Fax: (416) 586-8397
Phase 2 study of preoperative image-guided intensity-modulated radiation therapy to reduce wound and combined modality morbidities in lower extremity soft tissue sarcoma†
Article first published online: 19 FEB 2013
Copyright © 2013 American Cancer Society
Volume 119, Issue 10, pages 1878–1884, 15 May 2013
How to Cite
O'Sullivan, B., Griffin, A. M., Dickie, C. I., Sharpe, M. B., Chung, P. W. M., Catton, C. N., Ferguson, P. C., Wunder, J. S., Deheshi, B. M., White, L. M., Kandel, R. A., Jaffray, D. A. and Bell, R. S. (2013), Phase 2 study of preoperative image-guided intensity-modulated radiation therapy to reduce wound and combined modality morbidities in lower extremity soft tissue sarcoma. Cancer, 119: 1878–1884. doi: 10.1002/cncr.27951
Presented at the 52nd Annual American Society for Radiation Oncology (ASTRO) Meeting, November 2010, San Diego, California.
- Issue published online: 6 MAY 2013
- Article first published online: 19 FEB 2013
- Manuscript Accepted: 4 DEC 2012
- Manuscript Revised: 30 NOV 2012
- Manuscript Received: 23 AUG 2012
- image-guided radiotherapy;
- intensity-modulated radiotherapy;
- preoperative radiotherapy;
- phase 2 study
This study sought to determine if preoperative image-guided intensity-modulated radiotherapy (IG-IMRT) can reduce morbidity, including wound complications, by minimizing dose to uninvolved tissues in adults with lower extremity soft tissue sarcoma.
The primary endpoint was the development of an acute wound complication (WC). IG-IMRT was used to conform volumes to avoid normal tissues (skin flaps for wound closure, bone, or other uninvolved soft tissues). From July 2005 to June 2009, 70 adults were enrolled; 59 were evaluable for the primary endpoint. Median tumor size was 9.5 cm; 55 tumors (93%) were high-grade and 58 (98%) were deep to fascia.
Eighteen (30.5%) patients developed WCs. This was not statistically significantly different from the result of the National Cancer Institute of Canada SR2 trial (P = .2); however, primary closure technique was possible more often (55 of 59 patients [93.2%] versus 50 of 70 patients [71.4%]; P = .002), and secondary operations for WCs were somewhat reduced (6 of 18 patients [33%] versus 13 of 30 patients [43%]; P = .55). Moderate edema, skin, subcutaneous, and joint toxicity was present in 6 (11.1%), 1 (1.9%), 5 (9.3%), and 3 (5.6%) patients, respectively, but there were no bone fractures. Four local recurrences (6.8%, none near the flaps) occurred with median follow-up of 49 months.
The 30.5% incidence of WCs was numerically lower than the 43% risk derived from the National Cancer Institute of Canada SR2 trial, but did not reach statistical significance. Preoperative IG-IMRT significantly diminished the need for tissue transfer. RT chronic morbidities and the need for subsequent secondary operations for WCs were lowered, although not significantly, whereas good limb function was maintained. Cancer 2013. © 2013 American Cancer Society.
Contemporary preoperative radiotherapy (RT) protocols for extremity soft tissue sarcoma (STS)1 irradiate less normal tissue and require lower total doses in comparison to postoperative RT; this leads to decreased late tissue morbidities such as limb edema, joint stiffness, fibrosis,2 and fractures,3 but an increased risk of acute wound healing complications.4 In the National Cancer Institute of Canada (NCIC) randomized SR2 trial, wound complications were twice the rate for preoperative compared with postoperative RT (35% versus 17%, respectively) and, in the lower extremity, were 43% and 21%, respectively.4 These complications required wound management, including repeat surgical or invasive procedures and/or deep wound packing with disability often lasting 6 months or longer.5
Image-guided intensity-modulated radiotherapy (IG-IMRT) permits precise targeting of complicated volumes, while allowing dose reduction to normal tissue.6 Modeling studies have shown that it is theoretically possible to spare both bone and uninvolved soft tissues with IMRT for patients with lower extremity STS (LE-STS).7, 8 Subsequent retrospective studies have supported this in cases at relatively high risk for local recurrence.9, 10 We report a phase 2 study that prospectively evaluated if preoperative IG-IMRT can minimize dose to uninvolved tissues, including potentially reducing the incidence of acute wound complications (WCs) in adult patients with LE-STS.
