量子点标记分子探针影像技术研究乳腺癌侵袭行为

Quantum dots (QDs) conjugated molecular probes for the real-time, in situ, and co-location molecular imaging study and multi-spectral analysis technology have promising advantages in cancer research, especially in the field of cancer invasion and metastasis. Based on this technology, this project aims at accurate, sensitive and quantitative detection of key molecules in both cancer cells and their micro-environment related to the invasive behavior of infiltrative ductal carcinoma of the breast, and together with multi-spectral analyses and new algorithm, in order to establish a comprehensive platform to evaluate the invasive potential of breast cancer. Firstly, novel quantum dots with smaller particle size and better biocompatibility will be fabricated and modified. The quantum dots based molecular probes targeting key molecular events in breast cancer invasion are applied to cancer models and specimens so as to establish in-situ and multi-spectral imaging platform, which in combination with computer-based visual analysis, artificial intelligence, and logical algorithm, will set up a in-situ molecular imaging based, multi-spectral analysis technology. Secondly, this technology will be applied to breast cancer cell lines, animal models and clinical specimens of infiltrative ductal carcinoma of the breast, to delineate the spacial and temporal expression features of key molecules in both cancer cells and their micro-environment, related to cell proliferation, matrix degradation, inflammatory cells infiltration, cancer cell adhesion and migration, and tumor angiogenesis. A paranoma picture will be generated to simultaneously show these major molecular events in action, revealing the comlex inter- and intra-molecule interactions in breast cancer microenvironment during the invasive growth of cancer. Thirdly, with the help of high throughput bioinformatics study, general rules will be crystalized behind these seemingly haphazard changes, which will be finalized after trial and error test, could help predict breast cancer invasive potential and formulate individualized treatment strategy.

量子点标记分子探针技术和以此为基础的原位、实时、共定位分子成像及多光谱分析技术在癌症研究，尤其是癌侵袭转移领域具有明显优势。本项目以此技术为基础，准确、灵敏、定量检测分析乳腺癌局部侵袭进展过程中癌细胞和间质微环境中的关键分子，综合评估乳腺癌的侵袭潜能。通过建立新型量子点合成和表征技术，整合量子点标记分子探针技术，针对乳腺癌局部侵袭行为的关键分子事件，创建原位多光谱成像技术平台；整合计算机视觉分析、人工智能和逻辑算法，建立肿瘤多种分子影像学信息的分析技术体系；通过对不同侵袭行为的乳腺癌细胞模型、动物模型和临床标本研究，原位显示癌细胞内、细胞间和间质中与细胞增生、间质降解、炎细胞浸润、粘附运动、新生血管形成相关的多种关键分子的表达及空间分布特征，展现乳腺癌侵袭关键分子事件的全景图，探索乳腺癌局部侵袭过程中癌细胞与间质微环境中关键分子相互作用的规律，预测乳腺癌的侵袭行为，为个体化治疗提供依据。

本项目拟采用以量子点标记分子探针技术为主的多模式原位、实时、共定位分子成像及多光谱分析技术，研究乳腺癌侵袭力相关指标体系，旨在发展一套预测乳腺癌侵袭风险的新模型。项目首先建立了乳腺癌临床病理资料大数据库，对3844例乳腺癌手术标本进行大数据分析，总结出淋巴结阴性、浸润性导管癌为重点研究对象；然后发展了针对常规HE染色、免疫组化染色、量子点多色荧光染色和传统荧光素染色的多光谱成像和分子信息解析技术，解析出乳腺癌癌巢、癌细胞、癌细胞核及癌细胞关键分子HER2、Ki67的多维特征，形成多个分子靶标、各类细胞组分的定性、定量及空间形态特征的算法；利用这些计算机算法，提取乳腺癌病理图像中的各类数理参数和形态学特征，结合患者预后，筛选出基于乳腺癌图像信息水平的独立预后因子，在免疫组织化学图像上初步建立了乳腺癌侵袭预后模型，并在苏木素-伊红染色图像上对建立的乳腺癌侵袭预后模型进行进一步优化，形成基于癌巢特征和癌细胞核特征的综合预测模型。在分子水平上建立了量子点单分子/多分子成像技术，定性、定量、定位研究乳腺癌中Ki67、HER2、细胞角蛋白、IV型胶原等分子与乳腺癌患者预后之间的关系，评估上述分子量化表达、时空表达、交互表达与乳腺癌预后的关系。将上述组织水平、细胞水平和分子水平的预后预测指标进行逻辑整合，形成包括肿瘤大小、组织学特征、细胞分布特征、分子表达特征的乳腺癌侵袭力评估体系，最后得出基于肿瘤原位信息的乳腺癌侵袭力数学模型。小样本临床验证显示，该模型对判断乳腺癌侵袭行为的预测效能，优于现有的Nottingham预后分级模型。在研究过程中，课题组还深化了肿瘤侵袭的理论认识，创建了肿瘤侵袭模型的“脉冲模式”，为减慢肿瘤侵袭进展提供了理论支持。本项目所建立的量子点标记分子探针技术、多光谱成像技术、图像分割、多维信息提取技术、神经网络支持向量算法，对于发展形成人工智能深度学习技术，奠定了基础。