Abstract
Background: The data on 227 patients who had been managed for a chondrosarcoma at one institution were reviewed to determine the nature of the lesions, the predictors of outcome, and whether there were any ways to change the treatment approaches to improve the results.Methods: The patients were followed for a mean duration of six years (range, three to twenty-five years). The mean age of the patients was forty-seven years (range, nine to eighty-four years). The most prevalent sites of the tumors were the femur (seventy-eight), the pelvis (fifty-one), and the humerus (thirty-nine). The tumors were divided into two groups according to histological grade. Eighty-six tumors (sixteen atypical enchondromas and seventy grade-1 chondrosarcomas) that were locally destructive but were associated with a low likelihood of metastasis were considered to be low-grade. The remaining 141 lesions, which were locally destructive, potentially metastatic, and capable of causing death, were thought to be high-grade. One hundred and three of these 141 lesions were grade 2, and thirty-eight were grade 3 (eighteen of the thirty-eight were grade 3 only, and twenty were both grade 3 and dedifferentiated). Two hundred and twenty-four patients were managed with resection and a limb-sparing procedure; the remaining three patients had an amputation. Postoperative adjuvant radiation was used for fifty-six patients; chemotherapy, for thirty-five; and both radiation and chemotherapy, for nineteen. Flow cytometric patterns were analyzed for 105 patients.Results: The patients who had a high-grade tumor were older than those who had a low-grade tumor (mean age [and standard deviation], 50 ± 17.0 years compared with 40 ± 15.9 years; p < 0.001). Pathological fracture, metastasis, local recurrence, and death were more prevalent in the group that had a high-grade lesion (p < 0.001). Predictors of metastasis and death in that group of patients included local recurrence, a pelvic location of the tumor, a tumor that was more than 100 cubic centimeters in size, a ploidic abnormality (aneuploidy coupled with a high mean DNA index), a histological grade of 3, and a dedifferentiated type of tumor (p < 0.001).Conclusions: Although the data are suggestive, with the numbers available for study we could not detect a significant difference in the rates of pulmonary metastasis and death between the patients who had a grade-3 lesion and those who had a grade-3 lesion that was also dedifferentiated. However, the interval between diagnosis and death was 32 ± 22.8 months for the patients who had a grade-3 lesion compared with 5 ± 3.7 months for those who had a grade-3 lesion that was also dedifferentiated (p < 0.001). Overall, patients who had had a resection with wide margins (margins extending outside the reactive zone) had a longer duration of survival than did those who had had a so-called marginal resection (margins extending outside the tumor but within the reactive zone) or an intralesional resection (margins within the lesion) (p < 0.04). Adjunctive chemotherapy or radiation, or both (which, it must be noted, was used, without a protocol, in a relatively small number of patients), after an intralesional resection, for recurrent disease, or for distant metastasis did not appear to alter the outcome.
Chondrosarcoma is the second most frequent primary malignant tumor of bone4,5,14,15,19,32. The rate of recurrence, metastasis, and death after operative treatment has varied and has not always been predictable on the basis of the characteristics of the original tumor1,3,6,14,15,20,23,26,28-31,33,34,36,37,41. Although patients who have a high-grade lesion are considered to be at high risk for metastasis, chondrosarcomas generally are treated with operative resection only1,4,5,7,16,17,23,27,35. Adjuvant radiation and chemotherapy have been reserved for patients who have a mesenchymal chondrosarcoma or a dedifferentiated chondrosarcoma or for those who have had inadequate operative treatment10,28,31,32,37,38,42.
Planning the extent of operative treatment is sometimes a problem as it is often difficult to make a histological distinction between low-grade malignant and benign cartilaginous lesions. It is certainly possible that previous studies of the treatment of low-grade malignant cartilaginous lesions included some benign lesions, which may lack the capacity for metastasis. We believe that the rarity of this tumor, problems with definition of imaging studies, and questions about grading and staging make it difficult to define the level of malignancy of the lesion, and over the years these problems have led other investigators to use a broad array of methods to assess the virulence of the tumor2,3,8,11-13,17,18,21,24,25,39,40,43,44. Because of these issues, the gathering of data regarding the outcome for a patient who has a chondrosarcoma is a challenge.
The current study represents an attempt to meet this challenge. The study population consisted of 227 patients who had been followed for three to twenty-five years after the diagnosis of a chondrosarcoma. The purpose of the analysis was threefold: (1) to distinguish between low-grade and high-grade sarcomas, (2) to determine if there were factors that were predictive of failure of treatment of high-grade chondrosarcomas with our current protocol, and, perhaps most importantly, (3) to try to identify, on the basis of our findings, any obvious changes that we could make to reduce the number of patients who have metastasis or who die of the disease.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Hospital for Sick Children, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada.
‡Orthopaedic Oncology Unit, GRB 606, Massachusetts General Hospital, Boston, Massachusetts 02114.
§Mount Sinai Hospital, 5 East 98th Street, Box 1188, New York, N.Y. 10029.
A total of 293 patients who had a chondrosarcoma of bone were managed, between November 1972 and September 1994, at the Orthopaedic Oncology Unit of the Orthopaedic Service at Massachusetts General Hospital. This population included patients who had a hyaline, myxoid, fibrous, dedifferentiated, or clear-cell chondrosarcoma in a central, or intramedullary (enostotic), location or in a parosteal, or surface (exostotic), location. We excluded eight patients who had a mesenchymal chondrosarcoma and fifty-eight patients who had come to our institution for a second opinion or had been followed after operative resection at another center. This left 227 patients who were followed closely for a mean duration of six years (range, three to twenty-five years), and they are the subject of this report.
