Marginal miss or radioresistance? The pattern of local recurrence after operation and 3D planned radiation treatment in soft tissue sarcoma of the extremities and the limb girdles; an analysis based on image fusion

Abstract

Background. Most local recurrences have developed in the clinical target volume in previously published series after combined modality treatment for soft tissue sarcoma. However, marginal misses were seen in almost 20% of the patients. The aim of the present study was to determine the location of the recurrence and the total dose at the centre point of the local recurrence for future radiation therapy planning. Material and methods. We included only patients with images in digital form, during 1999–2006 (n = 17), treated for soft tissue sarcoma with combined surgical therapy and radiotherapy at Helsinki University Central Hospital. Image fusion was used to determine the location of the recurrence in relation to radiation therapy target. Results. In the present study utilising digital image fusion, in patients with 3D CT-based radiation treatment planning the risk of marginal miss was low as only one patient of 17 relapsed outside the target. Estimated mean radiation dose at the site of local recurrence was 49.1 Gy in patients with positive margins and 48.1 Gy in patients with negative margins. Conclusion. The risk of marginal miss in soft tissue sarcoma is low after modern 3D planned radiation treatment combined with surgery. More generous use of boost might improve in-target local control.

Radiation therapy improves local control after limb-sparing surgery in soft tissue sarcoma (STS) [1]. It is, however, not without side effects such as oedema, skin reactions, reduced limb strength and range of motion [2]. To the best of our knowledge no randomised studies investigating the effect of radiation dose or target margins on local control have been published in STS. Several retrospective studies have analysed the association of local control in STS with radiation therapy field size and dose [3–6]. None of the previous studies have used three-dimensional (3D) treatment planning in all patients nor digital image fusion for analysing the exact location and radiation dose at the site of recurrence. This method has previously been successfully used in a number of studies of recurrence pattern in head and neck carcinomas after radiotherapy [7–9], but to our knowledge never before in STS.

In 1987 a multidisciplinary STS group started a treatment protocol for STS at Helsinki University Central Hospital. A prospective treatment protocol was set up. For this study, we analysed all patients with STS of the extremity and the limb girdles, who received conventional radiation therapy with computed tomography (CT)-based treatment planning and who experienced a local recurrence. The aim was to determine the location of the recurrence and the total dose at the centre point of the local recurrence by using digital image fusion.

Material and methods

The study was approved by the Ministry of Social and Health Affairs and the ethical committee. The treatment protocol is described in detail elsewhere [10]. Surgical resection was the primary treatment for all patients, and patients with intra-lesional or marginal surgery were offered radiation therapy. Specimens were fixed in formalin, surfaces were painted, specimens dissected and margins measured from histological slides. Smallest margins and locations were reported. This method is recommended and proved reliable by the Scandinavian Sarcoma Group [11]. Wide margin was defined as the smallest margin being ≥ 2.5 cm (smaller margin is accepted if uninvolved fascia is removed). If marginal or wide surgical margins were not achievable preoperative treatment with radiation alone or combined with chemotherapy was considered. For the present study, we included only patients with images in digital form from year 1999 onwards to 2006 treated for soft tissue sarcoma with combined surgical therapy and radiotherapy at our institution.

