Year: 2017
Type: article
Abstract: The tumor suppressor PTEN is deregulated in approximately one third of glioblastoma (GBM). PTEN has both cytoplasmic and nuclear localizations and functions. Cytoplasmic PTEN functions as a lipid phosphatase to counteract the activation of PI3K. Nuclear PTEN acts via protein-protein interactions to regulate a variety of functions in a lipid phosphatase-independent manner. A recent publication in Nature Communications sought to discover new nuclear PTEN interactions and modes of action and uncovered a chromatin-associated role of PTEN in glioblastoma tumor suppression. Using an in silico analysis followed by experimental verification, Benitez et al. found that PTEN binds the histone chaperone protein DAXX. As DAXX binds and facilitates the deposition on chromatin of histone H3.3, they performed experiments in GBM cells that demonstrated that PTEN, DAXX, and H3.3 are found in one tripartite complex and that the DAXX-H3.3 interaction is regulated by PTEN in a lipid phosphatase-independent manner. They subsequently showed that PTEN represses the expression of oncogenes CCND1, MYC, FOS and BCL2 by regulating H3.3 association and deposition on chromatin. Further, they showed that DAXX inhibition in PTEN-deficient GBM cells induced global changes in H3.3 genomic distribution, leading to oncogene repression and tumor suppressor gene upregulation. These latter changes were associated with inhibition of malignancy endpoints, including suppression of GBM-PDX xenograft growth and prolongation of the survival of mice bearing PTEN-deficient xenografts. Finally, analysis of TCGA and REMBRANDT data revealed that DAXX expression was upregulated in gliomas and inversely correlated with PTEN expression. Based on their findings described above, the authors proposed the following model. In PTEN-deficient tumor cells, DAXX removes H3.3 from chromatin, probably by competing for chromatin binding. Inhibition of DAXX restores H3.3 on the chromatin and inhibits oncogene expression. The anti-tumor effects of DAXX inhibition do not work in PTEN-expressing cells because PTEN might also bind to H3.3, leading to attenuation of H3.3 chromatin binding. One implication of the model is that inhibition of the DAXX-H3.3 interaction or inhibition of DAXX expression could potentially be used for the therapy of patients with PTEN-null GBM tumors. The previously developed PANWARDS nomogram for predicting absolute risk of intracranial hemorrhage (ICH) in non-cancer cohorts makes use of the following criteria: platelets, albumin, no congestive heart failure, warfarin, age, race, diastolic blood pressure, and stoke. Its applicability to patients with gliomas, one of the most pro-thrombotic tumors, however, had not been studied formally until now. A recent publication by Mantia and collaborators in Blood seeks to address the utility of the PANWARDS score for the development of major intracranial hemorrhage in patients with glioma and raises questions as to whether modified anticoagulant approaches are warranted in these patients.1 The authors performed a matched, retrospective cohort review with blinded radiologists to assess the rate of ICH in patients with high-grade glioma treated with enoxaparin, a low-molecular-weight heparin (LMWH) preferred as an anticoagulant agent for patients with cancer due in part to its less frequent monitoring requirements and fewer drug interactions.2 Interestingly, the administration of anticoagulation for venothromboembolism to patients with brain metastases does not appear to increase the risk for ICH,3,4 yet a number of studies to date have now reported an increased risk of ICH among anticoagulated patients with glioma. These studies, however, don’t take into account the nuances of ICH, which may range from trace radiologic evidence of blood products to obvious hemorrhage with massive effect to the expected postoperative hemorrhage after surgery. This study addresses the confounders encountered in prior studies by including pre-defined ICH criteria. The outcomes of this study do in fact raise concern for the increased likelihood of developing a major ICH among those receiving enoxaparin compared with those with spontaneous ICH in the untreated cohort. Moreover, overall survival was shorter for patients suffering a major ICH on enoxaparin compared with controls. Given the specific risk factors facing patients with brain tumors, is there a need to develop a scoring system or at least a modified version of the currently