Integrated GC-EI-HRMS and LC-ESI-HRMS workflow along with molecular networking for untargeted identification of new psychoactive substances.

Untargeted toxicological screening (UTS) remains challenging due to the structural diversity of xenobiotics and current limitations of analytical platforms. On one hand, LC-HRMS workflows are considered the gold standard in UTS despite the lack of comprehensive spectral databases of xenobiotics, especially with new psychoactive substances (NPS). On the other hand, GC-MS benefits from extensive and highly reproducible EI spectral libraries, but its application to untargeted workflows remains limited in complex biological matrices, as deconvolution must be performed on unit-mass data, which restricts selectivity thus impacting sensitivity. Furthermore, in both platforms, dedicated tools for MS data organization are needed to facilitate the identification step. To overcome these limitations, we implemented and assessed an integrative workflow combining GC-EI-HRMS and LC-ESI-HRMS with molecular networks (MN). As a first step, we leveraged high resolution mass measurements to increase the selectivity of deconvolution allowing to provide accurate-mass EI spectra that can be directly matched to conventional libraries. As a second step, as LC-ESI-HRMS allows access to precursor ions and MS/MS spectra under soft ionization, such an analysis allows to confirm proposed structures and enables phase I and II metabolite annotation. In addition, the annotations proposed through the UTS performed using GC-EI-HRMS may thus be regarded as anchor strategy for MN built using LC-ESI-HRMS data. We applied this combined approach to two poisoning cases involving arylcyclohexylamines and cathinones and showed that MN built from GC-HRMS data can refine ambiguous spectral matches, improve annotation consistency, and support the identification of previously unreported metabolites. In conclusion, this dual-platform strategy addresses several limitations of LC-HRMS-only workflows and highlights the value of integrating GC- and LC-based MN to achieve more confident xenobiotic identification, opening new perspectives for broad exposomic applications.