PSII-1 Blood metabolomics for biomarker discovery and the diagnosis of Bovine Respiratory Disease
Abstract Bovine Respiratory Disease (BRD) is the leading cause of morbidity and mortality in Australian feedlot cattle. Diagnosis for BRD is based on visual scoring of illness and the use of rectal temperature above a defined level to trigger treatment protocols. These methods often have a low accuracy at diagnosing BRD. Blood metabolomics monitors alterations in small metabolites in the body and can be used to indicate the presence of disease. The aim of the current study was to search for biomarkers for BRD and develop alternate diagnosis methods for BRD using the blood metabolome profile of feedlot steers. Visually BRD affected (n = 148) and visually healthy (n = 152) steers were removed from their group pens for clinical assessment and blood sampling for metabolomics analysis. Lung lesions indicative of BRD were scored for all trial animals upon slaughter. A non-targeted metabolomics approach based on nuclear magnetic resonance (NMR) spectrometry was used to search for blood biomarkers using classification and regression trees. The data were split into training and validation datasets for model development. Visual diagnosis (VD), visual + clinical diagnosis (VCD; visually sick and elevated rectal temperature or lung auscultation score), and lung lesion diagnosis (LLD; lung consolidation ≥ 10% or pleurisy score ≥ 2) were used as reference diagnosis methods for BRD. Metabolomics demonstrated a high accuracy at detecting BRD in the validation dataset when using the VD (Acc=0.85, Se=0.82, SP = 0.87) and VCD (Acc=0.81, Se=0.88, SP = 0.74), but was less accurate at detecting animals defined as sick using the LLD (Acc=0.74, Se=0.38, SP = 0.89) (Table 1). The models selected nine metabolites important in differentiating sick and healthy animals. The results suggest the blood metabolome is a useful indicator of BRD status and could therefore be used for confirmation of BRD in feedlot cattle.