A rare case of paediatric astroblastoma with concomitant MN1‐GTSE1 and EWSR1‐PATZ1 gene fusions altering management

Brain tumours are the commonest childhood neoplasm, with a worldwide incidence of 29.9-47.1/million (1). Childhood brain tumours carry substantial morbidity/mortality and are the largest cause of paediatric cancer deaths (2). Historically, classification was largely based on histological features. In recent years, expansion of high-resolution genomic, epigenetic and transcriptomic profiling has led to improved molecular understanding and categorisation, as well as targeted therapies (3). Consequently, the 2016 WHO classification incorporated molecular features in some brain tumour entities (4). According to this classification, astroblastomas are extremely rare, not formally graded, and listed under 'other gliomas'. Astroblastomas are generally treated by surgery alone but can display intermediate behaviour with high recurrence rates and unpredictable behaviour (5, 6). Controversy exists as to whether astroblastomas are a truly distinct entity as they have histological features in common with both astrocytomas and ependymomas (7-9). Diagnosing astroblastomas is therefore challenging and misclassification can alter subsequent management (8). The present case demonstrates how recent molecular advances identified a novel gene fusion for this patient, confirmed a more precise tumour diagnosis and guided subsequent management decisions. The findings here have general importance to other rare paediatric brain tumour entities.

