Analysis of the diagnostic accuracy of the gamma interferon assay for detection of bovine tuberculosis in U.S. herds

The goal of this study was to evaluate the test sensitivity (SE) and specificity (SP) of the gamma interferon (G-IFN) assay used for the detection of bovine tuberculosis (bTB) in U.S. cattle herds. In addition, the study assessed the association between G-IFN test results and bTB status of cattle, and explored different cut off values for classification of test results in adult cattle using receiver operating characteristics (ROC) curve analysis. Test SE was estimated using a population of 87 confirmed infected cattle from 14 herds distributed in 6 states. Test SP was estimated using a population of 4123 cattle representing 3000 premises in 3 states. These animals were from bTB free areas, accredited bTB free herds, or herds that were historically bTB free based on the absence of lesions found at slaughter and historical records of negative tests performed for bTB surveillance. The distribution of G-IFN results and its association with bTB infection status was also explored in a group of 914 exposed cattle in which infection was not confirmed. The results showed that the SE of the G-IFN for a cut-off value ≥0.1 was 83.9% (76.1, 91.6). The SP of the G-IFN was 90.7% (95% CI: 89.8, 91.6), 97% (95% CI: 96.5, 97.5), and 98.6%(95% CI: 98.2, 98.9), for cut off values of 0.1, 0.3, and 0.5, respectively. For a cut off value ≥0.1, the likelihood ratio of a positive G-IFN test was 9.03 (95% CI: 7.90, 10.31), and the likelihood ratio of a negative G-IFN test was 0.18 (95% CI: 0.11, 0.29). The area under the ROC curve was 0.976 (95% CI: 0.97, 0.98), characteristic of a highly accurate test. ROC analysis also showed that lower cut-off values, such as 0.1, have high SE with suitable SP for use in parallel testing, while cut-off values ranging between 0.3 and 0.6 provide the high SP desired in series-testing protocols with lower SE values. Findings from this study indicated that the G-IFN performs with high accuracy in the field, yielding SE and SP estimates comparable to those reported in previous evaluations ( Ryan et al., 2000; Ameni et al., 2000; de la Rua-Domenech et al., 2006; Gormley et al., 2006).