MATERIALS AND METHODS
The study was conducted with ethics approval and informed consent. Eligibility included histologically proven LE-STS appropriate for preoperative RT and surgery. Exclusion criteria included metastatic disease, patients who required chemotherapy, concurrent or prior malignancy, or age < 16 years. The primary endpoint was an acute wound healing complication within 120 days of resection, as previously defined.4 Secondary objectives included acute and late toxicities (limb edema and fibrosis, bone fracture, limb function, overall patient function). Overall survival, cause-specific survival, and local control were assessed. Sample size was estimated from the lower extremity WC rate differences between both arms of the NCIC SR2 trial (43% versus 21%).4 A total of 53 patients evaluable for the primary endpoint were required, and 10% was added to account for unplanned amputations, toxicities, or death. This provided an adjusted sample size of 59 patients.
RT Planning and Volume Definition
Total dose was 50 Gy in 25 daily fractions over 5 weeks prescribed to the International Commission on Radiation Units and Measurement (ICRU) reference point (ie, treatment isocenter). The treatment objectives specified that 95% of the prescribed dose should encompass the planning target volume (PTV), and the maximum dose should not exceed 107%, according to ICRU Report 50.11 Beam energies were 6 MV, and bolus was not used. Patients were positioned for computed tomography (CT) simulation in a neutral position that facilitated IMRT using daily 3-dimensional (3D) image guidance. The affected limb was immobilized in a custom device.12
RT avoidance structures included the skin/subcutaneous tissues required to close the future resection site (virtual surgical skin flaps), bone, and normal musculature. The “future scar” was outlined by the surgeon and highlighted with radio-opaque markers to guide delineation of the “future surgical skin flaps” in the RT planning system (Fig. 1). The surgeon later contoured the proposed surgical flaps on the RT planning CT data set, considering tissues to be elevated for either exposure or wound closure, while taking account of tissues requiring excision. The radiation oncologist defined the gross tumor volume (GTV) using CT and magnetic resonance imaging (MRI), and the clinical target volume (CTV) for microscopic elective treatment. The CTV extended 4 cm proximally and distally from the GTV and 1.5 cm radially, restricted at uninvolved anatomic barriers to tumor spread (eg, bone), and included peritumoral edema13; target coverage was prioritized over normal tissue (Table 1). Daily cone-beam CT image-guided radiotherapy permitted a 0.5-cm CTV 3D expansion for PTVs.14 IMRT dose objectives for uninvolved soft tissues (including virtual surgical flaps) and bone are shown (Table 1).
|IMRT Objective||Population Mean|
|Mean dose to bone||<37 Gy||26.2 ± 8 Gy|
|Maximum dose to bone||<59 Gy||50.6 ± 5 Gy|
|% Bone receiving ≥ 40 Gy||<64%||32.6 ± 19%|
|Normal musculature/ Tissue dose||<20 Gy Maximum, 21 Gy||8.4 ± 5 Gy|
|GTV dose||Minimum, 50 Gy||51.1 ± 0.9 Gy|
|CTV dose||Uniform 50 Gy||50.8 ± 0.6 Gy|
|Target coverage||0.977||0.991 ± 0.04|
Prior to surgery, the dosimetric information was transferred from the RT planning system for reproduction in the operating suite via an optical localization system.15 At resection, the actual flaps were developed for tumor exposure and wound closure using this information. The intent of surgical resection was complete tumor removal with a margin of normal tissue surrounding the tumor while preserving as much functional tissue as possible. Primary closure was the goal but was at the discretion of the treating surgeon.
Daily IMRT Delivery
IMRT was delivered using daily image-guided radiotherapy with online 3D cone beam CT image guidance and an automated bone matching algorithm. Set-up errors exceeding 1 mm in any of the cardinal directions were corrected by automated couch translation; maximum allowable rotational deviation was 3 degrees at the isocenter. Tissue changes (including tumor growth/shrinkage) exceeding a 1-cm predetermined threshold in any dimension required intervention by the radiation oncologist, including potential resimulation and replanning.