Brain tumours are the commonest childhood neoplasm, with a worldwide incidence of 29.9-47.1/million [1]. Childhood brain tumours carry substantial morbidity/mortality and are the largest cause of paediatric cancer deaths [2]. Historically, classification was largely based on histological features. In recent years, the expansion of highresolution genomic, epigenetic and transcriptomic profiling has led to improved molecular understanding and categorisation, as well as targeted therapies [3]. Consequently, the 2016 WHO classification incorporated molecular features in some brain tumour entities [4].
According to this classification, astroblastomas are extremely rare, not formally graded, and listed under 'other gliomas'. Astroblastomas are generally treated by surgery alone but can display intermediate behaviour with high recurrence rates and unpredictable behaviour [5,6]. Controversy exists as to whether astroblastomas are a truly distinct entity as they have histological features in common with both astrocytomas and ependymomas [7][8][9]. Diagnosing astroblastomas is therefore challenging and misclassification can alter subsequent management [8]. The present case demonstrates how recent molecular advances identified two gene fusions for this patient, confirmed a more precise tumour diagnosis and guided subsequent management decisions. The findings here have general importance to other rare paediatric brain tumour entities. Histological examination demonstrated a cellular neuroepithelial tumour with a perivascular growth pattern forming vague pseudorosettes ( Figure 1E). In some areas, there was marked stromal and vascular collagenisation. There was no mitotic activity and no microvascular proliferation. The tumour had a generally noninfiltrative border with surrounding brain tissue. There were occasional eosinophilic granular bodies and no Rosenthal fibres or ganglion cell components. The tumour had focal strong staining for GFAP and only weak staining for synaptophysin ( Figure 1F/G). There was only very sparse mitotic activity and a low Ki67 proliferation index (2.5%), as evidenced by MIB1 staining ( Figure 1H). The histological and immunohistochemical features were most consistent with a diagnosis of astroblastoma.
Methylation profiling of the tumour [10] using the Illumina EPIC array platform and analysed using the DKFZ Heidelberg classifier, gave a very low calibration score of <0.9 (specifically 0.0475 at the time of reporting using classifier version MNPv11b4, and 0.0764 with the most recent version MNPv11b6). These scores were too low for reliable classification despite excellent probe hybridisation (only 0.09% of probes failed), suggesting that the tumour was unclassifiable compared with currently recognised tumour entities.
This was confirmed by comparing the tumour's methylation profile with other CNS tumour entities (Figure 2A). Regarding genomic alterations, inter-chromosomal gene fusions involving the MN1 gene (MN1-BEND2 and MN1-CXXC5) have previously been reported in CNS high-grade neuroepithelial tumours (HGNETs) [11], and recently it has been reported that a proportion of these CNS-HGNET-MN1 tumours exhibited histological features compatible with astroblastomas [12]. Following informed consent, ; and (C) Axial FLAIR sequences showed a well-circumscribed solid-cystic tumour in the right cingulate gyrus and corpus callosum, which crossed the midline and extended into the right lateral ventricle. The lesion had a local mass effect causing a large area of surrounding vasogenic oedema and secondary hydrocephalus. The solid components enhanced and contain some areas of low T2W signal intensity from calcification. The large anterior tumoural cyst was of high signal on FLAIR because of proteinaceous content. To assess the vascularity of the tumour preoperatively, digital subtraction angiography was also performed which demonstrated displacement of the anterior cerebral arteries to the left side, consistent with mass effect secondary to tumour, but minimal tumour 'blush' following contrast phasing through the brain centrally. Consequently, no embolisation was performed. gene fusion was also detected. This similarly involved inversion of chromosome 22 and a recent study suggests it may be consistent with a new glioneuronal tumour entity [14]. As a result of this unexpected additional finding, we explored this in more detail. Three Fusion genes involving MN1 and PATZ1 (MN1-PATZ1) have also very recently been described in a malignant paediatric brain tumour [15]. We carefully explored our data for evidence of rearrangements (including translocations and tandem duplications) that could similarly adjoin MN1 and PATZ1 in functional orientation. We did not find evidence supporting this proposition. However, amidst the rearrangement complexity, we cannot exclude the presence of undetected cryptic events, as have been observed to generate oncogenic fusions in human cancer [16].
Astroblastomas are associated with high recurrence rates and an unpredictable biological course [5,6], with initial peri-tumoural oedema, as was present in this case, associated with early recurrence [17]. Consequently, with the molecular confirmation of an astroblastoma, the multi-disciplinary-team outcome was to undertake MRI Here, we describe a patient with a rare astroblastoma, where additional molecular study assisted patient management. Neuropathologically, the differential diagnoses were an astroblastoma or glioneuronal tumour. Precise diagnosis is critical as it influences postoperative management decisions. Here, the molecular confirmation of the novel MN1-GTSE1 gene fusion, diagnostic of astroblastoma, supported the team in recommending 3-monthly follow-up, rather than the standard 6-monthly approach which would be more typical for a glioneuronal tumour.
Astroblastomas typically arise at supratentorial sites, although occasional infratentorial cases occur [18,19]. On MRI, they appear as well-demarcated solid-cystic lesions and typically extend from cortex to periventricular regions [20]. Commonly, calcification is seen in addition to a 'bubbly' appearance, arising from signal voids due to tumour angioarchitecture [21,22]. The lesions are typically hyperintense to white matter (FLAIR/T2-weighted sequences) and show heterogeneous enhancement with rim enhancement on contrastenhanced CT and T1-weighted MRI [9,22]. Our case showed many of these neuroradiological features ( Figure 1A-C).
Histologically, astroblastomas are characterised by perivascular hyalinisation and pseudo-rosettes [8]. While these pseudo-rosettes contribute to the diagnostic criteria, they may be completely undetectable [12]. A lack of fibrillary background, hyalinised vessels, and a compressive, rather than infiltrative, margin is also commonly observed [20]. Immunohistochemically, astroblastomas are typically positive for GFAP, vimentin and S-100 protein [20], although there is variability in staining [8]. Although there is currently no formal grading for astroblastomas (based on the 2016 WHO classification), a putative informal system has been suggested, with 'high-grade' tumours showing high cellularity, anaplastic nuclear features, vascular proliferation, high mitotic rates and necrosis [12], which may be considered for inclusion in future classifications.
Molecular data for astroblastomas are limited [23]. This is consistent with methylation profiling being uninformative in this case, despite its overall utility for paediatric brain tumours [24].
Interestingly, despite re-running the classifier again in 2020, the calibration score of 0.0764 was again too low for reliable classi-  [11,12]. Thus, it is plausible that the genetic basis of astroblastomas may involve MN1 rearrangement with fusion partner genes, such as BEND2, CXXC5 and other unidentified genes [12]. This is supported by another studying showing MN1 alterations in four of eight tumours diagnosed histologically as astroblastomas [25]. In our patient, we confirmed a novel MN1-GTSE1 gene fusion, which to our knowledge, is the first such description in an astroblastoma. Interestingly, we also observed an EWSR1-PATZ1 fusion generated through the rearrangement complexity emanating on chromosome 22, which has been described in glioneuronal tumours and sarcoma [14,26].
Whilst it is unusual to find two oncogenic gene fusions in the same tumour, this is not unprecedented [27]. It may represent the development of two parallel clones with uniform (i.e. nonbiphasic) histology or a double hit within the same clone.
From a diagnostic perspective, our observation of two fusions in this case are intriguing, as without WGS neither fusion would have been detected and our retrospective, integrated interpretation, requiring up-to-date molecular subtyping for context, was critical, despite the astroblastoma histology. We believe that this case is unlikely to represent a typical CNS-HGNET-MN1 tumour, due to the dissimilarity in methylation profile compared with reference HGNET-MN1 cohorts, the similarity with PATZ1-fusion cases and the intact X chromosome seen here (frequently disrupted in HGNET-

CO N FLI C T S O F I NTE R E S T
TSJ is Editor-in-Chief of the journal. All other authors have no conflicts to declare. The Editors of Neuropathology and Applied Neurobiology are committed to peer-review integrity and upholding the highest standards of review. As such, this article was peerreviewed by independent, anonymous expert referees and the authors (including TSJ) had no role in either the editorial decision or the handling of the paper.