Recent Advancements in Ependymoma: Challenges and Therapeutic Opportunities

Background: Ependymoma is one of the most common malignant pediatric brain tumors and can be difficult to treat. Over the last decade, much progress has been made in the understanding of the underlying molecular drivers within this group of tumors, but clinical outcomes remain unchanged. Summary: Here, we review the most recent molecular advances in pediatric ependymoma, evaluate results of recent clinical trials and discuss the ongoing challenges in the field and questions that remain. Key Messages: The field of ependymoma has vastly changed over the last several decades with ten distinct molecular subgroups now described, but much progress needs to be made in developing new therapeutic strategies and targets.


Introduction
Pediatric brain tumors are the leading cause of cancerrelated death in children, despite aggressive management with multimodal therapies [1].Ependymoma is one of the most common malignant pediatric brain tumors and can be found anywhere throughout the central nervous system, in both children and adults.Historically, ependymoma was thought to be one entity, but over time, has evolved into ten molecularly distinct groups, with further subdivisions that have emerged [2].While these subgroups of ependymoma are molecularly heterogenous, they also have different demographics and clinical features, thus studying and understanding the biological drivers is essential to develop targeted therapies, and appropriate stratification in clinical trials [2].Major advances in understanding the drivers of ependymoma have been made over the last several years, but breakthroughs to improve clinical outcomes are still desperately needed.This review focuses on more recent areas of research, ongoing challenges in the field, and opportunities to uncover new therapies.Supratentorial Ependymoma: A Disease Driven by ZFTA (C11orf95) and YAP1 Fusions?Supratentorial ependymoma is comprised of two main molecular groups -ZFTA fusion (formerly C11orf95), and YAP1 fusion tumors.Over 70% of supratentorial ependymoma carry the ZFTA-RELA gene fusion, a potent driver of cancer capable of transformation alone when expressed in neural stem cells.RELA is a transcription factor central to the NF-kB pathway, and NF-kB target genes are enriched in ZFTA-RELA tumors.However, expression of RELA alone cannot initiate tumorigenesis, thus research efforts have shifted toward understanding the function of ZFTA (formerly C11orf95).Supporting the critical role of ZFTA are recent findings that demonstrate ZFTA fuses with other partners, such as MAML2/3, NCOA1/2, and CTNNA2, often transcription factors or co-activator proteins [3].These distinct ZFTA-fusion partners are also drivers of ependymoma and robustly transform cells with implanted in mice [3].
ZFTA-fusion variants molecularly resemble ZFTA-RELA tumors; however, they form distinct subtypes when profiled by Illumina DNA methylation profiling.Interestingly, several groups have described distinct histological features of ZFTA-fusion variants [3][4][5].These studies highlight the functional importance of ZFTA in the neoplastic process of supratentorial ependymoma and underscore the additional tumor heterogeneity that is seen within ST ependymoma.The clinical significance and outcomes of ZFTA-fusion variants remain unclear.
In a multicenter collaboration across three independent studies, critical insight was recently revealed about the pathogenesis of ZFTA-fusion-driven ependymoma.It was shown that the tumor epigenome is modulated through de novo binding of DNA mediated by ZFTA at specific PLAGL transcription factor motifs, which is distinct from the NFkB pathway.ZFTA-RELA binding activates gene expression programs through recruitment of transcriptional co-activators such as histone acetylation readers (BRD4) and writers (Ep300 and CBP), along with RNA polymerase 2 [6].These findings along with the identity of ZFTA-fusion partners point to a key function of ZFTA acting as an oncogenic transcription factor [4,6].This is perhaps analogous to other gene fusions and potent oncogenic drivers delineated in other cancers, such as EWS-FLI1 in Ewing's Sarcoma and PAX3-FOXO1 fusions in rhabdomyosarcoma.An area of active investigation is whether epigenetic co-regulators serve as effective targets in fusion-driven ependymoma, which can be effectively inhibited with small molecules (i.e., BET inhibitors).
Collectively, these recent findings place a spotlight on ZFTA as a central component of neoplastic transformation, and recruitment of other gene partners necessary for transcriptional activation.Critically, loss of a zinc finger binding domain of ZFTA, completely abrogates tumor formation across all ZFTA-fusion variants.This highlights the relevance of aberrant nuclear translocation and gene activation, toward initiation of ependymoma [7].However, key future experiments will reveal whether ZFTA-fusion variants are still required for tumor progression or function only during the initiation process (i.e., a hit-and-run event).As such, future research is also needed on assessing the growth dependencies of ZFTAfusion ependymoma in relevant patient-derived xenograft models.These efforts may be complemented with research to delineate the function of additional drivers such as CDKN2A loss, which is observed in ~30% of ST-RELA ependymomas.
YAP1 fusions are much less common in ST ependymoma and are associated with favorable clinical outcome (with a caveat that these clinical findings have been in small retrospective cohorts).Although clinical outcomes are generally better than ZFTA-fused tumors, YAP1 fusions are also potent drivers of transformation.YAP1 tends to be fused with MAMLD1, although have also been found to be fused with FAM118B [4,8].YAP1-MAMLD1 fusions have been shown to be sufficient to drive oncofusion nuclear localization and tumorigenesis in vivo [8].Through recent studies, it has been suggested that this fusion acts through recruitment of TEADS (TEA Domain Transcription Factors), which may drive tumorigenesis via hyperactivation of YAP signaling [8].The molecular and transcriptional basis of YAP1 fusions remains unresolved; however, these data may provide rationale for testing of YAP1 and next-generation TEAD inhibitors and degraders in ependymoma.

