Clinical Features (see Appendix)
The study included nineteen female patients (66%) and ten male patients
(34%); the gender of one patient was unknown. The average patient age at the
time of presentation of the first tumor was twenty-one years (range, eleven to
sixty-two years). Seventeen (59%) of the patients were younger than twenty
years of age; thirteen of the seventeen were less than sixteen years of age,
and thirteen were female. Twenty-eight percent (eight) of the patients in the
series were in the third decade of life, 10% were in the fourth decade, and
only one patient was older than forty years of age at the time of the initial
presentation. A total of ninety-four tumors were identified in the thirty
patients, who had an average of three tumors (range, two to nine tumors) each.
Thirteen patients (43%) had only two tumors. Eleven patients (37%) had
synchronous tumors, and seven of them (23% of the series) had multiple tumors
discovered at the initial presentation. The average time between the diagnosis
of the first tumor and the discovery of the second in the nineteen patients
with metachronous tumors was 5.3 years (range, one to sixteen years). In six
patients, a metachronous lesion developed within two years after the first
tumor. The maximum time span over which any one patient had the development of
two or more tumors was twenty-three years.
Location
Most tumors arose in long bones; 31% of the tumors developed around the
knee, with almost equal involvement of the distal part of the femur and the
proximal part of the tibia (Fig.
1). Eight patients (27%) had ipsilateral involvement of the distal
part of the femur and the proximal part of the tibia, and in six of them these
were synchronous tumors. None of these tumors extended across the knee joint.
The next most frequent locations were the distal part of the radius (10%),
proximal part of the humerus (8%), distal part of the tibia (8%), distal part
of the ulna (7%), proximal part of the femur (5%), and proximal part of the
fibula (4%). Six tumors involved vertebral bodies, and five of them were
localized to the cervical spine. Eight tumors in seven patients occurred in
the bones of the hands and feet. One of these patients had involvement of two
bones of the same hand over a nine-year period. Few tumors arose in the pelvic
bones, and the ribs and skull bones were rarely involved. Rarely, more than
one tumor was discovered in a single bone; one patient had three separate
lesions involving the left ulna over a twelve-year period.
Radiographic Findings
Tumors in the long bones usually manifested as an expansile lytic mass
extending from the metaphysis through the closed growth plate and into the
subchondral region of the epiphysis (Figs.
2 and
3). The lesion margins were
generally well-defined (a narrow zone of transition) and lacked sclerosis. The
cortices were either expanded or destroyed by tumor, which frequently extended
into the adjacent soft tissues. Twelve tumors were either
diaphyseal-metaphyseal or metaphyseal in location. Eight of these were in
skeletally immature patients with open growth plates, and seven exhibited
prominent sclerosis and mineralization, which caused them to mimic a
fibro-osseous lesion such as fibrous dysplasia. The distal part of the femur
was the most common location for such unusual lesions (Figs.
4 and
5).
Tumors of the metacarpals, metatarsals, and phalanges usually were purely
lytic and extended to the end of the bone. Tumors arising in other small bones
of the hands and feet and those in the pelvis, skull, and vertebrae presented
as nonspecific lytic lesions.
Histologic Findings
All of the multicentric giant cell tumors contained foci that exhibited the
morphologic features diagnostic of giant cell tumor. These areas consisted of
sheets of uniform oval to polygonal mononuclear cells admixed with numerous
large osteoclast-like giant cells. The nuclei of the mononuclear and giant
cells were morphologically similar to one another and were round to oval with
fine to slightly vesicular chromatin and with distinct nucleoli. The giant
cells frequently contained more than twenty nuclei. The cytoplasm of the
mononuclear cells was eosinophilic and the cell membranes were not
well-defined, creating a syncytial appearance
(Fig. 6). In some areas, the
mononuclear cells were spindle-shaped, resembling fibroblasts, and were
enmeshed in a collagenous stroma that also contained scattered aggregates of
foamy histiocytes. The spindle cells were frequently arranged in a storiform
pattern and were most prominent in the periphery of the tumors. Nine patients
had tumors with extensive fibroblastic and fibrohistiocytic regions, and one
of those patients had tumors involving the distal part of the femur and the
patella that were composed entirely of the fibrohistiocytic component
(Fig. 7). One tumor exhibited
limited cytologic atypia in which the round and oval mononuclear cells had
enlarged hyperchromatic nuclei with irregular nuclear borders. These atypical
cells were only focally present within the tumor, were not mitotically active,
and often were associated with areas of hemorrhage. Because of these features,
such changes were considered to be degenerative in nature. This tumor did not
recur or metastasize following curettage.
