The purpose was to define the rate of local recurrence (LR) and identify prognostic factors for LR in patients with extremity soft-tissue sarcoma (STS) treated with limb-sparing surgery and a pathologically negative re-resection specimen without radiotherapy (RT).
A review of the prospective sarcoma database identified 200 patients with primary, nonmetastatic, extremity STS treated with limb-sparing surgery between June 1982 and December 2002 who had a pathologically confirmed negative re-resection. None of the patients had adjuvant RT. Univariate and multivariate analyses were performed to determine clinicopathologic factors associated with LR.
With a median follow-up of 82 months the 5-year actuarial LR rate was 9%. Factors associated with higher LR rates on univariate and multivariate analysis were older age, stage III presentation, and histology. The 5-year LR rate was 5% for those <50 compared with 15% for those ≥50 (P = .001). For patients with stage III the LR rate was 26% versus 7% for those with stage I/II (P < .001). On multivariate analysis only age ≥50 (relative risk [RR] 3.3; P = .02) and stage III disease (RR 3.4; P = .01) remained significant predictors for LR. When the cohort of patients was divided into 3 groups based on the number of risk factors present, the 5-year LRs were as follows: no risk factors (stage I/II and <50 years old) 4%, 1 risk factor (stage III or ≥50) 12%, and 2 risk factors (stage III and ≥50) 31% (P < .01).
The importance of obtaining pathologically negative margins in the surgical management of extremity soft-tissue sarcoma (STS) has been well documented.1–7 Margin status has major implications for the extent of surgical resection and impacts subsequent adjuvant treatment. Yet there are equivocal data on what constitutes an optimal or acceptable margin in sarcoma surgery. Some authors have suggested that wide margins of >1 cm may decrease the risk of recurrence and obviate the need for adjuvant radiotherapy (RT).5, 8, 9
Obtaining a pathologically negative re-excision, ie, no tumor cells in the specimen, may represent the ‘ultimate’ negative margin. Many patients with extremity STS initially undergo a limited excision without wide margins for a presumptive benign diagnosis. Once the diagnosis of sarcoma is established, patients are often referred to large specialty centers for further management.10, 11 This has resulted in a growing number of patients undergoing re-resection after an initial unplanned excision to obtain negative margins. It is estimated that at the time of re-resection 40% to 75% of patients will not have evidence of residual sarcoma on pathologic assessment.5, 12, 13 It is generally assumed that such patients have a low rate of local recurrence (LR) and may constitute a subset of patients who do not require adjuvant radiation; however, the data supporting this are scant.14, 15
The purpose of this study was to determine the feasibility of omitting adjuvant radiation in a group of patients with primary extremity STS treated with limb-sparing surgery who were found to have a pathologically negative re-resection specimen.
MATERIALS AND METHODS
Review of our prospective sarcoma database identified 200 patients with primary, nonmetastatic, extremity STS treated with limb-sparing surgery between June 1982 and December 2002 who had a pathologically confirmed negative re-resection. Patients who underwent amputation and those with desmoid tumors and dermatofibrosarcoma were excluded. Patients with metastatic disease were excluded. All patients had routine blood work and chest x-ray. Patients with large, high-grade tumors underwent CT scan of the chest to rule out pulmonary metastases. For patients with low-grade tumors and small, high-grade tumors, a CT scan of the chest was usually not performed given their low risk of metastases.16
Initial resection (before re-resection) for all patients was a gross total resection with ‘inadequate gross margins.’ The majority of patients had the unplanned local excision before referral to Memorial Sloan-Kettering Cancer Center (MSKCC) and a few patients had their initial resection at MSKCC. All patients underwent a second operation at MSKCC that involved re-resection of the tumor bed, surgical scar, and drain sites based on the inadequacy of the initial surgery. The median time period between the local excision and second wide resection was 1 month (range, 0-6).
Patients who received adjuvant RT were excluded because the main purpose of the study was to determine whether it was feasible to withhold RT in this group of patients. In this retrospective review having a pathologically negative re-excision may have not been the only determining factor in forgoing radiation but most likely it contributed to the decision to omit it.
