Annals of Surgical Oncology

, Volume 18, Issue 3, pp 720–725

Myxofibrosarcoma: Prognostic Factors and Survival in a Series of Patients Treated at a Single Institution

Authors

    • Department of Cancer MedicineIstituto Nazionale Tumori
  • Rosalba Miceli
    • Unit of Clinical Epidemiology and Trial OrganizationIstituto Nazionale Tumori
  • Federica Grosso
    • Department of Cancer MedicineIstituto Nazionale Tumori
    • General Hospital
  • Marco Fiore
    • Department of SurgeryIstituto Nazionale Tumori
  • Elisa Puma
    • Department of Cancer MedicineIstituto Nazionale Tumori
  • Elisabetta Pennacchioli
    • Department of SurgeryIstituto Nazionale Tumori
  • Marta Barisella
    • Department of PathologyIstituto Nazionale Tumori
  • Claudia Sangalli
    • Department of RadiotherapyIstituto Nazionale Tumori
  • Luigi Mariani
    • Unit of Clinical Epidemiology and Trial OrganizationIstituto Nazionale Tumori
  • Paolo G. Casali
    • Department of Cancer MedicineIstituto Nazionale Tumori
  • Alessandro Gronchi
    • Department of SurgeryIstituto Nazionale Tumori
Bone and Soft Tissue Sarcomas

DOI: 10.1245/s10434-010-1341-4

Cite this article as:
Sanfilippo, R., Miceli, R., Grosso, F. et al. Ann Surg Oncol (2011) 18: 720. doi:10.1245/s10434-010-1341-4

Abstract

Background

First described in 1977, myxofibrosarcoma is one of the most common sarcoma subtypes of the elderly. Until some years ago, myxofibrosarcoma was diagnosed as “myxoid malignant fibrous histiocytoma.” The aim of this retrospective case series analysis was to investigate prognostic factors and the clinical outcome of a cohort of patients with myxofibrosarcoma treated at a single institution.

Methods

We reviewed 158 patients with localized myxofibrosarcoma who underwent surgery at the Istituto Nazionale Tumori of Milan, Italy, over 15 years. Local recurrence, distant metastases, and survival were analyzed.

Results

One hundred twenty patients had primary tumors, while 38 patients had locally recurrent tumors. Five-year overall survival was 77%. Tumor size, grade, and margins were statistically significant predictors of survival. Five-year local recurrence and distant metastases rate were 18% and 15%, respectively. Surgical margins were the only statistically significant prognosticator of local relapses. Patients treated with radiotherapy had the same prognosis as nontreated patients, but likely they had worse local presentations. The histological grade correlated with distant recurrences but not with local relapses. The value of adjuvant chemotherapy could not be determined.

Conclusions

Patients with myxofibrosarcoma have a better disease-specific survival than other sarcoma subtypes, but also a higher local relapse rate. This is likely related to the peculiar local growth pattern of these tumors. Adequate surgery should be pursued, while the role of adjuvant therapies need to be investigated.

Myxofibrosarcoma is one of the most common sarcomas of the elderly. It was first described in 1977, and it represents a spectrum of myxoid fibroblastic sarcomas, the high-grade end of the spectrum of what were formerly labeled as “myxoid malignant fibrous histiocytoma”.1,2 Currently, the World Health Organization recognizes myxofibrosarcoma as a distinct entity, marked by reproducible cytoarchitectural features and distinctive clinicopathologic characteristics.3 Most of these lesions arise in the extremity, especially in the lower limbs, but they can occur in the trunk and in head and neck region.4,5 Origin in the retroperitoneum and in the abdominal cavity is extremely rare: a myxofibrosarcoma-like dedifferentiated liposarcoma should be always ruled out.6,7 Compared to other sarcoma subtypes, the prognosis of myxofibrosarcoma is relatively good, with an overall risk of metastases ranging between 20% and 25%.15,8,9

Few large series of myxofibrosarcoma have been reported so far. We therefore analyzed all cases of myxofibrosarcoma treated consecutively at our institution from 1994 to 2009 in an attempt to better understand the natural course of disease and to investigate putative prognostic factors.

