Researchers have discovered a pair of drugs that work together to kill cancer cells


Researchers have devised a new plan of attack against a group of deadly childhood brain cancers collectively called diffuse midline gliomas (DMG), including diffuse intrinsic pontine glioma (DIPG), thalamic glioma and spinal cord glioma.

Scientists at the National Institutes of Health, Stanford University, California, and Dana-Farber Cancer Institute, Boston, identified a drug pair that worked together to both kill cancer cells and counter the effects of a genetic mutation that causes the diseases.

The researchers showed that combining the two drugs – panobinostat and marizomib – was more effective than either drug by itself in killing DMG patient cells grown in the laboratory and in animal models.

Their studies also uncovered a previously unrecognized vulnerability in the cancer cells that scientists may be able to exploit to develop new strategies against the cancer and related diseases. The results were published November 20 in Science Translational Medicine.

Matrix screening delivers insights, increases options

DMGs are aggressive, hard-to-treat tumors that represent the leading cause of brain cancer-related death among U.S. children. DMGs typically affect a few hundred children a year between ages 4 to 12; most children die within a year of diagnosis.

Most cases of DMG are caused by a specific mutation in histone genes. Histones are protein complexes in the cell nucleus. DNA wraps around histones to form chromatin, which packages DNA in the nucleus.

How DNA winds and unwinds around histones is influenced by enzymes, including histone deacetylases.

These enzymes add or remove chemical tags, which indirectly controls if genes are turned on or off.

In an earlier study, Stanford neuro-oncologist Michelle Monje. M.D., Ph.D., and her colleagues showed that panobinostat, which blocks key histone deacetylase enzymes, could restore the DIPG histone function to a more normal state.

While panobinostat is already in early clinical testing in DIPG patients, its usefulness may be limited because cancer cells can learn to evade its effects. So Monje’s team wanted to identify other possible drugs – and combinations of them – that could affect the cancer.

“Very few cancers can be treated by a single drug,” said Monje, a senior author of the study who treats children with DIPG and other diffuse midline gliomas.

“We’ve known for a long time that we would need more than one treatment option for DIPG. The challenge is prioritizing the right ones when there are thousands of potential options. We’re hopeful that this combination will help these children.”

Monje and the National Cancer Institute’s Katherine Warren, M.D., now at Dana-Farber Cancer Institute and Boston Children’s Hospital, collaborated with Craig Thomas, Ph.D., and his colleagues at the NIH’s National Center for Advancing Translational Sciences (NCATS). Thomas and his team used NCATS’ drug screening expertise and matrix screening technology to examine drugs and drug combinations to see which ones were toxic to DIPG patient cells.

NCATS’ robotics-enabled, high-throughput screening technologies enable scientists to rapidly test thousands of different drugs and drug combinations in a variety of ways.

Scientists can examine the most promising single drugs and combinations, determine the most effective doses of each drug and learn more about the possible mechanisms by which these drugs act.

The NCATS researchers first studied the effects of single approved drugs and investigative compounds on DIPG cell models grown in the laboratory from patient cells.

They focused on agents that could both kill DIPG cells and cross the brain’s protective blood-brain barrier, a necessity for a drug to be effective against DIPG in patients. The team then tested the most effective single agents in various combinations.

“Such large, complex drug screens take a tremendous collaborative effort,” said Thomas, also a senior study author. “NCATS was designed to bring together biologists, chemists, engineers and data scientists in a way that enables these technically challenging studies.”

While there were multiple, promising outcomes from these screens, the team focused on the combination of histone deacetylase inhibitors (like panobinostat) with drugs called proteasome inhibitors (such as marizomib). Proteasome inhibitors block cells’ normal protein recycling processes. The panobinostat-marizomib combination was highly toxic to DIPG cells in several models, including DIPG tumor cell cultures that represented the main genetic subtypes of the disease and mice with cells transplanted from patient tumors.

The combination also reduced tumor size in mice and increased their survival. A similar response was found in spinal cord and thalamic DMG models developed from cells grown in culture from patient cells.

Mechanisms at play

The screening studies also provided important clues to the ways the drugs were working. Building on these data, the collaborative team subsequently conducted a series of experiments that showed the DIPG cells responded to these drugs by turning off a biochemical process in the cell’s mitochondria that is partly responsible for creating ATP, which provides energy to cells.

The drug combination essentially shuts down tumor cell ATP production.

“The panobinostat-marizomib drug combination exposed an unknown metabolic vulnerability in DIPG cells,” said first author Grant Lin, Ph.D., at Stanford University School of Medicine. “We didn’t expect to find this, and it represents an exciting new avenue to explore in the development of future treatment strategies for diffuse midline gliomas.”

Plans are underway for clinical trials of the drug combination and of marizomib alone.

“Many drugs that we test have multiple effects on DIPG cells,” said Warren, a senior study author. “Panobinostat, for example, inhibits a specific enzyme, but it has other mechanisms working in tumor cells that may contribute to its effectiveness. We’re still trying to understand the various Achilles heels in these cancer cells. This work is an important step in translating our preclinical data into patients.”

Monje stressed the panobinostat-marizomib combination might be an important component of a multitherapy strategy, including approaches that harness the immune system and those that disrupt factors in the tumor microenvironment that the glioma cells depend on to grow. Like Warren, Monje emphasized the need to better understand how drugs target and impact the DIPG cells’ vulnerabilities.

