential, especially given the potential consequences of a diagnosis of bovine tuberculosis (bTB).
As the Enferplex test uses separate spots for each antigen, as opposed to a mixed combination as with other commercially available serological tests for bTB, the number of spots that are considered to have reacted can be quantified. This allows the test to be read at different levels of sensitivity (measure of the probability that if the animal tested has the disease, the test will be positive) and specificity (measure of probability that if the animal does not have the disease, the test will be negative). The principle being that the higher the number of individual spots that are required to be positive before an animal is diagnosed as positive to bTB, the lower the sensitivity of the test but the higher the specificity will become.
The BAS/BLS/BCL/VLA study1 (hereafter referred to as the BAS trial for simplification only) showed that using the four-antigen cut-off, the Enferplex test achieved, when used in isolation, close to 100% specificity i.e. an extremely low risk of a false positive diagnosis (see table two).
For routine surveillance, then use of the two-antigen cut-off will also be used. This will reduce the specificity to an estimated 97%1. However, the increased risk of false positives this will generate will be accounted for within a statistical package that Synergy Farm Health has developed and been approved by DEFRA.
Based on the number of animals tested, this tool will determine a threshold below which any animals that react at the two antigen level will be initially treated as potential false positives and, as such, DEFRA will NOT be notified of the suspicion of the presence of bTB in the herd. However, this policy is based on the understanding that these animals may be true positives and as such they should be re-tested within 30-60 days of the initial test. Failure to do so will obligate SureFarm to notify DEFRA of the existence of these individual animals.
Table 1From the samples in the BAS trial, the sensitivity and 95% confidence interval for the Enferplex test using the 4-antigen rule for high specificity have been calculated. Martin Vordermeier (AHVLA) has used the ROC curves to work out the sensitivity and confidence interval for the IDEXX and DPP tests at high specificity. The results are as follow:
|Sensitivity (%)||95% Cl||Specificity (%)||95 % Cl|
|Idexx||46.20||32.2 - 60.5||99.70||98.2 - 99.99|
|DPP||45.65||30.9 - 61.0||99.67||98.2 - 99.99|
|Enfer||55.1||41.1 - 69.2||99.99*||99.5 - 100.3|
*The specificity for the Enferplex test was set at 99.99% rather than 100% in order to make the CI calculation possible.
Cross Reactivity with M. Tb complex species
While the above clearly demonstrates the high specificity of the test though there are a level of expected differences in terms of antigen reaction the test cannot determine accurately the difference between M bovis and M microti.
- Need for optimal sensitivity in confirmed cases
In the event of disease being confirmed, then the need to optimise sensitivity while maximising specificity is required. The BAS trial suggested that there was no statistical difference between the other available serological tests. However, Enfer Diagnostics and MV Diagnostics (Enfer diagnostic’s partner in the United Kingdom) contend that the study had methodological flaws (though not deliberate) that down-graded the potential of the Enferplex test i.e. it was reasonable to postulate that the Enferplex test may have demonstrated better sensitivity under different study designs. These contentions were based on: –
- Inherent bias due to pre-selection of samples by the Statpak test prevented Enferplex showing what it can do (because the results all converged on the Statpak antigens which are common to all the tests). Data from cattle studies illustrated in table 2 below demonstrate that this theoretical concern is seen in practice in cattle studies.
- Only 48 test-positives were used in the study, which is too low a number to allow maximum discrimination between tests.
- Samples from two herds (known to us as Herd 15 and Herd 44) were selected by APHA for inclusion in the trial because the animals had visible lesions. The rest of the animals from these 2 herds, which were in-contact with VL the animals and would therefore be expected to be exposed to M. bovis challenge, were not included in the trial. However, all 59 samples from these herds have been tested for antibodies by both Enferplex and Stat Pak test. The Stat Pak test gave 23 as positive (39.0%), and the Enferplex test gave a positive result in 29 (49.2%). Of the 23 Stat Pak positive animals 21 were positive by Enferplex (relative sensitivity of Enferplex to Stat Pak of 93.1%) i.e. the Enferplex test missed 2 Stat Pak positive samples. However, of the 36 samples negative by the Stat Pak test, 10 were positive in the Enferplex test giving a relative sensitivity of Stat Pak to Enferplex of 72.4% i.e. Stat Pak missed 10 Enferplex test positive samples. This suggested, albeit from small numbers of animals, that Enferplex is more sensitive than Stat Pak. (Note that the Stat Pak test missed 5 VLs while the Enferplex test only missed 2 VLs). These findings may not have been apparent from the BAS trial due to the pre-selection of a large proportion of the samples using the Stat Pak test.
On a practical level, the fact that the test can switch from a “standard – four antigen” interpretation to a “severe-two antigen” interpretation without the need to re-sample animals was also seen as a significant advantage.
Table 2Cattle data showing Enferplex positives to antigens not present in the StatPak test
|Total number of samples from infected cattle||Enferplex positive to Statpak antigens||Enferplex positive to unique antigens*||Reference|
ST+, Lesion or culture+
|0/88||54/88 (61%)||Data from Whelan et al 2008|
ST+, Lesion +
|0/13||4/13 (31%)||unpublished data|
ST+ Lesion +
|0/22||22/22 (100%)||Unpublished data|
ST+ Lesion +, Culture +
|0/6||5/6 (83%)||Data from Whelan et al 2010|
* Positive by the 2 Antigen rule- The results show that significant numbers of samples do not contain antibodies to the antigens present in the Statpak test (i.e. ESAT-6, CFP10, MPB83), but contain antibodies to others antigens present in the Enferplex test
The Camelid societies recognise the development and understanding of serological testing for bTB in camelids is still in its infancy. There is a lot more to learn. As the Enferplex is based on quantitative, rather than qualitative outputs, and relies on seven individual antigens that are monitored separately, the potential to refine the test, to both maximise sensitivity and specificity, is far higher than with the other tests. For instance it should be possible to develop the test to allow:-
- Inclusion of additional antigens to increase sensitivity.
- Inclusion of antigens that potentially allow discrimination between M. bovis and M. microti infections.
- Improved understanding of the optimal cut-offs for the currently used antigens to improve both sensitivity and specificity (would rely on good information exchange with APHA).