In a monumental paper written in 1931, Codman originally
described an "epiphyseal chondromatous giant cell tumor of the proximal
humerus."3 Although the diagnosis
was subsequently correctly changed to chondroblastoma of bone by
Jaffe and Lichtenstein in 19429,
a tumor located at that site in children is still known as Codman's
tumor2 (Fig. 1). As noted by
many authors, the tumor most frequently presents in adolescence
and seems to arise from secondary centers of ossification2,5,6,8,9,12,17,19. Most authorities
agree that the cell of origin arises from the epiphyseal plate or
some remnant of it2,8,9,17. The
lesion is rare, accounting for approximately 1 to 2 percent of all
benign bone tumors2,5,17,19.
Most patients with chondroblastoma present with pain, which is
often quite severe, and limitation of movement of the adjacent joint2,3,5,8,17,18,19. Radiographic and
special imaging studies demonstrate a destructive lesion of the
epiphyseal site, often with some expansion of the bone and a sclerotic
margin2,5,12,18,20. Within the
cartilaginous mass that contains the tumor, there is often punctate
calcification2,12,19,20. Histologically,
the lesion is quite distinct from other cartilage-containing tumors;
it shows well defined chondroblasts with benign-appearing nuclei,
multinucleated giant cells, chondroid and sometimes hyaline foci,
and a pattern of tiny calcific bodies that surround the cells and
hence have been named "chicken-wire calcification" (Fig. 2)2,5,9,17.
Treatment has been highly variable but currently usually consists
of curettage and packing with bone graft2,5,18,19.
A relatively high rate of recurrence (10 to 35 percent) has been
reported in the past, and metastasis has been described but is extremely
rare2,5,10,11,14-16. Risk factors
for these two forms of aggressive behavior (recurrence and metastasis)
have not been clearly defined.
The purpose of this article is to describe a series of forty-seven
cases of chondroblastoma, all diagnosed, treated, and followed by
our group, and to further define which factor or factors should
be considered as possible evidence of more aggressive behavior and
therefore conceivably dictate improved plans for management.
We reviewed the charts, imaging studies, and histological studies
of seventy-three patients with chondroblastoma who had been treated
by the Orthopaedic Oncology Service at the Massachusetts General
Hospital, Boston, Massachusetts, and Boston Children's Hospital
between 1977 and 1998. Of this group, twenty-six patients either were
followed for too short a period or had received a major part of
their care at another institution. We were able to obtain historical
data, imaging studies, histological findings, and adequate follow-up
by means of a personal visit or telephone communication to determine
the results of treatment for the remaining forty-seven patients,
who constituted our series. We identified the age at presentation,
gender, anatomical site, presenting symptoms, physical findings,
and treatment methods. By reviewing the imaging studies, we were
able to define the location within the bone (the epiphysis or the
metaphysis, or both), the size of the tumor, the presence or absence
of growth-plate closure, and the amount of cortical destruction.
The histological findings were reviewed, and the presence or absence
of an aneurysmal-bone-cyst component was particularly noted. Operative complications,
whether there had been a recurrence, current status in relation
to the tumor, and functional status were also determined for each
of the patients in the study. Recurrence rates were analyzed in
relation to all of the above parameters, and statistical analyses
with use of chi-square methods and power studies were performed.
The average age at presentation (and standard deviation) was
22 9.9 years (range, eleven to fifty-six years). Twenty-nine (62
percent) of the patients were between ten and twenty years of age. The
average duration of follow-up was 88 35 months (range, twenty-six
to 240 months).
Eleven tumors were located in the proximal part of the tibia
(Fig. 3);
ten, in the proximal part of the humerus (Fig. 4); and four each,
in the proximal part of the femur, the greater trochanter, the hemipelvis
(Fig. 5-A),
and the distal part of the femur. Two lesions each were located
in the calcaneus (Fig. 5-B), the talus, and the scapula, and
one each was located in the patella, the distal part of the radius,
the distal part of the humerus, and the proximal part of the fibula.
Twenty-nine of the thirty-two tumors in the long bones were in
the epiphyseal or apophyseal part, even when the epiphysis was closed.
Twelve of the epiphyseal tumors involved the metaphysis as well.
Only one of the lesions in the long bones was metaphyseal.
The physis remained open and presumably functional in twelve
patients, who had ages ranging from eleven to fourteen years. Five
tumors, all in the proximal part of the humerus or the proximal part
of the tibia, had violated the growth plate.
All forty-seven tumors had the typical histological appearance
of a chondroblastoma, and the histological features of the seven
recurrent tumors and the two that metastasized were indistinguishable from
those of the other tumors. Five tumors had violated the articular
cartilage. An aneurysmal-bone-cyst component was identified in eleven
of the tumors4,19 (Fig. 6).
