Abstract
Vitamin A levels in fattening Japanese Black cattle affect meat quality; therefore, it is important to monitor serum retinol concentrations. To simplify and accelerate the evaluation of serum retinol concentrations in cattle, we developed a new predictive method using excitation-emission matrix (EEM) fluorescence spectrophotometry. For analytical comparison, the concentration of serum retinol was also measured using the conventional HPLC method. We examined excitation (Ex) and emission (Em) wavelengths of cattle serum, which were 250–450 and 250–600 nm, respectively. Parallel factor analysis separated four components from EEM data, one of which was related to retinol. Next, a partial least square regression model was created using the obtained EEMs as explanatory variables and accrual measurement values as objective variables. The determination coefficient value ( R 2 ), root mean squared error of prediction (RMSEP), and the ratio of performance to deviation (RPD) of the model were determined. A comparison with reference values found that R 2 , RMSEP, and RPD of the calibration model were 0.95, 6.4 IU/dl, and 4.2, respectively. This implies that EEM can estimate the serum retinol concentration with high accuracy. Additionally, the fluorescent peaks that contributed to the calibration, which were extracted from the regression coefficient and variable importance in projection plots, were Ex/Em = 320/390 and 330/520 nm. Thus, we assume that this method observes not only free retinol, but also retinol-binding protein. In conclusion, multidimensional fluorescence analysis can accurately and quickly determine serum retinol concentrations in fattening cattle.