Novel Markers for Disease Outcome in Breast Cancer

在与贝勒医学院Sreekumar实验室正在进行的合作研究中，我们的小组一直在研究乳腺肿瘤的代谢组。对于发现，我们使用Metabolon公司的服务，通过非靶向质谱分析方法，测量了67例人类乳腺肿瘤和65例邻近非癌组织中352种已知代谢物和184种未知代谢物的丰度。在70例雌激素受体阴性肿瘤和36例邻近非癌组织中，通过基于多重反应监测的靶向检测，验证了几种关键代谢物的定量差异。我们目前的研究结果表明，乳腺肿瘤具有内在的代谢物特征，随着疾病的进展而持续存在。我们没有发现代谢物将肿瘤分为基因表达定义的亚型（管腔A和B，基底样，her2富集），这表明乳腺肿瘤代谢物组可能描述了该疾病的不同特征，正如其他人所建议的那样。由于我们的研究人群具有种族多样性，并且肿瘤ER状态和三阴性/基底样疾病的种族/种族匹配，因此我们分析了非裔美国患者（n = 32）与欧美患者（n = 35）的组织代谢物差异。所有患者都进行了血统分型，非洲裔美国患者的平均西非血统为83.2%，欧美患者的平均欧洲血统为97.6%。使用基于296种代谢物丰度的无监督分层聚类方法，雌激素受体（ER）阴性肿瘤和三阴性肿瘤被分成亚簇，这些亚簇不是随机代表两组患者。相反，在非裔美国人或欧美患者中出现了不同的集群，这在三阴性/基底样肿瘤类别中最为引人注目。值得注意的是，许多线粒体代谢物以及溶酶（磷酸）脂、肉毒碱和花生四烯酸在非裔美国患者的肿瘤中更为丰富，而组胺明显减少。我们的数据表明，非裔美国人和欧裔美国人的肿瘤可能存在代谢差异。然而，这些发现还有待证实。我们目前正在从各种来源收集更多的乳腺肿瘤进行独立验证。未来的研究将对两组患者之间的代谢物差异进行更深入的分析，包括对受这些差异影响的关键途径的描述，以及它们如何影响疾病结局和对治疗的反应。Sreekumar实验室也可以在裸鼠身上获得人类移植肿瘤。该系统将用于检查移植代谢组与治疗反应之间的关系，并允许实验性组织代谢组修饰来测试这些变化是否能改善治疗反应。我们的数据中另一个有趣的方面是2-羟戊二酸（2HG）水平在er阳性和er阴性肿瘤中升高，而在大多数邻近的非癌组织样本中可以忽略不计。2HG在胶质瘤和白血病中积累，细胞质或线粒体异柠檬酸脱氢酶（IDH1和IDH2）发生热点突变，导致酶产生2HG。乳腺肿瘤的测序未发现IDH1和IDH2突变的催化结构域存在突变，这与其他未能在人类乳腺肿瘤中发现这些突变的研究一致。由于异常积累的2HG改变了DNA甲基化，我们使用Illumina Human methylation 450 BeadChips检测了67个肿瘤的整体DNA甲基化模式。该分析导致发现了一种新的不良预后肿瘤亚型，具有高组织2HG浓度，独特的全基因组DNA甲基化特征和干细胞样转录特征。我们的数据将这种分子亚型描述为一类倾向于具有高2HG的肿瘤，其中包括不成比例的非裔美国患者。因此，非裔美国患者可能比欧美患者更频繁地发生甲基化定义的亚组III肿瘤，这与一项早期研究观察到这两组患者之间DNA甲基化差异一致。我们未来的研究将集中在2HG与甲基化定义的预后不良的乳腺癌亚型（亚组III）的发展之间的关系。在第二个项目中，我们将评估压力相关暴露对乳腺癌患者肿瘤生物学的影响，追求压力暴露、社会隔离和歧视与预后基因表达特征和肿瘤儿茶酚胺水平升高相关的假设。我们还假设这种压力诱发的特征在非裔美国人的乳腺肿瘤中比欧美患者更普遍，并且是儿茶酚胺信号的生物标志物，可用于选择患者进行干预治疗（β受体阻滞剂，压力管理）。巴尔的摩市的文化、种族/民族和社会人口的多样性以及来到马里兰大学的患者，使该地区成为评估歧视和压力对乳腺癌生物学影响的理想研究地点，正如我们的研究所提出的那样，并建立压力相关暴露与特定肿瘤特征发展之间的联系(例如，TAM浸润和微血管密度增加（存在明显的不良预后基因表达特征），非裔美国患者可能比欧美患者更频繁地受到影响。我们将分析手术切除癌变乳房的乳腺癌患者的肿瘤。这些患者将在手术前完成研究问卷（评估压力）。根据目前的设计，我们将首先分析40名非裔美国人和40名欧美乳腺癌患者的80个新鲜冷冻乳房肿瘤，这些患者患有TNM I期或II期疾病。我们的目标是纳入40例er阳性和40例er阴性肿瘤。这些肿瘤将从马里兰大学医学中心计划招募的多达150名乳腺癌患者中挑选出来。目前，我们已经完成了18例患者的招募。我们将从病理和医疗记录中获取临床病理信息，包括肿瘤受体状态（ER、PR、HER2/neu）和通过国家死亡指数（National Death Index）进行生存随访。我们将测定肾上腺素和去甲肾上腺素浓度，并对肿瘤和邻近正常组织进行基因表达谱和免疫组织化学分析。我们将根据感知到的压力、社会隔离和歧视对患者进行分层，将识别与压力相关暴露和肿瘤儿茶酚胺水平相关的基因特征，并利用许多公开可用的乳腺癌数据集评估这些基因特征作为患者生存和/或治疗反应的预测因子。我们还将研究应激相关暴露与选定蛋白质标志物（巨噬细胞浸润[CD68作为标记]和肿瘤微血管密度[CD31作为标记]）的关系。这是我们设计的优势，我们可以使用肿瘤儿茶酚胺水平的测量作为自我报告压力暴露的验证工具。我们的临床研究将得到实验室研究的支持，实验室研究评估去甲肾上腺素对人类乳腺癌细胞的细胞表型、基因表达和DNA甲基化模式的影响。如果我们发现去甲肾上腺素可以塑造DNA甲基化模式，我们将分析乳腺肿瘤中DNA甲基化与压力暴露之间的关系。

In ongoing collaborative studies with the Sreekumar laboratory at the Baylor College of Medicine, our group has been investigating the metabolome of breast tumors. For discovery, we used the services of the company, Metabolon, to measure the abundance of 352 known and 184 unknown metabolites in 67 human breast tumors and 65 adjacent non-cancerous tissues with an untargeted mass spectrometry-based profiling approach. Quantitative differences for several key metabolites were validated in 70 estrogen receptor-negative tumors and 36 adjacent non-cancerous tissues with Multiple Reaction Monitoring-based targeted assays by the Sreekumar laboratory (validation set). Our current findings indicate that breast tumors have intrinsic metabolite signatures that persist with disease progression. We did not find that metabolites classify tumors into the gene expression-defined subtypes (luminal A & B, basal-like, HER2-enriched), suggesting that the breast tumor metabolome may describe different traits of the disease, as was suggested by others. Because our study population was ethnically