Radiation Therapy Techniques in the Management of Locally Advanced, High-Grade, Soft-Tissue Sarcoma

Limb-salvage therapy using external-beam radiation therapy (EBRT) to reduce local recurrence (LR) allows surgeons to perform wide-excision of soft-tissue sarcoma (STS), allowing less radical surgery such as amputation. ^1,2 In 2011, the Radiation Therapy Oncology Group (RTOG) published guidelines for designing clinical target volume (CTV) margins in the treatment of extremity soft-tissue sarcoma (ESTS), which was verified by subsequent publications, where longitudinal proximal/distal CTV recommendations were 3 to 5 cm, but radial CTV was 1.5 cm. ^3-6 One question was whether these guidelines can be applied to STS involving subcutaneous tissues, and the trunk, where space is more limited. Large treatment volumes can increase a patient’s risk of long-term complications. ⁷ However, this must be balanced against risks of LR when treatment volumes are too small, and thus our goal was to assess prognostic factors associated with higher LR rates and wound complications of locally advanced, high-grade, STS, and see which CTV margins may be optimal.

Retrospective analysis was performed on data from 100 adult patients from January 2005 to December 2018, who had primary high-grade STS of the extremity or trunk, > 5 cm, localized to the muscle or subcutaneous tissue, that were either synovial (16 patients), dedifferentiated liposarcoma (16), myxofibrosarcoma (13), round cell liposarcoma (7), undifferentiated pleomorphic sarcoma (39), or undifferentiated sarcoma not otherwise specified (9). We defined locally advanced as > 5 cm in size. Patients were T2 if the tumor was > 5 cm in maximal dimension, T2a for subcutaneous tumor, and T2b for muscle-invasive tumor. ⁸ Patients had to have their radiation fields and dose plans available for review, with a minimum of 1 year of follow-up. Bone sarcomas, gastrointestinal stromal tumors, pediatric sarcomas, and retroperitoneal sarcomas were excluded from this analysis. Central pathology review was performed by our soft-tissue pathologist (S.S.) and graded according to the Federation Nationale des Centres de Lutte Contra le Cancer (FNCLCC); ⁹ our study included only grade 3 or high-grade sarcomas. Radiation therapy was delivered using megavoltage photons, using either 6 MV or 15 MV, with the majority (78%) undergoing preoperative radiation to 4400-5000 cGy in 200 cGy fractions to the gross tumor with margin, followed by wide excision, which occurred about 6 weeks after radiation. Twenty-two percent underwent postoperative radiation to 4500-5040 cGy with a shrinking field boost to a total of 5940-6660 cGy in 180-200 cGy fractions. Additional boost was not performed on preoperative radiation patients who had a positive margin after wide excision. ¹⁰ All patients underwent immobilization using a polystyrene-filled vacuum cradle. Gross tumor volume (GTV) contours included T1-weighted images with gadolinium enhancement of the tumor on MRI, and T2-weighted edema included in the CTV. CTV volumes were determined by the prescribing physician. CTV to planning target volume (PTV) expansion was 1 cm, as image guidance was not performed. A bolus was used in select cases at the discretion of the prescribing physician due to concern about scar/cutaneous recurrence. During the timeframe these patients were treated, there was no effort to treat the skin as an organ at risk (OAR). Three-dimensional conformal radiation therapy (3DCRT) was performed in 91 patients, and intensity-modulated radiation therapy (IMRT) was performed in 9 patients. The radiation planning images (digitally reconstructed radiographs) and isodose plans were scaled and compared to MRI evidence of LR, which was categorized as out of field if > 80% of the volume was outside of the irradiated volume, in field if > 80% of the volume was inside the irradiated volume, and marginal if recurrence was at the edge of the irradiated volume between the out-of-field and in-field parameters, listed above. Binary classification of CTV margins was performed. The first was a CTV of 1.5 cm. The second was a CTV margin ≥ 2 cm or if the CTV extended to a fascial boundary, as a fascial boundary is equivalent to a wide margin. For postoperative patients, CTV margins were recorded based on the original fields taken to 4500-5040 cGy, and not based on the boost margins.

The Charlson Comorbidity Index score is a method of categorizing comorbidities of patients based on the International Classification of Diseases, and was assessed for each patient, and tabulated in Table 1 . ¹¹ After therapy, patients without wound complications were generally followed every 6 months for 5 years, and yearly thereafter, with MRI of the local area and computed tomography (CT) of the chest.

