Fluorescence-guided multi-sampling and spatial multi-omics reveal functional lipid vulnerabilities in glioblastoma

Background: Glioblastoma (GB) is the most common and aggressive primary brain tumor in adults. Tumor recurrence occurs in most cases, typically within 2 cm of the resection margin, despite chemo and radiotherapy. This highly invasive nature and the marked heterogeneity of the tumor microenvironment make it challenging to treat. Fluorescence-guided multiple sampling of patient samples is being used to study intra-tumor heterogeneity. In this study we combined mass spectrometry imaging (MSI) with multiplex immunofluorescence (mIF) and histology to identify new therapeutic vulnerabilities based on spatial tumor biology. We developed a same slide workflow applying all 3 technologies to the same section to identify functional lipids that can be manipulated as less toxic therapeutic options. Methods: Twenty GB patients underwent craniotomies and tumor tissue was surgically resected from multiple different fluorescing and non-fluorescing brain locations. MSI was carried out using a Bruker Solarix FTICR mass spectrometer, the same section then underwent mIF for cell and organelle phenotyping, followed by H&E-staining for histopathology. The Leica Stellaris 5 and GT 450 slide scanner were used for confocal and histological image acquisition, respectively. Data analysis was carried out using the SciLs lab, Leica LAS X and Imaris software. Processed mIF, H&E and MSI data were integrated in SciLS Lab for multimodal analysis on the same section. Results: Integration of spatial lipidomic, proteomics and histopathology identified functional lipids that correlated with different pathological presentations and cellular neighborhoods within each sample. We focused on two multifunctional lipids: cardiolipins (CLs) and gangliosides (GGs). CLs are unique mitochondrial lipids that contain four acyl chains, whose structure is known to regulate mitochondrial function. In GB, we detected differential accumulation based upon fatty acid residue composition, which correlated with mitochondrial numbers and phenotype. Peritumoral regions showed interconnected mitochondria, while tumor regions with vascular proliferations displayed abundant fragmented mitochondria, indicative of altered metabolism and dynamics. GGs are involved in cell-cell interaction, immune modulation, and tumor growth. Anti-GD2 immunotherapy is FDA-approved for neuroblastoma and in clinical trials for other solid tumors. In GB, we identified over 30 GGs species across the GM1-3, GD1-3 and GT1,3 series, which segmented with different pathological regions based on acyl chain and sialic acid composition. Conclusions: We optimized a same section spatial multi-omics workflow that identified tumor-specific lipid pathways as therapeutic vulnerabilities in glioblastoma. Ongoing studies are using patient-matched 3D models to determine the functional relevance of these findings and for drug screening studies.