Malignant Giant Cell Tumor of Synovium (Malignant Pigmented Villonodular Synovitis): A Histopathologic and Fluorescence In Situ Hybridization Analysis of 2 Cases With Review of the Literature Open Access

In 1941, Jaffe et al1 characterized a variety of fibrohistiocytic tumors that occurred within the large joints and termed them pigmented villonodular synovitis (PVNS). He considered these lesions to be nonneoplastic inflammatory reactions to an unknown agent or agents. Although PVNS was capable of local recurrence and considerable destruction of the joint and surrounding soft tissue, it did not appear to be capable of distant metastasis. Since that time, many reports have confirmed the basic clinical and epidemiologic features of PVNS as first described by Jaffe et al,2–13 including its benign nature. Recently, distinctive chromosomal aberrations have been described for PVNS and giant cell tumor of tendon sheath,14–17 and a few case reports have indicated the potential for metastases.18–23 These aggressive cases have generally been designated as malignant giant cell tumor of tendon sheath or malignant PVNS.18–23 In general, these neoplasms had been present for several years before malignant degeneration occurred, and the patient died either following local recurrence or with pulmonary metastases. Bertoni et al23 reported a series of 8 cases and set forth criteria for the diagnosis of malignant PVNS. We report the histopathologic and fluorescence in situ hybridization (FISH) findings in 2 cases of PVNS that pursued an aggressive course and compare the findings in our own cases with those reported by Bertoni et al23 and other investigators.

A 65-year-old woman was admitted to Duke University Medical Center, Durham, NC, in November 1987 for right femoral head replacement and excision of an intra-articular and soft tissue mass involving the right hip. A previous biopsy had disclosed PVNS. Operative margins appeared free of PVNS. In June 1988, a second biopsy of the right hip revealed recurrent PVNS, and the tumor mass was irradiated by external beam. The patient tolerated radiation therapy well, but in January 1989, she complained of fever, night sweats, weakness, and hematuria. Studies at that time were consistent with a vasculitis and immune complex glomerulonephritis. She was treated with prednisone followed by azathioprine. No significant improvement was detected, and in May 1989, she started cyclophosphamide therapy. She was readmitted in July for treatment of increasing renal failure, at which time she started a trial of antithymocyte globulins without response and was readmitted 2 weeks later for a trial of cyclosporine. Following discharge, she died at home in early August 1989, secondary to severe bilateral bronchopneumonia.

Postmortem examination revealed a 10 × 15 × 8-cm soft tissue mass in her right hip and thigh. This mass extended into the gluteus maximus, rectus femoris, and the area of the right hip prosthesis. The mass was histologically interpreted as a malignant fibrous histiocytoma. Autopsy did not reveal evidence of distant metastases.

In 1944, a 24-year-old white man sustained an injury to his right knee while on active duty in the US Air Force. Because of pain and persistent joint effusion, he underwent synovectomy. Pathologic examination disclosed PVNS. The slides representative of the resected material are unavailable for current review, and margin status is unknown. He received a course of radiation therapy of unknown duration and extent to lessen the chance of recurrence. During the next 3 decades, he experienced increasing pain and limitation of motion, finally requiring crutches for ambulation in 1981. In 1983, skin nodules appeared over the anterior aspect of the right knee, and knee arthroscopy was performed, revealing extensive PVNS. Because of the extensive nature of the process, total knee arthroplasty was felt inadvisable. In 1984, he underwent an above-the-knee amputation for persistent PVNS. Examination of the amputation specimen demonstrated PVNS penetrating through the retinaculum and into the popliteal fossa, with destruction of the distal femur. Operative margins were reported to be free of PVNS. In 1988, a recurrent nodule was detected at the above-the-knee amputation stump site. Surgical excision with histopathologic evaluation confirmed recurrent PVNS. Between mid-1988 and November 1990, 3 additional stump site recurrences were resected, each showing similar histologic findings. A computed tomographic scan performed in November 1990 revealed a large tumor in his right buttock. Radiographic imaging studies did not demonstrate evidence of direct extension from the above-the-knee amputation stump site. Histologic study at the time of hemipelvectomy (December 1990) disclosed PVNS. The tumor showed evidence of venous invasion. Operative margins were free of neoplastic change. Five months later, in April 1991, recurrent PVNS was found within the skin and subcutaneous tissue overlying the right buttock. The patient was subsequently lost to follow-up.