The patient population was typical for this type of tumor. There were more male than female patients (139 compared with eighty-eight). The mean age (and standard deviation) was 47 ± 17.2 years (range, nine to eighty-four years) (Table I); fifteen patients were less than twenty years old, seventy-three were twenty to less than forty years old, eighty-two were forty to less than sixty years old, and fifty-seven were sixty years old or more. The principal sites that were affected were the femur (seventy-eight patients; 34 percent), the pelvis (fifty-one patients; 22 percent), the humerus (thirty-nine patients; 17 percent), the tibia (fifteen patients; 7 percent), and the spine (fifteen patients; 7 percent) (Table I). Eight patients had a tumor in the foot or the ankle; seven, in the thorax (four in the ribs and three in the sternum); six, in the forearm or the hand; five, in the scapula; and three, in the fibula.
Staging for the entire series was done according to a modification of the system of Enneking and the Musculoskeletal Tumor Society13. With use of this system, benign lesions are stage 0; although that stage was not included in the original report13, it is currently used frequently by tumor surgeons. Low-grade lesions (those that are locally destructive but are associated with a low likelihood of metastasis) are stage IA or IB; A and B refer to whether the lesion remains within or extends outside of an anatomical compartment. The designations IIA and IIB are used for high-grade lesions (those that are locally destructive, potentially metastatic, and capable of causing death). Stage III describes any tumor that has metastasized. At the time of enrollment in the study, five patients were thought to have a stage-0 lesion (considered by the pathologists to be an atypical enchondroma); thirty-six, a stage-IA lesion; seventy-four, stage-IB; eleven, stage-IIA; ninety-two, stage-IIB; and nine, stage-III.
The size of the tumor, recorded for 211 patients at the time of review of the operative specimen by the pathologists, was a mean (and standard deviation) of 344 ± 1004 cubic centimeters (range, five to more than 9000 cubic centimeters). The mean size on imaging studies, performed for sixty-two patients, was 70 ± 91 square centimeters (range, five to 490 square centimeters). No association could be found between these two sets of data as studies such as computed tomography and magnetic resonance imaging were not performed for patients who were managed before the early 1980s; therefore, we chose to use the size of the pathological specimen as a determinant of outcome because that parameter has been consistent since the study began in 1972.
The histological grading system that we used was based on that described by Rosenthal et al.36 and was modified by one of us (A. E. R.). A grade of 1, 2, or 3 is assigned on the basis of the cellularity of the lesion (ranging from hypocellular to hypercellular), changes in the matrix (ranging from abundant and hyaline to sparse and myxoid), the character of the cells (ranging from small and minimally pleomorphic with little nuclear detail to markedly pleomorphic with bizarre forms, giant cells, and hyperchromatic nuclei), and replicative activity (ranging from infrequent binucleate forms to multiple, bizarre mitotic figures).
Histologically, eighty-six tumors were low-grade; these included seventy lesions that were classified as histological grade 1 and sixteen that were designated as grade 0. It should be noted that the grade-0 (so-called atypical) enchondromas were difficult to distinguish from low-grade chondrosarcomas, and we, like physicians at most treatment centers, depended on the size, location, findings on imaging, and biological behavior in order to make this determination. All of the atypical enchondromas were distinguished from benign enchondromas on the basis of the increased cellularity and bizarre morphology of the cells, radiographic evidence of local invasiveness, and unusual pain. Two patients who had an atypical enchondroma had metastasis, and one of them died of the disease, which confirmed the logic of including patients who had such a lesion in the series. Four of the eighty-six low-grade chondrosarcomas were of the clear-cell type. The remaining 141 tumors were classified as high-grade: 103 were grade 2 and thirty-eight were grade 3. The grade-3 tumors included twenty that were designated as dedifferentiated chondrosarcomas by the pathologists5,10,28.
Two hundred and twenty-four patients were managed with resection of the tumor followed by a reconstructive procedure to preserve the structural and functional integrity of the limb. The remaining three patients had an amputation as a limb-sparing procedure was not possible because of encasement of neurovascular structures by the tumor. Adjuvant radiation was administered to fifty-six patients; chemotherapy, to thirty-five; and both radiation and chemotherapy, to nineteen (Table II). At the time of presentation, forty-six patients had a pathological fracture, defined as a definite fracture line throughout the bone, discontinuity of at least three cortices as seen on anteroposterior and lateral radiographs, or a fracture line seen on computed tomographic scans (Table I).
The DNA content of the cells of cartilage and other types of tumors, as measured with flow cytometry and cytofluorometry, has been shown to vary according to the malignancy of the tumor2,3,8,18,43,44. An aneuploid pattern and a high mean DNA index are considered to be signs of a malignant lesion, whereas a diploid pattern and a low mean DNA index are more consistent with a lesion that has a lower potential for metastasis2,3,8,23-25. We started using flow cytometry in 1982, and we were able to study the flow cytometric patterns of samples from 105 patients in the current study. It should be noted that no selection process was used to determine which samples to evaluate with flow cytometry. Samples were excluded from this analysis on the basis of a variety of factors, including a lack of cellularity, an inadequate size of the specimen, the time of the day when the specimen was obtained, and so on. Fifty-one samples (49 percent) had an aneuploid pattern, and fifty-four (51 percent) had a diploid pattern. The mean DNA index, determined by calculating the mean for the number of cells in terms of the concentration of DNA, was 1.3 ± 0.31 for the entire group, reflecting the described ploidic distribution (Table I).
