The anatomic location of an extremity soft tissue sarcoma may influence the patient's presentation, management, and local and systemic recurrence rates. The objective of this study was to compare the presentation, management, and outcome of patients with soft tissue sarcomas of the upper extremity and the lower extremity.
Prospectively collected data from patients who underwent limb-sparing surgery for extremity soft tissue sarcoma between January, 1986 and April, 1997 were analyzed. Local recurrence free rates and metastasis free rates were calculated using the method of Kaplan and Meier. Univariate and multivariate analyses of potential predictive factors were evaluated with the log-rank test and the Cox proportional hazards model.
Of 480 eligible patients, 48 patients (10.0%) had a local recurrence, and 131 patients (27.3%) developed distant metastasis. The median follow-up of survivors was 4.8 years (range, 0.1–12.9 years). Patients with upper extremity tumors had smaller lesions (6.0 cm vs. 9.3 cm; P < 0.001), more often underwent unplanned excision before referral (89 patients [64.0%] vs. 160 patients [46.9%]; P < 0.001), and less often received radiotherapy (98 patients [70.5%] vs. 289 patients [84.8%]; P < 0.001). The 5-year local recurrence free rate was 82% for patients with sarcomas of the upper extremity and 93% for patients with sarcomas of the lower extremity (P = 0.002). The 5-year metastasis free rate was 82% for patients with sarcomas of the upper extremity and 69% for patients with sarcomas of the lower extremity (P = 0.013).
Factors known to influence local recurrence after treatment of an extremity soft tissue sarcoma include a positive surgical margin, presentation with a local recurrence, and age > 50 years.1–3 The development of metastasis is associated with large, high-grade, deep tumors.4–8 However, because tumor size, the width and quality of the surgical margin, and the use of adjuvant radiotherapy all may vary with anatomic location,9 the site of a tumor may be an important determinant of oncologic outcome. A comparison of extremity and retroperitoneal soft tissue sarcomas illustrates this point. Compared with extremity tumors, retroperitoneal soft tissue sarcomas tend to be large at presentation, difficult to excise with a wide margin, and difficult to irradiate because of adjacent vital structures. Proximity to vital structures also means that retroperitoneal sarcomas more often cause death by local recurrence, whereas extremity sarcomas are fatal through the development of pulmonary metastases.10
Clearly, there is wide variation in both the anatomic location of a soft tissue sarcoma and in the anatomy of the upper and lower extremities. The question of how this anatomic variation influences patient presentation, treatment, and outcome has not been addressed directly in the literature. However, some differences have been reported. Lower extremity tumors are more common,11–15 larger,9 and more likely to lead to metastasis4, 9, 16 compared with tumors of the upper extremity. Conversely, patients with upper extremity tumors are more likely to have undergone unplanned excision before referral.12, 17 The purpose of this study was to determine whether there were differences in the presentation and management of patients upper extremity soft tissue sarcomas and lower extremity soft tissue sarcomas among our patient population and, subsequently, whether there were differences in the local recurrence free rate and the metastasis free rate.
MATERIALS AND METHODS
Since January, 1986, a record of all patients who underwent surgery in our center for an extremity soft tissue sarcoma has been kept in a password-protected, computerized data base. Data are collected prospectively in outpatient clinics and at a weekly multidisciplinary meeting. This data base was used to identify and retrieve information about patients who had an extremity soft tissue sarcoma and underwent limb-sparing surgery between January 1986 and April 1997 and, thus, had the potential for a minimum follow-up of 3 years. The upper extremity was defined as commencing at the medial border of the scapula, including the pectoral, periscapular, and latissimus dorsi muscles. The lower extremity was defined as commencing at the iliac crest. Patients with metastatic disease at the time of presentation or with the histologic types dermatofibrosarcoma protruberans, primitive neuroectodermal tumor, or rhabdomyosarcoma were excluded.
