Feasibility and long-term outcomes of surgery for primary thoracic synovial sarcoma.

OBJECTIVES: Primary thoracic synovial sarcoma (SS) is a rare, high-grade, malignancy. Involvement of vital organs is frequent and may decrease the benefits of surgical resection. We reviewed our practice at a highly experienced thoracic-surgery centre to assess early- and long-term outcomes after surgery.
METHODS: We conducted a retrospective, observational, single-centre study of patients undergoing curative-intent surgery for primary thoracic SS between 1 January 2000 and 31 January 2021 as part of a multidisciplinary management. We assessed demographics, medical history, histopathology and follow-up information.
RESULTS: We enrolled 20 patients (13 males) with a median age of 40 years old and a median tumour size of 11 cm. Neoadjuvant chemotherapy was administered to 13 patients. Surgery consisted in extrapleural pneumonectomy (n = 7), extrapleural lobectomy (n = 5), chest wall resection (n = 4) or tumour resection (n = 4). R0 resection was achieved in 16 (80%) patients. Adjuvant therapy was given to 13 patients. 6 patients developed postoperative complications. The median hospital stay was 11.5 days. Overall survival at 2 and 5 years was 51% and 22%, respectively; median overall survival was 25 months and median disease-free survival was 8.5 months. Relapses occurred in 15 patients. By univariate analysis, incomplete resection was the only significant predictor of survival (P = 0.01).
CONCLUSIONS: Primary thoracic SS is an aggressive disease. Surgery included in a multimodal treatment may contribute to achieving a good outcome, providing that an R0 resection is obtained. Given the considerable technical challenges of surgery, patient selection and referral to an experienced centre are crucial to minimize morbidity and mortality.

INTRODUCTION

Synovial sarcoma (SS) accounts for only 5–10% of all soft tissue sarcomas [1] and usually develops in the extremities. SS is a high-grade malignancy classified as a mesenchymal spindle-cell tumour of uncertain differentiation [2] and is characterized in over 90% of the cases by the t(X; 18) (p11; q11) translocation, which is pathognomonic [3, 4]. Primary thoracic SS (PTSS) is extremely rare, contributing to 4–14% of all SSs, and mostly affects adults in the third and fourth decades [5].

The treatment of SS is frequently multimodal. Neoadjuvant chemotherapy is recommended for non-resectable disease, while adjuvant chemotherapy may be administered to high-risk patients [6, 7]. Surgical resection should be considered in all patients with localized SS [8]. However, because of the potential involvement of vital organs and need for extensive resection, surgery can be very demanding and may be followed by major complications [9]. If resection is incomplete, adjuvant radiotherapy should be given to improve local control [8].

Five-year overall survival (OS) of patients with SS arising from any site ranges from 52% to 66% [10]. However, location within the thorax is associated with a far poorer prognosis, with only 30% of patients being alive 5 years after surgery [3, 11].

As PTSS is rare, experience on treatment, notably surgery, is limited. Only case reports and small cohort studies are available [9, 12], and data on early- and long-term outcomes are particularly scarce.

The primary aim of our study was to evaluate OS in patients undergoing curative-intent surgery for PTSS. The secondary aims were to assess the feasibility of surgery based on the hospital stay duration, postoperative complications and 30-day mortality and to determine disease-free survival (DFS).

PATIENTS AND METHODS

We conducted a retrospective, single-centre study of consecutive patients who underwent curative-intent surgery for PTSS between 1 January 2000 and 31 January 2021 at our thoracic surgery department. Patients were identified by searching the histopathology laboratory database. We excluded patients with thoracic metastasis of a primary SS of another site and patients who underwent only diagnostic procedures or first surgical treatment at another institution.

We collected demographics, preoperative data (risk factors, comorbidities, clinical and radiological presentation, preoperative histological diagnosis, neoadjuvant therapy and functional preoperative evaluation), details on the surgical procedure, postoperative data [intensive care unit (ICU) admission, mechanical ventilation duration, postoperative complications classified according to Clavien–Dindo [13] and hospital stay length] and follow-up data (relapses and subsequent treatments, vital status at last follow-up and cause of death).

As our centre is exclusively surgical, follow-up was not performed in our hospital. For the purpose of the study, we collected information on follow-up by a combination of contact with patients or primary physicians and oncologists and data linkage to a national death register. We did not have any missing data.

