Short summary

Gliomas are brain tumors with diffuse growth, making surgical removal challenging. In a project with 48 patients, we used MALDI-MSI and machine learning to distinguish tumor from healthy and necrotic tissue at high resolution. As part of a new project, we now aim to include blood samples from glioma patients to explore lipid biomarkers as potential tools for monitoring tumor burden. The project and biobank are approved by the ethics committee

Rationale

1. Background Gliomas are rare, with an incidence ranging from 4.80 to 7.70 per 100,000 person-years (1, 2). Brain metastases (BM) are approximately 10 times more common than primary malignant brain tumors, occurring in 10% to 27% of all cancer-related deaths, making BM the most prevalent brain malignancy (3-5), with an increasing incidence (4). Both glioma and BM are devastating diseases that causes substantial risk of reduced lifespan and quality of life(6-10). Gliomas comprise a highly heterogeneous group of primary central nerous system (CNS) tumors, traditionally classified based on histological type and malignancy grade (WHO grading 1-4). Founded on the revised WHO classification from 2016 (updated in 2021) regarding tumors in the CNS, several of the diagnosis for CNS tumors is now based on an integrated diagnosis where both histological and molecular characteristics are used (11, 12). These molecular characteristics has demonstrated high prognostic value and is now used to optimize treatment approaches and predict patient outcomes (13, 14). Adult-type diffuse gliomas WHO grade 3 and 4, including glioblastomas(GBM), are the most common malignant tumors originating from the central nervous system. Infiltration of tumor cells over large distances into the CNS parenchyma and the resistance to treatment means, that adult-type diffuse gliomas (high grade) are widely regarded as uncurable. (11, 15-17). In contrast, brain metastases (BM) are typically considered well-demarcated; however, infiltrative growth patterns have been documented (18). Tumor resection is a ground pillar in treatment of both glioma and BM to relieve symptoms and prolong survival(19-21). In order to increase survival, the surgical goal is to safely resect as much tumor tissue as possible while preserving neurological functions (22). In addition, the response to radio- and chemotherapy with temozolomide for GBM patients is further improved when maximal resection is achieved(23). GBM typically grows in nodular tumor patterns on T1-contrast MRI, while tumor infiltration can be assessed with FLAIR sequences as a peritumoral hyperintense signal. Removing both contrast-enhancing tissue and surrounding FLAIR-hyperintense areas can be referred to as a supramaximal resection of GBM(24). Supramaximal/supramarginal resection of gliomas and brain metastases has shown to further improve survival rates(24-27). The explanation may lie in the removal of infiltrating cells, which can act as a seed for recurrence, since glioma and brain metastasis often recur locally(28, 29). However, tumor microinfiltration remains undetectable by current technologies such as magnetic resonance imaging (MRI), positron emission tomography (PET), and fluorescence imaging(30). Gangliosides are sialic acid-containing lipid species anchored by their hydrophobic lipid tails, primarily located in the outer cell membranes, where they usually cluster in lipid raft microdomains(31). The human brain contains 10-30 times more gangliosides compared to any other tissue in the body, with the majority being complex species(31). Studies have shown that GBMs contain five times less gangliosides and exhibit a shift towards simpler gangliosides normally involved in brain development(31). Among these, some, for instance, GD3, have been shown to promote gliomagenesis in GBM(31). Significant differences in ganglioside expression are observed in both primary brain tumors and brain metastases compared to normal brain tissue(32), making gangliosides an hypothetical species of molecules for differentiating non-tumor from tumor tissue, making detection of tumor microinfiltration possible. Matrix-Assisted Laser Desorption Ionization (MALDI) time-of-flight (TOF) mass spectrometry imaging (MSI) offers label-free identification of numerous molecules (including proteins, lipids, and metabolites) from a single tissue slice, facilitating the discovery of unknown structures, in contrast to antibody-based methods (33-35). The intensity of each m/z feature or molecule is represented as a spatial heatmap, a so-called ion image, displaying the spatial distribution of a molecule. MALDI-MSI is composed of an ionization source (MALDI), a mass analyzer (TOF), and a detector. Before MALDI-MSI analysis, the tissue section is coated with a matrix that extracts molecules, facilitating both desorption and ionization by the laser. Furthermore, as only the matrix is targeted by the laser, subsequent histological or immunohistochemical staining can be performed on the underlying tissue for further examination and comparison (36-38). In this pilot study, we test MALDI-MSI ability to discriminate tumor and brain tissue, in order to identify tumor infiltration, which for GBM(IDH-wildtype) has no specific immunostaining. The technology has before proven to be a powerful tool for objectively exploring tumor margins, including identifying residual disease traits in phenotypically normal cells surrounding the tumor border zone(39, 40), metabolite imaging of GBM (41-43) or delineating pituitary tumors and cutaneous squamous cell cancer (44, 45).

Description of the cohort

Patients with all intracranial tumors from department of neurosurgery, OUH.

Data and biological material

Blood and resected tissue samples

Collaborating researchers and departments

department of biochemistry and molecular biology university of southern denmark

• Ole Nørregaard Jensen