Genomic Analysis Does Not Support Malignant Transformation of Osteoblastoma to Osteosarcoma

INTRODUCTION

Although osteoblastomas and osteosarcomas are both bone-forming tumors, they represent entirely different clinical and histologic entities. Nevertheless, rare reports exist of malignant transformation, which have raised the possibility that genetic similarities underlie their development. In a search of the literature dating back 50 years, only 24 cases have been reported. Given the rarity of this clinical phenomenon and lack of objective genetic data, it remains unclear whether these cases are indeed transformational events or simply coincidence. We have identified a single patient who was initially diagnosed with a benign osteoblastoma and subsequently presented years later with a high-grade osteosarcoma in the same anatomic region. This scenario provided an opportunity to compare and contrast the genomic landscapes of these two tumors.

CASE REPORT

A 16-year-old male presented in 2010 with a 6-month history of lower back pain. Magnetic resonance imaging demonstrated an infiltrative 3.3-cm soft tissue mass that arose within the posterior aspect of S2 and extended within the neural foramina (Fig 1). A biopsy sample of the mass demonstrated conventional-type osteoblastoma with focal epithelioid features for which the patient underwent intralesional curettage (Fig 2). Subsequent imaging demonstrated a persistent abnormal signal in the sacrum, and a plan of continued surveillance was agreed upon. Over the ensuing 4-year period, imaging failed to demonstrate appreciable radiographic change, and the patient’s clinical course remained stable.

FIG 1.

(A) Magnetic resonance imaging in 2010 of the lumbar spine. The axial, T1-weighted, fat-suppressed image demonstrates a destructive, infiltrative lesion that extended into the canal at the level of S2. (B) Magnetic resonance imaging in 2010 of the lumbar spine. The axial, T1-weighted, postcontrast, fat-suppressed image demonstrates significant enhancement after contrast administration.

FIG 2.

(A) A 2010 hematoxylin and eosin–stained low-power (10×) micrograph that demonstrates findings consistent with osteoblastoma, including trabecular, calcified osteoid, and polygonal cells with prominent cytoplasm. (B) A 2010 hematoxylin and eosin–stained low-power (40×) micrograph that demonstrates relatively bland-appearing spindle cells with small nuclei and occasional giant cells.

Five years later, the patient returned with increasing pain. Imaging demonstrated an increase in the size of the mass, and an urgent open sacral biopsy was performed (Fig 3). Histologic findings were consistent with high-grade osteosarcoma with secondary aneurysmal bone cyst features (Fig 4). The prior biopsy sample was re-evaluated in light of these findings, confirming that the two tumors represented distinct entities. Staging studies performed at that time were negative for distant disease.

FIG 3.

Magnetic resonance imaging in 2015 of the pelvis where (A) the axial, T1-weighted image demonstrates enlargement of the sacral mass now measuring 7.3 cm in the maximal transverse dimension; (B) the axial, T2-weighted image demonstrates heterogeneity and fluid-fluid levels; and (C) the axial, T1-weighted, postcontrast, fat-suppressed image demonstrates extensive enhancement after contrast administration.

FIG 4.

(A) A 2015 hematoxylin and eosin–stained low-power (10×) micrograph demonstrates findings consistent with high-grade osteosarcoma, including high cellularity, multiple foci of tumor necrosis, and osteoid formation. (B) A 2015 hematoxylin and eosin–stained high-power (40×) micrograph that demonstrates pleiomorphic and bizarre cellular features.

Neoadjuvant chemotherapy was started in accordance with Children’s Oncology Group protocol AOST0331; however, because of severe pain, the patient required semi-urgent surgical intervention and did not complete planned neoadjuvant treatment. The patient underwent an R0 sacrectomy that extended from S2 to S5 followed by adjuvant chemotherapy. Two years postresection, a 0.6-cm left-upper-lobe lung nodule was detected on computed tomography. The nodule was resected, with histology confirming metastatic osteosarcoma. The decision was made to manage conservatively, and as of July 2019, surveillance demonstrated no evidence of disease.

METHODS

Patient consent and institutional review board approval were obtained before study initiation. Whole-genome sequencing was performed on samples obtained from a single patient, including the osteoblastoma, the osteosarcoma, and a matched germline sample obtained by buccal swab. Sequencing and data analysis were performed by Novogene Technology (Beijing, China) to evaluate for structural variants, copy number variation, single nucleotide variants, and insertions/deletions. Germline-matched comparison from each tumor was undertaken to evaluate for somatic changes.