The following information was extracted from the data base: age at surgery, gender, unplanned excision before referral, presentation with a local recurrence, greatest tumor dimension (in centimeters), histologic type and grade, surgical margin status, adjuvant radiotherapy or chemotherapy, local recurrence, time to local recurrence after surgery, systemic recurrence, and time to systemic recurrence after surgery. An unplanned excision was defined as an excisional biopsy or unplanned resection that was performed without adequate preoperative staging or consideration of the need to remove normal tissue around the tumor.18 A pathologist with an expertise in sarcoma determined the histologic type and graded tumors as 1, 2 or 3.19 The status of the surgical margin was recorded as positive if the pathologist had identified tumor, viable or not, at the inked resection margin or if there had been intraoperative exposure of tumor. A local recurrence was defined as tumor within or at the edge of the previously treated field at any time after treatment. Metastatic disease was defined as tumor identified on chest X-ray or computed tomography scan during follow-up or the development of lymphadenopathy containing tumor. Wherever possible, after multidisciplinary discussion, patients were offered limb-sparing surgery. Amputation was performed when limb-sparing surgery would not have adequately resected the tumor or would have resulted in a limb without useful function. Adjuvant radiotherapy was given to patients with high-grade tumors who underwent resection with less than wide surgical margins or patients with low-grade tumors who underwent deliberately marginal surgical excision.
The following data relating to anatomic location were retrieved: extremity, location within the extremity, and depth. Location within the extremity was recorded in relation to the closest major joint as the shoulder, elbow, wrist/hand, hip, knee, and foot/ankle. Tumors above and that did not involve the investing fascia of the limb were considered superficial, and other tumors were considered deep.
Data were used first to define the characteristics of the whole group and the anatomic distribution of tumors. Next, differences in presenting features, tumor characteristics, and treatment between the extremities were determined. The local recurrence free rate in the extremities was compared using the methods of Kaplan and Meier and the log-rank statistic. Then, the Cox proportional hazards model was used to investigate first whether greatest tumor dimension was predictive of time to local recurrence and then whether surgical margin status, extremity, and unplanned excision before referral were predictive of time to local recurrence. Thereafter, exploratory analyses examined the influence of unplanned excision before referral, surgical margin status, and radiotherapy on the rate of local recurrence. The metastasis free rate was compared using the methods of Kaplan and Meier and the log-rank statistic. The Cox proportional hazards model was used to investigate whether histologic grade, greatest tumor dimension, depth, and extremity were predictive of time to metastasis.
Differences in the mean values between groups were compared using one-way analyses of variance and the Tukey post-hoc test. Differences in proportions were compared using the Pearson chi-square test. When constructing Kaplan–Meier curves and in the Cox proportional hazards model, patients were censored at the time of last follow-up or death. In the Cox proportional hazards model, the hazard ratio for each covariate was plotted to ensure proportionality over time. Throughout, results with a P value < 0.05 were considered significant. Analyses were performed using SPSS software for Windows (release 10.0.5; SPSS Inc., Chicago, IL).
Five hundred sixty-six patients were entered into the data base between January 1986 and April 1997. There were 6 primary amputations in 168 patients with upper extremity tumors (3.6%) and 22 primary amputations in 398 patients with lower extremity tumors (5.5%). Another 58 patients were excluded: 34 patients who presented with metastatic disease, 12 patients who presented with a diagnosis of dermatofibrosarcoma protruberans, 7 patients who presented with primitive neuroectodermal tumors, and 5 patients who presented with rhabdomyosarcoma. Four hundred eighty patients remained in the study.
Characteristics of the Whole Group
There were 261 men (54.4%) and 219 women (45.6%) with a median age of 56 years (range, 15–96 years). Malignant fibrous histiocytoma (163 patients; 34.0%) and liposarcoma (107 patients; 22.3%) were the most common histologic types. Tumors were Grade 1 in 83 patients (17.3%), Grade 2 in 172 patients (35.8%), and Grade 3 in 225 patients (46.9%).
