In 1927, Kolodny1 was the first to describe a bone tumor presenting as a "giant cell variant." A few years later, Codman2 used the term "epiphyseal chondromatous giant cell tumour." The term benign chondroblastoma was proposed by Jaffe and Lichtenstein in 1942 to describe this rare, benign, and distinctive tumor composed of immature chondroblasts with a scant chondroid matrix3. This benign cartilaginous lesion represents 1% of all primary bone tumors and appears to arise from secondary centers of ossification4-7 such as epiphyses and apophyses. Chondroblastomas most commonly originate from the epiphyses of long bones, particularly from the epiphyses of the proximal and distal parts of the femur, the proximal part of the humerus, and the proximal part of the tibia7,8.
This benign cartilaginous lesion most commonly occurs during childhood or adolescence. Clinical symptoms include pain, local tenderness, swelling, and limited motion of the adjacent joint4-7. Rarely, an asymptomatic chondroblastoma is discovered incidentally on a plain radiograph.
Surgery is recommended for the treatment of chondroblastoma6-8. The gold standard for surgical treatment is complete and meticulous curettage of the lesion followed by bone-grafting8.
The reported rates of local recurrence after surgical treatment have varied considerably, ranging from 10% to 35%8. The factors that are predictive of local recurrence have not been well defined and remain controversial. The presence of an aneurysmal bone cyst in association with the chondroblastoma was thought to be associated with local recurrence4,9, but this has not been confirmed in recently published series10. It also has been suggested that certain locations (the proximal part of the femur and the pelvis)8 or a young age7,11 at the time of diagnosis may be associated with a higher risk of local recurrence. Metastasis has been described but is extremely rare12.
To our knowledge, there have been no published series exclusively focusing on cases of chondroblastoma in a pediatric population, with the bone lesions being diagnosed and treated while the patients had active open physes. Epidemiologic characteristics and predictors of local recurrence of chondroblastoma during growth are not known. Patients with partially closed or closed physes were not included in the present study.
The objectives in the present retrospective study were to analyze the epidemiologic characteristics of chondroblastoma in children and to identify the factors that may influence local recurrence after surgical treatment. Functional outcome was also evaluated.
We performed a retrospective multicenter cohort study of eighty-seven children with histologically proved benign chondroblastoma of bone who were managed between 1950 and 2005. All eighty-seven children who were included in the study had open physeal plates at the time of diagnosis and treatment. Patients were managed in the Departments of Pediatric Orthopaedic Surgery affiliated with SOFOP (the French Pediatric Orthopaedic Society). The present study was approved by the local institutional review board.
Medical charts, radiographs, and histological studies were reviewed for each patient. The age at the time of presentation and the sex of the patient were identified, and the surgical and pathological records were reviewed. The initial clinical symptoms and the findings of the first clinical examination were noted. The intervals between the first symptoms and the times of diagnosis and recurrence (if applicable) were recorded.
Radiographs that were made at the time of presentation were reviewed for the location of the lesion in relation to the physes, and the activity of the lesion was classified as latent, active, or aggressive with use of the system of Springfield et al.7. Latent lesions were defined as being confined to bone, with a well-defined, intact reactive sclerotic rim surrounding the radiolucent chondroblastoma. Active lesions were confined to bone but had an incomplete sclerotic rim or were contained within a margin of reactive periosteal bone. Aggressive lesions had a poorly defined edge, with minimal or no intraosseous reaction and an extraosseous component that was not surrounded by periosteal bone.
For patients with more than twenty-four months of follow-up, the chondroblastomas were grouped on the initial radiographs as being close to more-active physes (that is, close to the knee or far from the elbow) or close to less-active physes (that is, close to the elbow or far from the knee) as the level of activity of the growth plates may influence the event of recurrence. Growth in the limbs has been studied extensively, and previous authors have agreed that 70% of the growth of the femur occurs in the distal physis and almost 60% of the growth of the tibia occurs in the proximal physis13. For the upper limb, 80% of the growth of the humerus occurs at the proximal physis and 80% to 85% of the growth of the forearm occurs at the distal physes of the radius and ulna14.