validated system to take into account such factors as histology, tumor size and location, and the potential effect of other anti-cancer agents? Moreover, is there any benefit of reduced-intensity anticoagulation? The authors point out that among the over 3000 patients followed in the Computerized Registry of Patients with Venous Thromboembolism who received less aggressive anticoagulation with LMWH below the daily recommended dosing, there were fewer cases of major hemorrhages and less fatal pulmonary emboli reported. Also, is there a role for direct oral anticoagulants, purported to result in less frequent major hemorrhages compared with warfarin and less bleeding-related mortality in non-cancer patients? Most importantly, this study draws our attention to the observation that major ICH incidents are significantly greater in enoxaparin-treated patients with glioma and highlights the need for larger studies to address these very questions. Glioma patients with the IDH-mutant (IDH-MUT)/glioma CpG island methylator phenotype (G-CIMP) show highly improved prognosis compared with those with wild-type IDH/non-G-CIMP.1 However, IDH mutations are rare in the more aggressive grade IV glioblastomas, and introducing IDH mutations in mouse neural cells does not induce gliomagenesis. This is a paradox, given that IDH mutation is considered an early oncogenic event, and the mutated IDH produces the oncometabolite 2-hydroxyglutarate (2-HG).2 The reasons behind the differences in prognosis is unknown, and identifying the cause is further hampered by the lack of mouse models that compare IDH-WT versus IDH-MUT gliomas in the same genetic background. To address this, Amankulor et al 3 developed a pair of isogenic mouse models using the RCAS/tva system with ectopic expression of mutant (R132H) or wild-type IDH1 in PDGF-driven gliomas. The authors first examined the differences in human gliomas and found generally reduced immune cell types, including microglia, macrophages, dendritic cells, B cells, and T cells in IDH-MUT gliomas when compared with IDH-WT. TCGA analyses confirmed the generally lower immune infiltration in IDH-MUT gliomas, which the authors attributed to lower expression of chemotaxis-associated genes in these tumors. In line with the human situation, mice bearing IDH-MUT gliomas had an improved overall survival compared with the IDH-WT mice in both Ink4a/Arf+/+ and Ink4a/Arf+/-, but not in the Ink4a/Arf-/- backgrounds, and the tumors in IDH-MUT mice contained increased 2-HG and hypermethylation. Importantly, a similar pattern of reduced immune cell infiltration was observed in murine IDH-MUT gliomas. Microarrays revealed differential expression of genes associated with the immune system as well as leukocyte and neutrophil migration. The authors postulated that the reduced expression of chemotaxis-associated genes in the IDH-MUT tumor cells functionally inhibits immune cell migration. Consistently, using Boyden chamber assays, they found reduced migration of neutrophils towards both lysates and conditioned media from murine IDH-MUT tumors. The gene expression of chemoattractant cytokines was also reduced in IDH-MUT tumors, confirming their hypothesis. Finally, treatment of IDH-WT, but not IDH-MUT, tumor-bearing mice with an antibody that depletes neutrophil populations caused a significant improvement in survival. Collectively, this study elucidates a key mechanism that may underlie differential prognosis between IDH-MUT and IDH-WT patients, and the mouse models developed in this study will be invaluable for future studies comparing these two tumor types. An article in by Chen et al in Neurology explores the observed relationship in gliomas that seizures are more common in low- and intermediate-grade gliomas and that pre-glioma treatment seizures predict longer survival.1 The authors posit these observations are reflective of IDH1-mutated gliomas and its inherent biology. Using a retrospectively collected cohort of adult WHO Grade II-IV gliomas (n=712) from 3 institutions, pre-therapy seizure frequency was noted to be higher in IDH1-mutated gliomas compared with IDH1 wild-type gliomas (59–74% vs. 18–34%). WHO Grade II and III gliomas were shown to be enriched in IDH-mutated gliomas (Grade II 80.5%, Grade III 54.7%, and Grade IV 8.9%), as previously demonstrated, which underpins the relationship of increased seizures in lower-grade gliomas. There was no difference in seizure frequency between IDH1-mutated 1p/19q-codeleted gliomas and IDH1-mutated 