A reevaluation of the primary diagnosis of hemangiopericytoma and the clinical importance of differential diagnosis from solitary fibrous tumor of the central nervous system

1. Introduction 

Hemangiopericytoma (HPC) used to be classified as a hemangiopericytic variant of angioblastic meningioma. There were two reasons for doing so: HPCs generally resemble meningiomas, and lesions with features common to both HPCs and meningiomas were presumed to exist [1], [2], [3]. Solitary fibrous tumor (SFT) of the meninges has also been recently recognized as a main differential diagnosis from HPC [4].

Central nervous system SFT is a rare neoplasm which usually develops as a dura-based mass fed by the eternal carotid arteries. The typical histologic features are spindle cells arranged in a patternless architecture or in interlacing fascicles, prominent collagen bands, and branching vascular channels with thin walls. SFT can mimic hemangiopericytoma or angiomatous meningioma in neuroradiological or pathological studies, and thus is sometimes misdiagnosed as such. SFTs are benign tumors with good prognosis after surgical resection, and seldom do they recur or metastasize [5], [6], [7], [8], [9].

In contrast to SFTs, HPCs often have high incidences of local recurrence and metastasis, even if they have post-surgical radiation therapies [10]. In light of the difference in prognosis between HPCs and SFTs, the differential diagnosis between these tumors should be carefully considered.

Central nervous system SFT is a relatively new entity and presumably has been difficult to distinguish from HPC and angiomatous meningioma until recently. Now, however, newer immunohistochemical evaluations for SFT, HPC, and meningioma can be effectively used for the different diagnosis of these three tumors. To our knowledge, there has been only one retrospective study to reevaluate HPCs by reattempting differential diagnosis from the newly emerged SFTs of the central nervous system [11]. In this study we used the new immunohistochemical methods to reassess all the HPCs diagnosed between 1970 and 2006, in order to determine whether there were SFTs that had been misdiagnosed. We also evaluated the prognosis of each case after the re-evaluation to reaffirm that SFT has a more benign behavior than HPC.

2. Materials and methods 

All intracranial dural based tumors diagnosed as hemangiopericytoma (HPC) were retrieved from the files of the Kanazawa University Hospital and Kumamoto University Hospital, Department of Neurosurgery. 1018 cases were diagnosed as meningioma at the time of admission between 1970 and 2006, and 13 cases of those were diagnosed as HPCs with histological examination after surgical tumor resections. All 13 cases were evaluated in this study with routine HE staining, reticulin staining, and specific immunohistochemical staining including CD34, Bcl-2, Epithelial membrane antigen (EMA), Vimentin, and S100 protein. In addition, MIB-1 labeling index (LI) was measured.

Histological sections were cut from formalin-fixed paraffin-embedded tissue blocks. Sections were examined using the avidin–biotin (ABC) technique with appropriate negative and positive controls. Sections from a meningioma that showed +4 positivity were used as a positive control for EMA. All other control tissues were selected as per manufactures’ recommendations. The sources and dilutions of antibodies used for immunohistochemical analysis were as follows: CD34, DakoCytomation (Glostrup, Denmark), monoclonal (1:100); Bcl-2, DakoCytomation, monoclonal (1:100); EMA, Dako Corporation (Carpinteria, CA), monoclonal (ready to use); Vimentin, DakoCytomation, monoclonal (ready to use); S-100, Dako Corporation, polyclonal (1:6000); MIB-1 (Ki-67), DakoCytomation, monoclonal (1:200). Immunohistochemical staining was graded as follows: 0, no staining; +1, weak focal; +2, weak; +3, strong focal (<50%); +4, strong diffuse reactivity (>90%). The MIB-1 labeling index (LI) was measured by counting the percentage of immunopositive tumor nuclei within a high magnification field in which the staining was highest.

3. Evaluation criteria 

Based on our reevaluation in this study, we propose that SFT be diagnosed in cases that meet all four of the following criteria: (1) strong immunohistochemical staining for CD34 (score of over 3+); (2) strong immunohistochemical staining for Bcl-2 (score of over 3+); (3) sparse pattern of reticulin staining around cell clusters; (4) negative immunohistochemical staining for EMA. In our view, these are the optimal evaluation criteria for diagnosing SFT as an entity distinct from HPC [4], [5]. Unfortunately, however, there are still no direct methods by which to distinguish intracranial SFTs from HPCs. Moreover, we cannot rule out the possibility that some HPCs will meet all four criteria as false positives.