Patients and Methods

Between May 1994 and February 2009, a total of 158 consecutive patients with histologically proven localized myxofibrosarcoma underwent surgery with an eradicating intent at Istituto Nazionale Tumori, Milan, Italy. Histologic criteria for myxofibrosarcoma are those reported in the 2002 World Health Organization classification of soft tissue tumors, wherein is recognized the existence of a myxoid sarcoma showing fibroblastic differentiation and forming a spectrum from low to high grade. Important diagnostic clues are represented by the presence of a myxoid background, capillary-sized, archiform blood vessels, and pseudolipoblasts. Unequivocal nuclear atypia represents a key diagnostic clue for diagnosis. The high-grade end of the morphologic spectrum of myxofibrosarcoma corresponds to what has been formerly reported as myxoid malignant fibrous histiocytoma.

Clinical data were retrieved from a prospective database of all adult patients with soft tissue sarcoma and included sex, age at diagnosis, site, size, depth, histological grade, status of margins, adjuvant treatments, time of recurrences, death, or last follow-up. Tumors were labeled as superficial or deep according to the involvement of the investing fascia. All cases were pathologically diagnosed at Istituto Nazionale Tumori. The French Federation of Cancer Centers grading system was applied to the untreated primary tumors.10 Histological grading of recurrent tumors was performed on the slides from the primary, untreated tumor seen in consultation.

An attempt was made to perform a complete resection with negative margins in all patients. The surgical specimen was always examined in the presence of the operating surgeon. Margins were inked and separately sampled. All macroscopically complete resection were classified according to the closest surgical margin, which was microscopically categorized as positive (tumor within 1 mm from the inked surface, R1) or negative (absence of tumor within 1 mm from the inked surface, R0). Whether to provide radiotherapy was decided by both the operating surgeon and the radiation oncologist; a higher-than-average risk of recurrence was supposed on clinical grounds. However, no prospectively selected criteria were used to this end. All clinical decisions were made within the multidisciplinary sarcoma service. External beam radiation was provided at doses ranging from 45 to 70 Gy (median, 60 Gy). Radiotherapy was provided at the same rate for patients with primary tumors and patients with recurrent tumors. Chemotherapy was provided at the discretion of our multidisciplinary team or as part of clinical trials. Anthracycline-based regimens were used, which, in most patients, incorporated ifosfamide.

The study end points were overall survival (OS), sarcoma-specific mortality, local recurrence, and distant metastases. The analysis of sarcoma-specific mortality was performed in a competing risks framework, with deaths due to conditions unrelated to sarcoma regarded as competing events.11 Likewise, when analyzing local relapses or distant metastases, competing events were deaths without evidence of disease and distant metastases or local relapse, whichever occurred first. Concomitant local relapses and distant metastases were computed as distant metastases. Times to event were computed from the date of surgery at our institution or were censored at the date of last follow-up assessment for event-free subjects. As regards OS, descriptive analysis was based on Kaplan–Meier method for survival curves estimation, whereas for sarcoma-specific mortality, local recurrence, and distant metastases, descriptive analyses were based on the estimation of crude cumulative incidence curves.

Multivariable analyses of OS, local recurrence, and distant metastases were based on cause-specific hazards and therefore carried out by Cox regression models. The following putative prognostic factors were studied: tumor size, depth, French Federation of Cancer Centers grade, histotype, surgical margins, and provision of pre- and postoperative chemotherapy and radiotherapy. All factors were modeled as categorical variables by dummy variables.

We considered as significant two-sided P values below the 5% conventional threshold. All statistical analyses were performed by SAS software (SAS Institute, Cary, NC) and R software (R Development Core Team, http://www.r-project.org).

