Felipe A. Calvo and Claudio V. Sole contributed equally to this work.

Successful treatment of high-risk soft tissue sarcomas (STS) of the extremities remains challenging in the 21st century [1]. External-beam radiation therapy (EBRT) combined with limb-sparing surgery local control (LC) rates are comparable to those achieved with amputation [2]. The benefit of adding EBRT to limb-sparing surgery has been addressed in two randomized trials; both showed that combined treatment reduced the risk of local recurrence by 20–25 % when compared to limb-sparing surgery alone [3, 4]. Thus, the use of adjuvant EBRT maximizes functional and cancer outcomes without the significant morbidity and cosmetic deformity of radical surgery [4]. Margin status has been reported to be the most important prognostic factor for LC even in patients treated with combined surgery and radiotherapy [5]. Complete negative margin resection can sometimes not be achieved due to close proximity or proven invasion into adjacent unresectable structures [6, 7]. Therefore, higher EBRT doses are sometimes used to compensate for close or positive margins.

Total escalated doses of radiotherapy that can be delivered even with the most sophisticated and updated EBRT precision techniques is limited by the presence of dose-limiting surrounding organs or structures in the planning treatment volume (PTV) [8]. Accordingly, high radiation therapy doses delivered with EBRT have often been associated with significant late toxicity [9]. Because a cumulative EBRT dose of 64 Gy or greater is needed in the positive and/or close margin setting [10], this scenario is an ideal situation to consider intraoperative electron-beam radiation therapy (IOERT) as a component of treatment, as this modality has the advantage of delivering a high boost dose to deep-seated sarcoma residues or risk surgical bed areas adjacent to radiosensitive critical organs by mobilizing these structures temporarily out of the radiation field [11]. Since 1986, three Spanish institutions have approached the treatment of patients with extremity STS using an IOERT-boost component in high-risk areas (post-resection and pre-reconstruction), EBRT and limb-sparing surgery. In this study, a joint analysis of data from three institutions was performed in order to evaluate, on a large and mature cohort of patients, evidence of the contribution of an IOERT-containing multimodality approach in promoting LC with acceptable tolerance.

Materials and methods

Patient characteristics and staging evaluation

From June 1986 to April 2012, patients aged ≥ 16 years (Karnofsky performance status ≥ 70) with pathologically confirmed [macroscopically resected (non-R2)] nonmetastatic extremity STS were eligible for multimodal treatment. Patients (n = 159) with primary (nonrecurrent) tumours, with either close (< 1 cm) and/or positive surgical margins (limb-preserving surgery) underwent EBRT and IOERT. Additionally, during the study period, 95 patients (with margins ≥ 1 cm) were treated exclusively with surgical resection and postoperative EBRT. Patients with the diagnosis of desmoid tumour, dermatofibrosarcoma protuberans, rhabdomyosarcoma and peripheral neuroectodermal tumour were not included in the study, the former two because of the often indolent clinical course, and the latter two from the well-known radioresponsiveness. Pretreatment evaluation consisted of a complete history and physical examination, complete blood count, renal and liver function tests, chest X-ray, and computerized tomography (CT) or magnetic resonance imaging (MRI) of the tumour site, chest and abdomen. Data were prospectively collected and retrospectively analysed at the time of scheduled follow-up. Patients were reclassified according to the 7th AJCC/UICC staging system for the analysis. Patient and treatment characteristics are listed in ◉ Table 1; there were no significant differences in baseline variables between the patients treated for lower and upper extremity STS. The protocol followed the recommendations of the Declaration of Helsinki. The Institutional Ethics Committee approved the protocol, and signed informed consent was obtained from all patients.

