Year: 2019
Type: article
Abstract: Isocitrate dehydrogenase 1 (IDH1) mutations are found in the majority of low-grade gliomas and secondary glioblastoma. IDH1 mutants convert 2-oxoglutarate (2OG) to 2-hydroxyglutarate (2HG), which inhibits 2OG-dependent dioxygenases and leads to changes in the epigenetic landscape of gliomas. However, cells also have other 2OG-dependent enzymes, including transaminases that catalyze the production of glutamate needed for the production of the antioxidant glutathione. Previous work showed that IDH mutant cells have decreased levels of branched-chain α-ketoacids, which are made from the branched-chain amino acids (BCAA) by the transaminases BCAT1 and BCAT2, suggesting that these transaminases are inhibited in IDH mutant cells. Based on these findings and the dependence of BCAT1 and BCAT2 on 2OG, a recent study published in the journal Cell hypothesized that IDH1 mutant-derived 2HG inhibits BCAT1 and BCAT2, leading to a decrease in glutamate and glutathione and thus creating metabolic vulnerabilities that could be exploited therapeutically.1 To verify the hypothesis, the authors first demonstrated that 2HG inhibits BCAT1 and BCAT2 by showing that IDH1 mutants increase BCAAs and decrease glutamate in glioma cells and that 2HG competitively inhibits BCAT1 and BCAT2 activities in a recombinant BCAT activity assay. They then demonstrated that IDH mutations impair BCAA catabolism and glutamate biosynthesis by showing that IDH1 mutants inhibit BCAA leucine transamination and suppress glutamate levels in a 2HG- and BCAT-dependent manner. Since cells can also make glutamate from glutamine using glutaminase (GLS), and since human brain tumors use large amounts of glutamine, the authors hypothesized that GLS might compensate for the loss of BCAT activity in IDH mutant cells. They proved this hypothesis by experimentally showing that IDH mutant cells and tumors display increased reliance on GLS to produce glutamate and that GLS inactivation leads to a decrease of glutamate in IDH mutant but not IDH wildtype experimental brain tumors. They then showed that inhibition of glutamate synthesis by IDH mutants leads to a reduction of glutathione levels and that this reduction was mediated by BCAT inhibition and rescued by GLS overexpression. Since GLS compensates for IDH mutant-induced inhibition of the antioxidant glutathione, the authors then speculated that GLS inhibition might have anti-glioma effects under oxidative stress. Accordingly, they showed that GLS inhibition with the drug CB-839 in combination with oxygen-, tert-butyl oxidant-, or radiation-induced oxidative stress leads to a strong inhibition of the growth of IDH mutant but not IDH wildtype glioma cells and experimental tumors. In summary, the study showed that IDH mutation-induced 2HG inhibits BCAT transaminases, leading to impaired glutamate and glutathione synthesis. As activation of glutaminase can compensate for BCAT inhibition, inhibition of glutaminase in combination with induction of oxidative stress represents a new therapeutic strategy for IDH mutant gliomas. In a current translational paper published in Nature Medicine, a multi-center group led by Peter Fecci from Duke postulates sequestration of T cells in the bone marrow as a reason for lymphocytopenia in patients with intracranial neoplasms including glioblastoma.1 The current obstacles for an effective anti-glioma immunotherapy are not fully elucidated. The main factors usually discussed are the low number of mutations limiting the amount of putatively more immunogenic neo-antigens.2 Further, glioblastomas are considered immunologically resistant and “cold”, with limited intra-tumoral immune cell infiltration.3 Other factors may include the prevalent use of steroids with their T cell cytotoxic properties and reduced lymphocyte counts exerted by the cytotoxic treatment. Already some years ago Grossman et al. speculated about the relevance of treatment-associated immunosuppression in patients with glioblastoma.4 They described starting levels of 664 CD4+ cells/mm3 in 96 patients, with a severe and persistent drop during treatment. The current work by Chongsathidkiet et al excels the prior work on many levels.1 The authors describe a 15% cohort of patients with glioblastoma (and tumor-bearing mice) to have extremely low CD4 counts (≤200 cell/mm3). In their search for the cause of this pretreatment lymphocytopenia, they observed T cells sequestered in the bone marrow. This phenomenon is not only naturally present in some patients, but is inducible in mice by implantation of glioblastoma or other cancer cells into the intracranial compartment, but not into the periphery. Although a functional link between T cell sequestration and the neoplastic brain lesion was not established, the authors refer to a prior observation of sphingosine 1-phosphate (S1P) / S1P