CT-based treatment planning and target volume delineation

Only patients with a CT-based treatment planning were included in this study. Individual fixation methods either with polyurethane or vacuum cushions were used. The PTV included the whole involved muscle compartment or the tumour and surgical cavity with at least 2–3 cm margin in the transversal direction, and with a longitudinal margin of at least 5 cm. All patients in this series were treated with 3D conformal radiation therapy using ICRU50 criteria as guidelines for treatment planning. Our treatment guidelines recommend surgery followed by postoperative radiation therapy in case of inadequate margin with or without chemotherapy. The radiation dose was 50/2 Gy (2 Gy/day) for patients receiving no chemotherapy. However, one patient received 52.2/1.8 Gy and one patient 50.4/1.8 Gy (Table I). A 10–20 Gy boost was delivered to a smaller area to the surgical bed after microscopically positive and/or marginal margins. At present, patients less than 70 years and with adequate performance status are offered adjuvant chemotherapy if the tumour malignancy grade is high and the tumour fulfils at least two of the following criteria: more than 8 cm in size (in synovial sarcomas 5 cm), necrosis, or vascular invasion. Adjuvant chemotherapy consists of six cycles of doxorubicin and ifosfamide at three-week intervals. As part of combined chemotherapy and radiation therapy patients received hyperfractioned radiation therapy of 42/1.5 Gy twice a day between chemotherapy cycles: 30 Gy was usually given between the second and third chemotherapy cycle and 12 Gy after the third cycle. This protocol was based on our favourable experiments in Ewing's sarcoma [12]. The advantage of hyperfractionated radiotherapy sandwiched between chemotherapy cycles is that neither chemotherapy nor radiation needs to be postponed in patients with large recurrence risk where undue delay may be detrimental to tumour control. A similar but not identical treatment protocol has recently been documented to be feasible also in non-Ewing soft tissue sarcoma [13].

Table I. Patient, tumour, treatment and local recurrence characteristics.

Patient	Sex	Age (years)*	Histology	Grade^†	Size (cm)	Site	Smallest surgical margin/cm	Timing of RT	Dose/Fraction (Gy)	Boost (Gy)	Number of recurrence foci	Location of Recurrence(s)	Dose(s) at center of recurrence(s)	Dose relative to total prescription dose (%)	V(95%) initial, boost (cm^3)
1	M	44	MFH	IV	20	groin	Intralesional	PRE	42/1.5 BID	12/2	2	Boost, boost border	50.8, 50.7	94, 93	8335, 5038
2	F	45	LS	III	10	thigh	Intralesional	POST	50/2	10/2	1	Boost border	57.7	96	1510, 1141
3	M	46	SS	IV	15	groin	Intralesional	PRE	42/1.5 BID	18/1.5	4	Boost border, border, border, outside	56.3, 40.5, 38.0, 28.9	93, 67, 63, 48	12799, 7106
4	M	48	MFH	IV	14	gluteal	Intralesional	POST	50/2	10/2	2	Boost, boost border	59.1, 59.2	98, 98	4034, 2043
5	M	52	MPNST	II	7.7	knee	Intralesional	POST	50/2	16/2	1	Boost	66.1	100	1526, 895
6	M	53	MFH	IV	14	thigh	Intralesional	POST	42/1.5 BID	18/1.5	1	Boost border	49.7	82	2454, 875
7	M	47	MFH	IV	12	upper arm	Intralesional	POST	50/2	No	1	Target	49.0	98	2225
8	M	49	SS	IV	3	knee	Intralesional	POST	42/1.5 BID	18/1.5	1	Target, outside boost	42.4	70	2910, 955
9	M	51	Sarcoma NOS	IV	17	thigh	Intralesional	PRE	50/2	No	1	Target	49.2	98	5412
10	M	54	Sarcoma NOS	II	11.5	thigh	Marginal/not reported	POST	50/2	No	1	Target	52.8	105	1771
11	M	56	MFH	IV	15	thigh	Marginal/0.1	POST	42/1.5 BID	No	2	Target, target	45.6, 46.2	108, 110	3523
12	M	59	MFS	II	no data	elbow	Marginal/0.5	POST	50/2	No	1	Target	50.8	101	326
13	M	60	LS	I/III^‡	9.5	thigh	Marginal/0.7^‖	POST	50.4/1.8	No	1	Target	52.0	103	4367
14	M	58	Sarcoma NOS	IV	25	thigh	Marginal/0.5	POST	52.2/1.8	No	1	Border	49.2	94	8551
15	M	57	MFH	IV	10	elbow	Marginal/0.3	POST	50/2	No	1	Border	47.6	95	1590
16	M	55	LS	III	7	axilla	Marginal/0	POST	50/2	10/2	1	Outside	0	0	1993.1149
17	M	43	Sarcoma NOS	III	6	lower arm	Wide/0.1^¶	PRE	42/1.5 BID	No	1	Target	40.6	96	1107

