Development of innovative analytical strategies for the identification of New Psychoactive Substances

New Psychoactive Substances (NPSs) are not regulated under the Single Convention on Narcotic Drugs (1961) or the Convention on Psychotropic Substances (1971), yet they pose a serious threat to public health, particularly due to the unknown purity and composition of many such substances. The detection and identification of NPSs present significant challenges in clinical, toxicological and forensic fields, due to their rapid evolution and structural similarities [1][2], which make them often undetectable through usual analytical methods. This scenario highlights the urgent need for new analytical strategies to identify these compounds, even though it is complicated by the absence of available analytical standards. In this context, the present study, conducted in collaboration with the Italian Scientific Police Service, aimed to develop an advanced UHPLC-HRMS workflow for the targeted and untargeted determination of emerging substances of abuse in law enforcement seizures. UHPLC-HRMS has been employed for targeted analyses in both Full Scan mode and MS/MS to detect 176 standards analytes from common illicit substance classes, including synthetic cannabinoids, synthetic opioids, stimulants, dissociatives, hallucinogens, sedatives/hypnotics. The data obtained from these analyses were utilized to develop a chemometric model, leveraging the fragmentation spectra and neutral losses observed across a range of collision energies, tailored to the chemical properties of the analyzed molecules. The application of an initial exploratory method, such as PCA, revealed the presence of sample clusters differentiated based on their structural chemical characteristics, more or less influenced by variables such as retention time, mass fragments, and the corresponding neutral losses, considered above a relative intensity threshold of 10%. This allowed us to apply a linear classification method (SIMCA) with the aim of predicting the class membership of unknown NPSs, which are structurally related to existing ones, but potentially also new. The classification model was validated both on a test set (internal validation) built from available standards and on analytes found in seizures (external validation).