Effect of tuberculin skin testing on serological results against Mycobacterium avium ssp. paratuberculosis (MAP): Evidence of distinct effects in MAP-infected and noninfected cows

Johne's disease and bovine tuberculosis are diseases of economic, public health, and animal welfare importance. The single intradermal cervical comparative tuberculin (SICCT) test, which is used to determine bovine tuberculosis status as part of eradication schemes in the United Kingdom and some other countries, has been reported to interfere with the results of the widely used ELISA to detect antibodies against Mycobacterium avium ssp. paratuberculosis (MAP) in milk. Better understanding of the relationship between SICCT and MAP tests can improve management and control of Johne's disease. The aim of this study was to characterize the relationship between SICCT testing and milk ELISA performance and to assess whether the immunological response to the SICCT test is different for MAP-infected cows and noninfected cows. We used repeated MAP milk ELISA test results of a cohort of 805,561 cows in the United Kingdom between 2010 and 2018 that had milk ELISA tests within 90 d of SICCT testing to identify cows likely to be infected. We then assessed, separately, for cows deemed to be MAP-infected and noninfected, the association between MAP test results and proximity to SICCT testing by means of survival analysis and generalized additive mixed models. The results were used to quantify the effect SICCT testing may have on performance of milk ELISA tests conducted soon after SICCT testing. At high prevalence levels (20%) of MAP in the infected herd, overall accuracy of the milk ELISA is not reduced when testing occurs within 14 d from SICCT testing. Milk ELISA values of cows deemed to be infected were highest when MAP testing was closer in time to SICCT testing, suggesting the SICCT test enhances antibody response for MAP in infected cows. This corresponds to higher sensitivity of the MAP milk ELISA when testing within 30 d of the SICCT test. For cows deemed to be noninfected, the effect of previous SICCT testing was delayed compared with infected cows, with MAP milk ELISA values peaking at around 15 d post-SICCT testing. For both, MAP-infected and noninfected cows, interference from SICCT test diminished 30 d after SICCT testing, suggesting post 30 d to be the most appropriate time for evaluating the milk ELISA for MAP after SICCT testing. Our results provide strong evidence that the effect of the SICCT test on serological response against MAP is different for MAP-infected versus noninfected cows and that, as a result of this distinct effect, it is possible to impr