We performed a retrospective study of eighty-five patients in whom a grade-I central chondrosarcoma of a long bone had been treated in our hospital between 1994 and 2005. Patients from hospitals in the surrounding areas who were suspected of having chondrosarcoma were referred to our musculoskeletal oncology department. All patients underwent a preoperative gadolinium-enhanced magnetic resonance imaging (Gd-MRI) scan prior to surgery.
The indication for surgery was the likely presence of low-grade chondrosarcoma26,27 located in one of the long bones on the Gd-MRI scan. The average age at surgery was 47.5 years (range, 15.6 to 72.3 years). Surgery consisted of an oncologically safe biopsy, followed by intralesional curettage immediately or two weeks later.
There were eighty-five patients in this series. Frozen-section biopsy followed by curettage was performed during the same operation in twenty-five cases, whereas sixty patients had a two-stage procedure. In thirteen of these sixty patients, the biopsy had already been done by the referring physician. All biopsy results were reviewed again by an experienced pathologist at our hospital (P.C.W.H.) who specializes in the pathology of bone and soft-tissue tumors. The lesions were histologically classified according to the recently published consensus criteria28 and graded according to the system described by Evans et al.13. Patients were included in this study on the basis of a histological diagnosis of grade-I central chondrosarcoma located in a long bone (Fig. 1).
The initial volume of the tumor was measured preoperatively with use of dynamic Gd-MRI scans. Due to the difficulty in measuring a three-dimensional structure on two-dimensional MRIs, all lesions were measured by projecting an imaginary cylinder. The average maximal radius r (anterior-posterior and medial-lateral) and maximal height of the tumor h (craniocaudal) were used as parameters to calculate the volume of a cylinder (πr2h). Depending on the largest diameters, different sequences of the MRIs were used.
Depending on the site of the tumor, patients received general or regional anesthesia for definitive treatment. Following preoperative identification of the precise location of the bone window on MRI, a small incision was made without the use of any Hohmann retractors to avoid possible tumor spill to other anatomical compartments. A window was thus created in the middle of the length of the tumor. The chondrosarcoma was removed macroscopically with use of small curets. No high-speed burr was used. The mechanical extension of the margin was determined by both the cortical border and the intramedullary canal. If there was doubt about whether all of the cartilage had been removed, fluoroscopy was used to detect any calcified cartilage. A solution of 85% phenol (Liquid Phenol, Ph Ned Ed VI quality; BUFA bv, Pharmaceutical Products, Uitgeest, The Netherlands) was applied for a period of five minutes to the interior of the remaining bone cavity with a surgical swab. The phenol was subsequently rinsed with a 96% ethanol solution. Finally, the bone cavity was filled with deep-frozen, nonirradiated allograft bone chips derived from donor femoral heads (Bio Implant Services, Leiden, The Netherlands). During surgery, the bone window was submerged in a phenol solution and then rinsed with ethanol. Following placement of the bone graft, the bone window was replaced.
Prior to discharge, a postoperative radiograph was made for all patients to ensure that there were no postoperative complications. Additional radiographs were obtained six and twelve weeks after the procedure to establish the extent to which the patient could safely resume normal activities.
Patients were also scheduled for a dynamic Gd-MRI scan27 six months after the procedure. The first scan was used as a baseline so that, with the following scans, a distinction could be made between the postoperative effects and the possible recurrence of chondrosarcoma. Dynamic Gd-MRI scans, in addition to radiographs, were repeated six months later and then periodically (Fig. 2). Patients were evaluated clinically on an annual basis. The average duration of follow-up was 6.8 years (range, 0.2 to 14.1 years).
If a recurrence of the cartilaginous tumor was suspected on evaluation of the Gd-MRI scan, and the recurrent lesion was ≥10 mm, curettage, phenol application, and bone-grafting, as described for the index procedure, was repeated and the curetted tissue was evaluated histologically. With small lesions (<10 mm), a computed tomography-guided biopsy was performed and the specimens were evaluated histologically. In the same procedure, radiofrequency ablation was performed, with acceptance of overtreatment in the cases in which no recurrence of tumor would be diagnosed. Following these procedures, patients with recurrent chondrosarcomas were treated according to the postoperative protocols described above, including follow-up Gd-MRI scans.
