Modelling mammary metabolism in the dairy cow to predict milk constituent yield, with emphasis on amino acid metabolism and milk protein production: model evaluation.

A model of mammary metabolism has been constructed and parameterized, with milk protein synthesis represented as a function of five essential amino acids (EAA) (Hanigan et al., 2001). Herein the model is evaluated using both the data used to construct the model (reference data) and an independent data set (literature data), and sensitivity to inputs and parameter estimates is assessed. The model predicted metabolite removal well for the reference data with exceptions for glutamate, glucose, and acetate. However, predictions of milk protein synthesis exhibited significant mean positive bias, which apparently was associated with the representation of milk protein synthesis. Adjustment of model parameters removed the mean bias, however, prediction accuracy was still inadequate. Simulation of the single reference experiment containing all critical inputs resulted in predictions of milk protein output that explained 53% of the observed variation, suggesting that the limited accuracy of the model when applied to the entire reference data set was due to assumptions regarding missing inputs. Mammary removal of glutamate, isoleucine, lysine, phenylalanine, tyrosine, valine, glycerol, beta -hydroxybutyrate (BHBA), and acetate were predicted less accurately when simulations of the independent data set were conducted. Twenty-five percent of the observed variation in milk protein yields for the independent data set was explained by the model. Refitting parameters for removal of isoleucine, lysine, phenylalanine, tyrosine, valine, glycerol, BHBA, and acetate raised the variation explained to 43%. Sensitivity analysis indicated that milk protein synthesis was responsive to only the five EAA used in its determination, with sensitivity to any single EAA falling to zero as supply of the EAA exceeded protein synthetic needs. Similarly, milk protein synthesis was readily affected by parameters associated with removal and metabolism of the five EAA. Milk lactose was found to be sensitive to glucose input as well as to similar parameters and inputs as milk protein. It is concluded that representation of the milk protein synthesis process as a function of a single limiting EAA may not be adequate and might be better represented by simultaneous consideration of multiple EAA. Additional work on the description of energy metabolism is also suggested.