Materials from cases 1 and 2 were routinely formalin fixed and paraffin embedded. The sections for light microscopy were stained with hematoxylin-eosin. FISH studies were performed on isolated nuclei obtained from paraffin-embedded block material.

Briefly, 50-μm-thick sections were cut from the formalin-fixed, paraffin-embedded blocks for each patient. To isolate the nuclei from the paraffinated tissue, the following technique was used. Each section was first cut into smaller pieces and placed in 15-mL tubes with 5 mL of HemoDe for 20 minutes. After spinning, the supernatant was carefully removed (this procedure was repeated 3 times). Subsequently, 10 mL of 100% ethanol was added to the tube for 10 minutes (this procedure was repeated twice). Then, 2 mL of pepsin solution (40 mg of pepsin in 10 mL of 0.9% NaCl, pH 1.5) was added to the tube, which was placed in a 37°C water bath for 2½ hours. The digested tissue was then filtered using a 1-mL syringe, 40-μm-mesh, and 18-gauge needle. The resulting suspension was then placed in the refrigerator overnight with 10 mL (1×) phosphate-buffered saline. The next day the suspension was treated again with freshly made pepsin for another 2½ hours at 37°C, followed by the addition of 100 μL of pepstatin. This suspension was filtered and placed in a tube with (1×) phosphate-buffered saline. After centrifugation, these cells were resuspended with (1×) phosphate-buffered saline and dropped onto slides.

These slides were used for FISH. This procedure was performed as previously described,24 with modifications. Briefly, patients and control slides (peripheral blood lymphocytes from a normal individual) were hybridized using a centromeric probe for chromosome 7 and a region-specific probe for chromosome 5 (5p15.2) (Vysis, Inc, Downers Grove, Ill). Both probes were directly labeled, and for the purpose of visualizing them simultaneously in the same nuclei, a SpectrumOrange 7 probe and a SpectrumGreen 5 were used. Slides were denatured in 70% formamide and 2× SSC, pH 7.0, at 70°C and dehydrated in ice-cold ethanol series. The probe mixture was denatured at 70°C and applied to prewarmed (37°C) slides. Slides were covered with a glass coverslip, sealed with rubber cement, and incubated overnight at 37°C in a moist chamber. Posthybridization washes were performed to remove unbound probe using 50% formamide and 2× SSC, pH 7.0. The nuclei were stained using DAPI counterstain and viewed under a fluorescence microscope equipped with a CCD camera and a Vysis Imaging System.

The resection specimen obtained in November 1987 showed on gross examination that the tumor diffusely involved the synovium and extended into the surrounding soft tissues. Histologic sections revealed a heavy infiltrate of mononuclear histiocytes having a granular-to-foamy cytoplasm. Most cells grew in a sheetlike pattern, although rare clefts and villiform structures were seen. The infiltrate extended into the surrounding soft tissues. No maturational zoning toward the periphery or necrosis was seen. Admixed with the mononuclear histiocytes were a moderate number of multinucleated histiocytic giant cells. The nuclei of the giant cells were identical in appearance to those of the mononuclear cells. The histiocytic giant cells had between 4 and 20 nuclei. The individual nuclei had a finely clumped chromatin with some chromatin clearing (Figure 1). Nucleoli were generally single, distinct, and eosinophilic. The nuclear outline was smooth without nuclear folds. Mitotic activity was approximately 1 mitotic figure per 10 high-power fields in the most active areas. Occasional large zones of pink hyaline material surrounded and entrapped groups of the mononuclear and multinuclear histiocytes. Hemosiderin, when present, was scanty.