In general, the mononuclear cells, regardless of their morphology,
exhibited occasional mitoses; mitotic figures were numerous in only a minority
of the tumors. Other secondary findings in the tumors included aneurysmal bone
cyst-like changes, which were present in nine tumors from six patients, and
reactive bone formation, which was present in most tumors but was especially
prominent in areas associated with fibrohistiocytic or aneurysmal bone
cyst-like changes. Tumors that resembled a fibro-osseous lesion
radiographically were among the tumors with prominent fibrohistiocytic
components and abundant reactive bone with or without aneurysmal bone
cyst-like changes (Figs. 8-A
and 8-B).
Treatment and Recurrence
The majority of tumors (62%), particularly those involving long bones or
bones of the hands and feet, were initially treated with curettage and
bone-grafting. A tumor in the proximal part of the tibia was treated with
radiation therapy and curettage, and a tumor in the acetabulum was treated
with cryotherapy and curettage. Methylmethacrylate was used following
curettage of one tumor of the proximal part of the tibia. Use of adjuvant
therapies was not documented for any of the remaining tumors treated with
curettage. Twenty-one tumors (22%) were initially treated with wide excision.
These lesions generally involved bones that could be sacrificed without a
substantial loss of function, such as the distal part of the radius or the
proximal part of the fibula. Two tumors in the cervical spine were treated
with radiation therapy alone. Two sphenoid tumors were treated with partial
excision followed by adjuvant radiation therapy, and one skull tumor was
treated with partial excision, adjuvant radiation therapy, and chemotherapy.
One tumor in the left lateral cuneiform bone was initially treated with
below-the-knee amputation, possibly because it was thought to be a more
aggressive tumor on the basis of clinical findings.
Patients were generally followed with routine periodic plain radiography of
the affected bone to look for tumor recurrence. Twenty-four (26%) of the
ninety-four tumors recurred, and twenty of the thirty patients had recurrence
of at least one tumor. The average time to recurrence was 2.5 years (range,
two months to nine years), and twelve patients had a recurrence in less than
two years. Fourteen patients had only one recurrence of a single tumor. One of
these tumors was a C2 vertebral lesion that had been treated initially with
radiation therapy alone. Another was a distal ulnar tumor that had been
treated initially with excision; it recurred in the soft tissues. The
remaining twelve tumors were initially treated with curettage, and all but one
involved a long bone. In three other patients, one tumor recurred multiple
times following the initial curettage. These cases included five recurrences
of a C2 vertebral tumor (Case 10; see Appendix), each treated with curettage
over a thirteen-year period; two recurrences of a distal radial tumor (Case
11), one treated with curettage with wrist fusion and the subsequent
recurrence treated with excision over a seven-year period; and two recurrences
of a metatarsal tumor (Case 26), one treated with curettage and the subsequent
recurrence treated with amputation over a five-month period. Multiple tumors
of long bones recurred in three patients (Cases 2, 6, and 23). All but one of
these recurrences were treated with either excision or amputation. One of
these patients (Case 23) had a soft-tissue recurrence of a distal femoral
lesion following an initial curettage and had multiple recurrences of an
ipsilateral proximal femoral tumor following several curettage procedures.
None of the three tumors treated with adjuvant cryotherapy or
methylmethacrylate packing recurred.
Seven patients with a total of eighteen tumors for which there were
follow-up data had no recurrences. The follow-up period for these patients,
beginning from the discovery of the initial tumor, ranged from one year and
two months to sixteen years (average, 7.7 years). These tumors involved a
variety of bones, including long bones, bones of the hands and feet, pelvic
bones, and vertebrae. All of these tumors were initially treated with
curettage or excision, except for one tumor of the left lateral cuneiform,
which was treated with amputation.
No follow-up data were available for three patients who presented with
synchronous tumors (Cases 18, 20, and 27; see Appendix).
Outcome
Follow-up data beginning from the discovery of the most recent tumor or
recurrence were available for twenty of the thirty patients. The follow-up
period ranged from two months to thirty-five years (average, six years). Other
than a patient who died of osteosarcoma (as described below) and a patient
with pulmonary metastasis who died of influenza, these patients had no
additional recurrences or new tumors.