4. Results 

The tumor cells in all 13 of the cases contained elongated nuclei with dispersed chromatin, an absence of prominent nucleoli, and a vascular structure with a remarkably prominent “staghorn” pattern. These microscopic features were the bases for the original diagnoses of HPC. In all tumors, mitoses ranged from 0 to 2/10400× fields, and all were diagnosed as grade II.

Immunohistochemical stains showed strong (+3 or +4) reactivity for both CD34 and Bcl-2 in four cases (case nos. 1, 6, 8, and 9). All 4 of these cases also exhibited sparse reticulin staining patterns around cell clusters and an absence of EMA staining. Based on these immunohistochemical features, all four were newly diagnosed as SFTs.

Tumor staining with CD34 and Bcl-2 was weak or absent (0, +1, or +2) in eight cases (case nos. 2, 3, 5, 7, 10, 11, 12, and 13). All eight of these cases were negative for EMA, and seven of the eight exhibited a fine reticulin-staining pattern around individual cells. These immunohistochemical features strongly supported the original diagnoses as HPCs.

One case (case no. 4) was negative for both CD34 and Bcl-2 staining, but exhibited strong and diffuse positive staining with EMA. This case was newly diagnosed as meningioma.

All tumors but one (case no. 4) were negative for S-100, and the one exception was strongly positive for S-100.

All tumors but one (case no. 6) were strongly positive for vimentin.

All 13 cases were studied by immunohistochemistry for the Ki-67 (MIB-1) antibody. MIB-1 was 4% in the case reevaluated as meningioma (case no. 4) and 8% in the case reevaluated as HPC (case no. 13), and less than 1% in all of the other cases.

The above results from immunohistochemical staining are summarized in Table 1 and in Fig. 1.

Table 1. Immunohistochemistry and reevaluation of 13 cases

	Case	Age/sex	CD34	Bcl2	EMA	Vimentin	S-100	Reticulin	MIB-1 (%)	diagnosis
	1	44/M	4+	4+	0	3+	0	Sparse	<1	SFT
	2	61/M	1+	0	0	4+	0	Sparse	<1	HPC
	3	48/M	0	0	0	3+	0	Fine	<1	HPC
	4	74/F	0	0	4+	4+	4+	Sparse	4	Meningioma
	5	64/F	2+	1+	0	4+	0	Fine	<1	HPC
	6	56/M	4+	4+	0	1+	0	Sparse	<1	SFT
	7	45/M	1+	1+	0	4+	0	Fine	<1	HPC
	8	58/F	4+	3+	0	4+	0	Sparse	<1	SFT
	9	35/M	3+	3+	0	3+	0	Sparse	<1	SFT
	10	30/M	0	2+	0	4+	0	Fine	<1	HPC
	11	64/M	0	2+	0	3+	0	Fine	<1	HPC
	12	84/M	0	2+	0	3+	0	Fine	<1	HPC
	13	35/M	2+	0	0	4+	0	Fine	8	HPC

HPC, hemangiopericytoma; SFT, solitary fibrous tumor Staining score: 0, no staining; +1, weak focal; +2, weak; +3, strong focal (< 50%); +4, strong diffuse reactivity (> 90%).

View Large Image Download to PowerPoint

Fig. 1. Photomicrographs showing the results of immunostainings. Each column showing representative case of reevaluation, right panel: hemangiopericytoma (HPC); center: solitary fibrous tumor (SFT); left: meningiomas, and all panels with original magnification 200×. (A) HE staining. Each case had rich vascular formations like a so-called stag-horn shape. (B) Immunostaining with CD34. Only the center panel (case no. 6) showed diffuse positive staining in tumor cells, and other two panels (right panel, case no. 3, and left panel, case no. 4) showed positivity in only vascular pericytes. (C) Immunostaining with Bcl-2. Only center panel showed diffuse positive staining in tumor cells, and other two showed negative. (D) Immunostaining with EMA. Only the left panel showed diffuse positive staining, and other showed negative. (E) Reticulin staining. The right panel showed fine reticulin pattern, and other panels showed sparse pattern. From above staining patterns, the right panel was diagnosed as HPC, the center panel was newly diagnosed as SFT, and the left panel was newly diagnosed as meningiomas.

