We thank Nicole Moraco and Alisa Pinkhasik for data management support.
The reported high rate of local recurrence (LR) in myxofibrosarcoma raises the question of whether this sarcoma histology should be considered radioresistant. In this study, the authors compared rates and patterns of LR of high-grade (HG) myxofibrosarcoma with rates and patterns of HG leiomyosarcoma, which was chosen because of the similarity in incidence and general treatment approach.
Two hundred two patients with primary, nonmetastatic extremity myxofibrosarcoma (n = 114) and leiomyosarcoma (n = 88) underwent limb-sparing surgery and were followed prospectively. All 202 patients had HG tumors, and 138 patients (68%) received adjuvant radiation therapy.
The groups were comparable in terms of age, sex, and receipt of chemotherapy. Compared with leiomyosarcoma, myxofibrosarcoma presented more frequently with tumors >5 cm (P < .001), deep location (P = .036), and upper extremity site (P = .015). In addition, rates of positive/close margins (P < .001) and the receipt of radiation therapy (P < .001) were significantly higher in the myxofibrosarcoma group. The 5-year overall LR rate was not significantly different according to histology (14.6% for myxofibrosarcoma, 13.2% for leiomyosarcoma; P = .594). The only predictor of LR for the whole cohort of patients was positive/close margins (P = .01). Of 17 myxofibrosarcoma LRs, 8 (47%) occurred out of field, versus 1 of 12 (8%) leiomyosarcoma LRs (P = .04). Leiomyosarcoma more commonly recurred distantly (54.1% vs 24.3% at 5 years; P = .014).
Over the past several years, there have been improvements and greater consensus in the histopathologic classification of soft tissue sarcoma (STS).1 These efforts, combined with advances in molecular genetics, have led to an improved understanding of the pathogenesis of STS and, in some patients, better prediction of biologic behavior and response to therapy.2-4 One example for which improved molecular analysis has impacted multidisciplinary management is myxoid liposarcoma histology. Myxoid liposarcoma is a common variant of liposarcoma with a characteristic reciprocal translocation, a unique predilection for nonpulmonary skeletal metastases,5, 6 and an exquisite radiosensitivity.7-9 Notwithstanding these advances, in most patients, it remains largely unclear how the histopathologic classification of sarcomas should impact local management.
Myxofibrosarcoma is 1 of the most common histologic types of STS and frequently involves the extremities. Myxofibrosarcoma has previously been reported to recur significantly more frequently than other STS histologies.10 Various series have reported local recurrence (LR) rates ranging from 32% to 60%.10-12 The perceived high propensity for LR in myxofibrosarcoma raises the important question of whether myxofibrosarcoma histology should be considered radioresistant. Therefore, we examined the prevalence and pattern of recurrence in high-grade (HG) extremity myxofibrosarcoma and compared these findings with those in patients who had HG leiomyosarcoma, which is another common extremity STS. Leiomyosarcoma was chosen because of similarities in incidence and general treatment approach.13-15 Other histologies were considered as comparison groups, including malignant peripheral nerve sheath tumor (MPSNT) and HG liposarcoma. These were not selected, because myxoid liposarcoma is exquisitely radiosensitive.7-9 In addition, other HG liposarcomas and MPSNTs are rare compared with myxofibrosarcoma.
MATERIALS AND METHODS
Patient and Tumor Characteristics
From April 1991 to December 2006, 887 consecutive patients aged ≥16 years underwent definitive management of STS at Memorial Sloan-Kettering Cancer Center (MSKCC) and met the following criteria: primary nonmetastatic presentation, extremity site, HG histologic features, and limb-sparing surgery. Of these 887 patients, 114 consecutive patients with HG myxofibrosarcoma and 88 consecutive patients with HG leiomyosarcoma were followed prospectively and were the focus of this study. All patients were reviewed by an STS pathologist before entry into our prospective database. The 114 patients with myxofibrosarcoma were derived from our prospective database by querying the different malignant fibrous histiocytoma (MFH) histologic variants. Patients with pleomorphic, storiform, and MFH not otherwise specified were excluded. Of the 114 patients, 80 were entered with a diagnosis of myxofibrosarcoma. The remaining 34 patients were reclassified with myxofibrosarcoma based on a pathologic diagnosis of MFH, myxoid variant.16 Exclusion criteria included patients who underwent amputation (including ray amputation for hand sarcomas), patients with recurrent tumors, patients with low-grade histologic features, patients with distant metastases at presentation, and patients who underwent surgical resection outside of MSKCC.
