| Summary: | 碩士 === 臺北醫學大學 === 藥學系 === 91 === Abstract
The purpose of personalized drug therapy is to identify not only the best drug to be administered to a particular patient, but also the most effective and safest dosage from the outset of therapy. It is known that genetic variability in drug response occurs as a result of molecular alterations at the level of drug metabolizing enzymes, drug targets/receptors, and drug transport proteins. It would play a crucial role to correlate the relationship between the genotyping and the phenotyping in successfully accomplishing personalized drug therapy.
In this study, we are intent to construct population pharmacokinetic (PK) model to correlate the relationship between genotyping and phenotyping for the purpose of personalizing drug therapy. At first, dextromethorphan as a probing drug in plasma was analyzed by high-performance liquid chromatography. A sensitive and reproducible HPLC assay was developed for the simultaneous determination of dextromethorphan and its metabolites in plasma. Limits of quantitation in plasma were 1nM for all compounds. In the clinical trial, plasma concentrations of 12 volunteers were quantified with the HPLC method and to evaluated CYP2D6 activity. On the other hand, the inhibition of CYP3A4 by co-administration of grapefruit juice was explored. Genotyping for CYP2D6 was conducted in genomic DNA by polymerase chain reaction (PCR) and Restriction fragment length polymorphism (RFLP) for the allelic frequency distribution of single nucleotide polymorphisms (SNPs) at the same time. It was incorporated the number of the grades of CYP2D6 genotype and physiologic factors (age, height) as covariances into modeling study. The pharmacokinetic model was a 1-compartment model with first order absorption and elimination. The simulation database was generated by SAS nonlinear regression. Using our experimental data and simulation database to regression based on population model with covariances by WinNonMix, respectively.
The optimal model of experimental data was Model 4 and a full matrix structure (Block) for Between-Subject Covariance. The regressed data was achieved convergence and the minimum value of objective function (OFV) was 197.31, -2*ML Log Likelihood (-2LL) was 419.68. On the other hand, the proper model of SAS-simulation database considered GENE population distribution was settled on Model 1 with Block/ Sigma Squared error structure. And the following OFV was 197.31; -2LL was 419.68. Based on the above Model 4 and a full block matrix for Between- Subject Covariance, the full plasma concentration profile would be predicted by using single-point plasma concentration. However the output information was strained and sometimes could not achieved convergence. In addition, the plasma concentration of co-administration with grapefruit juice was regressed by the same model condition. The results appeared that the PK parameters were regressed easily and more achieved convergence.
Taken together, the population PK model could be improved by increasing subject number and CYP2D6 alleles influenced metabolism, altering the model equations, and even modifying the error structure of model. Based on individual genotype, the fittest population PK model will be constructed for personalized drug therapy.
|
|---|