Table 1 Patient, tumour and treatment characteristics
Full size table

Treatment characteristics

Details of EBRT technique, IOERT and adjuvant chemotherapy (CT) followed previously described standards [11]. A total median EBRT dose of 45 Gy (range 40–54 Gy; 1.8–2.0 Gy/5 days/week) was applied postoperatively [79 %, 45 Gy (range 40–54 Gy)] or preoperatively [21 %, 45 Gy (range 40–50 Gy)] and delivered with megavoltage equipment (6 to 15 MV) using a three-dimensional (3D) conformal field technique. The technique for the EBRT component consisted of conventional (2D-RT) EBRT for patients treated between 1986 and 1992 (n = 28, 18 %) and conformal (3D-CRT) EBRT for patients treated after 1992 (n = 131, 82 %). PTV for 2D-RT was defined as tumour bed plus 3 cm in the radial directions in all cases, and for the longitudinal directions a 5 cm margin was applied. Clinical target volume (CTV) for the 3D-CRT technique included the surgical tumour bed plus a 2 cm margin in the radial directions and a 3 cm margin in the longitudinal (proximal and distal) directions, while the PTV was defined as CTV plus a 1 cm margin in the longitudinal and radial directions. At least one third of the circumference of the extremity was spared from irradiation to prevent development of chronic lymphedema. Surgical procedures (4–6 weeks before postoperative or after preoperative treatment) were categorized as marginal resection [(n = 19, 12 %) defined as resection through the tumour pseudocapsule or surrounding reactive tissue] or wide [(n = 140, 88 %) resection including normal tissue]. In all, 40 patients (25 %) underwent a tumour bed re-excision after prior excision, excisional biopsy, or intralesional surgical procedure. The remaining 119 patients (75 %) underwent a single attempt at definitive resection after incisional or core needle biopsy. For patients who had more than one procedure, the most radical procedure is listed. The IOERT program was performed in a non-dedicated linear accelerator with outpatient radiotherapy activity by the three institutions. After surgery and before reconstruction, 10–20 Gy (median 12.5 Gy) were delivered in a single fraction to one- (n = 128, 81 %) or two-field (n = 31, 19 %) PTVs, using a median energy of 6 MeV (range 4–20 MeV) (◉ Table 2). Dose was prescribed to the 90 % isodose line, covering the entire surgical bed. The intraoperative margin status was assessed using frozen pathologic sections. The IOERT dose was chosen according to the margin status and surgical bed volumes. Beveled (15–45º) Lucite circular applicators (size range 5–15 cm) were adjusted to collimate the target surface air gap, allowing dosimetric adaptation and uniform dose distribution. CT-guided treatment has been available since 2008 for IOERT planning [12]. Patients with higher histologic grade (grade 3) and tumour size (≥ 5 cm) were offered adjuvant CT (most commonly CT consisted of 4 or 5 cycles of doxorubicin 75 mg/m2 and ifosfamide 5 g/m2, every 3 weeks).

Table 2 Correlations between macroscopic/microscopic pathology characteristics and IOERT technical parameters
Full size table

Follow-up and toxicity evaluation

All patients were required to be followed according to a common protocol every 3 months after treatment completion for the initial 3 years and every 6 months for 3 additional years thereafter. Patients were restaged 4 weeks after EBRT and routinely every 6 months with chest X-ray, and CT or MRI of the initial tumour site. Acute and late toxicities were evaluated according to Radiation Therapy Oncology Group/European Organization for Research and Treatment of Cancer score [13].

Statistical analysis

The collected data were analyzed using SPSS (version 19.0) statistical software. The primary endpoint was LC. Secondary endpoints included IOERT in-field control, disease-free survival (DFS) and overall survival (OS).

The Kaplan–Meier method was used to estimate LC, IOERT in-field control, DFS and OS probabilities (all time-to-event end points were defined as the time from treatment initiation to event or the day of last follow-up). For survival outcomes potential associations were assessed in univariate and multivariate analysis using the Cox proportional hazards model. Based on, first, p values ≤ 0.10 in univariate analyses, and second, on clinical relevance, multivariate analysis was performed using a stepwise regression model to identify variables that have an effect (two-sided p test ≤ 0.05) on survival outcomes.

Results

Median follow-up time for all patients was 53 months (range 4–316 months). A total of 118 patients were alive at the time of analysis. Median follow-up for surviving patients was 67 months (range 4–316 months). Of the 41 deceased patients, 37 (90 %) died from cancer progression, and 4 (10 %) died from causes unrelated to their tumours or treatment. Crude local relapse (LR) rate was 16 % (n = 25), IOERT in-field rate was 12 % (n = 19) and 30 % (n = 48) developed distant metastases [most commonly pulmonary (n = 29, 60 %)]. Of the 25 patients who had local progression, 8 (5 %) were rescued with extremity amputation. Actuarial 5-year amputation-free survival was 94 %. The other 17 patients (15 had synchronic distant metastases) with local relapse underwent wide excision (n = 1), received chemotherapy alone (n = 14), or received no therapy (n = 2).