Data analysis was performed with Excel (Microsoft, Redmond, Washington), SPSS (version 17.0 for Windows; SPSS, Chicago, Illinois), and R (version 2.10.0; R Foundation for Statistical Computing, Vienna, Austria).
We used the Kaplan-Meier product-limit estimator to analyze the survival rate, with recurrence as the end point. We treated patients who were lost to follow-up as censored at their last recorded visit. One patient died of an unrelated cause during the follow-up period, and we treated this death as a competing risk.
Source of Funding
No external funding sources were used for this study. The authors have no personal or institutional interest in any of the drugs, materials, or devices described in this article.
General
The average duration of follow-up for the patients was 6.8 years (range, 0.2 to 14.1 years), with five patients being followed for less than two years. One patient, followed for two months, came from abroad to have surgery in our clinic and returned to his home country after surgery. One female patient, seventy-two years of age at the time of surgery, lived at a substantial distance from the hospital; given her age, she was referred to a hospital near her home for follow-up. Three other patients were lost to follow-up. With use of the Kaplan-Meier estimator, the patients who were followed for a limited duration were censored at their last recorded visit.
Patients were admitted to the hospital for one to three days, depending on the site of the chondrosarcoma. Postoperative management depended on the tumor site and the size of the bone window. Patients with a chondrosarcoma in the upper extremity were managed with a sling for two to six weeks postoperatively. Following curettage in the lower extremities, patients were either non-weight-bearing or partially weight-bearing for six weeks and used crutches once they were mobile. None of the patients were treated with internal fixation, and casts were not necessary because of the less invasive and limited nature of our surgical procedure compared with wide resection and reconstruction of the long bone29.
Tumor Volume
The preoperative mean volume of the lesions, measured on MRI scans, was 23.7 cm3 (range, 1 to 104 cm3). The median volume was 18.8 cm3 (Fig. 3).
Histological Findings
All patients were diagnosed with a grade-I central chondrosarcoma, according to the recently published consensus criteria28 and the system described by Evans et al.13 (Fig. 4).
Comorbidity
Four (7%) of the fifty-five women had a history of breast cancer or developed this tumor during the period of follow-up of the cartilage tumor.
Complications
One patient developed a superficial wound infection postoperatively, which resolved with antibiotics. Two (5%) of the thirty-nine patients treated for a tumor in the femur experienced a femoral fracture, which was likely due to the bone window, within six weeks after surgery. One of these patients was treated with open reduction and internal plate fixation, and the other was treated with an intramedullary nail. Gd-MRI performed five years after removal of the nail did not show any sign of tumor recurrence. Patients with chondrosarcoma of the femur appear to have a higher fracture risk, which can be addressed with a hip spica cast and non-weight-bearing with two crutches. The prophylactic use of internal fixation should be avoided, to allow follow-up MRI and to avoid a second surgical procedure for removal of the metallic implants.
Repeat Interventions
Remaining tumor was suspected in eleven patients on the basis of postoperative Gd-MRI scans. All of these patients underwent repeat procedures. Depending on the size of the lesion, biopsy followed by radiofrequency ablation (for tumors of <10 mm) or repeat curettage (for those of ≥10 mm) was performed. All tissue obtained with biopsy prior to radiofrequency ablation or with curettage was sent for histological analysis. These analyses showed recurrence of the grade-I chondrosarcoma in three of the six patients who underwent radiofrequency ablation and no signs of recurrence in the other three. Of the five patients who underwent repeat curettage, two were found to have recurrence of the grade-I chondrosarcoma and three had no signs of recurrence (Table I). The recurrence rate in this series was 5.9% (95% confidence interval [CI], 0.9% to 10.9%).
The serial Gd-MRI scans did not show any signs of tumor recurrence in the remaining seventy-four patients. With regard to the ability of the Gd-MRI to predict recurrence of chondrosarcoma in this series, the positive predictive value was 45% and the negative predictive value was 100% (Table II).
