Malignant Bone Forming Neoplasm with NIPBL::BEND2 Fusion

Abstract

Conventional high-grade osteosarcomas are characterized by aggressive radiologic features, cytologic pleomorphism and complex genomics. However, rare examples of osteosarcomas remain challenging due to unusual histology, such as sclerosing or osteoblastoma-like features, which may require molecular confirmation of their complex genetic alterations. We have encountered such a case in a 17-year-old man, who presented with a third metatarsal sclerotic bone lesion, found incidentally in the workup of a foot trauma. The initial imaging revealed a lesion with sclerotic/blastic features proximally and lucent/lytic portion distally, findings interpreted consistent with osteoblastoma. The lesion was managed intra-lesionally with currettings and cryoablation, however, the microscopic findings were non-specific, showing a bland osteoblastic proliferation embedded in a densely sclerotic matrix. Subsequently, the patient developed two rapid recurrences; the first recurrence was treated similarly despite its associated soft tissue extension radiographically, and the histologic findings remained non-specific. The 2^nd recurrence showed a large mass, with bone destruction and soft tissue extension and an open biopsy revealed features of osteosarcoma with lace-like osteoid deposition, albeit with uniform cytomorphology. The subsequent below knee amputation showed features compatible with high grade osteosarcoma, including solid growth of uniform epithelioid cells, with vesicular nuclei and scant cytoplasm, set in a lace-like meshwork of osteoid matrix. There was significant mitotic activity and tumor necrosis. Tumor cells were positive for SATB2. Further molecular work-up was performed showing an unexpected NIPBL::BEND2 fusion, which has been previously reported in 2 cases of phosphaturic mesenchymal tumor (PMT). FGF23 (ISH) was performed and was negative. By DNA methylation profiling, unsupervised clustering and UMAP dimensionality reduction revealed grouping with high grade osteosarcomas and not with the PMT group. The patient received chemotherapy post-amputation and is alive without evidence of disease, with 10-month follow-up. We report an aggressive, overtly malignant acral bone-forming tumor, harboring a NIPBL::BEND2 fusion. Further studies are needed to evaluate the recurrent potential of this fusion in osteosarcomas and its relationship with PMT.

1. INTRODUCTION

Conventional high grade osteosarcoma is the most common primary bone malignancy in children and adolescents¹. Genomically, it is characterized by complex copy number alterations, chromotrypsis and mutations in tumor suppressor genes^2–4. Herein, we report a unique bone tumor harboring a NIPBL::BEND2 fusion, showing a monomorphic histology, osteoid matrix deposition and overt malignant features. Clinically, the patient followed an aggressive clinical course, and overall findings were in keeping with a high grade osteosarcoma. Recently, two cases of phosphaturic mesenchymal tumor (PMT) were identified with a similar NIPBL::BEND2 fusion^{5, 6}. Neither of these tumors were described as malignant. Histologic features of PMT or clinical and imaging evidence of tumor induced osteomalacia (TIO) were absent in the current case. Using methylation profiling, we explored the relationship of this tumor with a group of mesenchymal tumors, including high grade osteosarcoma and PMT.

2. MATERIALS AND METHODS

2.1. Histologic Features and Immunohistochemistry

Histologic slides obtained from formalin-fixed paraffin-embedded (FFPE) blocks of tissue were reviewed. Morphologic features, such as architectural pattern, infiltrative growth, epithelioid vs spindle cell cytology, nuclear features, cytoplasm, mitotic figures and necrosis, were assessed. Immunostain for SATB2 was performed using clone EP281 (Cell Marque) according to manufacturer’s manual.

2.2. FGF23 In Situ Hybridization Assay

Probe to FGF23 (ready to use, Advance Cell Diagnostic (ACD), RNAscope DAB kit) were utilized. In situ hybridization for FGF23 mRNA was performed in FFPE tissue and immunodetection was completed according to the manufacturer’s instructions.