Date of surgery was established as time zero, and Kaplan-Meier estimates were used to calculate local recurrence-free survival (LRFS). ¹² Characteristics of patients were reported as percentages for categorical factors, and median with ranges was used for continuous factors. Fisher’s exact tests were used to test for categorical differences in treatment groups. Wilcoxon rank-sum tests were used to calculate differences on continuous factors. Multivariable analysis (MVA) of prognostic factors using the Cox proportional hazards models was used to estimate hazard ratios on LR. ¹³ Multivariate logistic regression was performed and odds ratios were calculated to detect the factors associated with the probability of wound complications or the presence of a wound vacuum device at 3 months after surgery. The forest plot was produced to display the results graphically. The Wald test was used to calculate P value for the odds ratio. ¹⁴ Statistical significance utilized a 2-sided P < 0.05. All analyses were conducted using SAS EG 7.13 (SAS Institute Inc.).

Median follow-up was 5.8 years (range 1.0 to 15.0 years). Ifosfamide-doxorubicin was given in 31 patients, and 69 received radiation therapy alone as adjuvant therapy. Ifosfamide-doxorubicin was usually prescribed neoadjuvantly for 5 cycles followed by radiation, then followed by wide excision; whereas for postoperative radiation, ifosfamide-doxorubicin was administered after the completion of radiation. All patients underwent en bloc wide excision, but 6 patients had unintended excision, where the surgeon had violated the tissue planes. These 6 patients subsequently underwent en bloc wide re-excision.

Cohort characteristics using univariate comparisons revealed that trunk location (P = 0.001), longitudinal-distal CTV 1.5 cm (P = 0.019), and radial CTV 1.5 cm (P < 0.001) were associated with higher rates of LR, but longitudinal-proximal CTV 1.5 cm did not quite make statistical significance (P = 0.058) ( Table 1 ). MVA of these factors revealed that trunk location (P = 0.048), and radial CTV 1.5 cm (P = 0.006) were independently associated with higher rates of LR ( Table 2 ). No significant difference in LR was found with preoperative radiation, margin status, tumor size, unintended initial surgery, use of bolus, or chemotherapy ( Table 1 ). Of the patients with LR, 10 out of 15 had a trunk location, and 10 of 15 were muscle-invasive (T2b). In relation to the field of radiation, 10 out of 15 were marginal at the radial edge of the field, 4 out of 15 were in-field recurrences (1 of which had a positive margin), and 1 in 15 recurred marginally at the longitudinal-proximal edge of the radiation field in which the CTV margin was 2 cm ( Table 3 ). None of the LRs were completely outside the field of radiation. Most local recurrences (14 out of 15) correlated with the depth of their initial stage, and only 1 patient with T2a disease experienced a cutaneous recurrence, which had also recurred at the marginal edge of the radial margin, despite having a bolus applied. Most of the patients were treated with 3DCRT (91%), and only 9 (9%) patients were treated with IMRT.

There were 15 patients with LR, with an estimated 5-year LRFS of 83.0%. Sixty-nine percent of patients had a wide radial margin since the field edge was beyond the fascial boundary with 7.2% LR. Of these with a wide fascial margin, those with a positive margin had 20% LR, whereas those with a negative margin had 5.1% LR. Patients with a 1.5 cm radial CTV that was not beyond the fascial boundary experienced 45.4% LR, but those with radial CTV ≥ 2.5 cm did not experience LR, despite not having a fascial boundary ( Table 4 ).

Requirement of a wound vacuum device and/or open wounds occurred in 27 (27%) patients, ranging from 3 days to 29.7 months, with the median duration of wound vacuum devices/open wounds being 4.7 months. The odds ratio for open wounds at 3 months or the presence of a wound vacuum device showed a higher risk for patients with larger tumor size (P = 0.02), larger field width (P = 0.02), and use of a bolus (P = 0.02). Patients with muscle-invasive disease had a lower odds ratio of wound complications compared with subcutaneous disease (P = 0.008). Also, patients with a smaller proximal CTV of 1.5 cm (P = 0.046) and those who underwent chemotherapy (P = 0.044) experienced a lower risk of wound complications ( Figure 1 ).

Longitudinal proximal/distal CTV margins of 3 cm seem adequate, but high-grade STS ≥ 5 cm may benefit from increased radial CTV margins of 2-2.5 cm in the absence of a fascial boundary, although larger CTV may increase nonhealing wound rates. Bolus techniques may increase wound complications in T2a-b STS, and should not be routinely employed.