Two pediatric supratentorial ependymal tumors with novel PLAG1 fusions

Abstract

The 2021 World Health Organization (WHO) Classification of Central Nervous System (CNS) classification formalized routine molecular profiling, and DNA-methylation studies have since delineated PLAG-family–altered CNS entities: PLAGL1 fusion–positive supratentorial neuroepithelial tumors (NET_PLAGL1) and embryonal tumors with PLAGL1/PLAGL2 amplification. PLAG1 fusions with diverse partners have been reported in CNS embryonal tumors, but have not been described in supratentorial neuroepithelial tumors to date. We describe two pediatric supratentorial ependymal tumors with novel PLAG1 fusions that do not match any 2021 WHO-defined entity or any PLAG-family–related entity recently reported in the literature. Case 1 (4-year-old boy) had a 7.6-cm left lateral ventricular mass with edema and heterogeneous enhancement; gross total resection was performed without adjuvant therapy (alive at 8 months). Histology showed diffusely low cellularity with small- to medium-sized round nuclei, minimal atypia, and focal calcifications; no ependymal rosettes, branching vessels, or clear-cell change. Tumor cells were positive for GFAP and H3K27me3, and negative for Desmin, EMA, OLIG2, L1CAM, NF-κB and H3K27M. The MIB-1 labeling index was ~ 5%. RNA-seq identified TNC::PLAG1 fusion; FISH showed PLAG1 rearrangement. DNA methylation clustered with spinal subependymoma, despite supratentorial location. Case 2 (4-year-old girl) had a 1.4 × 1.1 cm left parahippocampal lesion; resected without adjuvant therapy (alive at 4 months). Histology showed low-to-moderate cellularity with diffuse microcystic change, focal clear/vacuolated cells, and delicate branching vessels. Tumor cells were positive for GFAP, EMA, L1CAM, H3K27me3 and ATRX, and negative for EMA, Desmin, NF-κB, OLIG2, H3K27M, and CD34 (MIB-1 ~ 2%). RNA-seq identified TXNIP::PLAG1 fusion. Methylation did not reach a class threshold. Both cases ultimately warrant a final diagnosis of ependymal tumor, not elsewhere classified. To our knowledge, TNC::PLAG1 and TXNIP::PLAG1 are first-ever fusions reported in any tumor type. They also represent the first PLAG1 fusions identified in pediatric supratentorial ependymal tumors. These cases highlight the value of integrating histology, methylation profiling, and fusion detection, and suggest a new candidate supratentorial ependymal subtype with PLAG1 fusions, pending validation in larger series.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40478-026-02242-w.

Introduction

The 2021 World Health Organization (WHO) Classification of Central Nervous System (CNS) Tumors is the first to integrate molecular features with classical histopathology, emphasizing the routine use of methods such as next-generation sequencing and DNA methylation [1]. Ependymal tumors (EPN) are a heterogeneous group of glial neoplasms that account for 4.6% (1132 of 24,776) of pediatric CNS tumors [2]. According to the current WHO classification and DNA methylation profiling, EPNs are categorized by location into supratentorial, posterior fossa, and spinal ependymomas, as well as subependymomas [3]. Supratentorial ependymomas are further subdivided into supratentorial ependymoma, ZFTA fusion-positive and supratentorial ependymoma, YAP1 fusion-positive. Posterior fossa ependymomas are further subdivided into posterior fossa group A and posterior fossa group B by DNA methylation profiling, underscoring the diagnostic importance of methylation-based classification. Subependymoma is a rare, typically indolent variant, characterized by clusters of uniform to mildly pleomorphic nuclei embedded in an abundant fibrillary matrix with frequent microcystic changes.

