A microRNA-based Johne’s disease diagnostic predictive system: preliminary results

Background Johne’s disease, caused by Mycobacterium avium subsp. paratuberculosis (MAP), is a chronic enteritis impacting welfare and productivity in cattle. Screening and animal removal are common for disease management, but efforts are hindered by low diagnostic sensitivity. Expression levels of small non-coding RNA molecules involved in gene regulation (microRNAs) altered during mycobacterial infection may present an alternative diagnostic method. Methods Levels of 24 microRNAs affected by mycobacterial infection were measured in sera from MAP-positive (n=66) and MAP-negative samples (n=65). They were used to train a collection of statistical and machine learning models to identify an optimal classifier for diagnosis. Results The best-performing model provided 72% accuracy, 78% AUC, 73% sensitivity and 71% specificity on average. Limitations Although control samples were collected from farms nominally MAP-free, low sensitivity in current diagnostics means animals may be misclassified. Conclusion MicroRNA profiling combined with advanced predictive modelling techniques accurately diagnosed Johne’s disease in cattle.