The biopsy specimen obtained in June 1988, revealed an almost identical cellular pattern, with most cells being mononuclear in form. Intracellular and extracellular hemosiderin appeared increased but still was found in a minority of the tumor volume. Mitotic activity remained low, with approximately 3 mitotic figures per 10 high-power fields in the most mitotically active areas. Although villiform structures were not apparent, occasional tissue clefting was seen. The nuclear morphologic features were similar to those seen in the specimen obtained in 1987, but there appeared to be a mildly increased coarseness of the nuclear chromatin pattern. The nuclear membranes remained smooth, and the nucleoli appeared identical to those found in 1987.

The material obtained at autopsy (following radiation therapy) revealed a high-grade spindle, polygonal, and giant cell neoplasm. The neoplasm grew in a solid or nodular pattern and invaded the surrounding soft tissues. The cells were predominantly mononuclear in form, with most being polygonal in shape. Approximately 10% revealed a “spindle cell” morphologic structure. The cytoplasm was moderate to abundant, eosinophilic, and granular. Multinucleated giant cells were present but in reduced numbers. The multinucleated giant cells had between 2 and 6 nuclei, with an appearance similar to that seen in the mononuclear cells. The nuclei of both cell populations were enlarged and hyperchromatic and frequently contained a coarse chromatin pattern (Figure 2). Chromatin clearing was prominent in some cells, as were irregularities of the nuclear membranes. Anaplastic nuclei were frequent. The mitotic rate was approximately 4 per 10 high-power fields in the most mitotically active areas. Although the neoplasm contained numerous capillary-sized vessels, clefted spaces and villiform structures were not apparent. Hemosiderin was focally prominent, both extracellularly and within the mononuclear cell component.

Material obtained in 1984 revealed a proliferation of mononuclear histiocytes and multinucleated histiocytic giant cells, expanding and distorting the synovial tissue and infiltrating into the subsynovial soft tissues of the joint. Most cells were polygonal and mononuclear in form with ovoid nuclei, occasionally showing clefting. Chromatin clearing was prominent, and 1 to 3 small but distinct nucleoli were present in each cell. Multinucleated giant cells were frequent, and the mitotic rate was approximately 10 per 10 high-power fields in the most mitotically active areas. The mitotic figures were all symmetrical in form. The multinucleated histiocytic giant cells had between 3 and 20 nuclei per cell. The nuclei were identical in appearance to those in the mononuclear cells. Hemosiderin was abundant both intracellularly and extracellularly (Figure 3). No nuclear anaplasia was identified. The cellular infiltrates were often subdivided by bands of spindle cells with a bland nuclear morphologic structure or by collagen bundles.

The 1988 and 1990 stump recurrences revealed nodules composed of a cellular infiltrate nearly identical to that seen in the 1984 specimen. Specifically, there did not appear to be an increase in nuclear atypia, and the mitotic activity was quantitated at between 2 and 4 mitotic figures per 10 high-power fields in the most mitotically active areas. Multinucleated giant cells appeared to be present in proportionately greater numbers. Some cells contained nucleoli larger and more prominent than in the 1984 specimen. The hemipelvectomy specimen obtained in December 1990 contained cellular infiltrates similar in appearance to the earlier specimens obtained from the surgical stump. The number of multinucleated giant cells appeared somewhat increased but were otherwise morphologically identical to those seen in earlier biopsy specimens. The mononuclear component was now dominated by short spindle cells that contained nuclei similar to those seen in the multinucleated giant cells. Mitotic activity was in the range of 3 mitotic figures per 10 high-power fields. Material obtained in 1991 from the right buttock mass revealed cellular infiltrates of both mononuclear and multinuclear cells. The mononuclear cells were predominantly polygonal, although spindle forms were easily found. The nuclei appeared enlarged, with clearing of the chromatin giving a vesicular pattern (Figure 4). Nucleoli were more prominent. Some nuclei contained membrane irregularities in addition to the previously seen smooth nuclear clefts. Mitotic activity was increased, with 8 mitotic figures present per 10 high-power fields in the most mitotically active areas. The mitotic figures remained symmetrical in form.