Survival
The survival rate, with histologically proven recurrence of grade-I chondrosarcoma as the end point, was 91.3% (95% CI, 84% to 99.4%) at a mean of 6.8 years after the first surgery (Fig. 5).
Mortality
One patient died, due to an adenocarcinoma of the pancreas, during the follow-up period.
We describe a large group of patients with grade-I central chondrosarcomas of the long bones who were treated with intralesional curettage followed by application of phenol and ethanol as adjuvant therapy and then by bone-grafting. The use of phenol as an adjuvant in the treatment of bone tumors has been described only in small patient series that have included a variety of tumors, both benign and malignant20-32. Although efficacy of this treatment has been proven in vitro24, it has been difficult to convincingly demonstrate the efficacy of phenol as an adjuvant in humans.
This study is limited by its observational and retrospective design. We did not use a control group to compare the results. The ideal situation would be to perform a prospective, multicenter, randomized trial comparing phenol with cryotherapy or polymethylmethacrylate adjuvant treatment.
Four (7%) of the fifty-five women in our series had also been diagnosed with breast cancer in the past or during the follow-up period. Odink et al.33 described an odds ratio of 7.62 to be diagnosed with both breast cancer and a cartilaginous tumor. Therefore, physicians should be aware of this combination of diseases whenever either a cartilaginous tumor or breast cancer is diagnosed in a female patient33-35.
Performing follow-up only with radiographs to detect local recurrence overestimates the disease-free survival10,18. We therefore use Gd-MRI to follow our patients. On the first postoperative scan, it is sometimes difficult to distinguish between the normal postoperative appearance as a consequence of the use of bone grafts from femoral heads and the presence of postoperative edema at the surgical site36. A second scan is therefore often decisive because the postoperative changes are lessened and the lesion suspected of containing residue is still clearly enhanced within ten seconds on the dynamic series of the Gd-MRI scan27,37. In retrospect, the cause of the residual tumor in the two recurrent femoral cases in our series was surgical in nature. The residual tumor remained as a result of incomplete curettage, primarily as a consequence of a bone window that was too small or had been placed in a suboptimal location; this is particularly a risk for femoral diaphyseal lesions. No significant correlation was seen between preoperative tumor size and recurrence rates (Table I).
The distinction between benign and malignant cartilaginous tumors is often subject to discussion. To improve the reliability of the diagnosis of these lesions, Eefting et al. performed a study on interobserver variability28. With use of the recently proposed consensus criteria, 94.7% of their cases were diagnosed correctly (sensitivity, 95%; specificity, 95%). Eighteen pathologists from Europe and the United States participated in that study. In our study, all specimens were reviewed again by an experienced pathologist who is familiar with the above criteria.
In our series, recurrence was identified in five patients (5.9% [95% CI, 0.9% to 10.9%]). None of the recurrent tumors had a higher histological grade than the original tumor. However, eleven patients underwent repeat surgery because residual lesion was suspected. Regarding the use of Gd-MRI to predict recurrence of tumor, the positive predictive value was 45% and the negative predictive value was 100%. In this series, postoperative Gd-MRI overestimated the number of recurrent lesions. We started using this new method of treating low-grade central chondrosarcoma in 1994, at which time the technique was deemed controversial in Europe because of concerns about its oncological safety. In light of this, the threshold for surgery in the event of a suspected lesion on postoperative Gd-MRIs was low. Throughout the past fifteen years, we have gained greater insight into the benign nature of these residual or recurrent lesions and we now take a more conservative approach toward repeat surgery for small suspected lesions. Despite the large number of false-positive results in the past, Gd-MRI remains the most sensitive tool for detection of small residual or recurrent lesions.
The use of phenol as an adjuvant as described in this study has potential advantages for the patients. In contrast to cryosurgery, following which up to 14% of patients sustain a postoperative fracture18, there was no need for prophylactic implant placement to prevent fractures. Moreover, joints adjacent to the surgical site are not impaired by this procedure.