Actuarial local control for the study population at 5 and 10 years was 82 and 81 % (Supplemental Fig. 1a). Univariate Cox proportional hazard analyses showed that an R1 resection (Fig. 1e (p = 0.001)) and marginal excision (p = 0.05) were associated with a higher probability of LR (◉ Table 3). After adjustment for other covariates only R1 resection (p < 0.001) remained significantly associated with LR (◉ Table 4). We then evaluated patients with upper and lower extremity STS separately. For the subset with lower extremity STS (86 %), patients with R1 resections experienced a significantly higher risk of LR in univariate analysis (HR 3.67; 95 %CI 1.54–8.76; p = 0.003). Alternatively, for the subset with upper extremity STS (14 %), univariate analysis did not show that patients with R1 resection status had an increased risk of LR (HR 2.02; 95 %CI 0.54–7.02; p = 0.35). Actuarial IOERT in field-control at 5 and 10 years was 86 and 85 % (Supplemental Fig. 1b). Univariate analyses showed that R1 resection (p = 0.004) was associated with a higher probability of IOERT in-field relapse (◉ Table 3). An IOERT boost dose ≥ 12.5 Gy (Fig. 1f; p = 0.01) was associated with a lower probability of IOERT in-field relapse. In multivariate analysis an R1 resection (p = 0.001) and higher IOERT dose (≥ 12.5 Gy) retained a significant association with regard to IOERT in-field relapse (◉ Table 4). When IOERT dose was evaluated separately in patients with R0 and R1 resection margin status, we found that only for the subset of patients with R0 resections (84 %), receiving an IOERT dose ≥ 12.5 Gy was associated with a lower probability of IOERT in field relapse (HR 0.14; 95 %CI 0.02–0.98; p = 0.05). Actuarial DFS at 5 and 10 years was 62 and 57 % (Supplemental Fig. 1c). Univariate Cox proportional hazard analyses showed that stage III (p = 0.02) and R1 margin status (p = 0.004) were associated with a higher probability of overall metastases (◉ Table 3). After adjustment for other covariates stage III (p = 0.008) and R1 resection (p = 0.001) retained a significant association with DFS (◉ Table 4). Actuarial OS at 5 and 10 years was 72 and 64 % (Supplemental Fig. 1d). On univariate analysis, only stage III patients (p = 0.04) were at a significantly higher risk of overall death (◉ Table 3). We found on multivariate analysis that stage III (p = 0.04) and age ≥ 50 (p = 0.05) were significantly associated with OS (◉ Table 4).

Table 3 Univariate analyses of associations between the patient, tumour and treatment with locoregional control, IOERT in-field control, disease-free survival and overall survival
Full size table
Table 4 Factors associated with locoregional control, IOERT in-field control, disease-free survival and overall survival in multivariate analyses
Full size table

Overall 23 patients (14 %) had grade ≥ 3 acute toxicity [severe skin reactions (n = 14, grade 3) and wound-healing disturbances (n = 8, grade 3; n = 1, grade 4)]. Sixteen patients (10 %) developed grade ≥ 3 chronic toxicity [neuropathy (n = 4, grade 3; n = 2, grade 4), necrosis/fistula/ulcer (n = 1, grade 3; n = 1, grade 4), joint function impairment due to fibrosis (n = 4, grade 3) and severe chronic lymphedema (n = 2, grade 3)]. No perioperative or long-term death from treatment occurred. In relation to acute [14 % (n = 19) vs. 17 % (n = 4); p = 0.41)] and chronic toxicity [10 % (n = 13) vs. 13 % (n = 3); p = 0.65)] no differences between patients with upper and lower extremity STS were observed.

Discussion

To our knowledge, this is the largest reported study that focuses on the outcomes of patients with primary STS treated with IOERT and EBRT. Discrimination between patients receiving treatment for an initial diagnosis and recurrence is important because it has been consistently reported that patients treated for LR have worse overall outcomes (◉ Table 5).

Table 5 Series of extremity soft tissue sarcomas treated with external beam radiotherapy and intraoperative radiation therapy. Stratified by margin (R0 vs. R1) and disease status (primary vs. recurrent)
Full size table

Our relevant findings can be summarized as follows. First, in a group of patients with high-risk features for local relapse (all incomplete or close margin resections), the 5-year LC and OS rates of 82 and 72 % compare well with more favourable cohorts of patients treated with limb-preserving surgery and EBRT without IOERT [5-year OS (71–87 %) and LC (72–96 %)] [2–6]. Second, we found that an IOERT dose ≥ 12.5 Gy reduces the risk of IOERT in-field relapse. Interestingly, this maintained significance when patients with complete resection (R0) were analysed separately. Finally, we found that patients with an R1 resection had an increased probability of local and distant relapse that could not be compensated by a moderate IOERT boost to the high-risk region. In a subgroup analysis margin status retained significance with regard to LC only for patients with lower extremity tumours.

