LAUSR

Predictive models have been developed for the early identification of novel anabolic androgenic steroids and to obtain information on their molecular structure. To this purpose, gas-chromatographic and mass spectrometric characteristic parameters of 136 anabolic androgenic steroids have been specifically considered. Starting from Principal Component Analysis, different chemometric methods were applied, such as classification and clustering techniques, outlining a spectral and structural characterization for each steroid subclass, and considering the contribution of more than 30 variables. Mass spectrometric data on the TMS-derivatives of the target steroids were obtained by gas chromatography coupled to quadrupole-time of flight mass spectrometry using electron ionization. Steroids included in the training set were grouped in 5 subclasses according to their structural similarity, and the experimental data, processed by the chemometric models, allowed the identification of class-specific common fragments and spectral trends. The results of this study, validated on a test set of 21 steroids, have confirmed that the proposed approach allows tracing novel "designer anabolic steroids", including those previously unknown new structures that may have been designed and illicitly synthesized to be invisible to the current anti-doping tests.