Despite our desire to resect the tumors as widely as possible, twenty-six patients had resection with intralesional margins, leaving behind gross tumor. Most of the intralesional margins were unintentional; however, a few were the result of a debulking procedure, performed in patients who had a chondrosarcoma involving the spinal column or the pelvis or adjacent vital structures, or both. Preoperatively, some of these patients had stated a preference for preserving the neurovascular structures rather than having an ablative procedure. Twenty-eight patients had marginal margins—that is, the resection was performed through the reactive zone; theoretically, all gross tumor is removed with this type of procedure, but microscopic tumor may be left behind. One hundred and seventy-three patients had the more desirable, wide margins, meaning that the resection included a sizable cuff of normal, healthy tissue (generally agreed to be one centimeter or more); this procedure, at least in theory, results in the removal of all tumor with the exception of skip lesions. Of the 141 patients who had a high-grade lesion, 107 had wide margins, nineteen had marginal margins, and fifteen had intralesional margins (Table II).
We compared the data for the 141 high-grade malignant chondrosarcomas (histological grade 2 or 3 or stage IIA, IIB, or III according to Enneking's system13, or both) with those for the entire series of 227 tumors and with those for the eighty-six low-grade lesions (Tables I, II, and III).
Age, Gender, Anatomical Site, and Stage-III Classification at the Time of Presentation
With respect to gender, no difference was noted between the patients who had a high-grade tumor and those who had a low-grade tumor (Table I). On the average, the patients who had a high-grade tumor were older than those who had a low-grade tumor (p < 0.001). With the numbers available for study, the difference in the anatomical distribution between the high-grade and low-grade tumors could not be shown to be significant (Table I). Only nine patients had a stage-III tumor at the time of presentation, and none of these lesions were low-grade.
Pathological Fracture
Thirty-eight (27 percent) of the 141 patients who had a high-grade tumor had a pathological fracture compared with only eight (9 percent) of the eighty-six patients who had a low-grade tumor (p < 0.001) (Table I). Most of these fractures were treated with protective immobilization until the operation was performed, within a week or two. Although the presence of a pathological fracture is suggestive, it did not affect the outcome. Of the forty-six patients who had this complication, ten had a local recurrence, sixteen had metastasis, and nine died; however, with the numbers available, these rates were not significantly different from those for the patients who did not have a pathological fracture.
Size of the Tumor
Measurements of the pathological specimen were evaluated for 211 patients. These data did not reveal a significant difference between the mean sizes of the low-grade and high-grade tumors. It should be noted, however, that the mean volume of the pelvic tumors was considerably greater than that of the tumors in other locations (Table IV). Of the seventy-five patients who had a high-grade tumor that was 100 cubic centimeters or less, eighteen (24 percent) had metastasis compared with twenty-six (45 percent) of the fifty-eight patients who had a high-grade tumor that was larger than 100 cubic centimeters (p < 0.04, Mantel-Cox test) (Table V).
Location of the Tumor
Of the thirty-two patients who had a high-grade lesion in the pelvis, seventeen (53 percent) had metastasis compared with thirty (27 percent) of the 109 patients who had a high-grade tumor in another location (p < 0.0001, Mantel-Cox test) (Fig. 1). The rate of survival of the patients who had a high-grade lesion in the pelvis was 63 percent (twenty of thirty-two) compared with 85 percent (ninety-three of 109) for the patients in whom the lesion was in another location (p < 0.0001, Mantel-Cox test) (Fig. 2). The rate of local recurrence of high-grade pelvic lesions was higher than that of low-grade pelvic lesions (nine [28 percent] of thirty-two compared with two of nineteen; p < 0.04). With the numbers available for study, no significant difference was detected between the results for high and low-grade lesions at any other location.
As noted earlier, a chondrosarcoma was considered exostotic (surface or parosteal) when the main tumor mass was outside the cortex. If the main tumor mass was located within the medullary cavity, the lesion was defined as an enostotic, or central, chondrosarcoma. There were 206 enostotic and twenty-one exostotic lesions. All but four of the 141 high-grade tumors were enostotic (intramedullary), and thirty-eight (27 percent) of them were either dedifferentiated or grade 3, or both. The exostotic lesions, which included five chondrosarcomas associated with hereditary multiple exostosis, three juxtacortical chondrosarcomas, and thirteen primary exostotic chondrosarcomas, were clearly of a lower grade (only four were high-grade) and were associated with a considerably lower risk of metastasis (p < 0.005) or death (p < 0.05) (Table III).