To evaluate for pathogenic variants, somatic analysis on small variants focused on changes that would elicit a coding or splice site change and were not common polymorphisms (allele frequency < 0.01 in 1,000 Genomes and ExAC databases). Furthermore, we evaluated whether any variants were found in cancer mutational hotspots (recurrence of more than five in COSMIC [Catalog of Somatic Mutations in Cancer]) or were a truncating variant in a known tumor suppressor gene (Cancer Gene Census). To evaluate for potentially pathogenic germline variants, we filtered common polymorphisms in the same way and manually reviewed findings annotated by ClinVar as being pathogenic or likely pathogenic for a cancer predisposition syndrome.

CONTEXT

• Key Objective

• Does genomic analysis of a patient who was initially diagnosed with a benign osteoblastoma and subsequently presented years later with a high-grade osteosarcoma in the same anatomic region demonstrate evidence of malignant transformation?

• Knowledge Generated

• There was near-zero overlap in the somatic small variants, somatic copy number variation pattern, and predicted structural variants in the osteoblastoma compared with the osteosarcoma. Findings from this study argue against malignant transformation as an evolving or stepwise process and conversely support two distinct neoplasms with dissimilar genetic makeups.

• Relevance

• This study performed an in-depth genetic characterization of two distinct tumors that historically have been believed to be along the same spectrum of disease. In addition, a germline mutation in BRIP1 was discovered, which lends itself to additional investigation.

RESULTS

Whole-genome sequencing resulted in high-quality data, which generated an average 63 million reads per tumor, with more than 98% usable on the basis of quality metrics. After filtering and duplicate removal, mapping to hg19 resulted in an average sequencing depth of 29× and 32× for the benign and malignant tumors, respectively, and at least 4× coverage in more than 99% of the genome.

Somatic variant analysis revealed approximately 98,000 and 141,000 high-confidence small variants in the benign and malignant tumor, respectively, which corresponded to somatic mutational burdens of 33 and 47 mutations/megabase (Mb). Of these, 327 (osteoblastoma) and 359 (osteosarcoma) variants were predicted to elicit a coding or splice site change and were not common polymorphisms. Of note, zero overlap was found at the position level in these somatic coding/splice site variants between tumors. Evaluation for potential driver variants in the osteoblastoma did not identify any cancer hotspot somatic mutations or truncating mutations in known tumor suppressors. In the osteosarcoma, a single variant recurrent in COSMIC of unclear oncologic significance, ZNF429 p.N426K, was found as well as truncating mutations in two tumor suppressor genes, DAXX and NCOR1.

Structural and copy number variation analysis revealed significant alterations in both tumors. The osteoblastoma demonstrated large chromosomal areas of copy number loss (monosomy), including all of 1p as well as 9.8-Mb, 15.9-Mb, and 3.8-Mb areas on chromosome (chr) 6, chr15, and chr22, respectively. In the osteosarcoma, chromosomal gains were frequent, which resulted in near-complete trisomy, with monosomy of chr9 also seen. Several areas of focal copy number gain were observed in both tumors, although none in an area of a known oncogene. A striking number of somatic inversion events were found in both tumors (approximately 1,750 per tumor), particularly compared with other structural changes, such as translocations, deletions, or tandem duplications (each approximately 10 to 30 per tumor). Similar to somatic small variants and copy number variation, there was near-zero overlap in the predicted structural variants in the osteoblastoma compared with the osteosarcoma.

Germline analysis using the methods described identified a single pathogenic germline mutation, BRIP1 R798X, identified with high confidence in all three samples. The BRIP1 truncating variant was predicted to be heterozygous in all samples, with no evidence of loss of heterozygosity in either tumor. A summary of genomic findings is listed in Table 1.

TABLE 1.

Summary of Genomic Findings

DISCUSSION

The notion that osteoblastoma can undergo malignant transformation to osteosarcoma dates back to an original report in 1967.¹ Since then, 24 cases have appeared in the literature, none of which investigated the transformation event on a genomic level^1-20 (Table 2). Given the lack of objective genetic data and short time to transformation in some cases, there is a distinct possibility that some of these reports represent coincidence or diagnostic error.

TABLE 2.

