Postoperative radiotherapy in patients with R0 resection of soft tissue sarcoma: results from the European sarcoma CONTICABASE analysis

Abstract

Objectives

We aim to report the outcomes of soft tissue sarcoma (STS) patients with surgical R0 margin size (large: ≥5 mm or close: ≥1 to <5 mm) tumour, treated or not with postoperative radiotherapy (RT).

Methods

The Connective Tissue Cancer Network database CONTICABASE collected data from adult patients diagnosed from 1999 to 2016 for histologically-proven locally advanced STS of trunk or limbs, with surgical R0 margin size (large: ≥5 mm; close: ≥1 to <5 mm), who did not receive neoadjuvant/adjuvant chemotherapy. A propensity score matching analysis was performed to account for potential treatment selection biases. We reported local control (LC), disease-free survival (DFS), overall survival (OS), RT impact in subgroup analyses, and performed multivariable analyses adjusted for classical prognostic factors.

Results

Among 738 patients with STS eligible from 11 sites, 524 patients had R0 surgical margins, either treated with postoperative RT (N = 374, 71.4%), or not (N = 150). Besides similar characteristics regarding sex, age, margin size, and preoperative MRI assessment, more tumour localizations in lower limbs (62.8% vs 46%), larger tumours (≥5 cm) (68.4% vs 46%), deeper tumours (75.9% vs 56%), and more grade ≥3 tumours (50.4% vs 32.7%) were reported in patients having received adjuvant RT. The median follow-up was 74.7 [95% CI 71.8-83.0] months, 10-year-LC, -DFS, and -OS were 77.5% [95% CI 71.2-82.6], 51.0% [45.1-56.7], and 69.6% [63.7-74.6], respectively. The multivariable analysis showed that patients with older age and larger tumours are more at risk in terms of LC, DFS, and OS. Preoperative MRI assessment (HR 0.43, 95% CI 0.22-0.83; P = .012) and RT (HR 0.27, 95% CI 0.15-0.49; P < .001) were significantly associated with better LC. However, RT did not impact OS. Subgroup analyses showed that RT also benefit to grade 1 and 2 tumours.

Conclusion

This series shows that adjuvant RT improved LC in all patients with STS with surgical R0 resection.

Advances in knowledge

This series did not identify subgroups in whom RT can be deleted.

Introduction

Soft tissue sarcomas (STSs) are a heterogeneous group of rare tumours accounting for 1% of adult malignancies, with different locations and over 100 histological subtypes, making the most appropriate therapeutic strategy difficult to define. Sarcoma management should be performed in expert reference centers involving dedicated multidisciplinary tumour boards.¹ Standard management of high-grade (grades 2-3) STSs of the extremities and trunk include surgery and RT. Early management with preoperative MRI in reference centers is highly encouraged in patients with sarcoma. MRI provides a valuable support for preoperative local staging modalities,¹ and contributes to improve the quality of surgery and radiotherapy (RT). Neoadjuvant chemotherapy with anthracycline and ifosfamide for STS of the limbs and trunk is a valid option.^2–4 Preoperative RT is used in routine management in most of the centres,⁵ and considered as an option more relevant than postoperative RT in tumours with borderline operability, counteracting the negative impact of anticipated R1 margins.² However, preoperative RT is associated with more acute postoperative complications and higher risks of complications in patients with proximal thigh/groin or axillary locations or in patients with diabetes or obesity have been reported. Some patients at higher risk of wound complications and those with tumours that cannot be safely or effectively managed with preoperative RT (fungating, bleeding, painful) need to be operated before RT.^6,7 To note, postoperative RT in low-grade liposarcomas is not anymore indicated in the recent years. Indeed, even if patients with liposarcoma having received postoperative RT reported improved local control (LC) with rare metastatic events, the national sarcoma group recommended surveillance exclusively, and resection is exclusively decided in the case of recurrence.⁸

