INTRODUCTION

Interindividual and intraindividual variability in drug response can lead to insufficient therapeutic efficacy or life-threatening adverse events (Kaddurah-Daouk et al., 2014). In this context, precision medicine aims to improve therapeutic outcomes by integrating the entire genetic and phenotypic knowledge specifically related to an individual. Pharmacogenomics and pharmacometabolomics are both major and complementary approaches to precision medicine (Beger et al., 2016).
Pharmacogenomics is the use of patient-specific information associated with the genome to study individual response to drugs, while pharmacometabolomics focuses on the metabolome (profile of low molecular weight molecules within a biological system) (Schrimpe-Rutledge et al., 2016; Pang et al., 2019; Wake et al., 2019). Metabolomics allows identification and understanding of pathways involved in drug-response variation (Kaddurah-Daouk et al., 2014). It is also an important tool in the discovery of biomarkers that can be applied to personalized medicine (Villaseñor et al., 2014; Jensen et al., 2017; Ivanisevic and Thomas, 2018; Yeung, 2018). Biomarkers help monitor the evolution of a disease and the corresponding response to drugs, as well as better predict the clinical outcomes (Kohler et al., 2017). For instance, testosterone glucuronide, when normalized by androsterone glucuronide, can be used as a urinary biomarker of an androgen- and drug-metabolizing enzyme (i.e. UGT2B17), as recently shown through targeted metabolomics analysis (Zhang et al., 2020). Five ω- and (ω-1)-hydroxylated medium-chain acylcarnitines have also been identified as novel CYP3A biomarkers using an untargeted metabolomics approach (Kim et al., 2018).
The cytochrome P450 2D6 (CYP2D6) is responsible for the metabolism of around 25% of all drugs used in clinical practice including antidepressants, analgesics, β-blocking agents and antipsychotics (Gaedigk, 2013). Prescribing CYP2D6 drug substrates is often challenging for physicians because of the large variability in the activity of this enzyme. CYP2D6 is a highly polymorphic gene locus and genotyping assays can be used to predict enzyme activity (Nofziger et al., 2020). However, relying only on genotyping has several limitations. First, it does not take into account environmental factors such as concomitant medications, food intake and disease-related factors (Gaedigk et al., 2018). Second, depending on the technology and database used, some of the rare variants may not be screened or even identified, and an allele may be erroneously categorized as functional (Gaedigk et al., 2018). And third, when duplication or multiplication is detected, a majority of copy number tests do not distinguish which of the two allele has several copies (Langaee et al., 2015; Shah et al., 2016). Therefore, in clinical practice, precision medicine must rely on both real-time phenotyping and genotyping in order to provide the best possible recommendations. Currently, CYP2D6 phenotyping requires the administration of an exogenous probe drug specifically metabolised by this isoenzyme (Samer et al., 2013; Magliocco et al., 2019). Microdosing of the probe drug and enhanced detection capacities of mass spectrometry have lowered the risk of probe-related side effects. However, potential iatrogenic harm (allergic reaction, dosing errors) would only be totally eliminated if endogenous probes were available (Magliocco et al., 2019; Magliocco and Daali, 2020). A recent review summarized human endogenous compounds that have been tagged as potential CYP2D6 markers (Magliocco et al., 2019). One of them stands out. It is a very promising urinary biomarker named M1 (m/z 444.3102). It was characterised, but not yet structurally identified in a non-targeted metabolomics study (Tay-Sontheimer et al., 2014). Some in vitro and animal studies have also demonstrated that CYP2D6 metabolizes the endocannabinoid anandamide (Snider et al., 2008).
Our main objective in this study was to explore the presence of CYP2D6 biomarkers in human urine and plasma, using an untargeted metabolomics approach. For this purpose, healthy volunteers were invited to two sessions (control vs inhibitory). Prior to the inhibitory session, volunteers received over 7 days, a daily dose (10 or 20 mg) of the strong CYP2D6 inhibitor paroxetine. The CYP2D6 genotype and phenotype were also integrated in the data analysis.