Predicting calcium status in early lactation multiparous Holsteins using milk constituent analysis
J. A. Seminara1, K. R. Callero1, C. R. Seely1, M. Van Althuis3, S. An3, C. M. Salpekar3, D. M. Barbano2, and J. A. A. McArt1
1 Department of Population Medicine and Diagnostic Sciences, College of Veterinary Medicine
2 Department of Food Science, College of Agriculture and Life Sciences
3 College of Agriculture and Life Sciences
Cornell University, Ithaca, NY 14853
Subclinical hypocalcemia is common among early lactation dairy cows. Cows that have not recovered from this condition by 4 DIM develop dyscalcemia, a metabolic state associated with negative health and production outcomes. Despite relatively high prevalence, identifying cows with dyscalcemia remains a challenge on dairies. Our objective was to develop an algorithm for predicting dyscalcemia status using Fourier-transform infrared spectroscopic (FTIR) analysis of milk samples at 3 and 4 DIM, through a prospective cohort study. We collected blood from multiparous Holsteins (n = 453) on 4 herds in NY at 4 DIM for total calcium concentration (tCa) analysis and classified cows as eucalcemic (tCa ≥ 2.2 mmol/L; n = 353) or dyscalcemic (tCa < 2.2 mmol/L; n = 100). Proportional milk samples were collected at 3 and/or 4 DIM and analyzed via FTIR for the following constituents: lactose, protein, fat, de novo fatty acids (FA), mixed origin FA, and preformed FA; relative percentages of each FA group; individual FA including C16:0, C18:0, and C18:1 cis:9; milk urea nitrogen, milk acetone, milk β-hydroxybutyrate, and milk predicted blood non-esterified FA. Milk weights were also collected at each milking. Partial least squares regression models were fit for each DIM separately and were used to predict calcium status at 4 DIM using individual daily estimated milk constituents, daily milk yield and parity group (2, 3, and ≥4). Receiver operating characteristic curves were generated using predictions from each model. Area under the curve was 0.76 and 0.81 for 3 and 4 DIM respectively. Sensitivity, specificity, positive and negative predictive values for the models were 0.57, 0.84, 0.51, and 0.88 respectively at 3 DIM, and 0.62, 0.86, 0.55, and 0.89 respectively at 4 DIM. Our results indicate that FTIR milk constituent analysis at 3 and 4 DIM can accurately identify cows that are eucalcemic at 4 DIM and therefore may be useful as a tool to optimize the management of early lactation cows to improve health and production outcomes. Key words: dyscalcemia, milk Fourier-transform infrared spectroscopy, subclinical hypocalcemia