One of the key issues in the management of STS is to achieve LC. The quality of surgical resection is significantly associated with recurrence rates.^9,10 Indeed, complete macroscopic surgical resection with negative margins (R0 resection) drastically decreased the risk of local recurrence. In addition, two prospective randomized trials showed that adjuvant RT after limb-sparing surgery significantly decreased the rates of local recurrence.^11,12 However, no improvements in survival have been demonstrated so far with postoperative RT. Moreover, potential long-term radiation-induced toxicities should be anticipated to reasonably define the most adequate risk-adapted therapy and adequately preserve patient quality of life. Notably, increased rates of fibrosis and lymphedema and bone fractures^13–15 advocated for surveillance without adjuvant RT in selected patients presenting low-grade and superficial sarcomas and treated with compartmental surgery. Furthermore, the benefit from RT remains also controversial in some selected patients with high-grade (grade ≥3) deep sarcomas after complete surgical resection (even non-compartmental surgery), that is, initial surgical resection with large margins.^16–21 Indeed, decisions for postoperative RT in sarcomas were based on rather old prospective randomized studies, and the benefit from RT need to be re-assessed considering the recent improvements in surgical management and in pathological characterization during the last decades. The study SARC-01 (NCT00870701) initiated in 2009 aimed to clarify the place of adjuvant RT in patients with STS involving limbs and trunk wall with R0 resection in the modern era. The suboptimal recruitment of about 70 patients obliged to premature discontinuation of the trial in 2018. Expert radiation oncologists from the group consequently performed a retrospective study using the Connective Tissue Cancer Network database CONTICABASE in patients with selection criteria similar to those used in the SARC-01 trial, who did not receive chemotherapy to prevent additional confounding biases. This study aimed to report LC, disease-free survival (DFS), and overall survival (OS) in patients with complete macroscopic resection (R0) treated or not with postoperative RT in routine practice.

Methods

This retrospective multicentric study included data from adult patients ≥18 years with histologically-proven STS diagnosed between January 1999 and September 2016, after R0 primary tumour resection performed in one of the 11 participating referral centers for STS. Data were collected in the European retrospective clinicopathologic database on sarcomas and Ethics approval was obtained from the appropriate committees.

Expert radiation oncologists reviewed surgical R0 margin reports from expert anatomo-pathologists and margin sizes were categorized as large (≥5 mm), or close (≥1 to <5 mm) margins.¹⁰ Patients with an interposition of an anatomical barrier (aponevrosis) were classified in the large margin group. Margins are defined according to the R classification.¹⁰ Data regarding early management with preoperative MRI were collected.

Exclusion criteria were patients with alveolar or embryonal rhabdomyosarcoma, retroperitoneal, head and neck or intra-thoracic or breast sarcomas, well-differentiated liposarcoma (WDLPS), dermatofibrosarcoma protuberans (DFSP), desmoid tumours, patients with node or visceral metastases, no neoadjuvant, concomitant or adjuvant chemotherapy, brachytherapy, or preoperative RT. Radiation oncologists reviewed patient health electronic records, and more specifically RT features and histopathological reports to ensure appropriate data quality control. Data were updated by a dedicated database research assistant in each center. The data cut-off was March 15, 2018.

Statistics

Continuous variables were presented as median with range (min-max) and categorical variables as frequencies and percentages. Comparisons between groups were performed using Mann-Whitney test for continuous variables and χ² or Fisher’s exact test for categorical variables.

Survival data including LC, DFS, and OS were estimated using the Kaplan-Meier method with 95% confidence intervals (CIs) from the date of surgery to local recurrence for LC, recurrence (local or distant) and death from any cause for DFS and death from any cause for OS. Patients with no event were censored at the last follow-up. Duration of follow-up was estimated using the reverse Kaplan-Meier method. An additional analysis of OS calculated from the date of local recurrence was performed in the 71 patients with a local recurrence.

A propensity score matching analysis was performed to account for potential treatment selection bias. The propensity score defined as the probability of receiving RT was estimated using a multivariable logistic regression model including age at diagnosis in years, tumour localization (trunk vs lower limb vs upper limb), grade (1 vs 2 vs 3), size in mm, and depth (superficial vs deep). Patients treated with RT were matched to patients without RT by 1:1 nearest neighbour matching using a 0.05 calliper.

