Mass spectral differentiation of positional isomers using multivariate statistics

Abstract

Mass spectral analysis is considered to be a confirmatory method of analysis for the purpose of identifying controlled substances. However, the spectra of positional isomers are generally too visually similar to allow for differentiation. This study of fluoromethcathinone (FMC) and fluorofentanyl has shown that multivariate statistical analysis offers a feasible means of differentiating between electron ionization mass spectra of positional isomers. Three positional isomers of each compound (2-FMC, 3-FMC, 4-FMC, meta-fluorofentanyl, ortho-fluorofentanyl, and para-fluorofentanyl) were analyzed twice daily using gas chromatography/electron ionization mass spectrometry (GC/EI-MS) on six separate instruments over five days. This resulted in 60 mass spectra collected for each positional isomer. Principal component analysis (PCA) followed by linear discriminant analysis (LDA) was performed and successful differentiation of positional isomers was achieved. Leave-One-Sample-Out Cross Validation showed no errors for either group of isomers. Nineteen blind study samples were analyzed for each group of positional isomers with no misclassifications. Furthermore, data from previous case samples was analyzed using this method and in all cases the samples were properly attributed to the correct positional isomer. In addition to visual inspection of the LDA plots, objective classifications were conducted using the resulting posterior probabilities generated when LDA was performed. The results indicate multivariate statistical analysis is a promising addition to the analytical scheme of the identification of positional isomers. This would allow for higher confidence in the final identification of a compound without the need for additional instrumental analysis, saving laboratories both time and money.