ZFTA::YAP1-rearranged ependymoma: is still ZFTA or already YAP?

Abstract

Under the auspices of the World Health Organization Classification of CNS Tumors (5th Edition), supratentorial ependymomas are divided into 2 main molecular groups, comprising tumors that are either ZFTA- or YAP1-fused with the majority harboring ZFTA::RELA fusion. ZFTA-rearranged ependymomas with non-canonical 3´ fusion partners have significant clinical, radiological, and histopathological variability. We report a 5-year-old girl with a left frontal lobe tumor harboring a ZFTA::YAP1 fusion. The patient presented with recurrent absence episodes aggravating to epileptic seizures. The histological appearance of the tumor was inconclusive and shared morphological features of ependymal and glioneuronal neoplasms. Despite the nominal presence of a ZFTA alteration typically associated with aggressive ependymal tumors, the lesion demonstrated prolonged indolent biological behavior. Gene expression assay revealed paradoxical activation of YAP/TAZ signaling and signature unambiguously correspondent to YAP1-activated ependymoma. To our knowledge, this is the first case report demonstrating the molecular background of ZFTA::YAP1-rearranged ependymoma. The case denies the binary classification of supratentorial ependymomas into ZFTA and YAP1 subtypes based entirely on the identity of 5′ partner gene involved in the rearrangement. The findings elaborate the concept of molecular pathogenesis for YAP1-rearranged tumors.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40478-026-02352-5.

Introduction

Among neoplasms of the central nervous system (CNS), ependymal tumors (ependymomas, EPN) are particularly challenging in terms of differential diagnosis. A recent molecular classification of ependymomas provides a better grip on the biological complexity and variable clinical behavior of these tumors by accounting for anatomical site combined to molecular driver identity. Supratentorial ependymomas (ST-EPN), typically localized in the frontal or parietal lobes of the brain, are classified into ZFTA- and YAP1-rearranged subtypes — respectively, ST-EPN-ZFTA and ST-EPN-YAP1 [1].

The more prevalent ST-EPN-ZFTA tumors harbor eponymous chimeric transcripts with a 5′ coding fragment of ZFTA fused to another gene, typically RELA; the oncoprotein constitutively activates the NF-κB signaling pathway. The canonical ZFTA::RELA fusions result from chromothripsis at 11q12.1-11q13.3 [2]. In 15–20% of ST-EPN-ZFTA (non-RELA) the fusions involve MAML2, NCOA1, NCOA2 or DUX4 transcription coactivator genes, which suggests a common mechanism of aberrant gene activation via chromatin remodeling and similar epigenetic profiles [3–9]. The rarer ST-EPN-YAP1 subtype, accounting only for ~ 6–7% of ST-EPN, is characterized by predilection to young age, female gender, and indolent clinical course. These tumors are driven by recurrent gene fusions with 5′ involvement of either YAP1 (YAP1::MAMLD1, YAP1::FAM118B) or TEAD1 (TEAD1::NCOA2) resulting in YAP/TAZ-Hippo signaling dysregulation [10, 11]. ST-EPN-YAP1 tumors are clustered in close epigenetic proximity and independently of the particular genetic event triggering the YAP/TAZ signaling dysregulation.

The recurrent and mutually exclusive ZFTA and YAP1 rearrangements found in the vast majority of ST-EPN are often regarded as supratentorial localization markers for ependymoma, although isolated cases of infratentorial ependymal tumors with such rearrangements have been reported [12, 13]. Here we describe a pediatric ST-EPN with ZFTA::YAP1 fusion suggesting an ST-EPN-ZFTA, but revealing a transcriptomic signature and activated YAP/TAZ-Hippo signaling characteristic of ST-EPN-YAP1. Remarkably, the case denies the ST-EPN-ZFTA vs. ST-EPN-YAP1 binary classification.

