Radiotherapy and hyperthermia with curative intent in recurrent high risk soft tissue sarcomas

Abstract

Purpose: Radiotherapy before or after resection is one of the pillars of treatment for localised high risk soft tissue sarcomas. Treatment intensification has been described with concurrent chemotherapy and hyperthermia. The aim of this study is to assess local control after multimodal treatment, focussing on the treatment of local recurrences after surgery only.

Patients and methods: Of 42 patients treated in a prospective protocol with radiotherapy and hyperthermia, nine were treated for isolated local recurrences without metastatic spread. Most patients were treated with trimodal therapy including chemotherapy with ifosfamide and underwent resection whenever possible. Median follow-up was 1.4 years.

Results: The treatment was well tolerated. Estimated disease free survival, distant metastases free survival and local control for the whole cohort after 1.5 years were 66, 73 and 88%, respectively. Neoadjuvant vs. adjuvant treatment influenced local control with a trend to statistical significance. Resection status did not influence local control. The cohort of patients treated for local recurrence after surgery alone had a significantly impaired local control compared to multimodal treatment at primary diagnosis (100 vs. 52%, p < 0.001).

Conclusions: With multimodal therapy including radiotherapy and hyperthermia local tumour control is achievable even in locally recurrent tumours. The clear-cut difference of the treatment of local recurrence in contrast to primary diagnosis might either reflect difficulties in diagnosis and treatment of local recurrences or biological aggressiveness of recurrent tumours. However, we recommend to consider multimodal treatment at primary diagnosis of high risk soft tissue sarcomas.

Keywords:

• Sarcoma
• radiotherapy
• hyperthermia
• recurrence
• local control

Introduction

The treatment of localised high-risk adult-type soft tissue sarcoma poses major challenges and requires close cooperation of a multidisciplinary team. High risk situations are defined by stage (grading, size, location) and in some studies also by age at diagnosis [1]. The main pillars of treatment are complete en-block resection and pre- or postoperative radiotherapy [2]. Yet, treatment results still give room for improvement. Local control is a major issue whenever complete resection with negative microscopic margins is not achievable with limb sparing surgery.

Whereas surgical therapy has become less aggressive over the last three decades, developing from amputation to limb-sparing approaches [3,4], efforts have been undertaken to intensify pre- and postoperative therapy.

As distant metastases are still the main type of recurrence, the addition of chemotherapy seemed promising to improve disease-free and overall survival. The agents most commonly used are doxorubicin and ifosfamide. Yet, the results of large multicentric randomised studies are conflicting. A pooled analysis of two STBSG-EORTC studies showed a significant benefit for disease-free survival [5], whereas a large randomised trial published in 2012 did not point toward an effect [6]. Thus, current recommendations state that neoadjuvant and/or adjuvant chemotherapy is a treatment option in high risk soft tissue sarcomas. Yet, the conflicting data warrant stringent benefit-risk analyses and open discussions with the patients. Chemotherapy is also applied in hyperthermic isolated limb perfusion at specialised centres [7,8]. A randomised study of neoadjuvant or adjuvant chemotherapy with or without regional hyperthermia showed a significant increase in disease free survival for combination treatment [9].

Pre- or postoperative radiotherapy has been established in multimodal treatment of high risk soft tissue sarcomas of the extremities [10]. Postoperative radiation with >60.0 Gy improves local control in high risk soft tissue sarcoma [11,12]. Preoperative irradiation with 45.0 –50.0 Gy leads to more acute wound healing complications after surgery, yet seems to have a benefit for late complications like fibrosis and joint stiffness in comparison to adjuvant treatment [13–15]. Advanced radiation techniques such as intensity-modulated radiotherapy (IMRT) and image guided treatment (IGRT) seem to further reduce the risk of late complications [16–18]. The role of radiotherapy is less clear for retroperitoneal sarcomas as there are no large randomised studies reported up to now. Single centre and database analyses seem to indicate a benefit for preoperative radiotherapy [19–21].

Radiotherapy combined with concurrent chemotherapy and/or hyperthermia was first described in 1999 by Sauer et al. [22]. Since then, several groups reported small retrospective series of concurrent radiochemotherapy either with interdigitated MAID chemotherapy and radiation [23,24] or with ifosfamide and radiotherapy [25–27], partly in combination with hyperthermia. The regimens seem feasible, yet conclusions about the efficacy of the treatment are limited by patient numbers and retrospective data collection.