Histologically, a typical chondroblastoma shows proliferation of two different types of cells: mononuclear cells and giant cells. The mononuclear cells have well-defined cytoplasmic boundaries and an oval-to-elongated nucleus with a characteristic longitudinal groove, producing a "coffee bean" appearance. Cartilage matrix calcification with a "chicken wire" pattern is typically observed. Chondroblastoma cells also express protein S-100. Despite the fact that other features occasionally can be seen in a classic chondroblastoma, we are not aware of any cases in which mitotic activity, tumor necrosis, or nuclear atypia have been described. Histological analysis was done with special emphasis as to whether there was evidence of an associated cystic component.
The diagnosis of local recurrence was always confirmed on the basis of histological analysis of the lesion. When recurrence was diagnosed, the location, the type of treatment performed, and the pathological findings were recorded. Epidemiologic, radiographic, and surgical data from the recurring chondroblastomas were analyzed separately in an attempt to identify factors leading to recurrence.
A functional outcome scale was developed from the information available in the charts at the time of the latest follow-up. The result was classified as good if the patient was pain-free and had unrestricted activities (including sports), normal joint mobility, and intact joint surfaces radiographically (radiographic scars from surgical treatment did not exclude patients from this category). The result was classified as fair if the patient had unrestricted activities but had occasional pain (less than once per week), asymmetric joint mobility, a limb-length discrepancy (due to partial or complete epiphyseodesis), and/or radiographic evidence of loss of sphericity of the joint surfaces or of morphological modifications such as coxa vara or varus or valgus deformity. The result was classified as poor if the patient had a modification of his or her daily activities for any reason, such as chronic pain, deformity, or limp.
Statistical Analysis
Recurrence of chondroblastoma after surgical treatment was the end point studied. Categorical variables that initially had more than two classes were merged into two classes. Tumor location was separated into two categories: (1) close to the most-active physis or (2) far from the most-active physis. Univariate associations between predictors and the recurrence of chondroblastoma were evaluated with the chi-square test, Fisher exact test, or Student t test, as appropriate. A log-rank test was also used to deal with time-dependent data.
Multiple logistic regression was used to find independent predictors of recurrence. Covariates with a significance level of <0.2 were entered in the multiple regression analysis. Forward stepwise variable selection was performed with a selection procedure based on a significance value of 0.1. The goodness of fit and consistency with logistic function were evaluated with the Hosmer-Lemeshow and C.C. Brown tests, respectively.
Source of Funding
There was no external funding source for this study.
The study group included fifty-three boys and thirty-four girls. The mean age of the patients at the time of presentation was 12.5 years (11.5 years for the girls and 13.5 years for the boys). The presenting symptoms were pain in seventy-three patients (84%), limping in nine (10%), and local swelling in three (3%). An asymptomatic lesion was discovered on a radiograph made for another reason in two patients (2%). The mean delay between the onset of symptoms and diagnosis was 6.5 months (range, 0.5 to thirty-six months). In eleven patients, the duration of symptoms was more than one year.
The chondroblastoma was located within the epiphysis in fifty-nine patients (68%) (Fig. 1), in the metaphysis in four (5%), and in an apophysis (the greater trochanter) in two (2%). In the remaining twenty-two patients (25%), the chondroblastoma was metaphyseal-epiphyseal because it crossed the physis as seen radiographically.
The most commonly involved anatomical site was the proximal part of the tibia (Fig. 2) (twenty-four patients; 28%), followed by the proximal part of the femur (twenty-three patients; 26%), the proximal part of the humerus (nineteen patients; 22%), the distal part of the femur (eight patients; 9%), the tarsal bones (four patients; 5%), the distal part of the tibia (three patients; 3%), and the pelvic bones (two patients; 2%). There was one lesion each in the distal part of the radius, the proximal part of the fibula, the scapula, and the patella (Fig. 3).