Results

The main characteristics of 158 study patients are summarized in Table 1. The median age at presentation was 64 years, and there was a slight male predominance (56%). One hundred twenty patients (76%) presented with primary tumors, and 38 patients (24%) presented with locally recurrent tumors, 26 of which were making their first recurrence. Sixty-one percent of tumors were located in the lower limbs, 24% in the upper extremities, 15% in the trunk, and 1% in the head and neck region. Negative margins were achieved with surgery in 130 patients (82%). Fifty-six patients (35%) received only radiotherapy as complementary treatment to surgery, and 22 patients (14%) received both radiotherapy and chemotherapy. In our series, size and margins were positively correlated, in that patients with positive margins tended to have tumors of greater size (median, 9.5 cm; interquartile range, 5–16.5 cm) than patients with negative margins (5 cm; 3–7 cm; P < 0.001 at the Wilcoxon–Mann–Whitney test for comparison between the two groups).
Table 1

Patient and disease characteristics

Characteristic

Value

Total

158

Median age (y) (IQ range)

64 (54–72)

Sex, n (%)

 Female

69 (44)

 Male

89 (56)

Presentation, n (%)

 Primary

120 (76)

 Recurrent

38 (24)

Depth, n (%)

 Superficial

69 (44)

 Deep

89 (56)

Tumor size (cm)

 Median (IQ range)

5 (3–9)

 <5, n (%)

67 (42)

 ≥5, n (%)

91 (58)

Tumor site, n (%)

 Upper extremity

37 (24)

 Lower extremity

95 (60)

 Trunk

24 (15)

 Head and neck

2 (1)

Surgical margins, n (%)

 Negative

130 (82)

 Positive

28 (18)

FNCLCC grade, n (%)

 GI

35 (22)

 GII

67 (42)

 GIII

56 (36)

Surgical procedure, n (%)

 Limb-sparing resection

157 (99.4)

 Amputation

1 (0.6)

RT, n (%)

 Not provided

77 (49)

 Provided

81 (51)

CT, n (%)

 Not provided

136 (14)

 Provided

22 (86)

IQ interquartile, FNCLCC French Federation of Cancer Centers, RT radiotherapy, CT chemotherapy

OS and Sarcoma-Specific Mortality

As of February 2009, median follow-up was 53 months (interquartile range, 25–99 months).

In the 158 investigated patients, 33 deaths were recorded, 21 of which were due to sarcoma. Two patients died of local relapse. Figure 1 shows the OS curve for the whole series. Five-year OS was 77% (95% confidence interval [95% CI], 69.4–85.5%). Figure 2a shows the sarcoma-specific mortality crude cumulative incidence curve. Five-year estimate was 16.8% (95% CI, 11.1–25.3%).
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-010-1341-4/MediaObjects/10434_2010_1341_Fig1_HTML.gif
Fig. 1

Kaplan–Meier overall survival curve

https://static-content.springer.com/image/art%3A10.1245%2Fs10434-010-1341-4/MediaObjects/10434_2010_1341_Fig2_HTML.gif
Fig. 2

Crude cumulative incidence of sarcoma-specific mortality (a) and local recurrence and distant metastases (b)

The Cox multivariable model for OS (Table 2) yielded tumor size, histological grade, and status of surgical margins as statistically significant prognostic factors. Figure 3 shows curves according the histological grade. The 5-year estimates in estimates in GI, GII and GIII subgroups were 77.5% (95% CI, 61.7–97.3%), 86.4% (95% CI, 76.6–97.4%), and 64.5% (95% CI, 51.1–81.4%), respectively.
Table 2

Multivariable Cox model analysis of prognostic factors for overall survival

Characteristic

HR

95% CI

Pa

Presentation, recurrent vs. primary

1.6

0.7–3.6

0.291

Depth, deep vs. superficial

1.0

0.4–2.4

0.988

Tumor size (cm), ≥5 vs. <5

3.6

1.3–9.7

0.011

Tumor site, other vs. extremity

2.0

0.6–6.3

0.260

Surgical margins, positive vs. negative

3.1

1.3–7.1

0.009

FNCLCC grade, III vs. I–II

4.2

1.9–9.1

<0.001

RT, yes vs. no

1.5

0.7–3.3

0.342

CT, yes vs. no

0.8

0.3–2.2

0.619

HR hazard ratio, 95% CI 95% confidence interval, FNCLCC French Federation of Cancer Centers, RT radiotherapy, CT chemotherapy

aTwo-sided Wald test

https://static-content.springer.com/image/art%3A10.1245%2Fs10434-010-1341-4/MediaObjects/10434_2010_1341_Fig3_HTML.gif
Fig. 3

Kaplan–Meier overall survival curves according to histologic grade

Local Recurrence

Thirty patients experienced local recurrences after surgery at our institution, including 27 as their first event (20 patients with primary disease and 7 patients with local recurrences at presentation). Fifteen had only one local recurrence, and 14 patients had two or more local recurrences. Twenty-four patients had local recurrences only, whereas 3 patients had both local recurrences and distant metastases. Time to local recurrence varied from 2 months to 8 years, with a median time to recurrence of 27 months. Figure 2b shows the local recurrence crude cumulative incidence curve; 5-year estimate was 18.2% (95% CI, 12.3–26.9%).