We focussed our analysis on nine of 42 patients followed prospectively after treatment for localised high risk soft tissue sarcomas who were treated for local recurrences without metastatic spread after surgery alone. The results were compared with the whole cohort and patients treated multimodally at primary diagnosis.

Patients and methods

Patients with high risk soft tissue sarcoma treated with a combination of radiotherapy and hyperthermia were followed prospectively from 2012 to 2016. All patients were treated in a multidisciplinary setting in the Center for Soft Tissue Sarcoma, GIST and Bone Tumors of the Comprehensive Cancer Center Tübingen. Patients treated with combined radiotherapy and hyperthermia for better local control of single disease sites in a metastatic setting were excluded from the analysis as were patients who did not finish the planned treatment at least to a radiation dose of 30.0 Gy. All patients had pre-treatment staging consisting of at least history and physical examination, biopsy, cross-sectional imaging of the tumour region and laboratory data. Pulmonary metastases were excluded by chest computed tomography (CT).

42 patients with a median age of 56 years (range 19–76 years) were treated in the prospective protocol. Nine patients were treated multimodally for recurrences after surgery only without radiotherapy, whereas 33 received multimodal treatment at initial diagnosis. Six presented with tumours smaller than 5 cm and 36 with larger tumours. Median tumour size at diagnosis was 8.7 ± 0.7 cm. Seven patients were treated for sarcomas of the upper extremities, 16 for lower extremity tumoursand 19 for tumours of the trunk. Histopathology showed one well differentiated sarcoma, 11 intermediate differentiated tumours and 30 high grade sarcomas (according to FNCLCC). AJCC stage for all patients was IIa, IIb or III. Histopathological subtypes covered the whole spectrum of adult type soft tissue sarcomas. Patient and tumour characteristics are summarised in Table 1.

Table 1. Patient characteristics of the whole cohort of 42 patients, nine were treated for isolated local recurrences after surgery alone.

Age (years)
Median	56
Range	19–76
Gender (n, %)
Male	21	50%
Female	21	50%
Location (n, %)
Upper extremity	7	17%
Lower extremity	16	38%
Trunk	19	45%
Tumour size (n, %)
<5 cm	7	17%
5–10 cm	22	52%
>10 cm	13	31%
Grading according to FNCLCC (n, %)
G1	1	3%
G2	11	26%
G3	30	71%
Histology (n, %)
Liposarcoma	8	19%
Synovial sarcoma	5	12%
Leiomyosarcoma	3	7%
NOS	13	31%
Other*	13	31%
Manifestation (n, %)
Primary disease	33	79%
Recurrent disease	9	21%

Other* – Angiosarcoma, epitheloid sarcoma, malignnant peripheral nerve sheath tumour (MPNST), myofibroblastic sarcoma, sarcoma with myogeneic differentiation, myxofibrosarcoma.

Our institutional treatment guidelines follow the published IAWS protocol [28] for radiotherapy and chemotherapy of localised high risk soft tissue sarcomas. Patients with large, poorly differentiated tumours (mostly in the neoadjuvant setting) under the age of 65 were offered pre-irradiation chemotherapy with 4–6 cycles of doxorubicin (60 mg/m² day 1) and ifosfamide (3,000 mg/m² days 1–3), repeated on day 22. Most patients were treated with two cycles of a reduced dose of ifosfamide without doxorubicin concomitantly during radiotherapy. The same ifosfamide regimen (3,000 mg/m² days 1–2) was applied in patients treated with concurrent radiochemotherapy without sequential chemotherapy.

Hyperthermia was offered to all sarcoma patients scheduled for radiotherapy presenting without exclusion criteria such as severe cardiovascular diseases, metallic implants or claustrophobia and was performed as MR-controlled partial body hyperthermia (PBH, 20 patients), regional deep hyperthermia (RHT, 12 patients) or superficial hyperthermia (10 patients). Quality assurance guidelines according to the recommendations of the European Society for Hyperthermic Oncology were considered [29]. For PBH the MR-guided hyperthermia unit BSD 2000/3 D MRI (Pyrexar Medical, formerly BSD medical corporation, Salt Lake City, UT) with Sigma-Eye or Sigma-30 applicator was used. For 14 of 20 PBH patients hyperthermia treatment planning was done using the software SigmaHyperPlan^® V2.01 (Dr. Sennewald Medizintechnik GmbH, Munich, Germany) to optimise steering of phases and/or amplitudes of the applicator. MR-temperature control every 10 min with SigmaVision^® V1.01 (part of SigmaHyperPlan^®) was used for online adaptation of steering parameters. Patients with RHT were treated with the hyperthermia system BSD 2000/3 D (Pyrexar Medical, formerly BSD medical corporation, Salt Lake City, UT) with Sigma-30 or Sigma-60 applicator. For superficial hyperthermia the same system with the spiral applicator SA-115 was used. Hyperthermia was performed twice weekly. Depending on the tumour localisation temperature probes were located on the skin, intraluminally in bladder, rectum and vagina or in rare cases interstitially in the tumour. Target temperature was between 40 and 43 °C for at least 60 min and the maximum treatment duration per hyperthermia session was 90 min.

