A Robust Chemometrics Approach Towards Trace Biometal And Associated Speciation Analysis In Soft Tissue Utilizing Energy Dispersive X-Ray Fluorescence And Scattering Spectrometry

Abstract

The goal of this study was to utilize a robust multivariate chemometrics approach towards direct, rapid and accurate quantitative determination of Na, Mg, Zn, Fe, Cu, Mn and Co and speciation of Cu, Mn and Fe in complex light element matrix materials in this case soft body tissues. This study has relevance in disease diagnostics in body tissues utilizing trace biometals as the disease biomarkers. The technique in use is Energy Dispersive X-ray Fluorescence and Scattering (EDXRFS) spectrometry. Direct, rapid and simultaneous determination of trace biometal concentrations and their speciation in human body tissues utilizing Energy Dispersive X-ray fluorescence and Scattering (EDXRFS) spectrometry is challenging. This is because the spectra are characterized by analyte peak overlaps, weak fluorescence peak signals and extreme matrix effects due to the predominance of low-Z elements (dark matrix), poor signal-to-noise ratio (SNR) of the analyte peaks, and imprecise sample geometry. Moreover, there is not yet available a direct method for speciation analysis in EDXRF spectroscopy. The utility of chemometrics in EDXRF analysis is however still under development. High-noise and low-resolution EDXRF spectra are amenable to multivariate chemometric methods for qualitative and quantitative analysis. Simple univariate, classical linear regression and multivariate regression methods are restricted to linear relationships and are therefore inapplicable. Samples of soft body tissues prepared as thin (5μm), intermediate thick (10μm) and thick (20 μm) were analyzed. Robust chemometrics methods namely combined use of wavelet transform (WT), principal component analysis (PCA), independent component analysis (ICA) for spectral processing and artificial neural network (ANN) and partial least squares (PLS) for multivariate calibration based on the use of paraffin wax ‘standards’ spiked with Fe, Cu, Mn, Zn, Na, Mg and Co were utilized. WT was used in de-noising and resolution enhancement of the spectra to optimize them for the determination of the elemental concentration of Na, Mg, Mn, Fe, Cu, Co and speciation of Fe, Mn, Cu. When WT and PCA were combined, there was improved signal–to–noise ratio (SNR) of analyte as well as scatter peaks. ICA was used for both pattern recognition (classification) of tissues into those with lower and higher speciation of Fe, Mn, Cu and for spectral preprocessing and/or dimension reduction when combined with WT to optimize the spectra for vi qualitative analysis. PCA was used for reduction of spectral data dimension and pattern recognition. The preprocessed spectra were used as an input to artificial neural networks (ANN) and partial least squares (PLS) models for development of calibration strategy for direct quantitative analysis using fluorescence spectral signature regions and Compton scatter peaks. Both ANN and PLS calibrations gave results for trace element concentration better than when raw spectra was utilized i.e. (R2 values for the elements Fe, Mg, Mn, Na, Co, Zn and Cu were; R2 ~ 0.889 – 0.951 before and ~ 0.989 - 0.997 after preprocessing for ANNs; and R2 ~ 0.876 - 0.931 before, and ~ 0.969 – 0.993 after preprocessing for PLS). The results also indicate that there was improvement in determination of low–Z elements (Na and Mg) when the preprocessed spectra of both fluorescence and scatter regions were utilized simultaneously i.e. (R2 = 0.976 for featured fluorescence and R2 = 0.994 for both featured and Compton scatter for Na; and R2 = 0.932 for featured fluorescence and R2 = 0.995 for both featured and Compton scatter for Mg utilizing ANN model). Normally Na and Mg cannot be analyzed by classical EDXRF spectroscopy. Quantitative analysis Oyster tissue by the analytical approach was in agreement with certified values of the analytes in the standard reference material (≤ 6 % or less for most elements). The results are independent of sample thickness. Quantitative analysis of dog tissues indicate that mammary and prostate cancer tissues were dominated with high concentration of Zn, Fe, Mg and Cu as compared to healthy mammary and prostate tissues. The results of speciation analysis indicate that mammary and prostate cancer tissues were rich in high oxidation state (Cu2+, Fe3+, Mn7+). The analytical approach reported here is novel for direct rapid analysis of concentration levels and speciation alterations of selected trace elements (Fe, Cu, and Mn) in the context of cancer characterization.