Pharmacometabolomics-based Prediction Model Of Bile Acid Pharmacokinetics

The presence of endogenous baseline levels and the individual variances has brought lots of difficulties to the behavior assessment of endogenous drugs in vivo.It is preferable to explore a strategy that can effectively predict the variability in endogenous drugs metabolism and disposition,which was an important subject in clinical medication safety.For this purpose,we evaluated the potential of a pharmacometabolomics approach to predict individual variances in the pharmacokinetics(PK)of cholic acid.The stable isotope-labeled cholic acid was selected as the substitute analyte of cholic acid to ensure the accurate blood concentration measured.Ultra performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry(UPLC-Q-TOF/MS)analysis was performed on the rat predose urine samples to get the large-scale metabolite profiling.After oral administration of deuterium 4-cholic acid(d4-cholic acid),the plasma concentrations were determined using liquid chromatography-hybrid triple quadrupole linear ion trap mass spectrometry(LC-QTRAP/MS)and the pharmacokinetic parameters were calculated by WinNonlin software.Partial least squares(PLS)modeling was calculated based on the data of predose urine metabolites to predict the pharmacokinetic parameters of d4-cholic acid and to select the metabolites that substantially contributed to such prediction.At the same time,network biology analysis was applied to establish the relationship between predose urine metabolites and cholic acid,give the biological interpretation of the individual variances in cholic acid metabolism and disposition.1.Metabolic profiling of predose urine samplesSince there is rich metabolome information in urine samples,large-scale metabolite profiling is essential for screening the urinary metabolites that can predict of individualized PK.LC-MS analysis can provide a wide coverage of metabolome required to achieve this goal with its high sensitivity.In the study,UPLC-Q-TOF/MS analysis was performed on the predose urine samples collected over 24 h of the 28 rats before they were dosed with d4-cholic acid,so as to generate global urine metabolic profiles from them.3,510 metabolic features in positive mode and 1,923 metabolic features in negative mode were detected from the 28 datasets,respectively.These major peaks cover a broad range of metabolites,including amino acids,organic acids,lipids,nucleosides,and other urinary metabolites.2.A quantitative method was developed and validated for the determination of d4-cholic in rat plasma.In this study,in order to determine the accurately concentration of cholic acid after drug administration,the stable isotope-labeled cholic acid was selected as the substitute of cholic acid.The plasma concentrations were determined using liquid chromatography-hybrid triple quadrupole linear ion trap mass spectrometry(LC-QTRAP/MS).Chromatographic separation was achieved on a BEH C18 column(100 mm × 2.1 mm i.d.,1.7 ?m)by a gradient elution at a flow rate of 0.40 mL/min.All analytes were monitored by multiple reaction monitoring(MRM)mode with negative electrospray ionization.The calibration curves of d4-cholic acid is linear(r>0.9981)over wide concentration range.The intra-and inter-day precision ranged from 2.0?9.9%and 4.4?9.5%,while the accuracy ranged from 1.3?2.7%and-1.1?5.9.The recovery of d4-cholic acid was ranged from 81.86?96.35%,while the recovery of IS was 94.48%.The matrix effects were in the range of 90.54?103.09%,which could be neglected.3.Pharmacokinetic analysisTo evaluate the pharmacokinetic response,the stable isotope-labeled cholic acid was selected as the substitute of cholic acid,and the plasma concentration of d4-cholic acid was measured at various time points after oral administration(60 mg/kg).The result shows that the plasma concentration profiles of d4-cholic acid versus time in the 28 rats,with the mean profile overlaid,revealing a high degree of individual variation regarding pharmacokinetic responses to d4-cholic acid.Using these profiles,the pharmacokinetic parameters AUC0-72 for d4-cholic acid were estimated using WinNonlin software(Pharsight,USA),which its maximum and minimum values differed by approximately 10-fold.multiple peaks plasma concentration-time profiles were observed following the oral administration of d4-cholic acid.This phenomenon might be explained by the mutual transformation of bile acids,enterohepatic recirculation and other complex reasons.4.Prediction of individualized PK bansed on pharmacometabolomicsA two-stage PLS analysis was employed to build a statistical mode that can predict individualized PK effectively using the measured metabolic data.The first stage PLS analysis was to reduce the dimensionality of X data;the second stage was to predict the pharmacokinetics,as well as select the key metabolites contributing to that prediction.PLS analysis is a multivariate framework that can be used to find the relationships between two groups of variables(here,metabolites and AUC)and that can build a supervised model(PLS model).The PLS model can efficiently screening the metabolites(X variables)that make a large contribution to the ability to predict the response variable(Y variable,AUC).The initial PLS analysis was performed on all 3,510 and 1,923 peak intensities(X block,prediction variables)related to the AUC(Y block,response variable)in positive and negative mode,respectively.On the basis of this analysis,we selected the metabolites(X variables)that made a large contribution to predicting the AUC(Y variable).We then performed a second PLS analysis to predict individual AUC values,using the selected metabolites.According to the initial PLS analysis,116 and 74 variables of high VIP values(VIP>1.5)were highly relevant to AUC in positive and negative mode,respectively.On the basis of these selected variables,we built the second PLS model to be used for predicting the AUC,and a set of 17 and 14 key metabolic features in positive and negative mode were selected,respectively,which characterizing individualized PK,represented by the AUC as those with VIP>1 in the second PLS model.In which 28 of those selected metabolites were identified,those identified metabolites and the AUC are shown strong correlations(r>0.6).The results of linear correlation regression analysis revealed that the metabolites that selected from the second stages PLS analysis correlated well with AUC.In this paper,in order to illustrate the relationship between the selected metabolites and cholic acid in molecular level,a metabolic network was constructed using network biology analysis to describe the selected metabolites interactions with cholic acid.The results shown that the variation in the baseline level of sarcosine and s-adenosyl-l-homocysteine is significantly correlated to the pharmacokinetic response after cholic acid treatment,the use of the two metabolites allows us to predict the individualized PK of cholic acid before administering the drug to a rat.