The aggressiveness of the lesion was classified, according to the Springfield system, as latent in twenty-five patients (29%), active in fifty-one (59%) (Fig. 1), and aggressive in eleven (13%).
All chondroblastomas were treated surgically. In 27% of the patients, a biopsy was performed first as an independent procedure because the radiographic appearance or location of the bone lesion was unusual. The surgical treatment consisted of complete curettage followed by autogenous bone-grafting in fifty-five patients (63%), curettage alone in twenty-four (28%), and en bloc resection in five (6%). In this last category, two patients had a primary joint arthrodesis and one had a primary hip arthroplasty following the en bloc resection. Finally, three patients were managed with curettage and filling of the cavity with cement.
Histological analysis revealed that 28% of the chondroblastomas had an associated cystic component presenting as an aneurysmal cyst (Fig. 4). S-100 protein was found to be present in all but one of the twenty-four chondroblastomas for which the test was performed.
All tumors had typical histological features of a chondroblastoma, with a proliferation of mononuclear cells and giant cells. No atypical mitotic activity, tumor necrosis, nuclear atypia, or other signs of malignancy were found.
The mean duration of follow-up was 62.5 months (range, four to 300 months). Among the eighty-seven patients, seventy-six had at least twenty-four months of follow-up. The statistical analysis evaluating the risk factors for recurrence was performed with the data from these seventy-six patients.
Recurrence
At a minimum of twenty-four months of follow-up, twenty-four (32%) of the seventy-six patients had a local recurrence (see Appendix). The group with recurrence included sixteen boys and eight girls with a mean age of 12.6 years (range, 4.5 to sixteen years). In the group of tumors that were located near the most-active physis, twelve of forty-eight chondroblastomas recurred. In the group of tumors that were located near less-active physes, twelve of twenty-eight recurred. The rates of recurrence were the highest for lesions involving the proximal part of the femur and the tarsal bones (Table I). The mean time to recurrence was eleven months (range, three to twenty-four months) postoperatively. Recurrence was always local, involving the initial site of the chondroblastoma. There were no cases of metastases.
For chondroblastomas of the proximal part of the femur, two different approaches were used for curettage: (1) a direct Hueter anterior approach or (2) a transcervical approach through the lateral cortex of the proximal part of the femur. With the transcervical approach, the epiphyseal lesion was reached with a trephine that was inserted through the lateral cortex of the proximal part of the femur just beneath the greater trochanter and was passed up through the femoral neck. The curettage was performed under fluoroscopy with use of long curettes. At the time of the latest follow-up, 50% of the lesions recurred after transcervical curettage whereas 31% recurred after curettage through a direct approach.
Twelve recurrent chondroblastomas were treated with repeat curettage followed by additional autologous bone-grafting (nine) or filling of the cavity with cement (three). Five recurrent chondroblastomas were treated with en bloc resection, two were treated with simple curettage, and two were treated with soft-tissue-mass resection. Finally, at the time of the present report, three recurrent lesions had not been treated with additional surgery and were being followed with clinical examination and radiographs because they had remained small and the patients were free of symptoms.
Histological analysis revealed an aneurysmal-like cystic component in seven (29.2%) of the twenty-four recurring lesions, compared with 28.8% of the nonrecurring lesions; this difference was not significant (p = 0.97) (Table II).
With the numbers available, we could not identify a significant correlation between recurrence and epidemiologic data (p = 0.37 for sex, and p = 0.94 for age), between recurrence and the radiographic grade according to the scale of Springfield et al. at the time of diagnosis (p = 0.33), or between recurrence and the type of treatment performed (p > 0.05) (Table II). With regard to the location of the chondroblastoma relative to the physeal plate, the risk of recurrence was significantly higher (p = 0.01) for lesions that were located only within the epiphysis as compared with those that were metaphyseal, apophyseal, and epiphyseal-metaphyseal (Table II). Finally, logistic regression analysis showed that recurrence was significantly influenced by the location of the chondroblastoma relative to the most-active physis and relative to the epiphysis. The risk of recurrence was significantly higher for epiphyseal lesions as compared with apophyseal, metaphyseal, and epiphyseal-metaphyseal lesions (p = 0.004) and was significantly lower for lesions located close to the most-active physes (p = 0.04) (Table III).