Case presentation

In a previously healthy 5-year-old girl, the disease manifested with few-second absence episodes recurring 1–2 times per two weeks; the patient was unresponsive to verbal communication, with her eyes open. On examination, a skull deformity in the frontal region was noticeable. Magnetic resonance imaging of the brain revealed a large, 8 × 6 × 5 cm, mass-forming lesion with ill-defined outline, infiltrating the gray and white matter in the lateral frontal lobe and the insular lobe of the left hemisphere, and extending into the basal ganglia. The lesion was mildly intensified in T2-weighted and fluid-attenuated inversion recovery images, iso/hypointense in native T1-weighted images and without signal changes in diffusion-weighted images, with low-intensity contrast accumulation (Fig. 1A-D). The midline was not displaced, no signs of hydrocephalus were observed, and there was no compelling evidence of metastatic involvement of the spinal cord or the meninges.

Fig. 1.

Magnetic resonance images of the tumor lesion at the time of presentation (A-D) and after continued growth (E-H). Primary tumor presented with the solid mass in the left brain hemisphere extending into the basal ganglia. The cranial vault bone deformation was clearly present on the coronal projection (D). The tumor at progression moderately increased in size and showed significant cystic degeneration

A biopsy of the tumor was performed and no further specific treatment was administered. Seven months after biopsy, an epileptic episode occurred during febrile illness. A resting electroencephalography revealed an epileptiform focus in the left frontocentral area. A sleep electroencephalography revealed intermittent regional slowing in the frontocentral area with predominance on the left side. With anticonvulsant medication administered, the episodes recurred up to 1 time in 6 months, invariably on a background of pyrexia.

Three and a half years after biopsy, slow continued tumor growth was noted. The lesion had increased in size to 10 × 6.5 × 5 cm, with a significant cystic component now observable within its structure (Fig. 1E-H). A second surgery was performed amounting to subtotal resection, with substantive volume of infiltrative tumor tissue (> 1.5 cm³) remaining in the adjoining deep brain regions. A radiation therapy regimen was administered, on the resected tumor bed and the residual component in single focal doses ranging from 1.8 Gy, to a total focal dose of 54 Gy. Two years after treatment completion, the patient maintains remission.

Histological examination of specimens from the initial biopsy and repeated surgical intervention revealed tumor tissues of identical structure, of moderate-to-high cellularity with focal microcystic changes. The cells closely packed in cords, moderately pleomorphic with round nuclei containing finely dispersed salt-and-pepper chromatin, and pale eosinophilic or clear cytoplasm. Ganglion cells were observed focally as large polymorphic elements with abundant eosinophilic cytoplasm and normochromic oval or irregular nuclei with inconspicuous nucleoli, sometimes surrounded by clustering smaller tumor cells (a satellitosis tendency) (Fig. 2A). The mitotic activity constituted 1 mitotic figure per 10 high-power fields (3.2 mm²), with Ki67 index ≤ 5% on average and up to 10% focally. Immunohistochemistry revealed expression of GFAP (Fig. 2C) and S100, and the absence of OLIG2 and EMA. The ganglion cells expressed Chromogranin A (Fig. 2D), Neu-N, and Synaptophysin. The residual axonal fibers formed unevenly spaced network expressing NF. The findings suggested a diffuse low-grade glial/glioneuronal tumor. Routine PCR testing revealed no BRAF aberrations.

Fig. 2.

A, tumor tissue of high cellularity, cells polymorphous with rounded nuclei and finely dispersed salt-and-pepper chromatin, x200; the insert shows a focal inclusion of ganglion cells, x400; B, ultrastructural analysis of tumor tissue reprocessed from FFPE demonstrating close cell arrangement with tight junctions and zonula adherens (black arrows) and glial intermediate filaments at the periphery of tumor cells (arrowhead), x1000; C, heterogeneous expression of GFAP, bright in residual astrocytes, x200; D, expression of Chromogranin A in ganglion cells, x400; E, F, total diffuse cytoplasmic expression of L1CAM and nuclear expression of YAP1 in tumor cells, x200; G, unsupervised hierarchical clustering analysis of mRNA levels for genes involved in major oncogenic signaling pathways demonstrates notable segregation among CNS tumors. The current case with ZFTA::YAP1 fusion is clustered together with ST-EPN-YAP1 and separately from the canonical and non-canonical ST-EPN-ZFTA clusters

