A bayesian analysis of phage-based qPCR and liquid culture for the early detection of Mycobacterium avium subsp. paratuberculosis in young dairy calves

Abstract

Bovine paratuberculosis (PTB), caused by Mycobacterium avium subsp. paratuberculosis (MAP), is an endemic infectious disease causing significant economic losses, with infection typically occurring in the first months of life. However, early diagnostic in subclinical infected calves is challenging due to the paucibacillary nature and intermittent shedding of the pathogen, which limits the sensitivity (Se) of existing diagnostic tools. This study aimed to estimate and compare the diagnostic Se and specificity (Sp) of a Phage-Magnetic Separation (PhMS) qPCR assay against a qPCR-confirmed liquid culture for the early detection of MAP infection in young dairy calves. Using a Bayesian Latent Class Model (BLCM) that accounted for conditional dependence, we analyzed 528 fecal samples from dairy calves < 60 days of age from 39 commercial herds in Chile. The model revealed that both tests have comparable moderate median Se (PhMS = 45 %; culture = 51 %) and high median Sp (PhMS = 98 %; culture = 100 %) for MAP detection. However, tests were strongly negatively correlated in MAP-infected animals (correlation coefficient, ρDPhMS,Cult = -0.66; 95 % Posterior probability interval [PPI]: -0.91 to -0.12). This strong negative correlation indicates that the tests tend to detect different subsets of the infected population, making them complementary when used in a parallel testing strategy to maximize joint sensitivity (Sej). The estimated mean true within-herd prevalence in calves was 13 % (95 % PPI: 8-22 %). Additionally, the Sp of PhMS-qPCR decreased in high-prevalence herds, possibly due to passive shedding of the pathogen. Critically, leveraging their complementary nature, a parallel interpretation of the tests achieved a high combined Se of 92 % (95 % PPI: 53-100 %), while maintaining a high Sp of 97 %. In conclusion, PhMS-qPCR and qPCR-confirmed liquid culture are complementary diagnostic tools. Their combined use in a parallel testing strategy could provide a highly effective screening algorithm, making it a powerful approach for research settings aiming to maximize case detection.

Keywords: Bovine; Conditional dependence; Fecal shedding; Latent class model; Parallel testing; Paratuberculosis; Viable but non-culturable.