R software version 4.1.0 was used for the statistical analysis (http://www.cran.r-project.org/doc/FAQ/R-FAQ.html#Citing-R). Data were described as the median [interquartile range] or absolute frequency (percentage). Kaplan–Meier plots were used to describe OS and DFS. Cox regression was chosen for the univariate analysis to identify factors associated with outcomes. Comparing the results with the log-rank test, we did not find any difference between the 2 methods. As the sample size is small, we did not perform a competing risk analysis nor a multivariable analysis as they would not have a statistical power. P-Values smaller than 0.05 were considered statistically significant.

The study protocol was approved by the ethics committee of the French society for cardiovascular and thoracic surgery (Société Française de Chirurgie Thoracique et Cardio-Vasculaire, CERC-SFCTCV-2021-02-22). In keeping with French law about retrospective studies of anonymized patient data, informed consent was not required.

RESULTS

Through the histopathology database, we identified 65 patients. Twenty patients met our selection criteria (Fig. 1). In 17 cases, the diagnosis was confirmed by identification of the t(X; 18) translocation. In the remaining 3 patients, tissue alterations by the fixative precluded fluorescence in situ hybridization, but the diagnosis was confirmed by retrospective review of the slides by expert histopathologists.

Figure 1:

Patients’ selection flow chart.

13 patients were males and 7 females, the median age was 40 [28-54] years old. 16 patients had no history of major comorbidities at the diagnosis, one patient had atrial fibrillation, 1 chronic obstructive pulmonary disease, 1 diabetes and 1 a history of breast cancer treated by radiotherapy on the same side of the PTSS. Nine (45%) patients had a history of smoking.

The tumour was symptomatic in 18 patients, causing chest pain (n = 11), dyspnoea (n = 5), haemoptysis (n = 3), cough (n = 3), dysphagia (n = 2), thoracic bulge (n = 1) and/or arm swelling (n = 1). In 2 cases, the diagnosis was incidental.

The initial tumour was pleuropulmonary in 11 patients (55%), some of which had a chest wall infiltration. Other locations were the mediastinum (n = 4), chest wall (n = 4) and trachea (n = 1). The median tumour diameter was 11 [8-15] cm.

Preoperative diagnosis was obtained in 18 patients, by computed tomography (CT)-guided percutaneous biopsy (n = 10) or surgical biopsy (n = 8). The histopathological diagnosis was SS in 12 patients, sarcoma other than SS in 4 patients and undetermined malignancy in 2 patients.

Neoadjuvant treatment, consisting of 4–6 cycles of chemotherapy, was given to 13 patients. Surgery consisted of extrapleural pneumonectomy (EPP) (n = 7), extrapleural lobectomy (n = 5), chest wall resection (n = 4) and tumour-mass resection (n = 4). Surgery was performed by excising the tumour en bloc with all the structures and/or tissues macroscopically infiltrated. For chest wall resections, a macroscopical margin of at least 3 cm was considered safe for complete resection. If a preoperative biopsy was performed, the biopsy tract and the scar were safely removed.

All patients underwent surgery by an open approach. EPP, extrapleural lobectomies and posterior mediastinum resections were performed by a posterolateral thoracotomy. EPP was systematically associated with a diaphragm resection and reconstruction with a Goretex mesh as well as a pericardium resection and reconstruction with a Vycril mesh (Fig. 2). In the extrapleural lobectomy, as the tumour was adherent to the chest wall, the parietal pleura was excised en bloc.

Figure 2:

Example of a patient with a pulmonary and pleural tumour. (A and B) Twenty-six-year-old female with a 10 cm × 12 cm × 12 cm right intrathoracic mass, pleural lesions and a pleural effusion. (C) She had a partial response to 5 chemotherapy cycles then underwent a right EPP. The postoperative course was uneventful. (D) Operative specimen: histology showed an R0 resection of a biphasic synovial sarcoma that measured 10.5 cm × 7.5 cm × 3 cm and partially infiltrated the diaphragm; the tumour contained a necrotic haemorrhagic component, cystic areas and microcalcifications.

An anterior transclavicular approach was preferred for tumours involving the apex, as it allowed optimal control of the vascular and nervous structures of the thoracic inlet. Because of the number of resected ribs (maximum 4 ribs) and the location of the chest wall defect, chest wall resections did not need reconstruction. The only case requiring osteosynthesis was a sternal resection. Reconstruction was made with 3 Titanium bars and a Goretex mesh.

Histopathological examination showed a complete resection (R0) in 16 patients. R1 was found in 3 patients:

In a right EPP with chest wall resection, R1 was detected on a chest wall nodule and on the superior pulmonary vein.

In a right EPP, the resection appeared to be R1 on the pulmonary artery.

In the resection of a posterior mediastinal tumour, R1 was found on the aortic and the oesophageal adventitia.

