Improving genomic prediction models in breast cancer

描述(申请人提供)：乳腺癌是一种不同的疾病，具有不同的生物学特性和经常不可预测的临床结果。基因组技术已经显示出更好地区分相似临床病理阶段的患者，并在某些情况下预测临床结果的前景。我们和其他人已经发表了基于阵列的乳腺癌预后预测模型的研究，包括ER状态、淋巴状态和疾病复发。虽然这些初步研究为基于基因组的个性化结果预测带来了希望，但在基于表达阵列的预后可以纳入临床医学之前，必须回答一些重要的问题。例如，乳腺肿瘤基因表达的正常变异是否会影响基于表达的结果预测？我们最近的数据表明，从同一乳腺肿瘤中提取的相似肿瘤的复制样本在两个不同的时间显示出一致的全球基因表达模式。另一方面，不同细胞异质性的复制肿瘤标本或处于不同缺血时期的复制肿瘤标本在整体基因表达上也存在显著差异。重要的是，这种变异可以合理地预期发生在常规的临床条件下，在某些情况下给出了不同的结果预测模型。我们现在认为，了解基因表达的变异，并将关于这种变异的知识纳入基于基因组的模型中，可能会为提高这些模型的准确性提供有用的信息。在这项提案中，我们将检验这一假设，即确定乳腺癌基因表达的生物和过程变异性将提高基于基因表达的模型预测乳腺癌结果的能力，包括ER状态、腋窝淋巴结转移和疾病复发。具体来说，我们的目标是： 具体目的1：确定不同细胞含量的核心活检标本的基因表达数据在基于基因表达的乳腺癌ER状态、腋窝淋巴结转移和复发预测模型中是否具有相似的预测作用。 具体目的2：确定乳腺癌组织缺血时肿瘤基因表达的变化，并确定这些变化对乳腺癌ER状态、腋窝淋巴结状态和复发的基于基因表达的预测模型的影响。 具体目标3：测量绝经前和绝经后T1N0M0激素受体阳性乳腺癌中基因表达的变异性，并确定这种变异性是否改变了基于表达的乳腺癌ER状态、腋窝淋巴结状态和复发的预测模型。

DESCRIPTION (provided by applicant): Breast cancer is a heterogeneous disease with variable biology and often-unpredictable clinical outcomes. Genomic technologies have shown promise to better discriminate between patients of similar clinicopathologic stage and in some cases predict clinical outcomes. We and others have published studies of array-based predictive modeling of breast cancer outcomes, including ER status, lymph node status, and disease recurrence. While these initial studies raise hope for genomic-based personalized outcome predictions, significant questions must be answered before expression array-based prognostics can be incorporated into clinical medicine. For example, does normal variation in breast tumor gene expression affect expression-based outcome predictions? Our recent data suggest that replicate samples of similar tumor taken from the same breast tumor at two different times show consistent patterns of global gene expression. On the other hand, replicate tumor specimens of different cellular heterogeneity or those subjected to different periods of ischemia have significant differences in global gene expression as well. Importantly, this variation, which can reasonably be expected to occur under routine clinical conditions, gives different model predictions of outcome in some cases. We now believe that understanding variation in gene expression and incorporating knowledge about this variation in genomic-based models may provide useful information to improve the accuracy of these models. In this proposal, we will test this hypothesis that ascertainment of biologic and process variability in breast cancer gene expression will improve the ability of gene expression-based models to predict breast cancer outcomes, including ER status, metastasis to axillary lymph nodes and disease recurrence. Specifically, we aim to: Specific Aim 1: To determine whether gene expression data from core biopsy specimens of different cellular content give similar predictions in a gene expression-based predictive model of ER status, axillary lymph node metastasis and recurrence in breast cancer. Specific Aim 2: To identify changes in breast tumor gene expression with tissue ischemia and determine the impact of these changes on gene expression-based predictive models of ER status, axillary lymph node status and recurrence in breast cancer. Specific Aim 3: To measure variability in gene expression in T1N0M0 hormone receptor positive breast cancer in pre- and postmenopausal women and ascertain whether this variability changes expression-based model predictions of ER status, axillary lymph node status and recurrence in breast cancer.