Five samples from 2 different patients (2 from patient 1 and 3 from patient 2) were analyzed by FISH. For each case, 100 nuclei were scored, and the FISH results were compared with normal control values obtained from normal synovial tissue (A. Meloni-Ehrig, PhD, unpublished data, 1997). Normal values for trisomy 7 are 0.8% ± 0.2%; normal values for trisomy 5 are 0.97% ± 0.3%; and normal values for trisomies 5 and 7 in the same nucleus are 0.3% ± 0.1%. These values were comparable with those obtained in the peripheral blood lymphocyte controls used in this study.

Patient 2 showed trisomy 7 in 8% to 11% of nuclei (Figure 5), whereas trisomies 5 and 7 together were seen in 1% of the nuclei. No nuclei with trisomy 5 alone were found. The first biopsy specimen showed 10% of nuclei to be +7, material from the second biopsy showed 11% to be +7, and the final biopsy tissue had 8% of nuclei +7.

Patient 1 showed trisomy 7 in 5% of the nuclei (Figure 6, A), trisomy 5 in 5% of the nuclei (Figure 6, B), and trisomies 5 and 7 together in 1% of the nuclei in the initial biopsy specimen, whereas the autopsy tissue displayed 8% trisomy 7, 1% +5, and 1% +5 and +7 together.

A number of purported malignant giant cell tumors or examples of malignant PVNS have been documented in the English-language literature,18–23 but many of these cases do not meet the criteria for diagnosis of malignant giant cell tumor of tendon sheath established by Enzinger and Weiss25 or as modified by Bertoni et al.23 Several of the reported malignant giant cell tumors most probably represent other types of sarcoma that contain a component of giant cells.20,22 Enzinger and Weiss25 require the coexistence of frankly malignant areas with a typical benign giant cell tumor or a malignant-appearing recurrence following treatment of a typical benign giant cell tumor before they accept a lesion as a bona fide example of malignant giant cell tumor. Although Enzinger and Weiss accepted the case reported by Carstens and Howell21 as an example of malignant giant cell tumor, they rejected the case reported by Nielsen and Kiaer.22 

Bertoni et al23 reported 8 cases of malignant PVNS and reviewed the literature. Three of their cases appeared to be secondary, arising in previously unremarkable PVNS, whereas 5 were apparently examples of primary malignant PVNS without antecedent lesions. He stipulated criteria for the diagnosis of malignant PVNS represented by the following: (1) a nodular solid infiltrative growth pattern, (2) large plump round or oval cells, (3) cells with large nuclei containing prominent nucleoli, (4) necrotic areas, and (5) absence of a zonal pattern of maturation.23 The findings in the first of our set of secondary lesions were similar except for an absence of significant necrosis. A nodular or solid invasive pattern was a prominent finding. Many of the cells had a plump eosinophilic appearance and a granular eosinophilic cytoplasm similar to that reported by Bertoni et al.23 In accordance with their criteria, we were unable to detect a maturational pattern in our cases. The mitotic rates in our cases were variable both between cases and between recurrences in the same patient, but mitotic rates were generally between 3 and 10 mitoses per 10 high-power fields in agreement with the rates reported by Bertoni et al.23 

Both our cases appear to qualify as examples of malignant giant cell tumor of tendon sheath (malignant PVNS) under the stringent criteria established by Enzinger and Weiss25 and the subsequent modified criteria developed by Bertoni et al.23 Both our patients' lesions were initially characterized as classic PVNS (giant cell tumor of tendon sheath, diffuse), which over a course of years displayed multiple recurrences, finally culminating in either a histologically malignant (case 1) or clinically malignant (case 2) lesion. We believe the buttocks mass in patient 2 represented a metastasis in that no direct extension between the stump recurrences and the right buttocks mass could be demonstrated by radiographic imaging studies or pathologic examination of the resection specimen. Both our patients underwent radiation therapy for treatment of recurrent or extensive PVNS. Unfortunately, pre–radiation therapy histologic specimens were not available for review in patient 2. The pre–radiation therapy specimens from patient 1 did not reveal significant differences from one specimen to another, nor were features present to distinguish them from classic PVNS. Specifically, there was no evidence of increased cellularity, nuclear atypia, or mitotic activity. Carstens and Howell21 reported a decrease in the number of multinucleated giant cells among clinically malignant forms of PVNS. There did not appear to be a decrease in the number of giant cells present in the material obtained before radiation therapy from our first patient. Subsequent to radiation therapy, the neoplasm in patient 1 had the appearance of a high-grade sarcoma similar in appearance to the neoplasm reported by Nielsen and Kiaer.22 Our example contained both anaplastic multinucleated giant cells and mononuclear forms. A storiform pattern was not found in either the material from our patient or in the case reported by Nielsen and Kiaer.22 