Univariable and multivariable analyses were performed using Logrank test and Cox proportional hazards model, respectively. Hazard ratio (HR) was estimated with 95% CI. In addition, a competing risk analysis was performed with the first local recurrence as event of interest and distant metastases and death as competing events. Univariable and multivariable analyses were performed using the fine and gray model. The sub-distribution HR (sHR) was estimated with 95% CI. The univariable analysis for OS, DFS, LC, and competing risk analysis included the following variables: age at diagnosis, tumour characteristics (size, location, grade, depth, and margin size), preoperative MRI, and RT. Variables significant at 10% level in the univariable analysis and tumour characteristics were included in the multivariable analysis. A univariable subgroup analysis was performed to assess the impact of RT on LC in each subgroup according to age at diagnosis and tumour characteristics, and results were summarized in a forest plot. The P-value of interaction and HR (95% CI) were calculated using a Cox proportional hazards model.

All statistical tests were two sided and a P-value <.05 was considered statistically significant. Statistical analyses were carried out using STATA software version 16 (StataCorp LP, College Station, TX).

Results

Patients and treatments

Expert radiation oncologists reviewed reports from 738 patients operated in 11 anticancer centers or university hospitals. To note, 214 patients with no margin size available after review were excluded. The 524 patients with margins available had tumour samples recovered after initial surgery (n = 221, 42.2%), or after re-excision (n = 303, 57.8%) (Figure 1) A propensity score matching analysis including 260 patients (RT: N = 130; No RT: N = 130) was performed to account for potential treatment selection bias.

Figure 1.

Flow chart. Margin size (mm) recovered after review (N = 141) (≥1 to <5 mm: N = 4; ≥5 mm: N = 137). *To note, 221 (42.2%) patients had one surgery, 303 (57.8%) patients had re-excisions.

Patient characteristics are presented in Table 1. The patient population included 247 (47.1%) women and 277 (52.9%) men. The median age at diagnosis was 57 (18-92) years. A total of 150 patients were treated by definitive surgical resection, and 374 (71.4%) patients were treated with postoperative RT, with median dose of RT of 50 (20-70) Gy (<50 Gy: 3.9%; 50-60 Gy: 91%; >60 Gy: 5.1%) delivered in fractions of 1.8-3 Gy. Characteristics of patients in both the groups (RT, No RT) were well balanced regarding gender, age, median age at diagnosis, surgical margin size, and use of preoperative MRI (Table 1). Patients having received RT showed some imbalances in tumour characteristics (lower limb location, RT: 62.8% vs No RT: 46%; size ≥5, RT: 68.4% vs No RT: 46.6%; depth, RT: 75.9% vs No RT: 56%; grade 3: RT: 50.4% vs No RT: 32.7%).

Table 1.

Demographics and tumour characteristics in patients with STS, treated or not with radiotherapy (N = 524).

	Total	No RT	RT	P-value
	(N = 524)	(N = 150)	(N = 374)
Gender				.1083
Female	247 (47.1%)	79 (52.7%)	168 (44.9%)
Male	277 (52.9%)	71 (47.3%)	206 (55.1%)
Age at diagnosis				.1178
<60 years	297 (56.7%)	77 (51.3%)	220 (58.8%)
≥60 years	227 (43.3%)	73 (48.7%)	154 (41.2%)
Median age at diagnosis	57 (18-92)	58.5 (18-92)	57 (18-92)	.1112
Tumour location				.0011
Trunk	126 (24.0%)	50 (33.3%)	76 (20.3%)
Lower limb	304 (58.0%)	69 (46%)	235 (62.8%)
Upper limb	94 (17.9%)	31 (20.7%)	63 (16.8%)
Tumour size (cm)—median	6.0 (0.6-40.0)	4.0 (0.6-18.0)	6.0 (0.8-40.0)	<.0001
Q1-Q3	(3.5-9.5)	(2.5-7.0)	(4.0-10.0)
Tumour size (cm)				<.0001
<5	197 (37.8%)	79 (53.4%)	118 (31.6%)
≥5	324 (62.2%)	69 (46.6%)	255 (68.4%)
Not specified	3	2	1
Measurable margin size (mm)
Close (<5)	229 (43.7%)	63 (42%)	166 (44.4%)	.6188
Large (≥5)	295 (56.3%)	87 (58%)	208 (55.6%)
Depth of tumour				<.0001
Superficial	156 (29.8%)	66 (44.0%)	90 (24.1%)
Deep	368 (70.2%)	84 (56.0%)	284 (75.9%)
Grade of tumour				<.0001
1	69 (13.5%)	32 (21.8%)	37 (10.1%)
2	211 (41.2%)	67 (45.6%)	144 (39.5%)
3	232 (45.3%)	48 (32.7%)	184 (50.4%)
Not specified	12	3	9
Histology
Undifferenciated sarcoma	167 (31.9%)	48 (32.0%)	119 (31.9%)
Myxofibrosarcoma	73 (14.0%)	18 (12.0%)	55 (14.7%)
Leiomyosarcoma	76 (14.5%)	35 (23.3%)	41 (11.0%)
Liposarcoma	106 (20.3%)	24 (16.0%)	82 (22.0%)
Synovialosarcoma	49 (9.4%)	10 (6.7%)	39 (10.5%)
MPNST	13 (2.5%)	3 (2.0%)	10 (2.7%)
Other	39 (7.5%)	12 (8.0%)	27 (7.2%)
Missing	1	0	1
Preoperative MRIa				.1281
No	235 (53.2%)	72 (59.0%)	163 (50.9%)
Yes	207 (46.8%)	50 (41.0%)	157 (49.1%)
Not specified	82	28	54

