Detection of subtype-specific breast cancer surface protein biomarkers via a novel transcriptomics approach.
Detection of subtype-specific breast cancer surface protein biomarkers via a novel transcriptomics approach.
复制标题
DOI：
10.1042/bsr20212218
复制
发表时间：
2021-12-22
影响因子：
4
通讯作者：
Giorgi FM
中科院分区：
文献类型：
--
作者：
Mercatelli D;Formaggio F;Caprini M;Holding A;Giorgi FM
背景：细胞表面蛋白已广泛用作癌症研究中的诊断和预后标志物以及抗癌药物开发的靶标。迄今为止，很少有尝试来表征乳腺癌患者的表面组，特别是与当前分子乳腺癌（BRCA）分类相关的尝试。鉴于此，我们开发了一种新的计算方法，从转录组数据推断细胞表面蛋白质活性，称为“SURFACER”。方法：使用 GTEx 的基因表达数据构建正常乳腺网络模型作为输入，以推断从 TCGA 检索的 BRCA 组织样本与正常样本相比细胞表面蛋白活性的差异。根据 PAM50 转录亚型（Luminal A、Luminal B、HER2 和 Basal）对数据进行分层，同时应用无监督聚类技术根据细胞表面蛋白活性定义 BRCA 亚型。结果：我们的方法确定了 213 个 PAM50 亚型特异性失调的表面基因，并定义了 5 种 BRCA 亚型，通过生存分析评估其预后价值，确定风险增加的细胞表面活性配置。通过评估 11 种不同机器学习分类算法的性能，测试了 SURFACER 方法在 BRCA 基因分型中的价值。结论：BRCA 患者可分为五个表面活性特异性组，有可能识别亚型特异性的可操作靶点，以设计定制的靶向治疗或用于诊断目的。 SURFACER 定义的亚型还显示出预后价值，可识别较高风险的表面活性特征。
Background: Cell-surface proteins have been widely used as diagnostic and prognostic markers in cancer research and as targets for the development of anticancer agents. So far, very few attempts have been made to characterize the surfaceome of patients with breast cancer, particularly in relation with the current molecular breast cancer (BRCA) classification. In this view, we developed a new computational method to infer cell-surface protein activities from transcriptomics data, termed ‘SURFACER’. Methods: Gene expression data from GTEx were used to build a normal breast network model as input to infer differential cell-surface proteins activity in BRCA tissue samples retrieved from TCGA versus normal samples. Data were stratified according to the PAM50 transcriptional subtypes (Luminal A, Luminal B, HER2 and Basal), while unsupervised clustering techniques were applied to define BRCA subtypes according to cell-surface proteins activity. Results: Our approach led to the identification of 213 PAM50 subtypes-specific deregulated surface genes and the definition of five BRCA subtypes, whose prognostic value was assessed by survival analysis, identifying a cell-surface activity configuration at increased risk. The value of the SURFACER method in BRCA genotyping was tested by evaluating the performance of 11 different machine learning classification algorithms. Conclusions: BRCA patients can be stratified into five surface activity-specific groups having the potential to identify subtype-specific actionable targets to design tailored targeted therapies or for diagnostic purposes. SURFACER-defined subtypes show also a prognostic value, identifying surface-activity profiles at higher risk.