Follow-up for acute WCs occurred weekly for the first postoperative month, biweekly for postoperative month 2, and at postoperative month 3. The final assessment occurred 120 days after resection by a surgeon and a radiation oncologist not directly involved in the patient's care. WCs were defined using established criteria (Table 2) and followed the same assessment protocol used in the NCIC SR2 trial.
|• Secondary operations required for wound treatment (debridement, secondary closure procedures such as rotation flaps, free flaps, or skin grafts);|
|• Readmission to hospital for wound care;|
|• Invasive procedures required for wound care (drainage of hematoma, seroma or infected wound collection, use of vacuum-assisted closure therapy);|
|• Deep wound packing required at any time (deep packing defined as packing deep to dermis in an area of dehisced wound) to an area of the wound measuring at least 2 cm in length;|
|• Prolonged dressing changes, including packing of the wound for greater than 6 weeks from wound breakdown;|
|• Repeat surgery for revision of a split thickness skin graft or requirement for wet dressings for longer than 4 weeks. (It is permissible for a patient to protect a totally epithelialized skin graft with a dry dressing without declaring a major wound complication)|
Acute, Late Radiation Toxicity
Acute skin toxicity using the Radiation Therapy Oncology Group (RTOG) Acute Radiation Morbidity Scoring Criteria,16 late morbidity to skin, subcutaneous tissue, bone and joints according to the RTOG/European Organization for Research and Treatment of Cancer (EORTC) Late Radiation Morbidity Scoring Scheme,16 and the Common Toxicity Criteria, version 3.017 were documented. Limb edema was classified according to Stern.18 Morbidity was assessed at scheduled surveillance visits by a clinician not directly involved in the patient's treatment.
Limb Function and Overall Patient Function
Limb function and overall patient function determined by the Musculoskeletal Tumor Society Rating Scale (MSTS)19, 20 and the Toronto Extremity Salvage Score (TESS)21, 22 were assessed preoperatively at 3 and 6 months and at postoperative years 1, 2, 3, and 5. These and other secondary endpoint assessments also followed the SR2 processes.
Overall survival, local recurrence-free survival, and metastasis-free survival were measured from the date of resection. Clinical local recurrence was confirmed with CT/MRI and biopsy and defined as in-field, out-of-field, or marginal relative to the RT target volumes. Surveillance for local and distant recurrence involved 3-monthly clinical assessments and chest radiographs for 2 years, 6-monthly between years 2 and 5, and yearly thereafter (distant recurrence was confirmed by CT). Date of recurrence was the first date of observation.
Baseline characteristics are presented as proportions for discrete variables and means for continuous variables. The primary endpoint, major WC, was presented as a proportion and compared to the assumed population rate of 43%, using one-sample normal approximated z-test or binomial exact test. The 95% confidence intervals were calculated to infer the population rate. The impact of the baseline characteristics on complications was explored and tested using Fisher's exact test. Multivariate regression modelling was used to examine risk factors for developing a wound complication.
Acute and late radiation toxicity rates were tabulated. Limb function and overall patient function were analyzed as continuous variables. Mean or median scores and their 95% confidence intervals are presented. Overall survival, local recurrence-free survival, and metastasis-free survival were estimated by Kaplan-Meier method.23
A total of 70 cases were enrolled between July 2005 and June 2009 and 59 were evaluable for the primary endpoint. Six were excluded: 2 withdrew consent prior to treatment, 1 was benign on final pathological review, 1 underwent immediate surgery for unmanageable pain, 1 manifested metastases during the RT planning phase, and 1 had a prior cancer. Five eligible patients were not evaluable for the primary endpoint: 2 required amputations, 1 died of rapid systemic relapse after surgery, and 2 did not undergo surgery due to either development of metastases during preoperative RT or death from medical comorbidity.
There were 29 females and 30 males with mean age of 56 years (range, 26-86; Table 3). Mean tumor size was 10.6 cm, median size was 9.5 cm (range, 1.5-25), and 44% (26 of 59) were large (> 10 cm). Almost all tumors were deep (58 of 59), and 29 were grade 3 (49%), 26 were grade 2 (44%), 4 were grade 1 (7%). Most were classified as malignant fibrous histiocytoma/undifferentiated pleomorphic sarcoma (n = 21, 35.6%) and myxoid liposarcoma (n = 19, 32.2%) with a variety of other pathological subtypes. Four had microscopically positive resection margins. No patients received a postoperative RT boost.
