Proton nuclear magnetic resonance spectroscopy of plasma lipoprotein: technical problems and potential interest in cancer disease

Abstract

This paper reviews several methods presently available for analysing lipoprotein NMR spectra. Two main steps can be distinguished: NMR signal processing and data analysis. Time domain (wavelet transform) and frequency domain (curve fitting) signal processing methods are compared. Statistical methods of data analysis (Ascending Hierarchical Classification, Correspondence Analysis and Principal Component Analysis) have been tested on simulated NMR data of plasma lipoprotein with different numbers of sampling points and different noise levels. These few examples clearly attest that the NMR approach to complex "mixture" (such as body fluids) analysis is emerging from its infancy. New interest in plasma lipoprotein analysis in cancer biology is finally discussed in the light of previous clinical and experimental results and of understanding of lipid metabolism in cancer.