receptor (S1P1) gradient disruption, which prevents lymphocyte egress from lymphoid organs5 (and now also the bone marrow). In the present paper, Chongsathidkiet et al. demonstrate that S1P1 is lost from the T cell surface with tumor inoculation into the intracranial compartment. Genetic inhibition of S1P1 internalization prevents sequestration in this model. In addition to the fine experimental data, this work introduces T cell sequestration as a potential additional mode of T cell dysfunction in brain tumors. The link to S1P1 loss on naïve T cells provides a novel target for immune intervention in glioblastoma and putatively brain metastases. This potential use, however, needs to be put into perspective of other treatments, as the mono intervention of preventing S1P1 internalization had no relevant impact on the brain tumors in the model described, as is to be expected from interfering with the number of naïve T cells. The S1P1 internalization blockade had an impact on the efficacy of checkpoint inhibition as well as CD137 agonism. As pharmacological tools to prevent S1P1 down-regulation are lacking, the use of G-CSF, as described in the current article, may be attractive and should be considered when discussing immunotherapy adjunct measures. Secondary glioblastoma arises from lower-grade glioma and comprises about 10% of glioblastoma. While it often has IDH1 or IDH2 mutation and improved prognosis, it is still universally lethal. Therapies targeting mutant IDH1/2 are being developed, but they have not yet shown significant efficacy against secondary glioblastoma, and other treatment approaches are desperately needed. A new report from Hu et al. provides important new biological information on secondary glioblastoma that gives rise to a new therapeutic strategy.1 The authors performed DNA sequencing of a large set of 188 secondary glioblastomas, providing the most extensive genetic landscape of these tumors to date. They identified a number of expected elements, including IDH1, TP53, and ATRX mutations, but they also noted genetic aberrations not previously observed in secondary glioblastoma—prominently including lesions in the MET oncogene. Three were identified: 1) skipping of MET exon 14 in the transcript; 2) fusions of the PTPRZ1 and MET genes; and 3) MET amplification. MET exon 14 skipping was by far the most common, in approximately 14% of secondary glioblastoma, and was associated with poor prognosis. MET exon 14 skipping occurred through more than one mutational mechanism and was found to prolong MET protein half-life. MET fusions frequently occurred simultaneously with exon 14 skipping. Interestingly, the MET exon 14 skipping and MET fusions were associated with an increase in tumor-associated macrophages. To test whether these might be targetable driver lesions in secondary glioblastoma, the authors used a novel MET kinase inhibitor with good blood-brain barrier permeability, PLB-1001. PLB-001 showed preclinical promise against MET-driven glioma lines in vitro and in vivo, prompting a phase I trial in patients with pretreated, recurrent secondary GBM with MET lesions. This trial showed not only safety but also some preliminary signs of efficacy, with two patients with advanced disease showing partial responses that lasted for 20 weeks. This report establishes MET as a potential driver in at least a substantial subset of secondary glioblastoma, yielding valuable novel biological information about these lethal brain tumors and also providing a new therapeutic option. Other MET inhibitors with good central nervous system penetration are already FDA-approved for non-neuro-oncological indications or are in the pipeline, suggesting therapeutic alternatives to PLB-001 for targeting MET—although data from this report suggest that PLB-001 may be superior. Notably, MET inhibition has thus far proven disappointing in trials for patients with glioblastoma, but these trials included an unselected patient population and would have had few if any patients with MET-altered secondary glioblastoma. This report suggests the need for larger trials of MET inhibitors in MET-altered secondary glioblastoma and raises other questions as well. It will be important to determine if responses are durable or last only months, as was seen in this study and as is often the case with precision medicine approaches against adaptable cancers. It should also be evaluated whether there is combinatorial activity in combining MET inhibition with blockade of other potential drivers of secondary glioblastoma, e.g. mutant IDH1, or of potential resistance mediators such as PI3 kinase. Molecular profiling in recent years has dramatically clarified the landscape of pediatric brain tumors, refining established classification schemes while also designating a number of novel disease entities and subtypes. Capturing the full extent of this