Metastatic disease to the lungs developed in three patients (Cases 11, 13,
and 16; see Appendix). The time from discovery of the initial giant cell tumor
to the detection of the pulmonary metastases ranged from five to eight years
(average, 6.3 years). These patients had a total of eight tumors, in the long
bones, pelvic bones, and ribs; five tumors were treated with curettage, and
three were excised. All three patients had had a local recurrence of one of
the tumors; one of those recurrences was in the soft tissues and followed
excision of a distal ulnar tumor. In all three patients, the initial pulmonary
metastatic tumor was excised and histologically confirmed to be giant cell
tumor. One patient had new pulmonary metastases two years after excision of
the original tumor. These lesions remained stable until the patient died of
influenza six years later. Another patient was disease-free at the time of
follow-up, 1.5 years after the excision of the lung metastasis. The third
patient was lost to follow-up. None of the patients had metastatic disease
beyond the confines of the lungs.
Osteosarcoma developed in one patient thirteen years after the diagnosis of
multicentric giant cell tumor. This patient had presented, at the age of
thirty-five years, with synchronous tumors involving the distal part of the
femur and the proximal part of the tibia, which were treated with excision of
the femoral tumor and curettage of the tibial lesion. No other tumors or
evidence of metastatic disease were documented at the time of the initial
presentation. The patient was clinically stable until thirteen years later,
when he presented with additional synchronous tumors involving the distal part
of the ulna, the second phalanx, and the proximal part of the fibula. The
fibular lesion showed areas of giant cell tumor and high-grade osteosarcoma.
The patient died one year later of metastatic osteosarcoma to the lungs. In
another patient, an intermediate-grade fibrosarcoma developed in a recurrent
tibial tumor.
Multicentric giant cell tumor is rare and accounts for <1% of all cases
of giant cell tumor of
bone3-10,12-35.
We believe that we have reported on the largest series of multicentric giant
cell tumors to date, as most published reports on this condition have been
single case
descriptions17,19,27,28,31,33,34
or studies of small
series22,25,32.
In our cohort, female patients outnumbered male patients by a ratio of
almost two to one. This gender distribution is similar to that previously
reported in the literature for both
multicentric22,25,32,33
and solitary giant cell
tumor4,9,13,35,38,39.
The peak incidence of solitary giant cell tumor is in the third to fifth
decades of life, with approximately 80% of patients being older than twenty
years of age at the time of
diagnosis4,13,35,39.
In our series, the average age at presentation was twenty-one years, and 59%
were younger than twenty years of age. Other investigators have also noted
that patients with multicentric giant cell tumor are considerably younger than
those with a solitary giant cell
tumor22,25,32;
however, this has not been a consistent
finding33. One of
our patients was a girl who presented with tumors at the ages of eleven and
twelve years; to our knowledge, she was the youngest patient with multicentric
giant cell tumor reported to date.
Eleven (37%) of our patients presented with synchronous tumors. Six of
these cases involved the knee (the distal part of the femur and proximal part
of the tibia of the same limb). The prevalence of synchronous tumors and their
predilection for the knee region have not been previously recognized, to our
knowledge. Interestingly, we found that patients who had multicentric giant
cell tumor generally had it relatively early in the course of the disease, as
56% of our patients had more than one tumor at the initial presentation or had
the development of additional tumors within two years after the discovery of
the first giant cell tumor. However, there may be a lag time of as long as
sixteen years between the first and second tumor, and additional lesions can
develop for as long as twenty-three years. Overall, patients with multicentric
giant cell tumor generally have two or three lesions, but in our experience
patients have had as many as nine. At least one patient with ten tumors has
been described in the
literature40.
The anatomic distribution of multicentric giant cell tumors in this series
(Fig. 1) is similar to that of
solitary giant cell tumors, with a few
exceptions4,8,13,35,38,39.
Most of the tumors in our patients arose in the long bones of the lower
extremity, predominantly around the knee, followed by the proximal part of the
humerus and the distal part of the radius. Previous reports have emphasized
the high frequency at which multicentric giant cell tumor affects the
hands12,22,33.
In fact, in some series, as many as 20% to 39% of cases of multicentric giant
cell tumor have had at least one focus in the hand and 18% of patients with
giant cell tumor affecting the hand have had polyostotic
lesions12,32,33.
