Optical genome mapping identifies a novel pediatric embryonal tumor with a ZNF532::NUTM1 fusion

Abstract

Molecular characteristics of pediatric brain tumors have not only allowed for tumor subgrouping but have introduced novel treatment options for patients with specific tumor alterations. Therefore, an accurate histologic and molecular diagnosis is critical for optimized management of all pediatric patients with brain tumors, including central nervous system embryonal tumors. We present a case where optical genome mapping identified a ZNF532::NUTM1 fusion in a patient with a unique tumor best characterized histologically as a central nervous system embryonal tumor with rhabdoid features. Additional analyses including immunohistochemistry for NUT protein, methylation array, whole genome, and RNA-sequencing was done to confirm the presence of the fusion in the tumor. This is the first description of a pediatric patient with a ZNF532::NUTM1 fusion, yet the histology of this tumor is similar to that of adult cancers with ZNF::NUTM1 fusions reported in literature. Although rare, the distinct pathology and underlying molecular characteristics of the ZNF532::NUTM1 tumor separates this from other embryonal tumors. Therefore, screening for this or similar NUTM1 rearrangements should be considered for all patients with unclassified central nervous system tumors with rhabdoid features to ensure accurate diagnosis. Ultimately, with additional cases, we may be able to better inform therapeutic management for these patients.

Introduction

Embryonal tumors are a class of highly aggressive and heterogenous malignant central nervous system (CNS) tumors that primarily occur in infants and young children [1]. They are characterized histologically as undifferentiated or poorly differentiated tumors of neuroepithelial origin and have high cellular activity and rapid growth. There are several types of embryonal tumors including medulloblastoma and other embryonal tumors including atypical teratoid rhabdoid tumor (AT/RT) and embryonal tumors with multilayered rosettes (ETMR) [2,3]. To assist in distinguishing between the different classifications of embryonal tumors, genetic characterization, which has rapidly become part of the diagnosis workflow, is essential. Medulloblastoma is further subdivided into four categories: Wnt-activated, SHH-activated, Group 3 (non-Wnt/non-SHH), and Group 4 (non-Wnt/non-SHH)[4]. ETMR tumors, which are rare and generally arise in infants, are mostly classified through the presence of multilayered neuroepithelial cells that resemble rosettes with or without the presence of C19MC alteration [5]. For AT/RTs, SMARCB1 (INI1) or less commonly SMARCA4 (BRG1) loss of function is a criterion for diagnosis [6]. CNS neuroblastoma, FOXR2- activated and CNS tumor with BCOR internal tandem duplication are embryonal tumors that have recently been added to the 2021 WHO classification guidelines based on their molecular signatures [3]. Tumors that have histologic features of AT/RT, but do not have either INI1 or BRG1 loss were classified as CNS embryonal tumors with rhabdoid features based on the 2016 WHO guidelines, and now are included in the CNS embryonal tumors (not otherwise specified) category [1–3]. The tumors in this last subgroup do not have known/common molecular features.

Standard of care therapy for patients with embryonal tumors generally includes a combination of surgery, chemotherapy and radiation therapy [7–9]. Historically, therapeutic management primarily depended on patient characteristics such as age, tumor location, presence of metastatic disease and tumor histology (i.e. medulloblastoma versus AT/RT, etc). More recently, molecular characteristics of the tumors have introduced novel treatment options for patients with specific tumor alterations. For example, hedgehog pathway inhibitors are being included in clinical trials for patients with SHH-activated medulloblastoma, lower radiation doses are being trialed in patients with lower-risk Wnt-activated medulloblastoma, and intensified therapies are being trialed in patients with high-risk embryonal tumors [9]. An accurate histological and molecular diagnosis of embryonal tumors is therefore very important for both risk stratification and treatment planning.

In this case, the patient presented with an unusual tumor that had histopathologic features best aligning to a CNS embryonal tumor with rhabdoid features (or CNS embryonal tumor, NOS based on the 2021 WHO guidelines). Optical genome mapping (OGM) revealed a novel ZNF532::NUTM1 fusion that has not been described previously in children with any cancer type or adult brain tumor but has been identified in adults with tumors of the lung, mandible, parotid gland, and pelvic bone that have similar histologic features as the patient’s tumor. This case report describes clinical and genomic features of a patient with a novel pediatric embryonal tumor type.

