Multiscale computational models for developing combination cancer therapy

描述（由申请人提供）：开发有效的癌症联合治疗是具有挑战性的，因为许多癌症药物作用于交叉的信号通路，这些信号通路可以相互干扰。例如，众所周知，药物-药物相互作用可以发生巨大变化，例如，根据治疗条件（药物浓度、治疗时间和药物排序），从协同作用到拮抗作用。此外，在体内条件下，药物浓度随时间变化（即药代动力学或PK），不同药物具有不同的PK，这使得很难在药物浓度通常保持不变的培养细胞中翻译研究结果。我们建议开发多尺度、可推广的计算PK和药效学（PD）模型来解决这些挑战。首先，我们将开发单一药物的体外PD预测模型。这些PD模型采用确定性模型（根据药物作用和细胞周期定位指定单个细胞的命运）和概率模型（确定所有细胞的命运）的组合。这些模型共同描述单个细胞的反应和整个细胞群的总体反应（作为单个细胞的集体反应），作为治疗（药物浓度，治疗时间）和细胞的化学敏感性的数学函数。其次，我们将开发联合治疗的体外PD预测模型。药物可以在细胞周期分布（细胞周期相互作用）和分子靶点（分子相互作用）两个层面上相互作用。我们将扩展上述单一药物的方法，为三种类型的组合开发两种药物联合模型：(a)仅具有细胞周期相互作用的药物，(b)具有分子相互作用的药物，其中两种药物都具有细胞毒性作用，以及(c)具有分子相互作用的药物，其中一种药物本身没有细胞毒性，但可以增强或降低另一种药物的活性。第三，我们将开发将体外PD转化为体内PD的方法。我们将解决两个问题，即体外条件下（恒定C）到体内情况下（改变C）的转换，并扩展体外PD模型，以包括体内肿瘤中存在的非循环G0细胞。最后，我们将整合体外PD模型和体内PK模型，并评估整合模型的性能。我们将开发综合的PK-PD模型来描述单一药物的体内效应，然后是它们的组合模型。在荷瘤动物中评估模型性能。所提出的模型是同类中第一个，将能够计算不同药物和/或不同体内治疗方案/序列的潜在组合的结果。这种预测模型可以减少结果的不确定性和实验的数量，从而加速开发有效的联合癌症治疗方法。

DESCRIPTION (provided by applicant): Development of effective combination therapy for cancer is challenging because many cancer drugs act on intersecting signaling pathways that can interfere with each other. For example, it is well established that drug-drug interactivity can change drastically, e.g., from synergy to antagonism, depending on the treatment conditions (drug concentrations, treatment time, and sequencing of the drugs). Further, under in vivo conditions, drug concentrations change with time (i.e., pharmacokinetics or PK) and different drugs have different PK, which make it difficult to translate the findings in cultured cells where drug concentrations are typically kept constant. We propose to develop multiscale, generalizable computational PK and pharmacodynamic (PD) models to address these challenges. First, we will develop predictive in vitro PD models for single agents. These PD models employ a combination of deterministic models (that designate the fate of a single cell based on drug actions and cell cycle location) and probabilistic models (that determine the fate of all cells). These models jointly depict the response of an individual cell and the overall response of whole cell population (as the collective response of individual cells), as mathematical functions of a treatment (drug concentrations, treatment time) and chemosensitivity of a cell. Second, we will develop predictive in vitro PD models for combination therapies. Drugs can interact on two levels, i.e., cell cycle distribution (cell cycle interactivity) and molecular targets (molecular interactivity). We will extend the above approaches for single agents to develop two-drug-combination models for three types of combinations: (a) drugs with only cell cycle interactivity, (b) drugs with molecular interactivity where both drugs have cytotoxic effects, and (c) drugs with molecular interactivity where one drug does not have cytotoxicity on its own but can enhance and reduce the activity of the other drug. Third, we will develop methods to convert in vitro PD to in vivo PD. We will address two issues, i.e., conversion of CxT under in vitro conditions (constant C) to in vivo situations (changing C), and extend the in vitro PD models to include the non-cycling G0 cells present in vivo tumors. Lastly, we will integrate in vitro PD models with in vivo PK models and evaluate the performance of the integrated models. We will develop integrated PK-PD models to describe the in vivo effects of single agents, followed by models for their combinations. Model performance is evaluated in tumor-bearing animals. The proposed models are first-of-its-kind and will enable the computation of outcomes of potential combinations of different drugs and/or different in vivo treatment schedules/sequences. Such predictive models can reduce the uncertainty in outcomes and the amount of experimentation and thereby accelerate the development of effective combination cancer therapies. PUBLIC HEALTH RELEVANCE: We propose to develop multiscale, computational pharmacokinetic-pharmacodynamic (PK-PD) models to link systemic PK with cell cycle-directed drug actions and time-dependent changes in cellular response, in order to predict the outcome of a given treatment in vivo. Such predictive models can reduce the uncertainty in outcomes and the amount of experimentation and thereby accelerate the development of effective combination cancer therapies.