Robust Clinical Translation of CT Imaging Biomarker in COPD for EBV Patient Selection

项目摘要 慢性阻塞性肺疾病（COPD），包括肺气肿等疾病，是第四种 美国的主要死因。虽然COPD的进展可以通过生活方式来减缓 变化，这种疾病是无法治愈的。最近出现了新的治疗/管理选择， 获批支气管内瓣膜（EBV）作为治疗重度COPD的突破性器械。EBVs 提供了一种新的治疗途径，但它们的植入与发病率和死亡率风险相关。在 此外，并非所有接受植入的患者都证明了该器械的预期受益。 试验数据表明，成像数据可用于帮助识别EBV有效的患者， 特别显示，在具有完全裂缝的异质性COPD患者中， 在目标肺叶中。EB病毒被正式用于“肺气肿的严重区域， 但对这些标准的解释是不精确和定性的。这个目标 一项建议是制定更统一、更有力和更量化的标准， 可能对EBV植入有反应的患者。确定的三个最具预测性的变量 到目前为止的临床试验一直是COPD的严重程度，相邻COPD之间的异质性， 叶和裂的完整性，所有这些都可以通过CT成像进行评估。然而，COPD的量化 成像中使用的CT采集和重建参数选择可能会混淆严重程度 议定书这些也会影响评估裂缝完整性的能力。此外，传统的 COPD严重程度（即肺气肿评分）不能区分肺破坏与空气潴留，这可能是 在评估患者是否是EBV的良好候选人方面很重要。在这份提案中，我们将调查 标准化CT成像数据的方法，这些方法将允许更稳健的定量图像分析，并应用于 由此产生的方法，以制定一个更强大的标准，为患者选择的应用EBV设备。 因此，我们的具体目标是：（1）发展稳健的，定量的CT评估肺的方法 密度和裂隙完整性评分（FIS）。这将涉及应用各种方法， CT扫描仪和采集/重建参数的差异以及机器学习方法， 更准确地评估肺裂的完整性;（2）开发和测试EBV预测模型， 改进措施。使用目标1的肺密度和FIS的改进估计，我们将开发和 测试区分EBV应答者与非应答者的预测模型;以及（3）为了延长EBV应答者的寿命， 预测模型纳入其他生物标志物。我们将扩展EBV评分模型， 研究其他生物标志物，可以提供补充信息，如区分空气滞留 和肺血管分布的评估。我们评估这个扩展模型的能力，以提高 我们区分EB病毒应答者和非应答者的能力。

PROJECT SUMMARY Chronic obstructive pulmonary disease (COPD), which includes conditions such as emphysema, is the fourth leading cause of death in the United States. Although progression of COPD can be slowed through lifestyle changes, the disease is incurable. New treatment/management options emerged recently when the FDA approved endobronchial valves (EBV) as a Breakthrough Device for the treatment of severe COPD. EBVs provide a new treatment pathway but their implantation is associated with morbidity and mortality risks. In addition, not every patient who undergoes implantation has demonstrated the expected benefits from the device. Trial data indicate that imaging data can be used to help identify patients in whom an EBV will be effective, showing specifically that effectiveness increases in patients with heterogeneous COPD with complete fissures in the targeted lobes. The EBV is officially indicated for “severe emphysema in regions of the lung that have little to no collateral ventilation,” but interpretation of these criteria is imprecise and qualitative. The goal of this proposal is to develop more unified, robust, and quantitative criteria that will more successfully identify patients that are likely to respond to EBV implantation. The three most predictive variables identified by clinical trials to date have been the degree of severity of COPD, the heterogeneity of COPD between adjacent lobes and fissure completeness, all of which can be assessed by CT imaging. However, quantification of COPD severity can be confounded by the CT acquisition and reconstruction parameter selections used in the imaging protocol. These can also affect the ability to assess fissure completeness. In addition, traditional measures of COPD severity (i.e. Emphysema score) do not distinguish lung destruction from airtrapping, which may be important in assessing whether a patient is a good candidate for an EBV. In this proposal, we will investigate methods to standardize CT imaging data that would allow for more robust quantitative image analysis and apply the resulting methods to develop a more robust criteria for patient selection for the application of EBV devices. Therefore, our Specific Aims are: (1) To develop methods for robust, quantitative CT estimates of lung density and the Fissure Integrity Score (FIS). This will involve the application of methods to account for differences in CT scanners and acquisition/reconstruction parameters as well as machine learning methods to more accurately assess lung fissure integrity; (2) To develop and test an EBV predictive model using improved measures. Using the improved estimates of lung density and FIS from aim 1, we will develop and test a predictive model that distinguishes EBV responders from non-responders; and (3) To extend the EBV Prediction Model to incorporate additional biomarkers. We will extend the EBV scoring model by investigating other biomarkers that may provide complementary information, such as distinguishing airtrapping from COPD and an assessment of lung vascularity. We evaluate the ability of this extended model to improve our ability to distinguish EBV responders from non-responders.