Ensuring food safety requires robust screening approaches capable of detecting the administration of growth-promoting agents in livestock. While previous metabolomics studies have demonstrated proof-of-concept, most were limited to a single class of compounds and offered only partial metabolome coverage due to the use of individual analytical platforms. The present study aimed to develop a comprehensive classification model through a more global exploration of the metabolome. This integrative workflow represents a major step towards implementing effect-based metabolomics screening within official food control frameworks. A total of 502 urine samples from six experiments (n = 59 cattle) involving β-agonists, steroids, and SARMs were analysed using four complementary LC-HRMS platforms. This multi-platform approach expanded metabolome coverage by capturing a wide range of molecular fingerprints. Multiblock Consensus-OPLS analysis was applied to integrate the four datasets into a joint modelling strategy, thereby enhancing data interpretation. This strategy provided an innovative framework that increased the predictive power of the classification model and underscored the complementarity of the LC-HRMS techniques. Overall, this comprehensive workflow enabled the efficient classification of samples from animals subjected to multiple anabolic treatments within a single analysis. Such advances not only strengthen detection capabilities but also offer a versatile tool to address key public health concerns.
Keywords: Multiblock discriminant analysis; anabolic compounds; biomarkers of effects; classification model; doping practices; metabolomics.