Flow Cytometry
Although the mean DNA index did not differ between the thirty-five low-grade tumors and the seventy high-grade tumors that were studied with use of flow cytometry, an aneuploid pattern was found much less frequently in the low-grade tumors than it was in the high-grade tumors (ten [29 percent] of thirty-five compared with forty-one [59 percent] of seventy; p < 0.007). The rate of metastasis associated with the seventy high-grade tumors was lower for lesions that had a lower mean DNA index coupled with a diploid pattern (eleven of twenty-nine; 38 percent) than it was for lesions that had a higher mean DNA index and an aneuploid pattern (seventeen of forty-one; 41 percent); however, with the numbers available, this difference could not be shown to be significant. The grade-0 and grade-1 tumors that had a diploid pattern on flow cytometry had a rate of metastasis of 4 percent (one of twenty-five) (Table VI).
Rate of Metastasis
Of the entire group of 227 patients, fifty (22 percent) had metastasis. Of the 141 patients who had a high-grade lesion (including nine who had metastasis at the time of presentation), forty-seven (33 percent) had metastasis compared with three (3 percent) of the eighty-six who had a low-grade lesion; this difference is significant (p < 0.0001, chi-square test) (Table II and Fig. 3). No significant difference with regard to the time to metastasis was detected between the two groups, with the Student t test. The patients who had an enostotic lesion had a higher rate of metastasis than did those who had an exostotic lesion (p < 0.005) (Table III).
Rate of Death
Twenty-nine (13 percent) of the 227 patients died of the disease, at a mean of 42 ± 48 months. Twenty-seven (19 percent) of the 141 patients who had a high-grade tumor died compared with two (2 percent) of the eighty-six patients who had a low-grade tumor (p < 0.001, Fig. 4). The time to death was not found to be significantly different between the high and low-grade groups, as determined with the Student t test. Among the patients who had a high-grade lesion, the rate of death was significantly lower for those who had a grade-2 lesion than it was for those who had a grade-3 or a dedifferentiated lesion (seventeen [17 percent] of 103 compared with ten [26 percent] of thirty-eight; p < 0.001) (Fig. 4). With the numbers available for study, the rate of death was not found to differ significantly between the patients who had a grade-3 lesion and those who had a dedifferentiated lesion (seven of eighteen and eight [40 percent] of twenty, respectively). In contrast, the mean time to death did differ significantly between these two groups (5 ± 3.7 months for the patients who had a dedifferentiated lesion compared with 32 ± 22.8 months for those who had a grade-3 lesion; p < 0.001).
Margins of Resection
The findings related to the margins of operative resection were remarkable. The patients who had been managed with a wide resection had a significantly higher rate of survival than did those who had been managed with either a marginal or an intralesional resection (p < 0.04) (Fig. 5). All nineteen patients who had a high-grade lesion and had been managed with a marginal resection died. With the numbers available for study, we could not detect a significant relationship between the margins of resection and the rate of either metastasis or death in the group of patients who had a low-grade lesion.
Local Recurrence
Of considerable importance was the frequency of local recurrence. Fifty-five (24 percent) of the 227 patients had a local recurrence after the operation, including thirteen (15 percent) of the eighty-six who had a low-grade lesion and forty-two (30 percent) of the 141 who had a high-grade lesion; this difference is significant (p < 0.03) (Table II). The frequency with which this complication occurred at various anatomical sites was also of interest. Although most local recurrences were in the pelvis, femur, and humerus, the rates among the sites did not differ appreciably (range, none of three tumors in the fibula to two of six in the hand or the forearm). The most frequent site of local recurrence was the pelvis (ten of fifty-five recurrences); it should be noted that nine of these ten recurrences were in patients who had a high-grade tumor, whereas only one was in a patient who had a low-grade lesion; this difference is significant (p < 0.04, chi-square test). With the numbers available for study, the rate of local recurrence was not found to be related to the operative margins in the patients who had a low-grade tumor; however, such a relationship was detected in the patients who had a high-grade tumor (Table VII). Twenty-seven (25 percent) of the 107 patients who had had a resection with wide margins had a local recurrence compared with six of the nineteen patients who had had a resection with marginal margins and nine of the fifteen who had had a resection with intralesional margins.
Effect of Local Recurrence on the Outcome
The relationship between local recurrence and metastasis in the entire series of patients was highly significant. Of the fifty-five patients who had a local recurrence, twenty-seven (49 percent) had metastasis (p < 0.0001). The rate of death of the patients who had a local recurrence before and after twelve months was nine (26 percent) of thirty-four and ten (48 percent) of twenty-one, respectively (p > 0.05). Of the 172 patients who did not have a local recurrence, only twenty-seven (16 percent) had metastasis (p < 0.00005, Mantel-Cox test). The rate of death of the patients who had a local recurrence was thirteen (24 percent) of fifty-five compared with only sixteen (9 percent) of 172 for the patients who did not have a local recurrence; this difference is significant (p < 0.0004, Mantel-Cox test). Local recurrence was not found to have a significant effect on the rates of metastasis and death in the group of patients who had a low-grade chondrosarcoma, but it was associated with a significant increase in the rates of both metastasis and death in the group of patients who had a high-grade tumor (p < 0.0005 and p < 0.0004, respectively) (Figs. 6 and 7). In the 141 patients who had a high-grade tumor, the eleven metastases that followed a local recurrence (no stage-III lesions were included) were noted at a mean of 12 ± 13.8 months (range, zero to forty-seven months). If death ensued, it did so at a mean of 25 ± 23.6 months (range, one to sixty-two months).
Determinants of the Outcome
The data clearly show that the eighty-six patients who had a low-grade lesion had very low rates of metastasis and death. These data should help treating physicians to predict the outcome, and such lesions generally should cause only limited concern.