The pleomorphic adenoma gene (PLAG) family of transcription factors, including pleomorphic adenoma gene 1 (PLAG1), pleomorphic adenoma gene-like 1 (PLAGL1), and pleomorphic adenoma gene-like 2 (PLAGL2), encodes zinc-finger proteins with highly conserved DNA-binding domains that regulate proliferation, differentiation, and apoptosis [4, 5]. Recently, DNA methylation profiling has delineated PLAG-family–altered entities in the CNS, with PLAGL1 fusions defining supratentorial neuroepithelial tumors (NET_PLAGL1) and PLAGL amplifications (chiefly PLAGL1 or PLAGL2) characterizing CNS embryonal tumors [6–8]. Emerging reports also describe PLAG1 fusions in embryonal tumors, in which reported partner genes include EWSR1, ASAP1, ADGRG1, TMEM68, NCALD, HNRNPK, CHD7, TCF4, and LOC105378102 [6, 9]. But, unlike PLAGL1, PLAG1 fusions have not been reported in supratentorial neuroepithelial tumors to date. Outside the CNS, PLAG1 fusions are well documented in pleomorphic adenomas, cutaneous mixed tumors, and uterine myxoid leiomyosarcomas, with common partners including CTNNB1, LIFR, FGFR1, NFIB, CHCHD7, and TGFBR3 [10–14].

Herein, we report two pediatric intracranial ependymal tumors, each harboring a distinct PLAG1 fusion: tenascin C (TNC)::PLAG1 and Thioredoxin-Interacting Protein (TXNIP)::PLAG1. To our knowledge, these are the first two reported pediatric intracranial ependymal tumor harboring PLAG1 fusions. Moreover, both TNC::PLAG1 fusion and TXNIP::PLAG1 are first-ever descriptions in any tumor type. DNA methylation profiling clustered the TNC::PLAG1 fusion tumor with the spinal subependymoma subtype, highlighting a novel molecular alteration and expanding the spectrum of ependymal tumor.

Materials and methods

Case selection

Two cases of neuroepithelial tumor barboring PLAG1 fusion encountered at the Pathology Department of West China Hospital were collected for review. Radiologic and clinical data were obtained from the electronic medical records. This study was approved by the institutional review board. Informed consent for translational research was obtained from each patient’s legal guardian.

Immunohistochemistry

The following primary antibodies were used: alpha-thalassemia/mental retardation syndrome X-linked protein (ATRX) (rabbit polyclonal, ZA-0016, ready-to-use, ZSGB-BIO, Beijing, China;), CD34 (rabbit monoclonal, ZA-0550, ready-to-use, ZSGB-BIO), Desmin (mouse monoclonal, MX046, 1:200, MAIXIN-BIO, Fuzhou, China), Epithelial membrane antigen (EMA) (mouse monoclonal, ZM-0095, 1:100, ZSGB-BIO), glial fibrillary acidic protein (GFAP) (rabbit monoclonal, clone EP13, 1:200, ZSGB-BIO), Ki-67 (mouse monoclonal, clone MIB-1, ZM-0167, 1:200, ZSGB-BIO), H3 K27M (rabbit monoclonal, ZA-0321, ready-to-use, ZSGB-BIO), H3 K27me3 (rabbit monoclonal, ZA-0327, ready-to-use, ZSGB-BIO), L1 cell adhesion molecule (L1CAM) (rabbit monoclonal, GT245802, ready-to-use, GENE TECH, Shanghai, China), Neurofilament (NF) (mouse monoclonal, 2F11, 1:50, MAIXIN-BIO), Nuclear factor kappa-B (NF-κB) p65 (rabbit monoclonal, D14E12, 1:800, Cell Signaling Technology, Danvers, MA, USA), and OLIG2 (rabbit monoclonal, clone EP112, 1:200, ZSGB-BIO). All immunohistochemical staining was performed using the Roche BenchMark ULTRA automated staining system (Roche, Basel, Switzerland) according to the manufacturer’s protocols.

Targeted next-generation sequencing

Both RNA and DNA targeted next-generation sequencing (NGS) were performed by Nanjing Simcere Diagnostics Laboratory (SimcereDx), Nanjing, China. Total RNA extracted from formalin-fixed, paraffin-embedded (FFPE) tumor tissue was submitted for NGS on the MGI DNBSEQ-T7 platform (MGI Tech, Shenzhen, China). A targeted panel covering 267 tumor-related genes (Supplementary Table 1) was used, including analysis of gene rearrangements (fusions) within the target regions. Genomic DNA was extracted from FFPE tissue using a Concert™ Genomic DNA Extraction Kit (Avantor, Radnor, PA, USA) following the manufacturer’s protocol. DNA quantity was measured with a Qubit™ 2.0 fluorometer (Life Technologies, Carlsbad, CA, USA), and purity was assessed by the A260/280 ratio. Qualified DNA was sheared to 250–300 bp, and libraries were prepared with the KAPA HyperPlus Kit and KAPA HyperCapture Reagent Kit (Roche, Basel, Switzerland) for the SimcereDx NeuroOnco360 panel (Supplementary Table 1). Paired-end 150-bp sequencing was performed on the MGI DNBSEQ-T7 platform (MGI Tech, Shenzhen, China).