Summary of the Literature

A few prior reports have attempted to document malignant transformation through a limited analysis of sequential biopsies samples.^4,18,21 However, these tumors were characterized only in terms of ploidy, thereby limiting understanding of the transformation process.^4,18,21 Although these reports suggested that transformation is associated with an increase in DNA content, only one patient’s tumor showed an observable change in DNA content over time. Moreover, in the current study, we have demonstrated the benign and malignant tumors to have distinct genomic profiles with almost no overlap in somatic changes. Genomic analysis fails to support malignant transformation in this patient. Whether these tumors are driven in part by an unrecognized anatomic predisposition remains speculative.

We also identified a germline mutation in the BRIP1 tumor suppressor gene, a member of the RecQ helicase family associated with BRCA1.^22,23 BRIP1 mutations are associated with a predisposition to breast cancer, ovarian cancer, and Fanconi anemia,^22,24,25 and BRIP1 is known to play a critical role in DNA repair, with knockdown leading to chromosomal instability.²⁶ Of note in our patient, both the benign and the malignant tumors exhibited a higher-than-typical somatic mutation rate compared with most pediatric malignancies²⁷ as well as a high number of structural variants with a preponderance of inversion events, which suggests a functional role of this germline finding. We have speculated that these changes may have accumulated because of BRIP1 partial dysfunction that led to chromosomal instability and impaired DNA crosslink repair.²⁶ Admittedly, definitive conclusions cannot be drawn on the basis of a single patient, and additional investigation seems warranted.

Although osteoblastoma infrequently has been associated with mutations involving RB, TP53, Fos, FosB, d-jun, and MDM2, the genetic landscape of this tumor is still largely unknown.^28-31 The osteoblastoma in this report is similar to previous reports in that we observed many structural arrangements and some degree of aneuploidy but no clearly oncogenic or transformative point mutations. Many osteosarcomas demonstrate chromosomal abnormalities that commonly involve tumor suppressor genes or DNA helicases.^32,33 However, osteosarcoma generally is recognized as exhibiting tremendous genetic complexity with no clearly identifiable genetic driver. In some cases, this heterogeneity may stem from chromothripsis,^34-36 markers of which have been described in 2% to 3% of all cancers and up to 33% of osteosarcomas. Although the current case did not exhibit evidence of chromothripsis, it demonstrates an otherwise typical genomic profile for osteosarcoma with a high degree of chromosomal instability, increased ploidy, and truncating point mutations in several tumor suppressor genes.

In conclusion, although malignant transformation of osteoblastoma to osteosarcoma historically has been accepted, review of the literature has revealed a paucity of convincing evidence, and controversy has persisted. To our knowledge, this study is the first to report an in-depth genetic characterization of two distinct tumors within the same patient and the same anatomic location. Findings from this study argue against malignant transformation as an evolving or stepwise process and conversely supports two distinct neoplasms with dissimilar genetic makeups. This report also has uncovered a germline truncating mutation in BRIP1 in this patient, which raises the question of whether this serves as an underlying predisposition for both tumors. Additional work toward a better understanding of the role of BRIP1 seems warranted.

AUTHOR CONTRIBUTIONS

Conception and design: David S. Geller, Nicole L. Levine, Bang H. Hoang, Rui Yang, Daniel Weiser, Jonathan Morris, Richard Gorlick, Jonathan B. Gill, Janet Tingling, Andrew S. Brohl

Financial support: David S. Geller, Jonathan Morris

Administrative support: Bang H. Hoang, Janet Tingling

Collection and assembly of data: David S. Geller, Nicole L. Levine, Jonathan Morris, Andrew S. Brohl

Data analysis and interpretation: David S. Geller, Nicole L. Levine, Bang H. Hoang, Rui Yang, Daniel Weiser, Jonathan Morris, Richard Gorlick, Jonathan B. Gill, Michael E. Roth, Andrew S. Brohl

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

David S. Geller

Honoraria: Hangar

Patents, Royalties, Other Intellectual Property: Royalty agreement unrealized at this time

Richard Gorlick

Research Funding: Eisai (Inst)

Jonathan B. Gill

Consulting or Advisory Role: Celgene (I), Puma Biotechnology (I), Legend Biotech (I)

Michael E. Roth

Research Funding: Eisai

Andrew S. Brohl

Consulting or Advisory Role: Bayer AG

Expert Testimony: GlaxoSmithKline (I)

No other potential conflicts of interest were reported.