Blood RNA test accurately detects bovine tuberculosis

Mycobacterium bovis, the zoonotic bacterium that causes bovine tuberculosis (bTB), remains difficult to eliminate in countries where current tests miss infected animals. It is closely related to Mycobacterium tuberculosis, the main cause of human tuberculosis (hTB), which has motivated work on blood-based biomarkers in people. Building on that idea, researchers led by David E. MacHugh asked whether a similar blood test could spot bTB in cattle. They used peripheral blood mRNA as the source of information and applied RNA-seq to measure gene activity across animals. To test whether host-response signals in blood could reliably distinguish infected and uninfected cattle, the team combined RNA-seq with machine learning and analyzed data collected from three countries: Ireland, the UK and the US. The aim was to find a compact and robust signature of infection that could improve or augment existing diagnostics and help identify animals that current tests fail to detect.

The study relied on RNA-seq data from peripheral blood mRNA and machine learning methods to search for classifiers that separate bTB-positive from bTB-negative cattle. From this analysis the researchers identified two powerful RNA-based classifiers: a 30-gene signature and a 273-gene elastic net classifier. Performance was quantified using area under the curve (AUC) values in both training and testing sets: the 30-gene signature achieved AUCs of 0.986 in training and 0.900 in testing, while the 273-gene elastic net classifier achieved AUCs of 0.968 in training and 0.938 in testing. Both classifiers produced high sensitivity and specificity values (≥ 0.853 for both metrics) in the testing set. The classifiers also proved robust when distinguishing bTB+ animals from those infected with other bacterial or viral pathogens, with AUC ≥ 0.819, indicating they detect patterns specific to Mycobacterium bovis infection rather than general illness.

These results show that peripheral blood mRNA, read out by RNA-seq and interpreted with machine learning, can yield accurate and robust signatures of Mycobacterium bovis infection. The two classifiers described — a compact 30-gene signature and a larger 273-gene elastic net classifier — both delivered strong accuracy and useful sensitivity and specificity in independent testing, and they separated bTB from other bacterial and viral infections. Because the analysis used data from Ireland, the UK and the US, the findings suggest the approach can work across different populations of cattle. The authors conclude that RNA-based classifiers accurately diagnose bTB and differentiate bTB from other diseases, representing a promising tool to augment current diagnostics. If translated into practical assays, such classifiers could strengthen bTB surveillance and control in endemic regions.