Reverse transcription polymerase chain reaction (RT-PCR) and Sanger sequencing

RNA was extracted from FFPE tumor tissues using a RNApure FFPE Kit (CW0535S, CWBIO, China). Reverse transcription was performed using ExonScript RT SuperMix with dsDNase (Chengdu Rongwei Gene Biotechnology, China) according to the manufacturer’s instructions. The primers of RT-PCR for TNC::PLAG1 were FP: CAGCCTGCCTACTGTCACC and RP: TCGTGGAAGAGAGTGGAATC. The primers of RT-PCR for TXNIP::PLAG1 were FP: GCCAGCCAACTCAAGAGACA and RP: CAGGTGATCCTTTCGCCCAA. The reaction mixture was subjected to an initial denaturation at 95 °C for 5 min, followed by 35 cycles of amplification (denaturation at 95 °C for 30 s, annealing at 60 °C for 15 s, and extension at 72 °C for 20 s). PCR products were sequenced in both the sense and antisense directions by Sangon Biotech (Chengdu, China).

Fluorescence in situ hybridization (FISH)

A Locus-Specific Identifier dual-color break-apart PLAG1 probe (GSP PLAG1, Anbiping) was used for FISH according to the manufacturer’s instructions. Red and green signals flanking the PLAG1 locus were assessed for split patterns or signal imbalance. The nuclei were counterstained with DAPI. A nucleus was considered positive for rearrangement when a clear split of the red/green signals or isolated 3′/5′ signal consistent with break-apart was observed in more than 20% cells [15].

DNA methylation profiling

DNA methylation profiling was performed by Guangzhou Huayin Medical Laboratory Center. Raw IDAT files were processed in R 4.1.0 using minfi (v1.40.0) under Bioconductor 3.14 to extract signal intensities, followed by preprocessQuantile normalization [16]. Methylation sites on the X and Y chromosomes were excluded, as well as non-informative sites from the 935 K array. The top 20,000 most variable methylation sites were selected for clustering analysis. The t-distributed stochastic neighbor embedding (t-SNE) method was applied using the Rtsne package. Classifier modeling used a training cohort of 2,801 samples. Raw IDAT files were also uploaded to the DKFZ/Heidelberg CNS Tumor Methylation Classifier v12.8 for reference cohort projection; classifier outputs were provided in the Supplementary 2 and Supplementary 3.

Case presentation

Case 1

A 4-year-old boy presented with acute onset of right-sided limb weakness and fever for one day and was admitted to our hospital on April 22, 2025. Neurological examination revealed drowsiness, decreased muscle strength in all extremities (grade IV), neck stiffness, and a positive Babinski sign. Magnetic resonance imaging (MRI) of the brain revealed a large infiltrative mass located in the left trigone of the lateral ventricle, measuring approximately 7.6 × 7.6 × 6.0 cm. The lesion was associated with extensive surrounding vasogenic edema, significant mass effect on adjacent structures, and a midline shift. On imaging, the mass appeared predominantly hypointense on T1-weighted sequences (Fig. 1A), markedly hyperintense on T2-weighted sequences (Fig. 1B), and showed heterogeneous hyperintensity on FLAIR images (Fig. 1C). After administration of gadolinium contrast, the lesion demonstrated irregular, heterogeneous enhancement (Fig. 1D). The patient underwent intracranial tumor surgery. The patient did not undergo any radiotherapy or chemotherapy. As of the most recent follow-up on October 17, 2025, the patient remains alive, with an overall survival of eight months since diagnosis.

Fig. 1.

