Meningeal Hemangiopericytomas and Meningomas: a Comparative Immunohistochemical and Genetic Study

Background: The meningeal hemangiopericytoma (MHPC) is a vascular tumor arising from pericytes. Most intracranial MHPCs resemble meningiomas (MNGs) in their clinical presentation and histological features and may therefore be misdiagnosed, despite important differences in prognosis. Materials and Methods: We report 8 cases of MHPC and 5 cases of MNG collected from 2007 to 2011 from the Neuro-Surgery and Histopathology departments. All 13 samples were re reviewed by two independent pathologists and investigated by immunohistochemistry (IHC) using mesenchymal, epithelial and neuro-glial markers. Additionally, we screened all tumors for a large panel of chromosomal alterations using multiplex ligation probe amplification (MLPA). Presence of the NAB2-STAT6 fusion gene was inferred by immunohistochemical staining for STAT6. Results: Compared with MNG, MHPCs showed strong VIM (100% of cases), CD99 (62%), bcl-2 (87%), and p16 (75%) staining but only focal positivity with EMA (33%) and NSE (37%). The p21 antibody was positive in 62% of MHPC and less than 1% in all MNGs. MLPA data did not distinguish HPC from MNG, with PTEN loss and ERBB2 gain found in both. By contrast, STAT6 nuclear staining was observed in 3 MHPC cases and was absent from MNG. Conclusions: MNG and MHPC comprise a spectrum of tumors that cannot be easily differentiated based on histopathology. The presence of STAT6 nuclear positivity may however be a useful diagnostic marker.


Introduction
Meningeal hemangiopericytoma (MHPC) is a malignant neoplasm which originates from meningeal capillary pericytes (Stout and Murray, 1942). It has been reported to represent 2 to 4% of meningeal tumors, comprising less than 1% of all intracranial tumors (Guthrie et al., 1989). Epidemiological studies reports that MHPC are rare tumors with lower average than meningiomas (MNGs) (Jazayeri et al., 2013) . Most intracranial MHPCs mimic MNGs in their clinical and histopathological presentation, as well as their immunohistochemical profile (Rajaram et al., 2004), and are hence often misdiagnosed.
Meningiomas are neoplasms arising from meningothelial cells of the meninges (Larijani et al., certain diagnosis limitations (Wu et al., 2013).
However, MHPC metastasize outside the CNS in 25%-60% of cases, whereas meningiomas metastasize only occasionally (Rajaram et al., 2004). Given this important prognostic difference, the correct diagnosis and thus the appropriate treatment strategy should be adopted.
The recent WHO classification describes HPC, solitary fibrous tumours (SFT) and MNG as a new biomarkers has prompted us to look for potential differences in IHC. According to S. Barthelme et al. (2014) presence of the NAB2-STAT6 fusion gene distinguishes SFT from MHPC (Barthelmess et al., 2014). This fusion is not well described in MHPCs and has not been reported in the specific comparison between MNG and MHPC. Investigation of additional molecular markers in this differential diagnosis has also been lacking.

Materials and Methods
We investigated a cohort of 8 MHPCs and 5 MNGs from adults. MHPC were provided from 3 women and 5 men; MNG patients were from 3 women and 2 men. Patients' ages ranged from 25 to 81 years. 7 of the 8 MHPC locations were supratentorial: 4 occipital, 2 frontal and 1 temporo-occipital. Only 1 MHPC was located in the cerebellum. MNG tumor location was supra-tentorial in 4 cases and infra-tentorial in 1 case. Clinical symptoms at diagnosis were variable: hemiparesis, disorientation, decreased visual acuity, and headache (Table 1).

Standard histology
Formalin-fixed paraffin-embedded tissue was sectioned at 4 µm and stained with haematoxylin and eosin and reviewed by two pathologists.

Immunohistochemistry
An immunohistochemical study was performed on formalin-fixed, paraffin-embedded tissue cut at 4µm using polyclonal antibodies (Table 2).
Immunohistochemistry for STAT6 was carried out as a surrogate for the NAB2-STAT6 fusion. Antigen retrieval was performed at 98°C with DAKO antigen retrival solution pH 6 for 20 min. Endogenous peroxidase activity  was blocked with 3% hydrogen peroxide in methanol. The detection system used was Novolink Polymer (Leica Microsystems, Newcastle Ltd) with diaminobenzidine as a chromogen. The slides were counterstained with hematoxylin.

DNA extraction
Tumor genomic DNA was extracted from either fresh tissue using a standard phenol:chloroform protocol, or FFPE tissues according to The QIAamp DNA FFPE Tissue kit according to the manufacturers protocol (QIAGEN).

Multiplex ligation probe amplification (MLPA)
MLPA is a multiplex PCR technique in which up to 45 specific sequences are simultaneously quantified in a single reaction. For this multigenic technique, a specifically designed set of probes was used to test for chromosomal abnormalities -SALSA MLPA Kits P105, P370 and P088 (MRC Holland, Amsterdam, The Netherlands).
Fragments were separated by electrophoresis on the ABI Prism 310 capillary genetic analyser. Data analysis was performed with "GeneMarker ® " software (SoftGenetics). Thresholds to detect losses and gains of genetic material were set respectively at 0.75 and 1.25.

Histology
Microscopically, MHPCs showed a variable cell proliferation rate and focal necrosis. All studied MNG showed a high cellular proliferation rate and are thus classified as a high grade MNG. The MHPC group showed oval and atypical cells with high vascularity. Areas presenting surrounding branched vessels exhibiting a staghorn pattern were specifically seen in MHPCs.
Mitoses were frequently observed in MNG but were rare in MHPC (Figure 1, suplementary Table).
Based on morphological and immunohistochemical staining features, a final diagnosis of anaplasic meningeal hemangiopericytoma was made in all 8 studied tumors.