Of much greater concern are the outcomes for the 141 patients who had a high-grade lesion. Several factors appeared to be associated with the increased rates of metastasis and death in this group. The grade of the tumor was clearly a determinant. Although grading is somewhat subjective and there is some question about the reproducibility of the classification of a lesion as a grade-2 or grade-3 chondrosarcoma, the findings in this series were striking. The patients who had a grade-3 lesion had a rate of metastasis of 55 percent (twenty-one of thirty-eight), whereas the patients who had a grade-2 lesion had a rate of 25 percent (twenty-six of 103).
Similarly, 39 percent (fifteen) of the thirty-eight patients who had a grade-3 lesion died compared with only 12 percent (twelve) of the 103 who had a grade-2 lesion; this difference is significant (p < 0.0001, chi-square test). The rate of death was similar for the eighteen patients who had a grade-3 lesion and the twenty who had a grade-3 lesion that was also dedifferentiated lesion (p > 0.05); however, the actual duration of survival for this latter group was significantly shorter (mean, 5 ± 3.7 months compared with 32 ± 22.8 months; p < 0.001).
The margins of operative resection had an effect on survival and particularly on local recurrence. The patients who had been managed with a wide resection had a higher rate of survival at five and ten years than did those who had been managed with either a marginal or an intralesional resection (p < 0.04). Local recurrence had a profound effect on the rates of both metastasis and death (Figs. 6 and 7).
The smaller high-grade lesions were considerably less likely to metastasize than were the larger high-grade lesions (p < 0.001).
Pelvic lesions were associated with higher rates of both metastasis and death than were lesions in long bones (Figs. 1 and 2).
An aneuploid pattern and a high mean DNA index were associated with a higher rate of metastasis but not a higher rate of death.
Cox Regression Analysis
It was evident that some of these factors were not independent. Cox regression analysis9 showed that the only factors that had a significant independent effect on metastasis, according to the chi-square test, were local recurrence (p < 0.0001), a pelvic location (p < 0.01), and a histological grade of 3 (p < 0.001). The same three factors had a significant effect with regard to death (p < 0.001 for local recurrence, p < 0.009 for a pelvic location, and p < 0.001 for grade-3 and dedifferentiated lesions), as did the margins of resection (p < 0.04).
Effect of Adjuvant Radiation and Chemotherapy
Adjuvant therapy was given to seventy-two patients as determined by the surgeon and the medical oncologist. The reasons for adjuvant therapy included inadequate margins of resection or the development of local recurrence or metastasis during the course of treatment. It was believed that statistical analysis would not be helpful because of selection bias. Of the 141 patients who had a high-grade lesion, thirty-two received chemotherapy, forty-seven received radiation, and sixteen received both. Of the sixty-three patients who were so managed, eight had had metastasis (stage-III disease) at the time of presentation. It is difficult to assess the reason for the adjuvant treatment or when it was instituted in the course of the disease. However, the administration of adjuvant therapy to patients who had inadequate margins of resection or a failure of control of the disease did not seem to alter the outcome. Twenty-five of the sixty-three patients had a local recurrence, thirty-six had metastasis, and twenty-four died; these numbers are all greater than those for the seventy-eight patients who did not have adjuvant treatment. It should be noted that seventeen of the sixty-three patients had a pelvic lesion and seven of them had a local recurrence.
Outcome According to Period of Treatment
There have been no changes, since the time that the study began twenty-five years ago, in the characteristics or management of patients who have chondrosarcoma or in the rate of failure of treatment.
The rate of local recurrence in the present study (forty-two [30 percent] of the 141 high-grade tumors) has remained constant (approximately 30 percent per year). However, the seventy patients who were managed for chondrosarcoma during the last five years of the study (1990 through 1994) had a rate of failure of 16 percent, whereas the rates for the patients seen during the two previous five-year periods were 24 and 29 percent. Although these numbers suggest a trend, with the numbers available they were not found to be significant according to the chi-square test.
The rate of metastasis has remained virtually fixed at approximately 33 percent per year, with the exception of the last five-year period of the study (1990 through 1994); this finding is perhaps attributable to the sometimes long delay between metastasis and the death of patients who have a chondrosarcoma. Given the mean time to metastasis of twenty months, with a wide standard deviation and an even longer time to death, the rates of metastasis and death for the patients managed during the last five years may increase with time.
The original purpose of this study was to attempt to distinguish between low-grade and high-grade chondrosarcomas, to determine if there were factors that could predict failure of treatment with our current protocol for high-grade chondrosarcoma, and to determine if changes could be made that would help to reduce prevalences of metastasis and death. We seem to have been successful with regard to the first and second objectives, particularly in identifying the high-risk population, but we were not successful with regard to the third objective as we identified few suggestions for diminishing the rates of local recurrence, metastasis, and death in these high-risk patients.