“Our work with NCATS showed the need to gather more preclinical data in a systematic, high-throughput way to understand and prioritize the strategies and agents to combine,” Monje said. “Otherwise we’re testing things one or two drugs at a time and designing clinical trials without preclinical data based on hypothesized mechanisms of action. We want to move past this guesswork and provide preclinical evidence to guide clinical decisions and research directions.”

This shows cancer cells

Diffuse intrinsic pontine glioma (DIPG) is a rare, deadly childhood brain cancer. Here, a confocal micrograph shows DIPG cells, grown from patient cells, in culture. The cells are expressing several proteins, including: DAPI (blue, in the cell nuclei); vimentin (red, a protein marker of neural stem/precursor cells and glioma cells); and Nestin (green, also a protein marker of neural stem/precursor cells and glioma cells). The image is credited to Shawn Gillespie, Monje Laboratory, Stanford Medicine.

Lin added, “The idea is to get as many effective tools as possible to work with that can have an impact on patients.”

Funding: The research was funded by Alex’s Lemonade Stand Foundation, Izzy’s Infantry Foundation, McKenna Claire Foundation, Unravel Pediatric Cancer, Defeat DIPG Foundation, ChadTough Foundation, N8 Foundation, Kortney Rose Foundation, Cure Starts Now Foundation and the DIPG Collaborative, Sam Jeffers Foundation, Lyla Nsouli Foundation, Abbie’s Army Foundation, Waxman Family Research Fund, Virginia and D.K. Ludwig Fund for Cancer Research, National Institute for Neurological Disorders and Stroke (R01NS092597) and NIH Director’s Common Fund (DP1NS111132), Maternal and Child Health Research Institute at Stanford, the Anne T. and Robert M. Bass Endowed Faculty Scholarship in Pediatric Cancer and Blood Diseases, The DIPG All-In Initiative and the NCATS and NCI intramural programs.

Diffuse midline gliomas (DMG) are aggressive and incurable pediatric brain tumors, characterized by high levels of intrinsic and acquired resistance to therapy, as well as therapy failure due to an intact blood-brain barrier (BBB), leading to a dismal median survival of ˜11 months (Jones et al., 2017Veldhuijzen van Zanten et al., 2017). In general, chemoresistance to antitumor agents in DMG, brain tumors and various human cancers is a primary hindrance towards curative cancer therapy (Abdul et al., 2018Saleem et al., 2019Sharifzad et al., 2019van Tellingen et al., 2015). Hence, deciphering the molecular mechanisms underlying intrinsic and acquired cancer drug resistance (Mudduluru et al., 2016Wijdeven et al., 2016Zhitomirsky and Assaraf, 2016) is a major goal of cancer resistance, in order to develop novel modalities that may overcome multidrug resistance in cancer (Bar-Zeev and Assaraf, 2017Cui et al., 2018Levin et al., 2019Livney and Assaraf, 2013Mudduluru et al., 2016).

Formerly, DMG tumors were known as diffuse intrinsic pontine glioma (DIPG), or high grade gliomas (HGG) occurring in midbrain structures. The discovery of highly prevalent, oncogenic Histone 3 mutations in these tumors, prompted the World Health Organization in 2016 to redefine these tumors as diffuse midline gliomas (DMG) carrying H3K27 M mutations (Khuong-Quang et al., 2012Lewis et al., 2013Louis et al., 2016Wu et al., 2012). Until about a decade ago, very little research was performed on this type of cancer. This was mainly due to the lack of biological material, since the majority of children did not undergo biopsies or surgeries due to the presumed risk of operating in such delicate areas, and at the same time the possibility to confirm the diagnosis by magnetic resonance imaging (MRI). Central in DMG biology are the lysine-to-methionine substitutions, at position 27 in Histone 3 genes (H3K27M), identified in over 80% of cases (Bender et al., 2013Castel et al., 2015Chan et al., 2013Khuong-Quang et al., 2012Lewis et al., 2013Wu et al., 2012). The introduction of autopsy and biopsy studies has also led to a significant increase in the availability of cell culture and animal models of DMG, enabling preclinical therapeutic testing (Becher et al., 2010Grasso et al., 2015Hennika et al., 2017Misuraca et al., 2015Pathania et al., 2017Plessier et al., 2017Veringa et al., 2013). Consequently, a variety of therapeutic targets has been discovered, some of which have been the topic of intensive preclinical studies. However, currently there is no therapeutic trial in patients that has reported any convincing survival benefit of pharmacological interventions.

The current review summarizes the druggable targets that have been identified for the treatment of DMG, and provides a critical evaluation of the preclinical studies performed to validate the efficacy of inhibiting or modulating these putative targets. Currently, the main therapeutic targets that have been identified for DMG can be classified as targeting epigenetic modulators, receptor tyrosine kinases and their related signal transduction pathways, cell cycle checkpoints, the stem cell phenotype of DMG cells, DNA damage repair systems and targets for immunotherapy. Each of these groups of therapeutic targets will be discussed in detail in the following sections.

