Prediction of variability in CYP3A4 induction using a combined 1H NMR metabonomics and targeted UPLC-MS approach

Abstract

The activity of Cytochrome P450 3A4 (CYP3A4) enzyme is associated with many adverse or poor therapeutic responses to drugs. We used (1)H NMR-based metabonomics to identify a metabolic signature associated with variation in induced CYP3A4 activity. A total of 301 female twins, aged 45--84, participated in this study. Each volunteer was administered a potent inducer of CYP3A4 (St. John's Wort) for 14 days and the activity of CYP3A4 was quantified through the metabolism of the exogenously administered probe drug quinine sulfate (300 mg). Pre- and postintervention fasting urine samples were used to obtain metabolite profiles, using (1)H NMR spectroscopy, and were analyzed using UPLC--MS to obtain a marker for CYP3A4 induction, via the ratio of 3-hydroxyquinine to quinine (3OH-Q:Q). Multiple linear regression was used to build a predictive model for 3OH-Q:Q values based on the preintervention metabolite profiles. A combination of seven metabolites and seven covariates showed a strong (r = 0.62) relationship with log(3OH-Q:Q). This regression model demonstrated significant (p < 0.00001) predictive ability when applied to an independent validation set. Our results highlight the promise of metabonomics for predicting CYP3A4-mediated drug response.