Function
The mean duration of follow-up was 62.5 months (range, four to 300 months). In the group of seventy-six patients who were followed for at least twenty-four months, the outcome was rated as good, fair, or poor on the basis of a combined functional and radiographic classification system.
Fifty-two patients (68%) eventually had a good outcome without pain or growth disturbance, and all returned to normal unrestricted activities. Eighteen patients (24%) had a fair outcome because of occasional pain, asymmetric range of motion, or radiographic joint changes without arthritis. The last six patients (8%) had a poor outcome because of chronic pain, radiographic findings of arthritis, loss of joint motion impairing normal life activities, or a limb-length discrepancy and limp. In the group of eighteen patients with a fair outcome, the most common complications were a limb-length discrepancy of 1 to 3 cm (six patients), occasional pain (six patients), and radiographic abnormalities (six patients). Two patients also had asymmetric joint motion (but with <15° of discrepancy). Finally, two patients complained of slight pain or discomfort during sports activities (running). In the group of patients with a poor outcome, four patients complained of a limp, three patients complained of pain more than once a week, two had a limb-length discrepancy, and two patients had a stiff joint. Thus, twenty-four (32%) of the patients had a fair or poor outcome with substantial clinical impairment and/or radiographic joint changes. In seven patients, a surgical resection that included the adjacent joint was performed. Among these patients, four had a poor clinical outcome (one ankle arthrodesis, one knee arthrodesis, one total hip arthroplasty, and one total knee arthroplasty), and three had a good outcome (one shoulder arthroplasty, one total hip arthroplasty, and one ankle arthrodesis).
Chondroblastomas located in the tarsal bones and the proximal part of the femur were associated with the worst functional outcome at the time of the latest follow-up. Of the four tarsal chondroblastomas, two eventually were treated with an ankle arthrodesis after en bloc resection following a recurrence. Nine (47.4%) of the nineteen lesions that were located in the proximal part of the femur (and were followed for at least twenty-four months) had a fair or poor outcome. Two patients with chondroblastomas located in the proximal part of the femur eventually had a total hip replacement, with one having a good outcome and the other having a poor outcome at the time of the latest follow-up. Of the two patients with apophyseal chondroblastomas involving the greater trochanter, one had a good outcome and the other had a fair outcome.
To our knowledge, the present report describes the first series to focus exclusively on cases of chondroblastoma that were treated and followed in patients with open physes. In 1998, Schuppers and van der Eijken reported an interesting case series of chondroblastoma in children and adolescents15. The average age of their patients was 14.4 years, but 34% of the patients had closed physes at the time of diagnosis and treatment. Whether the aggressiveness of the lesion depends on its development before or after physeal closure is not known. Considering the major role attributed to the physis in the development of this cartilaginous tumor7 and the high potential for recurrence, we thought that a pediatric cohort should only include patients with open physes at the time of diagnosis and treatment. Because of the physeal activity, it can be hypothesized that the behavior of a chondroblastoma would be different in skeletally immature patients than in older patients, particularly with regard to recurrence after curettage.
A few differences are noticeable when our results are compared with those reported in other series of cases in adults and adolescents. The two most frequent locations of the lesions in the present study were the proximal part of the tibia and the proximal part of the femur, whereas the most frequent locations in other series usually have been the proximal part of the humerus and the distal part of the femur6-8,16. The distal femoral epiphysis was the most common location in the series reported by Schuppers and van der Eijken15 and was the second most common in the series reported by Springfield et al.7 (see Appendix), but that location represented only 9% of the lesions in the present study. The distal femoral epiphyseal location seems to be less frequent in pediatric patients but, because of the small size of the samples in each bone location in the present study and in other published studies, statistical analysis was not possible. We found four lesions located in the metaphysis of long bones, a rare feature in all series17,18. Patients with pelvic or tarsal lesions tended to be older than those with long-bone lesions, which is consistent with the findings of previous reports8,17. With regard to sex distribution, male predominance was less pronounced in our pediatric series than in the adult population: we reported a 1.6 male:female ratio, whereas the reported ratios in the literature have ranged between 1.7:16,19 and 2:116.