*At the time of diagnosis of the primary tumour. ^†Scale I-IV. ^‡Grade I liposarcoma with a focus of grade III liposarcoma. ^‖To grade III focus. ^¶Uninvolved fascia removed. BID, two fractions daily; LS, liposarcoma; MFH, malignant fibrous histiosytoma; MPNST, malignant peripheral nervous sheath tumour; NOS, not otherwise spesified; RT, radiation therapy; SS, synovial sarcoma; V(95%), volume receiving 95% of prescription dose.

Image registration and dosimetric evaluation

The diagnostic magnetic resonance imaging (MRI) or CT images in which the local recurrence was first detected were imported into the treatment planning system (TPS: Eclipse 8.6, Varian Medical Systems Inc., Palo Alto, CA, USA). The visible volumes of local relapses (LR) were delineated in the diagnostic images in TPS by a radiologist together with an oncologist. The diagnostic images were manually registered with the CT images acquired for treatment planning of radiation therapy. The registration was performed using bony landmarks and visible soft tissue structures in the immediate surroundings of the LR. The LR contours were copied into treatment planning CT images for dosimetric evaluation. The slice thickness of treatment planning and diagnostic images was 0.25–2 cm. Four patients received preoperative radiation therapy and for these patients the registration was performed between preoperative treatment planning CT images and postoperative diagnostic MR or CT images. The maximum error in the localisation of the LR caused by geometrical uncertainty of the registration was assessed by visual evaluation, error ≥ slice thickness. The error was estimated to be within the range of 0.7–2.1 cm.

In general, the LR contours were nearly spherical, and the exact location of the relapse was assumed to lie at the centre of the LR structure. Hence, the total dose at the centre point of the LR was recorded (Table I). For multifocal LRs total dose at the centre of each LR was recorded. LRs were classified as “target” (LR completely inside the target volume), “boost” (completely inside the boost target volume), “border” (partly outside the target), or “outside” (completely outside the target). Here the target volume was defined as the volume enclosed by the 90% isodose surface in the planning CT. For each patient with lesions classified as geographical miss, we inspected the dose plans to ensure that planning target volumes were defined according to our treatment guidelines.

Results

During 1999–2006 radiation therapy was administered to 108 patients with STS. Eighteen patients with local recurrence were identified. One patient was excluded, because diagnostic CT images of the patient were unavailable for analysis. The mean age of the 17 patients at diagnosis of the primary tumour was 61 years, range of 43–60 years (Table I). Surgical margin was defined as intra-lesional with only microscopic residual in nine patients and of these patients seven received a radiation boost. Seven patients had marginal definitive surgery. One patient relapsed following wide surgery and radiation therapy.

Four patients had a multifocal local recurrence. All nine patients with intra-lesional surgery developed recurrence in the target volume. Three of these patients had multiple recurrences, of which some were outside the boost volume and one also outside the target. Two of the patients had recurrence exclusively at the boost border. Four patients of seven with marginal surgery had their local relapse in the target, two at the border and one outside the target. The only patient with wide surgery experienced an in-target relapse. In both patients with relapse at the border had most of the relapse volume within the treated volume and the longest distances from the 90% isodose were 1.3 cm and 0.5 cm, respectively. One patient had both the primary tumour and the relapse extracompartmentally and the other primary tumour intracompartmentally under the investing fascia which was removed during the operation and the relapse subcutaneously. Therefore both these patients had relapse in risk area.