Posterior Fossa Ependymoma: A Disease of Defective Development and Metabolism?
Posterior fossa ependymomas account for two-thirds of ependymomas and can be subdivided into two main entities that are epigenetically and transcriptionally distinct, named Posterior Fossa Group A (PF-A) and Posterior Fossa Group B (PF-B) [9].PF-A tumors tend to have a balanced genome, with the exception of 1q gain and/or 6q loss, which are infrequent events (15% and 9%, respectively) [10].While 1q gain has been shown in numerous prospective trials to be associated with poor prognosis [11,12], 6q loss has only been recently identified as a high-risk feature associated with rapid progression and virtually universal relapse, in a large retrospective study [10].The underlying biology of the driving features on either chromosome 1q or 6q remains unknown; however, they are critical markers to predict disease relapse.
Exhaustive sequencing efforts have shown that PF-A ependymomas have no recurrent fusion genes, recurrent mutations, or focal copy number alterations.This is counterintuitive given the aggressive nature of these tumors compared to other ependymoma subtypes.
Despite "quiet" genomes, ependymomas harbor defective epigenomes in the form of (1) global DNA methylation depletion, (2) CpG island hypermethylation, and (3) loss of the repressive chromatin mark, H3K27me3 [13,14].Indeed, these findings have highlighted ependymoma as a disease which may be entirely epigenetically driven.This has reshaped our approach to treatment of this disease, as traditional mutationdrug pairing (i.e., precision medicine) is not possible in PF-A (or PF-B) ependymoma.
PF-A tumors are characterized by global loss of the repressive histone 3 lysine 27 trimethylation (H3K27me3), caused by aberrant expression of Enhancer of zeste homolog inhibitory protein (EZHIP).EZHIP is a protein not normally expressed in the brain (at least defined by murine studies) and restricted to germ cells during development.The expression of EZHIP may be entirely different in human developing posterior fossa, which has not yet been defined.EZHIP functions to mimic the H3K27M mutation seen in DMG; a natural inhibitor that blocks function of the PRC2 complex which is critical for H3K27me3 deposition and gene repression [14,15].While there have been rare cases of EZHIP and H3K27M mutations seen in PF-A ependymoma, the lack of recurrent mutations supports a hypothesis of epigenetically reprogrammed cell types during development.
The unique epigenome of PF-A ependymoma cells presents potential therapeutic opportunities.Indeed, PF-A ependymomas have been shown to grow preferentially in hypoxic environments, which may reflect more accurate tumor microenvironments of tumor cells in vivo.Metabolic and epigenetic programs are integrated circuits and cofactors for epigenetic modifications are derived from metabolic inputs.PF-A ependymoma cells have been shown to be highly dependent on specific metabolic programs such as methionine, glucose, and glutamine regulation.Inhibition of these pathways using new drug compounds, i.e., MAT2A inhibitors or old drugs such as Metformin, has shown promise [16].However, rigorous testing of these findings in vivo has been hampered by a lack of robust PF-A genetic or PDX mouse models.Until paired advancements can be made in PF-A ependymoma mouse modeling, translation of novel treatments to clinical trials will remain a slow process.
Unlike PF-A ependymoma, PF-B tumors have favorable clinical outcomes, which have been reproduced across multiple independent clinical trial cohorts [11,12].PF-B ependymomas are characterized by chromosomal arm-level gains and losses, such as chromosome 2 loss, 5 gain, 17 loss, 1q gain, in addition to high levels of H3K27me3.Indeed, PF-B ependymomas are described as having a chromosomal instability phenotype (CIN),

