儿童药动学试验设计的样本量与采样点数的模拟探索

Requirement of sample size and sampling points in pediatric pharmacokinetic trial

目的:研究在保证模型参数一定准确度的条件下,儿童药动学试验中的样本量和采样点数设计的基本要求。方法:用群体药动学分析方法,以奥司他韦为工具药,进行药代动力学模拟,构建符合真实人群数据特点的儿童样本120例,按采样时间点数分别为2、3、4、5、6、8、10,同时每个采样时间点数下按样本量分别为20、40、60、80、100、120设计42种不同方案的采样,估算各方案下药时曲线下面积(AUC_(0→t)、AUC_(0→∞))、血药峰浓度(C_(max))参数,将各参数的估算值与初始值的比值范围作为预测准确度,评价适宜的儿童稀疏采样设计方案。结果:随着样本量和采样点数的增加,预测准确度不断改善,活性药物羧化奥司他韦的参数预测准确度相较于前体药物奥司他韦较低。以药动学参数估算值与初始值的比值在95%~105%内,同时其90%置信区间相比初始值在80%~120%为评价标准时,发现研究奥司他韦的代谢过程的最低采样设计要求为3个采样时间点,40个样本量,此时AUC_(0→t)、AUC_(0→∞)、C_(max)参数中位数预测偏差为0.96%、2.51%、3.8%,90%置信区间上下限与初始值预测偏差依次为-4.96%~6.66%、-7.970%~20.031%、-5.99%~16.13%;研究奥司他韦羧化代谢物的最低采样要求为5个采样时间点,60个样本量;此时3个参数中位数预测偏差为1.070%、1.203%、2.005%,90%置信区间上下限与初始值预测偏差依次为-7.69%~5.27%、-8.41%~6.69%、-16.98%~16.23%。结论:采样点个数对研究奥司他韦药物代谢过程有显著的影响。在确定合适的儿童临床稀疏采样设计方案中发现低于3个采样点的方案预测结果不可信,在预测奥司他韦羧化代谢物的参数时不低于5个采样点的预测结果稳定。

AIM： To explore the minimum requirements for sample size and sampling points designed to ensure a certain accuracy in pediatric phar macokinetic trial. METHODS： The pharmacokinetic simulation was carried out using oseltamivir as a tool for population pharmacokinetic analysis, to generate data of 120 children with the characteristics of real population which included covariates, PK parameters and plasma concentrations. Forty-two different sparse sampling schemes under different sampling size and sample points conditions was designed. The data of 120 children were sampled at the sampling time points of 2,3,4,5,6,8,10 and each sampling time points were by sample size were 20,40,60,80,100, 120 to estimate the area under the curve （AUCo→t, AUCo→∞） and peak blood concentration （ Cmax ）. The ratio of the estimated value of each parameter to the initial value as the prediction accuracy was calculated to evaluate the appropriate sparse sampling scheme for pediatric clinical trial. RESULTS：As the sample size and number of sampling points increase, the prediction accuracy improves continuously, and the parameter prediction accuracy of the active drug earboxylation of osehamivir was lower than that of the precursor drug oseltamivir. If we set a evaluation criteria that the ratio of the estimated pharmacokinetic parameters to the initial value within 95% to 105%, meanwhile the 90% confidence interval compared with the initial value of 80% to 120%, the study has found that the minimum sampling design for the study of oseltamivir metabolism was 40 samples at three sampiing time points. The predicted median deviations of AUCo→t, AUCo→∞ and Cmax were 0.96% and 2.51% 3.8%, 90% confidence interval deviations of AUCo→t, AUCo→∞ and Cmax were -4.96%-6.66%,-7.970%-20.031%,-5.99%-6.13% ; The minimum oseltamivir carboxylate metabolites sampling design was 60 samples at five sampling time points. The median deviation of three parameters was 1.070%,1.203%,2.005%, the 90% confidence interval deviation of three parameters was -7.69%-5.27%, 8.41% -6.69%, -16.98% -6.23%. CONCLUSION： The number of sampling points has a more significant effect on the prediction of the metabolic process of osehamivir. Scenario prediction results of less than 3 sampling points were found to be unreliable in identifying appropriate pediatric clinical sampling design programs. The predictions for oseltamivir carboxylate metabolites with no less than 5 sampling points were stable.