Radiation doses were estimated at the centre of all the 22 local recurrences recurring in boost, in-target or at the border. The mean radiation dose in patients with positive margins was 49.1 Gy, range 28.9–66.1 Gy and in patients with negative margins, mean 48.1 Gy, range 40.6–52.8 Gy (Table I). One patient with intra-lesional surgery recurring at the boost border had an estimated dose of 82% of the prescribed target dose in the boost. All the other patients had an estimated dose of more than 90% of the prescribed dose (mean 99%, range 93–110%) at the centre of the recurrence except for one patient with four recurrences in whom three lesions had lower doses.

Discussion

The questions of optimal radiation dose and target size in STS have never been addressed in prospective randomised trials. Thus one way of obtaining information of these issues is to analyse the pattern of local recurrence in patient series treated with radiation according to a well-defined treatment protocol. Our series is small although our institution is the largest tertiary STS referral centre in our country. Our treatment protocol was set up in 1987 and it emphasises aggressive local approach with skillful plastic surgeons performing the resection and also performing the necessary reconstruction. Radiation therapy is administered when surgical margins are inadequate.

Image fusion technique used in the present study has previously been successfully used in head and neck carcinomas in assessment of local recurrence pattern but to our knowledge not previously in soft tissue sarcomas [7,8]. Each LR was analysed in relation to radiation dose and classified in relation to radiation target volume. Especially in radiation therapy of extremities the fixation can be challenging and the treatment accuracy may be reduced due to rotation of the limb. The uncertainty in treatment setup of the extremities may easily be over 1 cm [14], which is comparable with our estimated uncertainty of image registration. Taking this geometric uncertainty into account, three LRs classified as “boost border” and three LRs classified as “border” in our analysis may have situated inside boost or target volume, respectively, and hence interpreted to belong into a wrong category.

Only one patient developed a local recurrence completely outside the radiation target volume, and may therefore be considered a marginal miss, although several patients with multiple recurrences had some lesions that potentially may have been geographical misses.

The treatment guidelines at the Massachusetts General Hospital recommended a gross tumour margin of 1–1.5 cm radially and 3.5 cm longitudinally with planning target volume expansion of 5–7 mm [3]. Doses varied from 44 to 50.4 Gy, with postoperative boost of 10–20 Gy after close or positive operative margins. All five local failures occurred within the clinical target volume (CTV), two also extending beyond the CTV. At the Princess Margaret Hospital the recommended radiation treatment margins were 5 cm longitudinally beyond gross tumour or surgical bed and at least 1–2 cm axially [4]. The majority of the patients (73%) received preoperative irradiation, 20% combined with postoperative boost. Forty-nine recurrences (82%) occurred within the target volume, nine out of target, and two at the target border. In a British study 10.5% of patients developed local recurrence [5]. Most patients had received 50 Gy in 2 Gy fractions to the whole involved muscle or muscle compartment followed by a 10 Gy boost to the original tumour extent with a 2-cm margin. Seventeen of the 25 patients with a local recurrence recurred within the radiation target volume, four within the boost area and four outside the irradiated area. The assessment of the location of local recurrence relative to the irradiated volume in these studies was done comparing simulation films with later imaging of the recurrence, which probably is a less accurate method than the image fusion used in our study. Despite this uncertainty the results of the above mentioned studies are remarkably consistent indicating that a minority of recurrences, less than 20%, develop outside the irradiated area after postoperative irradiation. In a Scandinavian retrospective study with similar treatment protocol to ours 14% of the local recurrences were out-of-field [6]. The low ratio of marginal misses (1 of 17) in our study where the majority of patients like in the Royal Marsden series was treated with postoperative radiation may at least partly be due to technical development of treatment planning technology because all our patients were treated relatively recently and all with 3D treatment planning technique and at least with weekly portal imaging in contrast to the two older series.

