Treatment strategies for pediatric soft tissue sarcomas differ somewhat from those for adults. In a child, avoidance of radiotherapy is always preferred and surgical excision is the major therapeutic modality. While radiation reportedly enhances local control [11], it is associated with greater risks including growth retardation and secondary radiation-induced malignancy in children compared with adults. Adjuvant or neoadjuvant chemotherapy improves overall survival and is considered the standard of care of patients with rhabdomyosarcoma and Ewing’s sarcomas/primitive neuroectodermal tumor (PNET) [3, 5], whereas the value of chemotherapy in adult sarcoma populations and pediatric nonrhabdomyosarcomas is less clear [10].

Sarcomas in extremities also have a differing spectrum of diagnoses and prognoses than those located elsewhere. For example, nonextremity sarcomas have higher mortality than those located in extremities [6]. When sarcomas are not surgical by limb preservation resections, radiotherapy and amputation are alternative choices for obtaining local control.

Because soft tissue sarcomas in children are rare and histologically heterogeneous, few large studies document clinical behaviors or outcomes; thus, it is difficult to ascertain standardized therapy for pediatric soft tissue sarcomas. There are several case series for patients with soft tissue sarcomas [11, 18, 19] but none that specifically determine what factors influence recurrences, metastases, and survival in childhood extremity sarcomas. However, based on the literature, we would expect (1) positive margins after surgical resection would be associated with an increase in local recurrence [1, 17] and that radiation therapy would reduce local recurrence [2]; that (2) adjuvant chemotherapy would reduce the rate of metastasis [14] and that local recurrence would be associated with an increase in the rate of metastasis [8]; and that (3), local recurrence [20] and metastasis at diagnosis [15] are risk factors for death.

Demographic data collected included gender, age at the time of diagnosis, and duration of followup. Tumor characteristics were tabulated with tumor size (less than 5 cm or 5 cm or greater), the site of primary tumor (upper proximal, upper distal, lower proximal, lower distal), depth (deep, superficial to fascia), invasiveness into bone or neurovascular structures (yes or no), histologic subtypes, and metastasis at diagnosis (yes or no). Tumor size was determined by the imaging studies at diagnosis using maximal diameter measured on CT or MRI. If patients had unplanned excision elsewhere, the tumor size was determined by outside images if available. Thirty-six percent of patients presented with tumors larger than or equal to 5 cm, whereas 64% had tumors less than 5 cm. We reviewed the pathology report when tumors were excised without preoperative imaging studies. We did not examine the histologic grade of the tumors, but the majority of the tumors were high-grade and 19 patients had low-grade tumors. We defined a tumor as invading bone or neurovascular structures if there was either gross or microscopic involvement by tumor at surgery or subsequent pathologic analysis. The most common histologic subtypes were synovial sarcoma (29%) followed by rhabdomyosarcoma (16%). Histologic diagnosis was confirmed by two pathologists (AR, AP-A) experienced in musculoskeletal pathology.

We recorded surgical procedures as well as adjuvant treatments administered. Data of unplanned tumor excision elsewhere before the initial visit (yes or no), surgical procedures (limb-sparing, amputation, none), chemotherapy (yes or no), and radiotherapy (yes or no) were collected. The indication for radiation varied over time and between institutions. Preoperative radiotherapy was used in instances when it was decided that a wide surgical margin could not be achieved without sacrificing neurovascular structures or important bony segments. Those who had surgical excision with positive margins were treated with either radiation or additional surgery. For patients who underwent definitive surgical resection of the primary tumor, we recorded microscopic margins (positive, negative). Surgery was the primary treatment in 94 patients. Limb-sparing surgery was performed in 90 patients (Table 2). Four patients were treated with amputations proximal to the wrist or ankle. Of 94 patients who were treated with surgical resection or amputation, 70 patients had microscopically negative margins. All patients treated by amputation had widely negative margins. Four patients did not receive surgical treatments, because one patient had unplanned excision with negative margins and three had metastasis elsewhere and surgery was not indicated.

Of 84 patients who did not have metastasis at initial diagnosis, seven patients developed distant metastasis. The median time to development of metastasis was 20.9 months (range, 6–142 months). Two of these patients had distant metastasis that occurred synchronously with or subsequent to local recurrence. More patients (p = 0.02) developed metastasis with tumor in lower extremities compared with those in upper extremities. In the univariate Cox regression analysis, local recurrence was a major adverse prognostic factor for metastasis with a rate ratio (RR) of 3.13 (95% confidence interval [CI], 0.58–16.9; Table 4); however, when controlling other patient characteristics, including categorical age and gender in the regression model, local recurrence did not increase the rate for metastasis.

