Livermore Laurent J, Isabelle Martin, Bell Ian M, Edgar Oliver, Voets Natalie L, Stacey Richard, Ansorge Olaf, Vallance Claire, Plaha Puneet
1Nuffield Department of Clinical Neurosciences, and.
3Department of Neurosurgery, Oxford University Hospitals NHS Foundation Trust, Oxford.
J Neurosurg. 2020 Oct 2;135(2):469-479. doi: 10.3171/2020.5.JNS20376. Print 2021 Aug 1.
Raman spectroscopy is a biophotonic tool that can be used to differentiate between different tissue types. It is nondestructive and no sample preparation is required. The aim of this study was to evaluate the ability of Raman spectroscopy to differentiate between glioma and normal brain when using fresh biopsy samples and, in the case of glioblastomas, to compare the performance of Raman spectroscopy to predict the presence or absence of tumor with that of 5-aminolevulinic acid (5-ALA)-induced fluorescence.
A principal component analysis (PCA)-fed linear discriminant analysis (LDA) machine learning predictive model was built using Raman spectra, acquired ex vivo, from fresh tissue samples of 62 patients with glioma and 11 glioma-free brain samples from individuals undergoing temporal lobectomy for epilepsy. This model was then used to classify Raman spectra from fresh biopsies from resection cavities after functional guided, supramaximal glioma resection. In cases of glioblastoma, 5-ALA-induced fluorescence at the resection cavity biopsy site was recorded, and this was compared with the Raman spectral model prediction for the presence of tumor.
The PCA-LDA predictive model demonstrated 0.96 sensitivity, 0.99 specificity, and 0.99 accuracy for differentiating tumor from normal brain. Twenty-three resection cavity biopsies were taken from 8 patients after supramaximal resection (6 glioblastomas, 2 oligodendrogliomas). Raman spectroscopy showed 1.00 sensitivity, 1.00 specificity, and 1.00 accuracy for predicting tumor versus normal brain in these samples. In the glioblastoma cases, where 5-ALA-induced fluorescence was used, the performance of Raman spectroscopy was significantly better than the predictive value of 5-ALA-induced fluorescence, which showed 0.07 sensitivity, 1.00 specificity, and 0.24 accuracy (p = 0.0009).
Raman spectroscopy can accurately classify fresh tissue samples into tumor versus normal brain and is superior to 5-ALA-induced fluorescence. Raman spectroscopy could become an important intraoperative tool used in conjunction with 5-ALA-induced fluorescence to guide extent of resection in glioma surgery.
拉曼光谱是一种生物光子学工具，可用于区分不同的组织类型。它是非破坏性的，无需样品制备。本研究的目的是评估拉曼光谱在使用新鲜活检样本时区分胶质瘤与正常脑组织的能力，并且在胶质母细胞瘤的情况下，比较拉曼光谱预测肿瘤有无的性能与5-氨基酮戊酸（5-ALA）诱导荧光的性能。
使用从62例胶质瘤患者的新鲜组织样本以及11例因癫痫接受颞叶切除术的无胶质瘤脑组织样本离体获取的拉曼光谱，构建了一种主成分分析（PCA）-线性判别分析（LDA）机器学习预测模型。然后使用该模型对功能引导下的超最大胶质瘤切除术后切除腔新鲜活检样本的拉曼光谱进行分类。在胶质母细胞瘤病例中，记录切除腔活检部位的5-ALA诱导荧光，并将其与拉曼光谱模型对肿瘤存在的预测进行比较。
PCA-LDA预测模型区分肿瘤与正常脑组织的灵敏度为0.96，特异度为0.99，准确度为0.99。超最大切除术后从8例患者（6例胶质母细胞瘤，2例少突胶质细胞瘤）获取了23份切除腔活检样本。拉曼光谱在预测这些样本中的肿瘤与正常脑组织方面显示出灵敏度为1.00，特异度为1.00，准确度为1.00。在使用5-ALA诱导荧光的胶质母细胞瘤病例中，拉曼光谱的性能明显优于5-ALA诱导荧光的预测值，后者的灵敏度为0.07，特异度为1.00，准确度为0.24（p = 0.0009）。
拉曼光谱可以准确地将新鲜组织样本分类为肿瘤组织与正常脑组织，并且优于5-ALA诱导荧光。拉曼光谱可能成为一种重要的术中工具，与5-ALA诱导荧光联合使用，以指导胶质瘤手术的切除范围