2.3. RNA targeted sequencing

For targeted RNA sequencing, the TruSight RNA Fusion Panel (Illumina, San Diego, CA) was performed, as previously described⁷. Briefly, RNA was extracted from FFPE tissue using RNeasy FFPE Kit of Qiagen (Hilden, Germany). RNA-sequencing libraries were prepared using 20 to 100 ng total RNA. The samples were subjected to targeted RNA sequencing on an Illumina MiSeq (~3 million reads per sample). All reads were independently aligned with STAR (version 2.3) and BowTie2 against the human reference genome (hg19) for Manta-Fusion and TopHat-Fusion analysis, respectively.

2.4. Sarcoma classification by DNA Methylation Profiling

2.4.1. DNA Methylation Data Set

The details of DNA methylation has been previously described^{8, 9}. Briefly, DNA was isolated from formalin fixed paraffin embedded tumor tissue that was macrodissected from prepared slides. DNA was treated with sodium bisulfite followed by whole genome amplification and processing on the Illumina methylation EPIC/850k bead array platform. Data was processed using the DKFZ random forest-based sarcoma (version 12.2) classifier package. To enhance our analysis, publicly available data sets were used. Raw IDAT files for samples from the Heidelberg sarcoma methylation classifier reference cohort (Gene Expression Omnibus (GEO) study accession number GSE140686). In addition, four cases of PMT from ArrayExpress, accession number: E-MTAB-9875 was used. Taken together, the dataset of DNA methylation arrays was composed of 9 Giant cell tumor of bone, 22 Ewing sarcoma, 8 BCOR- altered sarcoma, 11 epithelioid sarcoma, 7 undifferentiated sarcoma, 4 PMT, 13 ossifying fibromyxoid tumor, 10 extraskeletal myxoid chondrosarcoma, 8 chondrosarcoma (IDH group A), 9 chondrosarcoma (IDH group B), 7 clear cell chondrosarcoma, 6 sclerosing epithelioid sarcoma, 17 high grade osteosarcoma, 10 chondroblastoma, 11 CIC-rearranged sarcoma, and 8 mesenchymal chondrosarcomas. IDAT processing on the reference samples and the current cohort cases was performed using R software version 4.3.2 and the “minfi” package version 1.48.0¹⁰. Normalization was carried out using the preprocess Illumina function. Probes that had a detection p-value greater than .01, along with probes related to single-nucleotide polymorphisms and those located on sex chromosomes, were excluded. Methylation levels were assessed using β values. For unsupervised clustering, the T-distributed Stochastic Neighborhood Embedding (t-SNE) method was used for dimensionality reduction. The input data matrix was normalized by adjusting the mean of each column to zero. The “Rtsne” package (version 0.17) was used to analyze the 10,000 most variable CpGs, determined by variance and a t-SNE plot was constructed.

3. RESULTS

3.1. Case report

3.1.1. Clinical and Pathologic Features

A 17 year-old man presented with foot pain after trauma. Imaging studies identified an incidental, markedly sclerotic lesion in the third metatarsal. The lesion was confined to the bone without aggressive features radiographically, with a mixed blastic and lytic appearance (Supplementary Figure 1). The mass was treated intra-lesionally with curettings, cryoablation and bone grafting. The material showed mostly non-specific findings, with mineralized matrix deposition with associated osteoblastic rimming and significant sclerosis. Cytologic atypia was not appreciated. Six months later, imaging studies documented local recurrence, however, the overall findings were interpretated in the spectrum of osteoblastoma/osteoblastic neoplasm, despite its associated with soft tissue extension (Supplementary Figure 1). Biopsy of this recurrence was similarly limited, with scant tissue, showing rather nonspecific features, including bone trabeculae with osteoblastic rimming and fibrosis. Another round of cryoablation and bone grafting was performed. Within 5 months, the patient developed a second recurrence, characterized by a large mass, overt bone destruction and soft tissue extension (Supplementary Figure 1). An open biopsy was performed and the diagnosis of osteosarcoma was rendered. Subsequently, the patient underwent below knee amputation. On gross examination, the third metatarsal mass measured 6.3 cm in largest dimension, with extension into soft tissue, as well as into the second and fourth metatarsal and cuneiform bones. The cut surface was tan-white, with focal hemorrhage and gritty surface (Figure 1).

