Intracranial ependymomas: molecular insights and translation to treatment

Abstract

Ependymomas are primary central nervous system tumors (CNS), arising within the posterior fossa and supratentorial regions of the brain, and in the spine. Over the last decade, research has resulted in substantial insights into the molecular characteristics of ependymomas, and significant advances have been made in the establishment of a molecular classification system. Ependymomas both within and between the three CNS regions in which they arise, have been shown to contain distinct genetic, epigenetic and cytogenic aberrations, with at least three molecularly distinct subgroups identified within each region. However, these advances in molecular characterization have yet to be translated into clinical practice, with the standard treatment for ependymoma patients largely unchanged. This review summarizes the advances made in the molecular characterization of intracranial ependymomas, outlines the progress made in establishing preclinical models and proposes strategies for moving toward subgroup‐specific preclinical investigations and treatment.

Introduction

Ependymomas are rare neuroepithelial tumors, accounting for only 6.9% of primary central nervous system (CNS) tumors diagnosed annually. However, they are the third most common pediatric CNS tumor, constituting 10% of primary intracranial tumors 67. Ependymomas typically arise from the ependymal lining of the ventricles and central canal of the spinal cord 37, although they can also arise in the cerebral hemispheres. More than 60% of all ependymomas arise in the posterior fossa, with approximately 30% in the supratentorial region, and the remaining 10% in the spine 49. However, the location varies by age, with 90% of ependymomas in pediatric patients arising in the brain 44, and the majority of tumors in adults arising in the spine 12, 29, 73. The three regions in which ependymomas arise are shown in Figure 1.

Figure 1.

Regions of the CNS in which ependymomas arise. Illustration of the three distinct regions of the central nervous system (CNS) in which ependymomas arise; the supratentorial region (red), the posterior fossa (blue) and the spine (purple).

Grade I ependymal tumors tend to be well demarcated and complete surgical excision is typically curative 27, 49. Treatment for grade II and III intracranial tumors commonly involves maximal surgical excision, followed by radiotherapy 3, 60, 70, 99 for patients over the age of 3 98. However, in cases where grade II ependymomas are completely resected, radiation is not always administered 27, 99. For patients under 3 years, irradiation has typically been avoided or delayed 34, 99, due to the risk of damaging neurocognitive and neurological effects 13. However, more recently patients aged between 12 months and 3 years have also been treated with irradiation, with studies showing its favorable effects upon survival in this population 60, 80, 81. Second look surgery has also been utilized in pediatric cases where postoperative imaging reveals tumor capable of excision 30, 57. In cases of tumor dissemination via the cerebrospinal fluid, craniospinal irradiation is typically administered 99. As clinical trials utilizing various cytotoxic chemotherapy regimes, for both primary and recurrent ependymomas, have shown mixed results and limited improvements in survival 3, 103, chemotherapy is not part of the standard treatment for ependymomas 48, 51, 70. The majority of these trials have involved pediatric patients, with chemotherapy given as an adjuvant therapy to delay or avoid irradiation 33, 34, 90, prior to irradiation where there is residual tumor 23 and prior to further or second look surgery 19. There are also two ongoing phase II and III pediatric trials investigating maintenance chemotherapy (NCT01096368), combinations of radiotherapy and chemotherapy in patients over 12 months and chemotherapy alone in those 12 months and under (NCT02265770, SIOP Ependymoma II).

While the extent of surgical resection is strongly associated with progression free and overall survival 30, 46, 73, 91, 92, 104, the proximity of intracranial ependymomas to critical brain structures 48, 81 mean the benefits of resection have to be balanced with the potential for significant neurological deficits 61, 92. In light of this, complete surgical excision is only achieved in 40%–60% of patients 104. However, even in cases of complete resection and radiotherapy, up to half of pediatric patients will experience tumor recurrence 44, 86. Despite further surgery, irradiation/re‐irradiation, and trials of cytotoxic chemotherapy, long‐term control of relapsed ependymoma is rare, with around 90% of pediatric patients ultimately succumbing to the disease 56, 62, 79, 103.