|Sex||29 female/30 male|
|Mean age||56 years (26-86 years)|
|Median tumor size||9.5 cm (1.5-25 cm)|
|Tumor depth||1 superficial 58 deep|
|Undifferentiated pleomorphic sarcoma||21 (35.6%)|
|Myxoid liposarcoma||19 (32.2%)|
|Pleomorphic liposarcoma||6 (10.2%)|
|Synovial sarcoma||1 (1.7%)|
|Not otherwise specified||2 (3.4%)|
Mean pre-RT shift adjustments at set-up were 4 mm (range, 3-16 mm). Due to tumor growth and soft tissue changes exceeding 1 cm in any dimension after treatment planning, 8 patients were replanned of whom 5 developed WCs. Overall, WCs occurred in 18 of 59 patients (30.5%): 6 required a secondary operation (4 debridement, 1 free flap, and 1 rotation flap), 5 had an invasive procedure for seroma/hematoma drainage, 3 had infections debrided in clinic (with or without antibiotics), and 4 required dressing changes and/or deep packing beyond 4 months. This WC rate did not differ from the LE-STS preoperative RT arm of the NCIC SR2 trial (43%, P = .2, Fisher's exact test). All 18 experienced primary wound closure, including one with an intraoperatively repaired vascular injury but who later required hematoma drainage. The buttock was the most common site for WCs (5 of 11, 45%), followed by adductor (4 of 9, 44%), and hamstring (4 of 9, 44%) lesions.
Seven patients required vacuum-assisted closure for management of wound dehiscence, with 4 requiring additional invasive procedures (3 secondary operations, 1 seroma drainage). Vacuum-assisted closure was unavailable during the NCIC SR2 trial, and therefore a comparative assessment of the need for secondary operation for WC is compromised.
Primary wound closure was performed by the treating surgeon and was employed more frequently than in SR2 (55 of 59 [93.2%] versus 50 of 70 [71.4%]; P = .002). Of the 4 nonprimary closures, 3 were performed by the treating surgeon (2 gastrocnemius flaps and split-thickness skin graft, 1 split-thickness skin graft) whereas the fourth required the assistance of the plastic surgery service (a pedicled rectus abdominus rotation flap). No free flaps were required. The number of secondary operations for WC was numerically less than in the SR2 trial, although this was not a statistically significant difference (6 of 18 [33%] versus 13 of 30 [43%], P = .55).
There was a trend toward higher mean RT doses to the surgical skin flaps for those experiencing WC (Table 4). The mean percentage of flaps receiving ≥ 30 Gy and the flap/PTV overlap were also greater in the WC group (P = .043 and P = .003, respectively). On multivariate analysis, only flap/PTV overlap remained statistically significant (0.003). In 21 cases there was minimal overlap (< 1%) of the flaps and PTV. Three of these patients subsequently developed a wound complication (3 of 21, 14.3%), as compared with a rate of 39.5% (15 of 38, P = .04) in the remainder where there was greater overlap.
|Mean dose to flaps||32.4 Gy||34.3 Gy||.097|
|Mean volume of flaps||218.5 cm3||328.3 cm3||.005|
|Mean flap width||1.69 cm||2.01 cm||.040|
|Mean flap thickness||1.74 cm||2.0 cm||.235|
|Mean flap length||25.3 cm||27.0 cm||.202|
|Flap and PTV overlap||21.53 cm3||67.13 cm3||.003|
|% Flap treated ≥ 30 Gy||62.7%||69.8%||.043|
|Mean GTV volume||608.2 cm3||849.6 cm3||.302|
|Mean CTV volume||1178.0 cm3||1832.0 cm3||.020|
At median follow-up of 49.0 months, 4 developed local recurrence (6.8%), none near the surgical skin flaps. Three of these were in-field and 1 was out-of-field. Two were in patients with positive resection margins (1 out-of-field, 1 in-field). The 5-year local recurrence-free survival was 88.2%. Seventeen patients (28.8%) developed metastases: 4 are currently alive with disease, 12 died of disease, and 1 died of other causes. Estimated metastasis-free survival was 66.8% and overall survival was 74.6% at 5 years.
Fifty-one patients had functional evaluations performed at least 1 year after surgery. Mean TESS score was 83.1 (range, 33.3-100), mean MSTS87 was 31.5 (range, 21-35), and mean MSTS93 was 89.3 (range, 47-100). For each measure, the scores indicate a relatively high functional level for the average patient. No severe late radiation toxicities according to the RTOG/EORTC Late Radiation Morbidity Scoring Scheme16 (> grade 2) were seen among the 54 patients surviving longer than 2 years. One patient (1.9%) had moderate skin toxicity, 5 patients (9.3%) had moderate subcutaneous fibrosis, and 3 patients (5.6%) had moderate joint stiffness. Moderate edema occurred in 6 patients (11.1%) and mild edema in 17 patients (31.4%). No patient suffered a posttreatment fracture.
