Expanding the Molecular Genetic Spectrum of Bone and Soft Tissue Fibrosarcomas: An Institutional Experience

Introduction

Since their initial description, and until relatively recently, fibrosarcomas were considered the most common soft tissue sarcoma of adults comprising upwards of 65% of sarcoma diagnoses. ¹ Over the years, as our understanding of sarcomas has evolved and newer technologies have allowed us to more readily identify defining characteristics of discrete histopathologic entities, fibrosarcomas have become a rarity among a growing number of diagnoses and are now considered a diagnosis of exclusion.^1,2 To this point, several large studies investigating the incidence of fibrosarcomas have been conducted by Pritchard et al first in 1974 and again in 1989, in which they were able to reclassify 33% and 72% of previously diagnosed fibrosarcomas. These studies were then followed by the largest modern study to date by Bahrami et al ¹ in 2010, in which they were able to reclassify 84% of previously diagnosed fibrosarcomas and arrived at the conclusion that <1% or around 10,000 sarcoma cases diagnosed at their institution during the study time interval were fibrosarcomas. ² In the evaluation performed of fibrosarcomas in the aforementioned studies, the investigators used such tools as immunohistochemistry and targeted fluorescence in-situ hybridization (FISH) to reclassify the diagnosis. However, large-scale studies of fibrosarcoma looking for novel genetic fusions are lacking. In 2019, Yamazaki et al ³ described 2 cases of fibrosarcoma, 1 arising in soft tissue and 1 arising in bone, in which they described a novel neurotrophic receptor tyrosine kinase (NTRK)3 fusion in both tumors by ribonucleic acid (RNA) sequencing. Herein, and prompted by this index report, we investigate bone and soft tissue fibrosarcomas through the use of next-generation sequencing (NGS) RNA fusion analysis and DNA/RNA sequencing, for the presence of novel gene fusion events, including NTRK-related fusions, in an attempt to expand the genetic spectrum of fibrosarcomas and identify therapeutically targetable cases.

Materials and Methods

This study was deemed exempt by the Institutional Review Board of the University of Pittsburgh (MOD20080228-002). Institutional archives were searched for cases diagnosed as “fibrosarcoma” involving bone and soft tissue from 2000 to present. Strictly defined morphologic cirteria for “fibrosarcoma” including the classic cellular monomorphic fascicular proliferation of spindled cell with low to moderate degree of atypia or pleomorphism and nonspecific immunoprofile with absent or limted smooth muscle actin (SMA) expression was used as the main inclusion criteria. Morphologic mimics such as malignant peripheral nerve sheath tumor (MPNST), solitary fibrous tumor, synovial sarcoma, biphenotypic sinonasal sarcoma, leiomyosarcoma, Kaposi sarcoma, and sarcomatoid carcinoma were all excluded on the basis of negative immunoexpression for S100, CD34, epithelial membrane antigen (EMA), cytokeratin, desmin, and erythroblast transformation-specific [ETS]-related gene (ERG). Thirty cases were identified, 9 were excluded (2 metastatic disease, 3 recurrent disease, 1 re-excision of residual disease, 1 status postirradiation, and 2 did not meet the strict morphologic features of “fibrosarcoma” and had instead high degree of pleomorphism with high grade features). Of the remaining 21 cases, 10 cases had formalin-fixed paraffin-embedded tissue (FFPE) blocks available for molecular testing. One case, at the request of the submitting clinician, was sent for DNA/RNA sequencing at an outside facility while the 9 remaining cases underwent in-house molecular testing. Additionally, 1 case (case 1) was already subjected to FISH testing for the SS18 subunit of BAF chromatin remodeling complex (SYT) gene rearrangement and was negative.

RNA was extracted by the Clinical Laboratory Improvement Amendments (CLIA) certified College of American Pathologists (CAP) accredited molecular and genomic pathology laboratory at the University of Pittsburgh Medical Center (UPMC). Surgical FFPE samples were manually microdissected under hematoxylin and eosin (H&E) guidance, and total nucleic acid was isolated using the DNeasy Blood and Tissue Kit (Qiagen) using the QIAcube instrument (Qiagen). RNA was quantified using the Glomax Discover fluorometer (Promega).