The objective of the present study was to report the rate of recurrence and to identify the predicting factors for recurrence of chondroblastomas in children. None of the epidemiologic factors (age or sex) or therapeutic factors that were analyzed in this series were significantly correlated with recurrence.
The location of the chondroblastoma relative to the physis and the activity of the physis did have an influence on recurrence. Epiphyseal chondroblastomas were associated with a 4.4 times higher risk of recurrence (hazard ratio estimate associated with the event of recurrence) when compared with those not completely within the epiphysis at the time of diagnosis (metaphyseal, apophyseal, and epiphyseal-metaphyseal lesions). Conversely, lesions located close to the most-active physes (close to the knee or far from the elbow) were associated with a lower risk of recurrence (Table III). Performing a thorough curettage within the epiphysis in a growing child is more challenging than performing the same surgical procedure to remove a metaphyseal or a diaphyseal bone tumor. As stressed by Springfield et al.7, preserving the integrity of the physis is more technically demanding in cases of epiphyseal lesions and the risk is incomplete curettage and recurrence. In the present series, there was no histological difference between epiphyseal lesions and metaphyseal, apophyseal, or epiphyseal-metaphyseal lesions that could explain the discrepancy in recurrence rates. We also found that recurrence was more frequent for chondroblastomas located near the less-active physes. Recurrent lesions were either among the epiphyseal chondroblastomas or among the lesions located far from the most active physes. Recurrent chondroblastomas in this series were mostly lesions of the proximal femoral epiphysis and the tarsal bones.
Performing curettage while preserving the architecture of the small tarsal bones (particularly in young children) is quite challenging. Incomplete curettage or cortical fractures may occur. Recurrence may lead to destruction of the bone with spread to the soft tissues and finally to the need for a large surgical resection followed by an arthrodesis. Therefore, curettage should be as meticulous and thorough as possible when initially treating a chondroblastoma, particularly one located in a tarsal bone. Furthermore, we recommend en bloc resection in the case of a recurrence and when the tumor occupies all of a small bone and complete curettage will be difficult or impossible.
Resection of chondroblastomas located within the proximal femoral epiphysis requires a difficult surgical approach. In a large series of seventy-three cases, Ramappa et al. found that this location was the only factor that was significantly associated with a higher rate of recurrence8. Gaining surgical access to the proximal part of the femur may compromise the blood supply to the femoral head. A direct hip joint approach provides the surgeon with good access to and exposure of the lesion, but the risks include spreading tumor tissue into the hip joint and damaging the femoral head cartilage. Conversely, transcervical curettage may lead to spread of the tumor into the femoral neck and may lead to physeal closure. Moreover, through the transcervical approach, curettage is generally less complete than is the case with a direct approach. In our experience, the rate of recurrence was two of four in the transcervical group and four of thirteen in the direct anterior approach group. Although the difference was not significant, we advocate a direct approach for the treatment of chondroblastomas located in the proximal part of the femur in order to facilitate a complete and safe curettage, supporting the recommendation of Springfield et al. as reported in 19857.
Few cases of chondroblastoma of the pelvis have been reported in the literature4,5, but some authors have suggested an increased rate of recurrence of lesions in this location7,8. Neither of the two pelvic chondroblastomas in the present series recurred. Both of these lesions occurred in boys with ages of fifteen and seventeen years (one of whom had less than twenty-four months of follow-up). In both cases, a biopsy was first performed for diagnosis. Thorough curettage was then carried out with use of a direct anterior approach. Lin et al. reported that a pelvic location was associated with a worse prognosis and reported two cases of recurrence and metastasis20. They suggested that the high rate of recurrence reported may be related to the technical difficulty of curettage in this location but not to the inherent biologic aggressiveness of the tumor20.