STAT6 expression
As expected, the 5 MNGs were negative for STAT6 expression, used a surrogate for the NBA2-STAT6 fusion gene. Nuclear staining was exclusively present on 3 out of 8 MHPC specimens. 2 positive MHPCs specimens represent a first tumor and its relapse (tumor #4 and #6). The third positive MHPC for STAT6 was according to histology and IHC investigations set as an uncertain diagnosis case. The STAT6 labelling showed a focal area of nuclear positivity in addition to abundant cytoplasmic staining in the same tumor section (tumor # 8) (Figure 1).

Molecular analysis
We used a molecular approach based on copy     Table 3. Both MHPC and MNG presented alterations of 1p36 and ERBB2 regions, and had normal statuses of 3p, 7q3, 19q13. CDKN2A loss was exclusively detected on MNG samples. Of note, PTEN loss was present in both MHPCs (4 MHPC out of 8) and MNGs (3 MNG out of 5).

Discussion
Intracranial MHPC is a dural-based neoplasm that highly mimics MNGs. The World Health Organization has separated MHPCs from MNGs since 1993, and individualized MHPCs as a "mesenchymal, nonmeningothelial" tumor (Chan et al., 2012). With the exception of MHPC clinical outcome, there are few distinguishing features between MHPC and MNG.
Chiechi et al reported that MHPC occurs in locations similar to MNG (Chiechi et al., 1996;Akiyama et al., 2004). They are found mostly at supratentorial, less frequently at infratentorial, very rarely intraventricular and even intraparenchymal locations have been reported (Muttaqin et al., 1991;Abrahams et al., 1999;Alen et al., 2001;Kuzeyli et al., 2003). Similarly, most of MHPC as well as MNG in our cases are in a supratentorial location.
The pathological features crucial for the diagnosis also remain the most controversial. Further, the immunoprofiles of MHPC and MNG were reported as slightly different (Shoji et al., 2002). The most important IHC markers thought to be diagnostically useful are EMA and CD34. The EMA is an epithelial marker reported positive in MNG but focal staining occurring in some MHPCs led to diagnostic uncertainty (Rajaram et al., 2004). CD34 is thought to be specific for MHPCs but was also reported to be controversial with a positive staining noted in MNG (Rajaram et al., 2004). However, Bcl-2 has been reported as very sensitive marker to MHPCs (Rajaram et al., 2004).
In our study, EMA as a highly sensitive meningothelial marker was found to be positive in 60% of our MNGs. Nonetheless, an EMA focal staining was observed in 33% of MHPCs. Thus EMA expression cannot exclude the diagnosis of MHPC. Bcl-2 was positive in 62% of MHPCs, however a focal weak signal was also observed in 40% of MNG. As reported by other studies CD99 proves to be another marker that MHPC strongly express (Rajaram et al., 2004). In our series it proved to be a sensitive marker with MHPC (62%) but still remains positive in MNG (40%). In fact, both CD99 and Bcl-2 highlight the extensive overlap between the two tumor groups.
In accordance with Shoji et al, the majority of MHPCs strongly express p21, whilst it is almost entirely absent from MNGs p21, is the product of the WAF1/CIP1/SDI1 gene and an inhibitor of cyclin-dependent kinases. It has been clearly demonstrated that prognostic significance of p21 is enhanced in combination with other apoptotic factors (i.e. p53) and associated with a better outcome (Shoji et al., 2002). The absence of p21 expression in MNG may be associated with their high grade.
We noted that the genetic profiles of MHPC and MNG were very similar except for CDKN2A loss predominantly present in MNGs. CDKN2A loss was previously described in anaplasic tumors and it matched with the high-grade histological description in our MNG samples (Bostrom et al., 2001).
PTEN is located at 10q23.3, a chromosomal region frequently implicated in tumor malignancy. In contrast to a previous report (Peters et al., 1998), we found PTEN loss in 3 MNG. This gene loss was also detected in 4 MHPCs and is therefore not specific to either entity.
In our study, both IHC markers and genetic profiles did not discriminate MNG from MHPC. A fusion gene between NAB2 and STAT6 has been recently identified as potential marker specific for SFT (Barthelmess et al., 2014).
Based on Schweizer et al study, we use IHC staining with the STAT6 antibody that recognizes C-terminal portion of STAT6. STAT6 protein mainly localizes to in the cytoplasm whereas the NAB2 protein localizes to in the nucleus. In case of fusion between NAB2 and STAT6, the fusion protein preserves the C-terminal portion of STAT6 connected with the 3' portions of the NAB2 (UniProtKB Q15742). For this reason, if the fusion occurs we expect to find the STAT6 staining in the nucleus (Schweizer et al., 2013).
MHPC sections presented STAT6 nuclear staining in 3 samples, whereas MNG were all negative, adding more support to previous studies (Iwaki et al., 1988) that reported the presence of NAB2-STAT6 fusion gene more frequently in MHPCs then MNGs. This molecular difference could be attributed to distinctive tumor steam cell origin and/or tumorogenesis pathways. Our conclusions need to be followed up by an enlargement of our sample cohort.
Interestingly, a case with highly uncertain diagnosis presented high cell proliferation. According to the immunoprofile with CD34 and Bcl2 positivity, it is more likely consistent with MHPC but still uncertain. The STAT6 IHC results showed a focal area of nuclear staining in addition to abundant cytoplasmic staining in the same tumor section. This could provide a furthermore argument to propose the MHPC as a final diagnosis and show the utility of STAT6 in the differential diagnoses of such lesions.