Prediction of lameness in dairy heifers using untargeted metabolomics

Objectives

This study aimed to investigate the use of untargeted metabolomics (liquid chromatography- mass spectrometry (LC-MS)) to differentiate the metabolome of lame and non-lame first lactation dairy cows and evaluate the predictive accuracy of identified metabolites. Study objectives included 1) evaluating the predictive accuracy of the metabolome from urine collected at the time of lameness and 2) evaluate the predictive accuracy of the metabolome from urine collected during the transition period. If successful, these techniques could be utilised to develop tools for the early prediction of lameness as part of preventive strategies.

Materials and methods

Urine samples were collected from two separate cohorts of dairy heifers housed at a 300-cow research dairy herd; cohort 1 were current first lactation cows and cohort 2 were heifers recruited pre-calving (3 weeks) and monitored until 70 days in milk (DIM). Mobility scores were recorded weekly with lameness defined as two consecutive weeks ≥2 [0 – 3 scale (AHDB)]. Urine samples were collected from lame and match paired by DIM non-lame heifers and snap frozen in liquid nitrogen prior to storage at -80 C. For cohort 2, samples were collected pre- and post- calving (within 3 weeks) prospectively with samples being selected once the lameness outcomes were known. Samples were analysed using untargeted LC-MS (Q-Exactive Plus mass spectrometer equipped with Dionex U3000 UHPKC system) to generated semi-quantitative metabolomic data, which were analysed using a suite of machine learning methods to identify the models with the best performance in terms of prediction accuracy. [...]