Journal of Neuro-Oncology

, 105:317

Sarcoma arising as a distinct nodule within glioblastoma: a morphological and molecular perspective on gliosarcoma

Authors

  • Cynthia Jimenez
    • Department of Pathology, Neuropathology UnitUniversity of California at San Francisco
  • Martin Powers
    • Department of Pathology, Neuropathology UnitUniversity of California at San Francisco
  • Andrew T. Parsa
    • Department of Neurological SurgeryUniversity of California San Francisco
  • Christine Glastonbury
    • Department of RadiologyUniversity of California at San Francisco
  • Jill M. Hagenkord
    • Department of Pathology, iKaryos DiagnosticsCreighton University Medical Laboratories
    • Department of Pathology, Neuropathology UnitUniversity of California at San Francisco
Clinical Study – Patient Study

DOI: 10.1007/s11060-011-0593-6

Cite this article as:
Jimenez, C., Powers, M., Parsa, A.T. et al. J Neurooncol (2011) 105: 317. doi:10.1007/s11060-011-0593-6

Abstract

Gliosarcoma is a variant of glioblastoma and is characterized by distinct glial and sarcomatous components. Typically, there is no macroscopic boundary between the components and special stains are often required to distinguish the glial and sarcomatous elements. Some studies suggest similar genetic alterations in both components pointing to a common origin. We present an extreme case of gliosarcoma arising as a discrete fibrous nodule adjacent to a typical glioblastoma. A 65 year-old woman presented with progressive weakness, seizures and right-sided hemiparesis. CT scan demonstrated an irregular enhancing left frontal lobe mass and an adjacent discrete nodule with different imaging characteristics. The unique nature of this macroscopically biphasic neoplasm allowed us to compare the molecular characteristics of glial and sarcomatous elements which were strikingly similar except for small losses and gains in Chr 3. Studies are under way to determine the significance of chromosome 3 alterations in gliosarcomas.

Keywords

GliosarcomaSarcomatous gliomaGenotypeSNP

Introduction

Gliosarcoma was initially described by Stroebe in 1895 as a rare central nervous system tumor having both malignant gliomatous and sarcomatous components. Revisited by Feigin almost 60 years later, it was described as a tumor composed of two dissimilar neoplastic tissues, one resembling glioblastoma multiforme (GBM) and the other resembling fibrosarcoma [1]. Currently, the World Health Organization (WHO) defines gliosarcoma as variant of glioblastoma showing biphasic tissue pattern with alternating areas displaying glial and mesenchymal differentiation [2]. One missing piece of information in this definition is the minimum percentage of mesenchymal component required to consider a tumor gliosarcoma.

Gliosarcomas represent a small percentage of malignant gliomas and the mesenchymal component can include cartilaginous, osseous, myoblastic and lipomatous elements [3]. Controversy remains as to whether gliosarcoma has a different biological behavior or prognosis than glioblastoma or a distinct entity altogether [4]. We report an extremely unique presentation of a gliosarcoma as a radiologically and pathologically discrete, well-circumscribed sarcomatous nodule adjacent to a classical GBM.

Case report

A 65 year-old woman was evaluated at our institution for a 1 month history of right arm weakness and slurred speech. During this period, her weakness progressed rapidly until she was unable to move her entire right upper extremity. She also had difficulty in standing from a sitting position, and suffered a few episodes of spells suggestive of simple partial seizures. Neurological examination on admission confirmed right sided hemiparesis without cranial nerve abnormalities. The remainder of her physical examination was unremarkable. She had a history of hypertension and hyperlipidemia but no history of cancer or intracranial radiation therapy. The family history was significant for a first degree relative with breast cancer. Her past surgical history included cholecystectomy for cholelithiasis and hysterectomy for leiomyomata.

The initial radioimaging studies a month earlier identified a left frontal lobe intra-axial mass, which increased in size to a maximum dimension of 3.3 cm with peripheral enhancement (see “Radiological findings” section). The patient was admitted and underwent a left frontal craniotomy. Intraoperatively, the tumor exhibited irregular and indistinct margins and marked hemorrhage, typical of a malignant glioma. Frozen section from this tissue confirmed glioblastoma. During further exploration, an entirely well-circumscribed, pearly-white nodule was encountered in the most infero-posterior aspect of the tumor (Fig. 2). This mass was clearly distinct from but adjacent to the glioblastoma. It was not attached to dura and could be easily shelled out from the surrounding brain parenchyma. Frozen section from this mass was interpreted as a low grade spindle cell neoplasm distinct from the glioblastoma. A near gross total resection was also accomplished in the malignant glioma component with no visible tumor intraoperatively. Postoperatively, all cranial nerves were intact, sensory examination was normal, and she regained strength on the left side. One month after tumor resection her symptoms had improved dramatically. Residual symptoms included mild left hemiparesis, decreased coordination and fine motor movement in her right hand as well as mild right-sided spasticity. She began conformal radiation and adjuvant temozolomide therapy. Follow-up radiologic evaluation 5 months after initial surgery showed no evidence of tumor progression. At the time of this report, she was receiving temozolomide on a 5 day schedule with bi-monthly radiologic evaluations.