It was possible to estimate the sizes of forty of the forty-seven
tumors, which averaged 4.5 centimeters (range, one to eleven centimeters)
in their longest diameter.
The primary treatment for these lesions varied and included curettage
and packing with allograft (seventeen tumors); curettage and packing
with polymethylmethacrylate (eight); curettage and packing with
iliac crest autograft (seven); curettage alone (five); marginal
resection (four); resection and allograft replacement in the proximal
part of the humerus (two), the distal part of the femur (one), and the
glenoid (one); curettage and fat implantation (one); and marginal
resection without packing (one). The seven recurrences were treated
with intralesional resection (two), curettage and packing with iliac
crest autograft (one), resection and allograft implantation (two),
curettage and packing with allograft (one), and radiofrequency ablation (one).
Complications of Surgery
The complications of surgery were not insubstantial. One patient
required another operative procedure after no diagnostic tissue
was obtained initially. In another patient, the hip became painful after
curettage and bone-grafting of a proximal femoral lesion and subsequently
required an osteotomy. One patient sustained an undisplaced subtrochanteric
fracture in a postoperative fall. Four of the six patients who had
osteoarticular allograft as well as three others in whom hardware had
been implanted required hardware removal several years after the
surgery. One scapular lesion and both talar lesions ultimately resulted
in osteoarthritis of the adjacent joint. One patient had a limb-length
discrepancy and angular deformity of the affected limb postoperatively.
Recurrence
Forty of the forty-seven tumors did not recur. The average duration
of follow-up for this group was 84 51.3 months (range, twenty-six
to 196 months), and the average age was 21 9 years (range, eleven
to forty-six years). Twenty-four of these patients were twenty years
of age or younger, and sixteen were older. Twenty-five patients were
male and fifteen were female. The average size of the lesion was
3.7 3.4 centimeters (range, one to ten centimeters). The growth
plate was documented to be open in ten patients. Five tumors involved
only the epiphysis, and five crossed the physis.
Seven of the patients had a recurrence. The duration of follow-up
averaged 106 72.4 months (range, thirty-two to 240 months). The
average age for this group was 23 15.7 years (range, eleven to
fifty-six years). Five of the seven patients were twenty years of
age or younger. There were four male and three female patients,
and the average size of the lesion was 5.3 4.6 centimeters (range, two
to eleven centimeters). Two patients had open physes, and both tumors
involved a secondary center of ossification. Three of the tumors
were originally treated with curettage and allograft; two, with
curettage and iliac crest autograft; one, with intralesional resection;
and one, with curettage alone. Four of the tumors recurred a second
time: two of these recurrences were treated with curettage and allograft,
and one each was treated with intralesional resection and radiofrequency
ablation. Only one of the four recurred a third time. The third
recurrence was treated with curettage and osteoarticular allograft,
and the tumor had not recurred again at five years.
With the small numbers available, there was no significant difference
between the nonrecurrent and the recurrent tumors with regard to
the age or gender of the patient, size of the lesion, duration of follow-up,
or method of primary or secondary treatment. Only one feature stood
out: there were no recurrences in the eight patients treated with implantation
of polymethylmethacrylate, whereas the tumor recurred in three of
the seventeen patients in whom allograft had been implanted and
in two of the seven in whom iliac crest autograft had been used.
Although these data are suggestive, they were not significant on
chi-square analysis because a power study indicated that the total
sample size necessary to arrive at a 95 percent confidence level
would be eleven patients treated with cement and thirty-four treated
with graft.
Another finding that could potentially influence the recurrence
rate was the presence of an aneurysmal-bone-cyst component, which
was noted histologically in eleven of the forty-seven tumors. Two of
these eleven tumors recurred, compared with five of the thirty-six
that did not have a cystic component. This small difference was
not significant.
The only significant difference between the nonrecurrent and
the recurrent tumors was related to their locations. The most common
sites of recurrence were the proximal part of the femur and the greater
trochanter; four of eight tumors in those locations recurred. One
of the four tumors in the pelvis also recurred, so a total of five
of twelve lesions around the hip recurred. This recurrence rate
differs markedly from that at all of the other sites, for which
the rate was two (6 percent) of thirty-five. This difference was
significant according to the Fisher exact test at the p < 0.03
level.
Current Status of the Patients
Metastases developed in two patients in our series. A twenty-nine-year-old
man presented with a pelvic lesion in 1981, which was treated by
intralesional resection. The tumor recurred locally three years
later and was again treated by resection. A second recurrence two
years later was treated by wide resection. In 1994, thirteen years
after the primary tumor was discovered, the patient had had metastases
to the sternum, multiple long-bone sites, and the lungs, and he
died of the disease one year later. The metastatic lesions were
identified histologically as chondroblastoma not unlike the primary
tumor of the pelvis. The second patient was initially seen, at another
center, at the age of forty-nine years with a chondroblastoma of
a rib, and a resection was performed. A recurrence was noted one
year later, and it was also resected. Six years later, at the age
of fifty-six years, he presented to our institution with a chondroblastoma
of the proximal part of the femur and he subsequently had an additional
recurrence at that site as well as metastases to the scalp and neck
and later to the pubis. All of the soft-tissue lesions were excised, and
the pubic lesion was treated with radiofrequency ablation. At the
time of writing, the patient was disease-free but was still being
followed closely.
