Abstract
The detection and quantification of xenobiotics and their metabolites in man is important for drug dosing, therapy and for substance abuse monitoring where longer-lived metabolic products from illicit materials can be assayed after the drug of abuse has been cleared from the system. Raman spectroscopy offers unique specificity for molecular characterization and this usually weak signal can be significantly enhanced using surface enhanced Raman scattering (SERS). We report here the novel development of SERS with chemometrics for the simultaneous analysis of the drug nicotine and its major xenometabolites cotinine and trans-3'-hydroxycotinine. Initial experiments optimized the SERS conditions and we found that when these three determinands were analysed individually that the maximum SERS signals were found at three different pH. These were pH 3 for nicotine and pH 10 and 11 for cotinine and trans-3'-hydroxycotinine, respectively. Tertiary mixtures containing nicotine, cotinine and trans-3'-hydroxycotinine were generated in the concentration range 10(-7)-10(-5) M and SERS spectra were collected at all three pH values. Chemometric analysis using kernel-partial least squares (K-PLS) and artificial neural networks (ANNs) were conducted and these models were validated using bootstrap resampling. All three analytes were accurately quantified with typical root mean squared error of prediction on the test set data being 5-9%; nicotine was most accurately predicted followed by cotinine and then trans-3'-hydroxycotinine. We believe that SERS is a powerful approach for the simultaneous analysis of multiple determinands without recourse to lengthy chromatography, as demonstrated here for the xenobiotic nicotine and its two major xenometabolites.