Materials and methods

Ethics approval and consent to participate

Written informed consent for participation and publication was obtained from the parents of the patient through IRB approved Children’s National Hospital protocol, Pro#1339, PI Javad Nazarian, PhD.

Sample processing

Ultra-high molecular weight (UHMW) DNA was extracted following manufacturer’s guidelines (Bionano Genomics Inc, San Diego, CA, USA) from flash frozen brain regions (superior frontal gyrus and primary visual cortex) as well as pelleted frozen Peripheral Blood Mononuclear Cells (PBMCs). In brief, a total of 15–20 mg of brain tissue or 1.5–2 million PBMCs were homogenized in cell buffer and digested with Proteinase K. DNA was precipitated with isopropanol and bound with a Nanobind magnetic disk (Bionano Genomics Inc). Bound UHMW DNA was resuspended in the elution buffer and quantified with Qubit dsDNA assay kits (ThermoFisher Scientific, Waltham, MA, USA). Total RNA was extracted using a Qiagen RNeasy Kit following manufacturer’s guidelines (Qiagen, Hilden, Germany). RNA sequencing was performed by Novogene Co, Ltd., Durham, NC. mRNA was selected from total RNA using poly-T oligo-attached magnetic beads and sequenced on an Illumina short-read instrument with 85 million reads (Novogene Co, Ltd.). Genome sequencing (50X, 2×150bp) was also performed by Novogene Co, Ltd..

DNA labeling was performed following the manufacturer’s protocols (Bionano Genomics Inc). Direct Labeling Enzyme 1 (DLE-1) reactions were carried out using 750 ng of purified UHMW DNA. Labeled DNA was loaded on Saphyr chips (Bionano Genomics Inc) for imaging. The fluorescently labeled DNA molecules were imaged sequentially across the Saphyr chip’s nanochannel arrays on a Saphyr instrument (Bionano Genomics Inc). Effective genome coverage of greater than 500X was achieved for all samples. All samples also met the following quality control metrics: labelling density of ~15/100 kbp; filtered (>15kbp) N50 > 230 kbp; map rate > 70%.

Optical genome mapping analysis

Genome analysis was performed using software solutions provided by Bionano Genomics Inc. Automated, OGM specific, pipelines – Bionano Access and Solve (versions 1.7.1.1 and 3.7.1, respectively), were used for data processing and variant calling.

De novo assembly

De novo assembly was performed using Bionano’s custom assembler software program based on the Overlap-Layout-Consensus paradigm. Pairwise comparison of all DNA molecules was done to generate the initial consensus genome maps (*.cmap). Genome maps were further refined and extended with best matching molecules. Structural Variants (SVs) were identified based on the alignment profiles between the de novo assembled genome maps and the Human Genome Reference Consortium GRCh38 assembly. If the assembled map did not align contiguously to the reference, but instead were punctuated by internal alignment gaps (outlier) or end alignment gaps (endoutlier), then a putative SV was identified.

Rare variant analyses

Rare variant analyses were performed to performed to capture mosaic SVs occurring at low allelic fractions. Molecules of a given sample dataset were first aligned against GRCh38 assembly. SVs were identified based on discrepant alignment between sample molecules and reference genome, with no assumption about ploidy. Consensus genome maps (*.cmaps) were then assembled from clustered sets of molecules that identify the same variant. Finally, the cmaps were realigned to GRCh38, with SV data confirmed by consensus forming final SV calls.

Fractional copy number

Fractional copy number analyses were performed from alignment of molecules and labels against GRCh38 (alignmolvrefsv). A sample’s raw label coverage was normalized against relative coverage from normal human controls, segmented, and baseline copy number state estimated from calculating mode of coverage of all labels. If chromosome Y molecules were present, baseline coverage in sex chromosomes was halved. With a baseline estimated, copy number states of segmented genomic intervals were assessed for significant increase/decrease from the baseline. Corresponding copy number gains and losses were exported. Certain SV and copy number calls were masked, if occurring in GRC38 regions found to be in high variance (gaps, segmental duplications, etc.)

