METABOLITE BIOIMARKERS IN THE DEVELOPMENT OF ESOPHAGEAL ADENOCARCINOMA

描述（由申请人提供）：近几十年来，美国食管腺癌（EA）的发病率增加了300%，而中位生存期仍然只有10个月。已经为EA及其前体Barrett食管（BE）定义了几个主要的风险因素，包括反流，白人种族，男性，肥胖和吸烟，但关键的挑战仍然是识别最高风险的个体，因为大多数反流不会发展为BE，大多数BE不会进展为EA。虽然已经评估了许多基于遗传、表观遗传和蛋白质的生物标志物检测或预测疾病进展的能力，但由于所需测定的技术复杂性、标志物组的灵敏度/特异性有限以及缺乏验证研究，新型临床测试的开发受到阻碍。根据我们的初步数据，我们预计代谢组学，低分子量代谢物的定量，提供了一个有利的替代方法，用于确定有效的生物标志物，以帮助在风险分层的EA的风险。Numerou的研究已经确定，细胞能量代谢在广泛的癌症中发生了根本性的改变，越来越多的证据表明，与癌症相关的代谢变化在血液和尿液中是可检测的。在这个提议中，我们假设基于血清的代谢物可以区分疾病进展的多个阶段，沿着从胃食管反流（GERD）到BE到EA的途径。使用先进的分析方法和一个良好的生物储存库，这项合作提案旨在识别和验证与这些疾病状态相关的新代谢物标志物。在目标1中，将对分布在病例类型（GERD、BE、高度异型增生/EA）中的一组60份患者血清样本进行总体代谢分析，以确定在两种或所有这些疾病中存在显著差异的候选代谢物生物标志物的扩展池。在目标2中，将对240份患者样本的较大样本集进行这组代谢物加上先前确定的28份候选物的靶向分析。基于识别的经验证的候选标志物，将使用偏最小二乘判别分析构建样本分类的统计模型（BE-GERD和HGD/EA-BE），并通过Monte Carlo交叉验证估计模型性能。这些目标的成功完成将为产生新的非侵入性工具奠定基础，用于根据疾病进展的可能性对具有EA风险的患者进行分层。这些工具有可能通过集中对癌症风险最高的人进行密集的临床监测，增加早期发现并减少不必要的监测。这些研究将为未来更大规模的研究奠定基础，这些研究将利用国际巴雷特和食管腺癌（BEACON）联盟的广泛资源来验证我们在人群中的发现，并确定可能提高预测模型准确性和适用性的关键修正因素。

DESCRIPTION (provided by applicant): The incidence of esophageal adenocarcinoma (EA) in the US has increased 300% over recent decades while median survival remains an abysmal 10 months. Several major risk factors have been defined for EA and its precursor, Barrett's esophagus (BE), including reflux, Caucasian race, male gender, obesity, and smoking, but a key challenge remains the identification of individuals at highest risk, since most with reflux do not develop BE, and most with BE do not progress to EA. While a number of genetic, epigenetic, and protein-based biomarkers have been evaluated for their ability to detect or predict disease progression, the development of novel clinical tests has been hampered by technical complexities of the required assays, limited sensitivity/specificity of the marker panels and lack of validation studies. Based on our preliminary data, we anticipate that metabolomics, the quantification of low molecular weight metabolites, offers a favorable alternative approach for identifying effective biomarkers to aid in risk stratification for those at risk of EA. Numerou studies have established that cellular energy metabolism is fundamentally altered in a broad spectrum of cancers, and increasing evidence indicates that cancer-associated metabolic changes are detectable in blood and urine. In this proposal, we hypothesize that serum-based metabolites can distinguish multiple stages of disease progression along the pathway from gastroesophageal reflux (GERD) to BE to EA. Using advanced profiling methods and a well-characterized biorepository, this collaborative proposal seeks to identify and validate novel metabolite markers linked to these disease states. In Aim 1, global metabolic profiling will be conducted on a set of 60 patient serum samples distributed across case type (GERD, BE, high- grade dysplasia/EA) to identify an expanded pool of candidate metabolite biomarkers that differ significantly across two or all of these conditions. In Aim 2, this targeted profiling of this set f metabolites plus a previously-identified panel of 28 candidates will be conducted on a larger set of 240 patient samples. Based on validated candidate markers identified, statistical models for sample classification (BE-GERD and HGD/EA-BE) will be constructed using partial least-squares discriminant analysis, and model performance estimated by Monte Carlo cross-validation. Successful completion of these aims will lay the groundwork for generating novel non-invasive tools for stratifying patients at risk of EA according to their likelihood of disease progression. Such tools have the potential to increase early detection and reduce unnecessary surveillance, by focusing intensive clinical monitoring on those at highest risk of cancer. These investigations will set the stage for larger-scale future studies, which will leverage the extensiv resources of the international Barrett's and Esophageal Adenocarcinoma (BEACON) consortium to validate our findings across population groups and identify key modifying factors that may enhance the accuracy and applicability of our predictive models.