The median age at presentation of the 202 patients in the study was 63 years (range, 22-95 years). There were 95 males (47%) and 107 females (53%). Tumors were considered to be in the upper extremity if they were at or beyond the shoulder (n = 52; 26%) and in the lower extremity if they were at or beyond the groin (n = 150; 74.3%). The anatomic depth of each tumor was evaluated relative to the investing superficial fascia of the extremity. A deep location of a tumor (vs a superficial location) was defined as any invasion of or through the superficial fascia. Tumor size was defined as the maximal dimension of the tumor at pathologic analysis. At the time of microscopic examination of the specimen, the surgical margins were defined as positive if tumor cells extended to the margin and close when a tumor was within ≤1 mm of the surgical margin.
The surgical technique used in this study has been described previously.17, 18 In brief, all visible or palpable disease was resected in an en-bloc fashion. Previous biopsy scars and drain sites, when present, were included in the resection. When the tumor was intermuscular or intramuscular, resection included 1 or more of the involved muscle bundles. For tumors situated near major neurovascular structures, resection was performed with margins limited by the lack of expendable soft tissues.
Of the 202 patients in the study, 138 (68%) received adjuvant radiation therapy. Thirty-three patients (24%) received postoperative brachytherapy (BRT) alone. One hundred one patients (73%) received external-beam radiation therapy (EBRT) alone, which was received preoperatively in 7 patients (7%) and postoperatively in 94 patients (94%). Four patients (3%) received a combination radiation therapy approach with BRT followed by EBRT. At our institution, patients with small (<5 cm) superficial tumors are not routinely irradiated when negative pathologic margins are obtained.
The median dose of BRT alone was 45 gray (Gy) (range, 34-45 Gy), and the median dose rate was 0.41 Gy per hour. All 4 patients who received BRT as a boost received 20 Gy, and their median EBRT dose was 50.4 Gy (range, 45-50.4 Gy). For the purpose of this analysis, the 4 patients who received BRT as a boost were included in the BRT group, for a total in that group of 37 patients.
For the 94 patients who received postoperative EBRT alone, treatment was received 4 to 6 weeks after surgery to a median dose of 63 Gy (range, 50.0-70.4 Gy) at 1.8 to 2.5 Gy per fraction. The initial target volume included the tumor bed plus 5-cm to 10-cm margins to a dose of 45 to 50 Gy. This usually was followed by 1 or 2 cone downs to bring the median total dose to 63 Gy. The 7 patients who received preoperative EBRT alone received a median dose of 50 Gy (range, 46.8-50.4 Gy) in 25 fractions followed by surgery in 4 to 6 weeks.
At MSKCC, chemotherapy is not generally recommended for all patients with HG STS of the extremity. Doxorubicin-based chemotherapy was received by 23 patients (11%). Those who received chemotherapy generally were part of early in-house protocols that were trying to address the role of adjuvant chemotherapy or they had tumors that measured >10 cm.
Follow-up was calculated from the date of the first operation at MSKCC. The median follow-up was 48 months. In patients who were still alive at the last follow-up visit, the median follow-up was 56 months. For patients who underwent surgery alone, we defined an out-of-field LR as an LR outside the tumor bed; and, for those who received adjuvant radiation therapy, an out-of-field LR was defined as an LR outside the radiation therapy field.
The Fisher exact test was used to examine differences in the clinicopathologic categorical variables between the myxofibrosarcoma and leiomyosarcoma groups. The Wilcoxon rank-sum test was used to analyze continuous variables. The cumulative incidence function was used for the competing-risks analysis to describe LR and distant recurrence (death without recurrence was regarded as a competing risk).19 The time to recurrence was defined the time (in months) elapsed from the date of surgery to the date of recurrence, death, or last follow-up. The Kaplan-Meier method20 was used to describe overall survival, and the log-rank test was used to compare overall survival differences nonparametrically. Survival was defined as the time (in months) elapsed from the date of surgery to the death date or last follow-up. Regression analysis was performed using a Cox proportional hazards model21 for overall survival and a Fine and Gray model22 for competing-risks analyses.