Several aspects make comparisons of the present series and previous ones difficult. Some patients in the previous studies were treated without CT planning, while all our patients were treated according to 3D CT plans. Moreover, a large proportion of the patients in previous studies was treated with preoperative radiation while the guidelines at our institution recommend surgery followed by postoperative radiation. The results from the Massachusetts General Hospital on patients treated exclusively with preoperative radiation suggest that the proportion of marginal misses may be even lower than after postoperative radiation [3]. One reason for this may the more easy treatment planning in patients with an intact tumour. Moreover 15% of the recurrences in the Princess Margaret Hospital series were also outside the target although the majority were treated with preoperative radiation [4].

A recent meta-analysis based on several non-randomised studies suggests that the risk of local recurrence might be lower after preoperative than after postoperative radiation [15]. However, the only randomised trial published hitherto showed no evidence of improved local control with preoperative radiation [16]. Thus the seemingly better outcome after preoperative irradiation may be due to patient selection as well.

According to the guidelines by a multi-institutional expert team a planning target volume encompassing the tumour in preoperative radiation, and the tumour and surgical bed after postoperative treatment, with approximately 5 cm margin longitudinally and about 2.5 cm transversely is proposed [17]. The results of the present study as well as previously published series with similar local treatment guidelines indicate that these margins should be adequate in most cases, provided modern 3D planning and fixation technique are used. An ongoing randomised study will hopefully shed light on whether margins may be further reduced without compromising local control [18].

All patients in our series recurred in the target after intra-lesional surgery and most of them in the boost volume. Thus every effort should be made to achieve negative surgical margins before postoperative radiation. When patients with intra-lesional surgery are excluded six of seven (one with wide margins) patients recurred in the target volume and one at the border with a mean dose in the middle of local recurrence of 48.4 Gy, range 40.6–52.8 Gy. The results of the present study and previous studies indicate that local recurrence may be less related to microscopic disease outside the radiation target volume than to inadequate doses or intrinsic radioresistance. Two patients with intra-lesional surgery in our series relapsed exclusively in the boost border and might have benefited from a more generous boost volume. Recurrences in the target but outside the boost area seen in both our and the Royal Marsden series may indicate that dose matters. Thus treatment results may be further improved by more generous use of a boost, whenever normal tissue tolerance allows. A retrospective dose response study from the M. D. Anderson Cancer Centre found doses of ≥ 64 Gy to correlate independently with improved local control [19]. No such correlation was present in a Scandinavian Sarcoma Group study [6]. Compared to the present series the doses in the M.D. Anderson study were generally higher with a median dose of 64.0 Gy with a range of 40–75 Gy. The doses in our series were lower partly since many patients were treated with a combined chemotherapy-radiation protocol where radiotherapy was given in a bi-daily fractionated fashion, where the biological effect of radiation is estimated to be higher, especially since combined with radiosensitising chemotherapy. Our current guidelines based on the Scandinavian Sarcoma group guidelines recommend radiation boost after intra-lesional surgical margins but not if negative histological margins have been achieved. The results of present series as well as previous ones, however, suggest that a boost should be considered also in cases with marginal surgery and negative histological margins. A higher dose is often difficult due to large tumour and surgical bed size and proximity of radiosensitive structures like joints and major nerves. With intensity-modulated radiation therapy higher doses can be reached in the target volume with avoidance of excess dose to nearby structures [20]. With increasing use of MRI-based treatment planning delineation of anatomy can be accomplished more accurately, which may enable better shielding of radiosensitive structures.

In conclusion, marginal misses and problems with dose inhomogeneity seem to be rare with modern treatment planning and fixation technique. Radiation boost does not compensate for the tumour burden after intra-lesional surgery. For positive margins a boost is warranted. The fact that both after marginal and especially after intra-lesional surgery the tumour tends to recur in the target indicates that higher doses utilising modern treatment techniques like IMRT may be beneficial.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

The study was supported by EVO funds, Helsinki University Central Hospital and Finnish Cancer Registry. Study sponsors had no role in the study design, in the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication.