Complementary animal and computational models for biomarker identification in ascending thoracic aortic aneurysm

摘要 升胸主动脉瘤（ATAA）是一种主要的心血管健康问题，其特征在于： 一种最终可能分离或破裂的扩张的主动脉。ATAA提出了一个严重的挑战，因为手术是 困难和危险，因此动脉瘤修复标准必须平衡夹层和/或破裂的风险， 手术的风险。目前的手术指南是基于ATAA直径或生长速度，但高达60%的 ATAA患者在达到手术标准之前经历夹层，因此明确需要 动脉瘤衰竭的其他生物标志物。可能的生物标志物分为广泛的类别， 微观结构，几何和生物流体，但它是具有挑战性的，以获得足够的人类数据来计算和 将这些生物标志物与失败等关键结果相关联。很可能一个单一的生物标志物不是 足够的，但直观上不明显的复合生物标志物可能是重要预测所必需的 患者的结果。在本提案中，我们将使用模型的组合：1）ATAA小鼠模型 与马凡氏综合征相关，2）ATAA生长和重塑的多尺度、多物理模型，以及3） 从真实的患者成像数据导出的虚拟患者模型，以确定复合生物标志物， 预测ATAA的生长、进展和失败。我们的第一个具体目标是使用遗传小鼠模型的ATAA 与马凡氏综合征相关，以表征动脉瘤进展和失败， 详细信息，量化主动脉形状，组织成分，组织机械特性和血流动力学， 时间这种程度的细节在人类患者中是不可能的，并且对于验证和测试以下假设是必要的： 在我们的多尺度，多物理场模型在特定目标2的增长和重塑规则，并提供一个初步的 一组生物标志物，以评估我们在特定目标3中的虚拟患者。我们的第二个具体目标是开发一个 ATAA生长和重塑的新型多尺度、多物理场计算模型， 与特定目标1中的小鼠数据进行比较，并用于预测真实的和 《特殊目标3》中的虚拟人患者在我们的第三个具体目标，我们将使用现有的人类ATAA扫描 从马凡氏综合征患者生成一个统计形状模型的基础上ATAA几何，我们将 在此基础上生成虚拟患者，其TAA进程在整个进展和失败过程中将 由特定目标2中的模型创建，参数由已发表的文献和我们的小鼠确定 具体目标1。然后，将使用真实的和虚拟的患者数据来训练机器学习工具， 将复合生物标志物与重塑结果联系起来并预测失败风险。该计划综合了 多个最近的进展，并补充他们的新想法，以产生一个计算机系统，能够 为ATAA做出有用的故障预测。

ABSTRACT Ascending thoracic aortic aneurysm (ATAA) is a major cardiovascular health problem characterized by a dilated aorta that may eventually dissect or rupture. ATAA presents a serious challenge in that the surgery is difficult and dangerous, so aneurysm repair criteria must balance the risk of a dissection and/or rupture with the risk of surgery. Current surgical guidelines are based on ATAA diameter or growth rate, but up to 60% of patients with an ATAA experience a dissection before surgical criteria are reached, hence there is a clear need for additional biomarkers of aneurysm failure. Possible biomarkers fall into broad categories including genetic, microstructural, geometrical, and biofluids, but it is challenging to obtain enough human data to calculate and correlate these biomarkers with critical outcomes such as failure. It is likely that a single biomarker is not sufficient, but composite biomarkers that are not intuitively obvious may be necessary for significant predictions of patient outcomes. In this proposal we will use a combination of models: 1) a mouse model of ATAA associated with Marfan Syndrome, 2) a multiscale, multiphysics model of ATAA growth and remodeling, and 3) virtual patient models derived from real patient imaging data, to determine composite biomarkers that may predict ATAA growth, progression, and failure. Our first Specific Aim is to use a genetic mouse model of ATAA associated with Marfan Syndrome to characterize aneurysm progression and failure in previously unachieved detail, quantifying aortic shape, tissue composition, tissue mechanical properties, and hemodynamics over time. This level of detail is not possible in human patients and is necessary to validate and test hypotheses on the growth and remodeling rules in our multiscale, multiphysics model in Specific Aim 2 and to provide an initial set of biomarkers to evaluate for our virtual patients in Specific Aim 3. Our second Specific Aim is to develop a novel multiscale, multiphysics computational model of ATAA growth and remodeling to produce results that will be compared to the mouse data in Specific Aim 1 and used to predict remodeling progression in real and virtual human patients in Specific Aim 3. In our third Specific Aim, we will use available human ATAA scans from Marfan Syndrome patients to generate a statistical shape model basis for the ATAA geometry, and we will use that basis to generate virtual patients, whose TAA course throughout progression and failure will be created by the model in Specific Aim 2, with parameters determined from published literature and our mouse data in Specific Aim 1. Both real and virtual patient data will then be used to train a machine learning tool to relate the composite biomarkers to the remodeling outcomes and predict failure risk. This plan synthesizes multiple recent advances and supplements them with new ideas to produce a computer system capable of making useful failure predictions for ATAA.