Magnetic resonance imaging of brain. Case 1: A Axial T1‑weighted image showed a predominantly hypointense mass in the left trigone of the lateral ventricle. B Axial T2‑weighted image showed a markedly hyperintense lesion with extensive perilesional edema. C Axial FLAIR image demonstrated heterogeneous hyperintensity within the lesion. D Post‑contrast T1‑weighted image revealed heterogeneous enhancement of the mass. Case 2: E Axial T1‑weighted image showed a hypointense mass (arrow). F Axial T2‑weighted image showed a markedly hyperintense lesion in the left parahippocampal gyrus (arrow)

Histological examination of the tumor revealed a diffusely low-cellularity neoplasm composed of tumor cells with small- to medium-sized round nuclei and minimal nuclear atypia. (Fig. 2A, B). Small foci of calcification were identified in certain regions (Fig. 2C). No ependymal rosettes, branching blood vessels, clusters of pleomorphic nuclei, or microcystic changes were observed. The overall bland cytology, uniform nuclear contours, and diffuse growth within a fine fibrillary matrix were morphologically reminiscent of diffuse astrocytoma, MYB- or MYBL1-altered. Immunohistochemically, the tumor cells were positive for GFAP (Fig. 2D) and H3K27me3 (Fig. 2E), and negative for Desmin, EMA, OLIG2, L1CAM, NF-κB and H3K27M. NF immunostaining was negative within the tumor (Supplementary 6). The MIB‑1 labeling index was approximately 5% (Fig. 2F).

Fig. 2.

Histological and immunohistochemical features of the Case 1 and Case 2. Case 1: A Hematoxylin and eosin staining showed a diffusely low-cellularity neoplasm composed of tumor cells with small- to medium-sized round nuclei. B No definite mitotic figures were identified under high-power magnification. C Small foci of calcification were identified. D Immunohistochemistry for GFAP showed diffuse cytoplasmic positivity. E Immunohistochemistry for H3K27me3 demonstrated retained nuclear expression. F Immunohistochemistry for MIB‑1 highlighted a labeling index of approximately 5%. Case 2: G Hematoxylin and eosin staining showed a diffusely low-cellularity neoplasm composed of tumor cells. H Branching blood vessels were identified. I Vacuolated tumor cells were identified. J Immunohistochemistry for GFAP showed diffuse cytoplasmic positivity. K EMA immunostain showed focal paranuclear (dot-like) positivity. L Immunohistochemistry for MIB‑1 highlighted a labeling index of approximately 2%

By targeted next-generation RNA sequencing, the tumor demonstrated TNC fusion with PLAG1 as the 3’ partner, which was validated by RT-PCR and Sanger sequencing (Fig. 3A, B). The breakpoint fell within exon 1 in TNC (NM_002160.4) and exon 2 in PLAG1 (NM_001114634.2), resulting in an in-frame fusion. Split red and green signals were observed in a subset of tumor cells, indicating a rearrangement of the PLAG1 locus (Fig. 3C) by FISH analysis.

Fig. 3.

TNC::PLAG1 and TXNIP::PLAG1 fusions. Case 1: A, B A schematic of the genomic rearrangement resulting in a TNC::PLAG1 gene fusion, which was validated by RT-PCR. C Fluorescence in situ hybridization using a dual‑color break‑apart probe demonstrated split red and green signals in a subset of nuclei, indicating rearrangement of the PLAG1 locus. Case 2: D, E A schematic of the genomic rearrangement resulting in a TXNIP::PLAG1 gene fusion, which was validated by RT-PCR. F FISH showed several nuclei displaying a single split red or green signal (arrow) alongside a fused red-green pair, compatible with a PLAG1 rearrangement

In the DNA methylation clustering analysis shown in Fig. 4A, the case 1 was positioned in close proximity to the spinal subependymoma group, suggesting a strong similarity in methylation profiles between the present case and the spinal subependymoma subtype (full-field t-SNE plot provided in Supplementary 4). DKFZ/Heidelberg brain tumor classifier assigned case 1 to the spinal subependymoma methylation class (calibrated score 0.97; class family “Ependymal tumours” 0.99; Supplementary 2). Subtype scores were below the match threshold. Copy number profiling based on methylation array data revealed no significant genomic or chromosomal alterations (Fig. 5A). Hypermethylation of the O⁶-methylguanine methyl transferase (MGMT) gene promoter was absent.

