基于非染色衍射成像富集脑脊液肺癌细胞的新方法研究

脑转移癌可累及软脑膜、颅内硬脑膜和脑实质，症状复杂、进展迅速、早诊困难、死亡率高。脑脊液细胞学检查是诊断脑膜转移癌的金标准，其特异性高，灵敏度偏低，与肿瘤细胞在样本中的比例相关。本项目是以肺癌软脑膜转移为研究对象，探索、研究非染色的脑脊液肿瘤细胞富集方法，将富集后的样本用于经典脑脊液细胞学检测，以提高检出率。申请人在之前的研究中参与开发了一项基于衍射成像和机器学习技术的非染色流式细胞分析技术，已实现对多种细胞的识别和分类。本研究拟获取红细胞、白细胞、肺癌细胞株，和杂质的衍射图像，建立一个识别和富集肺癌细胞的模型。通过染色和显微镜观察分类后的样本进行方法评价和优化。最后，通过临床诊断或病理确诊为肺癌脑转移怀疑累及软脑膜的患者的脑脊液样本测试和优化模型。新方法以相干散射光携带的三维信息识别和富集肿瘤细胞，无需抗体标记，操作简单，成本低，易于在无分子病理检测基础的基层医院推广。

Brain metastasis is a fatal disease including leptomeningeal carcinomatosis (LMC), intracranial dural metastasis (IDM) and brain parenchymal metastasis (BPM). It is characterized by complex symptoms, rapid progress, and difficulty in early diagnosis. The testing result is positively related to the ratio of cancers in the sample. The cerebrospinal fluid (CSF) cytology is a gold standard in the diagnosis of LMC. It shows high specificity but has low sensitivity. The aim of this project is to study the feasibility of improving the positive rate of CSF cytology in lung cancer leptomeningeal metastases diagnoses by enriching cancer cells using a label-free method. The applicant has been participated in developing a cell analysis system based on diffraction imaging flow cytometry and machine learning techniques, which can successfully recognize and classify many different cell types and apoptotic cells. In this study, diffraction images of white cells、red cells、lung cancer cells and other impurities will be acquired and analyzed to build and train classification models. The microscope and cell stain technique will be applied to observe the cells after classification to verify the method. The CSF samples from the patient who has been clinically diagnosed or highly suspected as lung cancer leptomeningeal metastases will be used in method verification. This method identifies and enriches cells based on 3D information carried by coherent scatter light. It has advantages in antibody label-free, simple operating, no specification training, shot time for testing and low cost. It is applicable in primary local hospitals without molecular pathology detection capability, which is good for early diagnosis.