A macroscopic incomplete resection (R2) was found in a patient who underwent a resection of a posterior mediastinal mass. R2 was confirmed on the specimen in multiple intra-pericardiac zones.

Seven patients required ICU admission with a median ICU stay of 4 days. Only 1 patient required postoperative mechanical ventilation. The median hospital stay was 11.5 [8-14] days.

Postoperative complications occurred in 6 patients, some of whom experienced >1 complication. They are summarized in Table 1. No patients died within 30 days after surgery.

Table 1:

Postoperative complications, according to the Clavien–Dindo classification

Grade of complication	Number of events and type of complication	Treatment
I	0	–
II	Pleural or pericardial bleeding in 2 patients Acute pulmonary ooedema and gastric ulcer in 1 patient	Blood cell transfusion Medical treatment
IIIa IIIb	Delayed pneumothorax in 1 patient, prolonged air leak in 1 patient Bronchial obstruction by secretions in 1 patient Haemothorax in 1 patient	Chest drainage Bronchoscopy Surgery
IV	0	–
V	0	–

Adjuvant treatment was given to 13 patients: 11 received radiotherapy and 2 received chemotherapy. Globally, 11 patients received both neoadjuvant chemotherapy and adjuvant radiotherapy, 2 only neoadjuvant chemotherapy, 2 only adjuvant chemotherapy and 5 only surgical treatment.

A local recurrence was diagnosed in 15 patients, of whom only 1 also had a distant metastasis (liver). Recurrences were treated by chemotherapy in 7 patients, and pazopanib in 2 patients. Three patients received palliative care. Surgery was performed in 2 patients. Unfortunately, in both cases, the resection was incomplete (R1). One of the patients received second-line chemotherapy and died 1 year after the second surgery, and the other received chemotherapy and radiotherapy and died 2.5 years after the second surgery.

Two- and five-year OS was 51% and 22%, respectively (Fig. 3). The median OS was 25 months. Two-year DFS was 24% (Fig. 4), with a median of 8.5 months. All patients presenting a recurrence died from the disease. In the 16 patients with R0 resection margins, 5-year OS and DFS were both 29%. Five patients were still alive at the time of the last follow-up with no evidence of disease. Three of them had long follow-ups of 10, 11 and 18 years, with a median follow-up of 127 months.

Figure 3:

Kaplan–Meier estimate of overall survival after the diagnosis of primary thoracic synovial sarcoma.

Figure 4:

Kaplan–Meier estimate of disease-free survival after the diagnosis of primary thoracic synovial sarcoma.

By the univariate analysis, incomplete resection was the only variable significantly associated with lower OS (P = 0.01) (Fig. 5). Age, tumour size, additional treatment (either neoadjuvant or adjuvant or both) and type of surgery did not appear to influence survival (Table 2).

Figure 5:

Kaplan–Meier estimate of overall survival after the diagnosis in the groups with tumour-free resection margins (R0) versus contaminated resection margins (R+).

Table 2:

Univariate analysis to identify factors associated with overall survival

Factors	N (%)	Hazard ratio	95% CI	P-Value
Age	Not applicable	1.02	0.99–1.06	0.07
Tumour size
<11 cm	10 (50)	Ref
>11 cm	10 (50)	0.44	0.28–2.68	0.16
Neoadjuvant treatment
No	7 (35)	Ref
Yes	13 (65)	0.86	0.39–0.56	0.80
Adjuvant treatment
No	7 (35)	Ref
Yes	13 (65)	1.17	0.39–3.56	0.77
Type of surgery
Other than EPP	13 (65)	Ref
EPP	7 (35)	1.53	0.47–4.92	0.48
Resection margins
R0	16 (80)	Ref
R1 and R2	4 (20)	0.14	0.03–0.62	0.01

CI: confidence interval; EPP: extrapleural pneumonectomy; Ref: reference value.

DISCUSSION

PTSS is a very rare malignancy, and surgically resectable patients represent an even smaller subgroup of PTSS patients. In the current study, we found that in highly selected patients, surgery included in a multimodal strategy might contribute to achieve a good survival. The main determinant to obtain this result was complete resection without perioperative mortality, underlining the importance of performing these procedures in expert thoracic surgical centres.

The median age of 40 years old in our population is consistent with earlier reports, while the predominance of males in our study contrasts with the usually more even sex distribution in previous works [14, 15]. Also, in keeping with published data, the symptoms were not specific and consisted chiefly of chest pain, dyspnoea, cough or haemoptysis [16, 17]. PTSS is typically diagnosed as a large mass, with a median diameter of 11 cm in our cohort and up to 13 cm in earlier studies [18].