Radiologic findings

Magnetic resonance (MR) imaging was performed on a 3-Tesla magnet demonstrating significant interval growth in a right posterior frontal mass as compared with an outside hospital MR examination from 1 month prior to admission. The ill-defined tumor now had a maximal diameter of 3.3 cm, expanded the left precentral gyrus and was associated with vasogenic edema but no midline shift (Fig. 1). The tumor had both cystic and solid regions, with the solid areas showing marked enhancement following gadolinium administration. At the infero-posterior aspect of this heterogeneous mass, a round, well-defined intensely enhancing solid nodule was evident. Dynamic contrast-enhanced T2*-weighted perfusion imaging demonstrated markedly elevated cerebral blood volume in the solid components of the main mass, and in the secondary round nodule as compared to normal white matter. The perfusion characteristics were somewhat different between the irregular, infiltrative neoplasm and the solid nodule. MR spectroscopy showed reduction in N-acetylaspartate (NAA) and elevated choline typically found in high grade primary cerebral neoplasms. The difference between the solid and the infiltrative components could not be interpreted with certainty due to overlapping voxels.
https://static-content.springer.com/image/art%3A10.1007%2Fs11060-011-0593-6/MediaObjects/11060_2011_593_Fig1_HTML.jpg
Fig. 1

a Sagittal FLAIR image demonstrates heterogeneous mass (arrow) expanding posterior frontal lobe with associated vasogenic edema. At posterior inferior margin of mass is a solid, well-defined round nodule (open arrow). b Axial post-gadolinium T1-weighted image shows solid, predominantly peripheral enhancement of the glioblastoma component with ill-defined margins, typical of a high-grade glial neoplasm. The round nodule (open arrow) appears distinctly better defined and more markedly enhancing. All enhancing areas showed markedly elevated cerebral blood volume on MR perfusion interrogation

Pathologic findings

Two separate gross specimens were received in pathology and had markedly distinct macroscopic appearances. The first specimen consisted of multiple, irregular and hemorrhagic fragments of ill-defined brain tissue. In contrast, the second specimen consisted of a discrete, spherical, glistening pearly-white, rubbery nodule. It measured 1.5 cm × 1.5 cm × 1.1 cm (Fig. 2). Microscopical examination also revealed two distinctive morphologic appearances between the two macroscopically distinct tumor components. The glial component was composed of highly cellular infiltrative pleomorphic astrocytic cells with conspicuous eosinophilic cytoplasm and marked nuclear atypia. The tumor cells exhibited brisk mitotic figures and classical vascular endothelial proliferation and palisading necrosis, typical of a glioblastoma (Fig. 3a, c). The discrete nodule showed a relatively less cellular spindle cell proliferation. The tumor cells in the nodule were less pleomorphic and had features of mesenchymal cells with uniform pale cytoplasm and vesicular, ovoid to cigar-shaped nuclei within a dense collagenous and focally myxoid stroma similar to a fibrosarcoma (Fig. 3b, d). There were no clearly identifiable glial elements or trapped neuropil in the sarcomatous nodule. Scattered mitotic figures were also present in this component, but vascular endothelial proliferation or necrosis of any type was distinctly absent.
https://static-content.springer.com/image/art%3A10.1007%2Fs11060-011-0593-6/MediaObjects/11060_2011_593_Fig2_HTML.jpg
Fig. 2

Macroscopic appearance of the sarcomatous nodule. The largest dimension of the nodule was 1.5 cm and the appearance was reminiscent of a benign fibrous neoplasm. It was easily shelled out in surgery

https://static-content.springer.com/image/art%3A10.1007%2Fs11060-011-0593-6/MediaObjects/11060_2011_593_Fig3_HTML.jpg
Fig. 3