The median age at the time of diagnosis was 46 (range, 16-91) with 39% (n = 77) of patients ≥50 years old. The clinical, pathologic, and treatment characteristics are listed in Table 1. Men constituted 54% (n = 107) of the patient population. Truncal, pelvic, and abdominal tumors were excluded. The tumor site was upper extremity (at or beyond the shoulder) in 35% of patients (n = 69) and lower extremity (at or beyond the groin) in 65% (n = 131). Of the 200 patients, 47% (n = 93) had low-grade (LG) tumors and 53% (n = 107) high grade (HG). Tumor size was defined according to the pathologic report as the maximum diameter of the tumor. Tumors >5 cm were present in 28% of patients (n = 55) and ≤5 cm in 72% (n = 145). Tumors were staged using the 2002 AJCC classification; the AJCC stage grouping was as follows: 93 stage I (LG), 85 stage II (≤5 cm HG or >5 cm superficial HG), and 22 stage III (>5 cm deep HG). The most common histologic subtype in this cohort was liposarcoma (n = 53, 27%) followed by malignant fibrous histiocytoma (n = 46, 23%), leiomyosarcoma (n = 43, 21%), synovial sarcoma (n = 16, 8%), and fibrosarcoma (n = 12, 6%). For the purpose of statistical analysis, the least common histologies were grouped together; these included epithelioid (n = 4), malignant peripheral nerve sheath tumor (n = 4), rhabdomyosarcoma (n = 3), clear cell (n = 3), hemangiopericytoma (n = 3), extraskeletal chondrosarcoma (n = 3), undifferentiated (n = 2), Ewing/PNET (primitive neuroectodermal tumor) (n = 2), osteosarcoma (n = 1), leiomyoblastoma (n = 1), spindle cell (n = 1), hemangioendothelioma (n = 1), angiosarcoma (n = 1), and alveolar-soft part sarcoma (n = 1).
Table 1. Patient and Treatment Characteristics
AJCC indicates American Joint Commission on Cancer.
Malignant fibrous histiocytoma
Chemotherapy was given at the discretion of the treating oncologists. Seventeen patients (9%) received doxorubicin-based chemotherapy. It was delivered before surgery in 3 patients and after surgery in 14 patients.
Follow-up and Statistics
Actuarial local control, distant metastases-free survival, and overall survival were calculated from the date of re-resection and were estimated using the Kaplan-Meier method. For overall survival, loss to follow-up was censored and death from any cause was considered an event. The median follow-up for the entire cohort was 82 months (range, 1-289) calculated from the date of re-excision. Actuarial rates were calculated using the Kaplan-Meier survival curves. Comparisons of survival curves were performed using the log-rank test (Mantel-Cox). Factors significant on log-rank test were entered into a stepwise Cox proportional hazards regression model. The following factors were studied as prognostic factors for the clinical outcomes of interest: age, sex, location, histology, AJCC stage, and adjuvant chemotherapy.
Patterns of Relapse
Of the 200 patients, 20 (10%) developed LR occurring between 3 and 98 months after the date of re-resection, with a median time to recurrence of 22 months. Fifteen of the 20 recurrences (75%) occurred within the first 3 years. All 20 patients with an LR underwent limb-sparing surgery as salvage therapy; no patient required amputation for salvage. Ten patients received adjuvant RT in addition to resection for the recurrence; 6 patients received brachytherapy and 4 received external beam RT. Of the 20 patients with an LR, 6 developed distant metastases. Three patients developed a synchronous distant relapse (2 with synovial sarcoma and 1 with liposarcoma); 1 patient developed pulmonary metastases 8 months after developing an LR (malignant fibrous histiocytoma) and 2 patients developed metastases approximately 2 years after salvage therapy (synovial sarcoma and malignant peripheral nerve sheath tumor [MPNST]). All 6 patients with metastases eventually died of their disease. Of the remaining 14 patients with LR, 3 died of unrelated causes: 1 patient died of unknown causes 15 months after successful salvage therapy; 1 patient died of unknown causes 4 years after definitive surgery with LR; and a third patient died of unrelated causes approximately 10 years after successful salvage therapy—he was without evidence of disease at the time of death. The remaining 11 patients with LR were alive and without evidence of disease at the time of last follow-up (median follow-up, 95 months).
A total of 21 patients (11%) developed a distant recurrence with a median time to recurrence of 30 months. In 13 of these 21 (62%) patients, metastases developed within the first 3 years after surgery. Six of these patients also had LR as previously described. Only 1 patient with distant metastases remains alive; the remaining 20 died of their disease.