There are major obstacles for direct thermometry in sarcoma patients. Placement of direct thermometry probes bears the risk of implant metastases [30,31]. In addition, with the trimodal therapy as applied in most of our patients, chemotherapy might lead to leukopenia and thus increase the risk of infectious complications endangering the completion of radiotherapy. Thus, for quality assurance, indirect MR-thermometry [32] was analysed retrospectively for four patients treated with MR-guided PBH for at least three treatments per patient with SigmaVision^® V1.01. Automatic temperature evaluation in SigmaVision^® V1.01 reports T90 for the contoured volume, however the given values include the 30 min heating time until the treatment temperature was reached (defined as T90*). As heatmaps were inhomogenous throughout the tumour volume due to MRI artefacts, small intratumoral volumes with homogenous MR signals were contoured manually for each treatment. Consequently, given values might correspond best to T10 to T20 of the whole tumour. Mean T90* values over the whole treatment time were calculated as well as CEM43° [33,34] per treatment.

All patients eligible for surgery underwent wide resection. In the case of extremity sarcomas, limb sparing surgery was performed in all patients. Most patients referred to our centre for diagnosis and treatment planning were scheduled for neoadjuvant (chemo) irradiation with hyperthermia. Details about the treatment are given in Table 2 and Figure 1.

Figure 1. Schematic overview of the treatment in the cohorts of patients treated for local recurrences after surgery alone and patients treated multimodally at primary diagnosis, as well as patients screened but not eligible for the analysis.

Table 2. Treatment characteristics of the whole cohort.

Surgery (n, %)
R0	27	64%
R1	9	21%
Macroscopic tumour	2	5%
Missing data	4	10%
Sequential Doxorubicin/Ifosfamide (n, %)
Chemotherapy	19	45%
No chemotherapy	23	55%
Radiotherapy (n, %, median (range) dose)
Neoadjuvant	26	62%	50.4 (45.0–50.4) Gy
Adjuvant	14	33%	60.0 (50.0–66.4) Gy
Defintive	2	5%	65.0 (60.0–70.0) Gy
Hyperthermia (n, %, median number (range) of treatments)
Superficial hyperthermia	10	24%	10 (6–13)
Regional deep hyperthermia	12	28%	9 (2–12)
MR-guided partial body hyperthermia	20	48%	10 (6–13)
Concomitant ifosfamide (n, %)
Chemotherapy	35	83%
No chemotherapy	7	17%

Evaluated parameters included estimated disease-free survival, distant metastases-free survival and local control. Factors possibly influencing local tumour control were tested for their effect on recurrences. Patients receiving multimodal treatment for local recurrence without metastatic spread after surgery only were compared to patients treated multimodally at primary diagnosis. Statistical analysis consisted of Kaplan-Meier estimation for survival curves, which were compared with the log-rank test as well as in multivariate analysis. Means are given ± standard errors and were compared with two-sided Student’s t-test for independent samples. The statistical analysis was performed with the software package SPSS 24 (SPSS Inc., Chicago, IL). Statistical significance was defined for a p values <0.05, a trend to statistical significance for p values between 0.05 and 0.10.

Results

Overall the treatment was well tolerated. Radiotherapy was completed in all but one patient. The treatment termination in this case was not indicated for toxicity, but for massive progression of the tumour (trunk, undifferentiated sarcoma, NOS, grade 3) with new metastatic manifestations. The patient was started on palliative systemic therapy with trabectedin, which has stabilised the disease for 1.5 years. The patient is still on treatment. Full dose concomitant chemotherapy with ifosfamide was administered in 26 of 33 patients for whom the treatment had been planned. 33 patients completed at least one cycle of chemotherapy (50% of the planned dose). Hyperthermia was terminated for severe skin toxicity due to the combination of radiotherapy and hyperthermia, oedema of extremity or lack of tolerability of the bolus pressure or circulatory problems during treatment. 28 patients completed all planned treatments (ten or more) and 34 patients completed at least eight treatments (Figure 2).

