Metabolomics and Transcriptomics of Low- and High-Grade Gliomas

Diffuse gliomas, both lower-grade gliomas (LGG, WHO grade 2) and higher-grade gliomas (HGG, including grades 3 and 4 and glioblastomas (GBM, WHO grade 4), are highly infiltrative brain tumors that arise from glial progenitor cells and include astrocytomas, oligodendrogliomas, and GBMs. In the latest WHO 2021 classification of brain tumors, they are divided into three diagnostic and prognostic subtypes: i) isocitrate dehydrogenase 1 and 2 (IDH1/2) mutated and 1p/19q co-deleted (i.e., oligodendrogliomas), ii) IDH1/2 mutated and 1p/19q non-co-deleted (i.e., astrocytomas), and iii) IDH1/2 wildtype astrocytomas, including GBMs.

In current clinical practice, only IDH-mutated lesions are considered true LGG. Furthermore, some LGG can progress to WHO grade 4 (grade 4 astrocytomas) within a few months, while others remain stable and indolent for years, with an overall survival ranging from 1 to 15 years.

Therefore, being able to efficiently predict the clinical trajectories of LGG subgroups and personalize the care and treatment of these patients represents a significant unmet clinical need that requires accurate and robust patient stratification.

Similarly, recent studies have demonstrated the presence of molecular subgroups of GBM (transcriptional and metabolic in particular) that present different alterations in important pathways that can potentially be targeted pharmacologically, increasing the still too limited arsenal in the therapeutic management of these patients who still maintain an invariably poor prognosis.

Additionally, studies on gliomas rely on cellular cultures and animal models. These techniques, however, are effective for GBMs but are still severely unsuitable for generating LGG models, which nowadays are still lacking, despite being necessary to enable proper development of new medications.

The primary endpoint of this study is the global quantitative assessment of endogenous metabolites and transcripts in samples from patients with GBM and LGG, and the subsequent identification of integrated metabolic-transcriptional subtypes with clinically relevant prognostic and therapeutic implications. Since this is a novel exploratory study, for which it is not possible to define an a priori effect for the primary outcome, based on the prevalence of cancer patients attending the Oncological Neurosurgery Unit of the IRCCS Galeazzi Sant'Ambrogio Hospital (OGSA), the number of GBM and LGG patients to be enrolled is approximately 90 and 130, respectively. With a sample size of 90 GBM patients and 130 LGG patients, enrolled for study feasibility purposes, using a two-sided two-sample equal-variance t-test, assuming a very low significance level (0.0005) and hypothesizing a much higher number of comparisons between the differentially expressed metabolites and transcripts between the two patient groups, with an effect size of 0.6, it is possible to have a test power of 0.8.

To achieve the primary objective, samples collected in the operating room for metabolomic analyses will undergo untargeted enrichment for polar (amino acids, nucleotides, and sugars) and non-polar (fatty acids and membrane lipids) molecules and profile the levels of each metabolite. Significant metabolites for each experimental comparison (i.e., intra-sample and inter-sample comparison) (p \< 0.05 from the t-test performed in MarkerView) will be used to perform principal component analysis (PCA). The enriched metabolites and pathways resulting from each comparison (supervised and unsupervised) will be classified based on their False Discovery Rate (FDR) values. The metabolic signatures and independent clusters obtained will be combined with clinical, radiological, and histological data. The expression of genes involved in the identified metabolic pathways will be tested by quantitative polymerase chain reaction (qPCR), Western blot (WB), or immunohistochemistry (IHC) on the original human samples (frozen or paraffin-embedded) or on their derived models (cell lines/animal xenografts, if available).

Samples collected from the operating room for transcriptomic analysis will be undergo cDNA library generation using SMART technology and subsequent NGS sequencing with alignment to the GRCh38 human reference genome using STAR. Single gene copy counting will be performed using the DESeq2 pipeline (differential analysis of gene expression based on negative binomial distribution). The cutoff value imposed for differential gene expression will be the one suggested by the SEquencing Quality Control (SEQC) consortium, which defines a gene as differentially expressed when it has an associated raw P value less than 0.01 and, at the same time, the absolute value of its differential gene expression log2FC is greater than 1 (i.e., log2FC > 1 or log2FC \< -1). Unsupervised clustering analyses and studies of differentially expressed genes among these clusters will then be performed, correlated with clinical-radiological and histological data, to identify genes with a potential prognostic and therapeutic role. Supervised comparisons defined a priori based on clinical-radiological characteristics deemed relevant from a diagnostic, prognostic, and/or therapeutic standpoint will also be performed. Follow-up data will be collected at regular clinical follow-up and correlated with information derived from transcriptomic and metabolic analyses and used for clustering patients in possibly new molecular subgroups of clinical relevance.

Enzymes and/or other proteins characterizing rate-limiting pathway steps identified as significant by the integration of transcriptomic and metabolomic analyses will be identified as potential therapeutic targets. These proteins may be investigated in future studies through RNA-mediated expression interference (RNA knockdown/RNA interference) or pharmacological/pharmacometabolic inhibition to evaluate their potential therapeutic efficacy in preclinical models.

To achieve the Secondary aim (implementing preclinical models of LGGs and GBMs), the following procedure will be performed:

- Since we are interested in maintaining intratumoral heterogeneity of gliomas, the collected surgical samples will be immediately frozen in a special culture medium (fetal bovine serum, FBS, + 10% dimethyl sulfoxide, DMSO) with a gradual reduction in temperature using dedicated containers (Mr. Frosty). They will be stored at -80°C. The samples will then be thawed and enzymatically digested into single cells and cultured under the neurosphere assay conditions in an incubator under moderate hypoxia (5% O2). The generated neurospheres will then be collected, centrifuged, mechanically dissociated, and then seeded under the same conditions, allowing the culture to be propagated until the line is exhausted in the first 10-15 passages or a stable primary line is isolated. If this method is unsuccessful, in the presence of leftover material, individual cells will be plated on laminin/Matrigel in the same culture medium, or orthotopically implanted into the striatum of immunocompromised mice for in vivo propagation (IACUC already approved and in use for other studies).

Since most metabolites are taken up by cells through passive diffusion and/or transport mediated by carriers, any critical metabolites identified by metabolic analyses performed for the primary objective will be added to the culture medium to increase the efficiency of isolation of GBM and LGG lines, maintain heterogeneity, or select a specific subset. We intend to adapt the protocol for isolating primary cell lines from glial tumors to expand the available modeling capabilities.

In case of transplants in immunocompromised mice, at least four technical replicates will be used to minimize the number of animals used. In the event of periprocedural or postoperative death of one of the animals, at least three technical replicates will be possible to evaluate the growth and invasion characteristics of the cell lines into nervous structures. In the event of transplantation, the mice will be anesthetized with sevoflurane and fixed to a special stereotaxic device, followed by the injection of cell suspension via microsyringe, after a skin incision and micro-drilled hole. The animals will be monitored for good tissue perfusion and the absence of respiratory distress throughout the procedure (approximately 10 minutes in total, with an injection of 4-5 μL of cell suspension at 105 cells/μL, at 0.5 μL/min). At the end, the skin will be sutured with 5-0 non-absorbable suture. The mice will be clinically monitored for the development of neurological symptoms following transplantation and sacrificed with an overdose of Avertin® before reaching clinically poor conditions, in accordance with international guidelines for animal experimentation. Upon sacrifice, the brain will be removed for both histological analysis and possible in vitro culture.