Epigenetic modification

With the high prevalence of mutations in genes encoding Histone 3 in DMG, and their consequences for chromatin remodeling and global transcriptional patterns, epigenetic modification represents a logical and promising therapeutic strategy in these tumors (Table 1 and Fig. 1) (Aihara et al., 2014Bender et al., 2013Castel et al., 2015Chan et al., 2013Cordero et al., 2017Fang et al., 2018Hoffman et al., 2016Khuong-Quang et al., 2012Lewis et al., 2013Nagaraja et al., 2017Nikbakht et al., 2016Pathania et al., 2017Ryall et al., 2016Stafford et al., 2018Wu et al., 2012). In this regard, the most obvious strategy would be to restore the trimethylation of H3K27, which is reduced in a transdominant manner as a consequence of heterozygous H3K27 M mutations. As such, the histone demethylase inhibitors GSKJ1 and GSKJ4 (a prodrug of GSKJ1), targeting the histone lysine demethylase 6B (KDM6B/JMJD3), have shown promising activity in vitro as well as in vivo in patient-derived xenograft models of DMG (Grasso et al., 2015Hashizume et al., 2014). Recently it was shown that in H3K27 M mutated cells, H3K27 trimethylation is actually increased at certain specific loci in the genome, leading to repression of transcription (Chan et al., 2013Cordero et al., 2017Fang et al., 2018Mohammad et al., 2017Piunti et al., 2017Stafford et al., 2018). Based on these findings, inhibition of enhancer of zeste 2 (EZH2), the catalytic subunit of the polycomb repressor complex 2 (PRC2) – which is responsible for trimethylation of H3K27 – has demonstrated preclinical efficacy in vitro and in vivo, although one study failed to demonstrate such an antitumor effect (Mohammad et al., 2017Piunti et al., 2017Wiese et al., 2016Zhang et al., 2017).

Table 1. Epigenetic regulators as therapeutic targets in diffuse midline glioma.

TargetIn vitro efficacyIn vivo efficacyRemarksReferences
KDM6B++Synergy with HDAC inhibitionHashizume et al., 2014Grasso et al., 2015
PRC2++Synergy with BET inhibitionWiese et al., 2016Mohammad et al., 2017Piunti et al., 2017Zhang et al., 2017
HDACs++Synergy with KDM6B and BET inhibitionGrasso et al., 2015Hennika et al., 2017Nagaraja et al., 2017Pal et al., 2018
CDK7++Synergy with HDAC inhibitionNagaraja et al., 2017
BET++Synergy with EZH2 and HDAC inhibitionPiunti et al., 2017Zhang et al., 2017
PRC1+n/aKumar et al., 2017Filbin et al., 2018

n/a: data not available.

Fig. 1

Another consequence of H3K27 M mutations is an increase in acetylation of the remaining H3K27 residues, resulting in an open chromatin structure and superenhancer activation, and subsequent transcriptional activation, at these genomic loci (Fang et al., 2018Piunti et al., 2017). Inhibition of histone deacetylases (HDACs) by the pan-HDAC inhibitor panobinostat and the dual PI3K/HDAC inhibitor fimepinostat (CUDC-907) further enhanced this consequence of H3K27 M mutations, while at the same time indirectly restoring H3K27 trimethylation (Grasso et al., 2015Hennika et al., 2017Nagaraja et al., 2017Pal et al., 2018). These global chromatin alterations induced by HDAC inhibitors seem specifically cytotoxic to H3K27 M mutated cells and as a result have shown promising preclinical efficacy in treating DMG in vitro and in vivo (Grasso et al., 2015Hennika et al., 2017Pal et al., 2018). Nonetheless, DMG cells rapidly acquire resistance to panobinostat, suggesting that monotherapy with HDAC inhibitors is unlikely to exhibit curative potential (Grasso et al., 2015Hennika et al., 2017).

A less specific strategy for epigenetic-based therapy, which has proven to be effective in other cancer types with strong transcriptional dysregulation, is the disruption of RNA polymerase II (RNAPII)-mediated transcription. This can be achieved in two ways: either via inhibition of bromodomain and extra-terminal (BET) proteins, or via inhibition of cyclin-dependent kinase 7 (CDK7), which phosphorylates and alters the activity of RNAPII (Nagaraja et al., 2017Piunti et al., 2017Zhang et al., 2017). In this regard, preclinical efficacy has been shown for the BET inhibitors JQ1, I-BET151 and OTX-015, and the CDK7 inhibitor THZ1, both in vitro and in vivo, although resistance to JQ1 eventually developed as well, and cross-resistance between JQ1 and panobinostat has been reported, limiting the potential of this type of combination therapy (Nagaraja et al., 2017Piunti et al., 2017Zhang et al., 2017). Furthermore, the polycomb repressor complex 1 (PRC1) subunit BMI1 has been shown to be overexpressed in DMG, possibly as a result of H3K27 M mutations (Kumar et al., 2017). Although the role of BMI1 in this regard has not been extensively investigated yet, its inhibition by CRISPR knockout or the small molecule PTC-209 resulted in partial restoration of H3K27 trimethylation and in vitro antitumor efficacy (Filbin et al., 2018Kumar et al., 2017).

Importantly, combinations of multiple epigenetic modifiers, targeting different components of the chromatin remodeling machinery and transcriptional activation, has shown promising synergistic antitumor effects in several studies (Table 1). Therefore, epigenetic modification represents a highly promising target candidate for the treatment of DMG, although further preclinical and clinical studies are required to fully understand the underlying molecular mechanisms and allow for tumor-specific treatment.