As indicated in the ESMO-EURACAN guidelines, clinical staging should be completed by a CT scan of the abdomen and pelvis. Imaging of the brain is not standard for SS and should be performed based on the clinical presentation. Fluorodeoxyglucose positron emission tomography scan may be reserved for characterizing equivocal CT findings such as lymph node involvement [8]. In our series, the clinical staging was achieved by performing either a fluorodeoxyglucose positron emission tomography scan or a total body CT scan. Further examinations, such as a flexible bronchoscopy, a thoracic magnetic resonance imaging or a gastroscopy were performed in specific cases.

Preoperative histopathological diagnosis should be sought routinely to guide treatment decisions [19]. In our cohort, the presence of a malignancy was established in 90% of patients, but the diagnosis of SS was made in only 67%, illustrating the challenges raised by the differential diagnosis. Difficult cases must therefore be referred to specialized centres to ensure correct diagnosis. In our study, three cases needed confirmation of the diagnosis by our expert pathologists, who belong to the French soft tissue and visceral sarcoma pathology review network (RRePS).

The variability in treatment strategies in our cohort reflects the absence of consensus regarding the optimal management of PTSS. Induction chemotherapy was proposed when the tumour was not eligible for R0 resection, with the aim of obtaining a tumour shrinkage. However, the topic of neoadjuvant chemotherapy in soft tissue sarcoma remains controversial. Several ongoing trials are trying to address this question (NCT04307277 and NCT03805022).

Adjuvant chemotherapy was considered in patients aged <40 years and able to tolerate doxorubicin and/or ifosfamide and those with R1 resection [20].

In our series, 5 patients had a resectable disease at diagnosis and did not have the criteria for adjuvant therapy; therefore, they received an exclusive surgical treatment. It is important to underline that every decision of the treatment process was taken through a multidisciplinary team (MDT) dedicated to sarcoma and discussed with the patients.

The importance of MDT management has been widely established, notably for rare malignancies [21]. Recently, He et al. [11] published a study on the impact of MDT management for PTSS. Although MDT discussion was not independently associated with outcomes, the median OS of patients managed by an MDT was longer than that of patients who were not (46.0 vs 18.0 months). At our institution, the MDT included experts from national referral centres for sarcoma and thoracic surgery.

Surgery should be considered part of the multimodal treatment of PTSS. However, due to the aggressiveness of PTSS and possible infiltration of vital organs, surgery is often very demanding. Morbidity and mortality in nowadays oncological thoracic surgery have been widely investigated. The 30-day mortality rates and the postoperative complication rates for pneumonectomy range from 4% to 7% and 20% to 40%, respectively [22]. Turbendian et al. [23] described a morbidity of 37% and a mortality of 6% in patients undergoing surgery for mediastinal sarcomas. The only previous study detailing postoperative complications for PTSS had 15 patients, of whom 1 (6.6%) died postoperatively, 3 had major complications requiring reoperation and 5 had minor complications. [9]. The lower complication rates in our cohort may be explained by the high selection of patients and the expertise of the surgical team.

Referral of patients to tertiary-care centres with high thoracic-surgery volumes and considerable experience in performing complex procedures is crucial to minimize morbidity and mortality. An extensive body of literature has emphasized the impact of institutional and surgeon experience on outcomes. However, studies about lung resection surgery have shown conflicting results. An analysis of a large database found that lung cancer resection volume did not predict mortality [24]. Nevertheless, when analyzing more complex procedures such as pneumonectomies, surgery performed in high-volume centres seems to be associated with reduced odds of mortality and lower rates of failure to rescue after postoperative complications [25].

The poor prognosis of PTSS is well documented. A literature review of 15 clinical trials on chemotherapy as first-line treatment for locally advanced and metastatic SS showed an OS of 15 months and a progression-free survival of 6.3 months [26]. In our cohort, the median OS was 25 months, and OS at 2 and 5 years was 51% and 22%, respectively. Comparisons with earlier studies are difficult due to the variability in data-reporting methods, incomplete follow-up information and probable differences in patient selection for surgery [27, 28]. Table 3, summarizing previously reported long-term outcomes, shows similar OS to ours in some studies [9, 29] and better OS in others [3, 18]. The high frequency of recurrence shortly after surgery in our cohort is consistent with earlier data [30].

