SEARCH

SEARCH BY CITATION

Keywords:

  • fine-needle aspiration cytology;
  • histology;
  • electron microscopy;
  • hemangiopericytoma

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

Hemangiopericytoma (HPC) is a relatively rare neoplasm, accounting for approximately 2.5% of all soft tissue tumors. Its histopathology has been well documented but to the authors' knowledge reports regarding its fine-needle aspiration (FNA) cytology rarely are encountered. In the current study the authors report the cytologic findings in FNA specimens from nine confirmed cases of HPC and attempt to correlate the cytologic features with the biologic outcomes.

METHODS

FNA was performed with or without radiologic guidance. Corresponding sections of tissue were reviewed in conjunction with the cytologic preparations.

RESULTS

Nine FNAs were performed in 5 patients (3 men and 2 women) with an age range of 38–77 years (mean, 56 years). Two lesions were primary soft tissue lesions arising in the lower extremities; seven were recurrent or metastatic lesions from bone (one lesion), kidney (one lesion), pelvic fossa (one lesion), lower extremities (two lesions), trunk (one lesion), and breast (one lesion). All aspirates were cellular and were comprised of single and tightly packed clusters of oval to spindle-shaped cells aggregated around branched capillaries. Basement membrane material was observed in 6 cases (67%). The nuclei were uniform and oval, with finely granular chromatin and inconspicuous nucleoli in all cases except one. No mitotic figures or areas of necrosis were identified. A correct diagnosis of HPC was made on one primary lesion and all recurrent or metastatic lesions.

CONCLUSIONS

HPCs show a spindle cell pattern in cytologic preparations and must be distinguished from more common spindle cell lesions. The presence of branched capillaries and abundant basement membrane material supports a diagnosis of HPC. Immunohistochemistry and electron microscopy performed on FNA samples may be helpful in the differential diagnosis. FNA is a useful and accurate tool with which to confirm recurrent or metastatic HPC; however, prediction of the biologic behavior of HPC based on cytologic features is not feasible. Cancer (Cancer Cytopathol) 1999;87:190–5. © 1999 American Cancer Society.

Hemangiopericytomas (HPCs) are relatively uncommon tumors, accounting for 2.5% of all soft tissue neoplasms.1 They primarily occur in patients as painless masses without specific clinical or radiologic features.2, 3 HPCs may arise in any part of the body but occur most commonly in the lower extremities, retroperitoneum, and the head and neck region.2, 3

Although the histologic features of HPC are well known, to our knowledge reports regarding the cytologic findings are few and have been limited to case reports and small series.4–9 In the current study we report the cytologic findings of fine-needle aspiration (FNA) biopsy of nine confirmed cases of HPC and attempt to correlate the cytologic findings with the clinical outcomes.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Nine FNA biopsies from five patients with HPC were retrieved from the files of the Pathology Department of New York University Medical Center (Tisch and Bellevue Hospitals) between 1989–1997. All cases were confirmed after resection. One case has been reported previously.9 Five of the FNA biopsies were performed either with 25-gauge or 27-gauge needles by cytopathologists and four FNA biopsies were performed with 22-gauge needles by a radiologist using computed tomography guidance. Smears were air-dried and stained by the Diff-Quik technique (Baxter Diagnostics, McGaw Park, IL), a modified Giemsa stain or were fixed in alcohol and stained with the Papanicolaou stain or hematoxylin and eosin. In two cases additional material from the FNA biopsy was placed in 10 mL of RPMI-buffered medium for preparation of cell blocks. Surgical specimens were fixed in 10% buffered formalin, processed in the routine fashion, and embedded in semisynthetic paraffin. Five-μm sections were cut and stained with hematoxylin and eosin. Immunocytochemical studies with antibodies to S-100 protein (Dako Co., Carpinteria, CA), CD34 (BioGenex, San Ramon, CA), vimentin (BioGenex), desmin (BioGenex), and actin (Dako Co.) were performed on formalin fixed specimens using the avidin-biotin-peroxidase method in the two cases with cell block preparations and on surgical specimens in four cases. In one case, additional fresh aspirated material was fixed in 10% glutaraldehyde for electron microscopy. Electron microscopy also was performed in fresh tissue from the surgical specimens in two cases.