Variant analysis

Bionano Access (Bionano Genomics Inc) was used for SV annotation and filtering. Variants were filtered in access and nanotatoR (version 1.13.0) [10] based on the following criteria: for de novo, rare variant and copy number variant pipelines, SVs were filtered based on Bionano Genomics recommended size and confidence cutoff values (e.g., >500 bp/5 kbp size cut-off for de novo assembly and rare variant pipelines respectively for insertions and deletions (INDELs)). Rare SVs were selected by filtering out common variants with population frequency of >1% using Bionano Genomics’ database of SVs containing >300 healthy individuals. To select for potential clinically significant aberrations a gene list overlapping SVs was used.

Genome sequence analysis

FASTQ reads were aligned to GRCh38 reference genome using, BWA-MEM (version 0.7.17) [11], followed by processing of the aligned bam (for variant calling) using SAMtools [12] and Picard (Broad Institute; https://broadinstitute.github.io/picard/, last accessed July 2022). Next, for small nucleotide variant (SNV) and small INDEL calling we use Mutect2 (version 4.2.2.0) [13], followed by annotation using ANNOVAR (database version release March 2022) [14]. Due to absence of a non-tumor tissues from the same individual, we used a 1,000-genome panel of normal variant call file from the Broad Institute, as a proxy. Quality filtration for SNV/INDEL was performed using the Mutect2 function FilterMutectCalls. For larger structural variant calls, we used Manta (version 1.6.0) [15], followed by annotation using AnnotSV (vs 3.1.1) [16]. For the SV visualization Integrative Genome Viewer (IGV; version 2.10.0) was used.

RNA-sequence analysis

RNA-seq data was aligned to the GRCh38 reference genome, followed by fusion-calling using STAR-Fusion (version 1.10.1) [17]. Visualization of the fusion was performed using Clinker (version 1.3A) [18] and IGV.

EPIC methylation chip analysis

An input of 300 ng of DNA was bisulfite-converted using the DNA Methylation-Lightning kit (Zymo Research, Irvine, CA, USA). After whole-genome amplification and enzymatic fragmentation, samples were hybridized to BeadChip arrays using the Infinium Methylation EPIC BeadChip kit according to the manufacturer’s protocol (Illumina, SanDiego, CA, USA). Intensity values at the over 850,000 methylation sites on the BeadChips were measured across the genome at single-nucleotide resolution using iScan (Illumina). For classifying the tumors, the CNS and sarcoma tumor classification tool hosted at molecularneuropathology.org (CNS) and molecularsarcomapathology.org (Sarcoma) were used on the methylation signal files [19,20].

Immunohistochemistry (IHC)

Tissue was fixed in 10% formalin, embedded in paraffin wax. Sections were cut at 4-μm thickness and mounted on a positively charged glass slide. Ki67 and INI1 immunohistochemistry was performed in the Children’s National Hospital Pathology Department as a part of their standard of care procedure for clinical testing. Ki67 antibody (REF 790–4286; Ventana Medical Systems, Inc. Tucson, AZ, USA) or INI1 antibody (REF 760–4615; Cell Marque Corporation, Rocklin, CA, USA) were applied following the recommended staining protocol for the BenchMark Ultra instrument. A positive control was run with each antibody. NUT and BRG1 IHC was performed at Mayo Clinic Laboratory (www.mayocliniclabs.com; Test ID: NUT and BRG1) following their standard of care procedure for clinical testing.

Results

Case presentation

The patient was a 1.5-year-old male who presented to care with stalled development, drooling, left-sided face flushing, left-sided eyelid drooping and daily complex partial seizures. MRI imaging revealed a cystic mass with an enhancing nodular component measuring 11×13 cm in the temporoparietal region (Figure 1A,B). Diffusion weighted imaging was consistent with a cellular lesion. MRI of the spine was negative for metastatic disease. A gross total resection was performed without complication three days later (Figure 1C). At pathologic assessment, the tumor was a high-grade neoplasm with focal necrosis (Figure 1D). Undifferentiated cells with an embryonal morphology were the primary cell type, but there were also slightly larger cells with rhabdoid or epithelioid morphology (Figure 1E,F respectively). Mitotic figures were noted to be abundant (Figure 1G) and the tumor had a high proliferative index (Figure 1H). Immunohistochemistry (IHC) showed patchy staining for GFAP, EMA and cytokeratin. p53 IHC staining was consistent with wild type, and INI1 was retained (Figure 1I). Further IHC analysis showed retained BRG1 (Figure 1J), which ruled out an AT/RT. The patient was diagnosed with a WHO Grade IV CNS embryonal tumor with rhabdoid features based on the 2016 classification criteria [2]. This translates to WHO Grade IV CNS embryonal tumor (NOS) based on the 2021 classification criteria [3].