Figure 1:

Gross appearance shows a large tan-white tumor with hard, gritty cut surface, centered in the third left metatarsal. The tumor was associated with bone destruction, soft tissue extension and involvement of second and fourth metatarsal bones.

Histologically, the amputation mass showed a densely sclerotic neoplasm composed of round to epithelioid cells with scant eosinophilic cytoplasm and round nuclei with mild atypia, vesicular chromatin, and prominent nucleoli. The cells were set in a lace-like meshwork of osteoid matrix either wrapping haphazardly the overtly malignant tumor cells or arranged in more linear ray-like pattern (Figure 2). In some areas, the non-mineralized, hyalinized stroma divided the tumors cells into straight columns, highly reminiscent of sclerosing epithelioid sarcoma. The tumor showed dense sclerosis in the center, while the periphery appeared more cellular with solid sheets of compact tumor cells and minimal intervening osteoid matrix. These latter areas resembled a primitive round cell sarcoma (Figure 2). Mitotic figures were easily identified (10 mitotic figures per 10 high power fields) and tumor necrosis was present (about 30% of the specimen). Rare osteoclast-like giant cells were present. Cartilaginous matrix was not identified. Resection margins were negative and lymphovascular invasion was not identified. Tumor cells were diffusely positive for SATB2 (Figure 2). The patient received adjuvant MAP protocol (methotrexate, doxorubicin and cisplatin). Follow up information was available for 10 months, in which the patient remained alive without evidence of disease.

Figure 2:

A. Densely sclerotic tumor with ray-like arrangement of closely juxtaposed woven-bone trabeculae with intervening low cellularity. B. Higher power magnification showing lace-like osteoid matrix deposition with variable mineralization, wrapping around overtly malignant pleomorphic cells. C. Areas of marked increased cellularity devoided of prominent stroma, composed of sheets of uniform epithelioid cells with enlarged round nuclei, fine chromatin and prominent nucleoli. D. Other areas of the tumor show non-mineralized stromal collagen arranged in thick columns, with a distinctive ‘clinging’ pattern of the tumor cells. These areas are reminiscent of sclerosing epithelioid fibrosarcoma. E. Periphery of the tumor was composed of sheets of monomorphic small round cells with minimal intervening stroma, resembling Ewing sarcoma or small cell osteosarcoma. F. The tumors cells were diffusely positive for SATB2.

3.2. FGF23 ISH

FGF23 ISH was negative in tumor cells and no punctate intracellular staining was detected.

3.3. Molecular Findings

Targeted RNA sequencing was performed on the open biopsy of the second recurrence and revealed an in-frame gene fusion between NIPBL (NM_133433.4) exon 6 and BEND2 (NM_153346.5) exon 6 (Figure 3). The fusion product retained the two BEN protein domains of the BEND2 gene.

Figure 3:

Schematic representation of NIPBL::BEND2 gene fusion resulting from an in-frame gene fusion between NIPBL exon 6 (NM_133433) and BEND2 exon 6 (NM_153346). The projected chimeric protein retains the two BEN protein domains of the BEND2 gene. Chromosomal location of partner genes, exonic breakpoints and protein domain are depicted.

DNA methylation studies showed a low calibrated score and could not be classified into any of the methylation classes by the DKFZ classifiers for sarcomas (version 12.2). Unsupervised clustering using t-SNE analysis placed the tumor neatly with high grade osteosarcomas (Figure 4).