The aspirates were examined for the following features: 1) cellularity; 2) architecture; 3) characteristics of the cell clusters; 4) nuclear features including nuclear size and shape, chromatin pattern, and presence of nucleoli; 5) amount and character of the cytoplasm; 6) presence of mitotic figures expressed in number per 10 high-power fields (HPF); and 7) presence of necrosis. The cytologic findings were compared with the histologic findings. A histologic diagnosis of HPC was made if the tumor showed the characteristic branched vascular pattern with uniform tumor cells arranged around these dilated vascular channels3 with no other specific stromal features. Other factors assessed included foci of increased cellularity, necrosis, hemorrhage, or increased mitotic figures. Medical charts were reviewed and follow-up information was obtained for all five patients.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Clinical Data

Clinical data for the five patients are summarized in Table 1. There were 3 men and 2 women, with a mean age of 56 years (range, 38–77 years). One patient underwent FNA of 5 seperate sites over 11 years; the remaining patients had one site aspirated. Two biopsy specimens were from primary soft tissue lesions in the lower extremities. The other seven biopsy specimens were from recurrent or metastatic lesions: two from the lower extremities, one from the pelvic fossa, one from a kidney, one from the vertebral column, one from the trunk, and one from a breast. The breast lesion was reported previously.9 The size of the tumors ranged from 2.5–7.0 cm. Follow-up was available for all patients and ranged from 2–12 years (mean, 6.4 years). Three patients developed recurrent tumors or distant metastases.

Table 1. Summary of Clinical Data
CaseAge (yrs)GenderPrimarySite of FNAP vs. MF/up (yrs)Cytologic diagnosis
  1. FNA: fine-needle aspiration; P: primary; M: metastatic; F/up: follow-up; M: male; F: female: R: right; L: left.

177MThighThighP5Spindle cell neoplasm
248MThighThighP2Hemangiopericytoma
357FButtockBreastM12Hemangiopericytoma
R flankMHemangiopericytoma
R thighMHemangiopericytoma
KidneyMHemangiopericytoma
L thighMHemangiopericytoma
438FDuraDuraM4Hemangiopericytoma
563MPelvisPelvisM9Hemangiopericytoma

Cytologic Findings

All aspirates were moderately to markedly cellular and were comprised of tissue fragments, loose cellular aggregates, and single cells (Fig. 1). The tissue fragments often were numerous, with irregular shapes and considerable nuclear crowding and overlapping. The majority of the tumor cells were arranged haphazardly in these clusters although some were arranged in a radial fashion. Endothelial-lined capillaries were identified easily within the tissue fragments (Fig. 2), and these vessels usually were branched. The endothelial cells were elongated with hyperchromatic nuclei and chromatin clumping compared with the perivascular cells. In six aspirates, metachromatic basement membrane material was conspicious within the tissue fragments on Diff-Quik stained smears (Fig. 3). This material separated the tumor cells from the endothelial cells or presented as a central core surrounded by neoplastic cells.

thumbnail image

Figure 1. Smear from a hemangiopericytoma includes irregular fragments of tissue, loose cellular aggregates, and single cells (Diff-Quik, ×40).

Download figure to PowerPoint

thumbnail image

Figure 2. Uniform oval or spindle-shaped cells with a high nuclear to cytoplasmic ratio and indistinct borders. Endothelial-lined capillary vessels lie within the fragments of tissue. Endothelial cells are long and hyperchromatic compared with the neoplastic cells (Diff-Quik, ×400).

Download figure to PowerPoint

thumbnail image

Figure 3. Metachromatic basement membrane material within the fragments separates tumor cells from endothelial cells (Diff-Quik, ×400).

Download figure to PowerPoint

The tumor cells were oval or spindle-shaped with a high nuclear to cytoplasmic ratio and indistinct cellular borders. The cytoplasm was scant, pale, and delicate. No cytoplasmic vacuoles were identified. The nuclei were uniform in all FNA specimens except one. They were oval to elongated and approximately two–three times larger than that of a small lymphocyte. The nuclear contour generally was smooth and regular; occasional nuclear irregularities were noted in one FNA sample. The nuclei had finely granular, evenly distributed chromatin with indistinct nucleoli. An FNA biopsy from one case demonstrated small but distinct nucleoli and a coarsely granular chromatin pattern. No significant cytologic atypia, mitotic figures, or necrosis were noted in any of the aspiration biopsies.

Cell block preparations showed neoplastic cells with round to oval nuclei and a moderate amount of cytoplasm. Cells were arranged around vascular spaces of varied sizes. A cytologic diagnosis of HPC was made in one of the two primary tumors and in all recurrent/metastatic lesions. The other primary tumor was diagnosed as a spindle cell neoplasm, not otherwise specified. Electron microscopy and immunocytochemistry were performed in the one primary case, where a diagnosis of HPC was made by FNA biopsy.