RT: radiotherapy; MPNST: malignant peripheral nerve sheath tumour; Q: quartile.

^aTo note, recent management more frequently included pre-operative MRI (median year of surgery: No MRI: 2006, 1999-2014; MRI: 2009, 2000-2013).

(i) Before 2004: No MRI: 63, 26.8%; MRI: 26, 12.6%.

(ii) From 2004 to 2007: No MRI: 73, 31.1%; MRI: 42, 20.3%.

(iii) From 2008 to 2010: No MRI: 65, 27.7%; MRI: 94, 45.4%.

(iv) After 2010: No MRI: 34, 14.5%; MRI: 45, 21.7%.

Survival outcomes

The median follow-up was 74.7 [95% CI 71.8-83.0] months. A total of 116 (22.1%) patients died, and 176 (33.6%) patients relapsed, including 71 patients with local relapse. The first event was local relapse in 61 (11.6%) patients, and metastasis in 115 (21.9%) patients. In the global population, the 5 y-LC, DFS, and OS were 88.0% [84.5-90.8], 65.3% [60.8-69.4], and 80.1% [95% CI 76.1-83.6], respectively, and the 10 y-LC, DFS, and OS were 77.5% [71.2-82.6], 51.0% [45.1-56.7], and 69.6% [63.7-74.6], respectively. In the 71 patients with a local recurrence, the 5 y-OS from local recurrence was 63.2% [95% CI 48.5-74.8] (Supplementary Data S1).

Figure 2 shows LC, DFS, and OS according to RT in the overall population, and in the propensity score matching population. The propensity score matching analysis highlighted that RT significantly increased LC (HR 0.32, 95% CI 0.16-0.65; P = .001) and DFS (HR 0.50, 95% CI 0.32-0.76; P = .001), but no significant impact on OS was reported (HR 0.96, 95% CI 0.54-1.71; P = .88).

Figure 2.

Survival curves according to radiotherapy in the global population (N = 524) and after confounding adjustment using propensity score matching (N = 260). (A) Local control (LC), (B) disease-free survival (DFS), and (C) overall survival (OS) in overall population; (D) LC, (E) DFS, and (F) OS in matched population according to radiotherapy (RT).

Competing risk analysis

The competing risk analysis highlighted that RT was associated with a lower risk of local recurrence (sHR: 0.34, 0.20-0.58; P < .001). No significant effect of RT was reported on competing events (defined as metastatic recurrence or death before local recurrence). Large resection margins (≥5 mm) were not associated with a lower risk of metastatic recurrence or death, neither local recurrence. Moreover, older age (≥60 years), higher grade (grade 3), and deeper tumours were at higher risk of metastatic recurrence or death but no impact on local recurrence was shown, whereas large tumours (≥5 cm) were associated with a higher risk of local and metastatic recurrence or death (Supplementary Material S2).