Although we were unable to review preradiation histologic sections from case 2, the initial recurrences subsequent to radiation therapy had the appearance of typical PVNS. The recurrences around the knee and at the resection stump did not demonstrate any evidence of anaplasia and were consistent with typical PVNS (diffuse giant cell tumor of tendon sheath). Although histologically similar, we consider the buttocks mass to represent a metastasis, since no direct extension from the amputation stump could be demonstrated either pathologically or radiographically. Although nuclear anaplasia was not noted in the buttocks mass (patient 2), the nuclei did appear slightly enlarged, with more prominent nucleoli than had been noted in the previous recurrences. In addition, mitotic activity was significantly increased over that seen in the previous biopsy and resection specimens. Thus, we believe this deposit represented a malignant giant cell tumor of tendon sheath based on both clinical behavior and morphologic findings.

It is of interest that both our patients had received radiation therapy at the primary site before the development of clinically aggressive neoplasms. In one case, the time between radiation therapy and the development of a clinically aggressive lesion was less than 10 years, whereas in the second patient approximately 40 years elapsed between radiation therapy and the appearance of a metastasis. Most postradiation sarcomas occur 10 or more years following radiation therapy.24,26,27 Nonetheless, radiation therapy may have had an etiologic role in the development of aggressive behavior in one or both of these neoplasms. Radiation therapy is known to induce karyotypic abnormalities, but the trisomy of chromosome 7 in patient 1 was present both before and after radiation therapy, suggesting that the abnormality is intrinsic to the neoplasm and not the result of therapy.

A variety of chromosomal abnormalities have been reported to be associated with tenosynovial giant cell tumors.14–17,28 Localized giant cell tumors of tendon sheath (nodular tenosynovitis), along with their diffuse counterparts (PVNS), appear to manifest reasonably characteristic translocations.15,16 Three of 11 reported cases contained a translocation, t(1;2)(p11;q35–36), and an additional case revealed a variant translocation, t(1;5)(p11;q22). Dal Cin et al15,16 discussed the frequency of involvement of the 1p11–p13 region and abnormalities in the long arm of chromosome 5. They believe that cytogenetic aberrations in both the localized and diffuse forms of giant cell tumor of tendon sheath were similar, with at least 2 cytogenetic groups being distinguishable.15,16 One set was characterized by rearrangements of the 1p11–p13 region and the other demonstrated abnormalities involving the 16q24 band.15 In a summary of the literature presented by Dal Cin et al,15 all reported cases of localized giant cell tumor of tendon sheath (nodular tenosynovitis) were diploid or hypodiploid. The 3 cases that demonstrated a triploid karyotype were examples of diffuse nodular tenosynovitis (PVNS)15 and had been reported previously by Ray et al,28 Fletcher et al,14 and Mertens et al.17 In 2 patients, the triploidy was due to +7, whereas in the third, the triploid karyotype was +12. In the example of PVNS reported by Ray et al,28 a second sample of the neoplasm demonstrated a karyotype of 48 +5 +7. Both of our cases had the clinical and pathologic appearance of diffuse giant cell tumor of tendon sheath (PVNS) and disclosed trisomies for chromosomes 7 or 7 and 5. Hence, we are in agreement with Sandberg and Bridge29 in that the karyotypes of giant cell tumor of tendon sheath and localized (nodular tenosynovitis) and diffuse (PVNS) giant cell tumor of tendon sheath are somewhat different, with trisomies of chromosomes 5 or 7 being characteristic of PVNS but not seen in the reported examples of localized nodular tenosynovitis. Our finding of trisomy 7 in the neoplasms from both our patients has interesting implications. First, the presence of trisomy 7 has not (to our knowledge) been reported in the localized form of giant cell tumor of tendon sheath but has been documented in 4 of 6 examples of PVNS where chromosomal data have been reported.15 Trisomy 5 is a less common abnormality, occurring in 2 of 6 patients with PVNS but absent in all studied examples of localized giant cell tumor of tendon sheath.