Histologic Findings

In routine sections of tissue all tumors were highly cellular and were comprised of relatively uniform cells. All contained a prominent branched vascular pattern, with marked variation in the size of the vascular spaces. The neoplastic cells aggregated around the branched vascular spaces, which often had a staghorn appearance (Fig. 4). The neoplastic cells occasionally were arranged radially around vessels. The vascular spaces were lined by endothelial cells and ranged from large sinusoids to capillary-sized vessels. The individual neoplastic cells had round, oval, or elongate nuclei with a scant to moderate amount of cytoplasm and ill-defined cytoplasmic borders. All cases had solid cellular areas in which tumor cells compressed vascular spaces. However, the cells were uniform and showed no atypical features. The number of mitotic figures varied. In 4 cases, there were <4 mitotic figures per 10 HPF. In the remaining 5 cases, >4 mitotic figures per 10 HPF were noted. Necrosis was noted in one case in which the patient had been treated with arterial embolization prior to the resection of the tumor. Two cases contained areas of microcystic degeneration; however, no hemorrhagic areas or areas of necrosis were noted.

thumbnail image

Figure 4. A histologic section of tumor is comprised of uniform neoplastic cells around branched thin-walled vascular spaces, with a staghorn appearance (H & E, ×100).

Download figure to PowerPoint

Ancillary Studies

The tumor cells in all cases in which immunocytochemistry was performed showed immunopositivity for vimentin and CD34. CD34 highlighted the vascular spaces and also was positive within the neoplastic cells. The staining pattern with CD34 ranged from focal positivity in three cases to diffuse and stronger positivity in three cases. Staining for S-100 protein, desmin, and actin was negative within the tumor cells.

Ultrastructural analysis showed spindle-shaped cells with oval, irregular nuclei arranged around irregular vascular channels. Some cells contained long cytoplasmic extensions and numerous pinocytotic vesicles. Focally, dense material lined the outer membranes of the cells. Intermediate filaments were identified within the cytoplasm of some cells.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

HPC, first described by Stout and Murray in 1942,10 is an uncommon neoplasm of the soft tissues. It may arise in the majority of body sites,2, 3, 11, 12 but more often is found in the lower extremities (in particular the thigh), followed by the retroperitoneum and the head and neck region. It is a tumor of adult life with an equal gender distribution. In the majority of instances, both the clinical and radiologic findings are nonspecific. Despite many publications regarding HPC,2, 3, 11, 12 the existence of HPC as a distinct tumor entity recently has been questioned.13 The diagnosis of this tumor can be challenging because the neoplastic cells are spindle-shaped and can be difficult to distinguish from cells of endothelial, fibroblastic, or histiocytic origin. The hallmarks of HPC include a relatively uniform population of spindle-shaped cells arranged around vascular channels, which include dilated staghorn shaped vessels. These findings are not pathognomonic and similar architectural patterns may be observed in other soft tissue neoplasms, giving evidence to current skepticism regarding the entity. We believe that a small number of tumors with the characteristic microscopic appearance of HPC and the absence of any specific differential features justifies the diagnosis.

The use of FNA biopsy in the diagnosis of soft tissue tumors remains unsettled.14–16 FNA is useful for excluding nonmesenchymal lesions such as carcinoma and lymphoma and for identifying localized infections. In establishing a diagnosis of mesenchymal neoplasm, both the sensitivity and specificity approach 96%.17 The majority of HPCs reported in the literature4, 7, 17–19 and in our study were classified correctly as mesenchymal based on the cytologic findings. This information may guide subsequent clinical management. In our study the use of electron microscopy and immunocytochemistry on FNA samples along with the cytologic findings enabled us to make the correct diagnosis in one primary tumor. In the second primary tumor additional material for ancillary studies at the time of FNA was not available; thus, a definitive diagnosis was not made. In patients with histologically documented HPC, FNA is valuable and accurate in confirming recurrent or metastatic tumor. All our patients with recurrent or metastatic HPCs were diagnosed correctly based on routine cytology alone. Obviously, review of prior specimens is critical to diagnosis.

