Abstract

Background: The objective of this study was to develop a robust quantification method for simultaneously analyzing molecules in human plasma using the Fourier transform infrared (FT-IR) system with a partial least square (PLS) regression.

Methods: Plasma spectra were analyzed from 4000 to 500 cm⁻¹ (with 2.0 cm⁻¹ of resolution and 32 scans), and the molecule concentrations (IgA, IgG, IgM) were measured blindly by using a cross-validation model prepared by PLS analysis of data from 135 samples.

Results: There was a significant correlation between the FT-IR predicted concentration and the concentration obtained with the clinical reference method: R²=0.98 (IgA), R²=0.98 (IgG), and R²=0.97 (IgM). The root mean square error of prediction (RMSEP) was 0.05 g·L⁻¹ (IgA), 0.4 g·L⁻¹ (IgG), and 0.03 g·L⁻¹ (IgM). Variability of inter-experimenter reproducibility was less than 2%. The interchangeability of the two methods was studied by using the Bland-Altman method.

Conclusions: Together with PLS analysis, FT-IR spectrometry appears to be an easy-to-use and accurate method to determine multianalyte concentrations in dried human plasma. It could be an alternative tool for rapidly quantifying many molecules after developing a specific predictive model.

Clin Chem Lab Med 2009;47:83–90.