Bevacizumab, an antibody against vascular endothelial growth factor (VEGF), is a promising, yet controversial, drug in human glioblastoma treatment (GBM). Its effects on tumor burden, recurrence, and vascular physiology are unclear. We therefore determined the tumor response to bevacizumab at the phenotypic, physiological, and molecular level in a clinically relevant intracranial GBM xenograft model derived from patient tumor spheroids. Using anatomical and physiological magnetic resonance imaging (MRI), we show that bevacizumab causes a strong decrease in contrast enhancement while having only a marginal effect on tumor growth. Interestingly, dynamic contrast-enhanced MRI revealed a significant reduction of the vascular supply, as evidenced by a decrease in intratumoral blood flow and volume and, at the morphological level, by a strong reduction of large- and medium-sized blood vessels. Electron microscopy revealed fewer mitochondria in the treated tumor cells. Importantly, this was accompanied by a 68% increase in infiltrating tumor cells in the brain parenchyma. At the molecular level we observed an increase in lactate and alanine metabolites, together with an induction of hypoxia-inducible factor 1α and an activation of the phosphatidyl-inositol-3-kinase pathway. These data strongly suggest that vascular remodeling induced by anti-VEGF treatment leads to a more hypoxic tumor microenvironment. This favors a metabolic change in the tumor cells toward glycolysis, which leads to enhanced tumor cell invasion into the normal brain. The present work underlines the need to combine anti-angiogenic treatment in GBMs with drugs targeting specific signaling or metabolic pathways linked to the glycolytic phenotype.

L-Glutamine (Gln) functions physiologically to balance the carbon and nitrogen requirements of tissues. It has been proposed that in cancer cells undergoing aerobic glycolysis, accelerated anabolism is sustained by Gln-derived carbons, which replenish the tricarboxylic acid (TCA) cycle (anaplerosis). However, it is shown here that in glioblastoma (GBM) cells, almost half of the Gln-derived glutamate (Glu) is secreted and does not enter the TCA cycle, and that inhibiting glutaminolysis does not affect cell proliferation. Moreover, Gln-starved cells are not rescued by TCA cycle replenishment. Instead, the conversion of Glu to Gln by glutamine synthetase (GS; cataplerosis) confers Gln prototrophy, and fuels de novo purine biosynthesis. In both orthotopic GBM models and in patients, ¹³C–glucose tracing showed that GS produces Gln from TCA-cycle-derived carbons. Finally, the Gln required for the growth of GBM tumours is contributed only marginally by the circulation, and is mainly either autonomously synthesized by GS-positive glioma cells, or supplied by astrocytes.

Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.

Heterozygous mutations in NADP‐dependent isocitrate dehydrogenases (IDH) define the large majority of diffuse gliomas and are associated with hypermethylation of DNA and chromatin. The metabolic dysregulations imposed by these mutations, whether dependent or not on the oncometabolite D‐2‐hydroxyglutarate (D2HG), are less well understood. Here, we applied mass spectrometry imaging on intracranial patient‐derived xenografts of IDH‐mutant versus IDH wild‐type glioma to profile the distribution of metabolites at high anatomical resolution in situ . This approach was complemented by in vivo tracing of labeled nutrients followed by liquid chromatography–mass spectrometry (LC‐MS) analysis. Selected metabolites were verified on clinical specimen. Our data identify remarkable differences in the phospholipid composition of gliomas harboring the IDH1 mutation. Moreover, we show that these tumors are characterized by reduced glucose turnover and a lower energy potential, correlating with their reduced aggressivity. Despite these differences, our data also show that D2HG overproduction does not result in a global aberration of the central carbon metabolism, indicating strong adaptive mechanisms at hand. Intriguingly, D2HG shows no quantitatively important glucose‐derived label in IDH‐mutant tumors, which suggests that the synthesis of this oncometabolite may rely on alternative carbon sources. Despite a reduction in NADPH, glutathione levels are maintained. We found that genes coding for key enzymes in de novo glutathione synthesis are highly expressed in IDH‐mutant gliomas and the expression of cystathionine‐β‐synthase ( CBS ) correlates with patient survival in the oligodendroglial subtype. This study provides a detailed and clinically relevant insight into the in vivo metabolism of IDH1‐mutant gliomas and points to novel metabolic vulnerabilities in these tumors.

Angiogenesis is regarded as a hallmark of cancer progression and it has been postulated that solid tumor growth depends on angiogenesis. At present, however, it is clear that tumor cell invasion can occur without angiogenesis, a phenomenon that is particularly evident by the infiltrative growth of malignant brain tumors, such as glioblastomas (GBMs). In these tumors, amplification or overexpression of wild-type (wt) or truncated and constitutively activated epidermal growth factor receptor (EGFR) are regarded as important events in GBM development, where the complex downstream signaling events have been implicated in tumor cell invasion, angiogenesis and proliferation. Here, we show that amplification and in particular activation of wild-type EGFR represents an underlying mechanism for non-angiogenic, invasive tumor growth. Using a clinically relevant human GBM xenograft model, we show that tumor cells with EGFR gene amplification and activation diffusely infiltrate normal brain tissue independent of angiogenesis and that transient inhibition of EGFR activity by cetuximab inhibits the invasive tumor growth. Moreover, stable, long-term expression of a dominant-negative EGFR leads to a mesenchymal to epithelial-like transition and induction of angiogenic tumor growth. Analysis of human GBM biopsies confirmed that EGFR activation correlated with invasive/non-angiogenic tumor growth. In conclusion, our results indicate that activation of wild-type EGFR promotes invasion and glioblastoma development independent of angiogenesis, whereas loss of its activity results in angiogenic tumor growth.

Anti-angiogenic therapy in glioblastoma (GBM) has unfortunately not led to the anticipated improvement in patient prognosis. We here describe how human GBM adapts to bevacizumab treatment at the metabolic level. By performing ¹³C₆-glucose metabolic flux analysis, we show for the first time that the tumors undergo metabolic re-programming toward anaerobic metabolism, thereby uncoupling glycolysis from oxidative phosphorylation. Following treatment, an increased influx of ¹³C₆-glucose was observed into the tumors, concomitant to increased lactate levels and a reduction of metabolites associated with the tricarboxylic acid cycle. This was confirmed by increased expression of glycolytic enzymes including pyruvate dehydrogenase kinase in the treated tumors. Interestingly, l-glutamine levels were also reduced. These results were further confirmed by the assessment of in vivo metabolic data obtained by magnetic resonance spectroscopy and positron emission tomography. Moreover, bevacizumab led to a depletion in glutathione levels indicating that the treatment caused oxidative stress in the tumors. Confirming the metabolic flux results, immunohistochemical analysis showed an up-regulation of lactate dehydrogenase in the bevacizumab-treated tumor core as well as in single tumor cells infiltrating the brain, which may explain the increased invasion observed after bevacizumab treatment. These observations were further validated in a panel of eight human GBM patients in which paired biopsy samples were obtained before and after bevacizumab treatment. Importantly, we show that the GBM adaptation to bevacizumab therapy is not mediated by clonal selection mechanisms, but represents an adaptive response to therapy.