Pediatric methylation class HGNET-MN1: unresolved issues with terminology and grading

High-grade neuroepithelial tumor with MN1 alteration (HGNET-MN1) is a rare, recently described central nervous system (CNS) entity with a distinct methylation profile, which was formerly part of CNS-primary neuroepithelial tumors (PNET). HGNET-MN1 affects children (77% of reported cases [1‐5]) and are characterized by a recurrent fusion implicating the MN1 (meningioma 1) gene [1]. Limited histopathological and clinical data are available on HGNET-MN1 (44 cases proven by DNA methylation analysis) and a number of outstanding issues still exist [1‐5].

Firstly, the histopathological diagnostic criteria are not well established. In Sturm’s study, they were initially classified as astroblastomas (16/41, as 4 of our cases) but also as PNETs (12/41), ependymomas (9/41, as 6 of our cases), meningioma (1/41), glioblastoma (1/41), embryonal tumor with multilayered rosettes (1/41), and unclassified tumor (1/41) [1]. These data raise the question of an overlap between a histopathological entity (astroblastoma, considered as a glioma in the current World Health Organization –WHO- classification) and a molecular entity (HGNET-MN1). Secondly, the prognosis of HGNET-MN1 remains unclear. The term “HG” was employed because they were initially described within a cohort of malignant tumors [1]. To explore these issues, we screened our local and the French neuropathological network database (n = 10, five of them were briefly mentioned in [6]) and the literature (n = 34), for pediatric DNA-methylation proven HGNET-MN1 [1‐5]. Here, we describe the molecular, histopathological, clinical and imaging characteristics of pediatric HGNET-MN1.

Including our cases, HGNET-MN1 predominantly concern girls (90.9%, 40/44 cases) and are mainly supra-tentorial (95.3%, 41/43 cases) [1, 3‐5]. Radiologically, all our tumors were well-demarcated from adjacent parenchyma with a multi-nodular aspect. Perilesional edema was not constantly observed. None was calcified and one case was hemorrhagic. They consisted of very large lesions (size ranging from 4.5 to 11.0 cm, mean 7.1 cm) with prominent solid portions and multiple necrotic areas but a unique cystic component was not the main tumor feature (Additional file 1: Figure S1 A-D). Thus, the radiological analysis showed that HGNET-MN1 differed from ependymoma, with RELA-fusion, the main pediatric differential diagnosis [6].

In the literature, HGNET-MN1 has been identified in a portion of astroblastomas (39% in adults compared to 78% in children) and did not constantly harbour astroblastic features (one case of our series and 8.5% of reported pediatric cases) [1‐5]. In our series, the only constant feature was tumoral fibrous or sclerous stroma (Additional file 1 Figure S1 E-F). They were often macro- and microcystic (n = 7) (Additional file 1 Figure S1 G-H). They showed a large variety of architectural patterns (Additional file 2 Figure S2 A-E). Cellular atypia was mild except in two cases (Additional file 2: Figure S2 F). True ependymal rosettes, lymphocytic infiltrates, eosinophilic granular bodies, and Rosenthal fibers were absent. Mitotic count ranged from 1 to 48 for 10 high-power fields (HPF) and the MIB1 ranged from 2 to 40% (Fig. 1). Microvascular proliferation and necrosis were observed in 3 and 8 cases, respectively (Fig. 1). The immunohistochemical profile (summarized in Additional file 5 Table S1 and Additional file 3: Figure S3) is also heterogeneous. The literature review and our cases show that 72% of tumors express glial markers [1‐5] and 58% express neuronal markers [1, 5]. Our analysis revealed that 50% of cases express CK18, and, surprisingly for a primary brain tumor, CD56 was focally or not expressed in 5/10 tumors. Further fundamental studies and ultrastructural analyses are required to elucidate the cell of origin of HGNET-MN1. Thus, HGNET-MN1 present variable morphological and immunophenotypical features.

Our meta-analysis (summarized in Fig. 1, with statistical methodology detailed in Supplementary Data), revealed that during follow-up (mean 81.4 months, range 2.6–324.0), 14/25 patients (56.0%) experienced tumor progression (8 local, 1 leptomeningeal, and 5 without details) and 3 patients (12.0%) died (Fig. 1). The mean/median progression-free survival (PFS) were 43.9/34 months. The mean overall survival (OS) was 81.6 months and the median OS was not reached. Children with HGNET-MN1 present a favourable outcome with 88.0% survival (median follow-up of 65.5 months), although 48.0% presented with tumor progression. Three patients (cases 7, 18 and 44) treated only by surgery were alive without tumor progression: one with a long follow-up (229 months) and two with only a short follow-up period (20 and 21 months). Total resection was significantly associated with better OS (p = 0.032) and PFS (p = 0.029). In univariate analysis, we found significant differences in terms of OS (p < 0.001) and PFS (p < 0.001) between HGNET-MN1 and ependymomas with RELA-fusion, ependymomas with YAP1-fusion and HGNET-BCOR from previous studies [1, 6‐9]. Our study has several limitations inherent to its retrospective observational design, but the prognosis does not seem to be as dismal as the term “high-grade” could suggest. No grading criteria have been established for HGNET-MN1. However, the WHO classification mentions histopathologic criteria for malignant astroblastoma [10]: mitotic index > 5/10 HPF, microvascular proliferation and palisading necrosis. In univariate analysis, these criteria were not significantly, independently or in combination, associated with worse OS (p = 0.461, 0.423, 0.426 respectively and in combination p = 0.461) and PFS (p = 0.094, 0.379, 0.301 respectively and in combination p = 0.426) (Fig. 1, Additional file 4: Figure S4). Moreover, three patients (cases 4, 6 and 12) presenting tumors with these three criteria did not develop progression and are alive (median follow-up of 68.0 months) [5].

In summary, this is the largest series of pediatric HGNET-MN1 with clinical, radiological and histopathological characterization. They present a homogeneous MRI aspect, but a heterogeneous morphology, with inconstant astroblastic pseudorosettes. MN1 fusion must be confirmed to rule out the major differential diagnoses (pleiomorphic xanthoastrocytomas [2‐4] and ependymomas [3]). Nevertheless when confronted to our metaanalysis results, naming these tumors has proven difficult. Indeed, both “HGNET” and “astroblastoma” do not completely correlate with grading and morphology. “HGNET” is not a proper designation because the prognosis seems favourable in a subset of cases and further studies are needed to comfort this tendency. “Astroblastoma” is also not quite appropriate since astroblastic pseudorosettes are inconstant and not specific, and a frequent glioneuronal immunophenotype is observed. Our study showed that the two constant histopathological and molecular features of these tumors are the fibrous stroma and the MN1 fusion. Consequently, we would like to suggest the terms “NET, MN1 fusion-positive” or “Fibrous neuroepithelial tumors (FNET), MN1 fusion-positive”.