Modeling Neoplastic Progression in Barrett's Esophagus

描述(申请人提供)：与大多数癌前肿瘤一样，巴雷特食道(BE)的主要临床问题之一是预测哪些患者可能进展为癌症(即风险分层)。一旦我们确定了高危患者，我们就可以证明 预防癌症发展的干预措施。然而，肿瘤通过一个基本随机的体细胞进化过程发展为癌症，导致克隆在空间和时间上的异质性。最近的癌症基因组测序结果表明，许多不同的突变组合可以产生癌症。所有这些都使得开发可靠的软件变得困难 以及准确的突变面板，用于风险分层。作为进化生物学领域的领导者 癌症和基于试剂的肿瘤进展模型，通过获得来自世界上最好的BE预期队列的组织样本和数据，我们处于独特的地位，可以研究BE肿瘤进展的进化动力学，并将我们的结果转化为临床。我们的初步结果表明，通过测量体细胞进化的参数，而不是其产品，我们可以开发出适用于大多数肿瘤(如果不是所有肿瘤)的强大的风险分层生物标记物。具体地说，我们已经证明，作为自然选择的燃料，对遗传多样性的衡量可以预测BE的进展。我们还证明，我们可以在体内测量体细胞突变率，而且似乎可以预防BE癌症的非类固醇抗炎药与大多数患者的体细胞突变率降低一个数量级相关。这一新的进展动力学的初步观点揭示了对BE的风险分层和癌症预防非常重要的3个惊喜：(1)活检水平的突变率显然非常低：每个细胞谱系每年约有1个染色体损伤。目标1将测试隐窝内的突变率是否高于活组织检查中观察到的突变率，无论是在进展为癌症的患者中，还是在未进展为癌症的患者中。(2)绝大多数病变发生在第一次内窥镜检查之前，而第一次内窥镜检查往往伴随着质子泵抑制剂(PPI)等强力抑酸药物的开始。目标2将使用观察性交叉研究设计来测试PPI在活检和隐窝水平对突变率的影响。这将是首次对PPI进行癌症预防机制测试。(3)目前的理论认为，肿瘤进展是通过一系列(例如，~20)克隆扩张进行的。然而，在基于活检采样的156个病人年的监测中，我们只观察到了一次这样的扩大。目的3将开发从两个大的独立队列的BE细胞学刷子中以更精细的分辨率在单细胞中检测克隆扩张的方法，并测试克隆扩张的检测是否可以预测进展。我们还将使用这些队列作为独立的验证性研究，以测试细胞水平的遗传多样性是否预测进展。对于每个目标，我们使用计算模型在不同的假设下生成预期数据，并识别用于癌症预防的体细胞进化过程中最敏感的方面。

DESCRIPTION (provided by applicant): One of the main clinical problems in Barrett's esophagus (BE), like most pre-malignant neoplasms, is predicting which patients are likely to progress to cancer (i.e., risk stratification). Once we identify high risk patients, we can justify interventions to prevent development of cancer. However, neoplasms progress to cancer through a fundamentally stochastic process of somatic evolution leading to clonal heterogeneity in both space and time. Recent cancer genome sequencing results show that many different combinations of mutations can generate a cancer. All of this makes it difficult to develop reliable and accurate panels of mutations for risk stratification. As leaders in the evolutionary biology of cancer and agent-based modeling of neoplastic progression, with access to tissue samples and data from the world's best prospective cohorts of BE, we are in a unique position to study the evolutionary dynamics of neoplastic progression in BE and translate our results to the clinic. Our preliminary results show that by measuring the parameters of somatic evolution, rather than its products, we can develop robust biomarkers for risk stratification that may be universally applicable to most, if not all neoplasms. Specifically, we have shown that measures of genetic diversity, which is the fuel of natural selection, predict progression in BE. We also demonstrated that we can measure the somatic mutation rate, in vivo, and that non-steroidal anti-inflammatory drugs, which appear to prevent cancer in BE, were associated with an order of magnitude decrease in that mutation rate in most patients. This novel initial view of the dynamics of progression revealed 3 surprises that are important for risk stratification and cancer prevention in BE: (1) The mutation rate at the biopsy level is apparently very low: ~1 chromosomal lesion per year, per cell lineage. Aim 1 will test if the mutation rate within crypts is higher than can b observed in biopsies, both in patients that progressed to cancer, and those that did not. (2) The vast majority of lesions occur prior to the first endoscopy, which is often concomitant with the start of powerful acid suppressive medications such as proton-pump inhibitors (PPIs). Aim 2 will use an observational cross-over study design to test the effects of PPIs on mutation rate at the biopsy and crypt levels. This would be the first time that PPIs would be tested for a cancer prevention mechanism. (3) Current theory posits that neoplastic progression proceeds through a series of (e.g., ~20) clonal expansions. Yet, we only observed one such expansion in 156 patient years of surveillance based on biopsy sampling. Aim 3 will develop methods to detect clonal expansions in the much finer resolution available in single cells from cytology brushings of BE from two large, independent cohorts, and test if the detection of a clonal expansion predicts progression. We will also use these cohorts as an independent validation study to test if cell level genetic diversity predicts progression. For each aim, we use computational models to generate the expected data under alternative hypotheses and to identify the most sensitive aspects of the somatic evolutionary process for cancer prevention.