Advancements in Ependymoma
Pediatr Neurosurg 2023;58:307-312 DOI: 10.1159/000530868 which in other solid tumor contexts has been associated with better survival.This phenotype may be interconnected with potential distinct cellular origins of PF-B ependymoma, which are characterized as having strong enrichment of ependymal cell differentiation programs, as opposed to progenitor programs enriched in PF-A and ST tumors.Where 1q gain is important for prognosis in PF-A tumors that does not seem to hold true in PF-B ependymoma [17,18].Additional heterogeneity in the form of molecular subtypes is seen in both PF-A and PF-B tumors; however, the underlying molecular basis is unclear, as are the clinical ramifications [19].

Spinal Ependymoma
Spinal ependymomas are generally observed in older children through adulthood; however, myxopapillary ependymomas have been observed in aggressive subsets in the pediatric population.Spinal ependymomas can be subdivided into four distinct groups -Myxopapillary ependymoma (SP-MPE), subependymoma (SP-SE), ependymoma (SP-EPN), and MYCN amplified spinal ependymoma (SP-MYCN) [2].While a large percentage of spinal ependymomas (SP-EPN) harbor chromosome 22q loss, some mutations have also been identified, including NF2, RP1, and ESX1 [2].SP-MYCN tumors have emerged as a distinct group with MYCN amplification, with diffuse leptomeningeal spread, and worse progression-free (PFS) and overall survival (OS) [20].Myxopapillary ependymomas are distinct from other groups, as they have elevated protein expression of key metabolic factors, such as HIF1a, HK2, PDK1, and elevated lactate, suggesting that MPE may be driven by a Warburg phenotype [21].Interestingly, these pathways are also highly enriched in PF-A ependymomas, suggesting intersectional molecular programs.

Ependymoma Clinical Outcomes in the Molecular Era
Irrespective of molecular subgroups, the significant predictors of survival in ependymoma include extent of resection, radiotherapy, and location.Molecular subgroup classification (namely, DNA methylation profiling) has enhanced predictive measures to distinguish poor versus favorable outcomes of ependymoma patients.

Supratentorial Ependymoma
As a group, supratentorial ependymomas historically have a 5-year OS of around 70-75%.This further decreases to 50% by 10 years, with many late relapses [11,12].While ZFTA-driven tumors tend to occur in older children (median age 8 years), YAP1-driven tumors happen much earlier, with a median age of 1 year, and in retrospective review, tend to have a more favorable prognosis [8].As more fusion variants are annotated and this small entity gets further subdivided, larger cohorts will be needed to study these fusion variants.While there have been several small case reports on ZFTA-fusion variant outcomes, these have not been powered for statistical significance, given the rarity of these variants.A large prospective trial is needed to delineate if there are any differences within this group.
Posterior Fossa Ependymoma PF-A is inherently in younger children and exhibits more aggressive features, with worse clinical outcomes, with PFS around 60-70% and 10-year OS being only 56% [15].The gain of 1q has been shown across numerous studies to be associated with poor prognosis, and while it is only present in approximately 15-20% of PF-A patients, the PFS is 20% [10].In a recent retrospective study, loss of 6q was associated with even poorer prognosis, with a 5-year OS less than 10% [10].PF-B tumors are typically seen in older children and adults and have an excellent prognosis [10,15].