Forty-eight patients (10.0%) developed a local recurrence at a median of 1.4 years (range, 0.1–9.2 years). No patients died of local recurrence. One hundred thirty-one patients (27.3%) developed metastases at a median of 1.0 years (range, 0.1–12.8 years), and 86 of those 131 patients died of metastatic disease at a median of 1.5 years (range, 0.1–6.8 years). Nineteen patients died from causes other than metastatic disease at a median of 1.0 years (range, 0.0–7.7 years). Three of those 19 patients died in the immediate postoperative period after developing a pulmonary embolus. The median follow-up for the 375 surviving patients was 4.8 years (range, 0.1–12.9 years).
There were 139 tumors (29.0%) of the upper extremity and 341 tumors (71.0%) of the lower extremity (Table 1). Seventy-four tumors (15.4%) were located around the shoulder, 41 tumors (8.5%) were located in the elbow, 24 tumors (5.0%) were located in the hand or wrist, 154 tumors (32.1%) were located in the hip, 148 tumors (30.8%) were located in the knee, and 39 tumors (8.1%) were located in the ankle or foot (Table 2). Three hundred seventy-seven tumors (78.5%) were deep to or involved the investing fascia of the extremity.
Table 1. Summary of Results by Extremity
No. of patients
Age at Surgery (yrs)
Proportion of males (%)
Greatest tumor dimension (cm)
Deep to or involving investing fascia
Histologic grade (%)
Unplanned excision before referral (%)
Presentation with a local recurrence (%)
Positive surgical margins (%)
Adjuvant radiotherapy (%)
Follow-up of survivors (yrs)
No. of patients with local recurrence
Time to local recurrence (yrs)
Local recurrence free rate at 5 years (%)
No. of patients developing metastases
Time to metastasis (yrs)
Metastasis free rate at 5 yrs (%)
Table 2. Summary of Results by Location Within Each Extremity
No. of patients
Mean greatest tumor dimension (cm)
Unplanned excision before referral
Positive surgical margins
Adjuvant radiotherapy given
No. of developing metastases
Differences between the Extremities
Patients with tumors of the upper extremity more often underwent unplanned excision before referral compared with patients who had tumors of the lower extremity (89 patients [64.0%] vs. 160 patients [46.9%]; P < 0.001). The proportion of patients presenting with a local recurrence was similar in each extremity (18 patients in the upper extremity [12.9%] vs. 38 patients in the lower extremity [11.1%]; P = 0.576).
Upper extremity tumors were significantly smaller compared with lower extremity tumors (6.0 cm vs. 9.3 cm; P < 0.001) and less often were deep to or involved the investing fascia (97 patients [69.8%] vs. 280 patients [82.1%]; P = 0.003) (Table 1). The mean tumor dimension decreased from proximal to distal in both extremities (Table 2). There was a relative excess of synovial sarcomas, epithelioid sarcomas, and fibrosarcomas in upper extremity sites and of liposarcomas and myxoid liposarcomas in lower extremity sites (Table 3). Histologic grade did not differ significantly between upper extremities and lower extremities (P = 0.219) (Table 1).
Table 3. Distribution of Histologic Types by Extremity
Percentage within extremity
Percentage within extremity
Malignant fibrous histiocytoma
Liposarcoma, otherwise undesignated
Malignant peripheral nerve sheath tumor
The proportion of patients with positive surgical margins did not differ significantly between the upper extremities and the lower extremities (28 patients [20.1%] vs. 71 patients [20.8%]; P = 0.868). However, within the group of patients with upper extremity tumors, positive surgical margins were more frequent around the elbow (16 of 41 patients; 39.0%) compared with the wrist and hand (5 of 24 patients; 20.8%) or the shoulder (7 of 74 patients; 9.5%; P = 0.001).