FNA biopsy smears of HPCs usually are cellular and the perivascular proliferation of cells observed in sections are found in cytologic specimens. The neoplastic cells aggregate around capillaries, which often are branched, but dilated, staghorn vascular channels are not observed. Metachromatic basement membrane material often is observed within the cellular aggregates and in the perivascular areas. This can be highlighted by silver staining on cytologic smears and cell block preparations.7 Nickels et al.6 described the occurrence of endothelial-lined knobby projections of tumor cells in the FNAs of all their HPCs. Like other authors,5, 8 we were unable to identify such formations. The oval or fusiform neoplastic cells can be distinguished from the endothelial cells, which appear more slender and have more open chromatin. Focal acinar formations by the tumor cells have been described and may be misinterpreted as evidence for epithelial differentiation,5, 8 representing a diagnostic pitfall. Immunohistochemistry and electron microscopy are necessary to clarify the diagnosis, and these studies must exclude the several spindle cell neoplasms that show similar cytologic features.

The differential diagnosis includes soft tissue tumors, which focally may show a hemangiopericytomatous growth pattern. These include malignant fibrous histiocytoma (MFH), synovial sarcoma, mesenchymal chondrosarcoma, and leiomyosarcoma.12 In these tumors, the hemangiopericytomatous pattern occurs with other findings characteristic of these specific lesions. Nuclear pleomorphism and the presence of multinucleated giant cells favor MFH, whereas a biphasic pattern of spindle-shaped and epithelioid cells suggests synovial sarcoma. The findings of chondrocytes and chondroid matrix point to mesenchymal chondrosarcoma. Leiomyosarcoma usually contains spindle-shaped cells with blunt-ended nuclei and cytoplasmic staining for actin. Distinguishing these lesions on FNA smears may be difficult, especially in smears comprised of bland spindle-shaped cells. Lesions such as monophasic synovial sarcoma, low grade MFH, and dermatofibromasarcoma protuberans must be considered in these cases; ancillary procedures such as electron microscopy and immunocytochemistry may lead to the diagnosis of HPC. The neoplastic cells often are immunoreactive to vimentin, factor XIIIa,20 and CD3421 as observed in our cases. Such findings are nonspecific; thus, the lack of immunostaining with epithelial, endothelial, muscular, and neural markers is important in excluding other neoplasms. Ultrastructurally, the findings of pinocytotic vesicles and multilayered basal lamina separating the individual tumor cells from one another and from the endothelial cells ultrastructurally favor a diagnosis of HPC.22

In a report of 136 lesions of soft tissue, 1 of the 2 HPCs was diagnosed correctly by FNA.17 Kumar and Misra7 correctly diagnosed two aspirates of primary HPC. Similarly, HPC was diagnosed correctly in one of the two primary HPCs in our study. In other reports Rydholm et al.18 and Akerman et al.19 described two HPCs interpreted as mesenchymal tumors, not otherwise specified. In another case, a cytologic diagnois of “mesenchymal tumor” was given to a breast lump from a 40-year-old man that was proven to be an HPC.4

Pathologists have difficulty predicting the behavior of HPC based on the clinical and pathologic features. Reports of the incidence of metastasis in HPC vary from 11.7% to 56.3%,10, 14, 23 and several microscopic criteria have been described as helpful in predicting the behavior and in distinguishing benign from malignant HPCs.12 In the majority of benign cases, < 2 or 3 mitotic figures per 10 HPF generally are noted and ≥4 mitotic figures are indicative of a tumor capable of recurrence and metastases. In addition, malignant HPCs tend to be larger and more cellular and display cellular atypia with areas of necrosis.

Nickels et al.6 reported their findings in five cases of malignant HPCs on FNA biopsy. Three of their cases showed obviously malignant features including large hyperchromatic nuclei with prominent nucleoli and numerous mitotic figures. The other two cases showed only benign neoplastic cells. In another report on the cytologic features of malignant HPCs, Geisinger et al.8 described a moderate degree of pleomorphism in approximately 50% of their aspirates and the presence of numerous mitotic figures in all patients. Except for its large size, the FNA smears of a malignant HPC reported by Nguyen et al.5 resembled a benign counterpart, and malignancy could not be determined. In our study, none of the “aggresive” HPCs had identifiable mitotic figures or nuclear pleomorphism cytologically. Obviously the presence of mitotic figures and cytologic atypia are indicative of a malignant tumor, but their absence does not exclude “aggressive” behavior.

In aspirates from cellular tumors occurring in the soft tissues of the body, especially the lower extremities, the findings of relatively uniform, oval, spindle-shaped neoplastic cells around capillaries and basement membrane material in perivascular areas suggest the diagnosis of HPC. Immunocytochemistry and electron microscopy can support a original diagnosis of HPC further. FNA biopsy is an accurate procedure for confirming recurrent or metastatic tumors. The outcome of HPC can not be predicted reliably, but the significant incidence of recurrence and metastases over long periods warrants long term follow-up of patients.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES