Near-infrared spectra of serum albumin and γ-globulin and determination of their concentrations in phosphate buffer solutions by partial least squares regression

Near-infrared (NIR) spectra have been measured for albumin and gamma-globulin in a powder state as well as in phosphate buffer solutions. The second derivative spectra of powder samples have been used to make assignments of NIR bands of the proteins. The second derivative spectra of albumin and gamma-globulin are significantly different from each other in the band frequencies and relative intensities. The NIR spectra in the 1300-1850 mt region of the solutions have been subjected to partial least squares (PLS) regression analysis to develop chemometrics models which predict the concentrations of the proteins. The calibration for the albumin solutions in the concentration range of 0.1-8.0 g/dl has yielded a correlation coefficient (R) of 0.9995 and a standard error of calibration (SEC) of 0.207 g/dl. For the gamma-globulin solutions in the concentration range of 0.1-6.0 g/dl, R of 0.9946 and SEC of 0.128 g/dl have been obtained. Regression coefficients (RCs) for the calibration models have been calculated for the first four factors. These RCs reflect the spectral variations in bands due to the proteins and in a water band near 1400 nm caused by the dissolution of the proteins. Moreover, the RCs have been compared with the NIR spectra of the proteins in the powder state. The positions of peaks in the RCs correspond well to those of bands in the NIR spectra of the proteins in the powder state. This suggests that the chemometrics models can pick up effectively the information about albumin and gamma-globulin, even if the models have been constructed from the spectra of the proteins in the dilute solutions. (C) 1998 Elsevier Science B.V. All rights reserved.