The 2 patient groups were comparable on the basis of age, sex, and receipt of chemotherapy (Table 1). Compared with patients who had HG leiomyosarcoma, however, patients who had HG myxofibrosarcoma presented more frequently with tumors >5 cm (78% vs 51%; P < .001), deep location (80% vs 66%; P = .036), and upper extremity site (32% vs 17%; P = .015). In addition, the rate of positive/close margins was significantly higher for the myxofibrosarcoma group (42% vs 18%; P < .001). Forty-one percent of patients in this study underwent excision at an outside hospital before undergoing en-bloc, definitive, wide local excision at MSKCC. The rate of prior excision was not significantly different according to histology (37% for myxofibrosarcoma vs 47% for leiomyosarcoma ; P = .195). Fewer patients with previously excised myxofibrosarcoma had negative pathologic specimens after undergoing definitive limb-sparing surgery at our institution (40.5% vs 80.5%; P < .001). The decision whether such patients should receive adjuvant radiation was based on the presence of adverse prognostic features. Thus, the rate of adjuvant radiation was significantly higher in the myxofibrosarcoma group than in the leiomyosarcoma group (80% vs 53%; P < .001). The type of adjuvant radiation therapy also differed, because significantly fewer patients with myxofibrosarcoma received BRT (20% vs 40%; P = .014).
Of all 202 study patients, 29 patients (14.3%) developed LRs (Table 2). The median time to LR was 8.2 months for the myxofibrosarcoma group and 12.9 months for the leiomyosarcoma group. The initial management of the 17 patients with myxofibrosarcoma who had an LR was biopsy only in 3 patients (18%), wide local excision alone in 6 patients (35%), wide local excision and radiation in 4 patients (22%), and amputation in 4 patients (22%). For the 12 patients with leiomyosarcoma who had an LR, the initial management was biopsy only in 6 patients (50%), wide local excision alone in 2 patients (17%), wide local excision and radiation in 3 patients (25%), and amputation in 1 patient (8%). Six of 17 patients (35%) with myxofibrosarcoma who developed LRs went on to develop subsequent and sometimes multiple LRs at the same site compared with no subsequent LRs in the leiomyosarcoma group (P = .056). In addition, 7 of 17 patients (41%) with myxofibrosarcoma that recurred locally eventually required amputation, compared with 1 of 12 patients (8%) with leiomyosarcoma (P = .09).
Table 2. Univariate Competing-Risks Analysis for Local Recurrence
At a median follow-up of 53 months for event-free survivors, the 5-year cumulative incidence for LR was 14% (95% confidence interval [CI], 8.9%-19%). The LR rate was 14.6% (95% CI, 7.6%-21.6%) for myxofibrosarcoma compared with 13.2% (95% CI, 5.8%-20.6%) for leiomyosarcoma (P = .47) (Fig. 1).
On univariate competing-risk analysis, the only significant predictor of LR for the whole cohort of patients (n = 202) was positive/close margins, with rates of 23.7% (95% CI, 12.5%-34.9%) for positive/close margins versus 9.5% (95% CI, 4.3%-14.7%) for negative margins (P = .016). None of the other factors that we analyzed, including age, tumor size, tumor site, depth, chemotherapy, radiation therapy, or radiation therapy type were significant predictors of LR. Patients who received adjuvant radiation therapy had a significantly higher rate of positive/close margins (39% vs 16%; P = .001), tumors >5 cm (82% vs 33%; P < .001), and deep location (86% vs 47%; P < .001) and were more likely to receive adjuvant chemotherapy (16% vs 2%; P = .003) than those who did not receive adjuvant radiation therapy. In the myxofibrosarcoma group, the type of adjuvant radiation therapy did not significantly impact LR. The 5-year LR rate was 17.6% for patients who received BRT compared with 11.7% for those who received EBRT (P = .359).
Patterns of Local Failure
Despite the lack of any difference in the LR rate according to histology, the patterns of LR did differ (Table 3). For those who underwent surgery alone, an out-of-field LR was defined as an LR outside the tumor bed; and, for those who received adjuvant radiation therapy, it was defined as an LR outside the radiation therapy field. Of 17 myxofibrosarcoma LRs, 8 (47%) were out of field, whereas only 1 of 12 (8%) leiomyosarcoma LRs occurred out of field (P = .04). Of the 8 out-of-field myxofibrosarcoma LRs, 4 occurred after EBRT, 2 occurred after BRT, and 2 occurred in patients who did not receive adjuvant radiation therapy.