Histopathological Classification

The World Health Organization's (WHO) classification of CNS tumors, categorizes ependymal tumors into five subtypes across three grades 49. Grade I comprises subependymomas and myxopapillary ependymomas, which are considered benign and rarely transform into grade II or III tumors 5, 50, 65, 89. However, there are reports of a small number of myxopapillary ependymomas which contain anaplastic features and display malignant behavior 41, 47. Grade II and III tumors are referred to as ependymomas and anaplastic ependymomas. However, the grading of these tumors, has been criticized for its subjective criteria and lack of reproducibility among pathologists. Further, due to tumor heterogeneity, there is little correlation between tumor grade and patient outcome in pediatric cases 17, 29. These issues are acknowledged in the 2016 WHO classification, which notes histopathological grading may soon become obsolete 49. It has been suggested molecular subgrouping provides a better predictor of patient outcome 36, 73, with proposals that histopathological grading no longer be used for grade II and III ependymomas, outside of clinical trials 36, 53, 72, 73, 95. Whether these proposed changes will extend to grade I ependymal tumors remains to be seen, given these tumors can be easily distinguished by pathologists 29, 70, and there is a lack of evidence to suggest subgrouping is more closely correlated to clinical outcomes for patients with grade I tumors.

Molecular Classification

The current molecular classification of ependymomas into nine subgroups, with three subgroups in each of the anatomical regions, is based upon whole genome DNA methylation profiling 73. Subependymomas arising in the supratentorial region, posterior fossa and spine comprise three of these subgroups, labeled ST‐EPN‐SE, PF‐EPN‐SE and SP‐EPN‐SE, respectively. The remaining posterior fossa subgroups are designated PF‐EPN‐A and PF‐EPN‐B, to highlight their consistency with the previously identified Group A and Group B posterior fossa tumors 93, 96. Building upon earlier findings of gene fusions in supratentorial ependymomas 75, testing for v‐rel avian reticuloendotheliosis viral oncogene homolog A (RELA) and yes associated protein 1 (YAP1) gene fusions was carried out on tumors in the remaining supratentorial subgroups. Approximately 88% of the larger group contained a RELA fusion transcript, with these tumors designated ST‐EPN‐RELA. While only 7 (54%) samples in the smaller subgroup were tested, each contained a YAP1 fusion gene, with this subgroup designated ST‐EPN‐YAP1. Interestingly, clustering of the samples by gene expression, resulted in similar subgroups, with the analysis of copy number alterations revealing significant differences between the subgroups 73. Correlation of clinical characteristics with the subgroups revealed patients with ST‐EPN‐RELA and PF‐EPN‐A tumors had a significantly worse progression free and overall survival compared to the other subgroups 73.

Recent analysis of methylation data from larger cohorts of PF‐EPN‐A and PF‐EPN‐B tumors, has identified additional subgroups and subtypes, within these major subgroups. In the case of PF‐EPN‐A tumors, two major subgroups with distinct copy number variations and transcriptomes have been identified and labeled PFA‐1 and PFA‐2 72. While the clinical characteristics of PFA‐1 and PFA‐2 tumors were similar, nine distinct subtypes within the groups [labeled PFA‐1(a‐e) and PFA‐2(a‐c)] were identified, with differences in age, gender, tumor grade and progression free and overall survival 72. For PF‐EPN‐B tumors, five subtypes, labeled PFB 1–5 were identified 11. While these subtypes had distinct copy number variations, gene expression and differences in patient age and gender, there were no significant survival differences.

While the current WHO classification 49 only recognizes ST‐EPN‐RELA ependymomas as a distinct subtype, the identification of the two distinct groups of grade II and III posterior fossa tumors in several studies 73, 93, 96 and conformation of these broader groups in more recent studies 11, 72, arguably warrants recognition of PF‐EPN‐A and PF‐EPN‐B as distinct subtypes in the next WHO classification.

Table 1 sets out the current molecular subgrouping classification of grade II and III intracranial ependymomas 73, 42, 93, 96, and the distinct molecular and clinical characteristics of each of the subgroups 42, 93, 96.

Table 1.

Molecular classification and characteristics of grade II and III intracranial ependymomas.