Of the entire cohort, 161 patients (81%) remain alive and 39 are deceased. The cause of death was recurrent sarcoma in 20 of the 39 deceased patients with a median time to death of 46 months. The remaining 19 patients died of other or unknown causes.
The 5- and 10-year actuarial LR rates were 9% (95% confidence interval [CI], 5%-13%) and 12% (95% CI, 7%-18%), as shown in Figure 1. On univariate analysis, risk factors associated with a statistically significant higher rate of LR were age ≥50, stage III presentation, and histology (Table 2). The 5-year LR rate was 15% (95% CI, 7%-26%) for patients ≥50 years of age compared with 5% (95% CI, 1%-9%) for those <50 years old, P = .001 (Fig. 2). For patients with stage III disease the LR rate was 26% (95% CI, 6%-46%) versus 7% (95% CI, 5%-9%) for those with stage I/II, P < .001 (Fig. 3). Histology was associated with a higher rate of LR (P = .02). The rate of LR ranged from 0% for leiomyosarcoma to 19% for synovial sarcoma, as listed in Table 2. Sex and tumor site did not have a significant impact on local control (Table 2). The selective use of chemotherapy in a small number of patients did not seem to affect the rate of LR. On multivariate analysis, age ≥50 (relative risk [RR] 3.3; 95% CI, 1-9; P = .02) and stage III disease (RR 3.4; 95% CI, 1-9; P = .01) retained their significance as independent predictors of local recurrence. Histopathologic subtype was not significant on multivariate analysis.
Table 2. Univariate Analysis for Predictors of Local Recurrence
No. of LRs
% 5-year LRs
LR indicates local recurrence.
6 of 123
14 of 77
13 of 178
7 of 22
0 of 43
5 of 53
3 of 30
Malignant fibrous histiocytoma
5 of 46
2 of 12
5 of 16
12 of 107
8 of 93
9 of 69
11 of 131
When the cohort of patients was divided into 3 groups based on the number of risk factors present, the 5-year actuarial LR rates were as follows: no risk factors (stage I/II and <50 years old) 4%, 1 risk factor (stage III or ≥50) 12%, and 2 risk factors (stage III and ≥50) 31%, as shown in Figure 4 (P < .001).
Distant Relapse and Survival
The 5- and 10-year actuarial distant metastases-free survival rates were 91% (95% CI, 89%-93%) and 88% (95% CI, 82%-93%). The 5- and 10-year actuarial overall survival rates were 88% (95% CI, 83%-93%) and 79% (95% CI, 72%-86%), as shown in Figure 1.
The majority of extremity soft tissue masses are benign and are usually managed with local excision. Only a small percentage of these masses will prove to be sarcomas.10, 11 Thus, it is not surprising that a large portion of STS of the extremity are initially resected with inadequate margins. Many publications have shown that the rate of LR after local excision alone is unacceptably high, ranging from 40% to 75%.17 For this reason, most patients who initially undergo an unplanned local excision will require a second resection to obtain wide negative margins.
The data regarding the prognostic implication of undergoing a re-resection are somewhat conflicting, in particular with regard to local control.2, 5, 12, 14 Lewis et al.12 found that patients who undergo a re-resection have improved disease-specific and metastases-free survival compared with those patients who are treated with a single definitive surgery. Re-resection, however, did not significantly decrease the rate of LR in that study. A report from Italy also showed no difference in local control for patients who had undergone re-resection compared with those who had a single definitive resection.14 In contrast, a study from the M. D. Anderson Cancer Center found that re-resection improved metastases-free survival, disease-specific survival, as well as local control.2
Among the patients who undergo a second surgery, there is a subset for whom the pathologic re-resection specimen shows no evidence of residual sarcoma. The rate of pathologically negative re-excision varies among published reports because of differences in inclusion criteria and the use of preoperative adjuvant therapy in some reports. Fiore et al.14 found that among 318 patients who underwent a re-resection at their institution, 76.4% had no microscopic residual disease. Other series have reported no residual sarcoma in 37% to 65% of patients at the time of re-resection.2, 15, 12, 18 Interestingly, little has been published on the prognostic significance of a negative re-resection, yet there seems to be an inclination to forego adjuvant RT in this group of patients under the assumption that the risk of local recurrence should be low, irrespective of other prognostic factors. Perhaps this is because many believe that the presence of a pathologically negative re-excision represents the ‘ultimate’ wide negative margin.