Subgroup and multivariable analyses

Subgroup univariable analyses were performed with the aim to identify subgroups of patients who might benefit from RT. Figure 3 explores the impact of RT on LC, and reports that RT benefit to the global population (HR 0.45, 95% CI 0.28-0.72) and to each subgroup of patients. The RT benefit is similar regardless margin sizes (interaction P-value = .89) and tumour depth (interaction P-value = .99). Whereas no interaction was significant, greater RT benefit was reported in subgroups of older patients (0.35, 95% CI 0.18-0.69), grade 1-2 tumour (HR 0.31, 95% CI 0.16-0.58), small tumours (HR 0.20, 95% CI 0.07-0.55), and sarcoma located in upper limbs (HR 0.13, 95% CI 0.04-0.44) or trunk (HR 0.28, 95% CI 0.10-0.81). RT was nevertheless identified as efficient in the subgroup of patients with larger tumours (HR 0.43, 0.24-0.78).

Figure 3.

Subgroup univariable analysis for local control.

Results from multivariable analyses for LC, DFS, and OS are shown in Table 2. In this series, after adjustment for confounding factors, patients with older age and larger tumours were more at risk in terms of LC, DFS, and OS; large (≥5 mm) resection margins were associated with a better OS (HR 0.64, 95% CI 0.43-0.94; P = .0225) and a trend for better DFS was observed (HR 0.84, 95% CI 0.63-1.12, P = .2392); grade 3 tumours were more at risk in terms of DFS (HR 1.69, 95% CI 1.25-2.30; P = .0008) and OS (2.14, 95% CI 1.41-3.25; P = .004). Performing a preoperative MRI was associated with better LC (HR 0.42, 95% CI 0.22-0.82; P = .0112). RT remained significantly associated with a lower risk of local recurrence (HR 0.25, 95% CI 0.14-0.45; P < .001), showed significant association with better DFS (HR 0.59, 95% CI 0.42-0.82; P = .0019). No OS benefit was observed (HR 0.95, 95% CI 0.58-1.55, P = .8328). Similar results were reported in the population excluding patients with grade 1 tumours (Supplementary Materials S3 and S4).

Table 2.

Multivariable analysis of local control (LC), disease-free survival (DFS), and overall survival (OS).

	Local control (LC)		Disease-free survival (DFS)		Overall survival (OS)
	HR [95%CI]	P-value	HR [95%CI]	P-value	HR [95%CI]	P-value
Age
≥60 years vs <60 years	1.77 [0.98-3.22]	.0586	1.84 [1.36-2.50]	<.0001	2.35 [1.57-3.53]	<.001
Tumour size (cm)	1.16 [1.08-1.24]	<.0001	1.10 [1.06-1.13]	<.0001	1.07 [1.03-1.11]	.003
Tumour location
Lower limb vs trunk	1.47 [0.74-2.93]	.2718	1.23 [0.86-1.75]	.2624	1.09 [0.68-1.75]	.7186
Upper limb vs trunk	1.53 [0.62-3.78]	.3566	0.89 [0.54-1.46]	.6321	0.85 [0.44-1.64]	.6379
Tumour grade
Grade 3 vs grades 1-2	0.78 [0.42-1.44]	.4262	1.69 [1.25-2.30]	.0008	2.14 [1.41-3.25]	.004
Tumour depth
Deep vs superficial	0.73 [0.39-1.37]	.3308	1.22 [0.84-1.77]	.2869	1.69 [0.98-2.89]	.0577
Preoperative MRI
Yes vs No	0.42 [0.22-0.82]	.0112	NI		NI
Margin size
Large (≥5 mm) vs close (<5 mm)	0.78 [0.44-1.37]	.3873	0.84 [0.63-1.12]	.2392	0.64 [0.43-0.94]	.0225
Radiotherapy
Yes vs No	0.25 [0.14-0.45]	<.0001	0.59 [0.42-0.82]	0.0019	0.95 [0.58-1.55]	.8328

HR [95%CI]: hazard ratio with 95% confidence interval. NI: variable not included in the multivariable analysis (P-value >.1 in univariable analysis). P-value <0.05 was considered statistically significant (bold).