Fig. 4.

t-SNE plot of DNA methylation-based classification of central nervous system tumors. Each dot represents a reference tumor sample, color-coded according to its methylation class. A The red-labeled sample corresponds to the TNC::PLAG1 fusion-positive case (case 1), which clusters closely with the spinal subependymoma (EPN_SPINE_SE) methylation class. B The red-labeled sample corresponds to the TXNIP::PLAG1 fusion-positive case (case 2), which clusters closely with the low-grade glioma, MYB (LGG_MYB) methylation class

Fig. 5.

Copy number variation profile. It revealed no significant genomic or chromosomal alterations on case 1 (A) or case 2 (B)

Case 2

A 4-year-old girl presented with paroxysmal upward gaze deviation lasting 1–2 min (up to 10–20 min) and was admitted to our hospital on August 25, 2025. MRI demonstrated a 1.4 × 1.1 cm nodular lesion with abnormal signal in the left parahippocampal gyrus. It was hypointense on T1-weighted (Fig. 1E) and hyperintense on T2-weighted images (Fig. 1F). The patient underwent intracranial tumor surgery. The patient did not undergo any radiotherapy or chemotherapy. As of the most recent follow-up on October 17, 2025, the patient remains alive, with an overall survival of four months since diagnosis.

Histological examination showed a diffusely low- to moderate-cellularity neoplasm composed of small round-to-oval tumor cells with mild nuclear atypia (Fig. 2G). The background was fine fibrillary with diffuse microcystic change; branching blood vessels (Fig. 2H) and focal clear/vacuolated tumor cells (Fig. 2I) were present. Those features were suggestive of diffuse astrocytoma, MYB- or MYBL1-altered morphology. No true ependymal rosettes or conspicuous perivascular pseudorosettes were identified, and no necrosis or marked microvascular proliferation was seen. The overall bland cytology, uniform nuclear contours, and diffuse growth within a fine fibrillary matrix were morphologically reminiscent of diffuse astrocytoma, MYB- or MYBL1-altered. Immunohistochemically, the tumor cells were positive for GFAP (Fig. 2J), EMA (Fig. 2K), L1CAM, H3K27me3 and ATRX, and negative for CD34, Desmin, NF-κB, OLIG2, and H3K27M. NF staining is diffusely positive within the tumor (Supplementary 6). The MIB‑1 labeling index was approximately 2% (Fig. 2L).

By targeted next-generation RNA sequencing, the tumor demonstrated TXNIP fusion with PLAG1 as the 3’ partner, which was validated by RT-PCR and Sanger sequencing (Fig. 3D, E). The breakpoint fell within exon 4 in TXNIP (NM_006472.6) and exon 4 in PLAG1 (NM_001114634.2), resulting in an in-frame fusion. Break-apart FISH detected approximately 5% of nuclei with a solitary red or green signal together with a retained red–green fusion signal; although this fell below the assay’s positivity threshold, the pattern was suspicious for a rearrangement of PLAG1 (Fig. 3F).

In the DNA methylation clustering analysis shown in Fig. 4B, the case 2 was close to the low-grade glioma, MYB group (full-field t-SNE plot provided in Supplementary 5). DKFZ/Heidelberg brain tumor classifier showed case 2 did not reach the classification threshold (max score 0.51 to the “Ependymal tumours” superfamily; Supplementary 3). Copy number profiling based on methylation array data revealed no significant genomic or chromosomal alterations (Fig. 5B). Hypermethylation of the MGMT gene promoter was absent.

NGS of DNA revealed no pathogenic alterations in both cases. Specifically, mutations in IDH1, IDH2, H3F3A, HIST1H3B, HIST1H3C, ATRX, TP53, and the TERT promoter (C228T/C250T) were not identified. There was no evidence of EGFR amplification or activating mutations, nor of BRAF mutations or fusions. Homozygous deletions of CDKN2A and CDKN2B were not observed. In addition, no gene fusions involving ZFTA or YAP1 were detected.