Subgroups	Supratentorial		Posterior fossa
	EPN‐RELA	EPN‐YAP1	EPN‐A	EPN‐B
Minor subgroups & subtypes			PFA1, PFA1(a‐e) 72	PFB (1‐5) 11
			PFA2, PFA2(a‐c) 72
Characteristics	RELA fusion gene (C11orf95‐RELA) 73	YAP1 fusion gene (YAP1‐MAMLD1 most common) 2, 54, 73, 75	Balanced genome 73, 96	Chromosomal instability (loss/gain whole chromosomes and chromosomal arms) 73, 96
	Chromothripsis (Chromosome 11) 73	Chromosome 11 aberrations around YAP1 locus 73	1q chromosome1 gain (25%) 73	Loss of chromosome 6 (82%) 73
	1q chromosome gain (24%) 73		CIMP positive 51	1q chromosome gain (18%) 73
	CDKN2A loss 73		DNA hypomethylation 6	CIMP negative 51
	Increased L1CAM & RelA/p65 expression 75		Reduced H3K27me3 6	H3K27me3 present 6
	Clear cell/vascular variant morphology common 54, 78		Lateral location 96	Midline location 96
Age (Median)	0–69 years (8 years) 73	0–51 years (1.4 years) 73	0 ‐ 51 years (3 years) 73	10–65 years (30 years) 73
Progression free survival	29% (5 years) 73	66% (5 year) 73	33% (5 year) 73	73% (5 year) 73
	19% (10 year) 73		24% (10 year) 73	56% (10 year) 73
Overall survival	75% (5 year) 73	100% (5 year) 73	68% (5 year) 73	100% (5 year) 73
	49% (10 year) 73		56% (10 year) 73	8% (10 year) 73

Grade II and III Intracranial Subgroups

RELA ependymomas

Approximately 72% of supratentorial ependymomas contain RELA gene fusions, falling within the ST‐EPN‐RELA subgroup 22, 73, 75, 78. These tumors are more common in children over the age of 4 and adults 42, 75 and have one of the poorest prognoses, among the subgroups 73.

RELA fusions are the result of structural variations, consistent with chromothripsis, in chromosome 11, resulting in the fusion of Chromosome 11 Open Reading Frame 95 (C11orf95) to RELA 73, 75. While little is known about the function of C11orf95, RELA codes for RelA/p65, which forms a subunit of the NF‐kB transcription factor complex in the canonical NF‐kB signaling pathway 39, 75, 77. The formation of this complex results in its translocation to the nucleus and regulation of genes involved in apoptosis, cell proliferation and immune responses 77, 78. Studies have shown the RELA gene fusion proteins translocate to the nucleus and upregulate NF‐kB signaling and NF‐kB targets 68, 75, 78. Based upon the formation of tumors with similar characteristics to human ST‐EPN‐RELA tumors in mice injected with a C11orf95‐RELA fusion transcript, it is suggested the fusion alone is sufficient for tumor formation 68.

YAP1 ependymomas

While less than 30 ependymomas with YAP1 fusions have been reported (ST‐EPN‐YAP1 subgroup) 2, 54, 73, 75, the available data suggest these tumors are more common in young children, and have a favorable prognosis 73.

At least 80% of YAP1 fusion ependymomas involve fusions between YAP1 and the mastermind like domain containing 1 (MAMLD1) gene 2, 54, 73, 75. YAP1 codes for the yes‐associated protein (YAP) 73, 75, an oncoprotein, which is a downstream effector of the HIPPO signaling pathway. Activation of this pathway results in the phosphorylation of YAP and its transcription co‐activator tafazzin (TAZ), preventing their translocation to the nucleus. In the absence of phosphorylation, YAP/TAZ translocate to the nucleus and interact with various transcription factors, including TEAD1‐4, resulting in increased expression of genes involved in cell proliferation and reduced expression of those promoting apoptosis and stem cell differentiation 87, 101, 106. Interestingly, when YAP was prevented from interacting with TEAD transcription factors, there was no tumor formation in mice, suggesting this interaction is critical for the formation of YAP1 ependymomas 71. The MAMLD1 gene, plays a key role in sexual development, with studies also showing it may be a co‐activator of the Hes family BHLH transcription factor 3 (Hes3) gene promoter, which is a target of the non‐canonical Notch pathway 9, 21. However, whether YAP‐MAMLD1 fusion proteins play a role in activating this promoter is unknown.