However, a compilation of some investigators' work has shown that only 0.4% to
3.9% of multicentric giant cell tumor lesions affected the
hands4,12,22,25,27.
In our series, four tumors of the hand (4%) and four tumors of the foot (4%)
occurred in three patients (10%) and four patients (13%), respectively. This
frequency of hand and foot involvement is similar to that reported by the
aforementioned investigators and is greater than the 2% frequency associated
with solitary
tumors4,8,12,13,35,38,39.
None of the tumors in our series arose in the sacrum, a site where 9% of
solitary tumors
originate38.
In most cases, individual lesions of multicentric giant cell tumor have
radiographic features that are essentially indistinguishable from those of a
solitary giant cell tumor. In long bones, multicentric giant cell tumors
generally present as eccentric lesions predominantly involving the metaphysis
and epiphysis with extension into the subchondral region of the bone (Figs.
2 and
3). Interestingly, our group
included eight patients with a total of twelve tumors (13% of all tumors) that
were restricted to the metaphyseal or diaphyseal-metaphyseal regions of long
bones (Fig. 4). The distal part
of the femur was the most common location for these unusual lesions. This is
the same anatomic site as that of solitary lesions that are confined to the
metaphysis13.
Tumors limited to the metaphysis account for 1% to 9% of all solitary giant
cell
tumors13,35.
In a previous report, metaphyseal lesions accounted for five (14%) of
thirty-five cases of multicentric giant cell
tumor25 (two of
these patients, who had a total of three tumors, were included in the present
series). The age range for our patients with metaphyseal lesions was twelve to
fifteen years, and all but one patient was female. In our series, four of six
patients with radiographically confirmed open growth plates had metaphyseal
lesions—that is, half of the patients with metaphyseal lesions were
skeletally immature as shown by imaging studies. Although the actual number
may be smaller, 5.7% of solitary giant cell tumors occur in skeletally
immature
patients41. In the
series reported by Kransdorf et
al.41, 96% of
solitary giant cell tumors in skeletally immature patients were predominantly
confined to the metaphysis, and others have demonstrated that metaphyseal
giant cell tumor can cross at least a partially open
physis42. As our
series demonstrates, lesions of multicentric giant cell tumor can be confined
to the metaphysis in either skeletally mature or immature patients and
metaphyseal lesions occur more frequently in patients with multicentric giant
cell tumor than in those with solitary giant cell tumor.
A small minority of tumors exhibited sclerosis and evidence of
mineralization, unusual radiographic features for a conventional giant cell
tumor, suggesting a fibro-osseous or bone-forming tumor (Figs.
3 and
4). The histologic correlate to
the sclerosis was abundant reactive bone that was prominent in
fibrohistiocytic and aneurysmal bone cyst-like areas (Figs.
7 and
8-A,
8-B). Such lesions in a patient
with polyostotic disease can result in diagnostic confusion, particularly with
entities such as fibrous dysplasia. This finding emphasizes the need for
histologic examination to establish the diagnosis of multicentric giant cell
tumor.
In general, multicentric giant cell tumor is histologically
indistinguishable from solitary giant cell tumor
(Fig. 6). All of our cases had
areas diagnostic of giant cell
tumor2,38;
however, some also contained fibroblastic and fibrohistiocytic areas, which
were usually only a minor component of the tumor and tended to surround
regions of classic giant cell tumor (Fig.
7). This same fibrohistiocytic proliferation is an accepted
component of solitary giant cell tumor, in which it is seen in varying
amounts. In one of our patients with multicentric giant cell tumor, however,
the fibrohistiocytic proliferation dominated the histologic picture of tumors
in the proximal part of the femur and the proximal and distal parts of the
tibia. This patient also had lytic lesions in the distal part of the femur and
in the patella that were composed solely of the benign fibroblastic and
fibrohistiocytic components. Such lesions have been reported in other patients
with multicentric giant cell
tumor22 and may
represent giant cell tumor in a stage of degeneration or involution in which
the diagnostic areas are obscured by the fibrohistiocytic tissue. One case
exhibited focal cytologic atypia with a benign clinical course. This type of
atypia is probably of a degenerative or "pseudoanaplastic" nature
and has been described in other benign bone
tumors43.
The overall tumor recurrence rate of 26% in our series was similar to that
reported in the literature for solitary giant cell
tumors13,39.
