乳腺癌ER状态、预后及他莫西芬耐药鉴别三联嵌套标志的识别

Many drug resistant predictive signatures for ER positive breast cancer patients treated with tamoxifen have been reported. However, there are still three main problems described as follows: (1)The Allred score of neoplastic cell positive staining percentage from immunohistochemistry technique can be affected by many factors and the cutoff of Allred score adopted to distinguish the ER status of breast cancer samples tested are still clouded of uncertainty. So it is necessary to identify objective molecular signatures for determining ER status; (2)As the high malignancy of tumor may result in poor prognosis of patients to a large extent, it is hard to distinguish whether the prognostic signatures have the ability to predict whether patients will benefit from drug therapy or not. (3) During the Cox proportional-hazard regression analysis, the risk scores of candidate signatures are vulnerable to variation of gene expression between individuals and batch effects of gene expression measuring. This results many prognosis molecular signatures reported are of low reproducibility.. Regarding these problems, in this subproject, we are aiming at: based on relative expression ordering of gene pairs within samples, (1) identifying robust molecular signatures to distinguish ER status of breast cancer samples. (2) identifying tumor prognosis signatures after re-distinguishing the ER status of ER+ breast cancer samples with surgical treatment only. (3) After identifying the samples with high recurrence risk from ER+ breast cancer samples treated with tamoxifen, we will identify drug resistance predictive signatures based on expression profiles of these samples. Finally, we apply the three predictive signatures to ER positive breast cancer samples obtained from TCGA database to identify tamoxifen-sensitive patients and tamoxifen-resistant patients , then we split the ER positive breast cancer samples obtained from TCGA database into tamoxifen-sensitive group and tamoxifen-resistant group. After analyzing the multi-omics datasets between the two groups, we can identify the resistant epigenomics signatures with high coverage and genomics abnormal signatures. Then we analyze Bayesian network derived from these multi-omics molecular signatures, which will provide candidate target signatures to reverse the resistance of ER positive breast cancers patients treated with tamoxifen.

已报道了许多ER+乳腺癌患者经他莫昔芬治疗的预后预测标志，但是存在如下三大问题：（1）根据半定量免疫组化技术检测ER的评分，判断ER状态存在较大偏差；（2）由于不良预后可能与药物无关，难以验证预后标志是否具有耐药预测能力；（3）目前基于风险得分的标志易受基因表达值个体间变异及检测的批次效应的影响，可重复性低。因此，本课题拟利用肿瘤样本内基因表达丰度相对大小关系，识别（1）可稳健鉴别乳腺癌样本ER状态的标志；（2）利用ER状态标志,对单纯手术治疗的ER+乳腺癌样本的ER状态重鉴定后，再识别预测术后复发风险预后标志；（3）利用预测术后复发风险标志，在接受他莫昔芬治疗患者的样本中，先识别出单纯手术后复发风险高的患者，在此基础上识别他莫昔耐药预测标志。最后，利用前述3组标志具有的跨数据集稳健性，在TCGA的ER+乳腺癌样本中鉴别耐药、敏感组，识别耐药组的多维组学高覆盖特征，寻找逆转耐药的候选靶标。

乳腺癌是女性中发病率最高的癌症，是主要的致死原因之一。其中约70%的患者呈雌激素受体阳性（ER+）。对乳腺癌样本ER状态的鉴别、早期ER+乳腺癌预后及他莫西芬耐药特征的研究具有重要意义。在前期预实验中，我们评估了ER阳性乳腺癌的术后复发风险，发现低复发风险组的乳腺患者经他莫西芬治疗后难获益且会产生不良的毒性反应，而高风险组患者则需要接受他莫西芬治疗以改善预后。本项目拟基于转录组定性特征，进一步开发了可客观鉴别乳腺癌组织样本ER状态的转录组标志，该标志可有效鉴别乳腺癌样本的ER状态，尤其是弱阳性/弱阴性样本；开发了可识别患者术后复发风险的转录组标志，并发现高风险组样本中的转录组不稳定性；开发了具有显著预后价值的可重新鉴别ER阳性乳腺癌组织学分级状态的转录组标志；开发了可整合不同实验室的数据集识别差异基因的算法RankComp，该算法可更有效的鉴别两种表型间的弱信号差异表达基因；评价细胞系对他莫西芬获得性耐药的乳腺癌组织表达谱的代表性，进一步识别他莫西芬获得性耐药相关的转录组标志。本项目对研究ER阳性乳腺癌患者的术后复发风险及经他莫西芬治疗后的耐药风险评估具有重要的理论研究和临床转化意义。