Second, the presence of rather characteristic karyotypic abnormalities in both the diffuse and localized variants of tenosynovial giant cell tumor supports the neoplastic as opposed to reactive or reparative nature of these lesions. All 11 tenosynovial giant cell tumors summarized by Dal Cin et al27 and the present 2 cases contain significant karyotypic abnormalities. As a group, tenosynovial giant cell tumors have cytogenetic aberrations divisible into 2 basic groups: one with rearrangements in the 1p11–p13 regions with either 2q35–36 and 5q22–5q35 involvement or an abnormality involving 16q24 band.15 In addition, most examples of PVNS demonstrated trisomies for chromosome 7. Unfortunately, we were unable to document the presence or absence of the translocations reported by Dal Cin et al15,16 because of the retrospective nature of our study, its restriction to paraffin-embedded material, and the use of FISH. Although initially believed to be nonneoplastic inflammatory reactions,1 other investigators using flow cytometric DNA analysis30 have proposed a neoplastic nature for PVNS. Abdul-Karim et al30 analyzed the DNA content and proliferative indices of 14 cases of diffuse giant cell tumor of tendon sheath and 11 cases of localized giant cell tumor of tendon sheath. Aneuploidy was documented in 3 cases of the diffuse type of giant cell tumor. They concluded that the presence of aneuploidy in some cases of diffuse giant cell tumor of tendon sheath, along with the reported occurrence of malignant transformation in rare lesions, supported a neoplastic character for these lesions. Our finding of trisomy 7 in both of our cases and trisomy 5 in one supports this interpretation.

Third, the presence and persistence of trisomy 7 in both our cases supports the interpretation that both malignant neoplasms are examples of malignant giant cell tumor of tendon sheath (malignant PVNS). In both cases, +7 persisted in a component of the cells throughout the disease despite the increasing aggressiveness of the lesions. The presence of trisomy 7 supports our interpretation of these lesions as examples of PVNS despite the lesions' aggressive courses. Although the diagnosis of malignant fibrous histiocytoma was a histologic consideration in patient 1, trisomy of chromosomes 5 or 7 is not characteristic of malignant fibrous histiocytoma.31 Hence, the chromosomal findings support our interpretation of the final neoplasm in patient 1 as a malignant example of PVNS. Also of interest is the apparent instability of trisomy 5 in the neoplasm in patient 1. The early biopsy material demonstrated 5% of the cells to be trisomic for chromosome 5, but the autopsy demonstrated only 1% of cells to show trisomy 5. A similar variability in the presence of trisomy 5 has been reported by Ray et al.28 Based on our findings of a persistent trisomy for chromosome 7, we agree with Bertoni et al23 that malignant PVNS is a true entity and represents the malignant counterpart of PVNS. In agreement with these authors, we believe that malignant PVNS (malignant giant cell tumor of tendon sheath) can arise in association with prior typical PVNS and that in a component of cases malignant PVNS may be difficult to differentiate from aggressive forms of typical PVNS. Finally, malignant PVNS is an extremely rare neoplasm that may be underrecorded when the criteria of Enzinger and Weiss25 are used.

We agree with the criterion defined by Bertoni et al23 for the diagnosis of malignant PVNS. Specifically, the presence of a nodular and solid invasive growth pattern coupled with large round or oval cells containing vesicular nuclei with prominent nucleoli and the absence of a zonal maturational pattern appears to define a set of these neoplasms. Necrosis was not a feature in our cases. The high-grade lesion that developed following radiation therapy in patient 1 revealed significant nuclear atypia and an increase in the proportion of spindle cells present. Further cases with well-documented histopathologic findings and clinical behavior are needed before concise criteria for the prediction of recurrence and metastasis in PVNS can be developed.