Duwal Sulav, von Kleist Max
Systems Pharmacology & Disease Control Group, Department of Mathematics and Computer Science, Freie Universität Berlin, Arnimallee 6, 14195 Berlin, Germany.
Eur J Pharm Sci. 2016 Oct 30;94:72-83. doi: 10.1016/j.ejps.2016.01.016. Epub 2016 Jan 18.
A major aim of Systems Pharmacology is to understand clinically relevant mechanisms of action (MOA) of drugs and to use this knowledge in order to optimize therapy. To enable this mission it is necessary to obtain knowledge on how in vitro testable insights translate into clinical efficacy. Mathematical modeling and data integration are essential components to achieve this goal. Two modeling philosophies are prevalent, each of which in isolation is not sufficient to achieve the above described: In a 'top-down' approach, a minimal pharmacokinetic-pharmacodynamic (PK-PD) model is derived from- and fitted to available clinical data. This model may lack interpretability in terms of mechanisms and may only be predictive for scenarios already covered by the data used to derive it. A 'bottom-up' approach builds on mechanistic insights derived from in vitro/ex vivo experiments, which can be conducted under controlled conditions, but may not be fully representative for the in vivo/clinical situation. In this work, we employ both approaches side-by-side to predict the clinical potency (IC values) of the nucleoside reverse transcriptase inhibitors (NRTIs) lamivudine, emtricitabine and tenofovir. In the 'top-down' approach, this requires to establish the dynamic link between the intracellularly active NRTI-triphosphates (which exert the effect) and plasma prodrug PK and to subsequently link this composite PK model to viral kinetics. The 'bottom-up' approach assesses inhibition of reverse transcriptase-mediated viral DNA polymerization by the intracellular, active NRTI-triphosphates, which has to be brought into the context of target cell infection. By using entirely disparate sets of data to derive and parameterize the respective models, our approach serves as a means to assess the clinical relevance of the 'bottom-up' approach. We obtain very good qualitative and quantitative agreement between 'top-down' vs. 'bottom-up' predicted IC values, arguing for the validity of the 'bottom-up' approach. We noted, however, that the 'top-down' approach is strongly dependent on the sparse and noisy intracellular pharmacokinetic data. All in all, our work provides confidence that we can translate in vitro parameters into measures of clinical efficacy using the 'bottom-up' approach. This may allow to infer the potency of various NRTIs in inhibiting e.g. mutant viruses, to distinguish sources of interaction of NRTI combinations and to assess the efficacy of different NRTIs for repurposing, e.g. for pre-exposure prophylaxis.
系统药理学的一个主要目标是了解药物的临床相关作用机制（MOA），并利用这些知识优化治疗。为实现这一使命，有必要了解体外可测试的见解如何转化为临床疗效。数学建模和数据整合是实现这一目标的重要组成部分。两种建模理念较为普遍，但单独使用其中任何一种都不足以实现上述目标：在“自上而下”的方法中，一个最小的药代动力学-药效学（PK-PD）模型从可用的临床数据中推导出来并进行拟合。该模型在机制方面可能缺乏可解释性，并且可能仅对用于推导它的数据所涵盖的场景具有预测性。“自下而上”的方法基于从体外/离体实验中获得的机制见解，这些实验可以在受控条件下进行，但可能无法完全代表体内/临床情况。在这项工作中，我们同时采用这两种方法来预测核苷类逆转录酶抑制剂（NRTIs）拉米夫定、恩曲他滨和替诺福韦的临床效力（IC值）。在“自上而下”的方法中，这需要建立细胞内活性NRTI-三磷酸酯（发挥作用的物质）与血浆前药PK之间的动态联系，并随后将这个复合PK模型与病毒动力学联系起来。“自下而上”的方法评估细胞内活性NRTI-三磷酸酯对逆转录酶介导的病毒DNA聚合的抑制作用，这必须结合靶细胞感染的情况来考虑。通过使用完全不同的数据集来推导和参数化各自的模型，我们的方法作为一种手段来评估“自下而上”方法的临床相关性。我们在“自上而下”与“自下而上”预测的IC值之间获得了非常好的定性和定量一致性，这支持了“自下而上”方法的有效性。然而，我们注意到“自上而下”的方法强烈依赖于稀疏且有噪声的细胞内药代动力学数据。总而言之，我们的工作让我们有信心可以使用“自下而上”的方法将体外参数转化为临床疗效的指标。这可能有助于推断各种NRTIs对例如突变病毒的抑制效力，区分NRTI组合的相互作用来源，并评估不同NRTIs用于重新用途（例如暴露前预防）的疗效。
