Optical Genome Mapping Identifies a Novel Pediatric Embryonal Tumor Subtype with a ZNF532-NUTM1 Fusion

Case presentation

The patient was a male between 0-5 years of age and 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 (GTR) was performed without complication three days later (Figure 1C). On pathology, 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 epithelioid or rhabdoid morphology (Figure 1 E&F). Mitotic figures were noted to be abundant (Figure 1E) and the tumor had a high proliferative index (Figure 1G). Immunohistochemistry (IHC) showed patchy GFAP, EMA and cytokeratin staining. p53 IHC staining was consistent with wild type and INI1 was retained (Figure 1H). Further IHC analysis showed retained BRG1, 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].

• Download figure
• Open in new tab

Figure 1: MRI and IHC images.

Axial T2+contrast (A) and coronal T1+ contrast (B) MRI imaging showed an enhancing supra-insular/inferior parietal mixed cystic/nodular neoplasm. The lesion was completely removed with surgical resection (C). (D) H&E staining revealed a high-grade tumor organized in sheets (purple) with areas of necrosis (pale pink) (scale bar = 0.5mm). (E) Mitotic figures were abundant (arrow heads; scale bar = 0.05mm). (F) The undifferentiated tumor cells had an embryonal morphology and scattered larger cells with a vague epithelioid or rhabdoid morphology (arrow in inset shows an epithelioid-like cell; rhabdoid not seen in picture; scale bar = 0.05mm). Ki67 proliferative index was more than 90% (G; scale bar 0.05mm). (H) INI1 was retained on IHC. (I) NUT antibody IHC (NeoGenomics) showed strong nuclear staining of NUT protein (scale bar = 0.1mm).

The patient was treated with standard 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, 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, whole genome sequencing, and RNA sequencing (See supplementary file for methods). The EPIC methylation array results were entered into the tumor methylation classifier (MolecularNeuropathology.org) but the tumor did not cluster with any of the commonly diagnosed embryonal tumors (not classifiable) [10]. Whole genome sequencing analysis for single nucleotides and small insertions/deletions did not reveal significant Tier1 or Tier 2 somatic mutations. The tumor mutation burden was calculated to be <1 mut/Mb (low) and a majority of the variants identified were noted to be intronic.

Optical genome mapping (OGM; Bionano Saphyr Instrument), which utilizes ultra-long DNA molecules to assess for structural variants, showed complex three-way rearrangements amongst chromosomes 12, 18 and 15 in the tumor (Supplementary Figure 1). 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 Figure 2). Subsequent SV analysis of the whole genome sequencing data and additional RNA-sequencing confirmed the presence of the ZNF532-NUTM1 fusion in the tumor (Figure 2B&C). No other clinically significant SVs or fusions were identified.

• Download figure
• Open in new tab

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

(A) OGM genome browser view of the identified fusion. Top: 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 label locations. The assembled sample map is displayed in yellow, the red labels in the middle don’t have alignment to chromosome 18, instead they align to chromosome 15. The total insertion size from chromosome 15 to 18 is approximately 200kb. The left breakpoints on both chromosomes is magnified to show annotations of 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 ZNF532 and exon 3 on NUTM1 genes, respectively. (C) RNA-sequencing alignments also confirm the exon-exon fusion between ZNF532 (exon 10) and NUTM1 (exon3). The split reads are designated by color, with part of the reads mapping to exon 10 of chromosome 18 and other part mapping to exon 3, chromosome 15. The red dotted lines differentiate the 2 chromosomes.

In order to better understand the role and frequency of ZNF532-NUTM1 fusions in pediatric brain tumors, a literature review using Pubmed 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 Table1). NUTM1 fusions in brain tumors are rare but have been identified in children and adults with supratentorial small-cell tumors similar to our patient (Table 1) [11–14]. 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-F) [15–19].

• Download figure
• Open in new tab

Figure 3: NUTM1 fusion tumors.

(A) Oncoplot showing cancers with ZNF-NUTM1 fusions. Ages are given as a 5 year range. (B) Kaplan-Meier curve showing overall survival of brain tumors with NUTM1- rearrangements (red line) and ZNF-NUTM1 cancers (blue line).

View this table:

• View inline
• View popup
• Download powerpoint

Table 1: Previous cases from brain tumors with NUTM1 fusions

The patients reported in literature had 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).

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 round cell tumors and rhabdoid/epithelial features would be helpful to determine the frequency of these fusions in pediatric brain tumors.

Sample processing

Ultra-high molecular weight (UHMW) DNA was extracted following manufacturer’s guidelines (Bionano Genomics Inc, USA) from flash frozen brain regions (superior frontal gyrus and primary visual cortex) as well as pelleted frozen PBMCs. Briefly, a total of 15-20mg 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 nanobind magnetic disk. Bound UHMW DNA was resuspended in the elution buffer and quantified with Qubit dsDNA assay kits (ThermoFisher Scientific). Total RNA was extracted using Qiagen RNeasy Kit following manufacturer’s guidelines (Qiagen, Germany). RNA sequencing was performed at Novogene Inc, 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, USA).

DNA labeling was performed following manufacturer’s protocols (Bionano Genomics, USA). Direct Labeling Enzyme 1 (DLE-1) reactions were carried out using 750 ng of purified UHMW DNA. Labeled DNA was loaded on Saphyr chips for imaging. The fluorescently labeled DNA molecules were imaged sequentially across nanochannel arrays (Saphyr chip) on a Saphyr instrument (Bionano Genomics Inc, USA). Effective genome coverage of greater than 500X was achieved for all samples. All samples also met the following QC 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 and 3.7, respectively), were used for data processing and variant calling. 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. 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 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 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 CN 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, CN 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 CN 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, USA) was used for SV annotation and filtering. Variants were filtered in access and nanotatoR [29] based on the following criteria: for de novo, rare variant and CNV pipelines, SVs were filtered based on Bionano Genomics recommended size and confidence cutoff values (e.g., >500bp/5kbp size cutoff for de novo assembly and rare variant pipelines respectively for 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 [30], followed by processing of the aligned bam (for variant calling) using SAMtools [31] and Picard (Broad Institute). Next, for small nucleotide variant (SNV) and small insertion and deletion (INDEL) calling we use Mutect2 [32], followed by annotation using ANNOVAR [33]. Due to absence of a non-tumor tissues from the same individual, we used a 1000 genome panel of normal (PON) variant call file from 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 [34], followed by annotation using AnnotSV [35]. For the SV visualization Integrative Genome Viewer (IGV) was used.

RNA-sequence analysis

RNA-seq data was aligned to GRCh38 reference genome, followed by fusion calling using STAR-Fusion [36]. Visualization of the fusion was performed using Clinker [37] and IGV.

EPIC methylation chip analysis

An input of 300 ng of DNA was bisulfite-converted using the DNA Methylation-Lightning kit (Zymo Research). 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). 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, CNS tumor classification tool hosted at molecularneuropathology.org, was used on the methylation signal files [10].