Figure 1. MRI and IHC images.

(A) Axial T2+contrast and (B) coronal T1+ contrast MRI imaging showed an enhancing supra-insular/inferior parietal mixed cystic/nodular neoplasm. (C) The lesion was completely removed by surgical resection. (D) H&E staining revealed a high-grade tumor organized in sheets (purple) with areas of necrosis (pale pink) (scale bar, 0.5 mm). (E,F) The undifferentiated tumor cells had an embryonal morphology and scattered larger cells with a vague (E) rhabdoid or (F) epithelioid morphology. Arrow in insets indicate (E) an epithelioid-like cell and (F) a rhabdoid cell with indented nucleus; scale bar, 0.05 mm). (G) Mitotic figures were abundant (arrow heads, arrow in inset indicates an example of a mitotic figure; scale bar, 0.05 mm). Ki-67 proliferative index was more than 90% (H; scale bar, 0.05 mm). (I) INI1 was retained as seen using IHC (scale bar, 0.05 mm). (J) BRG1 was retained as seen using IHC (scale bar, 0.05 mm). (K) A NUT antibody IHC (NeoGenomics) gave strong nuclear staining for NUT protein (scale bar, 0.1 mm).

The patient was treated with infant embryonal tumor chemotherapy (cisplatin, vincristine, etoposide, vincristine, cyclophosphamide, methotrexate) followed by 3 cycles of high-dose consolidation chemotherapy with stem cell rescue (carboplatin, thiotepa and etoposide). The patient had good response to the induction chemotherapy with no evidence of disease noted on the pre-consolidation MRI. During consolidation cycle #1, the patient developed hypotension and veno-occlusive disease of the liver. Despite maximum intervention, the patient passed away approximately one week later.

Because of the patient’s early presentation with a unique tumor, the cancer genetics team was consulted to rule out possible germline cancer predisposition. Family history was significant for a sibling who died in utero from presumed anencephaly, and distant history of leukemia, cervical cancer, colon and breast cancer on the maternal side of the family. A germline tumor panel including approximately 120 genes was performed and revealed a variant of unclear significance in the MC1R gene and a maternally inherited PTCH1 gene mutation that was initially a variant of unclear significance but has subsequently been reclassified as a likely benign variant. Neither of these germline gene mutations were thought to be associated with his tumor development.

Molecular characterization

Tumor analysis was performed on a research basis and included EPIC methylation array, optical genome mapping (OGM), genome sequencing, and RNA sequencing (see Materials and methods). The EPIC methylation array results were first entered into the CNS tumor methylation classifier (molecularneuropathology.org; MNP classifier), but the tumor did not cluster with any of the commonly diagnosed embryonal tumors (not classifiable) [19]. We then used the methylation classifier for sarcomas (molecularsarcomapathology.org), which showed identical results (not classifiable) [20]. Genome sequencing analysis for SNV and small INDELs did not reveal significant Tier 1 or Tier 2 somatic mutations. Specifically, a mutation in SMARCB1 was not identified. The tumor mutation burden was calculated to be <1 mut/Mb (low) and the majority of the identified variants were noted to be intronic.

OGM, which utilizes ultra-long DNA molecules to identify SVs, did not reveal a SV encompassing SMARCB1, and review of the CNV methylation plot did not demonstrate signal loss from SMARCB1. However, OGM showed complex three-way rearrangements amongst chromosomes 12, 18 and 15 in the tumor (supplementary material, Figures S1,S2). The most clinically significant event was an insertion that resulted in the fusion of ZNF532 on chromosome 18 with NUTM1 on chromosome 15 (Figure 2A). Additional rearrangements without clear clinical significance included (1) a translocation between chromosome 15 and 18; (2) an inversion on chromosome 18 near the same breakpoint as the translocation; (3) copy number variant change on chromosome 15 and (4) a small insertion on chromosome 18 derived from chromosome 12 (supplementary material, Figure S2). Subsequent analysis of the genome and RNA sequence datasets confirmed the presence of the ZNF532::NUTM1 fusion in the tumor (Figure 2B,C). No other clinically significant SVs or fusions were identified.