It should be evident, on the basis of our findings, that there are two basic grades of chondrosarcoma. Low-grade tumors are associated with very low rates of metastasis and death (three patients [3 percent] and two patients [2 percent] in the current series). These lesions, for the most part, can be effectively treated operatively, with a limited risk of local recurrence (thirteen [15 percent] of eighty-six patients had such a recurrence in the current series), and there is no evident need for adjuvant therapy or for resection with wider margins. These lesions can be graded histologically, and, with use of our system, are graded as 0 or 1. It should be noted that only sixteen patients who had a grade-0 lesion were enrolled in this study, and the lesion in each of them was further defined as atypical. Twelve of these patients had a lesion that was stage-IA (eight) or IIB (four) according to Enneking's system13 and as determined principally on the basis of imaging studies. Five of these patients had a local recurrence, and one patient died of the disease. We believe that the inclusion of these patients in a study of this sort was legitimate.
As noted earlier, the size of the tumor was difficult to assess on imaging. In the initial phases of the study, only radiographs and, occasionally, angiograms were available. Later, we were able to measure the lesion on computed tomographic and magnetic resonance images. As most of the imaging studies were performed in two dimensions, the values were recorded in square centimeters. The size of the specimen could be determined much more accurately, but it varied widely. Therefore, although the mean sizes of the high and low-grade lesions were different, the standard deviations were so wide that this difference was not significant. Nevertheless, the association between the size of the lesion and metastasis was significant (a 45 percent rate of metastasis for larger lesions compared with a rate of 24 percent rate for smaller ones; p = 0.02).
The patients who had a low-grade tumor were significantly younger than those who had a high-grade tumor, and flow-cytometric analysis showed a diploid pattern in twenty-five (71 percent) of the thirty-five patients who were studied and an aneuploid pattern in only ten. It seems relatively easy to assess the clinical stage of these lesions according to the relative ages of the patients, the findings on imaging studies (the lesion is considerably smaller in many of these patients), a histological grade of 0 or 1, and a diploid pattern on flow cytometry. A small lesion in a young patient who has a consistent histological and flow-cytometric profile is presumably low-grade and should be treated with an operation of sufficient extent and degree to prevent local recurrence; however, there appears to be little need for departure from conservative treatment (no radical operation and no radiation or chemotherapy).
High-grade lesions are easier to diagnose than are low-grade lesions, but they are much more difficult to treat4,10,14,16,28. The data suggest that it is possible to predict which patients have a higher risk of metastasis according to the grade of the tumor (grade-3 and dedifferentiated lesions); a pelvic location; local recurrence; a large size; an older age; and, less importantly, the flow cytometric finding of an aneuploid pattern coupled with a high mean DNA index. (The latter variable seems, for the most part, to parallel the histological grade but is still of some value.) The same information is helpful in determining the risk of death, although the rate of death of patients who have a high-grade tumor that is not treated with adjuvant therapy is relatively low. Perhaps the most startling findings were the high rates of death and metastasis for patients who had a local recurrence after seemingly effective operative treatment, as determined by examination of the specimen to assess the margins. The mean time to local recurrence was nineteen months, but the maximum interval was 9.8 years; similar maximum intervals for metastasis (7.6 years) and death (eighteen years) suggest that the short-term findings of the current study probably led us to be too optimistic.
It should be evident, on the basis of these data, that a patient who has a large grade-3 pelvic lesion with an aneuploid pattern and a high mean DNA index is at very high risk for metastasis and death despite treatment. Furthermore, if the margins are violated when the tumor is removed the risk is likely to be increased.
Our attempt to identify better ways to manage these patients also involved analysis of whether adjuvant radiation and chemotherapy had any effect on the outcome. In a study, such as this one, in which the adjuvant therapy was not randomized or administered according to a specific protocol, it is difficult to assess the value of these agents, but it seems clear that they do not have the same effect on these tumors as they do on Ewing sarcoma or osteosarcoma. The fact that radiation and chemotherapy did not result in improved outcomes is not surprising considering that they were used only in the most urgent circumstances, for patients who had a poor clinical outcome, rather than as an adjuvant treatment protocol. Our findings lend little support for more extensive use of these agents in an attempt to control high-grade chondrosarcomas; however, it seems very reasonable to use preoperative radiation in patients who have a grade-3 or dedifferentiated lesion because there is a high likelihood that operative treatment of such lesions will be followed by local recurrence. Similarly, if the lesion is stage III at the time of presentation, if it is both large and grade 3, or if it is a dedifferentiated pelvic sarcoma, the use of chemotherapy seems logical on the basis of our data. However, a prospective study is necessary to demonstrate the effect of these adjuvant agents on patients who have poor prognostic factors.
We believe that the single most important lesson that can be gleaned from these data is that the margin of resection should be as wide as is reasonable and should include a cuff of normal tissue. Failure to perform a wide resection is associated with a higher likelihood of local recurrence, which is in turn associated with very high rates of metastasis and death16,26.
It is distressing to note that more recent rates of local recurrence, metastasis, and death are no better than those for patients who were operated on twenty years ago. Perhaps the sometimes very long intervals between the operation and recurrence, metastasis, or death deluded us into thinking that we had accomplished more than we actually had, and our perception of early success may at times have been inappropriate. Chondrosarcomas must be studied and carefully graded at the time of presentation, and they must be treated effectively from the beginning. Patients must then be followed carefully for many years to be certain that a failure has not occurred.