Growth factor receptors

Among the first genetic aberrations and therapeutic targets identified in diffuse midline gliomas were amplifications and overexpression of receptor tyrosine kinases (RTKs, Table 2). Most importantly, the epidermal growth factor receptor (EGFR) and platelet-derived growth factor receptor α (PDGFRα) genes are frequently amplified and overexpressed in these tumors. At a lower frequency, gene amplification in other RTKs has also been identified, including the insulin-like growth factor 1 receptor (IGF1R), hepatocyte growth factor receptor (MET), stem cell-factor receptor (KIT) and vascular-endothelial growth factor receptor 2 (VEGFR2/KDR) (Buczkowicz et al., 2014abFilbin et al., 2018Fontebasso et al., 2014Gilbertson et al., 2003Grill et al., 2012Hoffman et al., 2016Mackay et al., 2017Paugh et al., 20112013Vinci et al., 2018Warren et al., 2012Zarghooni et al., 2010). Additionally, next generation sequencing (NGS) studies have identified activating mutations in PDGFRα, the activin receptor 1 (ACVR1) and fibroblast growth factor receptor 1 (FGFR1), the latter of which occurs mainly in DMG located in the thalamus (Buczkowicz et al., 2014abFontebasso et al., 2014Grill et al., 2012Hoffman et al., 2016Mackay et al., 2017Paugh et al., 2013Ryall et al., 2016Taylor et al., 2014Vinci et al., 2018). Based on these discoveries, inhibitors of these RTKs were tested in a series of preclinical studies (Fig. 2). Although inhibition of PDGFRα and MET effectively inhibited proliferation and migration of DMG cells in an in vitro study (Truffaux et al., 2015), the results from other studies were more ambiguous, and only showed relevant antitumor effects of multi-kinase inhibitors such as dasatinib and crizotinib (Grasso et al., 2015Meel et al., 2017). Interestingly, knocking out PDGFRα in H3K27 M DMG cells by CRISPR/Cas9 displayed a strong impact on cell survival (Filbin et al., 2018). Only one study tested dasatinib in vivo, employing a murine DMG model driven by PDGFB overexpression and loss of p53 (Mittapalli et al., 2016). Although some survival benefit of dasatinib was seen in this model, the efficacy was severely limited by multidrug efflux transporters overexpressed on the plasma membrane of tumor cells. Results of preclinical in vitro studies evaluating inhibitors targeted at EGFR, IGF1R, KIT and VEGFR2 yielded comparable results, with most efficacy observed with multikinase inhibitors and a global failure of more specific inhibitors (Grasso et al., 2015Meel et al., 2017). Only one other multi-targeted RTK inhibitor, BMS-754807, has been tested in vivo, using the same PDGFB;p53−/− murine DMG model as for dasatinib. Again, treatment with this multikinase inhibitor did not result in a survival benefit for the mice, although inadequate BBB penetration may have played a role here (Halvorson et al., 2015). Parallel to the preclinical studies, two clinical trials have been performed using multitargeted RTK inhibitors for the treatment of children with DMG. In line with the failure of these drugs to achieve relevant antitumor effects in vivo, no survival benefit was seen for RTK inhibition in these trials (Broniscer et al., 20132018). One phase 2 trial using nimotuzumab, a monoclonal antibody targeting EGFR, suggested the potential existence of a small subgroup of patients benefiting from this type of therapy, although these results could not be confirmed in a subsequent phase 3 trial (Bartels et al., 2014Fleischhack et al., 2019). FGFR1 mutations may still represent a valuable therapeutic target in a subset of DMG patients, but to date no preclinical or clinical studies have tested FGFR1 inhibition as a treatment strategy in these tumors. Nonetheless, a recent study identified FGFR signaling as a therapeutic target in DMG in vitro, independent of mutations in FGF receptors, emphasizing the need for further research on the potential of FGFR inhibition in DMG (Schramm et al., 2019). Finally, ephrin receptors are often activated in DMG cells, and in vitro experiments show antitumor efficacy of inhibitors of these receptors, even in the absence of mutations and gene amplifications (Meel et al., 2017Nagaraja et al., 2017Truffaux et al., 2015). Most importantly, ephrins have been shown to be essential for glioma cell migration and invasion, and their blockade inhibits these processes (Nagaraja et al., 2017Truffaux et al., 2015). Several explanations exist for the overall failure of targeting RTKs for the treatment of DMG. Firstly, recent studies have shown that gene amplification and mutations of PDGFRα are subject to intratumoral and subclonal heterogeneity, which may explain the lack of efficacy of PDGFRα inhibition (Filbin et al., 2018Hoffman et al., 2016Vinci et al., 2018). Although not explicitly studied, the same may be true for other RTKs, explaining the failure of therapeutic strategies based on RTK inhibition. Secondly, the presence of an intact BBB, as well as multidrug efflux transporters on the tumor cells, may contribute to the failure of RTK inhibition, simply because adequate drug concentrations are not achieved in tumor cells (Halvorson et al., 2015Mittapalli et al., 2016Veringa et al., 2013Warren, 2018). Thirdly, in vitro drug efficacy studies are prone to yield false positive or false negative results, as a consequence of the chosen culture methods of tumor cells, and because kinase inhibitors rarely target a single kinase, but are subject to off-target effects, which can skew the balance of efficacy versus toxicity in an unfavorable direction (Meel et al., 2017). Finally, tumor cells may rapidly develop escape mechanisms for targeted therapies by switching their dependency from one signaling pathway to another (Tan et al., 2017). To overcome these mechanisms of resistance and treatment failure of RTK inhibitors, one study suggested targeting NG2, a transmembrane protein that stabilizes a variety of RTKs, which is upregulated in DMG and essential for DMG cell survival (Fig. 2) (Yadavilli et al., 2015). However, evidence regarding the feasibility of such a therapeutic approach would have to be generated in vivo, which has not been done so far.

