Estimation of feasible solution space using Cluster Newton Method: application to pharmacokinetic analysis of irinotecan with physiologically-based pharmacokinetic models.
Estimation of feasible solution space using Cluster Newton Method: application to pharmacokinetic analysis of irinotecan with physiologically-based pharmacokinetic models.
复制标题
DOI：
10.1186/1752-0509-7-s3-s3
复制
发表时间：
2013-10-16
影响因子：
--
通讯作者：
Konagaya A
中科院分区：
文献类型：
--
作者：
Yoshida K;Maeda K;Kusuhara H;Konagaya A
为了促进新药开发，基于生理学的药代动力学 (PBPK) 建模方法作为充分理解和预测复杂药代动力学现象的工具受到越来越多的关注。由于 PBPK 模型中再现生理功能的参数数量往往很大，因此有效的参数估计方法至关重要。我们成功地应用了最近开发的算法来估计可行的解决方案空间，称为聚类牛顿法（CNM），以揭示两个癌症患者组中伊立替康药代动力学改变的原因。经过对原始 CNM 算法进行改进以保持参数多样性后，在 10 次迭代、3000 个虚拟样本和 15 分钟内（Intel Xeon E5-1620 3.60GHz × 1 或 Intel Core i7-870 2.93GHz × 1）成功估计出伊立替康 PBPK 模型中 55 或 56 个参数的可行解空间。在参数估计过程之后，控制参数或参数相关性得到澄清。提出了伊立替康药代动力学改变的可能原因，但还不是结论性的。 CNM的应用通过求解包含常微分方程(ODE)的系统的反问题获得了可行解空间。这种方法可以让我们在有限的信息下可靠地了解其他需要估计大量参数的复杂现象。设计前瞻性研究以进一步研究感兴趣的现象也很有帮助。
To facilitate new drug development, physiologically-based pharmacokinetic (PBPK) modeling methods receive growing attention as a tool to fully understand and predict complex pharmacokinetic phenomena. As the number of parameters to reproduce physiological functions tend to be large in PBPK models, efficient parameter estimation methods are essential. We have successfully applied a recently developed algorithm to estimate a feasible solution space, called Cluster Newton Method (CNM), to reveal the cause of irinotecan pharmacokinetic alterations in two cancer patient groups. After improvements in the original CNM algorithm to maintain parameter diversities, a feasible solution space was successfully estimated for 55 or 56 parameters in the irinotecan PBPK model, within ten iterations, 3000 virtual samples, and in 15 minutes (Intel Xeon E5-1620 3.60GHz × 1 or Intel Core i7-870 2.93GHz × 1). Control parameters or parameter correlations were clarified after the parameter estimation processes. Possible causes in the irinotecan pharmacokinetic alterations were suggested, but they were not conclusive. Application of CNM achieved a feasible solution space by solving inverse problems of a system containing ordinary differential equations (ODEs). This method may give us reliable insights into other complicated phenomena, which have a large number of parameters to estimate, under limited information. It is also helpful to design prospective studies for further investigation of phenomena of interest.