Targeted NGS was performed using the RNA primers of the Oncomine Comprehensive Assay v3 (ThermoFisher Scientific) according to the manufacturer's protocol. Briefly, 50 ng of mRNA was reverse transcribed to cDNA. Multiplex polymerase chain reaction was performed to amplify regions of interest. Amplicons were barcoded, ligated with sequencing adapters, and purified. NGS libraries were quantitated using the TapeStation 4200 (Agilent Technologies), normalized, and pooled for sequencing. Template preparation was performed on the Ion Chef instrument (ThermoFisher Scientific), and massively parallel NGS was performed on the Ion S5 GeneStudio instrument (ThermoFisher Scientific). Data were analyzed for gene fusions (gene fusion database in Supplemental Material) using Torrent Suite v5.12 (ThermoFisher Scientific) and an in-house developed bioinformatics program (Variant Explorer, UPMC). More than 50 fusion-spanning reads were required to call a sample fusion-positive. The limit of detection of the assay is 1% to 5% tumor cells.

Results

Clinicopathologic results, including the presence/absence of gene fusions, are summarized in Table 1. At the time of diagnosis the mean age was 54.5 (range 14-88) with a male to female ratio of 1.5:1. Locations included soft tissue of the lower extremity (3), trunk (2), pelvis (2), head (1), upper extremity (1), and bone (1). The morphology of all 10 cases fit the morphologic spectrum of classic fibrosarcoma and harbored cellular monomorphic fascicular proliferations of spindled cells with low to moderate degree of atypia or pleomorphism. Tumor grade included low grade (5) and intermediate grade (4). Two cases were regraded based on the FNCLCC grading system and were changed from high grde to intermediate grade fibrosarcoma. Of the 10 cases, 1 (case 2) demonstrated an FNDC3B-PIK3CA gene fusion (Figure 1A) and 1 (case 10) demonstrated a BRAF (p.G469A) mutation and loss of CDKN2A/B. Case 2 (FNDC3B-PIK3CA fusion) was of an 88-year-old female with a history of breast cancer who presented with a 5.5 cm mass of the right forearm that was composed of intersecting fascicles of variably atypical spindle cells with tapered nuclei and small nucleoli, eosinophilic cytoplasm and set in a collagenous stroma (Figure 2A and B) with only rare weak nonspecific CD34 immunostaining that was regarded as essentially negative and was diagnosed as an intermediate grade fibrosarcoma. Case 10 (BRAF p.G469A mutation and loss of CDKN2A/B) was of a 63-year-old female with no significant past medical history who presented with a 4.5 cm mass of the right lower extremity and tumor was also comprised of short fascicles of atypical spindle cells with tapered nuclei and small nucleoli with amphophilic cytoplasm within a collagenous stroma (Figure 3A); however, it did have a prominent vascular pattern with scattered hyalinized ectatic vessels (Figure 3B) and was diagnosed at the time as an intermediate grade fibrosarcoma. The remaining cases with no detectable gene fusions were of similar morphology and had no substantial morphologic features that set them apart from the 2 cases with molecular fusion/aberration. All cases were negative for S100, cytokeratin, EMA, CD34, desmin, and SMA (negative in most cases with 2 cases showing only rare focal staining) immunostains thus excluding, along with the confirmatory absence of the SYT gene fusion via NGS testing, all morphologic mimics of fibrosarcoma including MPNST, solitary fibrous tumor, synovial sarcoma, biphenotypic sinonasal sarcoma, leiomyosarcoma, Kaposi sarcoma, and sarcomatoid carcinoma.

OPEN IN VIEWER

Figure 1.

(A) Integrative genomics viewer (IGV) pileup of sequence reads of exon 3 of FNDC3B fuzed to exon 2 of PIK3CA. The blue bars at the bottom represent the exons of each fusion partner, with the breakpoint shown at the 2 bp junction in the middle. The reference sequence is displayed just above the blue bars. Sequence is 5′ to 3′ direction.

OPEN IN VIEWER

Figure 2.

(A, B) Histologic sections of case 2 showing intersecting fascicles of variably atypical spindle cells with tapered nuclei, small nucleoli, and eosinophilic cytoplasm within a collagenous stroma (H&E, 4× and 10×).

OPEN IN VIEWER

Figure 3.

(A) Histologic sections of case 10 demonstrating short fascicles of atypical spindle cells with amphophilic cytoplasm in a collagenous stroma (H&E, 20×) with (B) prominent vascular pattern (H&E, 10×).