Discussion

This series confirms that adjuvant RT improved LC in all patients with R0 resection of trunk and limb STS, but no subgroup of patients may be exempted from adjuvant RT, and clearly identified as eligible to surveillance exclusively. In the present study, 5 y- and 10 y-LC were 88.0% [84.5-90.8] and 77.5% [71.1-82.6], 5 y- and 10 y-DFS were 65.3% [60.8-69.4] and 51.0% [45.1-56.7], respectively, and 5- and 10 y-OS rates were 80.1% [76.1-83.6] and 69.6% [63.7-74.6], respectively. To note, the first relapse event was metastasis in 115 (21.9%) patients.

These results are consistent with recently published results although comparisons between series requires cautious interpretations. The risk of local relapse at 5 and 10 years reported in the literature is very variable, and ranged from 7% to 31%, and OS ranged from 57% to 85%. Despite improvements over time in terms of LC and OS,²² variations among studies still exist, according to the classical prognostic factors identified as age, tumour location, grade, size, depth, and treatments received including RT.^23–31

The present series identified RT, tumour size, and preoperative MRI assessment as independent prognostic factors for LC. In contrast, ≥5 mm resection margins were not significantly associated with improved LC. An early management in reference centers, including preoperative MRI, is highly encouraged in patients with sarcoma, and provides a valuable support for preoperative local staging modalities.¹ Recently, advances in MRI highlighted that tail signs may be suggestive for myxofibrosarcoma known to be at higher risk of local recurrence and therefore supported wider resection margins.²⁹ Treatment of these rare tumours in reference centers is highly encouraged.¹ However, myxofibrosarcoma only accounted for 14% in the current dataset.

The rational for RT in some selected STS patients is still debated, and management without RT has been reported.^{16,17,19–21} Published guidelines recommended the use of RT as standard treatment in patients with high-risk STS (stages 2 and 3).^32–34 However, RT is not required in low-grade, <5 cm (Stage IA) tumours, or after compartmentectomy.^6,32–34 In other situations, and specifically in low-grade, large tumours (≥5 cm, IB), or in high-grade tumour with large margins (>10 mm), referral to multidisciplinary tumour board in reference centers is required. R0 surgery, that is, with optimal tumour margin size was considered as a major prognostic factor for LC, and several series suggested that suboptimal surgical margin size^9,35 potentially associated with increased rates of local relapse may be counterbalanced by RT.^22,36 Other recent series reported that postoperative RT showed a significant benefit in patients with <10 mm surgical margins.³⁷ Baldini et al²⁰ and Austin et al³⁷ showed less recurrences in patients with margins ≥10 mm and consider large margins as the key selection factor for RT avoidance. Gundle et al reported a risk of relapse reduced to 1% in patients with large surgical margins (≥5 mm), treated with RT or not.¹⁰ In the present series, RT improves LC in patients with STS with surgical R0 resection, regardless the size of surgical margins (≥1 to <5 mm: HR 0.43, 95% CI 0.21-0.86; ≥5 mm: HR 0.45, 95% CI 0.24-0.86, interaction P = .89). In contrast, our series reported results from patients operated in reference centers, and only considered patients with margin size >1 mm.¹⁰ Our series showed that margin size are not significantly associated with better LC, neither better DFS. Prognosis related to margins not only result from margin size but also from the expertise of surgeons, and anatomo-pathologist interpretation. Our results encourage the recommendation of postoperative RT in all operated patients even in patients with large (≥5 mm) resection margins.

The impact of RT on LC was reported in subgroup univariable analysis in this series in older patients, small tumours, and low-grade tumours, and in upper limb and trunk primaries. The lower impact of irradiation on high-grade tumours may be partly explained by the occurrence of numerous competitive events metastases and deaths as revealed by the competitive risk analysis (Supplementary Material S2). Indeed, no conclusion can be ruled out and small size and/or superficial tumours benefit from RT and our results are not in favour of de-escalation.