Table 3:

Cohort studies on the treatment and long-term outcomes of primary thoracic synovial sarcoma

Author, year of publication	Sample size	Treatment	R0	FU available information (% of patients)	OS	DFS and information on recurrence
Zeren et al., 1995	25	100% surgery	NR	72%	40% died of disease within 1–7 years 16% died of unrelated causes	16% alive with disease, 16% alive without disease after 2–20 years
Gartner et al., 1996	5	100% surgery	NR	100%	80% died within 3 years	NR
Aubry et al., 2001	5	100% surgery	100%	80%	100% (median FU 9 months)	100%
Essary et al., 2001	12	8% NR 92% surgery	NR	100%	2.5-Year OS 58%	2-Year DFS 25%
Duran-Mendicuti et al., 2003	5	100% surgery	100%	100%	Median OS 22 months	80% recurrence at 2–14 months
Okamoto et al., 2004	11	10% NR 90% surgery	NR	91%	50% died of disease within 1–9 years	60% recurrence
Begueret et al., 2005	40	10% NR 82% surgery 8% CT ± RT	NR	83%	2-Year DSS 65.3% 5-Year DSS 31.6%	Median DFS 43 months
Suster et al., 2005	15	80% surgery 20% RT	83%	40%	NR	80% had recurrence 1–3 years after diagnosis
Hartel et al., 2007	60	8% NR 68% surgery 23% other treatment	NR	90% 63% information on recurrence	48% died at a mean of 23 months 46% died within 5 years	18% had local recurrence Mean DFS 17 months
Galetta et al., 2013	15	100% surgery	60%	100%	Median OS 27 months 10-Year OS 33.5%	75% recurrence 5-Year DFS 30%, median 15 months
Kim et al., 2015	14	7% NR 64% surgery 28% CT ± RT	NR	93%	21.4% died	57% recurrence 2-Year DFS 35.7%
Lan et al., 2016	26	77% surgery 12% CT 13% supportive care	NR	73%	Median DSS 14.5 months 2-Year DSS 27.7%	Median DFS in surgical patients 8.5 months
Terra et al., 2018	21	52% NR 43% surgery 5% CT + RT	NR	74%	69% died of disease within 5–32 months 24% alive with disease at 6–45 months	88% recurrence
He et al., 2021	13	77% surgery 23% CT ± RT	NR	100%	2-Year OS 58,3% 5-Year OS 30%	Median DFS 13 months
Present study	20	100% surgery	80%	100%	Median OS 25 months 2-Year OS 51% 5-Year OS 22%	Median DFS 8.5 months 2-Year OS 24%

CT: chemotherapy; DFS: disease-free survival; DSS: disease-specific survival; FU: follow-up; NR: not reported; OS: overall survival; R0: complete resection; RT: radiotherapy.

Nonetheless, long survivals have been achieved. In our series, 3 patients were alive 10, 11 and 18 years after initial surgery. Similarly, in his series of 25 patients, Zeren et al. [17] reported 3 long survivors, with an OS between 12 and 16 years.

However, no factors associated with long-term survival have been identified [4, 5, 18]. Age, sex, tumour size, preoperative and postoperative treatments, histological subtype, SSX-SS18 fusion type, Ki-67 expression and mitotic rate were prognostic factors for survival in some studies but not in others. In our study, the only factor significantly associated with OS by univariate analysis was incomplete resection, which was also consistently significant in other studies [9, 14]. More specifically, when analyzing the R0 patients’ subgroup, we found the same 5-year OS and DFS of 29%, which supports the conclusion that we should strive for an R0 resection, even if it can be very demanding, as it is the only factor that seems to make a difference in the outcome of these patients.

Limitations

A major limitation of our study is the retrospective design and the absence of a control group. Moreover, the high patients’ selection and the single-centre recruitment may have induced selection bias. As our institution is a purely surgical centre and does not have an oncological department, we did not have a group of patients undergoing medical treatment or palliative care to compare to our series.

Another major limitation is represented by the small sample size, which mainly depends on the rarity of this disease, thus precluding a multivariable analysis and limiting difference detection and results’ interpretation. Randomized trials comparing surgery to other treatments would be unethical, and multicentre observational studies in larger populations are therefore needed to provide more reliable results for the diagnosis and treatment of PTSS. Molecular targeted therapies and immunological strategies are under investigation, and trials evaluating genomic features as potential predictive markers are ongoing and may improve outcomes in the near future.

CONCLUSION

In conclusion, in patients with resectable PTSS, surgery as part of a multimodal strategy may contribute to achieve a good survival, provided R0 resection is obtained. Given the considerable technical challenges raised by complete PTSS resection, it is of utmost importance that patients are carefully selected and treated in experienced tertiary referral centres to minimize morbidity and mortality.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Conflict of interest: none declared.