OPEN IN VIEWER

Table 1.

Clinicopathologic Results.

Case	Age	Gender	Site	Diagnosis	Molecular testing	Genetic alteration
1	68	M	Right posterior thigh	Low-grade fibrosarcoma	NGS RNA fusion analysis	Fusion-negative
2	88	F	Right forearm	Intermediate-grade fibrosarcoma	NGS RNA fusion analysis	FNDC3B-PIK3CA fusion
3	67	F	Right knee	Low-grade fibrosarcoma	NGS RNA fusion analysis	Fusion-negative
4	66	M	Penis	Low-grade fibrosarcoma	NGS RNA fusion analysis	Fusion-negative
5	48	M	Paraspinal muscle/L2-S1	Intermediate-grade fibrosarcoma	NGS RNA fusion analysis	Fusion-negative
6	14	M	Anterior mandible	Fibrosarcoma of bone	NGS RNA fusion analysis	Fusion-negative
7	58	F	Right rectus muscle	Low-grade fibrosarcoma	NGS RNA fusion analysis	Fusion-negative
8	43	M	Left middle turbinate	Low-grade fibrosarcoma	NGS RNA fusion analysis	Fusion-negative
9	30	M	Right pelvis	Intermediate-grade fibrosarcoma	NGS RNA fusion analysis	Fusion-negative
10	63	F	Right lower extremity	Intermediate-grade fibrosarcoma	DNA/RNA sequencing (FoundationOne®)	BRAF (p.G469A) mutation CDKN2A/B loss Fusion-negative

Abbreviations: NGS, next-generation sequencing; RNA, ribonucleic acid.

Discussion

The incidence of fibrosarcomas has undergone a drastic decline in recent years, accounting for only 3.6% of sarcoma diagnoses according to data obtained from the Surveillance, Epidemiology and End Results for the years spanning 1978 to 2001.^1,2 This decline in incidence is likely due to a number of different factors as outlined by Bahrami et al. ¹ With ever-advancing knowledge and technological improvements, the diagnosis of fibrosarcoma has become one of exclusion necessitating ruling out a long list of potential histopathologic mimickers, and it has been recognized as such within the fifth edition of the World Health Organization (WHO) soft tissue and bone tumors which defines these tumors as rare monomorphic sarcomas of fibroblastic tumor cells of variable collagen production with no specific immunohistochemical profile or molecular alterations. ⁴ As such, as these advancements continue to be applied to those cases that would otherwise be diagnosed as fibrosarcoma, novel genetic findings will continue to alter our understanding of this entity.

In recent years, there has been growing interest in identifying and characterizing tumors with novel genetic fusions, particularly those with available therapeutic options, including NTRK gene fusion-related tumors. Tumors with fusions involving the NTRK gene are being identified at a rapid rate within disparate organ systems from glioblastomas to colorectal carcinomas and soft tissue sarcomas, among many others. ⁵ NTRK stands for neurotrophic receptor tyrosine kinase and is composed of 3 genes, NTRK1 to 3, which in turn encode 3 tropomyosin receptor kinases, TrkA-C. These kinases are normally expressed in neuronal tissue and are essential in the development and survival of the cells of the central and peripheral nervous systems. ⁵ Fusions of NTRK lead to oncogenic potential through overexpression and constitutive activation of the chimeric kinase, and as is the case in other tumors with kinase activity, there are specific inhibitors against the ATP binding site of the TRK protein which have shown clinical utility and are currently approved for therapeutic use. ⁵ As NTRK rearranged soft tissue tumors are being characterized, it has become apparent that this group spans a spectrum of histomorphologies from lipofibromatosis like neural tumor to those resembling MPNST. However, in the midst of these extremes are those tumors that are composed of monomorphic spindle cells in a fascicular to haphazard growth pattern with hyalinized stromal collagen, which resemble fibrosarcoma morphologically but have variable expression of CD34 and S100. Although seemingly nonspecific, the distinction and correct identification of such cases and their separation from fibrosarcoma is paramount due to the availability of specific targeted therapy. In recognition of these cases, the fifth edition of the WHO soft tissue and bone tumors has included NTRK-rearranged spindle cell neoplasm as an emerging entity. ⁶