The Cox multivariable regression model (Table 3) yielded significant result only for the status of surgical margins.
Table 3

Multivariable Cox model analysis of prognostic factors for local recurrence

Characteristic

HR

95% CI

Pa

Presentation, recurrent vs. primary

0.5

0.2–1.4

0.173

Depth, deep vs. superficial

1.1

0.5–2.8

0.793

Tumor size (cm), ≥5 vs. <5

0.7

0.3–1.7

0.437

Tumor site, other vs. extremity

1.0

0.3–3.6

0.999

Surgical margins, positive vs. negative

6.1

2.2–16.6

<0.001

FNCLCC grade, III vs. I–II

0.8

0.3–2.2

0.609

RT, yes vs. no

1.1

0.5–2.6

0.753

CT, yes vs. no

1.5

0.4–5.1

0.554

HR hazard ratio, 95% CI 95% confidence interval, FNCLCC French Federation of Cancer Centers, RT radiotherapy, CT chemotherapy

aTwo-sided Wald test

Distant Metastases

Twenty-three patients had distant metastases after surgery at our center, including 15 patients with primary disease and 8 patients with local recurrence at presentation. Distant metastases as a first event occurred in 12 patients with primary disease and in 8 patients with recurrence at presentation. Fifteen (65%) of 23 patients developed pulmonary metastases, 5 of which were associated with extrapulmonary metastases. Eight patients (3%) had extrapulmonary metastases only. The sites of extrapulmonary metastases were lymph nodes in 4 (20%) of 23 patients, bone in two patients, liver in one patient, and multiple sites (including soft tissue and abdomen) in 6 patients. The time to distant metastases varied from 2 month to 9 years, and the median time for those who developed distant metastases was 11 months.

Figure 2b shows the distant metastases crude cumulative incidence curve; the 5-year estimate was 14.6% (95% CI, 9.6–22.0%). By multivariable analysis (Table 4), only tumor grade was statistically significant. However, an unfavorable trend for status of surgical margins was also detected (P = 0.057).
Table 4

Multivariable Cox model analysis of prognostic factors for distant metastases

Characteristic

HR

95% CI

Pa

Presentation, recurrent vs. primary

2.0

0.7–5.2

0.178

Depth, deep vs. superficial

1.4

0.5–4.1

0.587

Tumor size (cm), ≥5 vs. <5

2.1

0.6–6.9

0.245

Tumor site, other vs. extremity

0.9

0.2–4.0

0.844

Surgical margins, positive vs. negative

2.9

1.0–8.6

0.057

FNCLCC grade, III vs. I–II

5.1

1.8–14.1

0.002

RT, yes vs. no

1.2

0.4–3.2

0.744

CT, yes vs. no

1.5

0.4–5.1

0.748

HR hazard ratio, 95% CI 95% confidence interval, FNCLCC French Federation of Cancer Centers, RT radiotherapy, CT chemotherapy

aTwo-sided Wald test

Discussion

In this series of 158 patients with myxofibrosarcoma surgically treated at our institution over a 15-year time span, the overall 5-year sarcoma-specific mortality, local recurrence, and distant metastases rates were 16.8%, 18.2%, and 14.6%, respectively. Size and histologic grade of the primary tumor as well as the status of surgical margins were independent predictors of survival. Local recurrences were predicted by quality of surgical margins and distant metastases by histologic grade.