Histological features of glial and sarcomatous components of the tumor. a Low grade magnification of the glial component with palisaded necrosis, typical of glioblastoma (H&E original magnification ×100). b Low grade magnification of the sarcomatous nodule demonstrating a spindle cell tumor with a vague myxoid background. (H&E original magnification ×100) c Medium power magnification of the glial component with abundant vascular endothelial proliferation (H&E original magnification ×200). d Medium power magnification of the sarcomatous nodule with a highly fascicular area, distinct from the glial component. (H&E original magnification ×200). e Immunohistochemical staining for GFAP in the glial component with diffuse strong positivity (original magnification ×200). f Immunohistochemical staining for GFAP in the sarcomatous nodule demonstrating focal positivity(original magnification ×200). g Immunohistochemical staining for EGFR showing negative staining in the glial component (original magnification ×200). h Immunohistochemical staining for EGFR showing strong diffuse positive staining in the sarcomatous nodule (original magnification ×200)

Immunohistochemical staining for the malignant glioma component was positive for glial fibrillary acidic protein-GFAP (DAKO, 1:3000; Fig. 3e), oligodendrocyte lineage transcription factor 2-Olig 2, (IBL Labs, 1:250), S-100 protein (DAKO, 1:2000), p53 protein (D0-7; Novocastra, 1:300), and was negative for epidermal growth factor receptor (EGFR, 528; Oncogene, 1:100. Fig. 3g) Collagen Type IV (DAKO, 1:50), Neu-N antibody (Chemicon, 1:2000), CD34 (QB-END/10; Novocastra, 1:400), Caldesmon (h-CD; DAKO, 1:200), Desmin (Cell Marquee, undiluted), Smooth Muscle Actin (HHF35; ENZO, 1:2), cytokeratins AE1/AE3 (DAKO, 1:100) and Cam 5.2 (Becton–Dickinson; 1:200). Neurofilament protein (Sigma; NN18, 1:20.000) demonstrated numerous positive axonal process within the glioma component but was virtually negative in the sarcomatous nodule. The sarcomatous nodule was only focally positive for GFAP (Fig. 3f), p53 protein and S-100. Unlike the glial component, this component was also strongly positive for EGFR (Fig. 3h) and Collagen type IV. The sarcomatous component was negative for Neu-N, Caldesmon, Desmin, cytokeratins, Olig 2, SMA, and CD34. Staining for EGFRvIII was negative in both components and less than 25% of cells showed nuclear PTEN immunopositivity in both components. The sarcomatous nodule was physically distinct from the surrounding neural tissue and did not contain any trapped islands from the malignant glioma component or reactive brain.

In addition we have also performed a 250 K Affymetrix NspI single nucleotide polymorphism (SNP) array on formalin fixed tissue of the sarcomatous and the glial component of this tumor since it was quite easy to segregate two components. Molecular analysis of both components were performed at the iKaryos laboratories in Omaha, NE, using methods previously described [5]. Tissue samples from both the sarcomatous and glioblastoma area demonstrated nearly identical profiles on SNP arrays (Table 1). Briefly, both components demonstrated loss of Chr 9p, 10 and 11q, three copies of Chr 7 and gains of Chr 19 and 20. The only significant difference between the two components was the presence of short gains and losses near the telomeric end of Chr 3q seen only in the sarcomatous component. Analysis of EGFR and PTEN genes using fluorescence in situ hybridization (FISH) with commercially available probes demonstrated deletion of the PTEN gene in both components. FISH analysis was compatible with the gain of the entire chromosome 7 in both the glial component (Average number of EGFR LSI signals: 2.6; Average number of Centromere 7 signals-CEP 7: 2.92) and the sarcomatous nodule (Average number of EGFR LSI signals: 2.42; Average number of Centromere 7 signals-CEP 7: 2.42). No EGFR amplification was seen in either component.
Table 1

Genomic aberrations in glial and sarcomatous components of the gliosarcoma using SNP array