Table 2. Growth factor receptors as therapeutic targets in diffuse midline glioma.

TargetIn vitro efficacyIn vivo efficacyRemarksReferences
PDGFRα+/-Ambiguous results, multikinase inhibition essentialGrasso et al., 2015Truffaux et al., 2015Mittapalli et al., 2016Meel et al., 2017Filbin et al., 2018
EGFR+/-n/aAmbiguous results, multikinase inhibition essentialBartels et al., 2014Grasso et al., 2015Meel et al., 2017Fleischhack et al., 2019
IGF1R+/-Ambiguous results, multikinase inhibition essentialGrasso et al., 2015Halvorson et al., 2015Nagaraja et al., 2017Meel et al., 2017
MET+/-Ambiguous results, multikinase inhibition essentialGrasso et al., 2015Halvorson et al., 2015Truffaux et al., 2015Meel et al., 2017
KITn/aMultikinase inhibition necessaryGrasso et al., 2015
VEGFR2+/-n/aMultikinase inhibition necessaryMeel et al., 2017
FGFR1+n/aEfficacy of inhibition not evaluatedSchramm et al., 2019
ACVR1+/-+In vitro results ambiguousTaylor et al., 2014,.Meel et al., 2017Hoeman et al., 2019
Ephrins+n/aMultikinase inhibition likely essentialTruffaux et al., 2015Nagaraja et al., 2017Meel et al., 2017
NG2+n/aYadavilli et al., 2015

n/a: data not available.

Fig. 2

The situation may be different for therapeutic strategies targeting ACVR1, which is mutated in a subset of DMG patients, mainly those carrying H3.1 K27M mutations (Buczkowicz et al., 2014abFontebasso et al., 2014Mackay et al., 2017Taylor et al., 2014). The same studies identifying intratumoral and subclonal heterogeneity for PDGFRα show that ACVR1 mutations are generally conserved among all tumor cells (Hoffman et al., 2016Nikbakht et al., 2016Vinci et al., 2018). Additionally, ACVR1 mutations are associated with the mesenchymal phenotype in DMG, and their role in the mesenchymal transition may contribute to therapy resistance in these tumors (Castel et al., 2015Hoeman et al., 2019Meel et al., 2018bPuget et al., 2012). Despite limited antitumor efficacy in vitro of monotherapy with ACVR1 inhibitors (Meel et al., 2017Taylor et al., 2014), efficacy could be demonstrated in vivo in genetically engineered mice carrying ACVR1 mutant murine DMG tumors (Hoeman et al., 2019). As such, combination therapies incorporating ACVR1 inhibition may be an effective approach in this subset of DMG.

Signal transduction pathways

Like many other types of cancer, DMG is characterized by frequent aberrations involving signal transduction pathways downstream of cell surface receptors. Although no single event is highly prevalent in these tumors, studies have identified mutations in both PIK3CA and PIK3R1, gene amplifications in HRAS and loss of PTEN, all of which resulting in aberrant phosphatidyl-inositol-3 kinase (PI3K) and/or mitogen-activated protein kinase (MAPK) pathway activation (Buczkowicz et al., 2014bFontebasso et al., 2014Grill et al., 2012Mackay et al., 2017Nikbakht et al., 2016Puget et al., 2012Taylor et al., 2014Vinci et al., 2018Warren et al., 2012). Consequently, preclinical studies have investigated the antitumor potential of inhibitors of core components of these pathways (Table 3 and Fig. 2). Based on the identification of activating PI3K mutations, several studies have been conducted to determine the preclinical efficacy of PI3K inhibitors, although all of these studies were performed using combination therapies or compounds with multiple targets. Two studies evaluated the dual PI3K/Akt inhibitor perifosine, and found it to be effective only in vitro (Becher et al., 2010) or in combination with the MEK inhibitor trametinib (Wu et al., 2017). In this respect, a phase 1 clinical trial did not demonstrate a survival benefit in DMG patients, although the number of patients included was very low (n = 3) (Becher et al., 2017b). In a functional drug screen, some in vitro efficacy was observed with the dual PI3K/mTOR inhibitor dactolisib (BEZ235), but this compound has not yet been evaluated in vivo (Grasso et al., 2015). Another recent study demonstrated promising preclinical efficacy of the dual PI3K/HDAC inhibitor fimepinostat (CUDC-907), both in vitro and in vivo, although the contribution of PI3K inhibition to the treatment efficacy cannot be deduced from this study (Pal et al., 2018).

Table 3. Signal transduction pathways as therapeutic targets in DMG.

TargetIn vitro efficacyIn vivo efficacyRemarksReferences
PI3K+/-+/-Monotherapy PI3K inhibition not tested, multikinase inhibition likely essentialBecher et al., 2010Grasso et al., 2015Wu et al., 2017Pal et al., 2018
AKT+/-Monotherapy AKT inhibition not tested, multikinase inhibition likely essentialBecher et al., 2010Hoeman et al., 2019
MTORC1/2+/-+/-Inhibition of both MTORC1 and 2 likely essentialGrasso et al., 2015Nagaraja et al., 2017Meel et al., 2017Miyahara et al., 2017Asby et al., 2018Flannery et al., 2018Tsoli et al., 2018
MEK1/2+/-n/aAmbiguous results, multikinase inhibition likely essentialBecher et al., 2010Grasso et al., 2015Meel et al., 2017Wu et al., 2017
ERK1/2+/-n/aAmbiguous results, multikinase inhibition likely essentialNagaraja et al., 2017
PP2A+n/aParadoxical efficacy by overactivation of PI3K and MAPK pathwaySchramm et al., 2019

n/a: data not available.