A pragmatic approach would be to identify specific histological subtypes for which RT avoidance should be considered and could be proposed with close surveillance. Indeed, some subtypes are more likely to develop local recurrences whereas metastatic recurrences are more frequently observed in other subtypes.^38,39 This heterogeneity probably contributes to limit the identification of statistically significant difference in survival. Namely, myxofibrosarcoma more likely relapse locally compared with other subtypes.³¹ Another option would be to consider indications for RT on a case-by-case basis, and the use of a nomogram may be helpful.^40,41 Cahlon et al reported a nomogram including risk factors for local recurrence after definitive surgery and highlighted that age, tumour size, margin status, histology, and grade allowed to estimate the individual risk of 3 y- and 5 y-local recurrence.⁴⁰ However, cautions should be taken, since low-grade liposarcomas were included, whereas they are not anymore considered as eligible for adjuvant RT. The PERSARC study included patients with pre- or postoperative irradiation or exclusive surgery.⁴¹ This nomogram was originally published in 2017 and retrospective series subsequently revealed that the local recurrence rates reported may have been overestimated.³ Prospective results are currently expected (NCT05741944).

A critical issue is the absence of RT impact on OS. Our results are consistent with previous studies and identify age at diagnosis, tumour size, grade, depth, margin size. The absence of RT impact on OS is consistent with results from the two randomized studies published to date.^11,12 It may result from a lack of power as suggested by Qu et al⁴² and Baldini et al⁴³ comparing with results in breast cancer⁴⁴ and considering that 4 local relapses need to be avoided to prevent one death. In contrast, several meta-analyses had reported that RT had a positive impact on survival particularly in subgroup of patients with high-grade sarcoma.^42,45

The present series showed that 71 (11.6%) patients had local recurrence. Among them, 23 (32.4%) patients were died at last follow-up. After local recurrence, 5 y-OS was 63.2% [48.5-74.8], which is quite satisfactory in comparison with other series showing a 5 y-OS of 29% in patients with local recurrence in the first year after treatment.²⁶

This emphasizes the critical value of detecting and treating local recurrences. In addition to metastasis and death related to local recurrence,⁴⁶ the associated risk of amputation or mutilating surgery also need to be considered.⁴⁷ These results suggested a clinically meaningful impact of postoperative RT on local recurrence. In the absence of demonstrated impact on survival, RT still need to be discussed in multidisciplinary board including anatomo-pathologists and surgeons to adequately assess the associated risk of local recurrence, metastatic event, or major surgery (amputation) related to local recurrence.

Conclusion

This propensity score matching highlights the role of adjuvant RT in LC and DFS but did not evidence any impact on survival in patients with R0 resection, regardless margin size. This study underlines the importance of preoperative MRI to ensure the high quality of care. Our series showed beneficial therapeutic effects of RT regarding LC in all patients. Cases need to be discussed in multidisciplinary board to decide whether additional factors such as histologic subtype, history of the patient, salvage surgery in relapsing tumours, and metastatic risks should be taken into account in conjunction with the classically used factors. Currently, sarcoma management favours the use of neoadjuvant therapy in patients with locally advanced or histologically aggressive tumours. Primary surgery is only indicated in sarcomas with a very good prognosis or at high risk for poor wound healing.

Author contributions

Conceptualization: Marie-Pierre Sunyach, Anne Ducassou, Jean-Yves Blay; Data curation: Françoise Ducimetière; Formal analysis: Amélie Lusque; Funding acquisition: Jean-Yves Blay; Investigation: Marie-Pierre Sunyach, Cécile Le Péchoux, Amélie Lusque, Paul Sargos, Sylvie Helfre, Juliette Thariat, Laurence Moureau Zabotto, Delphine Lerouge, Carmen Llacer, Augustin Mervoyer, Guillaume Vogin, Christine Chevreau, Jean-Yves Blay, Martine Delannes, Anne Ducassou; Methodology: Marie-Pierre Sunyach, Anne Ducassou; Project administration: Marie-Pierre Sunyach, Anne Ducassou, Françoise Ducimetière; Resources: Jean-Yves Blay; Software: Françoise Ducimetière, Amélie Lusque; Supervision: Marie-Pierre Sunyach, Anne Ducassou; Validation: All; Writing—original draft: Marie-Pierre Sunyach, Anne Ducassou; Writing—review & editing: All authors.

Supplementary material

Supplementary material is available at BJR online.

Funding

None declared.

Conflicts of interest

The authors declare that they have no competing interests.