We acknowledge limitations in this study. First is the limited number of patients with each histologic subtype. It would be ideal to evaluate different pathologic subtypes separately; however, pediatric soft tissue sarcomas are exceedingly rare and the numbers of each subtype are even rarer. We found the histologic subtypes did not predict local recurrence, metastasis, and survival but the chance of a Type II error is substantial. Nonetheless, ours is a relatively large series of soft tissue sarcomas in extremities in children and except for the caveat of histologic subtypes, 98 patients offered sufficient numbers to identify differences in other prognostic factors. Second, other reports have emphasized the histologic grade of the tumor; we did not use histologic grade as a potential prognostic factor. Histologic subtypes strongly correlate with grade. For example, all synovial sarcomas and rhabdomyosarcomas are considered high-grade. If we had used histologic grade as a prognostic factor as well, this would have induced collinearity in the analysis. Instead, we evaluated each histologic subtype as a variable because some histologic subtypes have known clinical outcomes that differ from others. For example, one adult study found leiomyosarcoma independently predicted metastasis and malignant peripheral nerve sheath tumor predicted overall survival [15]. Third, followup time for each patient varied in the study. This is because we evaluated risk factors for local recurrence, metastasis, and death using Cox regression and person-time data; we enrolled all patients seen between the study periods to avoid selection bias. The median time for followup was 61.5 months and it should be sufficiently long because the median time to develop local recurrence was 18.6 months. The median was 20.9 months for metastasis and 28.0 months for death.

We found local recurrence did not increase the rate for metastasis. The causal relationship between local recurrence and metastasis is controversial [23]. The relationship may be different for tumors in extremities than for tumors elsewhere because local recurrence can usually be treated with amputation for tumors in extremities, whereas radiation therapy may be the only modality for unresectable tumors in the trunk and other nonextremity sites. In our series, chemotherapy did not reduce the rate for metastasis.

We found only metastasis at diagnosis predicted death. Metastasis at presentation is reportedly a risk factor for death for synovial sarcoma [12] and adult soft tissue sarcoma of extremities [15]. One study suggested tumor size and positive surgical margin predicted death in truncal and retroperitoneal tumors [17]. In our series limited to the extremities, these and other factors may not have predicted death for two reasons. One is sarcomas in extremities can be excised with amputations if needed to achieve local control and that characteristics of the tumor itself do not affect the overall survival. Second is the limited power of this study to detect differences resulting from a small number of patients.

Radiation therapy reduced the rate of local recurrence. This result is consistent with previous literature [2]. Although we did not find any radiation-related complications in our patients, in a child, avoidance of radiotherapy is always preferred if adequate local control can be achieved by surgical resection alone. However, in the treatment of high-grade soft tissue sarcomas, especially those that are large and or inadequately resected, the addition of radiotherapy with surgery can maximize local control while avoiding amputation. Conformal techniques for treatment have been in use for many decades and these can limit toxicities. However, more technologic advances, including intensity-modulated radiation therapy and proton beam radiation therapy, can decrease further the late effects by minimizing the normal tissue exposure. Microscopically positive margins have been reported to increase the risk for local recurrence [1, 17]. In our study, positive margins did not increase the rate for local recurrence in multivariate analysis. We are relatively aggressive in using adjuvant therapies in patients with positive margins. Most such patients were treated with either another tumor bed excision or radiation therapy. In patients who received tumor bed excision or radiation therapy, a positive margin may lose its effect on local recurrence. Radiation therapy was effective in controlling local recurrence in our series. Surprisingly, those who received unplanned excision at outside facilities had a lower rate for local recurrence compared with those who received initial surgical treatment at our hospitals. In the literature, those who were referred to a cancer center and underwent a second operation had an improved disease-free survival rate [7]. This observation may be the result of the likelihood that large tumors in deep locations were more likely referred to us without any surgical treatments than smaller superficial tumors. Therefore, the data were reanalyzed adjusting for tumor size and depth in addition to age and gender and still patients who underwent unplanned excision experienced considerably fewer local recurrences (data not shown). Another possible explanation is most of patients who had unplanned excisions were treated with tumor bed excisions at our facilities, which achieved wider margins, but the precise explanation of this observation remains unclear. Of note, tumors located superficial to the fascia are not always benign as suggested by the observation that one-fourth of sarcomas located superficial to the fascia in our series were malignant. It is reported that duration of symptoms before diagnosis does not predict the sarcoma size and prognosis [16]. Although we did not find a higher local recurrence rate for those who had unplanned excisions, superficial tumors without longstanding symptoms should be treated with caution.