In Vitro Reaction Phenotyping and Projection of Drug Interactions Mediated by Cytochrome P450 Enzymes for Central Nervous System Drugs as Victims: Implications for Safety in Vulnerable Populations

Accurate prediction of cytochrome P450-mediated victim drug-drug interactions (DDIs) is critical for the safe use of central nervous system (CNS) drugs, particularly in populations at high risk of polypharmacy. This study evaluated the utility and limitations of in vitro human liver microsomal data for predicting victim DDIs of CNS compounds metabolized by CYP1A2, CYP2B6, CYP2D6, and CYP3A4. Reaction phenotyping with isoform-selective inhibitors quantified the fractional metabolic contribution (f_m,cyp) of each enzyme, and inhibition parameters for fluvoxamine (CYP1A2), ticlopidine (CYP2B6), cinacalcet/paroxetine (CYP2D6), and ketoconazole (CYP3A4) were incorporated into a mechanistic static model using unbound hepatic inlet concentrations as surrogates for site of inhibition exposure. Predictions for CYP2B6-, CYP2D6-, and CYP3A4-mediated interactions were generally within 3-fold of observed clinical values. In contrast, CYP1A2 substrates showed high sensitivity to f_m,cyp values exceeding 0.9, resulting in underprediction (eg, ramelteon-fluvoxamine, 66-fold) or overprediction (eg, tacrine). Contributing factors may include incomplete characterization of minor metabolic pathways, hepatic accumulation of inhibitors, and limitations of static modeling assumptions. Overprediction of select CYP3A4 interactions, such as lurasidone-ketoconazole, likely reflects underestimation of the fraction escaping intestinal metabolism. These findings indicate that human liver microsome-based reaction phenotyping provides valuable guidance for early DDI risk assessment, particularly for CYP2B6, CYP2D6, and CYP3A4 substrates, but small deviations in f_m,cyp for CYP1A2 substrates can lead to substantial underestimation of clinical risk. Integration with physiologically based pharmacokinetic modeling and refined f_m,cyp estimates may enhance prediction accuracy and better inform safe CNS drug use, especially in populations prone to polypharmacy. Significance Statement: Accurate prediction of P450-mediated drug-drug interactions is essential for safe central nervous system drug therapy, especially in patients at high risk of polypharmacy. This study shows that human liver microsome-based reaction phenotyping reliably predicts CYP2B6, CYP2D6, and CYP3A4 interactions, whereas small deviations in f_m,cyp for CYP1A2 substrates can cause substantial under- or overprediction. Incorporating physiologically based pharmacokinetic modeling and refined f_m,cyp estimates can enhance drug-drug interaction risk assessment and guide safer coadministration of central nervous system-active drugs with potent P450 inhibitors.