diverse and was race/ethnicity-matched for tumor ER status and triple-negative/basal-like disease, we analyzed tissue metabolite differences comparing the African-American patients (n = 32) with European-American patients (n = 35). All patients were ancestry-typed with an average West African ancestry of 83.2% among the African-American patients and an average European ancestry of 97.6% among the European-American patients. Using unsupervised hierarchical clustering based on the abundance of 296 measured metabolites, both estrogen receptor (ER)-negative tumors and triple-negative tumors separated into subclusters that did not randomly represent the two patient groups. Instead, distinct clusters emerged that were enriched for either African-American or European-American patients, which was most striking for the triple-negative/basal-like tumor category. Notable, many mitochondrial metabolites, and also lyso(phospho)lipids, carnitines, and arachidonic acid, were more abundant in tumors from African-American patients while histamine was markedly decreased in them. Our data indicate that metabolic differences may exist between tumors from African-American and European-American patients. However, these findings need confirmation. We are currently collecting additional breast tumors from various sources for independent validation. Future research will perform a more in-depth analysis of the metabolite differences between the two patient groups, including a description of the key pathways that are affected by these differences, and how they may influence disease outcome and response to therapy. The Sreekumar laboratory has also access to human transplant tumors in nude mice. This system will be used to examine the relationship between the transplant metabolome and response to therapy and allows for experimental tissue metabolome modifications to test whether the changes improve the therapy response. Another interesting facet in our data was the elevated level of 2-hydroxyglutarate (2HG) in a subset of ER-positive and ER-negative tumors, with negligible levels in most adjacent noncancerous tissue samples. 2HG accumulates in gliomas and leukemias with hotspot mutations in either the cytosolic or mitochondrial isocitrate dehydrogenase (IDH1 and IDH2), which cause the enzyme to produce 2HG. Sequencing of the breast tumors did not reveal the presence of mutations in the catalytic domains of IDH1 and IDH2 mutations, consistent with other studies that failed to find these mutations in human breast tumors. Because aberrantly accumulated 2HG alters DNA methylation, we examined the global DNA methylation pattern in the 67 tumors using the Illumina Human Methylation 450 BeadChips. The analysis led to the discovery of a novel poor outcome tumor subtype with a high tissue 2HG concentration, a distinct genome-wide DNA methylation signature, and a stem cell-like transcriptional signature. Our data describe this molecular subtype as a class of tumors that tends to have high 2HG and includes a disproportionately high number of African-American patients. Thus, African-American patients may develop more frequently a methylation-defined subgroup III tumor than European-American patients, consistent with one earlier study that observed DNA methylation differences between these two patient groups. Our future research will concentrate on the relationship between 2HG and