Overall, radiotherapy was given less often to patients with tumors of the upper extremity compared with its frequency among patients with tumors of the lower extremity (98 patients [70.5%] vs. 289 patients [84.8%]; P < 0.001). This was because patients with tumors around the shoulder were not treated with adjuvant radiotherapy (45 of 74 patients; 60.8%) as often as patients with tumors around the elbow (34 of 41 patients; 82.9%) or the wrist (19 of 24 patients; 79.2%; P = 0.048) (Table 2). The proportion of patients who received adjuvant chemotherapy did not differ significantly between those with tumors of the upper extremity or the lower extremity (2 of 139 patients [1.4%] vs. 8 of 341 patients [2.3%); P = 0.528).
The median follow-up for survivors was 4.9 years (range, 0.1–11.1 years) among patients with tumors of the upper extremity and 4.7 years (range, 0.2–12.9 years) among patients with tumors of the lower extremity.
There were 23 patients (n = 139 tumors) who developed local recurrences in the upper extremity and 25 patients (n = 341 tumors) who developed local recurrences in the lower extremity. The median time to local recurrence was 1.3 years (range, 0.2–9.2 years) for patients with tumors of the upper extremity and 1.6 years (range, 0.1–6.6 years) for patients with tumors of the lower extremity. The local recurrence free rate at 5 years was 82% for patients with tumors of the upper extremity and 93% for patients with tumors of the lower extremity (log-rank test; P = 0.002) (Fig. 1).
The difference in the local recurrence free rate between the extremities was investigated with a Cox proportional hazards regression model. First, recognizing the difference in tumor size between the extremities, regression was performed using the greatest tumor dimension as a single, continuous variable. This did not reach significance (P = 0.074), confirming our belief that tumor size is not a determinant of local recurrence. A second model was constructed using (in the following order) categorical variables for surgical margin status (negative or positive), extremity (lower or upper), unplanned excision before referral (no or yes), and an interaction variable of extremity and unplanned excision before referral. Plots for each variable confirmed proportionality of the hazard ratio over time. Surgical margin status reached significance in the model (P < 0.001) with a hazard ratio of 3.16 (95% confidence interval, 1.76–5.69) associated with a positive margin. Extremity, unplanned excision before referral, and the interaction variable of extremity and unplanned excision before referral did not reach significance (P = 0.127, P = 0.056, and P = 0.868, respectively).
An exploratory analysis of the correlation between an unplanned excision before referral, surgical margin status, and local recurrence compared crude local recurrence rates in four groups (Table 4). Among patients who underwent primary excision of a tumor with positive surgical margins, patients who had tumors of the upper extremity were more likely to develop a local recurrence compared with patients who had tumors of the lower extremity (3 of 13 patients [23.1%] vs. 2 of 46 patients [4.3%]; P = 0.032).
Table 4. Exploratory Analysis of the Relation between Unplanned Excision Before Referral, Surgical Margin Status, and Local Recurrence
P value for difference in local recurrence rates between extremities
No. of patients
No. oflocal recurrences (%)
Local recurrences not given radiotherapy
No. of patients
No. of local recurrences (%)
Local recurrences not given radiotherapy
Primary excision with negative surgical margins
Primary excision with positive surgical margins
Unplanned excision before referral, reexcision with negative margins
Unplanned excision before referral, reexcision with positive margins
A further exploratory analysis examined the effect of radiotherapy on local recurrence by calculating how many patients who developed local recurrences among these four groups had received radiotherapy (Table 4). Six of 11 patients with upper extremity tumors who underwent unplanned excision before referral and reexcision with negative margins and who developed a local recurrence had not received radiotherapy. In three patients, all with tumors located around the shoulder, this was because radiotherapy had been given on a previous occasion, precluding further radiotherapy. In two of the remaining patients, radiotherapy was not given, because no tumor was identified in the reexcision specimen. The sixth patient had a low-grade fibrosarcoma around the shoulder that was believed to have been completely excised.
