摘要目的 应用非线性混合效应模型(NONMEM)法分别建立丙戊酸在不同年龄阶段(婴幼儿、学龄前、学龄+青春期)的中国癫□患儿中的群体药代动力学(PPK)模型,为制定个体化给药方案提供一定参考.方法 收集518例(建模组404例,验证组114例)2010年至2013年在北京大学第一医院儿科 门诊应用丙戊酸治疗的癫□忠儿的临床资料,包括生物学资料、用药资料、血药浓度测定结果及同期生化指标.其中婴幼儿组191例,学龄前组172例,学龄+青春期组155例.采集有效血药谷浓度数据655个(建模组535个,验证组120个),其中婴幼儿组218个,学龄前组214个,学龄+青春期组223个.应用NONMEM软什,采用一级吸收和消除的一室模型,建立不同年龄组及全年龄组癫□患儿丙戊酸PPK模型.考察患儿的生物学信息以及联合用药对清除率(CL/F)和表观分布容积(V/F)的影响.采用1 000次自举法(Bootstrap)对模型进行验证.使用验证组数据来评价不同年龄组及全年龄组丙戊酸PPK模型的准确度和精密度,比较二者的平均预测误差(MPE)、平均绝对误差(MAE)和平均根方差(RMSE).结果 不同年龄组及全年龄组丙戊酸最终模型:1.婴幼儿期:CL/F (L/h)=0.154×(WEIG/11.5)1.31,V/F (L)=3.32×(WEIG/11.5)2.49;2.学龄前期:CL/F (L/h)=0.211×(WEIG/18) 0.534,V/F (L) =4.08;3.学龄+青春期:CL/F (L/h)=0.287×(WEIG/29.75)0.464,V/F (L) =7.05;4.全年龄组:CL/F (L/h)=0.197×(WEIG/18)0.648×1.09CEND,V/F(L) =4.25×(WEIG/18) 0.714 (CEND为性别,男=1,女=0;WE IG为体质量).Bootstrap法结果与模型拟合值相符.验证组数据计算不同年龄组与全年龄组模型MPE、MAE和RMSE,结果显示不同年龄组模型各值均小丁全年龄组模型,其具有更高的准确性和精确性.结论 应用NONMEM软什成功地建立了我国癫□儿章不同年龄阶段的丙戊酸PPK模型,其中体质量为影响清除率最重要的因素.不同年龄组模型相较丁全年龄组模型具有更好的预测性.

abstractsObjective To establish population pharmacokinetic (PPK) models with valproic acid (VPA) in different age groups (infant and toddler,preschool age,school age and adolescence) of Chinese epileptic children by nonlinear mixed effect modeling (NONMEM),and provide evidence for individualized drug therapy.Methods Cases of 518 epileptic children were collected from 2010 to 2013 in the Pediatric Outpatient Clinics of the First Hospital,Peking University.The clinical information included demographic data,medication,serum concentration data and blood biochemical parameters.The patients were divided into infants and toddlers (n =191),children at preschool age (n =172) and children at school and adolescence age (n =155).A total of 665 steady-state serum trough concentration samples were extracted.Population pharmacokinetic models of VPA were established by NONMEM program by using first-order absorption and elimination.Demography and drug combination was investigated for influence on apparent clearance (CL/F) and apparent volume of distribution (V/F).The reliability and stability of the PPK models were evaluated by 1 000 times of Bootstrap procedure.To assess the accuracy and precision of the different age groups models and whole age model,the average prediction error (MPE),average absolute error (MAE) and average root variance (RMSE) were compared.Results The final model of VPA in different age groups and whole age group were:1.the infant and toddler's age group CL/F (L/h) =0.154 × (WEIG/11.5) 1.31,V/F (L) =3.32 × (WEIG/11.5)2.49.2.preschool age group CL/F (L/h) =0.211 × (WEIG/18)0.534,V/F (L) =4.08.3.school and adolescence age group CL/F (L/h) =0.287 × (WEIG/29.75)0.464,V/F (L) =7.05.4.whole age group CL/F (L/h) =0.197 × (WEIG/18)0.648 × 1.09CEND,V/F (L) =4.25 × (WEIG/18) 0.714 (GEND was gender,male =1,female =0; WEIG was weight).The results of Bootstrap procedure were corresponding to the estimates from NONMEM.The values of the different age groups model MPE,MAE,RMSE were less than ones of the whole age group,which suggested the different age groups models had better precision and accuracy than the whole age group models.Conclusions PPK models of VPA in different age groups of Chinese epileptic children by NONMEM were successfully established.Weight was identified as a significant covariate on VPA clearance.Compared with the whole age group models,the different age groups models are more reliable.