Figure 2. Multimodal treatment with radiotherapy and hyperthermia complemented with concomitant chemotherapy in 33 patients. Radiotherapy was completed in all but one patient. 34 of 42 patients underwent eight or more hyperthermia treatments twice weekly. 26 of 33 patients completed the full two cycles of concomitant chemotherapy with ifosfamide.

For four patients undergoing neoadjuvant treatment with PBH, indirect MR-thermometry indicated that the treatment goal of 40–42 °C was reached, CEM43° was high overall, yet with rather large standard errors (Table 3).

Table 3. Indirect MR-thermometry data were evaluated for four patients in the neoadjuvant setting.

	T90* Mean ± SE	CEM43 °C/treatment Mean ± SE
Patient 1	42.05 ± 0.51	12.93 ± 3,56
Patient 2	41.86 ± 0,20	16.64 ± 2,96
Patient 3	41.44 ± 0,24	9.09 ± 2,60
Patient 4	41.31 ± 0,38	12.27 ± 4,57

T90 is analysed automatically by SigmaVision^® V1.01 for each manually contoured volume of interest and is calculated over the whole treatment time including the 30 min heating phase (T90*). As manually contoured volumes (in regions of homogenous MR signals without artefacts) were significantly smaller than the tumour volume, given values might best reflect T10 or T20 for the whole tumour. CEM43 °C was calculated with the mean T90* over the 60 min treatment time. Means ± standard error of the results of all evaluable treatments per patient are reported.

The median follow-up of the patients was 1.4 years. The estimated mean overall survival was 4.1 ± 0.13 years, the 1.5 year estimated overall survival was 96%. Mean disease free survival, distant metastases free survival and local control were 3.0 ± 0.29 , 3.3 ± 0.28 and 3.8 ± 0.21 years, respectively. The 1.5 year rates were 66%, 73% and 88% for disease free survival, distant metastases free survival and local control, respectively (Figure 3). All local failures were located inside of the irradiated volume.

Figure 3. Estimated overall survival (a), disease free survival (b), distant metastases free survival (c) and local control (d) are shown after a median follow up of 1.4 years.

Possible prognostic factors tested for their influence on local tumour control after multimodal treatment of the whole cohort were age at diagnosis, recurrent vs. initial diagnosis, localisation, grading, resection status, additional sequential chemotherapy and concurrent radiosensitizing chemotherapy. Tumour grading as well as localisation in the upper extremity vs. lower extremity vs. trunk were not prognostic. Adjuvant treatment yielded inferior results to neoadjuvant treatment, with a trend to statistical significance. Neither sequential nor concurrent chemotherapy influenced local tumour control. Local control was slightly worse for positive resection status compared to microscopically complete resection, yet without statistical significance. Local control was highly significantly worse after multimodal treatment of recurrent disease compared to patients treated with multimodal treatment at initial diagnosis (Figure 4). Disease free survival and distant metastases free survival were not significantly influenced by any of the tested factors (data not shown). Multivariate analysis of local control (recurrent vs. primary disease, neoadjuvant vs. adjuvant treatment and resection status) did not reveal independent prognostic factors.

Figure 4. Resection status did not influence local control (a). Adjuvant treatment in comparison to preoperative irradiation led to worse estimated local control with a trend to statistical significance (b). The most significant factor influencing local control was the treatment of local recurrences after surgery only in comparison to multimodal treatment at primary diagnosis (c). Multivariate analysis did not reveal independent prognostic factors among resection status, adjuvant vs. neoadjuvant treatment or treatment or recurrences.

The clear-cut difference between patients treated at primary diagnosis and with recurrent disease (although not significant in multivariate analysis) raised the question as how these cohorts were compared concerning known prognostic factors. Tumour size was significantly smaller for recurrent sarcomas (p = 0.04). Tumour grading was not different between the groups. On the other hand, more recurrent tumours were located at the trunk. Treatment characteristics were unfavourable in the recurrent tumours with more patients treated in the adjuvant setting or with macroscopic disease (Table 4).

Table 4. Comparison of known prognostic factors in the cohort of patients treated for local recurrences after surgery alone vs. patients treated multimodally at primary diagnosis.

