Preliminary investigation of a humoral and cell-mediated immunity ratio for diagnosis of paratuberculosis in beef cattle.

One thousand three hundred and twenty-four adult beef cattle were tested for paratuberculosis using 2 antibody enzyme-linked immunosorbent assays (ELISA), an interferon-gamma (INF-gamma) ELISA, and radiometric bacterial culture of feces from 5 populations. Two populations of cattle (n=226) had data available to calculate a ratio of humoral to cell-mediated immunity based on results from one antibody test and the INF-gamma ELISA. Latent class analysis was used to estimate accuracy of the 4 paratuberculosis assays within a Bayesian framework. Determination of test accuracy and paratuberculosis prevalence in the latent class analysis allowed for estimation of predictive value positive (PVP) functions. The estimated PVP functions were used to iteratively assign paratuberculosis status to sampled cattle. Accuracy of the immunity ratio, an antibody ELISA, and the INF-gamma ELISA were determined for multiple cutoffs based on probabilistically assigned paratuberculosis status. Area under the receiver-operating characteristic (ROC) curves (95% probability interval) were estimated as 0.78 (0.66, 0.89), 0.81 (0.68, 0.92), and 0.59 (0.47, 0.71) for the immunity ratio, antibody ELISA, and INF-gamma ELISA, respectively. The Youden index (sensitivity+specificity-1) peaked at immunity ratios of 0.5 (J=0.48) and 1.0 (J=0.46). Sensitivity and specificity (95% probability interval) at an immunity ratio cutoff of 0.5 were 0.65 (0.44, 0.85) and 0.83 (0.78, 0.88), respectively. Sensitivity and specificity (95% probability interval) at the 1.0 cutoff were 0.55 (0.33, 0.77) and 0.91 (0.87, 0.95), respectively. An immunity ratio could be used to diagnosis paratuberculosis in beef cattle but requires further investigation.