Comparing children and adults with synovial sarcoma in the Surveillance, Epidemiology, and End Results program, 1983 to 2005

An analysis of 1268 patients




Synovial sarcoma (SS) is a typical soft tissue sarcoma subtype crosswise between the pediatric and the adult age groups. Less satisfactory overall outcome has been recorded in adult series.


This study compares clinical features and outcomes of SS across the different age groups, by analyzing 1268 cases, 213 children/adolescents (≤18 years) and 1055 adults, registered in the Surveillance, Epidemiology, and End Results (SEER) 17 database from 1983 to 2005. Cancer-specific survival estimates were compared with univariate and multivariate models.


No major differences in stage distribution (localized, regional, and distant stage) were observed comparing the 2 age groups. The estimated 5-year cancer-specific survival was 83% for children/adolescents and 62% for adults (P < .001). Female sex, nonblack race, tumors located in the extremities, localized tumors, and tumors <5 cm in size were associated with better survival. In multivariate analysis, adult patients had significantly higher mortality rates than children after adjusting for other variables.


Children and adults with SS have a similar clinical presentation but a dissimilar outcome, suggesting that factors other than unfavorable clinical features might be involved in the unsatisfactory outcome of adult SS patients. It remains to be ascertained whether this difference is related to biological variables or to historically different treatment approaches adopted in pediatric versus adult patients. Cancer 2009. © 2009 American Cancer Society.

Soft tissue sarcomas (STS) are rare tumors; their annual incidence is 2-3 cases per 100,000 people, accounting for less than 1% of all malignancies, and about 8% of all pediatric cancers.1 The pattern of STS histotypes differs significantly between adults and children, an age group where rhabdomyosarcoma represents more than half of the cases.

Synovial sarcoma (SS) is a typical STS subtype that crosses the 2 age groups: its peak incidence is in the third decade, approximately 30% of cases occur in patients less than twenty years of age, and it has been reported as the most frequent nonrhabdomyosarcomatous STS in adolescents and young adults.2 It is a clinically, morphologically, and genetically distinct sarcoma, characterized by the specific chromosomal translocation t(X;18), which fuses the SYT gene from chromosome 18 with the SSX1 (about 2 of 3 of cases), SSX2 (about 1 of 3 of cases), or SSX4 (rare cases) genes from the X chromosome.3, 4 Classified among sarcomas of uncertain differentiation, SS may be subdivided into biphasic, monophasic, and poorly differentiated subtypes, the histological features being identical in children and adults. Although it can be graded according to the mitotic index, percentage of necrosis and tumor differentiation, SS should always be considered as a high-grade sarcoma, characterized by local invasiveness and a propensity to metastasize.5, 6

To our knowledge, no published data describe a different biology of SS when arising in adults as opposed to children,7 and the question of possible differences in behavior of SS in adults versus children is somewhat open.8 This is not a useless debate because up to recent times, completely different therapeutic strategies have been developed for pediatric and adult oncology protocols dealing with SS, in particular concerning the use of systemic therapy.9 Also, different overall outcomes have been reported by pediatric and adult groups.9

Despite these differences in treatment approaches, few studies have directly investigated this issue and compared adult with pediatric SS patients.9 Thus, it remains unclear whether the less satisfactory overall results documented in the adult series are due to differences in clinical prognostic features, tumor biology, or treatment strategies adopted.9

To better characterize the clinical features and outcomes of SS across the different age groups, we performed an analysis of all SS cases registered on the Surveillance, Epidemiology, and End Results (SEER) public-access database collected from various geographic areas in the United States.10


Data Source and Study Population

The data were obtained from the SEER 17 registries ( The SEER Program of the National Cancer Institute (NCI) is an authoritative source of information on cancer incidence and survival in the United States. SEER currently covers approximately 26% of the US population. The SEER open-access database has records of patients diagnosed since 1973 in the states of Connecticut, Iowa, New Mexico, Utah, and Hawaii and the metropolitan areas of Detroit and San Francisco-Oakland. Many regions joined the program in the subsequent years. In 1992, the SEER program was expanded to increase coverage of minority populations, especially Hispanics, by adding Los Angeles County and 4 counties in the San Jose-Monterey area south of San Francisco and in 2001, it added New Jersey, Louisiana, Kentucky, and the remaining counties in California (Greater California). Law in these regions mandated case reporting at different times as they joined the program. The database currently includes 5,306,606 tumors diagnosed from 1973-2005.10

We used Case listing session of the SEER*Stat 6.4.4 program to generate a matrix of all individuals reported with a diagnosis of SS from January 1983 to December 2005. A selection query was designed to retrieve tumors based on the International Classification of Diseases for Oncology, version 3 (ICD-O-3) morphology: 9040-3: Synovial sarcoma, not otherwise specified (NOS), 9041-3: Synovial sarcoma, spindle cell, 9042-3: Synovial sarcoma, epithelioid cell, and 9043-3: Synovial sarcoma, biphasic. Patients with autopsy/death certificate only and patients with no microscopic confirmation of diagnosis were excluded. Also, patients with a previous recorded malignancy were excluded.