1p/19q-intact gliomas. Older observations that oligodendroglial tumors have a comparatively higher incidence of seizures may reflect the co-relationship between oligodendoglial histology and IDH1 mutation. By multivariable analysis, the presence of IDH1 mutation independently correlated with seizures. Additionally, IDH1 mutation independently correlated with survival, a phenomenon previously shown in several prospective trials. The observation that patients with gliomas presenting with seizures survive longer is likely reflective of harboring an IDH1-mutated glioma, which predicts improved survival. Similarly, the clinical observation that patients with oligodendroglial tumors survive longer than those with astrocytic gliomas reflects that most oligodendroglial gliomas are IDH1 mutated. The authors suggest that the production of D-2-hydroxyglutarate (D-2HG), the oncometabolic product derived from the reduction of α-ketoglutarate by mutant IDH1, is a new and novel mechanism of seizure induction. The evidence provided for this novel mechanism of epileptogenicity was twofold. First, there is a strong chemical structural similarity of D-2HG to glutamate, the primary excitatory neurotransmitter in brain, suggesting D-2HG is a glutamate mimetic. Glutamate binds to and activates multiple neuronal receptors, including ligand-gated NMDA and AMPA ion channels. Upon activation, neuronal influx of sodium, potassium, and calcium occurs, resulting in depolarization and action potential generation. In an in vitro model using human kidney cells, D-2HG binds to NMDA but not AMPA receptors. Theoretically D-2HG could then activate neuronal NMDA receptors when released from tumor cells into the extracellular space and in proximity to cortical neurons. This hypothesis was tested by the authors using cultured rat neurons and multielectrode array (MEA) plates, a system widely used to measure neuronal activity experimentally. MEAs allow for non-invasive, simultaneous, and long-term recording from a population of neurons, permitting measurement of neuronal network activity and determination of changes in the inhibitory/excitatory balance that leads to seizures. The authors show D-2HG activates neuronal NMDA receptors and thereby increases synchronization of neuronal activity, a precondition for seizure-promoting agents. The release of D-2HG by IDH1-mutant gliomas into the extracellular space and subsequent excitatory activation of surrounding neurons appears to mechanistically explain the increased seizure frequency seen in IDH1-mutant gliomas. Anti-IDH1 small-molecule inhibitors already in clinical trials may therefore be an alternative anti-epileptic agent in IDH1-mutant tumors separate from their putative targeted anti-glioma therapy.2 In recent years, efforts have expanded to develop mathematical models of glioblastoma (GBM) to hammer out order from these chaotic brain cancers. Successful mathematical modeling of GBMs might enable better predictions of tumor behavior and patient prognosis, possibly even yielding insights that improve treatment. However, these endeavors have been hampered by the complexity of cancer in vivo, particularly GBM. Mathematical modeling depends on understanding factors affecting GBM growth, invasion, responses to stimuli and treatment, and system elements such as angiogenesis and regions of hypoxia. The report by Yan et al. in Cancer Research is likely the most extensive effort to date, and it yields a somewhat surprising new prediction concerning a key element in GBM treatment resistance.1 The authors articulate an elaborate mathematical model attempting to factor in GSCs (GBM stem/initiating cells), GECs (GBM cells transdifferentiated into vascular endothelial-like cells), secreted factors like Wnt activators and inhibitors, and angiogenesis and hypoxia (including secreted VEGF as a vascular stimulator). When the model was completed and run, it predicted that GBM resistance to standard treatments occurs at least in part through cooperation of GECs with GSCs. Cancer cell-derived endothelial-like cells were first demonstrated in melanoma many years ago, but in subsequent years they were identified in other cancers as well—prominently including GBM. GECs have not yet been confirmed to play an important functional role in GBM, but this report suggests that they might be critical not just in the GBM vasculature but also in GBM treatment resistance. As with any mathematical model, the output is only as strong as the assumptions provided as input. In