Abbreviations: LMS, leiomyosarcoma; LR, local recurrence; MFS, myxofibrosarcoma.
P value was obtained by using the Fisher exact test.
Out of field
Distant metastasis developed in 69 of 202 patients (34%). With a median follow-up of 55 months in event-free survivors, the 5-year cumulative incidence function for distant recurrence was 34% (95% CI, 27%-41%). The influence of histology on distant metastasis was as follows (Fig. 2): For myxofibrosarcoma, the 5-year cumulative incidence was 24.3% (95% CI, 15.8%-32.8%) compared with 45.1% (95% CI, 34.2%-56%) for leiomyosarcoma (P = .014). Other significant factors that were associated with distant recurrence on univariate competing-risks analyses (Table 4) were tumor size >5 cm (44.1% vs 14.7%; P < .001), deep location (37.9% vs 22.7%; P = .034), and positive/close margins (31.6% vs 18.4%; P = .024); whereas a negative pathologic specimen was associated with a reduced risk of distant recurrence (21.8% vs 37.9%; P = .009). On multivariate analysis, tumor size >5 cm with a hazard ratio of 4.05 (95% CI, 2.11-7.78; P < .001), and leiomyosarcoma histology, with a hazard ratio of 2.89 (95% CI, 1.78-4.68; P < .001), retained significant associations with distant recurrence.
Table 4. Univariate Competing-Risks Analysis for Distant Recurrence
At a median follow-up of 48 months, 92 deaths were recorded among the 202 study patients (45.5%). The 5-year survival rate for all patients was 62.4% (95% CI, 54.4%-69.3%). On univariate analyses, significant adverse prognostic factors for survival were tumor size >5 cm (53.4% vs 79.2%; P = .001) and positive/close margins (55.1% vs 65.8%; P = .034); whereas a negative pathologic specimen was associated with an improved 5-year survival rate (73.1% vs 58.9%; P = .034). Histology did not impact survival (Fig. 3), nor did patient age, tumor site, depth, sex, adjuvant radiation, or systemic chemotherapy (Table 5). On multivariate analyses, tumor size >5 cm, with a hazard ratio of 1.72 (95% CI, 1.03-2.87), remained a significant adverse feature.
The LR rates for STS of the extremity from previously reported experiences of pooled histologic subtypes ranges from 10% to 30%.14, 18, 23-26 In the current study, the 5-year incidence of LR among 114 patients with HG myxofibrosarcoma was 14.6% (95% CI, 8.9%-19%), which was not significantly different (P = .59) from the LR rate among 88 patients with HG leiomyosarcoma who were treated during the same era (13.2%; 95% CI, 5.8%-20.6%). The lack of any difference in LR was observed despite the presence of significantly more adverse features in the myxofibrosarcoma group, in which significantly more patients had positive/close margins (42% vs 18%; P < .001) and tumors >5 cm (78% vs 51%; P < .001) compared with the leiomyosarcoma group. Others, however, have reported substantially higher LR rates for myxofibrosarcoma, ranging from 32% to 60%.10-12, 27, 28
Perhaps the greater propensity for LR of myxofibrosarcoma reported in the literature may be related to the infrequent use of adjuvant radiation despite the adverse clinical and pathologic features of myxofibrosarcoma commonly observed at presentation.25, 29 In 1 series from Sweden, just 10% of 109 patients with myxofibrosarcoma who underwent primary surgical treatment received adjuvant radiation. At a median follow-up >5 years, the overall LR rate was 52%.27 In a recently published, large, contemporary series of extremity STS from the National Tumor Institute in Milan, Italy, myxofibrosarcoma histology was associated with the poorest local outcomes. The 10-year estimate for LR was 32%, and the hazard ratio for local relapse-free survival in patients with myxofibrosarcoma was 2.60 (95% CI, 1.38-4.88) relative to patients with leiomyosarcoma. Although those authors did not report the percentage of patients with myxofibrosarcoma who received adjuvant radiation therapy, only 45% of all patients in that study and only 56% of those who had positive or close margins received radiation therapy.10 In contrast, 80% of patients with myxofibrosarcoma overall and 90% of patients with myxofibrosarcoma who had positive or close margins received adjuvant radiation therapy in the current study, yielding a 14.6% 5-year LR rate.