Aboulafia, A. J.; Rosenbaum, D. H.; Sicard-Rosenbaum, L.; Jelinek, J. S.; and Malawer, M. M.: Treatment of large subchondral tumors of the knee with cryosurgery and composite reconstruction. Clin. Orthop.,307: 189-199, 1994.307189
1994
[PubMed]
Alho, A.; Connor, J. F.; Mankin, H. J.; Schiller, A. L.; and Campbell, C. J.: Assessment of malignancy of cartilage tumors using flow cytometry. A preliminary report. J. Bone and Joint Surg.,65-A: 779-785, July 1983.65-A779
1983
Alho, A.; Skjeldal, S.; Melvik, J. E.; Pettersen, E. O.; and Larsen, T. E.: The clinical importance of DNA synthesis and aneuploidy in bone and soft tissue tumours. Anticancer Res.,13: 2383-2387, 1993.132383
1993
[PubMed]
Bauer, H. C.; Brosjo, O.; Kreicbergs, A.; and Lindholm, J.: Low risk of recurrence of enchondroma and low-grade chondrosarcoma in extremities. 80 patients followed for 2—25 years. Acta Orthop. Scandinavica,66: 283-288, 1995.66283
1995
Campanacci, M. [editor]: Chondrosarcoma. In Bone and Soft Tissue Tumors, pp. 265-338. Translated by S. Notini. New York, Springer, 1990.
Capanna, R.; Bertoni, F.; Bettelli, G.; Picci, P.; Bacchini, P.; Present, D.; Giunti, A.; and Campanacci, M.: Dedifferentiated chondrosarcoma. J. Bone and Joint Surg.,70-A: 60-69, Jan. 1988.70-A60
1988
Chou, L. B., and Malawer, M. M.: Analysis of surgical treatment of 33 foot and ankle tumors. Foot and Ankle Internat.,15: 175-181, 1994.15175
1994
Coughlan, B.; Feliz, A.; Ishida, T.; Czerniak, B.; and Dorfman, H. D.: p53 expression and DNA ploidy of cartilage lesions. Hum. Pathol.,26: 620-624, 1995.26620
1995
[PubMed]
Cox, D. R.: Regression models and life-tables. J. Roy. Statist. Soc., Series B,34: 187-220, 1972.34187
1972
Dahlin, D. C., and Beabout, J. W.: Dedifferentiation of low-grade chondrosarcomas. Cancer,28: 461-466, 1971.28461
1971
[PubMed]
De Beuckeleer, L. H.; De Schepper, A. M.; Ramon, F.; and Somville, J.: Magnetic resonance imaging of cartilaginous tumors: a retrospective study of 79 patients. European J. Radiol.,21: 34-40, 1995.2134
1995
Dijkhuizen, T.; van den Berg, E.; Molenaar, W. M.; Oosterhuis, J. W.; Dam, A.; Wiersema, J.; Schraffordt Koops, H.; and de Jong, B.: Cytogenetics as a tool in the histologic subclassification of chondrosarcomas. Cancer Genet. and Cytogenet.,76: 100-105, 1994.76100
1994
Enneking, W. F.: A system of staging musculoskeletal neoplasms. Clin. Orthop.,204: 9-24, 1986.2049
1986
[PubMed]
Eriksson, A. I.; Schiller, A.; and Mankin, H. J.: The management of chondrosarcoma of bone. Clin. Orthop.,153: 44-66, 1980.15344
1980
[PubMed]
Evans, H. L.; Ayala, A. G.; and Romsdahl, M. M.: Prognostic factors in chondrosarcoma of bone: a clinicopathologic analysis with emphasis on histologic grading. Cancer,40: 818-831, 1977.40818
1977
[PubMed]
Gitelis, S.; Bertoni, F.; Picci, P.; and Campanacci, M.: Chondrosarcoma of bone. The experience at the Istituto Ortopedico Rizzoli. J. Bone and Joint Surg.,63-A: 1248-1257, Oct. 1981.63-A1248
1981
Gruber, H. E.; Marshall, G. J.; Kirchen, M. E.; Menendez, L. R.; and Schwinn, C. P.: Bone remodelling in the presence of chondrosarcoma: histomorphometry. Acta Anat.,148: 1-7, 1993.1481
1993
[PubMed]
Hasegawa, T.; Seki, K.; Yang, P.; Hirose, T.; Hizawa, K.; Wada, T.; and Wakabayashi, J.: Differentiation and proliferative activity in benign and malignant cartilage tumors of bone. Hum. Pathol.,26: 838-845, 1995.26838
1995
[PubMed]
Healey, J. H., and Lane, J. M.: Chondrosarcoma. Clin. Orthop.,204: 119-129, 1986.204119
1986
[PubMed]
Helio, H.; Karaharju, E.; Bohling, T.; Kivioja, A.; and Nordling, S.: Chondrosarcoma of bone. A clinical and DNA flow cytometric study. European J. Surg. Oncol.,21: 408-413, 1995.21408
1995
Inwards, C. Y., and Unni, K. K.: Classification and grading of bone sarcomas. Hematol./Oncol. Clin. North America,9: 545-569, 1995.9545
1995
Kaplan, E. L., and Meier, P.: Nonparametric estimation from incomplete observations. J. Am. Statist. Assn.,53: 457-481, 1958.53457
1958
Kreicbergs, A.; Zetterberg, A.; and Soderberg, G.: The prognostic significance of nuclear DNA content in chondrosarcoma. Analyt. and Quant. Cytol.,2: 272-279, 1980.2272
1980
Kreicbergs, A.; Boquist, L.; Borssen, B.; and Larsson, S. E.: Prognostic factors in chondrosarcoma: a comparative study of cellular DNA content and clinicopathologic features. Cancer,50: 577-583, 1982.50577
1982
[PubMed]
Kreicbergs, A.; Bauer, H. C. F.; Brosjö, O.; Lindholm, J.; Skoog, L.; and Söderlund, V.: Cytological diagnosis of bone tumours. J. Bone and Joint Surg.,78-B(2): 258-263, 1996.78-B(2)258
1996
Mankin, H. J.; Springfield, D. S.; Rosenberg, A. E.; Gebhardt, M. C.; and Schiller, A. L.: Chondrosarcoma of bone. In Surgery of the Musculoskeletal System, edited by C. McC. Evarts. Ed. 2, vol. 5, pp. 4895-4928. New York, Churchill Livingstone, 1990.