Component

Change

Chr

Start cytoband

End cytoband

Size Mb

Sarcomatous

Loss

3

q26.1

q26.1

6.311577

Sarcomatous

Loss

3

q26.31

q26.31

1.128813

Sarcomatous

Gain

3

q26.31

q26.33

5.77731

Sarcomatous

Loss

3

q26.33

q27.1

2.48717

Sarcomatous

Gain

3

q27.1

q27.3

4.14246

Sarcomatous

Loss

3

q27.3

q28

1.491838

Sarcomatous

Gain

3

q28

q28

0.936313

Sarcomatous

Gain

3

q28

q28

1.870645

Sarcomatous

Gain

7

p22.3

q36.3

158.6565

Glial

Gain

Sarcomatous

Loss

8

p23.3

p23.2

3.27502

Glial

Loss

Sarcomatous

Loss

9

p24.3

p13.2

36.26289

Glial

Loss

Sarcomatous

Loss

9

p21.3

p21.3

0.273208

Glial

Loss

Sarcomatous

Loss

9

q33.1

q33.2

1.468388

Glial

Loss

Sarcomatous

Loss

10

p15.3

q26.3

135.1078

Glial

Loss

Sarcomatous

Loss

11

p15.5

p15.4

3.202621

Glial

Loss

Sarcomatous

Loss

11

q12.3

q25

72.51496

Glial

Loss

Sarcomatous

Loss

17

p11.2

q12

8.095219

Glial

Loss

Sarcomatous

Gain

19

p13.3

q13.43

63.48835

Glial

Gain

Sarcomatous

Gain

20

p13

q13.33

62.35758

Glial

Gain

Discussion

Gliosarcoma is defined as a distinct variant of glioblastoma according to the current WHO classification [2]. The histological type and percentage of the sarcomatous are variable and there is no clear consensus as to how much of a mesenchymal component would qualify the neoplasm as a gliosarcoma [6]. As a rule, there is no macroscopic boundary between the two components of a gliosarcoma and the distinction requires microscopic examination. We have not been able to identify macroscopically distinct biphasic gliosarcomas in the literature.

The emergence of gliosarcomas is also a matter of controversy. The monoclonal origin hypothesis postulates that the sarcomatous component develops from neoplastic glial cells that acquired sarcomatous properties; thus, both the glial and sarcomatous cells arise from a common precursor [7, 8]. This hypothesis is supported by molecular and genetic studies showing that in some gliosarcomas both the glial and sarcomatous components share common genetic alterations [9, 10]. Identical TP53 mutations, PTEN deletions and CDKN2A deletions have been demonstrated to exist in both glial and sarcomatous components in 20–40% of such tumors [10]. In contrast, the polyclonal hypothesis proposes that the glial and sarcomatous components are thought to develop independently [15]. The origin of the sarcomatous component was initially thought to be hyperplastic endothelial cells due to stimulation by malignant astrocytes [1]. Some speculate that the origin of the sarcomatous component can be fibroblasts, cells of fibrohistiocytic lineage or primitive pluripotent stem cells. Some authors have used immunohistochemical and ultrastructural studies to suggest that the sarcomatous component is derived from undifferentiated mesenchymal cells associated with the vascular adventitia which dedifferentiate into smooth muscle and pericytic cell types [1113].

Most gliosarcomas are considered primary, while secondary gliosarcomas are those that emerge after irradiation for either glioblastoma or other neoplasms [14]. Gliosarcomas should also be distinguished from radiation-induced sarcomas, but this distinction still remains an academical exercise until more effective treatment options for malignant glial neoplasms are discovered. The critical differential diagnosis of gliosarcoma is often with a metastatic or a primary central nervous system sarcoma. While all of these neoplasms portend dismal prognoses, the distinction of sarcoma from gliosarcoma is important in terms of clinical management and also in terms of frequency and patterns of metastases.

Previous studies on gliosarcomas have often analyzed molecular markers or genes that are typically altered in malignant gliomas such as P53, PTEN and EGFR [9, 10, 15]. These earlier studies faced the challenge of reliably segregating two components for genetic analysis. Nevertheless, all such studies have found similar alterations and conclude based on these genetic commonalities, a single cell of origin for both components. Similarly in our case, we observed a remarkable extent of similarities in the genetic aberrations in numerous loci between two components. Some of these abnormalities are similar to those described earlier and some are newer observations. Nevertheless, the finding leaves little room for any other explanation than a common cell of origin.

It was equally interesting to identify chromosome 3 aberrations only in the sarcomatous component. Since we can clearly segregate two components of the tumor, this difference was not considered to be artifactual. We are not certain about the significance of alterations in the telomeric end of Chr 3q and currently we do not have any data to suggest that these small genetic alterations can account for the different morphology. Alterations in chromosome 3 were also reported in microdissected sections of gliosarcomas by Boerman et al. [16]. Their study found two of five cases showing gains of 3q on CGH but no imbalance of microsatellite alleles on PCR. While the authors found alterations in chromosome 3 “perhaps the most interesting finding”, they did not further elaborate on the reasons. Similar to this report, we do not have specific data as to why this alteration is or should be significant. Yet, it is possible that the alterations in this chromosome can underscore mesenchymal differentiation in malignant gliomas and may be associated with the sarcoma phenotype. We have listed some of the putative genes located in the altered sections of chromosome 3 that may be considered as targets for further research (see Table 2).
Table 2