Activation of the PI3K pathway results in downstream activation of the mammalian target of rapamycin complex (mTORC). As such, mTOR inhibitors have also been investigated for the treatment of DMG. Whereas in vitro efficacy of some mTOR inhibitors, especially those targeting both mTORC1 and mTORC2, has been reported by multiple groups (Asby et al., 2018Flannery et al., 2018Grasso et al., 2015Meel et al., 2017Nagaraja et al., 2017), in vivo efficacy has been observed only for the mTORC1/2 inhibitor TAK228 in a murine DIPG model, whereas treatment of human DMG xenograft-bearing mice with the mTORC1 inhibitor temsirolimus was ineffective (Miyahara et al., 2017Tsoli et al., 2018). The precise reason for this discrepancy may lie in a) the need to inhibit both mTORC1 and mTORC2, b) the differential BBB penetration of temsirolimus and TAK228, or c) an intrinsic difference in sensitivity to mTOR inhibition between the models used in these studies. Based on the lack of antitumor efficacy of temsirolimus in a phase 1 clinical trial (Becher et al., 2017a), further studies are required to assess the potential of mTOR inhibition for the treatment of DMG.

As a result of the activation of RTKs and HRAS gene amplification, the MAPK pathway is frequently activated in DMG as well. Therefore, inhibition of key components of this pathway, such as MEK1/2 and ERK1/2, has received attention in preclinical therapeutic studies in these tumors. Although in vitro studies show that DMG cells are, to some extent, sensitive to inhibition of MEK1/2 and ERK1/2, the results are ambiguous, and the observed efficacies may be a result of the tumor cell culture method used (Grasso et al., 2015Meel et al., 2017Nagaraja et al., 2017Wu et al., 2017). To this day, no studies have been performed that evaluate the antitumor efficacy of MEK or ERK inhibitors in vivo, and further research is needed to definitively assess the preclinical potential of these drugs.

In contrast to the traditional therapeutic strategies aimed at inhibiting oncogenic PI3K and MAPK signaling, a recent study identified protein phosphatase 2 (PP2A) as a therapeutic target in DMG in vitro (Schramm et al., 2019). PP2A is responsible for the dephosphorylation of AKT, RAF and MEK, among others, with its inhibition resulting in overactivation of these pathways and subsequent apoptosis, as demonstrated for DMG (Schramm et al., 2019). While interesting, the in vivo efficacy of such a therapeutic strategy is yet to be demonstrated in DMG, as well as potential combination therapies using other classes of cytotoxic agents.

Cell cycle checkpoints

The deregulation of cell cycle progression is a common feature of many types of cancer; as such, it is not surprising that DMG possesses several aberrations in cell cycle checkpoints which are potential therapeutic targets (Table 4 and Fig. 3). H3K27 M mutations, being the hallmark of DMG, result in epigenetic repression of p16INK4A (CDKN2A), an important regulator of the G1/S checkpoint (Cordero et al., 2017Mohammad et al., 2017). Furthermore, chromosomal gains of regions containing cyclin-dependent kinases 4 and 6 (CDK4/6) and associated cyclins, all essential at the G1/S checkpoint, have been detected in DMG patient specimens (Paugh et al., 2011). As a result, DMG cells appear to be sensitive to pharmacological inhibition of CDK4/6 in vitro (Asby et al., 2018Grasso et al., 2015). Moreover, one study showed a significantly prolonged survival of genetically engineered mice with DMG, upon treatment with palbociclib (PD-0332991) (Barton et al., 2013). However, the murine DMG model used carries a genomic deletion of the Ink4-ARF locus, which does not occur in DMG patients and may cause susceptibility to CDK4/6 inhibition. Nonetheless, based on the epigenetic repression of p16INK4A in DMG, CDK4/6 inhibition may represent a therapeutic strategy in these tumors that warrants further investigation.

Table 4. Cell cycle checkpoints as therapeutic targets for DMG.

TargetIn vitro efficacyIn vivo efficacyRemarksReferences
CDK4/6++/-Synergy with mTORC1/2 inhibition, in vivo data generated in not fully representative DMG modelBarton et al., 2013Grasso et al., 2015Asby et al., 2018
WEE1++Limited BBB penetration of inhibitor usedCaretti et al., 2013a2013b
Mueller et al., 2014De Gooijer et al., 2018
AURKB+/-AURKB in vivo only tested with multikinase inhibitor, AURKA inhibition possibly more effectiveBuczkowicz et al., 2013Grasso et al., 2015Halvorson et al., 2015
PLK1+n/aAmani et al., 2016Schramm et al., 2019
MELK++In vivo efficacy shown in Mdr1a/b−/−;Bcrp1−/− mice, as inhibitor did not cross BBBMeel et al., 2018a2018b

n/a: data not available.

