结合CT影像和呼出气体等多标识物的肺癌早期快速诊断及预测模型研究

The diagnosis of lung cancer at its early stage is a solution to reduce the mortality. Recently, National Lung Screening Trial (NLST) has shown that low-dose CT screening results in 20% mortality reduction in individuals at high LC risk. However, the diagnosis procedure after the abnormal CT still takes a longer time than it is expected, especially when the mass is small. Furthermore, invasive diagnosis procedures, such as biopsy and surgery, could bring an unnecessary harm to the person who did not have a lung cancer. And not all of the LC candidates can be diagnosed using these existing procedures. CT screening combined breath biomarkers is a novel method for a fast, non-invasive diagnosis of lung cancer at its early stage. We developed a model framework for using an existing carcinogenesis model, a model of the natural history of tumor growth and progression, a diagnosis model, and a survival model to predict the distribution of lung cancer survival. After simulating the survival distribution by using this model framework, we can estimate the diagnosis rate and survival rate of the population that were diagnosed through this novel method. This model framework could also be used to estimate the survival distribution of the population that were diagnosed through other detection methods, such as the biomarkers from the blood or from exhaled breath condensate (EBC).CT screening combined breath biomarkers and other potential biomarkers provides an in-time, non-invasive method for early LC diangosis.We expect this novel diagnosis model could help to improve the detection rate and potentially reduce the LC mortality.

如何更有效的发现早期肺癌病人是降低肺癌死亡率的关键。国际癌症机构(NLST)的最新研究结果表明使用螺旋CT对肺癌高危人群筛查可以降低20%的死亡率。然而从CT发现异常到活检确诊所花费的时间较长，并存在确诊时间不可控，诊断创伤较大及不能覆盖所有的影像学疑似病人等缺点。本研究结合CT筛查和呼吸标识物组合诊断的方法，构建新型的肺癌早期诊断模型。以医学仿真建模的方法，结合肺致癌模型，生长模型及生存率模型构建肺癌生存率仿真系统。从诊断率和生存率两方面研究该新型诊断模型在临床中的作用。所建立起的生存率仿真系统，进一步结合其它早期肺癌检测技术，如冷凝物或血液中生物标识物的检测技术，提供一种可预测其临床生存率的科学方法。本课题的创新之处在于把肺癌呼出气体及其冷凝物检测与CT影像学筛查相结合，构建新型的肺癌早期诊断模型，以实现对肺癌高危人群快速、实时、无损的诊断和预测，提高肺癌病人的整体生存率。

在全球范围内，肺癌已成为最常见的以及死亡率最高的癌症。目前仍然没有有效的早期诊断技术运用于肺癌的早期预防。筛查是一种很有前途的肺癌预防方法。美国National Lung Screening Trial (NLST)的对照研究表明通过应用低剂量计算机断层扫描（LDCT）技术筛查肺癌高危人群可有效降低该人群20%的死亡率。而呼出气体挥发性有机化合物（VOC）检测由于其非侵入性的特征，则是另一有前途的技术，可用于监视健康状况和疾病的进展。本项目验证了结合VOC标识物和CT成像的方法对于提高肺癌诊断率的可能性。基于此，我们建立了诊断模型并利用临床数据估计了模型参数。拟合后的模型结合致癌模型被用来重建肺癌病人在诊断时刻的年龄，性别，肿瘤的大小，和肺癌分期分布。比较从1988年到1999年SEER数据库中的独立数据，我们的模型精确地预测了性别分布和年龄分布，并合理地预测肿瘤的大小和肺癌分期分布。我们还构建了生存率仿真系统。它是结合了致癌模型，肺癌生长和转移模型，诊断模型和生存率模型的肺癌自然病程模型，用于了一个大型队列仿真研究中肺癌诊断效果的评估。仿真结果表明，这种改善的CT和VOC标识物组合的诊断模型将提高肺癌人群的生存率，特别是对5年和10年生存率的提高有统计学意义。生存率仿真系统这种模块化结构允许测试不同的肺癌诊断策略，这提供了一个可用于预测评估肺癌二级预防策略效果的平台。