Potential relevant genes from chromosome 3 aberrations in sarcomatous component

Gene

Change

Function

Claudin1

Loss

Integral membrane protein important in cell to cell adhesion

SOX2

Loss

Embryonic development and determination of cell fate

PDCD10

Loss

Cell apoptosis and vascular development

TP63

Loss

Development and maintenance of stratified epithelial tissue

FGF12

Gain

Development, cell growth, morphogenesis, tissue repair

MAP3K13

Gain

Activate MAPK8/JNK with role in JNK signaling (putative)

DVL3

Loss

Regulation of cell proliferation (putative)

TBL1XR1

Gain

Regulation of cell differentiation (putative)

Genetic aberrations in Chr3q include a loss at region 3q27.3–3q28 which contains the gene Claudin 1. This gene produces functional tight junction proteins involved in cell-to-cell adhesion and epithelial-to-mesenchymal transition (EMT) in breast cancer [17, 18]. Claudin 1 is a direct downstream target gene of Slug and Snail, transcription factors shown to repress E-cadherin, promote EMT and play an important role in the formation of the mesoderm [1921]. Specifically, Slug and Snail inhibit Claudin-1 expression in epithelial cells [22]. We identified a gain at region 3q27.1–3q27.3 which includes the somatostatin gene and prior studies have shown that high-grade osteosarcomas have increased somatostatin receptors [23, 24]. Loss at region 3q26.33–3q27.1 includes the Sox2 gene, a transcription factor involved in the regulation of embryonic development and the determination of cell fate [25]. Although not an exhaustive list, additional potential relevant genes identified at the location of the Chr 3q aberrations, solely in the sarcomatous component are included in Table 2.

An interesting molecular finding was the absence of difference in EGFR status on SNP array and lack of EGFR amplification on FISH, despite a striking difference in EGFR immunohistochemistry. The sarcomatous component was strongly positive for EGFR while the glial component was negative, both components being negative for EGFRvIII. Yet, neither of the tumor components demonstrated EGFR amplification and both had three copies of Chr 7. These findings highlight the complexities of the EGFR pathway and the challenges associated with the interpretation of immunohistochemical staining patterns. It is possible that the sarcomatous component may have altered EGFR function such as a novel activation mutation compared to the glial component. Yet, we cannot account for the differences in EGFR immunohistochemistry in both components. Even though we have repeated the stain a number of times, it is possible to include a false positive immunohistochemistry as the reason for this discrepancy. Such a discrepancy in EGFR immunohistochemistry can complicate treatment choices for glioblastoma patients and a clear understanding of this pathway is critical due to the number of treatment options targeting EGFR [2, 10].

While the conclusions from a single case should be drawn with caution, they confirm the current WHO approach that gliosarcomas need to be considered within the spectrum of glioblastomas. The diversity of phenotypical patterns, including the macroscopically distinct glial and sarcomatous components should be within the possible presentation patterns of this highly genetically unstable tumor. However, since the sarcomatous component can have different features than the gliomatous component, the approach to treating such tumors could be different from typical glioblastomas. Our recent studies in primary and secondary gliosarcomas have also highlighted the importance of identifying gliosarcomas particularly with reference to the radiological characteristics that can influence their outcome [26, 27].

Gliosarcomas and other variants of glioblastomas have numerous similarities clinically and radiologically and their survival characteristics are also indistinguishable [4]. Nevertheless, similar to other complex malignancies that contain de-differentiated components, these tumors may have different responses to targeted treatment modalities. With the increasing application of targeted chemotherapy, it will be necessary to distinguish genetic components of these complex tumors to better understand their biological characteristics and responses to emerging therapeutic modalities.

This case report of a unique and extreme example of sarcomatous nodule in a glioblastoma further supports the suggestion that both component arise from a common precursor and is not consistent with the suggestion that pericytic or fibroblastic cells may undergo neoplastic transformation under the influence of malignant astrocytic cells. In cases such as ours, it is easier to reach the conclusion of gliosarcoma, but it is still critical to establish clinically meaningful percentage of the sarcomatous/mesenchymal component to qualify a tumor for gliosarcoma in most cases.

Acknowledgment

We are deeply grateful to Dr. Shera Kash, PhD from the Molecular Pathology and Clinical Genomics Division of the Department of Pathology and Creighton Medical Labs for her expert contribution and the analysis of the 250 K Affymetrix NspI single nucleotide polymorphism (SNP) arrays.

Copyright information

© Springer Science+Business Media, LLC. 2011