this report, some of the assumptions might be questioned. For example, the authors assume a constant ratio of tumor cell to interstitial volumes, but tumor cell size changes, and vascular leakiness can affect interstitial volume and edema. The assumptions underlying modeling of secreted activators and inhibitors of differentiation is not well established and unavoidably oversimplifies this complex issue. Furthermore, there is substantial debate about whether the transdifferentiation to an EC-like fate is a major phenomenon in GBMs. A convincing prior report using staining for the IDH R132H mutation or EGFR amplification failed to find significant numbers of GBM cells incorporated into tumor vascular endothelium 2. However, the report by Yan et al. suggests that it may be important to perform such an assessment in post-treatment samples, as they might have a much higher proportion of GECs. Despite these questions about the input assumptions, this report provides an important step forward for mathematical modeling of GBM and a putative new mechanism for GBM treatment resistance—suggesting a need to target both GECs and GSCs. BRAF inhibitor monotherapy has previously been shown to have clinical activity and good tolerability in patients with BRAF-mutant melanoma who developed metastases in the brain. Although the BRAF and MEK inhibitor combination therapy has been recently reported to be superior to BRAF inhibitor monotherapy in patients with BRAFV600-mutant metastatic melanoma without brain metastases1, the clinical effect of this regimen on melanoma brain metastases has not been characterized. In the July issue of Lancet Oncology, Davies et al2 report the results from the multicenter, open-label phase 2 COMBI-MB trial evaluating oral dabrafenib plus oral trametinib in four patient cohorts with melanoma brain metastases enrolled from 32 institutions in Europe, North America, and Australia: (A) BRAFV600E-positive, asymptomatic melanoma brain metastases, with no previous local brain therapy, and an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1; (B) BRAFV600E-positive, asymptomatic melanoma brain metastases, with previous local brain therapy, and an ECOG performance status of 0 or 1; (C) BRAFV600D/K/R-positive, asymptomatic melanoma brain metastases, with or without previous local brain therapy, and an ECOG performance status of 0 or 1; and (D) BRAFV600D/E/K/R-positive, symptomatic melanoma brain metastases, with or without previous local brain therapy, and an ECOG performance status of 0, 1, or 2. The primary endpoint was investigator-assessed intracranial response in cohort A and secondary endpoints included intracranial response in cohorts B, C, and D. Between 2014 and 2016, 125 patients were enrolled: 76 in cohort A, 16 in cohort B, 16 in cohort C, and 17 in cohort D. After a median follow-up of 8.5 months, 44 (58%) of 76 patients in cohort A achieved an intracranial response. Intracranial response by investigator assessment was also achieved in nine (56%) of 16 patients in cohort B, seven (44%) of 16 patients in cohort C, and ten (59%) of 17 patients in cohort D. The most common serious adverse events were pyrexia for dabrafenib (6%) and decreased ejection fraction (4%) for trametinib. The findings of this study suggest that intracranial responses in patients with BRAFV600-mutant melanoma brain metastases were improved with dabrafenib and trametinib combination therapy compared with previously reported analyses of BRAF inhibitor monotherapy in this setting (16%-39%)3. However, the durability of clinical responses and disease control were relatively short compared with those in previous trials in patients without melanoma brain metastases. Median duration of overall response in this study (6.5 months in cohort A) was shorter than that observed in patients without melanoma brain metastases (12.0 months in the phase 3 study COMBI-d) 1. This is the first report of a phase 2 trial evaluating BRAF and MEK inhibitor combination therapy in patients with melanoma brain metastases. Limits of the study were the non-randomized design and small sample size for cohorts B, C, and D. Continued follow-up is needed to assess the full effect of dabrafenib plus trametinib on overall survival in all cohorts of patients. Moreover, combinatorial approaches with brain-direct treatments, such as stereotactic radiosurgery, should be explored.
Source: Neuro-Oncology
Language: English
Subfield: Genetics
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W4244227461