Spinal Ependymoma
Spinal ependymomas can be clinically heterogeneousmyxopapillary ependymomas are typically cured with surgery if a gross total resection is achieved and historically have not required radiation unless metastatic or with residual disease.On the other hand, SP-EPN that harbor MYCN amplifications can have worse clinical outcomes, despite aggressive treatment.As a group, ependymomas largely lack actionable targets for therapy and have historically been considered chemoresistant.Given that, neurosurgical intervention and complete (i.e., no residual nodular disease) resection are essential, as many studies have shown that extent of surgical resection is the strongest predictor of survival [11].Following surgery, focal radiation therapy for patients with nondisseminated disease is standard of care [8].
Upfront Ependymoma Trials and the Role of Chemotherapy ACNS0121 was a large cohort of newly diagnosed ependymoma patients, which estimated the event-free survival and OS of patients treated with surgery, radiation, and a selected group with chemotherapy.Most patients received 59.4 Gy, and those under 18 months received 54 Gy.The authors concluded that radiation should remain the mainstay of treatment for most molecular subgroups of ependymoma [11].SIOP Ependymoma I was an upfront trial of newly diagnosed nonmetastatic ependymoma patients that demonstrated that patients with gross total resection had the best outcomes, and 1q gain/H3K27me3 was associated with poor outcome [12].This protocol was the basis of the next trial, SIOP Ependymoma II, which looks to definitively report on the relevance of adding maintenance chemotherapy to eligible patients who have received radiation therapy [22].As we eagerly await the official results of ACNS0831, a trial that looked to definitively conclude whether maintenance chemotherapy after radiation would improve event-free survival and OS in patients, decisions must be made for the future direction of ependymoma trials.Given our knowledge that molecularly defined subgroups have distinct clinical outcomes, dividing these patients into molecularly stratified trials is essential but comes with the challenge of diminishing patient numbers in each strata and the ability to statistically power a trial in this capacity.

Conclusions -Challenges and Future Directions
There is currently a paucity of ependymoma-specific clinical trials, for both upfront and relapsed patients due to several factors, including the lack of drugs and druggability of known drivers, which in the case of ZFTA-RELA and EZHIP, are highly disordered proteins.Further complicating translation of drug targets is the relative lack of genetic and PDX models of ependymoma, and the models that do grow have significant periods of tumor development making preclinical studies very difficult.While multiple genetic models for ZFTA-RELA fusion-driven ependymoma are being developed, models for PF-A ependymoma have proved to be difficult.Many PF-A models are difficult to maintain in vitro and can take 6-8 months to grow in vivo.This proves to be challenging for drug screening and executing preclinical studies.As our understanding grows of the underlying biology and drivers of these entities, it is essential to develop more models that faithfully recapitulate these molecular entities, for preclinical studies and clinical trial development.As the field of immunotherapy starts to expand in the pediatric neuro-oncology field, we need to consider not only traditional targeted therapy but also Chimeric Antigen Receptor T-cell therapy that may be effective in subsets of ependymoma patients.
In the last 2 decades, the lack of progress on largescale drug screening efforts, and druggability of known drivers in ependymoma argues that we need to rethink how we identify new dependencies and drug targets.If models can be developed, can dependencies and targets be identified by functional genomics screening approaches?Can epigenetic and cellular programs be used to guide novel target discovery?A critical consideration is how effective are targets in a primary compared to a relapsed setting, the latter of which we know very little.Are we designing trials that are bound to fail by testing targets at the wrong stage of tumor development?
How can we better predict who will relapse and who will be salvageable?How can we leverage new technologies to be more specific or efficacious in ependymoma?Is there inherently process that causes resistance to chemotherapy compared to other tumor entities?While many questions in the ependymoma field remain, we must continue to push forward with our understanding of the biology of these tumors to interrogate new targets and therapies with nontraditional approaches to have a meaningful impact on the outcomes of our patients.