heterogeneity with tractable tumor model systems represents an important prerequisite for effective preclinical testing. In a recent Nature Medicine publication, Brabetz et al. report the characterization of 30 patient-derived orthotopic xenograft models and 7 cell lines that together represent 14 distinct molecular subgroups of pediatric brain tumors.1 In generating these models from freshly obtained tumor tissue collected between 2009 and 2014, the authors found a general engraftment rate of 37%, which varied between tumor type. After serial passaging, tumor latency became nearly identical between animals for each cell line, but varied considerably between lines based on histology and/or molecular subtype. Virtually all lines were found to regrow after cryopreservation in liquid nitrogen. Xenografts in the cohort resembled their parental tumor histopathologically. Moreover, global methylation profiling revealed that the xenografts clustered with tumors corresponding to the appropriate disease subclasses, with the exception of three cases that had been inappropriately classified initially on the basis of histopathology. The authors went on to subject their xenografts and cell lines to whole-exome sequencing and low-coverage whole-genome sequencing, finding that all models harbored genetic alterations appropriate for their molecular subtype. Additionally, ~98% of all single-nucleotide variants identified in parent tumors were also identified in their respective models. Similar results were obtained by DNA copy number, global methylation, and gene expression analyses, emphasizing the fidelity of the model systems. Of note, and perhaps not surprisingly, aggressive tumor types such as MYC-amplified group 3 medulloblastoma and MYCN-amplified sonic hedgehog (SHH) medulloblastoma were overrepresented in the model cohort. Finally, the authors demonstrated that their models could be effectively employed to identify therapeutic regimens targeted to specific molecular profiles. For instance, they found that two SHH medulloblastoma xenografts (each with a distinct PTCH1 mutation), but not a MYCN and GLI2-amplified xenograft, responded to treatment with the smoothened (SMO) inhibitor vismodegib. Analogous results with different drugs were also seen in high-grade glioma and group 3 medulloblastoma models. These findings are highly significant in that they present a well-validated cohort of viable disease models that are optimized for preclinical studies and readily available to the larger research community (http://www.r2platform.com/pdexplorer). Molecularly targeted therapy has been an area of active research in gliomas as exemplified by multiple studies targeting EGFR amplification and IDH mutations. Unfortunately, significant clinical benefit has been limited to date, but that may be changing. The VE-BASKET study was an international, open-label, nonrandomized study that enrolled multiple cancer types with BRAFV600 mutations, including 24 patients with gliomas. Kaley et al report the results of the glioma cohort, which enrolled a variety of histologies, including high-grade gliomas, pilocytic astrocytomas, anaplastic ganglioglioma, and pleomorphic xanthoastrocytoma (PXA).1 Patients were treated with vemurafenib until disease progression, withdrawal, or intolerable adverse effects. The results were promising, with a confirmed objective response rate of 25%, median progression-free survival of 5.5 months, and median overall survival (OS) of 28.2 months for all patients. Patients with PXA particularly benefited (median OS not reached yet), whereas those with higher-grade tumors experienced a more variable response (although these patients were also heavily pre-treated). There were no unexpected adverse events. Despite the small number of patients, this study is one of the few showing that targeting an oncogene may be efficacious in glioma. More work needs to be done to determine the sequencing of treatment; e.g. should BRAF targeting treatment be used earlier in the disease course? Most of these patients had been through several lines of treatment, and repeat tissue sampling was not routinely done to determine persistence of the BRAF mutation or accumulation of other mutations that may affect response to targeting a single mutation. Moreover, heterogeneity of target expression is a concern as is the loss of the target through Darwinian pressure or even in the absence of targeted therapy, as has been seen in prior studies.2 Treatment for melanoma has moved to combining RAF and MEK inhibition, and these studies are ongoing in glioma as well. In summary, this is a promising proof-of-concept study that opens up new treatment options for these patients with limited treatment choices.
Source: Neuro-Oncology
Language: English
Subfield: Genetics (clinical)
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W4252299596