Some authors have suggested that the association of a cystic component (aneurysmal bone cyst) with a chondroblastoma is correlated with a higher risk of recurrence4,9, but other series10,11 have failed to confirm this relationship. We could not identify any significant correlation between recurrence and the presence of a cystic component in the present series of chondroblastomas.
The gold standard for the treatment of chondroblastoma remains complete and meticulous curettage of the lesion, most frequently combined with bone-grafting. The cavity can also be packed with polymethylmethacrylate21. In our series, the rates of recurrence associated with simple curettage with no graft (nine of twenty-two) and with curettage and packing of the cavity with polymethylmethacrylate (two of three) were higher. These recurrence rates were not significantly different from those associated with other treatment methods, probably because of the small number of cases.
The overall rate of recurrence in the present series was 32%. This rate is among the highest in the literature. Biologic aggressiveness may explain local recurrence in some rare cases20,22, but the concept of malignant chondroblastoma remains controversial20,23. Various studies of chondroblastomas in adults and children have demonstrated recurrence rates ranging from 8.3% to 15%8,11,20. Open physes have been considered to be a risk factor for recurrence4,7,8,24. Suneja et al.11 analyzed pediatric and adult cases separately and found young age to be a predictive factor for recurrence; some authors have suggested that surgeons may perform an incomplete curettage in pediatric patients because they are concerned that aggressive curettage may damage the growth plate7,11,20. This remains a major concern to us, particularly when the chondroblastoma is located in the epiphysis, which is usually the case, and it is likely the explanation for the higher rate of recurrence associated with epiphyseal chondroblastomas in comparison with epiphyseal-metaphyseal, apophyseal, and metaphyseal lesions. It is also possible that this cartilaginous lesion originating near (or from) the physis may be more aggressive and may have a higher potential for local recurrence when the physes are still active. However, this is unlikely as in our series chondroblastomas located near the most-active physes exhibited a lower rate of recurrence.
Adjuvant treatments25 for pediatric chondroblastoma should be evaluated more extensively because the recurrence rate is high. For instance, radiofrequency heat ablation is another option26 for gaining access to difficult locations. A recent report demonstrated the success of this technique for the treatment of chondroblastoma of the femoral head27.
The present retrospective review of the results of surgery for the treatment of chondroblastoma in children demonstrated that the functional outcome after thorough curettage was good for 68% of the patients. In this review, fulfilling the criteria for a good result was difficult because the patient had to have a symmetric range of motion, had to have no joint changes on radiographs, and had to be symptom-free. In the remaining patients (32%), the result was graded as fair or poor. The group with fair results included patients who had an asymmetric range of motion but who returned to normal daily activities without pain. Two of our patients who needed a joint replacement were symptom-free and had a normal range of motion.
In conclusion, chondroblastoma in the growing child is most frequently located in the proximal part of the femur and the proximal part of the tibia. The lesion is associated with a high risk of recurrence (32%), possibly because less-extensive curettage is performed in young patients as a result of concerns about the adjacent open physis. Chondroblastomas located within the epiphysis are associated with a higher risk of recurrence than epiphyseal-metaphyseal, metaphyseal, or apophyseal lesions are. The most difficult lesions to treat are those that are located in the proximal femoral epiphysis and the tarsal bones because the rates of recurrence are particularly high and because the final outcome remains uncertain in these locations. Despite the proximity of the physis, we strongly recommend that thorough curettage be performed at the time of the initial operation in order to minimize the risk of recurrence.
Note: The following surgeons participated in this retrospective case series: Pr. Bernard Tomeno, Pr. Raphael Seringe, Pr. Gérard Bollini, Pr. Christophe Glorion, Pr. George Filipe, Pr. Jérôme Bérard, Pr. Rémi Kohler, Pr. Jérôme Sales de Gauzy, Pr. Jean-Michel Clavert, Pr. Jérôme Cottalorda, Pr. Henri Bracq, and Dr. Jean-Luc Clément.