4.1. Clinical features and reevaluations 

Of the eight cases reevaluated and diagnosed as HPC, seven had local recurrences and four had extra-cranial metastasis. Five of the eight cases died of the disease.

Only one of the four cases newly diagnosed as SFT had local recurrence and died of the disease. None of the four cases newly diagnosed as SFT had extra-cranial metastasis.

The one case newly diagnosed as meningioma, the case with relatively high MIB-1 LI (4%), had local recurrence and died of the disease (Table 2).

Table 2. Clinical features and reevaluation of 13 cases

	Case	Age/sex	Location	Operation	RTx	Rec.	Meta.	Follow up (M)	Outcome	Diagnosis
	1	44/M	Occipital	Gross total	+	−	−	87	Died of other cause	SFT
	2	61/M	Parietal	Gross total	−	+	−	–	Lost to follow-up	HPC
	3	48/M	Temporal	Gross total	+	+	−	54	Died of disease	HPC
	4	74/F	Temporal	Partial-subtotal	−	+	−	51	Died of disease	Meningioma
	5	64/F	Post. fossa	Gross total	+	−	−	51	No disease	HPC
	6	56/M	Temporal	Gross total	+	−	−	63	No disease	SFT
	7	45/M	Parietal	Gross total	+	+	+	120	Died of disease	HPC
	8	58/F	Occipital	Partial-subtotal	+	−	−	13	Alive with disease	SFT
	9	35/M	Post. fossa	Gross total	−	+	−	72	Died of disease	SFT
	10	30/M	Pineal	Gross total	+	+	+	56	Died of disease	HPC
	11	64/M	Frontal	Gross total	+	+	−	–	Lost to follow-up	HPC
	12	47/F	C-P angle	Partial-subtotal	+	+	+	84	Died of disease	HPC
	13	35/M	Suprasellar	Partial-subtotal	+	+	+	41	Died of disease	HPC

HPC, hemangiopericytoma; SFT, solitary fibrous tumor; RTx, radiation; Rec., recurrence; Meta., metastasis; M, months.

5. Discussion 

Hemangiopericytoma (HPC) is usually a dura-based tumor mimicking an angiomatous meningioma. Based on the new WHO classification of tumors of the central nervous system published in 2007, HPC is classified as a mesenchymal tumor that is non-menigothelial tumor and a different entity from meningiomas [1]. However, because both tumors are dura-based and usually fed from external carotid arteries, it is generally difficult to distinguish HPCs from meningiomas by neuroradiological examinations. HPCs are often remarkable for their “staghorn” vascular pattern, a histological pattern similar to that manifested by angiomatous meningioma (i.e., a distortion of the thin-walled blood vessels into thin, branching structures by encroaching tumor cells). Since HPC is usually EMA-negative and meningioma is usually EMA-positive, the best tool to distinguish HPC from angiomatous meningioma is EMA immunohistochemical staining. The high incidences of HPC metastasis and local recurrence call for careful following of patients even after gross surgical resection, as well as additional radiation therapy as an adjuvant treatment in some cases [10], [12].

SFT of the central nervous system (CNS) has recently emerged as a distinct clinicopathological entity among the dura-based CNS meningeal tumors. According to the new WHO classification, SFT is classified as one of the subsets of mesenchymal tumors that is the same subset as HPC [13]. SFT is a benign spindle cell tumor which presents similarly to HPC both clinically and radiologically. The final diagnosis relies heavily on histological features, especially those of a immunohistochemical nature. In contrast to HPCs, SFTs seem to be successfully managed by surgery alone and to have an excellent prognosis [4], [5], [7], [14]. Whereas a spectrum between the HPC and SFT has been suggested [7], in most cases, HPC is distinct from SFT and only a few SFTs have been reported to recur locally with malignant features, or to metastasize outside the CNS [6], [8], [9], [14]. A distinct differential diagnosis after the first operation should be important. If the case is diagnosed as HPC, meticulous and careful examinations for a local recurrence and remote metastasis should be necessary in relatively short term after the first operation.