Although the number of ependymomas with YAP1 fusions is relatively small, the identification of such tumors in several different sample cohorts arguably warrants recognition of these tumors as a distinct subtype in the next WHO classification.

Other supratentorial ependymomas

The current molecular classification 73 does not encompass grade II and III supratentorial tumors lacking RELA and YAP1 fusions, despite reports of at least 29 supratentorial tumors lacking any gene fusions 22, 54, 69 and 22 tumors with fusions involving different genes 22, 75. In analysing eight of the samples lacking gene fusions, using the DNA methylation‐based classifier of CNS tumors 10, 24, the authors found that three of the samples could not be classified, with the closest match for the remaining five samples including glioblastomas, neuroepithelial tumors and posterior fossa ependymomas 22. These results suggest supratentorial ependymomas lacking RELA and YAP1 fusions are unlikely to fit within a single subgroup. Nonetheless the molecular classifaction could benefit from the addition of a broad supratentorial subgroup, particularly if the ST‐EPN‐YAP1 subgroup is included as a subtype in the next WHO classification, to recognize that there are a number of supratentorial tumors which do fit within the current subgroups.

PF‐EPN‐A ependymomas

The majority of posterior fossa ependymomas fall within the PF‐EPN‐A subgroup. These tumors occur predominantly in younger children and have a poor prognosis 73, 93, 96.

Aside from gains of chromosome 1q, PF‐EPN‐A tumors have a balanced genome 73, 93, 96 global DNA hypomethylation and CpG island hypermethylation 6, 51. Many of the genes silenced by methylation are targets of the polycomb repressor complex 2 (PRC2) 51, which plays a key role in chromatin modification, through its trimethylation of lysine 27 on the histone H3 subunit (H3K27me3) 7, 55. Treatment of PF‐EPN‐A xenograft models with demethylating agents, and a drug which degrades PRC2 complex proteins, was found to decrease tumor volume and increase survival, suggesting hypermethylation of CpG islands and the lack of H3 trimethylation play an important role in maintaining these tumors 51 .

Subsequent investigations found more than 80% of posterior fossa tumors had reduced levels of H3K27me3 6, 74. Although loss of H3K27me3 is commonly seen in diffuse intrinsic pontine gliomas 43 and pediatric glioblastomas 84 as a result of H3 K27M mutations, such mutations are extremely rare in PF‐EPN‐A tumors 6, 25, 72, 83. While the precise mechanism for reduced H3K27me3 is unknown, a recent study found increased expression of CXorf67 in PF‐EPN‐A tumors, with CXorf67 binding to PRC2 in an PF‐EPN‐A cell line, with the authors suggesting this binding may interfere with the ability of PRC2 to methylate H3K27 72. Alternatively, it has been suggested hypermethylation of CpG islands prevents the recruitment of PCR2, resulting in reduced levels of H3K27me3 6. The latter suggestion is consistent with the inverse relationship between CpG island methylation and H3K27me3 levels in these tumors 6.

The identification of further subgroups and subtypes within PF‐EPN‐A tumors indicates there is substantial heterogeneity within this broader subgroup, both genetically and cytologically. This may in part be due to the region of the posterior fossa in which the tumors arise, with PFA‐1 tumors arising in a more caudal location than PFA‐2 tumors 72. PFA‐1 tumors have upregulation of genes involved in immune responses and angiogenesis, while PFA‐2 tumors have upregulation of genes involved with cilia. There are also cytological differences between these subgroups and subtypes, with 1q gain and the loss of 6q, 10q and 20q more common in PFA‐1 subtypes, and gains of chromosomes 2, 8, 9, 11 and 19 more common among PFA‐2 subtypes 72.

PF‐EPN‐B ependymomas

PF‐EPN‐B tumors tend to arise in older children and adults and based upon retrospective analyses of patient samples and clinical data, have a more favorable prognosis 73, 81, 96 with many patients with completely resected tumors experiencing favorable outcomes without radiotherapy 81. However, late relapses are relatively common, with a number occurring more than 10 years after diagnosis 11.

