论文部分内容阅读
目的:建立中国癫痫患儿口服丙戊酸的群体药动学模型,促进合理用药。方法:收集472例癫痫患儿口服丙戊酸后的血药浓度数据和临床资料。将患儿随机分成2组,PPK模型组(n=354):应用NLME程序建立一房室群体药动学模型(个体间变异采用指数模型,残差变异采用加法模型表示),考察各协变量对参数Ka、Vd和Cl的影响。用拟合优度、自举验证、直观预测检验(VPC)对最终模型的性能进行内部验证。PPK验证组(n=118):用最终模型预测验证组患儿的血药浓度,与实测值比较计算平均预测误差(MPE)、平均绝对预测误差(MAE)、平均预测误差平方(MSE)和均方根预测误差(RMSE)对最终模型进行外部验证。结果:PPK最终模型为:Ka=1.18×eηKa h-1;Vd=0.30×eηVd L·kg-1;Cl=0.008×(WT/18)-0.92×eηCl L·kg-1·h-1。体质量负相关影响Cl。拟合优度、自举验证和VPC的评价结果表明最终模型稳定、预测结果可靠。外部验证最终模型结果为:MPE=-0.09 mg·L-1,MAE=0.72 mg·L-1,MSE=0.76(mg·L-1)2,RMSE=0.87 mg·L-1。血药浓度实测值和最终模型的个体预测值(DV vs IPRED)的决定系数R2=0.998 1。外部验证说明最终模型预测准确度高。结论:本研究成功建立了中国癫痫患儿口服丙戊酸后的群体药动学模型。模型结构表明丙戊酸体重校正的清除率随患儿年龄和体质量的增加有下降趋势。
Objective: To establish a population pharmacokinetic model of oral valproic acid in children with epilepsy in China and to promote rational drug use. Methods: The data of plasma concentration and clinical data of 472 children with epilepsy after oral administration of valproic acid were collected. The children were randomly divided into two groups, PPK model group (n = 354): NLME program was used to establish a one-compartment group pharmacokinetic model (inter-individual variation using exponential model, residual variation using additive model), examine the covariates Effect on parameters Ka, Vd and Cl. Goodness of fit, bootstrap verification, visual predictive testing (VPC) were used to verify the performance of the final model. PPK validation group (n = 118): The final model was used to predict the plasma concentration of children in the validation group and calculate the average prediction error (MPE), mean absolute prediction error (MAE), mean prediction error square (MSE) The root mean square prediction error (RMSE) is used to verify the final model externally. Results: The final model of PPK was: Ka = 1.18 × eηKa h-1; Vd = 0.30 × eηVd L · kg-1; Cl = 0.008 × (WT / 18) -0.92 × eηCl L · kg-1 · h-1. Negative correlation of body mass. Goodness of fit, bootstrap verification and VPC evaluation results show that the final model is stable and the prediction result is reliable. The results of external validation of the final model were: MPE = -0.09 mg · L-1, MAE = 0.72 mg · L-1, MSE = 0.76 (mg · L-1) 2 and RMSE = 0.87 mg · L-1. R2 = 0.998 1, which is the coefficient of determination of the plasma concentration and the individual predictive value of the final model (DV vs IPRED). External validation shows that the final model has high predictive accuracy. Conclusion: This study successfully established a population pharmacokinetic model of epilepsy in children after oral administration of valproic acid. Model structure shows that clearance of body weight corrected valproic acid with the age and body weight of children with a downward trend.