Figure 2. A ZNF532::NUTM1 fusion identified by optical genome mapping and confirmed by short-read sequencing.

(A) OGM genome browser view of the identified fusion. Top: in silico G-band staining of chromosome 18, followed by copy number and structural variant tracks. The green line and adjacent purple dots indicate the location of the insertion and corresponding breakpoints aligning to chromosome 15. The reference chromosomes 18 and 15 are shown in blue, with black vertical lines showing the DLE1 (Direct Labeling Enzyme 1) label locations. The assembled sample map is displayed in yellow, the red labels in the middle do not have alignment to chromosome 18, instead they align to chromosome 15. The total insertion size from chromosome 15 to 18 is approximately 200 kb. The left breakpoint on both chromosomes is magnified to show annotations for ZNF532 and NUTM1 genes, approximate breakpoint location and exons fused to exons. (B) Short-read sequencing alignments to the breakpoint location indicated by OGM. Read-pairs maps to two different chromosomes (chr18 and chr15), to confirm the identified translocation between an intron near the gene ZNF532 and exon 3 of the NUTM1 gene, respectively. (C) RNA-sequencing alignments also confirm the exon–exon fusion between ZNF532 (exon 7) and NUTM1 (exon 3). The split reads are designated using color, with part of the reads mapping to exon 7 of chromosome 18 and other part mapping to exon 3, chromosome 15. The red dotted lines differentiate the 2 chromosomes.

To better understand the role and frequency of ZNF532::NUTM1 fusions in pediatric brain tumors, a literature review using PubMed (https://pubmed.ncbi.nlm.nih.gov/) was performed with a series of search strings including ‘ZNF532::NUTM1, NUTM1 AND brain tumor’, and ‘ZNF532 AND brain tumor’. The search was limited to English language articles concerning human studies exclusively and with publication dates going back 15 years from March 2022. Five additional cases with a ZNF::NUTM1 fusion (four with ZNF532, one with ZNF592) and five reports of patients with NUTM1 rearranged brain tumors were identified (Figure 3A and Table 1) [21–24]. The specific ZNF532::NUTM1 fusion identified in our patient has not been reported in a child or a patient with a brain tumor prior to this case. However, this has been associated with adult lung, mandible, parotid gland, and pelvic bone cancers that have round cell and/or undifferentiated epithelioid morphology with or without a rhabdoid cell component, comparable to our patient’s tumor (Figure 3A, Figure 1D–G) [25–29].

Figure 3. NUTM1 fusion tumors.

(A) Oncoplot showing a summary of cancers with ZNF::NUTM1 fusions. (B) Kaplan–Meier curve showing overall survival of patients with brain tumors harboring NUTM1 rearrangements (red line) and ZNF::NUTM1 cancers (blue line).

Table 1:

Previous cases from brain tumors with NUTM1 fusions

Reference	Age/Sex	Site	Histology	IHC positive	Treatment/Outcome	Fusion
Dickson et al [21]	3 yo male	Left Parietal	Small round cells. Epithelioid-polygonal cells with a reticular-alveolar pattern and prominent myxoid stroma. Nuclear molding, speckled chromatin and conspicuous mitotic activity.	GFAP (2+, focal), synaptophysin (1+), NUT (5+).	Surgery, Chemo, DOD (12 mo)	BRD4:: NUTM1
Sturm et al [24]	3 yo female	Temporal/Parietal	Small-cell phenotype, alveolar and fascicular growth	NUT (strong)	Unknown AWD (273 mo)	CIC:: NUTM1
Sturm et al [24]	2 yo female	Frontal/Parietal	Small-cell phenotype, alveolar and fascicular growth	NUT (strong)	Unknown	CIC:: NUTM1
Siegfried et al [23]	21 yo female	Frontal	Fascicular architecture and chondro-myxoid areas; some neuron-like tumor cells; large nucleoli	NUT, GFAP (strong), p53, CD56.	Surgery, NED (16 mo)	ATXN1::NUTM1
Ko et al [22]	29 yo female	Right Frontal/Temporal	Variegated tumor consisting mostly of small epithelioid cells with myxoid or fibrillar background	NUT, CD99, CD56, p53, GFAP (focal), neurofilament (focal).	Surgery, Chemo, DOD (1 mo)	PARD3B:: NUTM1
This case	1.5 yo male	Right Temporal/Parietal	Embryonal cell types, as well as epithelioid or rhabdoid-like cell types	GFAP (focal), Ki-67 high, NUT (strong), p53 wildtype	Surgery, Chemo, DOT (5 mo)	ZNF532::NUTM1

GFAP=glial fibrillary acid protein; Chemo=chemotherapy; DOD=died of disease; AWD=alive with disease; NED=no evidence of disease; DOT=died of treatment.