Genome-wide methylation profiling of the tumor revealed low-score matches with Ependymoma Family and EPN_ST_ZFTA_RELA_A (DKFZ/Heidelberg CNS tumor methylation classifier v12.8 calibrated scores, respectively, 0.48 and 0.15). Differential analysis of cytosine methylation across genomic regions by NCI/Bethesda Classifier v2.0 indicated similarity to Ependymal tumor superfamily (mean score 0.47 for Gene body, Shelf, Other genomic regions, Shore, Open Sea, and Open chromatin probes) and Low grade glial/glioneuronal tumors (mean score 0.39 for Island, 5’UTR, and Transcription Start Site), as well as EPN_ST_ZFTA_RELA_A and Ganglioglioma methylation classes (mean scores, respectively, 0.66 and 0.95) by the same genomic regions. In UMAP visualization of the methylation dataset the sample clustered in the immediate vicinity of diffuse astrocytoma, MYB- or MYBL1-altered; ependymomas of the posterior fossa group B; and ST-EPN-YAP1 (Supplementary Fig. 1A). Analysis of unbalanced chromosomal aberrations revealed the presence of trisomy 11 (Supplementary Fig. 1B). Whole-transcriptome RNA sequencing revealed a ZFTA (ENST00000433688.2, exon 5)::YAP1 (ENST00000615667.4, exon 3) chimeric transcript with the breakpoints at chr11:63763475 and chr11:102162454 in GRCh38 (Supplementary Fig. 1C). The identified chimeric transcript warranted immunohistochemistry with anti-L1CAM and -YAP1 antibodies, revealing diffuse cytoplasmic and nuclear expression of corresponding proteins (Fig. 2E, F).

With the new data, the diagnostic focus shifted toward ependymoma. Transmission electron microscopy was performed to confirm this hypothesis. Ultrastructural findings such as the arrangement of cells connected by adherent junctions (zonula adherens), fragmented microtubules, and glial intermediate filaments in the tumor cell cytoplasm (Fig. 2B). We also observed cells with intracytoplasmic cilia. These features suggested ependymal origin of the tumor.

Expression profiling was performed to predict the molecular group of ependymoma. Transcripts of genes involved in key oncogenic signaling pathways were assessed using the NanoString technology with nCounter PanCancer Pathways Panel (NanoString, USA) against a reference set of supratentorial CNS tumors with molecularly verified drivers (ZFTA::RELA ependymomas, n = 6, and ZFTA::non-RELA ependymomas, n = 4; ST-EPN-YAP1, n = 2; high-grade neuroepithelial tumor with MN1::BEND2, n = 5; pilocytic astrocytomas, KIAA1549::BRAF-positive, n = 3; pilocytic astrocytomas, NTRK-positive, n = 2; infant-type high grade glioma, TRK-fused, n = 3; pleomorphic xanthoastrocytoma, TRK-fused, n = 2; PATZ1-rearranged neuroepithelial tumors, n = 6; high-grade neuroepithelial tumor with BCOR alteration, n = 5; CNS Ewing sarcoma family tumor with CIC alteration, n = 4; and intracranial mesenchymal tumors, FET-CREB fusion-positive, n = 2). The tumor was clustered together with ST-EPN-YAP1, including YAP1::MAMLD1- and TEAD1::NCOA2-positive cases, and separately from ST-EPN-ZFTA with canonical (ZFTA::RELA) and alternative (ZFTA::SS18, ZFTA::MAML2) fusions (Fig. 2G).

The integrated evidence supported a diagnosis of supratentorial ependymoma with YAP1 dysregulation and ZFTA::YAP1 fusion.

Discussion and conclusions

Significant clinical, radiological, histopathological and immunohistochemical variability of ST-EPN reflects the diversity of molecular driver events for these tumors. In the current clinical case, large neoplastic lesion diffusely infiltrated local brain areas, projecting from the cortex to the basal ganglia. Despite the proximity to the left lateral ventricle, the tumor likely originated in cortical structures of the left frontal lobe, given the isolated manifestation with seizures. Regardless of clinical manifestation at 5 years of age, the cranial vault bone deformity suggest hidden slowly progressive growth.

