Cytochrome P450 2D6 (CYP2D6) is a major enzyme responsible for processing xenobiotics and contributes to the elimination of numerous therapeutic agents. Variation in the CYP2D6 gene produces substantial person-to-person differences in metabolic capacity and can influence how CYP2D6 probe compounds, such as dextromethorphan (DXM), are phenotyped. To explore these effects, we (i) assembled a large DXM pharmacokinetic dataset and (ii) created and verified a physiologically based pharmacokinetic (PBPK) model describing DXM and its metabolites—dextrorphan (DXO) and dextrorphan O-glucuronide (DXO-Glu). Drug–gene interactions (DGI) were incorporated by adjusting CYP2D6 kinetic parameters according to the activity score (AS), enabling simulation of individual genotypes through AS combinations. Variation in CYP3A4 and CYP2D6 activities was characterized using human liver microsome data. The resulting simulations closely replicate observed pharmacokinetics for CYP2D6 genotypes, AS-defined activity levels, and metabolic phenotypes (UM, EM, IM, PM). Using this framework, we examined genotype–phenotype relationships and assessed how CYP2D6 variants affect phenotyping, quantified by the urinary cumulative metabolic ratio (UCMR), defined as DXM/(DXO + DXO-Glu). Sensitivity analyses were used to determine which factors most strongly influence UCMR. Model outputs show high resilience to different dosing procedures (formulation, dose, dissolution, sampling schedule) and to physiological variability. The approach can estimate UCMR distributions in and across populations as a function of AS, and can also predict the likelihood of genotype–phenotype misclassification when allele frequencies are known. It supports individualized prediction of both UCMR and metabolic phenotype from CYP2D6 genotype. All datasets and model files are publicly accessible.