NUTM1 fusions in brain tumors are rare, and primarily identified in children and adults with supratentorial small-cell tumors as summarized in Table 1 [21–24]. These tumors have a myxoid and/or fascicular architecture, and a few (CIC::NUTM1) have been added to the diagnostic category CIC rearranged sarcomas in the 2021 WHO classification for Central Nervous System Tumors [3,24]. Publicly available methylation data from two CIC::NUTM1 and one ATXN1:NUTM1 tumors were analyzed by the sarcoma and brain tumor methylation classifiers for comparison along with the present case. Similar to our case, the ATXN1::NUTM1 tumor did not classify with a known tumor subgroup. The CIC::NUTM1 tumors classified with the CIC rearranged sarcoma on the brain tumor classifier, which was expected as these tumors served as part of the reference cohort for this subgroup [24].

An analysis of embryonal (n= 19) and AT/RT (n= 29) samples available through the Children’s Brain Tumor Network database did not reveal any additional tumors with a ZNF532::NUTM1 fusion or a NUTM1 or ZNF532 rearrangement. However, this is a limited dataset and none of these tumors had pathology similar to the patient in this report. Additional analysis of a larger cohort of samples with embryonal and rhabdoid/epithelial features that do not demonstrate loss of INI1 or BRG1 would be helpful to determine the frequency of these fusions in pediatric brain tumors.

The patients reported in the literature received a variety of different treatments. A multimodality treatment approach using surgery along with radiation, chemotherapy or both was most commonly employed, similar to the approach used with our patient. The one-year overall survival was 40% and 45% in the NUTM1 rearranged brain tumor group and ZNF532::NUTM1 cancer group respectively (Figure 3B).

Discussion and conclusions

Molecular characterization of pediatric brain tumors can help with both diagnostic and prognostic stratification. This has helped improve therapeutic strategies for patients with many tumors including embryonal tumors such as medulloblastoma and AT/RT. In this patient’s case, the tumor was described as an embryonal tumor with rhabdoid/epithelioid features that did not have genomic characteristics of an AT/RT and therefore was diagnosed under the umbrella term CNS embryonal tumor with rhabdoid features. Based on the 2021 WHO criteria, this diagnosis is even more general, falling into the category CNS embryonal tumor, which includes tumors without specific genomic alterations [3]. Through optical genome mapping and subsequent whole genome and RNA sequencing, we uncovered a ZNF532::NUTM1 mutation, which, based on previous reports in literature and the patient’s genomic and histologic findings, was most likely the genomic driver in this patient’s tumor.

NUTM1, is the gene NUT midline carcinoma gene family member 1 and is located on chromosome 15q14 [30]. NUT rearranged tumors have a chromosomal rearrangement resulting in the fusion of the NUTM1 gene on chromosome 15 with a gene involved in transcription regulation. The most common fusion partner is BRD4 (chromosome 19) but fusions involving BRD3 (chromosome 9), NSD3 (chromosome 8) and ZNF genes such as the ZNF532 (chromosome 18) seen in our patient have also been described [30]. NUT rearrangements are most commonly associated with sarcomas and hematologic malignancies in both children and adults, although a variety of other cancers have also been identified. Children and infants with NUTM1 fusions in B-cell ALL have favorable prognosis, while the clinical impact of this fusion on sarcomas is still unclear [30,31].

