Optimising and evaluating the characteristics of a multiple antigen ELISA for detection of Mycobacterium bovis infection in a badger vaccine field trial

Abstract

A long-term research programme has been underway in Ireland to evaluate the usefulness of badger vaccination as part of the national bTB (bovine tuberculosis) control strategy. This culminated in a field trial which commenced in county Kilkenny in 2009 to determine the effects of badger vaccination on Mycobacterium bovis transmission in badgers under field conditions. In the present study, we sought to optimise the characteristics of a multiplex chemiluminescent assay for detection of M. bovis infection in live badgers. Our goal was to maximise specificity, and therefore statistical power, during evaluation of the badger vaccine trial data. In addition, we also aimed to explore the effects of vaccination on test characteristics. For the test optimisation, we ran a stepwise logistic regression with analytical weights on the converted Relative Light Units (RLU) obtained from testing blood samples from 215 badgers captured as part of culling operations by the national Department of Agriculture, Food and the Marine (DAFM). The optimised test was applied to two other datasets obtained from two captive badger studies (Study 1 and Study 2), and the sensitivity and specificity of the test was attained separately for vaccinated and non-vaccinated badgers. During optimisation, test sensitivity was maximised (30.77%), while retaining specificity at 99.99%. When the optimised test was then applied to the captive badger studies data, we observed that test characteristics did not vary greatly between vaccinated and non-vaccinated badgers. However, a different time lag between infection and a positive test result was observed in vaccinated and non-vaccinated badgers. We propose that the optimized multiplex immunoassay be used to analyse the vaccine trial data. In relation to the difference in the time lag observed for vaccinated and non-vaccinated badgers, we also present a strategy to enable the test to be used during trial evaluation.

Figure 1. ROC curves of the logit obtained using either the optimised combination of antigens (blue line) or MPB83 (red line).

Each is based on converted RLU values, and the green line is included for reference.

Figure 2. Scatter plot and least square means of the probability of testing positive for the control (left graph) and vaccinated (right) groups in Study 2 by time since the start of the study.

Vertical reference lines showing the day of vaccination and challenge, and a horizontal reference line of the cut-off are included.

Figure 3. GEE predictions by time since challenge for vaccinated and non-vaccinated badgers.

Two reference lines are presented: a vertical line showing the minimum number of days to test positive for infected non-vaccinated badgers and a horizontal cut-off line.