With the advent of IG-IMRT, the opportunity emerged to redefine conventional target volumes with improved conformality, while significantly reducing dose to normal structures vulnerable to RT including bone and unaffected musculature in the radial dimension.8 A prospective trial with informed consent was designed for LE-STS with a defined endpoint to permit a realistic sample size to be calculated that could address the value of IMRT in improving outcome for this population. Moreover, tissues used for wound closure, bone, and uninvolved soft tissues were assigned dose objectives (Table 1). WCs within 120 days of surgery represents a meaningful endpoint to address morbidity in addition to other outcomes (fibrosis, edema, joint stiffness) to contrast with the outcomes of the SR2 trial, consistent with hypotheses generated from previous modeling studies and retrospective assessments of outcome following IMRT.
During the SR2 trial, technological limitations greatly prevented RT beam shaping or normal tissue sparing. Preoperative 3D IG-IMRT may overcome these problems, thus enabling dose-reduction to both skin/subcutaneous tissues required for wound closure and to bone and other normal tissues. The goal of the present trial was to reduce the LE-STS WC risk to the level in the postoperative RT arm of the SR2 trial. The incidence of WCs in the current study was 30.5%, representing a 12.5% reduction from the 43% seen with preoperative RT in the SR2 trial. Although not reaching our baseline goal of 21%, it may still be viewed as clinically, if not statistically, significant.
A principal reason for the absence of a larger reduction in WCs may be the fact that target coverage was prioritized in IMRT plan optimization at the expense of normal tissue avoidance. Therefore, it was not always feasible to spare the “virtual surgical skin flaps” as originally intended. Examination of the overlap between the “virtual surgical skin flaps” and the PTV showed that the volume of surgical flaps receiving the prescribed dose was higher for the WC than for the non-WC group (67.13 cm3 versus 21.53 cm3, respectively), which was significant on multivariate analysis (P = .003). In addition, only 14.3% of cases with minimal flap/PTV overlap developed WCs versus 39.5% in cases with greater overlap. Thus, PTVs that are, by necessity, more superficial lead to lower dose reduction to the flaps and a greater risk of WCs. Baldini et al24 also described increased WCs for tumors very close to the skin surface. Furthermore, 8 patients required RT replanning due to tumor growth and other soft tissue changes. All 8 cases that required replanning were due to tumor growth > 1 cm in any dimension, which would have compromised tumor coverage. All volumes were recontoured at the time of replanning using the original IMRT objectives with tumor coverage as a priority. This led to less sparing of the flaps in all cases. Previously spared tissues were transposed into the new PTV, thereby compromising the beneficial effect of dose minimization achieved in the initial IMRT plan; 5 of the 8 developed a WC. Because tumor/target coverage must always take priority over flap-sparing, the use of a tissue transfer should be considered in these instances, although rotational or free flaps do not guarantee avoidance of WCs4, 25 and have their own potential morbidities.
Secondary endpoints were also improved compared with that of the SR2 trial, although direct statistical comparisons were not possible due to the aggregated reporting of these upper and lower extremity endpoints in the preoperative RT group of the SR2 trial. Only 5 of 54 (9.3%) evaluable patients had moderate (grade 2) fibrosis at 2 years, with none rated severe. This compares favorably to 23 of 73 (31.5%) rated as grade 2 or higher in the preoperative arm of the SR2 trial.2 Moderate edema was present in 6 of 54 patients (11.1%), compared with 11 of 73 (15.1%), whereas moderate joint stiffness was found in only 3 of 54 (5.4%), versus 13 of 73 (17.8%) in the preoperative arm of the SR2 trial. In addition, investigators from the Memorial Sloan-Kettering Cancer Center reported the outcome of patients with extremity STS who were treated with IMRT, generally delivered postoperatively.10 Moderate edema was found to be similar to that of the current study (5 of 41, 12.2%), whereas moderate joint stiffness was higher (7 of 41, 17.1%) and 2 patients (4.8%) suffered a bone fracture. Subcutaneous fibrosis data were not provided. Wound complications were reported in 8 patients (19.5%), although it is uncertain whether these patients received preoperative or postoperative IMRT.