Although LR rates were similar for the myxofibrosarcoma and leiomyosarcoma groups in the current study, the locations of local relapse differed in relation to the surgical bed or radiation therapy field. For the patients who underwent surgery alone, an out-of-field LR was defined as a recurrence outside the tumor bed; and, for those who received adjuvant radiation therapy, it was defined as a recurrence outside the radiation therapy field. Of the 17 LRs in the myxofibrosarcoma group, 47% (n = 8) were out of field compared with only 8% (1 of 12) in the leiomyosarcoma group (P = .04). Another unique feature of myxofibrosarcoma LR was that, once an LR developed, the chance of a subsequent LR was high. In the current study, 6 of 17 patients (35%) with myxofibrosarcoma who recurred locally went on to develop subsequent and sometimes multiple LRs at the same site compared with none in the leiomyosarcoma group (P = .056). This explains, at least in part, why 7 of 17 patients (41%) with myxofibrosarcoma that recurred locally eventually required amputation compared with 1 of 12 patients (8%) who had leiomyosarcoma (P = .09). In addition, the 5-year rate of distant relapse was significantly less in the myxofibrosarcoma group compared with the leiomyosarcoma group (24.3% vs 45.1%, respectively; P = .014), further highlighting the importance of establishing local control. In future studies, more homogeneous populations will be required to identify which patients are most likely to benefit from chemotherapy. Our results suggest that patients with leiomyosarcoma histology may have more potential than myxofibrosarcoma to benefit from novel strategies aimed at reducing systemic relapse.
Taken together, the current results suggest that radiation is an effective adjuvant treatment for patients with myxofibrosarcoma and should be considered in the local management of this disease. Thus, the question is not whether myxofibrosarcoma is radioresistant but, rather, why there is a distinct pattern of LR.
Some experts have suggested that the perceived high propensity of myxofibrosarcoma to recur locally is caused by a characteristic infiltrative growth pattern with extension along vascular and fascial planes.10, 11, 30-32 Authors have called for careful attention to easily overlooked, tail-like extensions of myxofibrosarcoma disease on magnetic resonance images30 (see Fig. 4). In addition, some myxofibrosarcomas appear to be superficial on clinical examination or imaging, but they often involve the investing fascia and, thus, are classified as pathologically deep according to American Joint Committee on Cancer guidelines. In surgical planning for resection of myxofibrosarcoma, it is a challenge to delineate the scope of the disease along fascial planes as well as microscopic extension into the dermis and skeletal muscles.31, 33 Consequently, the entire extent of tumor infiltration may not be encompassed in the resection margin, resulting in high rates of positive margins, as reported here. Similarly, myxofibrosarcoma presents a targeting challenge for radiation oncologists who attempt to ensure that all potential microscopic disease is encompassed in the planning target volume. Recently, there has been increasing interest in using smaller radiation fields to treat STS of the extremity in an effort to decrease both acute and late radiation toxicity while preserving rates of local control.34, 35 These approaches have been facilitated by advances in the delivery of EBRT, particularly with the use of image-guided radiation therapy.36, 37 Probably because they had more adverse features, fewer patients who had myxofibrosarcoma received adjuvant brachytherapy compared with patients who had leiomyosarcoma in our study. Given the small number of highly selected patients with myxofibrosarcoma (n = 18) who received brachytherapy and the smaller volume treated with that technique, we cannot rule out the possibility that the finding of a 6% absolute difference in LR in favor of EBRT, although not statistically significant, could have important clinical implications. The characteristic growth pattern and propensity of myxofibrosarcoma to recur out of field indicate that a reduction in field margins may not be advisable in patients with extremity myxofibrosarcoma. In the future, molecular profiling may enable a better understanding of the patterns of growth and recurrence of this disease, which appears to represent unique tumor biology.38
In conclusion, in this study, compared with patients who had leiomyosarcoma, patients who had myxofibrosarcoma commonly had adverse clinical and pathologic characteristics, including a higher rate of positive margins after resection. Despite these differences, local control was comparable between the 2 groups, likely because of the frequent and effective use of adjuvant radiation. Our results suggest that clinical radioresistance is not inherent to extremity myxofibrosarcoma and that adjuvant radiation should play an important role in the management of this histology.
This study was supported in part by Specialized Programs of Research Excellence (SPORE) in Soft Tissue Sarcoma (Grant P50 CA 140146-01).