PF‐EPN‐B tumors have frequent cytogenic aberrations, involving whole chromosomes or chromosomal arms, and more genetic instability than any other subgroup 73. They also have markedly less CpG island methylation 51, stain positive for H3K27me3 and do not contain H3 K27M mutations 6.

The recent identification 5 PF‐EPN‐B subtypes, with distinct gene expression patterns, copy number alterations and chromosomal aberrations 11, confirms the presence of heterogeneity within this broader subgroup. While PFB1 and PFB5 differed most in terms of gene expression, all subtypes displayed distinct cytogenic changes, with the study finding loss of 13q may be a marker for poor prognosis in these tumors 11.

Transition Toward Subgroup‐Specific Targeted Treatments

Despite significant advances in the molecular characterization of ependymomas over the last decade, this knowledge has yet to be translated into subgroup specific targeted treatments, with surgery and adjuvant radiotherapy the only real treatment options 53, 70. Prognostic factors relating to subgroups of ependymoma have, however, led to some consensus recommendations regarding current treatments for patients over 12 months of age. These include a recommendation that treatment for PF‐EPN‐A tumors consist of maximal surgical resection followed by irradiation 70, 73, 81. It has also been agreed that a trial be carried out to compare adjuvant irradiation with observation alone, in patients with completely resected PF‐EPN‐B tumors 70, 81. In light of studies reporting more favorable outcomes for ST‐EPN‐RELA tumors than originally suggested 59, 73, 88, there are no current recommendations for these tumors, or for the small number of ST‐EPN‐YAP1 tumors identified, with further investigation of patient outcomes required 70. There are also no consensus recommendations for the treatment of ependymoma patients under 12 months of age 70.

With regards to intracranial ependymomas generally, it has been agreed that supratentorial and posterior fossa ependymomas are distinct diseases and that treatment decisions for grade II and III tumors should not be based upon classification and grading determined by histopathological characteristics alone 70. It is also agreed that both histopathological and molecular classifications should be used in all future trials 70. While the current consensus recommendations for patients over the age of 12 months, don't distinguish between pediatric and adult tumors 70, given clinical trials commonly involve either pediatric or adult patients, it is possible that future treatment recommendations will be both age and subgroup specific.

Currently, there are no subgroup‐specific clinical trials for ependymoma patients, and the rarity of the tumors and difficulties in establishing subgroup‐specific models to carry out preclinical investigations makes it unlikely that such trials will occur in the near future. Nonetheless, different strategies have been implemented to allow correlation of molecular subgroups with patient outcomes. A current ongoing trial is prospectively assessing various markers including NELL2 and LAMA2 for posterior fossa tumors and RELA fusions for supratentorial tumors, as well as undertaking DNA methylation analysis (NCT02265770, SIOP Ependymoma II). Molecular profiling using DNA extracted from tissue slides of patients involved in a trial which concluded more than a decade ago has been carried out in order to correlate subgroup and outcome 59. Consensus among researchers that all future trials collect fresh frozen tissue and blood samples 70, will further facilitate such analysis in the future. Moving forward, a staged approach to ependymoma trials may also be appropriate for particular drugs, whereby the drug is first tested in the particular subgroup(s) in which it is most likely to be effective, and in the event it is effective, the trial can then opened to other subgoups in a second stage, as is the case in a recently commenced ependymoma trial using marizomib (NCT03727841).

While progress has been made over the last few years to incorporate the molecular subgroups into current treatment recommendations and clinical trials, changes to patient treatments outside of clinical trials are unlikely to occur in the near future. In addition to further correlating subgroups and patient outcomes in response to current treatments, substantial work at both a preclinical and clinical level is required to develop models and identify and test targeted therapies in the various subgroups, before subgroup‐based targeted therapies can become a reality. However, the distinct genetic, cytogenetic and epigenetic nature of ependymomas, both within and between the anatomical regions in which they arise, suggests a shift toward subgroup‐specific targeted treatments may be the only way to improve patient outcomes.