Discussion and conclusions

EPNs are classified according to their location into supratentorial ependymoma, posterior fossa ependymoma, spinal ependymoma, and subependymoma [7]. In this report, we describe two pediatric supratentorial ependymal tumors with novel PLAG1 fusions. Case 1 represents the first pediatric intracranial ependymal tumor harboring a PLAG1 fusion, specifically the first description of a TNC::PLAG1 fusion in any tumor type, with DNA methylation clustering within the SP-SE class. Case 2 was diagnosed as ependymal tumor, not elsewhere classified (TXNIP::PLAG1 fusion). Remarkably, this case also represents the first description of TNC::PLAG1 and TXNIP::PLAG1 fusion in any tumor type, underscoring a novel molecular alteration and expanding the recognized spectrum of ependymal tumor biology.

Both tumors in this report showed notable morphologic and immunophenotypic resemblance to diffuse astrocytoma, MYB- or MYBL1-altered: low grade with low cellularity, minimal atypia, a fine fibrillary background, GFAP positivity, OLIG2 negativity, and a low MIB-1 index, but with discordant ancillary findings. Case 1 had an NF-negative tumor parenchyma, and methylation did not support this class. Case 2 mapped near the diffuse astrocytoma, MYB-/MYBL1-altered cluster on t-SNE, yet showed focal paranuclear (dot-like) EMA positivity, favoring ependymal lineage. Taken together with the absence of MYB/MYBL1 alterations and the presence of PLAG1 fusions, the integrated histologic–molecular profile supports classifying both lesions as ependymal tumors, after which the differential is considered within the ependymal spectrum.

According to the 2021 WHO CNS, supratentorial ependymomas are currently divided into two molecularly defined subgroups: ZFTA fusion–positive and YAP1 fusion–positive [1]. Histologically, the present tumor shows morphological features similar to those of supratentorial ependymoma with YAP1 fusion. Both entities arise within the lateral ventricle and typically occur in young children. Furthermore, supratentorial ependymomas with YAP1 fusion are generally considered WHO grade 2 or 3 tumors with a relatively favorable prognosis compared to ZFTA‑fusion ependymomas, whereas the present cases exhibited low‑grade histology and a low MIB‑1 index, features more in keeping with an indolent clinical course [7, 17]. In addition, Tauziède‑Espariat et al. reported that non-RELA ZFTA-fused supratentorial tumors with heterogeneous morphologies, frequent EMA, frequent L1CAM, and often negative NF-κB nuclear staining [18], an immunophenotypic pattern similar to the present case 2. However, RNA-seq in our cohort did not detect ZFTA/YAP1 fusions, and methylation profiling did not support the EPN-ZFTA or EPN-YAP1 class.

Apart from ZFTA- and YAP1-fusion ependymomas, NET-PLAGL1 have recently been recognized as a distinct subgroup of ependymomas based on DNA methylation profiling [1]. NET-PLAGL1 tumors are morphologically characterized by an ependymal component admixed with subependymal features, frequent true ependymal rosettes or perivascular pseudorosettes, occasional clear cell or papillary areas, and frequent microcystic changes within a fibrillary matrix. Immunohistochemically, most cases show diffuse GFAP expression, dot-like EMA positivity, low MIB-1 proliferation indices, and absence of neuronal marker expression [19]. The present cases shared several features with NET-PLAGL1. Notably, although PLAG1 and PLAGL1 belong to the PLAG1 gene family, DNA methylation–based classification placed these cases outside the canonical NET-PLAGL1 cluster. Nevertheless, given the overlapping morphological and immunophenotypic features, as well as the presence of rare PLAG1 fusions, the possibility that these tumors represent an NET-PLAGL1 variant cannot be excluded.

PLAG‑family amplification is a typical event in CNS embryonal tumors. Tauziède‑Espariat et al. first described two cases of CNS embryonal tumors harboring PLAG1 fusions, both classified within the methylation class of embryonal tumors with PLAG‑family amplification. Subsequently, Keck et al. described a pediatric cohort (median age ~ 5 years) of PLAG1-fused tumors epigenetically adjacent to embryonal tumors with PLAG‑family amplification, characterized by dense, primitive small round blue-cell morphology, frequent ependymoma-like pseudorosettes, diffuse GFAP with focal Desmin, and high Ki-67/MIB-1 (~ 10–20%) [6]. In contrast, our PLAG1-rearranged ependymal-lineage tumors lack embryonal cytology and the proliferative profile typical of the PLAG‑family amplification entity.

