Lipidomic Profiling of Milk Enables Identification of Novel Biomarkers for Subclinical Ketosis in Dairy Cows

Abstract

The current clinical test for ketosis in dairy cows, which is based on measuring blood β-hydroxybutyrate (BHB) concentrations, is an invasive procedure that impacts herd welfare. Alternative biomarkers, especially non-invasive ones, are urgently needed for subclinical ketosis (SCK) diagnosis. To address this, we examined the milk lipidomic profiles of healthy and SCK cows from a research dairy farm over two consecutive years. A total of 125 polar lipid species were quantified in milk samples using a targeted liquid chromatography–mass spectrometry (LC-MS) approach. Chemometric analysis identified four potential biomarkers (PC 28:0, PC 29:0, PC 30:1 and PC 31:0), which met the thresholds for p value (<0.0004) and fold change (FC > 2). Global fatty acid (FA) profiling by gas chromatography–flame ionization detection (GC-FID) after transesterification of all lipids revealed that the levels of C8:0, C10:0, C12:0, C14:0, C15:0 and C18:1t11 were significantly suppressed in the SCK group (15–52% lower compared to the control group, p < 0.01), and C15:0 and C18:1t11 were the most promising markers for SCK in dairy cows at the FA level (43–50% reduction compared to the control group, p < 0.0001). Tandem LC-MS and FA data generated by GC demonstrated that SCK also caused significant changes in regio-isomer and double bond isomer composition of a model phospholipid molecule (PC 34:1) and a model triglyceride molecule TAG 52:2; these fine-structural alterations could also be explored for discovering novel SCK biomarkers in dairy cows. The PLS-DA predictive model built from lipidomic data showed excellent accuracy (>95%) in predicting cows with SCK. All of the promising biomarkers identified and the predictive model built in this study need further validation using larger SCK cohorts from different herds.