Several groups have successfully implemented and reported combined management (IORT and EBRT) for patients with extremity sarcomas [5-year LC (73–95 %) and OS (70–80 %)] [14–18]. Although margin status is a common listed risk factor for local recurrence, what constitutes adequate surgical margins is not well defined. Positive surgical margins have been consistently reported as an adverse prognostic factor for LC (◉ Table 5). Oertel et al. [15] reported the largest single institution experience with IOERT plus EBRT for the management of extremity STS (n = 153). Although detailed data on the site of recurrence and rescue are not provided, LC was more favourable for patients receiving an IOERT dose ≥ 15 Gy (5-year LRC 85 vs. 50 %, p = 0.003) and complete margin resection (5-year LRC 85 vs. 60 %, p = 0.03). Azinovic et al. [16] analysed 45 patients with extremity sarcomas (58 % primary tumours) treated with postoperative EBRT (45 to 50 Gy) and IOERT. The 5-year local control was 87 % and margin status [negative or close margins vs. positive margins (p= 0.04)] significantly affected LC. Consistently, we also found that patients with an IOERT dose ≥ 12.5 Gy achieved better IOERT in-field control. In the current analysis positive microscopic resection margins (16 %) was the only factor that remained significantly associated with LC in the multivariate analysis. Additionally, the lack of the significant association of margin status with LC in the upper extremity and IOERT dose with IOERT in-field control in patients with R1 resection margin are likely due to small patient numbers in those subgroups. However, margin resection status may have a different prognostic impact in different settings [17]. Call et al. [17] analyzed 61 patients (treated with EBRT plus IORT) with upper extremity STS by margin status. The patients with positive margins had similar prognoses to patients with negative margins (5- and 10-year LC rates 100 % and 86 % vs. 89 % at both; p = 0.98). Likewise in the current analysis, margin status had no impact on LC for patients with upper extremity STS. In contrast for the subset of patients with lower extremity STS we found that R1 margins status was associated with an increased chance of LR.

Dickie et al. [19] examined the geometric relationship between LR and EBRT volumes of 768 STS patients treated with function-preserving surgery. Sixty (7.8 %) STS patients developed LR, 49 tumors relapsed in-field (6.4 % overall), 9 out-field (1.1 % overall) and 2 were marginal (0.3 % overall). Because the majority of STS tumours reoccur in-field, these data support that an accurate delivery of a higher radiation dose could potentially improve LC in select patients with limb STS. Al Yami et al. [7] reported no benefit for adding a postoperative EBRT boost for extremity STS patients treated with preoperative radiotherapy. This treatment strategy has several disadvantages such as a significant delay before boost delivery (potentially allowing tumour repopulation and systemic dissemination), diminished effectiveness due to tumour bed hypoxia after surgery and increased treated volume. An IOERT boost has several advantages over an EBRT escalated strategy such as a more precise delivery of radiation to a surgically identified high-risk area, mobilization of dose-sensitive critical organs temporarily out of the radiation boost field and to shorten overall treatment time (dose–dense radiotherapy). The observation that patients receiving a higher IOERT boost dose (≥ 12.5 Gy) had a decreased chance of IOERT in-field LR in the current analysis is an argument in favour of an IOERT dose escalation strategy, by implementing field within a field technique (Supplemental Fig. 2).

Distant metastases remain as the dominant pattern of progression for high-risk extremity STS [3, 20]. Although the effect of adjuvant CT on survival for resected soft-tissue sarcoma remains to be recognized [21], intensified local treatment needs to be tested in the context of more efficient concurrent, neo-, and adjuvant systemic therapy.

Concerning treatment-related toxicity, a treatment regimen that included IOERT for extremity sarcomas was tolerable for our 159 patients. The low rate of severe toxic events and the high limb preservation rates (95 % limb preservation) suggest that a multimodality approach with EBRT and an IOERT-boost component is feasible with acceptable risks and without prohibitive long-term side effects [22].

We acknowledge several limitations of our study. First, on average only two patients were treated per institution per year [study period is very long (almost 25 years)]. Second, the population was heterogeneous, receiving different treatment combinations, sequences and doses. Radiation therapy technology, treatment guidelines and surgical consensus have changed over time, and cannot be completely assessed in all three hospitals. Third, although we did observe a significant association between IOERT dose and IOERT in-field control after adjustment for several potential confounding factors, we certainly acknowledge the presence of a selection bias for patients referred for radiation therapy to a higher dose. Fourth, a systematic method of follow-up, including imaging, would be optimal to evaluate patterns of failure after radiation therapy. Given the retrospective nature of this analysis, consistent homogeneous imaging did not occur in a proportion of patients. Finally, we acknowledge the limitation that this series does not compare boost to no boost, and it is therefore difficult to assess the benefit of the IOERT boost.

Conclusion

We found that patients with extremity STS receiving EBRT and IOERT could be treated safely and had high LC rates. In addition, patients with radical resections experienced the largest benefit of a higher IOERT dose. A level of adverse prognostic features (R1 resections) might be compensated in upper extremity STS. Our results suggest that patients with close or positive margins could benefit from further intensified local treatment strategies.

Compliance with ethical guidelines

Conflict of interest

F. A. Calvo, C. V. Sole, A. Polo, M. Cambeiro, A. Montero, A. Alvarez, M. Cuervo, M. S. Julian, R. Martinez-Monge state that there are no conflicts of interest.

This work was supported in part by a grant from the Health Institute of Research Carlos III, Spanish Ministry of Science and Innovation (project code PI11-02908).