With the narrowing of RT margins around the tumor, and increased normal tissue sparing, no commensurate increase in local recurrence was observed. There were no recurrences manifested in the area of the protected surgical skin flaps. Estimated 5-year local recurrence-free survival remained high (88.2%), similar to our historical rate of 89.6% in extremity STS treated with surgery and external beam RT.26 Because all patients have been followed for at least 2 years, this rate is unlikely to change substantially. The Memorial Sloan-Kettering Cancer Center group also found no increase in local relapse risk with IMRT, with a 5-year estimated local control rate of 94%,10 decreasing to 92% when updated with additional patients in 2011.27
Assessment of the primary endpoint of acute wound healing complications used criteria that have been developed empirically. As such, there was little subjectivity in the assessment of acute wound healing complications. Moreover, this endpoint was evaluated independently in each patient by a surgeon and a radiation oncologist not directly involved in the patient's care. The assessment of late tissue morbidities was more subjective, but was assessed in the same manner as for acute wound healing complications.
Currently awaited are results of Radiation Therapy Oncology Group trial 0630, another image-guided preoperative RT study in extremity STS. Unfortunately, for several reasons, direct comparison between both studies may be problematic. In particular, the primary endpoints differ (late toxicity scores in the case of RTOG 0630 compared with WCs in the current study). Other differences include eligibility criteria (upper extremity included in RTOG 0630, which may alter the toxicity profile) and treatment (3D conformal RT, chemotherapy, and postoperative RT boosts may be components of RTOG 0630, in addition to target volume definitions and the type of image guidance).
In conclusion, the results of the current prospective trial indicate that IG-IMRT may ameliorate the risk and severity of WCs by reducing dose to the tissues to be employed for wound closure following STS resection, the need for tissue transfer at tumor resection, and the necessity for subsequent secondary operations in patients who experience WCs. Although these results did not reach statistical significance, it may relate to cases where the flaps could not be spared and received higher doses (due to tumor location or growth) resulting from mandated tumor coverage. A trial with larger sample size may yield more significant results and may further validate the results of the study. Moreover, the dose to other uninvolved tissues such as bone and unaffected musculature was reduced. This has resulted in excellent limb function measures with no bone fractures. The current approach at our institution remains preoperative IG-IMRT with attempt to minimize doses to uninvolved normal tissues, including subsequent surgical flaps where possible, but not at the expense of encroachment on the CTV.
This study received funds from the Ontario Institute for Cancer Research (OICR).
CONFLICT OF INTEREST DISCLOSURE
The authors made no disclosure.
- 11International Commission on Radiation Units and Measurement. ICRU Report 50: Prescribing, Recording, and Reporting Photon Beam Therapy. Bethesda, MD: ICRU; 1993.
- 14Measuring interfractional and intrafractional motion with cone beam computed tomography and an optical localization system for lower extremity soft tissue sarcoma patients treated with preoperative intensity-modulated radiation therapy. Int J Radiat Oncol Biol Phys. 2010; 78: 1437-1444., , , et al.
- 15Geometric performance and efficiency of an optical tracking system for daily pre-treatment positioning in pelvic radiotherapy patients. Technol Cancer Res Treat. 2011; 10: 163-170., , , , , .
- 18Clinical Examination: A Textbook for Physical Diagnosis. Chicago, IL: Year Book Medical Publishers; 1964..
- 19Modification of the system for functional evaluation of the surgical management of musculoskeletal tumors. In: , editor. Limb Salvage in Musculoskeletal Oncology. New York, NY: Churchill Livingstone; 1987., ed.
- 20A system for the functional evaluation of reconstructive procedures after surgical treatment of tumors of the musculoskeletal system. Clin Orthop Relat Res. 1993; 286: 241-246., , , , .
- 22Evaluating functional outcome in patients with lower extremity sarcoma. Clin Orthop Relat Res. 1999;( 358): 90-100., , , , .
- 23Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958; 53: 457-481., .
- 24http://www.ctos.org/meeting/2011/program.pdf. Accessed November 2, 2012., , , et al. Predictors for major wound complications following pre-operative radiotherapy and surgery for soft tissue sarcoma of the extremity and trunk: importance of proximity to skin surface. In: Proceedings of the 2011 Combined Meeting of the Connective Tissue Oncology Society and the Musculoskeletal Tumor Society; October 26-29, 2011; Chicago, IL; paper #22.