Fig. 3

Apart from deregulation of the G1/S checkpoint, DMG was found to overexpress several key players in G2/M progression as well including WEE1, maternal embryonic leucine zipper kinase (MELK), polo-like kinase 1 (PLK1) and aurora kinase B (AURKB), all of which have been studied as potential therapeutic targets (Amani et al., 2016Buczkowicz et al., 2013Caretti et al., 2013bMeel et al., 2018aMueller et al., 2014). Inhibition of WEE1 by the small molecule MK1775 (AZD1775) was shown to be effective in treating and radiosensitizing DMG cells. Importantly, treatment of mice carrying patient-derived xenografts of DIPG and other diffuse gliomas, resulted in improved survival, especially when combined with radiotherapy (Caretti et al., 2013aMueller et al., 2014). These results are in apparent discordance with the limited BBB penetration of AZD1775, implying that either the BBB is disrupted in the xenograft mouse models, or that very low levels are sufficient for an antitumor effect (de Gooijer et al., 2018). Further studies are therefore required to determine the definitive potential of WEE1 inhibition, especially in light of the development of novel WEE1 inhibitors with superior brain bioavailability (de Gooijer et al., 2018).

Inhibition of another G2/M checkpoint kinase, AURKB, resulted in some antitumor efficacy in DMG cells in vitro, but so far no successful in vivo trials using specific inhibitors have been performed (Buczkowicz et al., 2013). Only one in vivo trial has been performed with a multi-kinase inhibitor that has aurora kinases among its target, and this failed to show preclinical efficacy, possibly as a result of poor BBB penetration of the compound chosen (BMS-754807) (Halvorson et al., 2015). Strikingly, a functional drug screen failed to show efficacy of AURKB inhibitors in a panel of DMG cultures, whereas inhibition of its paralog AURKA seemed to be effective (Grasso et al., 2015). As such, the role of aurora kinases as therapeutic targets in DMG requires further studies, also to elucidate the potential translational relevance of aurora kinase inhibitors.

Although extensively studied in other types of cancer as a gatekeeper of the G2/M transition, only one study has described in vitro efficacy of direct PLK1 inhibition in DMG (Amani et al., 2016). As in vivo data are lacking, it is not possible to draw any conclusions regarding the potential of PLK1 inhibition for the treatment of DMG. As with PI3K and MAPK signaling, enhancing PLK1 activity by inhibiting PP2A has also been suggested as therapeutic strategy, although the contribution of PLK1 stimulation to the antitumor effect of PP2A inhibition is yet to be elucidated (Schramm et al., 2019).

Finally, a recent study by our group has identified MELK as a potential therapeutic target in DMG, showing in vitro efficacy of the small molecule OTSSP167, and shRNA targeting MELK, in a panel of primary DMG cultures (Meel et al., 2018a). Although this study did not address the influence of MELK inhibition on the cell cycle progression of DMG cells, MELK has been described as an important regulator of G2/M progression, which likely contributes to the antitumor efficacy of its inhibition. Unfortunately, brain pharmacokinetic studies showed that OTSSP167 does not cross the BBB due to the activity of multidrug efflux transporters. Nevertheless, treatment of Mdr1a/b−/−;Bcrp1−/− mice harboring patient-derived DMG xenografts with OTSSP167 resulted in a strong antitumor activity and increased survival, validating MELK as a therapeutic target in these tumors and encouraging the development of novel, MELK inhibitors that penetrate the BBB (Meel et al., 2018a).

Stem cell phenotype

Starting from the very first studies into the biology and oncogenesis of DMG, evidence has arisen that these tumors develop in primitive neural progenitor cells (NPCs) or oligodendrocyte progenitor cells (OPCs) in the fetal brainstem, resulting in stem cell-like tumor cells with high levels of therapy resistance (Anderson et al., 2017Funato et al., 2014Larson et al., 2019Monje et al., 2011Pathania et al., 2017). Consequently, attempts have been made to target the stem cell phenotype of DMG cells as a therapeutic strategy, as these stem cell characteristics are held responsible for at least part of the therapy resistance of DMG cells (Table 5).

Table 5. The stem cell phenotype, DNA damage repair pathway and cell surface antigens as (immune)therapeutic targets for diffuse midline glioma.

TargetIn vitro efficacyIn vivo efficacyRemarksReferences
SHH/PTCH1+Monje et al., 2011Grasso et al., 2015
NOTCH+Radiosensitization upon NOTCH pathway inhibitionTaylor et al., 2015
PARP++/-Mouse models used in in vivo study not fully representative of DMG, Radiosensitization by PARP inhibition, Possible PARP inhibition-sensitive subclones in DMGChornenkyy et al., 2015Vinci et al., 2018
PPM1D++Only effective in PPM1D mutated, p53 wild-type DMG; Radiosensitization by PPM1D inhibition in DMG cells with that specific genomic profileAkamandisa et al., 2019
GD2++CAR T-cell targetMount et al., 2018
B7-H3++/-CAR T-cell target, in vivo efficacy only in other pediatric brain tumor modelsMajzner et al., 2019

The first such study showed that DMG cells originate from Sonic hedgehog (SHH) expressing OPCs, and retain their dependence on hedgehog signaling, resulting in antitumor efficacy of hedgehog signaling inhibitors in vitro (Monje et al., 2011). Another study corroborated these observations by demonstrating overexpression of the SHH receptor Patched 1 (PTCH1) in a subgroup of DMG, and differential methylation of genes involved in hedgehog signaling compared to normal brain tissue (Saratsis et al., 2014). The in vitro efficacy of SHH inhibition has since been demonstrated once more, but in vivo trials studying the therapeutic potential of these drugs are lacking (Grasso et al., 2015). One study demonstrated overexpression of NOTCH and activated downstream signaling of the Notch pathway in DMG, which may also result in the stem cell phenotype of DMG cells (Taylor et al., 2015). Inhibition of Notch signaling resulted in decreased tumor cell viability and radiosensitization of DMG cells (Taylor et al., 2015). As with SHH inhibition, in vitro trials studying the therapeutic effect of Notch inhibitors are lacking, warranting further investigation.