Of the patients for whom follow-up information was available, 74% had
recurrence of at least one tumor. It appears that the single most important
factor related to the risk of recurrence in our patients was the
incompleteness of surgical removal, as the recurrence rate was highest (37%)
for lesions that had been initially treated with curettage. In comparison,
wide excision was associated with a recurrence rate of 5%. It is important to
note that adjuvant thermal or chemical therapies were not used for the vast
majority of patients in whom the multicentric giant cell tumor was treated
with curettage; this accounts for the slightly higher recurrence rate seen
with these tumors as compared with the recurrence rates of 8% to 34% seen with
solitary giant cell tumor for which adjuvant therapies were
utilized44.
Soft-tissue recurrence, which can occur many years after initial excision of a
solitary giant cell
tumor45, developed
in two of our patients. In general, in our series, recurrences were treated
with more aggressive surgical removal, including amputation; however, repeat
curettage was effective in controlling some tumors. Three patients had tumors
that were followed clinically without surgical intervention, and all of those
tumors remained stable. Radiation therapy, although reserved for lesions that
were not amenable to surgical removal, was generally ineffective in
controlling the tumors.
Pulmonary metastasis of benign giant cell tumor is a well-documented and
recognized phenomenon that occurs in <2% of patients with a solitary giant
cell
tumor35,46-49.
To our knowledge, pulmonary metastases have not been previously reported in
patients with multicentric giant cell tumor. Pulmonary metastases developed in
three (10%) of our patients, at an average of 6.3 years after the discovery of
the initial giant cell tumor. Combining our series of patients with those
previously
reported2-5,7-9,17,19,22,25,27,28,31-34,36,
we determined that metastatic disease has developed in approximately 4% of
patients with multicentric giant cell tumor. In our patients, the excised
metastatic tumors had histologic features of conventional giant cell tumor and
the patients had an indolent course similar to that of patients in whom a
solitary giant cell tumor metastasizes.
Malignant transformation of giant cell tumor can be defined as a sarcoma,
usually a fibrosarcoma or osteosarcoma, juxtaposed to areas of typical giant
cell tumor or arising at the site of a previously treated benign giant cell
tumor4,6,50,51.
Up to 5% of solitary giant cell tumors undergo such malignant change, which
may occur as many as twenty-five years after the primary
treatment50,51.
The overwhelming majority of these malignant transformations are associated
with previous radiation therapy. Two of our thirty patients had malignant
transformation with no previous radiation.
The clinical and radiographic similarities between solitary giant cell
tumor and multicentric giant cell tumor suggest that the lesions of
multicentric giant cell tumor arise independently, rather than being multiple
sites of metastatic disease that develop from a single tumor. Specifically,
individual lesions of multicentric giant cell tumor demonstrated
characteristic radiographic features of giant cell tumor (e.g., a
circumscribed lytic epiphyseal-metaphyseal lesion of a long bone), rather than
a destructive pattern of a metastasis. This characteristic is similar to that
of other polyostotic bone diseases in which separate lesions exhibit the
radiographic features of the solitary form. Furthermore, the biological
behavior of any given lesion in a patient was independent of the behavior of
the other lesions, with the outcome being determined by how each tumor was
treated. Disseminated metastatic disease did not develop even in patients with
numerous skeletal tumors. The tendency of multicentric giant cell tumor to
affect younger patients suggests that there may be a germ-line genetic
abnormality that predisposes them to the development of multiple tumors.
However, to our knowledge, familial forms of multicentric giant cell tumor
have not been reported.
In summary, multicentric giant cell tumor is rare and most commonly affects
long bones, particularly those around the knee. It tends to occur in younger
patients and frequently manifests as synchronous lesions. In addition, lesions
of multicentric giant cell tumor may have an unusual metaphyseal location and
have characteristics of a fibroosseous lesion on imaging studies.
Histologically, such tumors have abundant fibrohistiocytic and fibroblastic
areas as well as reactive woven bone formation. Virtually all tumors have
areas with typical histopathologic features of giant cell tumor. As is the
case with a solitary giant cell tumor, the most aggressive behavior of the
vast majority of multicentric giant cell tumors is local recurrence, although
there have been rare cases of metastasis to the lungs as well as of malignant
transformation. Because a variety of other primary bone lesions may also have
a polyostotic presentation, the correct diagnosis relies on correlation of
clinical and radiographic findings with confirmation of the diagnosis by
histopathologic examination.