the development of the methylation-defined breast cancer subtype with poor outcome (subgroup III). In a second project, we will assess the influence of stress-related exposures on tumor biology in breast cancer patients, pursuing the hypothesis that exposures to stress, social isolation, and discrimination are associated with a prognostic gene expression signature and increased tumor catecholamine levels. We also hypothesize that this stress-induced signature is more prevalent in breast tumors of African-American than European-American patients and is a biomarker for catecholamine signaling that can be used to select patients for intervention therapy (beta-blocker, stress management). The cultural, race/ethnic, and sociodemographic diversity of Baltimore city and of the patients coming to the University of Maryland make the area an ideal study location to evaluate the influence of discrimination and stress on breast cancer biology, as proposed by our study, and to establish a link between stress-related exposures and the development of specific tumor characteristics (e.g., increased TAM infiltration and microvessel density in presence of a distinct poor outcome gene expression signature) that may affect African-American patients more frequently than European-American patients. We will analyze tumors from breast cancer patients who had surgical resection of their cancerous breast. These patients will complete the study questionnaire (evaluating stress) prior to the surgery. Per current design, we will initially analyze 80 fresh-frozen breast tumors from 40 African-American and 40 European-American breast cancer patients who have TNM stage I or stage II disease. We aim to include 40 ER-positive and 40 ER-negative tumors. The tumors will be selected from a projected patient pool of up to 150 breast cancer patients that are being recruited into the study at the University of Maryland Medical Center. Currently, we have completed recruitment of 18 patients. We will obtain clinicopathological information from pathology and medical records including tumor receptor status (ER, PR, HER2/neu) and survival follow up through the National Death Index. We will determine adrenaline and noradrenaline concentrations and perform gene expression profiling and immunohistochemical analysis of tumors and adjacent normal tissue. We will stratify patients by perceived stress, social isolation, and discrimination, will identify gene signatures associated with stress-related exposures and tumor catecholamine levels, and evaluate these gene signature(s) as predictor of patient survival and/or response to therapy using the many publically available datasets for breast cancer. We will also examine the relationship of stress-related exposures with selected protein markers (macrophage infiltration [CD68 as marker] and tumor microvessel density [CD31 as marker]). It is the strength of our design that we can use the measurement of tumor catecholamine levels as a validation tool for self-reported stress exposures. Our clinical studies will be supported by laboratory research that evaluates the influence of noradrenaline on cell phenotypes, gene expression, and DNA methylation patterns in human breast cancer cells. If we find that noradrenaline can shape DNA methylation patterns, we will analyze the relationship between DNA methylation and stress exposures in the breast tumors.