Additionally, our study which identified MELK as a therapeutic target in DMG, further enforced the potential of targeting the stem cell phenotype of DMG cells, as MELK has been shown to be essential for cancer stem cell maintenance, and expression patterns of MELK during embryonic development correspond to the putative time of origin of DMG (Meel et al., 2018a). Finally, telomerase maintenance mechanisms, an important hallmark of both normal and malignant stem cells, have been proposed as therapeutic targets in DMG, although inhibition of telomerase and related enzymes has not been pursued for the treatment of DMG to date (Dorris et al., 2014).

DNA damage repair

Like the majority of other cancer types, DMG is characterized by a high prevalence of defects in DNA damage repair systems. Most importantly, p53 function is often compromised by loss-of-function mutations, genomic deletions or indirectly as a result of PPM1D mutations (Aihara et al., 2014Akamandisa et al., 2019Buczkowicz et al., 2014bFontebasso et al., 2014Grill et al., 2012Hoffman et al., 2016P. International Cancer Genome Consortium PedBrain Tumor, 2016Khuong-Quang et al., 2012Mackay et al., 2017Nikbakht et al., 2016Puget et al., 2012Saratsis et al., 2014Taylor et al., 2014Vinci et al., 2018Warren et al., 2012). Furthermore, mutations or losses involving ATM or ATRX are found in subsets of DMG patients (Aihara et al., 2014Buczkowicz et al., 2014bFontebasso et al., 2014Grill et al., 2012Hoffman et al., 2016P. International Cancer Genome Consortium PedBrain Tumor, 2016Khuong-Quang et al., 2012Mackay et al., 2017Nikbakht et al., 2016Saratsis et al., 2014Taylor et al., 2014Vinci et al., 2018). Excluding the use of traditional chemotherapeutic agents, a few studies have been performed that attempt to exploit these DNA damage repair deficiencies for the development of therapeutic strategies for DMG (Table 5). Nonetheless, inhibition of poly-ADP-ribose polymerase (PARP), involved in single strand break repair, may be effective in the presence of certain DNA damage repair deficiencies, especially as it is overexpressed in a subset of DMG (Chornenkyy et al., 2015Zarghooni et al., 2010). As such, PARP inhibition has been studied as a potential therapeutic intervention in DMG, resulting in radiosensitization and antitumor efficacy in vitro and in vivo, although the mouse models used in this study are not fully representative of DMG (Chornenkyy et al., 2015). Interestingly, a recent study has identified the potential presence of subclones in DMG that are highly responsive to PARP inhibition, warranting further investigation of these compounds in combination treatment strategies (Vinci et al., 2018). More specifically, a recent study has shown that targeting PPM1D effectively treats PPM1D mutated DMG in vitro and in vivo, especially when combined with radiotherapy (Akamandisa et al., 2019). These results indicate that PPM1D inhibitors represent a promising therapeutic strategy for the subset of DMG that carries PPM1D mutations in the presence of wild-type p53. As such, exploiting DNA damage repair deficiencies for the treatment of DMG may represent a promising strategy for the treatment of these tumors, despite the limited number of studies on this topic.

Targets for immunotherapy

In recent years, immunotherapy has emerged as a promising new treatment modality for various types of cancer, including adult glioma (Dunn-Pirio and Vlahovic, 2017Hodges et al., 2016Polivka et al., 2017). As a result, similar approaches are being explored for the treatment of DMG. However, the lack of immunocompetent mouse models of DMG hampers the study of immunotherapy for these tumors. An exception is formed by chimeric antigen receptor (CAR) T-cell therapy, in which T-cells are genetically engineered ex vivo to target surface antigens on tumor cells (Table 5). By using donor T-cells, these therapies can be tested in immunodeficient, xenograft-bearing mice, although the lack of a recipient immune system still imposes a limitation in the interpretation of the results of such trials. Nonetheless, impressive results have been obtained in the treatment of DMG xenograft-bearing mice with CAR T-cells targeted at GD2 or B7-H3 (Majzner et al., 2019Mount et al., 2018Zhou et al., 2013). Although the latter has not been specifically studied in DMG mouse models, the extensive data on using GD2 CAR T-cells in DMG, and B7-H3 in vitro and in other pediatric brain tumor models in vivo, makes it likely to be effective as well. A definitive assessment of the potential of CAR T-cell therapy and other forms of immunotherapy for the treatment of DMG requires the development of immunocompetent DMG mouse models. The development of these models is also essential to investigate the potential development of resistance to immunotherapy. Until th

Media Contacts:
Steven Benowitz – NIH/NCATS
Image Source:
The image is credited to Shawn Gillespie, Monje Laboratory, Stanford Medicine.

Original Research: Closed access
“Therapeutic strategies for diffuse midline glioma from high-throughput combination drug screening”. Michelle Monje et al.
Science Translational Medicine doi:10.1126/scitranslmed.aaw0064.


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.