PO.CL01.23 · 临床研究
ROR1和PMEPA1作为三阴性乳腺癌转移的联合预测生物标志物：多组学与临床验证
ROR1 and PMEPA1 as combined predictive biomarkers for metastasis in triple-negative breast cancer: Multi-omics and clinical validation
该海报暂无可下载的资料
AACR 官方页面
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
背景：三阴性乳腺癌（TNBC）仍是乳腺癌中最具侵袭性的亚型，其特点是高度异质性、早期转移和缺乏有效的靶向治疗。早期识别高危患者对改善预后至关重要。此前，我们鉴定出受体酪氨酸激酶样孤儿受体1（ROR1）是一种TNBC特异性的细胞外囊泡（EV）标志物，能够捕获肿瘤来源的EV（2025 AACR #摘要2005）。在此基础上，我们开展了一项整合的多组学和临床验证研究，以发现能够增强TNBC转移潜能预测的新型ROR1相关分子。
方法：使用EV蛋白质组学确认TNBC特异性标志物。分析了scRNA-seq（GSE176078）和TCGA-BRCA批量RNA-seq数据。hdWGCNA鉴定了上皮-间充质转化（EMT）高侵袭性TNBC细胞中的枢纽基因。应用LASSO、支持向量机（SVM）和随机森林算法筛选EMT相关的转移基因。TCGA数据按7:3拆分为训练集和验证集，采用带欠采样的AutoGluon机器学习构建转移预测模型。为进行临床验证，将TNBC患者按转移状态分类，对TNBC肿瘤组织进行PMEPA1和ROR1的免疫荧光染色，并用CKpan区分肿瘤实质与基质。
结果：蛋白质组学和计算分析将PMEPA1鉴定为关键的转移相关基因，在WeightedEnsemble-L2模型中显示出最高的预测性能（AUC）。在临床样本中，后来发生远处转移的患者表现出显著高于无转移患者的PMEPA1和ROR1表达。共定位分析显示两种蛋白均在CKpan+肿瘤区域富集。与单一标志物相比，ROR1和PMEPA1的组合提高了TNBC转移的预测性能，这与生物信息学预测一致。这些发现提示ROR1和PMEPA1在TNBC进展中存在协同参与。
结论：PMEPA1被鉴定为一种参与TNBC转移的新型ROR1相关分子。PMEPA1和ROR1的共表达是转移进展的稳健预测指标。将基于机器学习的转录组建模与临床验证相整合，为TNBC早期转移风险分层提供了一个有前景的框架。正在进行的研究旨在阐明ROR1-PMEPA1信号轴作为潜在治疗靶点的作用。
研究资助：国家自然科学基金（No. 82202603）。
查看英文原文 English abstract
Background: Triple-negative breast cancer (TNBC) remains the most aggressive subtype of breast cancer, characterized by high heterogeneity, early metastasis, and lack of effective targeted therapies. Early identification of high-risk patients is crucial for improving outcomes. Previously, we identified receptor tyrosine kinase-like orphan receptor 1 (ROR1) as a TNBC-specific extracellular vesicle (EV) marker capable of capturing tumor-derived EVs (2025 AACR #Abstract 2005). Building on this foundation, we conducted an integrated multi-omics and clinical validation study to uncover novel ROR1-associated molecules that could enhance the prediction of metastatic potential in TNBC.
Methods: EV proteomics were used to confirm TNBC-specific markers. scRNA-seq (GSE176078) and TCGA-BRCA bulk RNA-seq data were analyzed. hdWGCNA identified hub genes in epithelial-mesenchymal transition (EMT)-high invasive TNBC cells. LASSO, support vector machine (SVM), and random forest algorithms were applied to screen EMT-related metastasis genes. TCGA data were split (7:3) into training and validation sets, and AutoGluon machine learning with undersampling constructed the metastasis prediction model. For clinical validation, TNBC patients were categorized by metastatic status, immunofluorescence staining of PMEPA1 and ROR1 was performed on TNBC tumor tissues, with CKpan distinguishing tumor parenchyma from stroma.
Results: Proteomic and computational analyses identified PMEPA1 as a key metastasis-associated gene, showing the highest predictive performance (AUC) in the WeightedEnsemble-L2 model. In clinical samples, patients who later developed distant metastases exhibited significantly higher PMEPA1 and ROR1 expression than those who remained metastasis-free. Co-localization analysis showed both proteins enriched in CKpan + tumor regions. The combination of ROR1 and PMEPA1 improved predictive performance for TNBC metastasis compared to either marker alone, consistent with bioinformatics predictions. These findings suggest cooperative involvement of ROR1 and PMEPA1 in TNBC progression.
Conclusions: PMEPA1 is identified as a novel ROR1-associated molecule involved in TNBC metastasis. Co-expression of PMEPA1 and ROR1 serves as a robust predictor for metastatic progression. Integrating machine learning-based transcriptomic modeling with clinical validation provides a promising framework for early metastasis risk stratification in TNBC. Ongoing studies aim to elucidate the ROR1-PMEPA1 signaling axis as a potential therapeutic target.
Research Sponsor: National Natural Science Foundation of China (No. 82202603).
利益披露 Disclosure
H. Wang, None..
L. Xu, None..
S. An, None..
Y. Liu, None.