Xin Le, Zheng Wei, Lin Kan, Lin Shulang, Huang Zhiwei
Optical Bioimaging Laboratory, Department of Biomedical Engineering, College of Design and Engineering, National University of Singapore, Singapore 117576, Singapore.
National University of Singapore (Suzhou) Research Institute, Suzhou, Jiangsu 215123, China.
Anal Chem. 2025 Apr 15;97(14):7897-7905. doi: 10.1021/acs.analchem.4c07042. Epub 2025 Apr 4.
Metabolic dysregulation is a critical feature of various cancers, including brain tumors. Studying metabolic changes in tumor cells and tissues significantly improves our understanding of tumor development, progression, and treatment response. In this study, we utilize hyperspectral stimulated Raman scattering (SRS) imaging combined with biochemical spectral modeling to identify unique histological and molecular signatures linked to metabolic diversity across different glioma grades, without the need for labeling. By employing rapid label-free SRS histopathology and multivariate curve resolution analysis, we uncover changes in lipid profiles and varying levels of neuron demyelination from low-grade (LG) to high-grade (HG) gliomas. Quantitative analysis of key metabolites using non-negative least-squares regression spectral modeling reveals a significant increase in cellular proteins, DNA, and cholesterol levels, alongside a reduced redox ratio (flavin adenine dinucleotide (FAD)/NADH) in the glioblastoma (GBM, grade IV) tissue compared to pilocytic astrocytoma (PA, grade I) and healthy brain tissues, indicating a shift toward a pro-malignant metabolic state. A neural network diagnostic classifier, trained on 4547 SRS spectra (healthy: 1263; LG: 815; HG: 2469) from 45 patients with PA and GBM, achieves 99.6% accuracy in detecting and grading brain tumors. This study highlights the potential of hyperspectral SRS imaging for rapid, label-free, and spatially resolved analysis of metabolic heterogeneity in human gliomas, paving the way for metabolome-targeted therapeutic strategies in precision brain tumor treatment.
代谢失调是包括脑肿瘤在内的各种癌症的关键特征。研究肿瘤细胞和组织中的代谢变化显著增进了我们对肿瘤发生、发展及治疗反应的理解。在本研究中，我们利用高光谱受激拉曼散射（SRS）成像结合生化光谱建模，来识别与不同级别胶质瘤代谢多样性相关的独特组织学和分子特征，而无需进行标记。通过采用快速无标记SRS组织病理学和多元曲线分辨分析，我们发现从低级别（LG）到高级别（HG）胶质瘤的脂质谱变化以及不同程度的神经元脱髓鞘。使用非负最小二乘回归光谱建模对关键代谢物进行定量分析显示，与毛细胞型星形细胞瘤（PA，I级）和健康脑组织相比，胶质母细胞瘤（GBM，IV级）组织中的细胞蛋白质、DNA和胆固醇水平显著增加，同时氧化还原比（黄素腺嘌呤二核苷酸（FAD）/NADH）降低，表明向促恶性代谢状态转变。一个基于45例PA和GBM患者的4547个SRS光谱（健康：1263个；LG：815个；HG：2469个）训练的神经网络诊断分类器，在检测和分级脑肿瘤方面的准确率达到99.6%。本研究突出了高光谱SRS成像在快速、无标记和空间分辨分析人类胶质瘤代谢异质性方面的潜力，为精准脑肿瘤治疗中以代谢组为靶点的治疗策略铺平了道路。
