PhD Student/ Junior Neurosurgeon Imperial College London and National Physical Laboratory London, United Kingdom
Introduction: The extent of tumor excision is positively correlated with outcomes; however current intra-operative margin detection methods are limited. Rapid Evaporative Ionization Mass Spectrometry (REIMS) coupled with electrosurgery known as Intelligent Knife (iKnife) has been shown to provide in-vivo tissue characterization with accuracy comparable to histopathology. Mass spectrometry imaging methods like Laser Desorption Imaging-REIMS (LDI-REIMS) provide the means to detect molecules in a spatially resolved manner providing tumor metabolism insights. This study aims to enhance margin detection by building spatially resolved diagnostic ex-vivo models with LDI-REIMS to be used for in-vivo iKnife diagnosis and combining it with sub-cellular resolution imaging utilizing high-speed Line-scan Confocal Laser Microscopy (LS-CLM) for morphological assessment of tissue micro-architecture.
Methods: Fresh frozen brain tumor samples were obtained from consented patients (n=30) and sectioned (10 μm). The LDI-REIMS set-up comprised of a Xevo mass spectrometer with an OPO laser at 2.9µm wavelength (50 μm x 50 μm pixel size). A high-speed LS-CLM system that provides rapid (up to 120 fps) histopathology was used for tissue morphological assessment. H&E staining was also conducted for image co-registration.
Results: Multivariate data analysis (PCA and LDA) as well as univariate analysis utilizing database matching was conducted. Preliminary results reveal distinct classification of tumors with up to 98% accuracy on 20% out cross-validation. As data analysis continues, a full group out cross-validation will be achieved. Additionally, insights into intra-tumor heterogeneity were gained from LDI-REIMS by observing the spatial distribution of metabolites involved in glycolysis. LS-CLM yielded distinct morphological features of different tumor phenotypes.
Conclusion : There is early indication that LDI-REIMS is an appropriate approach to spatially resolve intra-tumor heterogeneity ex-vivo and has the potential to inform in-vivo diagnostics by differentiating biochemical signatures of tumors while LS-CLM differentiates morphology. Next steps include using a spectral identification algorithm based on LDI-REIMS models to characterize novel in-vivo samples.