Figure 4:

Methylation Profiling demonstrates that the sarcoma index case with NIPBL::BEND2 fusion (light pink) clusters close to the high grade osteosarcomas (gray), and away from the phosphaturic mesenchymal tumor cluster (orange). Color coded t-SNE plot, t-distributed stochastic neighbor embedding using the 10,000 most variable DNA-methylation probes of array-generated DNA-methylation profiles from the Illumina Infinium HumanMethylation450 or EPIC BeadChip (Illumina, San Diego, USA).

4. DISCUSSION

In this brief report, we present an unusual malignant bone-forming tumor characterized by a monomorphic cytomorphology and osteoid deposition, harboring a NIBPL::BEND2 fusion. The tumor occurred in the metatarsal bone of a young adult, which had a mixed sclerotic/blastic appearance proximally and a lucent/lytic portion distally, consistent with osteoblastoma radiologically. A definitive diagnosis could not be reached due to the limited biopsy material, sclerotic nature and lack of pleomorphism. From the subsequent 2^nd local recurrence amputation procedure, molecular analysis revealed an unexpected NIBPL::BEND2 fusion, pointing towards the possibility of a phosphaturic mesenchymal tumor (PMT). However, a PMT diagnosis was unlikely in the presence of overtly malignant features, such as increased mitotic activity and necrosis, findings more in keeping with a high-grade osteosarcoma. Thus, further DNA methylation profiling was performed which revealed close clustering with high grade osteosarcomas rather than PMT. Moreover, the patient followed an aggressive clinical course with multiple local recurrences ultimately requiring below knee amputation.

Based on the non-aggressive imaging findings and the limited initial biopsy of bone trabeculae with osteoblastic rimming, significant sclerosis and lack of cytological atypia, the initial consideration was that of a benign osteoblastic neoplasm. Osteoblastoma is a benign tumor, most commonly involving the posterior vertebral elements, but can occur at other sites, including pelvis, limbs and craniofacial bones^{11, 12}. Osteoblastomas of metatarsal bone are exceedingly rare and generally follow a benign course with rare instances of recurrence^11–15. In a large AFIP study of 329 osteoblastomas, 41 were located in the foot and ankle (12.5%). The majority were located in the talus with 7 tumors in metatarsal bones. All metatarsal osteoblastomas were morphologically benign, with exception of one tumor that showed focal atypical features (cellularity, nuclear pleomorphism and mitotic activity) in the background of usual osteoblastoma features ¹⁶.

Recently, the genetic hallmarks of osteoblastoma have been elucidated, with rearrangements of FOS and FOSB genes present in a large majority of cases ^{4, 17}. FOS/FOSB gene rearrangements were absent in the current case and areas of benign appearing osteoblastoma were not identified to suggest possible transformation from osteoblastoma to osteoblastoma-like osteosarcoma. The concept of malignant progression in osteoblastoma remains unresolved. Moreover, the recent characterization of a group of challenging sclerosing osteoblastic lesions, designated as ‘atypical sclerosing osteoblastic neoplasm’, brought an additional level of complexity to the differential diagnosis ¹⁸. These lesions encompass a group of unusual bone forming tumors that do not fit into conventional benign osteoblastoma, ‘aggressive’ osteoblastoma, and ‘osteoblastoma-like’ osteosarcoma. In that study of 8 cases, the initial diagnosis ranged from osteoblastoma, unusual bone forming tumor, low grade osteoblastoma-like osteosarcoma and osteosarcoma. The tumors appeared to involve the appendicular skeleton, with 3 of 8 cases arising within small metatarsal bones. Morphologically, at initial biopsy/curetting, all 8 tumors were characterized by the deposition of compact, sclerosing, sheet-like bone containing few osteoblasts. Where interface with host bone was available, these tumors appeared relatively well-circumscribed, at times showing maturation and without any evidence of permeation. Worrisome features for malignancy including sheets of osteoblasts without intervening stroma, significant nuclear atypia or tumor necrosis were absent. The mitotic count was <1 up to 2-3 mitotic figures per 10 high power fields. Local recurrence occurred in 71% patients and one patient died of disease status post amputation, with presumed lung metastasis. Morphologic features of recurrent tumors generally included no evidence of atypia, unchanged histologic features from initial biopsy, and one tumor with bone entrapment despite non-aggressive clinicoradiologic features. Only one recurrent tumor was distinct and showed findings of significant nuclear atypia, tumor necrosis, sheets of tumor cells without intervening stroma, soft tissue extension and telangiectatic features. This tumor was deemed to have undergone transformation or dedifferentiation to a high grade osteosarcoma. This is the patient who succumbed of disease status post amputation, with presumed lung metastasis (11 years after initial diagnosis, although he had other comorbidities). Of interest, this tumor occurred in third metatarsal of a 55 year-old man. Due to acid decalcification of tissue, molecular studies such as FISH, sequencing or methylation profiling could not performed in this cohort of challenging bone tumors. The relationship of this case to ours remains undetermined. This series highlights the uncertainty in diagnosis of unusual osteoblastic lesions in unusual locations. Although extended next generation sequencing is not performed in conventional high grade osteosarcomas, the current case was subjected to genomic and epigenetic testing as it had unusual morphologic features, such as monomorphic cytology and marked sclerosis, which made the final diagnosis quite challenging.