Metastasis free survival
Twenty-six patients with upper extremity tumors (n = 139 tumors) developed a metastasis compared with 105 patients with lower extremity tumors (n = 341 tumors). The median time to metastasis was 1.0 years (range, 0.1–6.7 years) after surgery for patients with tumors of the upper extremity and 1.0 years (range, 0.1–12.8 years) after surgery for patients with tumors of the lower extremity. The metastasis free rate at 5 years was 82% for patients with tumors of the upper extremity and 69% for patients with tumors of the lower extremity (log-rank test; P = 0.013) (Fig. 2). There was a trend toward fewer metastases in patients with distal tumors (Table 2).
To investigate the difference in metastasis free rates between the extremities, a Cox proportional hazards model was constructed using, in the following order, histologic grade (low [Grade 1] or high [Grade 2 or 3]), greatest tumor dimension (in cm), depth (superficial or deep), and extremity (upper or lower). Proportionality over time was confirmed for each variable. There was an increased risk of metastasis with high histologic grade (hazard ratio, 17.28; 95% confidence interval, 4.26–70.10; P < 0.001), greatest tumor dimension (hazard ratio, 1.05; 95% confidence interval, 1.02–1.08; per cm increase, P < 0.001), and deep location (hazard ratio, 1.93; 95% confidence interval, 1.07–3.48; P = 0.028). The variable for extremity did not reach significance (P = 0.211).
This study compared the presentation, treatment, local recurrence free rate, and metastasis free rate of upper and lower extremity soft tissue sarcomas in our patient population. We showed that patients with upper extremity tumors have smaller lesions, more often undergo unplanned excision before referral, have tumors of different histologic types, and less often meet our criteria for adjuvant radiotherapy. We also showed that, after treatment for extremity soft tissue sarcoma, the local recurrence free rates and metastasis free rates vary with anatomic location. In particular, local recurrence is more frequent after treatment of an upper extremity sarcoma. Conversely, lower extremity sarcomas are associated with an increased metastatic rate.
A number of factors may contribute to the higher local recurrence rate among patients with tumors of the upper extremity. These include differences in tumor characteristics, such as histologic type, and differences in treatment, such as the effect of unplanned excision before referral and the use of radiotherapy. The strongest predictor of local recurrence in many studies is a positive surgical margin, which implies the presence of residual disease.7, 14, 20, 21 The Cox model confirmed this correlation in our series and showed that, after incorporating the effects of a positive surgical margin, an unplanned excision before referral, or the location of a soft tissue sarcoma in the upper extremity rather than the lower extremity, they did not reach significance. Sadoski et al. suggested that the relation between a positive surgical margin and local recurrence is stronger in the upper extremity compared with the lower extremity.21 However, in our series, there was a higher local recurrence rate in the upper extremity in both patients with positive margins and patients with negative margins (Table 4).
The proportion of patients who had positive surgical margins was particularly high (16 of 41 patients) among those who had tumors around the elbow, which may reflect local anatomy. Despite this finding, the local recurrence rate in this group was not as high as it was among patients with tumors in other upper extremity sites (Table 2). The reason for this may be that four patients in this group had low-grade liposarcomas and, thus, were at low risk of local recurrence.
Although it was a subgroup analysis and, thus, should be treated with caution, it is interesting to note that there was a significant difference in local recurrence rates between the extremities among patients who underwent primary excision with positive margins. There are a number of possible explanations for this. We showed previously that an unexpected positive margin after a surgical error is associated with a higher risk of local recurrence compared with a planned positive margin against a critical structure (such as nerve, vessel, or bone) or a planned marginal resection of a low-grade liposarcoma.22 In 5 of 16 patients (31.3%) who underwent primary resection of an upper extremity sarcoma with a positive surgical margin, the positive margin was unexpected compared with 8 of 48 patients (16.7%) with lower extremity sarcomas, which may explain the higher local recurrence rates. This suggests that primary resection of a soft tissue sarcoma in the upper extremity is technically more demanding or that, to preserve function, the surgeon is prepared to operate closer to the tumor in the upper extremity. Another explanation for the difference in local recurrence rates is that there were more low grade liposarcomas in the lower extremity. These tumors are at low risk for local recurrence.23
The distribution of histologic types may have contributed to the difference in local recurrence rates between the extremities in another way. The histologic types angiosarcoma and malignant peripheral nerve sheath tumor have been associated with an increased risk of local recurrence.12, 24 Both of these types were relatively more frequent among tumors of the upper extremity compared with tumors of the lower extremity. Within the upper extremity, these histologic types were associated with 4 of 23 patients (17.4%) who developed local recurrences, compared with 2 of 25 patients (8.0%) who had tumors of the lower extremity (Table 3).
