论文部分内容阅读
目的利用来自恩泽医疗中心3家医院的成年万古霉素治疗患者的治疗药物监测数据研究其群体药动学(population pharmacokinetics,PPK)模型,并利用所建药动学参数模型和该类群体患者治疗过程中的单点谷浓度经贝叶斯反馈参数估算法计算个体化药动学参数,建立个体化给药方案。方法收集128例患者,共监测235个血清浓度,采用Kinetica软件的PPK模块中的一室静脉给药模型通过期望最大法(EM)和贝叶斯反馈拟合数据得到基础模型。利用逐步正向回归法研究消除速率常数(Kel和Vd)与患者个体协变量肌酐(Scr)、年龄(Age)、体质量(Wt)、性别(Sex)以及合并用药之间的关系,并拟合最终模型。利用内部自举法和外部验证法对模型进行评价。结果本研究中最终模型公式为Kel=θ_1×(Scr)~(θ_2)×(Sex)~(θ_3),V=θ_4×(Scr)~(θ_5)×(Age)~(θ_6)×(Wt)~(θ_7)(θ_1=0.18;θ_2=-0.19;θ_3=-0.31;θ_4=20.85;θ_5=-0.52;θ_6=-0.23;θ_7=0.98)。最终模型对应的CL和Vd群体典型值分别为5.0 L·h~(-1)和66.9 L,外部验证中贝叶斯预测平均误差分别为0.8 L·h~(-1)和9 L。结论本研究通过所建立的万古霉素PPK模型,较好的反应出中国成年患者的万古霉素PPK特征,贝叶斯单点反馈误差较低,为提高治疗效果、减少不良反应以及实现个体化给药提供了重要的理论实验参考依据。
OBJECTIVE: To study population pharmacokinetics (PPK) models using therapeutic drug monitoring data from adult vancomycin-treated patients in three hospitals in Enze Medical Center. The model of pharmacokinetic parameters and the treatment of patients in this group During the process of single-point trough concentration, individualized pharmacokinetic parameters were calculated by Bayesian feedback parameter estimation method to establish individualized dosing regimen. METHODS: A total of 128 patients were enrolled. A total of 235 serum concentrations were monitored. The baseline model was obtained by fitting the expected maximum method (EM) and Bayesian feedback fitting data using a one-compartment intravenous injection model in the Kinetica software PPK module. The relationship between the elimination rate constants (Kel and Vd) and the individual coxsarum creatinine (Scr), age, Wt, Sex and drug combination were studied by stepwise forward regression. The final model. Use the internal bootstrap method and external verification method to evaluate the model. Results The final model formula in this study is Kel = θ_1 × Scr ~ θ_2 × Sex ~ θ_3, V = θ_4 × Scr ~ θ_5 × Age ~ θ_6 × Wt ) ~ (θ_7) (θ_1 = 0.18; θ_2 = -0.19; θ_3 = -0.31; θ_4 = 20.85; θ_5 = -0.52; θ_6 = -0.23; θ_7 = 0.98). The typical CL and Vd population corresponding to the final model are 5.0 L · h -1 and 66.9 L, respectively. The average Bayesian prediction error of external validation is 0.8 L · h -1 and 9 L, respectively. Conclusions The vancomycin PPK model established in this study reflects the PPK characteristics of vancomycin in Chinese adult patients, and the single-point feedback error of Bayesian is lower. In order to improve the therapeutic effect, reduce adverse reactions and achieve individualization Administration provides important theoretical experimental reference.