Two recent studies identified a NIPBL::BEN2 fusion in two cases of benign phosphaturic mesenchymal tumor (PMT) of bone^{5, 6}. PMTs are rare mesenchymal tumors occurring in both bone and soft tissue. Morphologically benign, tumors are characterized by bland spindle to stellate cells set within a well-developed capillary meshwork. The histologic hallmark of PMT is the presence of grungy calcifications, sometimes forming flower-like crystals. Some cases exhibit larger caliber vessels with hemangiopericytoma-like pattern, adipose tissue, as well various forms of matrix deposition, such as peripheral shell of bone, immature cartilage or osteoid¹⁹. Rare cases of malignant PMT have been reported and demonstrate overt features of malignancy²⁰.PMTs account for the majority of mesenchymal tumor-associated osteomalacia²¹. The underlying mechanism is via excess production of FGF23 inhibiting renal reuptake of phosphate. About 50% of PMTs harbor FN1::FGFR1 fusions²², with a small subset showing FN1::FGF1 fusions²³. Fusion negative PMTs are shown to overexpress klotho²⁴. Sakai et al⁵ reported a case of osteoblastoma-like PMT of the fibula in a 12 year-old boy with one year history of muscle weakness and gait disturbance. Imaging studies showed deficiency of mineralization similar to Rickets and a radiolucent lesion in left fibula. Clinical labs detected elevated FGF23 and hypophosphatemia. A NIPBL::BEND2 fusion was identified by RNA sequencing, which involved exons 1-6 of NIPBL and exons 7-14 of BEND2. Forced expression of the fusion in osteoblastic cell line promoted cell proliferation with significant upregulation of MYC pathway; however, expression of FGF23 and FGFR1 was not significantly changed. The report includes very limited histologic description of the tumor, described as “irregularly deposited osteoid and osteoblast-like tumor cells scattered between the osteoid”. Post enbloc resection, levels of FGF23 dropped and there was no evidence of hypophosphatemia. After 21 months follow-up, the patient had no evidence of disease and no systemic symptoms. The second case of NIPBL::BEND2 fusion positive PMT was identified as part of a comprehensive RNA based sequencing of 76 cases⁶. The case occurred in the ilium of a 31-year-old male. There were no H&E photomicrographs included for this case. The fusion breakpoints were comprised of exon 6 of NIPBL and exon 6 of BEND2. In contrast to these cases, our case lacked classic PMT features as well as clinical or imaging evidence of tumor-induced osteomalacia.