For pediatric supratentorial tumors with seizure manifestations, the differential diagnosis typically includes ganglioglioma, dysembryoplastic neuroepithelial tumor and meningioma, but not ependymoma. However, atypical cortical ependymomas can initially manifest with seizures [14]. Convulsive symptoms in neuro-oncological diseases have been historically attributed to mechanical causes, with little consideration to molecular electrophysiology. The tumors have been thought to induce cortical hyperexcitability through compression and deformation of the cortex resulting from the volumetric impact of the tumor, as well as its invasive growth in surrounding brain areas and the peritumoral cerebral edema. Meanwhile, a growing number of clinical cases feature epileptogenic tumors with abnormally activated YAP/TAZ signaling pathway. Sporadic meningiomas with mutations in the YAP/TAZ signaling master regulator NF2 often manifest with seizures, although the mutation itself is not sufficient to cause seizures [15]. In isolated cases of YAP1-rearranged meningioma (YAP1::LMO1, YAP1::MAML2), young patients also presented with seizures, which disappeared after complete tumor resection [16, 17]. A case of low-grade glial/glioneuronal tumor with YAP1::FAM118B fusion reported by El-Dana et al. expands the histological spectrum of epileptogenic CNS tumors; importantly, the tumor matched none of the established methylation classes, suggesting a new group of epileptogenic supratentorial tumors with non-specific histological patterns and shared YAP/TAZ signaling activation [18]. An experimental study by Zhang et al. reveals a pathogenetic link between the dysergulated Hippo-YAP1-TAZ axis and epilepsy in rat model. The model involves activation of the Hippo signaling pathway in neonates induced by maternal immune stimulation; the resulting deficiency of full-length functional YAP1 leads to increased seizure susceptibility, as well as an endothelial-mesenchymal transition impairment manifesting with patent foramen ovale [19].

YAP1 is a key transcription coactivator of the highly conserved Hippo signaling pathway. It lacks a DNA-binding domain and activates its target genes through interactions with DNA-binding transcription factors — primarily the TEAD family proteins which interact with the amino-terminal TEAD-binding domain of YAP1 encoded by YAP1 exons 1 and 2. The central WW domains encoded by YAP1 exons 2–4 interact with transcription factors other than TEAD including p73, RUNX family proteins, and OCT4/KLF4. The number of WW domains in YAP1 depends on alternative splice variants either encompassing or skipping exon 4 [20]; therefore, the engagement of YAP1 in TEAD- vs. non-TEAD-dependent transcription regulation can be balanced via the 1-or-2 WW domain isoform ratio. The carboxy-terminal region of YAP1 rich in serine, threonine and acidic amino acids forms a strong transactivation domain.

YAP1 typically provides a 5′ partner in fusion oncogenes. Chimeric oncoproteins retaining the TEAD-binding domain of YAP1 have been identified in various pediatric CNS tumors, most notably in ST-EPN (YAP1::MAMLD1, YAP1::FAM118B), less typically in meningiomas and schwannomas with wild-type NF2 (YAP1::MAML2), and certain soft-tissue tumors, notably epithelioid hemangioendotheliomas (YAP1::TFE3) and sclerosing epithelioid fibrosarcomas (YAP1::KMT2A) [8, 21, 22], which share the aberrant activation of TEAD-dependent signaling. Incidentally, rearrangements involving 5´ regions of YAP1 and TEAD1 confer identical pathogenetic effects in ST-EPN, as indicated by highly similar epigenetic profiles of such tumors [11, 23].

