A stochastic finite element modelling framework to predict effect sizes on bone mechanics in preclinical studies

目前，在开发新药的过程中，其效果首先在动物群体上进行实验量化，然后在人类群体中进行量化。这一过程既耗时又昂贵。它也没有解释如何调整候选药物以获得最佳效果。因此，将一种新药推向市场需要15年时间，平均花费20亿加元。对于患有骨质疏松症的人来说，这种情况尤其站不住脚，因为目前可用的药物长期疗效不确定，并存在不典型骨折、坏死、癌症和中风等副作用的风险。预计在未来10年内，英国的骨质疏松症患者数量(目前为300万)和NHS治疗骨质疏松性骨折的年成本(目前为40亿GB)将增加高达30%。创新的方法，如预测候选药物的有效大小的计算模型，可以极大地减少药物开发所涉及的成本和时间。效应量是一种常用的药物有效性的统计量词。骨质疏松症药物的作用大小取决于接受该药物和安慰剂的个体组内和组之间的骨力学变化。这项拟议的研究旨在开发一种计算框架，通过量化这些变化来预测效果大小。过去的研究表明，使用有限元(FE)分析可以根据单个小鼠(英国研究中使用的所有动物的75%)的骨骼几何来准确地预测其骨骼力学响应。这项研究的新颖和创新贡献是应用随机有限元(SFE)方法来计算给予候选药物和安慰剂的小鼠组内和组之间骨力学的上述变化，从而预测药物的效果大小。通过采用已在骨骼研究中广泛使用的高效确定性有限元求解器，将开发一种SFE求解器。就其本身而言，这种适应将极大地提高当前SFE在分析具有非常大的模型尺寸的问题(通常在骨骼研究中)的能力。这一进展同时涉及多个工程领域。拟议框架的新颖预测能力将对使用MICE进行的研究产生重大影响。对于现有的药物，它将比实验中更完整地预测小鼠骨骼力学的诱导变化，因为在实验中使用的动物样本有限(通常是小样本)。它还将预测在过去研究中记录的骨骼几何变化范围内的任何骨骼几何变化(包括尚未观察到的变化)对骨骼力学的影响大小。这有助于在现有药物的空间内确定最佳效果的区域。通过与体外研究(如细胞培养、组织工程)提供的计算模型相结合，该框架可用于预测全新分子的效应大小。拟议研究的一个关键重点是详细评估将要开发的建模框架的可信度。这将通过使用在过去的实验中已经收集的小鼠骨骼几何数据进行一系列干预来实现。这一严格的可信度评估将支持该框架未来在临床背景下的适应，在那里可以预测候选药物对人类的影响大小，从而进一步降低药物开发所涉及的成本和时间。

Presently, in the process of developing a new drug, its effect is quantified experimentally first on groups of animals and then on groups of humans. This process is time consuming and expensive. It also does not explain how a candidate drug can be tuned for optimal effect. As such, it takes 15 years and costs £2 billion on average to bring a new drug to market. This situation is particularly untenable for people with osteoporosis, where currently available drugs have uncertain long-term benefit and pose risks of side-effects such as atypical fractures, necrosis, cancer and stroke. It is expected that over the next 10 years, the number of people in the UK with osteoporosis (currently, 3 million) and the annual costs to NHS for treating osteoporotic fractures (currently, £4 billion) will increase by up to 30%.Innovative approaches, such as computational modelling to predict effect sizes of candidate drugs, could dramatically reduce the cost and time involved in drug development. Effect size is a commonly used statistical quantifier of a drug's effectiveness. The effect size of an osteoporosis drug depends on the variation in bone mechanics within and between groups of individuals receiving the drug and a placebo. The proposed research aims to develop a computational framework that can predict effect sizes by quantifying such variations.Past research has shown that the bone mechanical response of an individual mouse (75% of all animals used in UK research) can be accurately predicted from its bone geometry using finite-element (FE) analysis. The novel and innovative contribution of the proposed research is to apply a stochastic FE (sFE) approach to compute the above-mentioned variations in bone mechanics within and between groups of mice given a candidate drug and placebo and thereby to predict the drug's effect size.An sFE solver will be developed by adapting a highly efficient deterministic FE solver that is already used widely in bone research. In itself, this adaptation will substantially advance current sFE capability in analysing problems with very large model size (typical in bone research). This advance is relevant to multiple engineering fields at once. The novel predictive capabilities of the proposed framework will have a major impact on studies using mice. For existing drugs, it will predict the induced variation in mouse bone mechanics more completely than is possible in experiments because of finite (and typically small) sample of animals used in the latter. It will also predict the effect-size on bone mechanics for any change in bone geometry (including those not yet observed) that falls within the range of bone geometry changes recorded in past studies. This can help identify regions of optimal effect within the space of existing drugs. By coupling with computational models informed by in vitro studies (e.g. cell culturing, tissue engineering) the framework can be adapted to predict effect sizes for entirely new molecules.A key focus of the proposed research is to assess in detail the credibility of the modelling framework to be developed. This will be achieved by using mouse bone geometry data already collected in past experiments for a range of interventions. This rigorous credibility assessment will support the future adaptation of the framework in the clinical context, where the effect size of candidate drugs on humans can be predicted, thus further bringing down costs and time involved in drug development.