PLAG1 rearrangements have been implicated in the pathogenesis of several mesenchymal tumors. For example, PLAG1 rearrangements leading to PLAG1 overexpression have been reported in approximately 25% of myxoid leiomyosarcomas and may serve as a diagnostic biomarker for this entity [20]. In addition, PLAG1 rearrangements have been observed in lipoblastomas across infants, children, and adults, and molecular detection of PLAG1 fusions can aid in the diagnosis of these tumors [21]. The most frequent PLAG1 fusion partners reported to date include CTNNB1, LIFR, FGFR1, NFIB, CHCHD7, and TGFBR3 [11, 12, 14]. These findings underscore the broader oncogenic role of PLAG1 and support the notion that PLAG1 fusions can be diagnostic markers in certain tumor contexts. In the present cases, the discovery of novel TNC::PLAG1 and TXNIP::PLAG1 fusions expands the known repertoire of PLAG1 rearrangements and, importantly, represents the first documentation of this fusion form in an intracranial ependymal tumor.

Subependymomas can arise throughout the central nervous system and are commonly found in the fourth and lateral ventricles. Histologically, they are typically well circumscribed, with minimal mitotic activity and a lack of conspicuous nuclear atypia, features that were also observed in the present case [22]. Although no subtype-specific driver alterations have been defined, DNA methylation profiling subdivides subependymomas into supratentorial (ST-SE), posterior fossa (PF-SE), and spinal (SP-SE) classes. Notably, case 1 clustered within the SP-SE class—which is typically spinal—despite its supratentorial (lateral ventricle) location. This discordance between anatomic site and methylation class underscores the complexity of epigenetic classification in ependymal tumors.

Taken together, Case 1 falls within the subependymoma spectrum by histology and methylation proximity, whereas both cases ultimately warrant a final diagnosis of ependymal tumor, not elsewhere classified. Since the TNC::PLAG1 and TXNIP::PLAG fusions have not reported, further supports that these cases may represent a distinct molecular variant within the spectrum of low‑grade ependymal tumors. Given the rarity of PLAG1 rearrangements in CNS tumors, these cases highlight the importance of comprehensive molecular characterization and long‑term follow‑up to determine the clinical significance of such alterations. Future studies and additional cases will be essential to clarify PLAG1‑rearranged tumors.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ATRX

Alpha-thalassemia/mental retardation syndrome X-linked protein

CNS

Central Nervous System

EMA

Epithelial membrane antigen

EPN

Ependymal tumors

FISH

Fluorescence in situ hybridization

FFPE

Formalin-fixed, paraffin-embedded

NET_PLAGL1

PLAGL1 fusion–positive supratentorial neuroepithelial tumors

NGS

Next-generation sequencing

PF-SE

Posterior fossa subependymomas

PLAG

Pleomorphic adenoma gene

PLAG1

Pleomorphic adenoma gene 1

PLAGL1

Pleomorphic adenoma gene-like 1

PLAGL2

Pleomorphic adenoma gene-like 2

RT-PCR

Reverse transcription polymerase chain reaction

SP-SE

Spinal subependymomas

ST-SE

Supratentorial subependymomas

TNC

Tenascin C

TXNIP

Thioredoxin-Interacting Protein

WHO

World Health Organization

Author contributions

Clinical data collection and analysis: LZ, JZ, JH, and XP. DNA methylation array analysis: TL and ZL. Conducted the molecular studies: LZ, ZS and YW. Pathology diagnosis: QZ and NC. Manuscript writing: LZ and NC.

Funding

The authors are supported by the National Natural Science Foundation of China (82472650, 82273047, 82273073) and Sichuan Provincial Natural Science Foundation—Youth Fund (2024NSFSC1893, 2025ZNSFSC1889).

Availability of data and materials

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

This study was approved by Clinical Ethics Committee of West China Hospital of Sichuan University and patient’s legal guardian gave consent.

Consent for publication

Patient’s legal guardian gave consent.

Competing interests

The authors declare that they have no competing interests.