Relevance of Earlier Studies and Trials

Although several studies have reported on molecular aberrations in ependymomas, their relevance in identifying targets for subgroup‐specific treatment is limited, in circumstances where the findings are based upon unsorted cohorts of ependymomas. This has been demonstrated in clinical trials of lapatinib, an ERBB1 and ERBB2 inhibitor, in which no objective responses or tumor inhibition was found 20, despite evidence of elevated expression of ERBB2 and ERBB4 receptors 28 in ependymal tumors. Similarly disappointing results occurred in trials involving the VEGF inhibitors, bevacizumab 15, 38 and sunitinib 94, despite evidence of elevated VEGF expression in ependymomas 45, 82. While it is possible such drugs may be effective in particular subgroups, subgroup‐specific investigations are required to explore this possibility.

It is also possible cytotoxic chemotherapies may be successful in treating particular subgroups, given recruitment for clinical trials has not been specific for tumor location or subgroup 73. Previous trials have reported more favorable outcomes in patients with supratentorial ependymomas, compared to those with posterior fossa tumors 90, suggesting cytotoxic regimes may be beneficial in some subgroups of ependymoma.

Identifying Subgroup‐Specific Markers, Targets and Treatments

A shift to subgroup‐specific treatments will require the identification of reliable and inexpensive markersthat can easily be assessed in a clinical setting to distinguish between the subgroups. For posterior fossa tumors, immunochemistry using NELL2 and LAMA2 or H3K27me3 antibodies has been shown to be fairly reliable in distinguishing between PF‐EPN‐A and PF‐EPN‐B tumors 74, 96. The use of L1CAM and RelA/p65 antibodies have also been shown to be relatively accurate in identifying ST‐EPN‐RELA tumors 18, 26, 54, 69. However, given the presence of a small number of false positives and negatives with these antibodies, whole genome DNA methylation analysis may still be required in the event significantly different treatments are recommended. The establishment of the DNA methylation‐based classification website for CNS tumors (currently in the development stage) 24, which allows the identification of ependymoma subgroups using raw DNA methylation data, is a big step toward making methylation analysis clinically accessible.

With regards to subgroup‐specific targets, gene enrichment and expression analysis has identified a number of processes, pathways and genes which are upregulated within the various subgroups 6, 11, 25, 31, 42, 51, 72, 73, 75, 96. The identification of oncogenes and tumor suppressor genes which differ across the three regions 64 also provide a basis for further subgroup‐specific investigations, as do the genes associated with the recently identified subgroup‐specific super enhancer regions of DNA and upregulated transcription factors 52.

High‐throughput drug screens also offer promise for identifying candidate drugs for testing. A large drug screen identified four compounds whose potency against the tumor cells was more than twofold greater than the potency against neural stem cells (NSC) 3. One of these compounds, 5‐FU, halved the rate of tumor growth and significantly prolonged survival 3. Although the response of patients in a subsequent clinical trial of 5FU was limited, only two of the enrolled patients had tumors of the same type as used in the screen 3, 97. Although both these patients experienced disease progression, the limited responses of the other patients may be due to fact they had different types of ependymal tumors 97. This trial highlights the importance of ensuring trials enroll patients whose tumors match the subgroup in which the drugs were found to be effective, so that drugs are not dismissed on the basis of limited efficacy across molecularly distinct tumors.

Smaller scale drug screens can also be of benefit, with a screen of two posterior fossa cell lines against 97 FDA approved compounds identifying nine compounds which were more effective in ependymomas 16.

Preclinical Models

Although the rarity of ependymomas, lack of established cell lines and difficulties in establishing representative mouse models has hampered subgroup‐specific investigations 100, there has been progress in establishing both in vitro and in vivo models over the last few years.

Despite difficulties in culturing many ependymoma samples beyond a few passages 44, several laboratories have now reported success in doing so, particularly when using samples from grade III recurrent tumors 1, 16, 35, 40, 51, 63, 76, 85, 102, 105. In 2018, the Brain Tumor Resource Laboratory also made available the first commercial posterior fossa ependymoma cell line 8, 66.