In extremely rare cases, YAP1 provides a 3′ partner in gene fusion rearrangements, corresponding to carboxy-terminal region in chimeric oncoproteins. Such products retain the WW domain(s), but not the TEAD-binding domain; accordingly, the pathogenesis involves alternative, TEAD-independent transcription networks. Isolated cases harboring ZFTA::YAP1 fusion have been incidentally identified in large cohorts of patients with ST-EPN, albeit not assigned to any molecular subtype [2, 24]. A unique case of ZFTA::YAP1-rearranged ependymoma of the spinal cord has been reported by Lim et al. The tumor expressed L1CAM and YAP1 proteins similarly with the current case, but has epigenetic homologies within the DNA methylation class of ST-EPN-ZFTA of SNUH (Seoul National University Hospital) methylation classifier. The radiography and histopathology findings indicated high malignancy grade, and local recurrence occurred shortly (8 months) after total resection [25]. A case of non-ependymal CNS tumor, rearranged with YAP1 in 3′ position (KDM2A::YAP1), histologically resembling BCOR internal tandem duplication tumors, has been recently described by Tauziède-Espariat et al. (2025). Noteworthy, the in-frame rearrangement involved exon 1 of YAP1, which preserved all functional domains of YAP1 in the chimeric oncoprotein [26].

Ependymomas with canonical ZFTA::RELA fusion typically reveal the characteristic histopathological features of ependymal tumors, with signs including perivascular preudorosettes, fibrillar matrix and ramified vessels, similarly with ST-EPN-YAP1. By contrast, ependymomas harboring ZFTA fusions with non-RELA partners have significant histopathological heterogeneity with a broad range of glial (pleomorphic xanthoastrocytoma-, astroblastoma-, high grade glioma-like) and mesenchymal neoplasias, and may show neuronal differentiation with a granular cell component and myogenic differentiation foci [3–7]. Accordingly, it has been suggested that the gene in 3´ position may define the tumor morphology and immunophenotype [4]. Thus, ZFTA-rearranged tumors with mesenchymal and ganglion cell differentiation patterns are more likely to involve, respectively, NCOA1/2 or DUX4 [4–7] and MAML2 [3]. Despite the phenotypical diversity, ST-EPN-ZFTA with non-canonical 3´ fusion partners are epigenetically vicinity and match the MC Ependymoma, supratentorial, ZFTA fusion-positive and subclasses C–E. By contrast to classic ZFTA::RELA, the alternative ZFTA::non-RELA fusions promote no aberrant activation of the NF-κB signaling cascade (a capacity of RELA domains) [4].

In the current case ZFTA was fused to YAP1, hence the expected lack of NF-κB signaling activation; at the same time, paradoxical activation of YAP/TAZ signaling was noted. Given the seeming contradictions between radiological, morphological and molecular findings regarding the glioneuronal vs. ependymal nature of the tumor, gene expression profiling was ultimately performed, encompassing a representative broad selection of brain tumor types in a single assay. The clustering of RNA signatures unambiguously revealed close proximity of the current case to YAP1-activated ependymoma, including both YAP1- and TEAD1-rearranged cases. The indolent clinical course also fits into the general concept of the prognostically favorable ST-EPN-YAP1 subtype.

Overall, the case denies the binary classification of ST-EPN into ZFTA and YAP1 subtypes based entirely on the identity of 5′ partner gene involved in the rearrangement. The findings expand the known spectrum of epileptogenic pediatric tumors and elaborate the concept of molecular pathogenesis for YAP1-rearranged tumors.

Authors’ contributions

Study conception: M. Z., A.D. Acquisition of data: M. Z., A.P., A.T., A.V., N.U., A.D. Analysis and interpretation of data: M. Z., A.P., A.T., A.D. Clinical coordination: L.P., G.N. Administrative support: N.G. Drafting of the manuscript: M. Z., A.D. Critical revision: A.P., A.D.

Funding

The study was supported by Foundation for support and development in the field of pediatric hematology, oncology and immunology “Science for Children”.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The Independent Ethics Committee and the Scientific Council of the Dmitry Rogachev National Medical Research Center of Pediatric Hematology, Oncology and Immunology approved the study.

Consent for publication

Written voluntary consent to the patient participation in the study was obtained from legal representative.

Competing interests

The authors declare no competing interests.