	Primary diagnosis	Recurrent disease
Tumor size
Mean ± STE	9.4 cm ± 0.8 cm	5.7 cm ± 1.6 cm
Grading according to FNCLCC (n, %)
1	0	2 (22%)
2	9 (27%)	5 (56%)
3	24 (73%)	2 (22%)
Localisation (n, %)
Upper extremity	5 (16%)	2 (22%)
Lower extremity	14 (42%)	2 (22%)
Trunk	14 (42%)	5 (56%)
Sequential doxorubicin/ifosfamide (n, %)
Chemotherapy	16 (48%)	6 (66%)
No chemotherapy	17 (52%)	3 (33%)
Radiotherapy (n, %)
Neoadjuvant	24 (73%)	2 (22%)
Adjuvant	9 (27%)	5 (56%)
Defintive	0	2 (22%)
Concomitant ifosfamide (n, %)
Chemotherapy	28 (85%)	6 (66%)
No chemotherapy	5 (15%)	3 (33%)

Tumor size favoured the recurrent tumours, whereas localisation and radiotherapy modality were unfavourable in the cohort of recurrences. Grading did not differ between the patient groups.

Discussion

Our results demonstrate a significantly worse local control in patients treated multimodally for isolated local recurrences after surgery only. However, interpretation of our results is limited by small patient numbers and retrospective definition of the compared cohorts of recurrent tumours vs. therapy at primary diagnosis. The interpretation of local tumour control might also be influenced by salvage systemic therapy for distant metastases.

Distant metastases and disease free survival are in line with Stubbe et al. describing neoadjuvant radiochemotherapy with ifosfamide and hyperthermia [27]. Local control of the mixed population in this cohort is between local control described for neoadjuvant treatment [25] and definitive treatment [26] with the same regimen. As expected, it is lower than for patient populations restricted to neoadjuvant treatment [27,35].

Whereas resection status is one of the strongest predictors of local control in soft tissue sarcomas undergoing resection only [36,37], results are conflicting when surgery is combined with irradiation [38,39]. Dagan et al. reported a 69% five-year local control rate for contaminated margins vs. 95% for marginal or wide excision [39]. Ahmad et al. analysed a large retrospective cohort of 384 patients and did not find a significant difference in local control for negative and positive pathological resection margins [38]. Our data suggest similar results like those reported by Ahmad et al., yet are limited by low numbers of local recurrences and comparatively short follow up.

The most striking predictor for local recurrence was multimodal treatment of local recurrences after surgery alone in comparison to multimodal treatment at primary diagnosis. Despite the fact that recurrence was not a significant prognosticator for local control in multivariate analysis, the comparison of known prognostic factors in the cohorts of recurrent tumours and sarcomas treated after primary diagnosis revealed conflicting trends. Whereas tumour size favoured the group of recurrent tumours, grading did not differ between cohorts and recurrences had unfavourable characteristics concerning localisation and treatment modality.

There is only scarce data about local control and survival after curative treatment of local recurrences of soft tissue sarcoma. Liu et al. analysed the significance of resection status at primary diagnosis and recurrence for extremity tumours and found a local control rate of appr. 80% at 1 year [40]. For retroperitoneal sarcoma a local control rate of appr. 50% at 1 year was reported for recurrent tumours [41]. Daigeler et al. specifically reported on 135 patients with recurrent soft tissue sarcoma. Yet, they focus on overall survival and do not report on local control after diagnosis of recurrence. In addition, the analysed population consisted of local recurrences with and without simultaneous metastatic spread [42]. In our cohort 1.5 year local control rate for patients treated with local recurrences of soft tissue sarcomas was 53%. The question arises, whether the striking difference in the treatment of primary and recurrent tumours is due to difficulties in the diagnosis and treatment of recurrent tumours or rather reflecting a more aggressive biology of recurrent tumours. On the one hand radiological diagnosis and treatment, e.g. planning of surgery and radiotherapy, are more challenging in pre-treated tissue. On the other hand, tumours recurring after aggressive resection might differ biologically from tumours that can be cured after primary diagnosis by surgery alone [43].

In conclusion, multimodal treatment with radiotherapy and hyperthermia, in most cases also in combination with surgery and concurrent ifosfamide, for high risk soft tissue sarcomas might be able to alleviate the negative prognostic value of microscopically positive resection margins. In addition, local control rate in recurrent disease after surgery is significantly lower than in primary disease. Despite the insecurities in the interpretation of the data concerning the reason for this difference, we recommend to consider multimodal treatment at primary diagnosis generously.

Acknowledgements

This study was partly funded by the “EKFS-Forschungskolleg Tübingen” (2015_Kolleg.14) of the Else-Kröner-Fresenius foundation.

Disclosure statement

The authors report no conflicts of interest.