Several authors have reported favorable results after wide resection with negative margins and without adjuvant RT in highly selected patients.13, 19–21 Khanfir et al.5 studied the role of adjuvant RT in 133 patients treated with wide excision and a negative microscopic margin. They showed that adjuvant RT improved local control exclusively in patients with a minimal resection (margin <1 cm). However, in patients with an optimal resection (>1 cm margin) RT did not have a significant effect on LR rates. Baldini et al.8 also found that the extent of the surgical margin impacted the rate of local control. In that study the local control rate for patients with a negative surgical margin of <1 cm was 87%, compared with 100% for patients with a histologic margin of ≥1 cm in a group of patients treated with surgery alone.
However, whether a wide negative margin and a negative re-resection are actually equivalent and should be managed in the same way has not been demonstrated. Fiore et al.14 did not find a difference in local control between patients with and without residual sarcoma in the re-excision specimen. This could suggest that a negative re-excision does not necessarily confer a local control advantage. However, because the authors in that study did not evaluate the impact of adjuvant RT, it could be argued that more patients with residual sarcoma received RT, whereas those with negative re-excisions did not, thus resulting in equivalent local control.
To our knowledge the current report is the first to focus only on the specific subset of patients who did not receive adjuvant RT after a negative re-resection. The 5-year rate of local recurrence rate in our report was 9%. This is comparable to the rate that has been achieved in most combined modality series, where local recurrence rates typically range from 10% to 15%.22, 23 In addition, the majority of recurrences occurred in the first 3 years after surgery. We believe this could have potential implications for surveillance imaging. This is an area that requires further investigation.
Although the overall rate of LR was low in the current series, we were able to identify risk factors that increased the risk of LR despite a negative re-resection. Patients with AJCC stage III tumors had an LR rate of 26% compared with 7% for those with stage I/II (P < .001) and the 5-year LR rate was 5% for patients <50 years old compared with 15% for those ≥50 (P = .001). Furthermore, if local recurrence was stratified according to the number of unfavorable risk factors present, the 5-year rate of local recurrence increased from 4% for those with no risk factors (stage I/II and <50 years old) to 12% for those with 1 risk factor (stage III or ≥50 years old) and to 31% for patients with both risk factors (stage III and ≥50 years old), P < .001. Therefore, a negative re-resection does not necessarily result in a low rate of LR and traditional prognostic factors such as age and stage need to be considered when making decisions regarding the need for adjuvant RT. Why the LR rate is not low across the board for patients with negative re-excision is an intriguing question that needs further study. One plausible explanation is that remaining microscopic disease was not detected at the time of re-resection.
We acknowledge that there are some limitations to the current study. The major limitation is the obvious selection bias in our series. The determination regarding adjuvant RT was made on an individual basis. Most likely, the decision to forego radiation was based partly on the negative re-resection specimen, in addition to other favorable tumor/patient characteristics. The favorable characteristics of this unique patient population likely contributed to the overall low rate of LR. Patients who undergo a local excision followed by an oncologic wide excision tend to have smaller, more superficial lesions; this is evident in Table 1, as 72% of patients had tumors ≤5 cm and 45% had superficial tumors. Another limitation is the wide range of histologies included. The goal of this study was to determine the local control rate for all patients with STSs treated at MSKCC with a negative re-resection who did not receive radiation. Therefore, we included histologies that are typically managed with multimodality therapy (PNET, Ewing, rhabdomyosarcoma) and which would be excluded from prospective surgery-alone protocols. A small number of these patients (6) over 2 decades at our institution did not receive radiation after having a negative re-resection, for various reasons; these patients were therefore included in the current analysis. We recognize that the only way to objectively determine whether adjuvant RT is needed in this group of patients would be to conduct a randomized trial; however, the rarity of this tumor is prohibitive.
In conclusion, this report suggests that young patients (<50 years old) with stage I/II disease have a sufficiently low rate of LR after a pathologically negative re-resection to consider omitting the routine use of adjuvant RT. But equally important is the finding that patients ≥50 years old and/or stage III disease have higher rates of LR, justifying the use of adjuvant RT.