Variations in the use of radiotherapy also may have influenced the local recurrence rate. In particular, patients with tumors around the shoulder did not meet our criteria for radiotherapy as often as patients who had tumors in other upper extremity sites. This probably is because a proportion of soft tissue sarcomas around the shoulder occur in muscle (such as deltoid), where they can be excised with an adequate margin more readily compared with tumors located in other upper extremity sites. Therefore, it is more likely that an unplanned excision will be successful in removing all identifiable tumor, and adjuvant radiotherapy is not indicated after reexcision. It also was more common for the shoulder to have been treated with radiotherapy previously, both for the presenting tumor and for unrelated conditions, such as Hodgkin disease.
An unplanned excision complicates the further management of patients with soft tissue sarcoma by obscuring tissue planes and potentially contaminating a large volume of tissue. Patients with tumors of the upper extremity more frequently undergo unplanned excision before referral, possibly because the tumors are smaller and more often are superficial. However, whether unplanned excision increases the risk of recurrence remains controversial,25 and we could not conclusively prove an adverse influence of unplanned excision on local recurrence in the Cox model.
The finding that lower extremity tumors are associated with a higher rate of metastasis confirms the reports of other authors.4, 9, 16 The Cox model suggested that that most of this difference is accounted for by recognized risk factors for metastasis (i.e., grade, size, and depth4, 5, 7, 8, 16) rather than another hitherto unrecognized feature of lower extremity sarcomas. The analysis makes the assumption that, after the treatment of an extremity soft tissue sarcoma, local recurrence and metastases are independent events and are not competing events. The length of follow-up of survivors was similar among patients with tumors of the upper extremities and patients with tumors of the lower extremities; therefore, we do not believe that the higher local recurrence rate among patients with tumors of the upper extremity was caused by an increased death rate and shorter follow-up of patients with tumors of the lower extremity.
There were more tumors of the lower extremities than tumors of the upper extremities in our series, an experience shared by other authors.11–15 This probably reflects the greater volume of mesenchymal tissue available to undergo malignant change in the lower extremity. Tumors in the upper extremity are smaller, probably because the smaller volume of the extremity allows the earlier detection of tumors or because the upper extremities are exposed, and abnormal swellings are identified readily. However, in common with others, we have shown that tumor size is not a strong predictor of local recurrence.4, 5, 7 We agree with the suggestion that the threshold above which a mass should be treated as potentially malignant should be lower for patients with tumors of the upper extremity.9 The larger size of lower extremity tumors probably explains why they more often involve the investing fascia of the limb and, thus, are classified as deep.
In conclusion, we have shown that, among our patient population, sarcomas of the upper extremity were associated with a higher local recurrence rate compared with sarcomas of the lower extremity because of differences in histologic type, the use of radiotherapy, and local anatomy. Although patients with upper extremity sarcomas more often underwent unplanned excision before referral, we could not conclusively prove that this had an adverse effect on local tumor control. In contrast, lower extremity sarcomas tend to be larger and more often are deep to fascias compared with upper extremity sarcomas; therefore, patients with lower extremity sarcomas have an increased risk of metastasis. This study also shows that, in our patient population, the higher rate of local recurrence in the upper extremity was not matched by a higher rate of metastatic disease. This supports our philosophy of preserving function in the upper extremity when possible. We do not believe that there is an underlying difference in the biological behavior of upper extremity soft tissue sarcoma and lower extremity soft tissue sarcoma, although the clinical course differs depending on the anatomic location.