At the time of the most recent follow-up, thirty-nine of the
remaining forty-five patients stated that they were pain-free and
had only minimal functional limitation. However, the two patients
who had had a talar lesion had peritalar arthritis. Two of the patients
with proximal femoral disease had a painful hip, and one of them
had had an osteotomy for the treatment of malalignment that resulted from
a pathological fracture through the lesion. One patient with a glenoid
lesion had a painful glenohumeral joint with limited motion, and
one patient with a pelvic lesion was receiving disability compensation
because of pain of unknown cause.
We are reporting on a large series of patients with chondroblastoma
who were treated operatively by one group of surgeons. Most of our
data coincide with the results of other studies1,5-8,12,13,17-20.
For example, male patients were affected more frequently than female
patients, and, except for three, all of our patients presented with
pain. The two most frequent sites of involvement were the proximal
part of the humerus and the proximal part of the tibia, as has been
reported in other series5,17,19.
All but one of the long-bone lesions in our patients arose from
a secondary center of ossification. One lesion was primarily metaphyseal,
a rare occurrence1,7,13. Four
of the lesions arose in the talus or calcaneus7.
Our patients were older than those in previous reports1,8,17,19. In our series, 62 percent
of the forty-seven patients were twenty years of age or younger,
whereas 82 percent of seventy patients were twenty years or younger
in the study by Springfield et al.19 and
73 percent of 231 were twenty years or younger in the composite
series reported by Huvos and Marcove8.
The average age at the time of presentation in our series was twenty-two
years, and, as might be expected, we had a lower percentage of patients
with open growth plates compared with the rates in other reports.
Furthermore, patients with a lesion that had not occurred in a long
bone tended to be older than those with a long-bone lesion. It is
unclear why our patient population was older, but the reason may
be related to the referral pattern for our center.
We attempted to identify risk factors for recurrence of chondroblastoma,
the rate of which was lower in our series than in those reported
by others2,5,8,17,19. The ages
of the patients with and without recurrence were similar. Although
the finding was not significant, the lesions that recurred were
on the average larger than those that did not (5.3 compared with
3.7 centimeters). Open growth plates have been considered to be
a risk factor for recurrence2,4,5,19.
It has been hypothesized that tumors in patients with open growth
plates are more likely to recur secondary to less aggressive surgical
curettage due to fear of injury to the physis19.
However, in our study, patients with open growth plates did not
have a significantly increased rate of recurrence. It is believed
that 20 percent of chondroblastomas have an aneurysmal cystic component,
which has been described as a precursor to recurrence4,5. This was not observed in our series.
The only factors that seemed to affect the recurrence rate were
anatomical location and, to a lesser extent, the method of treatment.
Five of twelve lesions in the proximal part of the femur, the greater trochanter,
or the pelvis recurred, and one of the tumors in the pelvis and
another that developed first in a rib and then in the pelvis metastasized. One
possible explanation for this phenomenon is difficulty with gaining
access to these lesions because of the proximal femoral and pelvic
anatomy combined with the surgeon's concern about compromising the
blood supply to the femoral head. However, this anatomical site
is known for harboring more aggressive lesions, particularly cartilage tumors2,18, and that may be the reason for
the higher recurrence rate.
Although the finding was not significant, perhaps because of
the small number of patients, the recurrence rate in our series
varied among the different treatment methods. Intralesional curettage
was associated with a higher recurrence rate (one of five) than
was marginal resection (none of four). Curettage and packing with
bone graft was associated with a higher recurrence rate (21 percent)
than was curettage and packing with polymethylmethacrylate (0 percent).
The heat of polymerization of the cement may destroy residual tumor
after curettage. On the basis of this finding, we now recommend
that polymethylmethacrylate be chosen over bone graft to fill defects
after removal of worrisome or recurrent lesions.
Two patients had metastases, and one died. One case involved
the pelvis, a site known to harbor more aggressive lesions. The
second case had some unusual characteristics, including older age at
presentation (forty-nine years), involvement of a rib (known to
be an uncommon site13), and the
development of a metachronous lesion some years later.
On the basis of this study, we concluded that although chondroblastomas
have a substantial recurrence rate and their juxta-articular nature
can lead to arthritic changes secondary to treatment, surgery that
is more aggressive than curettage and packing with polymethylmethacrylate
is not warranted for classic chondroblastomas and even local recurrences
should be treated in a similar fashion.