NIPBL gene (Nipped-B-like) is located at 5p13.2 and was cloned as the molecular basis of Cornelia de Lange syndrome (CDLS1)^{25, 26}. This syndrome follows an autosomal dominant inheritance and is characterized by dysmorphic facial features, intellectual disability, growth delay and limb defects. Mutations in NIPBL are the most commonly identified mutations in Cornelia Lange syndrome. NIPBL gene encodes the homolog of the Drosophila melanogaster Nipped-B gene product and fungal Scc2-type sister chromatid cohesion proteins. The encoded protein, delangin (preferred name nipped-B protein) is a cohesin regulatory factor and plays major role in sister chromatid cohesion, target gene expression, DNA repair and developmental regulation^25–29. NIPBL expression is also involved in human carcinogenesis. Data indicates mutations/deletions in NIPBL play a role in gastric and colorectal cancer^{30, 31}, acute myeloid leukemia³², urothelial carcinoma^{33, 34}, and gliomas³⁵. Only few examples of NIPBL-related fusions have been reported, including NIPBL::NACC1 fusion in cholangioblastic variant of intrahepatic cholangiocarcinoma³⁶, NIPBL::ERG in atypical tenosynovial giant cell tumor³⁷ and NIPBL::HOXB9^{38, 39} and NIPBL::TV6⁴⁰ in acute megakaryocytic leukemia.

BEND2 gene (BEN domain containing 2) is located at Xp22.13. The encoded protein has two BEN domain in C-terminus and such domains play a role in chromatin restructuring and transcription via protein and DNA interactions. Analysis of tissue-specific expression highlights strong expression of BEND2 protein in testicular tissue⁴¹. BEND2-related fusions have been reported in pancreatic neuroendocrine tumor (CHD7::BEND2⁴² and EWSR::-BEND2^{43, 44}), single case of salivary gland adenocarcinoma, not otherwise specified (EWSR1::BEND2⁴⁵), and astroblastoma/astroblastoma-like/ neuroepithelial tumors spectrum (MAMLD1::BEND2⁴⁶, EWSR1::BEND2^{47, 48}, MN1::BEND2^{49, 50}). Recently, a single case of low grade sinonasal sarcoma EWSR1::BEND2 ⁵¹ and an abdominal wall soft tissue sarcoma (MN1::BEND2⁵²) have been reported.

In summary, we present a unique case of a malignant acral bone forming neoplasm with features in keeping with osteosarcoma, albeit with monomorphic cytology, which displayed a NIPBL::BEND2 fusion. Methylation profiling studies supported the diagnosis as it showed clustering with high grade osteosarcomas. The shared gene fusion alteration with PMT remains puzzling, as both previously reported cases were benign and one had documented association with osteomalacia. Further studies are needed to determine if the NIPBL::BEND2 fusion defines a spectrum of both benign and malignant bone tumors, and to further clarify the relationship to PMT.

Supplementary Material

NIPBL-SF1-imaging

S Figure 1. Radiographic images of the initial presentation, first and second recurrences.

A. Initial diagnostic images (oblique radiograph) of the left foot demonstrates a proximal to mid 3^rd metatarsal lesion with both sclerotic (proximal, arrow) and lucent/lytic (distal, arrowhead) appearances. B. Sagittal CT confirms the 3^rd metatarsal lesion with mixed sclerotic (arrow) and lytic (arrowhead) portions. There is mild osseous expansion but no cortical breakthrough or soft tissue component. C. Images of the first recurrence include a sagittal CT reformation revealing interval development of cortical breakthrough with small soft tissue components of the blastic portion of the lesion (curved arrows) and enlargement of the lytic portion (arrowhead). D. Images of the second recurrence, leading to amputation: Anteroposterior radiograph of the left foot demonstrates marked extraosseous expansion with mineralized material (arrowheads), partially overlapping the intermediate and lateral cuneiforms and proximal 2^nd and 4^th metatarsals.

Data Availability Statement:

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.