In 2019, Yamazaki et al described 2 cases, 1 arising in bone and 1 arising in soft tissue, which were composed of monomorphic spindle cells with multiple growth patterns that otherwise fit morphologically with fibrosarcoma. However, both cases demonstrated widespread expression of CD34 with focal expression of SMA, and given the emergence of the aforementioned entities of NTRK-rearranged spindle cell neoplasms, immunohistochemistry for Pan-TRK was performed and was positive with subsequent RNA sequencing demonstrating NTRK3 fusions in both cases. This index report of novel NTRK rearrangements in what otherwise would be diagnosed as fibrosarcoma adds to the growing list of factors leading to the decline in incidence of this diagnosis. However, large studies investigating the prevalence of fusion events in fibrosarcomas are lacking.

Interestingly, neoplasms of the female gynecologic tract resembling soft tissue and bone fibrosarcoma have also been recently described to harbor genetic fusion events. In 2018, Chiang et al ⁷ described 3 cervical and 1 uterine corpus spindle cell sarcomas, with focal SMA and S100 immunopositivity in all 4 cases, that demonstrated either NTRK1 or NTRK3 rearrangements (TPM3-NTRK1, TPR-NTRK1, LMNA-NTRK1, and RBPMS-NTRK3). Similarly, in 2019 Croce et al ⁸ described an additional 7 cases of spindle cell sarcoma involving the cervix with CD34 and S100 coexpression that harbored either NTRK1 or NTRK3 fusions (TPM3-NTRK1 and EML4/NTRK3). In addition to NTRK fusion events, they described 3 cases of spindle cell sarcoma, 2 cervical and 1 uterine corpus, which were immunopositive for CD34 only and harbored COL1A1-PDGFB fusions similar to that seen in dermatofibrosarcoma protuberans.

In our study, 2 of 10 cases were found to harbor genetic alterations by NGS RNA fusion analysis and DNA/RNA sequencing. BRAF mutations (predominantly BRAF p.V600E) are known to occur in sarcomas with an ∼ mutation rate of 9% and include such entities as MPNST, gastrointestinal stromal tumor, Ewing sarcoma, among others. ⁹ In addition, a recent study demonstrated various BRAF gene fusions and point mutations (including p.L485F, p.V600D, and p.V600E) in spindle cell sarcomas resembling infantile fibrosarcoma. ¹⁰ However, the specific BRAF mutation identified in our study, BRAF p.G469A, has been reported most often in lung adenocarcinoma, and its prevalence in sarcoma is not well characterized. ¹¹ Similarly, CDKN2A/B aberrations are known to occur in sarcomas with a mutation rate of ∼16%, with the majority comprising copy number loss, and occur in such entities as MPNST, myxofibrosarcoma, undifferentiated pleomorphic sarcoma, fibrosarcoma, among others. ¹² To the best of our knowledge, the FNDC3B-PIK3CA fusion has recently been described as a novel fusion in a single case of uterine corpus endometrial carcinoma, but its prevalence in other tumor types is unknown. ¹³ Both cases that harbored genetic alterations, when compared together, had no significant morphologic or immunohistochemical differences with both of them displaying monomorphic cellular fascicular spindle cell proliferation and nonspecific immunoprofile, with the exception of the striking reactive glomeruloid vascular pattern noted in the BRAF p.G469A mutated case, the significance of which is uncertain at this point in time. Additionally, the remaining molecularly fusion-negative cases had no substantial morphologic features that would set them apart from the 2 molecularly positive cases.

In conclusion, our study demonstrated gene fusions (FNDC3B-PIK3CA) in 1 (10%) of 10 fibrosarcomas diagnosed at our institution in the past 20 years. Additionally, 1 case harbored BRAF (p.G469A) mutation and CDKN2A/B loss with no evidence of gene fusion. NTRK rearrangements were not detected in our cohort of fibrosarcomas. Although additional studies with a larger number of cases of fibrosarcoma would be required to determine the true nature of NTRK rearrangements within these tumors, our results would seem to indicate that it is a relatively rare occurrence. However, given that there is targeted therapy available for NTRK rearranged neoplasms, it is reasonable to consider screening such cases so that, if present, these patients are eligible to receive targeted treatment. Furthermore, the significance of the molecular aberrations detected in our cohort is not clear at this time, and future studies are needed to establish whether these findings carry any clinicopathologic significance.

Supplemental Material