Mankin, H. J.; Gebhardt, M. C.; Jennings, L. C.; Springfield, D. S.; and Tomford, W. W.: Long-term results of allograft replacement in the management of bone tumors. Clin. Orthop.,324: 86-97, 1996.32486
1996
[PubMed]
Mercuri, M.; Picci, P.; Campanacci, L.; and Rulli, E.: Dedifferentiated chondrosarcoma. Skel. Radiol.,24: 409-416, 1995.24409
1995
Morris, H. G.; Capanna, R.; Del Ben, M.; and Campanacci, D.: Prosthetic reconstruction of the proximal femur after resection for bone tumors. J. Arthroplasty,10: 293-299, 1995.10293
1995
[PubMed]
O'Connor, M. I., and Sim, F. H.: Salvage of the limb in the treatment of malignant pelvic tumors. J. Bone and Joint Surg.,71-A: 481-494, April 1989.71-A481
1989
Ozaki, T.; Hillmann, A.; Linder, N.; Blasius, S.; and Winkelmann, W.: Metastasis of chondrosarcoma. J. Cancer Res. and Clin. Oncol.,122: 629-632, 1996.122629
1996
Ozaki, T.; Lindner, N.; Hillmann, A.; Rodl, R.; Blasius, S.; and Winkelmann, W.: Influence of intralesional surgery on treatment outcome of chondrosarcoma. Cancer,77: 1292-1297, 1996.771292
1996
[PubMed]
Ozaki, T.; Hillmann, A.; Linder, N.; Blasius, S.; and Winkelmann, W.: Chondrosarcoma of the pelvis. Clin. Orthop.,337: 226-239, 1997.337226
1997
[PubMed]
Pritchard, D. J.; Lunke, R. J.; Taylor, W. F.; Dahlin, D. C.; and Medley, B. F.: Chondrosarcoma: a clinicopathologic and statistical analysis. Cancer,45: 149-157, 1980.45149
1980
[PubMed]
Quint, U., and Pingsmann, A.: Surgical treatment of enchondroma in long tubular bones. Preservation of function versus extensive excision in the humerus. Arch. Orthop. and Trauma Surg.,114: 352-356, 1995.114352
1995
Rosenthal, D. I.; Schiller, A. L.; and Mankin, H. J.: Chondrosarcoma: correlation of radiological and histological grade. Radiology,150: 21-26, 1984.15021
1984
[PubMed]
Sanerkin, N. G., and Gallagher, P.: A review of the behaviour of chondrosarcoma of bone. J. Bone and Joint Surg.,61-B(4): 395-400, 1979.61-B(4)395
1979
Sheth, D. S.; Yasko, A. W.; Johnson, M. E.; Ayala, A. G.; Murray, J. A.; and Romsdahl, M. M.: Chondrosarcoma of the pelvis. Prognostic factors for 67 patients treated with definitive surgery. Cancer,78: 745-750, 1996.78745
1996
[PubMed]
Simms, W. W.; Ordonez, N. G.; Johnston, D.; Ayala, A. G.; and Czerniak, B.: p53 expression in dedifferentiated chondrosarcoma. Cancer,76: 223-227, 1995.76223
1995
[PubMed]
Tsuchiya, H.; Ueda, Y.; Morishita, H.; Nonomura, A.; Kawashima, A.; Fellinger, E. J.; and Tomita, K.: Borderline chondrosarcoma of long and flat bones. J. Cancer Res. and Clin. Oncol.,119: 363-368, 1993.119363
1993
Tunc, M., and Ekinci, C.: Chondrosarcoma diagnosed by fine needle aspiration cytology. Acta Cytol.,40: 283-288, 1996.40283
1996
[PubMed]
van Loon, C. J.; Veth, R. P.; Pruszczynski, M.; Wobbes, T.; Lemmens, J. A.; and van Horn, J.: Chondrosarcoma of bone: oncologic and functional results. J. Surg. Oncol.,57: 214-221, 1994.57214
1994
[PubMed]
Welkerling, H.; Werner, M.; and Delling, G.: [Histologic grading of chondrosarcoma. A qualitative and quantitative analysis of 74 cases of the Hamburg Bone Tumor Register]. Pathologe,17: 18-25, 1996.1718
1996
[PubMed]
Werner, M.; Rieck, J.; Heintz, A.; Posl, M.; and Delling, G.: [Comparative DNA cytometric and cytogenetic ploidy determination of chondrosarcoma and osteosarcoma]. Pathologe,17: 374-379, 1996.17374
1996
[PubMed]