Ganoderma species discrimination by dual-mode chromatographic fingerprinting: a study on stationary phase effects in hydrophilic interaction chromatography and reduction of sample misclassification rate by additional use of reversed-phase chromatography.

Acetonitrile-water extracts of several Ganoderma species - a mushroom being used in Traditional Chinese Medicine - were analysed by liquid chromatography-UV detection in hydrophilic interaction chromatography (HILIC) and reversed-phase (RP) elution modes. A set of six polar stationary phases was used for HILIC runs. These columns had remarkably different separation properties under binary gradient conditions as evinced by hierarchical cluster analysis on retention patterns of seven test compounds. Complementary measurements of RP chromatograms were carried out on a C(18) packing. Injection precision (n=5) and intra-day precision (n=5) were each <2.0% RSD (HILIC) and <0.7% RSD (RP) for relative retention times of main characteristic peaks of a sample extract while for relative peak areas RSD values were max. 6.8%. Repetitive analysis (n=7) of a processed sample stored in the autosampler tray for 48h was used to confirm within-sequence sample stability. Eleven Ganoderma lucidum samples served as training set for the construction of column-specific simulated mean chromatograms. Validation with twelve samples comprising G. lucidum, Ganoderma sinense, Ganoderma atrum, and Ganoderma tsugae by correlation coefficient based similarity evaluation of peak patterns showed that a discrimination of G. lucidum from other Ganoderma species by means of chromatographic fingerprints is conceptually possible on all columns, except of a bare silica packing. The importance of the combined use of RP and HILIC fingerprints to improve the rate of correct sample classification was demonstrated by the fact that each one G. sinense specimen was wrongly assigned being G. lucidum by all HILIC fingerprints but not the RP fingerprint and vice versa. The present data revealed that (i) the analysis of complex biological materials by quasi orthogonal chromatographic modes such as HILIC and RP may deliver more discriminative information than single-mode approaches which strengthens the reliability of fingerprint-based sample classification and (ii) different retention and selectivity characteristics of polar bonded silica packings in the HILIC elution mode may only have a minor impact on chemometric sample discrimination capabilities in such kind of pattern-oriented metabolomics separation problems.