With regards to in vivo models, there has been success establishing supratentorial mouse models via the implantation of transduced murine NSCs with suspected driver mutations. Supratentorial and posterior fossa mouse models have also been established via the direct implantation of tumor cells. With regards to the former approach, a number of RELA fusion negative mouse models have been established 42, 64. There are also reports of RELA fusion mouse models established by transducing cells with RCAS vectors containing Cllorf95‐RELA fusion 1, before implantation into tv‐a transgenic mice 68. YAP1 mouse models, generated through the transduction of C11orf95‐YAP1 and YAP1‐MAMLD1 fusions into NSC 71, 75 have also been established.

In the absence of known genetic drivers for the majority of subgroups, the generation of patient‐derived xenograft (PDX) models, successfully reported in a number of studies 4, 14, 40, 63, 102 may be the best way forward. Although, it has been noted tumors in xenograft models are typically slow to grow 32, studies report quite variable time frames, ranging from 117 to 300 days 63, 102.

There are reports of a few distinct PDX posterior fossa mouse models which have been established 58, 85. Zhang and colleagues 105 also recently reported the development of six PF‐EPN‐A PDX models which replicate the pathogenic and morphological characteristics of these tumors. The Brain Tumor Resource Laboratory has also made available mouse ependymoma lines from two posterior fossa and 1 RELA fusion model 66.

Table 2 lists the reported intracranial ependymoma cell lines and mouse models, for which the tumor subgroup or anatomical region is known.

Table 2.

Grade II and III intracranial ependymoma cell lines and mouse models.

	Supratentorial		Posterior fossa
	EPN‐RELA	EPN‐YAP1	EPN‐A	EPN‐B
Cell lines (subgroup specific)	BXD‐1425EPN (pediatric, recurrent) 102		E520 (pediatric) 4
	DKFZ‐EPINS (grade III, recurrent) 63		EPD‐210FHTC (pediatric, grade III, recurrent) 66
	R254 (pediatric) 4		4423EPN (pediatric) 105
			MAF811 (pediatric, recurrent) 1, 16, 31
Cell lines (region specific)	nEPN1 (grade II, recurrent) 40		nEPN2 (pediatric, grade II) 40	EP1 (pediatric grade III) 85
			BT‐44 (pediatric, grade III) 35	EPP (pediatric, grade II) 85
			BT57 (pediatric, grade III) 35
Mouse models (subgroup specific)	PDX: IC1425EPN (BXD‐1425EPN cells – pediatric, recurrent) 102	Cllorf95‐YAP1 transduced Ink4a/Arf−/− NSC 75	PDX: E520 cells (pediatric) 4
	PDX: DKFZ‐EP1NS cells (grade III, recurrent) 63	YAP1‐MALD1 cloned into a luciferase vector 71	PDX: EPD‐210FH (pediatric, grade III, recurrent) 66
	C11orf95‐RELA transduced Ink4a/Arf−/− NSC 75		PDX: EPD‐710FH (pediatric, grade III, recurrent) 66
	C11orf95 ‐RELAFus1 RCAS vectors (N)/tv‐a, (G)/tv‐a, (B)/tv‐a mice 68		PDX: 4423EPN (pediatric) 105
	PDX: EPD‐613FH (pediatric, grade III, recurrent) 66		PDX: 2002EPN (pediatric) 105
Mouse models (region specific)	BCLC transduced Ink4a/Arf−/− NSC 64		PDX: nEPN2 cells (pediatric, grade II) 40
	RAB3A transduced Ink4a/Arf−/− NSC 64		PDX: BT44 cells (pediatric, grade III) 35
	ZNF668 transduced Ink4a/Arf−/− NSC 64		PDX: BT57 cells (pediatric, grade III) 35
	PDX: nEPN1 cells (grade II, recurrent) 40		PDX: EPP cells (pediatric, grade II) 85
			PDX: EPV cells (pediatric grade III) 85

Summary

While there has been real progress in establishing preclinical ependymoma models over the last few years, further models are required to investigate the myriad of potential therapeutic targets identified within the various ependymal subgroups. However, in moving toward targeted treatments care must be taken to ensure the effects upon NSC are minimized 3 and that use of molecules which target pathways essential for development are limited 98, given the majority of intracranial epenymomas arise in pediatric patients.

Despite the challenges involved, a shift toward targeted, subgroup specific preclinical investigations is imperative if new treatments are to be identified and patient outcomes improved.

Conflicts of Interest

The authors do not have any conflicts of interest.