In the four cases that were newly diagnosed as SFT, only one had local recurrence, and no case had extra-cranial metastasis during the follow-up periods. Among the eight cases reconfirmed to be HPC through the reevaluation, on the other hand, seven had local recurrence, four had extra-cranial metastasis after adjuvant radiation therapy, and five died from the disease (Table 2). The clinical courses of these cases seem to reaffirm that SFTs behave more benignly than HPCs, and they attest to the adequacy of our evaluation criteria based on immunohistochemical features.

The present study revealed relatively high incidence, four of 13 cases (31%), of newly diagnosed SFT among the tumors initially diagnosed as HPC. In the previous study, there was only one case (6%) reassessed as SFT among the 18 tumors initially diagnosed as HPC [11]. The difference incidence between the present study and the previous study may depend on the different diagnosis criteria for SFT used in the studies. Our diagnosis criteria for SFT included the tumor with strong focal CD34 positivity that may be excluded in the previous study. In addition, in the present study, we used bcl-2 immunohistochemistry for diagnosis of SFT that was not used in the previous study. Although there still be no definite criteria of SFT, our results of reevaluation may not be over estimated for SFT because the clinically courses of the tumors newly diagnosed as SFT was obviously better than that of HPCs.

One of the 13 cases was reevaluated and newly diagnosed as a meningioma based on EMA-positive staining. The original misdiagnosis of HPC would probably have been avoided if the EMA immunohistochemistry had been available when this case was first encountered. Only the case showing EMA-positive staining was also positive for S-100 immunohistochemistry. However, because the other tumors of central nervous system including schwannoma tend to be positive for S-100, the best tool for diagnosis of meningioma may be EMA immunohistochemical staining. As for the vimentin, the immunohistochemical activities were varied among the 13 tumors, and the immunohistochemistry for vimentin could not be useful for differential diagnosis in these three types of tumor.

There were two cases with a high MIB-1 LI level, one was meningioma (case no. 4, 4%), and the other was HPC (case no. 13, 8%). The relatively high MIB-1 LI might have been a sign of more active tumor cell growth and local recurrence, and the above two cases with relatively high MIB-1 LI had shorter survival period (51 months and 41 months, respectively) than mean survival time of the cases (68 months) that died of disease. MIB-1 LI generally indicates the growth potential of a tumor, and an extremely low index (<1%) may indicate a lower chance of tumor malignancy. In this study, however, almost all HPCs with recurrence or metastasis also had very low MIB-1 LI (<1%). In the previous studies, median MIB-1 LIs were from 2.9% to 10%, and each index was varied, ranged between less than 1% and 40% [15], [16], [17]. Even taking account of the variety of the MIB-1 LI, our series had lower LI than previous studies. The explanations of the difference may be that (1) all tumors in our series were lower grade (grade II) and too small sample size to compare, or (2) staining technique for MIB-1 had minor differences among each laboratory of study groups, leading to the varieties of LI. However, the conclusions derived from the result of MIB-1 immunohistochemistry in the present study were same as the previous studies, i.e., among the HPCs, no correlation was found between MIB-1 LI and tumor aggressiveness, and MIB-1 LI may not be predictive of poor prognosis with recurrence or metastasis.

Ten of 13 cases originally diagnosed as HPC after the initial surgeries were put through adjuvant radiation therapy to reduce what was believed to be a strong chance of local recurrence or remote metastasis. The three of those 10 cases who were newly diagnosed as SFTs based on the specific immunohistochemical stainings of the current study might not have required those adjuvant radiation therapies. Though SFTs have been reported to recur or disseminate through the CSF [8], these tumors are basically benign. A recommendable first step, in planning the treatment strategy of vascular-rich meningeal tumors is to make an accurate differential diagnosis between SFTs and HPCs. Though differential diagnosis will undoubtedly prove difficult in some cases, meticulous pathological examinations with CD34, Bcl-2, and reticulin staining will help to distinguish SFTs from HPCs. In addition, EMA immunohistochemistry studies should be initially undertaken to distinguish HPCs and SFTs from meningiomas.