Although the oncogenic impact of the ZNF532::NUTM1 fusion in CNS tumors has not been studied, functional studies of other cancers have shown that NUTMI fusion positive cells demonstrate lack of differentiation of epithelial cells and nuclear localization of the NUT protein, suggesting that NUTM1 fusions contributed to tumor development by associating with nuclear chromatin and interfering with cell differentiation [32,33]. Consistent with this finding, nearly all of our patient’s tumor cells displayed homogenous nuclear expression when using the monoclonal NUT antibody, confirming nuclear localization of the NUT protein in this tumor (NeoGenomics; Figure 1K). NUT rearranged tumors have also been described as having a lower mutation burden than other cancers, with an abundance of intronic mutations that do not affect canonical oncogenes or tumor suppressor genes [34,35]. This is consistent with the patient’s genome sequencing results that primarily showed intronic mutations with a low mutation burden and no other Tier1 or Tier2 pathogenic mutations.

Standard therapy for pediatric embryonal tumors includes a combination of surgery, radiation therapy and chemotherapy. The patient presented in this report had gross total resection followed by three cycles of induction chemotherapy and then one cycle of high dose chemotherapy followed by stem cell rescue before passing away from treatment complications. The patient had no radiographic evidence of disease at the time of passing. Upon review of CNS cases with NUTM1 fusions (Table 1) and patients with ZNF532::NUTM1 associated cancers (Figure 3A), most patients had similar multi-modality treatment including surgery, radiation therapy and chemotherapy. Surgery, with the goal for gross total resection and negative margins, was performed when possible and has been shown to have a positive impact on overall outcomes in some patients with midline (head and neck) NUT rearranged tumors [36–38]. Radiation therapy has also been shown to have some benefit in patients with NUT rearranged midline tumors and has primarily been used in older patients with localized disease [36]. Chemotherapy has had mixed responses and thus far has not been demonstrated to improve overall outcomes in patients with these tumors [30,37]. However, a standardized therapeutic approach has not yet been identified in children or adults with NUT rearranged tumors, so additional studies would be required to fully understand the role of surgery, radiation therapy, and chemotherapy in patients with these tumors.

For methylation analysis, we used the MNP classifier to determine which subgroups the tumors most closely clustered with. The CIC::NUTM1 fusion positive samples (Table 1) clustered with the CIC rearranged sarcoma subgroup. The ATXN1::NUTM1 tumor clustered most closely to the CIC::NUTM1 tumors (CIC rearranged sarcoma), with a score of 0.7, which suggests that this tumor may be related to CIC rearranged sarcomas. Our patient’s tumor clustered most closely to supratentorial ependymomas, however the score was low (0.15) and therefore this does not appear to cluster with any defined tumor types. Although all of the tumors identified on literature search had positive NUT staining on IHC, similar to the patient reported in this case report (Figure 3A, Table 1, Figure 1K), many of the CNS tumors had unique pathology and molecular characteristics, and therefore the molecular impact of these fusions in the tumor may not be the same as the ZNF532::NUTM1 fusion. Publicly available data from the ZNF532:NUTM1 non-CNS cases was not available, so it is unknown if the tumor presented in this case would have a similar methylation pattern as these tumors.

In this report, we have described a novel ZNF532::NUTM1 fusion associated with a CNS embryonal tumor that had rhabdoid/epithelial features on pathologic examination. Although we were not able to find another similar case, signifying the rare nature of this tumor, the distinct pathology and underlying molecular characteristics separates this tumor from other embryonal tumors, which may have an impact in terms of response to treatment and targeted therapeutic options. Therefore, screening of CNS embryonal tumors that cannot be classified as medulloblastoma, AT/RT, ETMR or another molecular subgroup for the NUT protein through IHC could be helpful to identify tumors with a NUTM1 fusion. If positive, additional fusion analysis through genome sequencing, RNA-sequencing or OGM can be used to identify the specific NUT rearrangement, which would be helpful for the identification of tumors with similar features. Methylation profiling may also be helpful in the future to distinguish tumors with different NUT rearrangements that have unique clinical behavior. Increased awareness and identification can ultimately result in additional studies with the goal to develop the best therapeutic options for these patients.

Supplementary Material

supinfo

Figure S1. Representation of the SVs identified in the case in the tissue sample and control blood sample

Figure S2. Complex rearrangements observed in the tumor sample

Data availability statement

Data generated and analyzed in this study is available in the following public databases: Whole Genome Sequencing Data SRA accession number PRJNA937173; RNA sequencing Data GEO accession number GSE225965; Methylation Data GEO accession number GSE226220; and Optical Genome Mapping Data DOI 10.5281/zenodo.7667681.