Multi-metal element analysis for the identification of foodborne pathogenic bacteria

Abstract

Mineral element contents, combined with multivariate analysis, were used for the identification and classification of foodborne pathogens from a common genus (Rhodococcus equi, Staphylococcus spp., Listeria spp., Salmonella spp., Shigella spp., Escherichia coli, Enterobacter sakazakii, Yersinia enterocolitica and Vibrio spp.). 45 macro- and trace mineral elements of 30 foodborne pathogens were determined by a semiquantitative inductively coupled plasma mass spectrometry (SQ-ICP-MS) technique. The elemental analysis identified 10 significant elements (²⁸Si, ⁴³Ca, ⁵⁷Fe, ⁴⁷Ti, ⁵²Cr, ⁵⁵Mn, ⁶⁶Zn, ⁸⁸Sr, ¹³⁷Ba and ²⁰⁸Pb) by ANOVA in different types of pathogens. Principal component analysis (PCA) reduced the 10 variables to 6 principal components which could explain 98.40% of the total variance. The classification models constructed by the Fisher linear discriminant analysis (Fisher LDA) and back-propagation artificial neural network (BP-ANN) achieved correctly classified rates of 86.9% and 91.3%, respectively. The results indicated that the combination of multi-metal element composition determination and multivariate analysis can be used as fingerprint to quickly identify and classify foodborne pathogens.