The incidence of local recurrences was definitely higher than that of all other soft tissue sarcomas, which we recently reported to be 10%.9 The most likely explanation is that myxofibrosarcomas frequently exhibit a multinodular infiltrative growth pattern with a tendency to quite extensively involve surrounding tissues. In this sense, gross identification of the true extent of disease may be difficult for the surgeon, so that surgery may prove inadequate. These tumors lack a pseudocapsule and may display a nonpalpable spread along muscle bundles and fascial planes, as well as superficial tissues. An accurate pathologic evaluation of the quality of surgical margins helps distinguish patients with a higher likelihood of local failure.1214

Tumor grade, as in most soft tissue sarcoma subtypes, predicted the risk of developing distant metastases, but not local failure. Actually, myxofibrosarcomas represent a spectrum of myxoid fibroblastic malignancies. At one end of this spectrum, well-differentiated forms have been suggested to be incapable of metastatic spread. Our findings confirm this view, as no G1 case in our series developed systemic metastases. Among patients who eventually developed metastatic disease, 14 had a G3 and 7 a G2 myxofibrosarcoma. The remaining two cases had a primary G1 tumor, but in these patients, distant spread was the consequence of three local recurrences, which displayed features of high-grade switch.

Interestingly, 20% of patients who developed metastatic disease had positive regional lymph nodes beforehand. In 4 patients with regional node disease, 3 developed distant metastases.

Mortality was obviously associated with histological tumor grade, but also to quality of surgical margins.1214 As suggested in some series, an explanation is that proximally located tumors can give rise to then-inoperable local regional recurrences, which in turn may lead directly to death.9 In this series, 2 of 21 patients died of local recurrence. To improve survival, it is imperative that surgical resection of these risky tumors achieves disease-negative microscopic margins. A multidisciplinary approach, resorting to plastic reconstructive surgery whenever needed, is then critical, particularly at presentation. In distal tumor sites, the quality of microscopic surgical margins might be less crucial for survival, but it still predicts the risk of further failures that may eventually lead to the loss of a limb. Finally, the potential for dedifferentiation should not be underestimated, given its implications for systemic risk.

As far as tumor grade is concerned, the main difference in outcome was observed between grade I and grade III tumors, while less relevant differences were evident between grade I and II tumors (Fig. 3).

Size did not correlate with the risk of local recurrence, even if there was a positive trend, likely for the correlation between size and quality of surgery.

In our series, radiotherapy seemed to have only a limited value in reducing the risk of a local recurrence. However, for obvious reasons, patients who received radiotherapy were unfavorably selected, the choice being a clinical one performed on an individual basis. In fact, they had worse prognostic factors, such as tumor size of >5 cm, deeper location, and high histological grade. Thus, we cannot rule out that radiotherapy did actually reduce the risk of local recurrences in a subgroup of patients.

Adjuvant chemotherapy (which was used in 14% of these patients) did not prove an independent prognostic factor. As is well known, its role in soft tissue sarcomas is still debated, and this series cannot help resolve this debate.15,16

Our data support the concept that myxofibrosarcoma carries a better prognosis when compared with other sarcoma subtypes.8,9 In particular, even grade III myxofibrosarcoma exhibits a better outcome that other high-grade pleomorphic sarcomas. In our series, the OS at 5 years approaches 65% for high-grade lesions. This confirms preliminary data published by Fletcher et al. in 2001.8 In fact, in his retrospective study for a reclassification of malignant fibrous histiocytoma, it was suggested that myxofibrosarcoma has a lower risk of metastases in comparison to other sarcoma subtypes. In the same study, it was first reported that myogenic differentiation in pleomorphic sarcoma is associated with a worse prognosis. All this further stresses the importance of a proper pathologic diagnosis of soft tissue sarcomas.

In conclusion, this study reinforces the notion that myxofibrosarcoma features a better prognosis, even when a high-grade component is present, as compared with other soft tissue sarcomas. In particular, the risk of distant metastases is definitely lower. By contrast, the rate of local recurrences is higher. Although the role of adjuvant treatments is undefined, efforts should be made to perform adequate surgical resections and balance benefits and sequelae. Achieving negative microscopic margins should be the primary goal of treatment, leaving the biology of the disease to exclusively influence the final outcome.

Conflict of interest

The authors declare no conflict of interest.

Copyright information

© Society of Surgical Oncology 2010