Data Analysis

The resulting matrix from SEER*Stat was transferred to MedCalc for Windows, version (MedCalc Software, Mariakerke, Belgium) to perform statistical calculations and generate survival curves. Primary site was relabeled using 5 categories. The SEER staging system is as follows: localized stage refers to an invasive neoplasm confined entirely to the organ of origin; regional stage refers to a neoplasm that has extended beyond the limits of the organ of origin directly into surrounding organs or tissues, has spread to regional lymph nodes by way of the lymphatic system, or both; and distant stage refers to a neoplasm that has spread to parts of the body remote from the primary tumor either by direct extension or by discontinuous metastasis to distant organs and tissues or via the lymphatic system to distant lymph nodes. All patients were also labeled according to initial tumor size (<5 cm vs ≥5 cm diameter).

According to the age at diagnosis, data were compared with “pediatric” cases (between 0 and 18 years of age) and adults (≥19 years). Pediatric cases were divided into children (0-9 years old) and adolescents (10-18 years old). Age cutoffs were chosen arbitrarily: the definition of pediatric patients (ie, ≤18 years old) was based on the assumption that most pediatric centers in the US treated patients within this age group, although this may not be true for all registered patients. Adult cases were similarly divided into 4 categories: 19-29, 30-39, 40-49, and ≥50 years old.

Cancer-specific mortality was chosen as an endpoint. Survival was calculated using the Kaplan-Meier method and the log-rank test was used to compare survival curves. Throughout the analysis, survival estimates are followed by standard errors. The chi-square test was used to compare sex, race, histology, primary site, stage, size, and status. The Cox proportional-hazards regression was used to calculate the hazard ratios (HR) and 95% confidence intervals (95% CI). The Mann-Whitney test was used to compare follow-up intervals. The rate session of the SEER*Stat program was used to calculate the rate and trend using SEER9 database and 2000 US standard population.


Patient Characteristics

In the SEER17 database, 1343 records with a reported diagnosis of SS were registered within the study period. Four patients were excluded because of the absence of histologic confirmation and a patient with autopsy/death certificate only follow-up was excluded. An additional 69 patients (1 child) were excluded because SS was not the first recorded malignancy. Among the 1268 patients we studied, the median age was 34 years (range, 2 to 94) and 17% were children/adolescents (Table 1, Fig. 1).

Figure 1.

(A) Presentation of 1268 patients with synovial sarcoma, diagnosed from January 1983 to December 2005, with bar histogram showing distribution of cases according to age at diagnosis. (B) Bar histogram showing 47,433 patients with soft tissue sarcoma registered on SEER 9 (1973 to 2005 excluding Kaposi Sarcoma, 3.6% are synovial sarcoma). The table shows * the synovial to soft tissue sarcoma ratio and ** the incidence rates of synovial sarcoma per million per year, age-adjusted to the 2000 US Standard Population (Census P25-1130).

Table 1. Patient Characteristics, Local Control, and Outcome: Data on 1268 Patients Diagnosed From 1983 to 2005
  • NOS indicates not otherwise specified; EBRT, external beam radiotherapy.

  • *

    Other locations include retroperitoneum, kidney, adrenal gland, liver, ileum, thymus, heart, unknown, and NOS.

Total1268 213 1055  
Age, y       
 0 to 932(2.5)32(15)   
 10 to 18181(14)181(85)   
 19 to 29292(23)  292(28) 
 30 to 39269(21)  269(26) 
 40 to 49208(16)  208(20) 
 ≥50286(23)  286(27) 
 Spindle cell342(26)50(24)292(28) 
 Epithelioid cell11(.9)0(.0)11(1.0) 
 Head and neck91(7)12(5.6)79(7.5).027
 Lungs and pleura52(4.1)2(.9)50(4.7) 
SEER stage       
 <5 cm358(28)75(35)283(27).068
 ≥5 cm592(47)95(45)497(47) 
Local therapy       
 Surgery performed1142(90)209(98)933(88)<.001
 EBRT given632(50)101(47)531(50).41
 No Local therapy73(5.8)2(.9)71(6.7).0016
 Cancer-related mortality394(31)35(16)359(34)<.001

The age-adjusted incidence rate was .81 per million in children (95% CI, .68 to .97) and 1.42 per million in adults (95% CI, 1.31 to 1.54). There was no significant change in incidence trend from 1983 to 2005 in both groups.

Sex and race were similarly distributed in children and adults. Half of registered tumors had no histologic subtype assigned and were, therefore, defined as NOS. This strongly limited any comparative analysis on histology; however, there were more biphasic tumors in children (35%) than adults (19%). As for tumor location, extremities were the most common site of origin (70% of cases). Adults had more tumors in the lung and pleura (4.7% in adults vs .9% in children). A small number of tumors were located in unusual sites, ie, retroperitoneum (n = 8), kidney (n = 6), adrenal gland (n = 1), heart (n = 7), thymus (n = 1), liver (n = 2), ileum (n = 1), and spinal cord (n = 1).

More than half (57%) of tumors were “localized”, while 23% had regional and 13% had distant SEER stage. No major differences in stage distribution were observed comparing the 2 age groups, though a quite different pattern was recorded just for patients younger than 10 years (Fig. 2). Tumor size was available for 950 patients, with approximately two third of cases larger than 5 cm in maximum diameter.

Figure 2.

Frequency charts showing relative frequencies of patients with different clinical characteristics distributed according to age categories. Patients were compared according to (A) primary tumor location, (B) histologic subtype, (C) SEER stage, and (D) primary tumor size.

The SEER database provides limited data on given treatment modalities: data on local therapy were partially available (but for instance the status of surgical margins after resection was not reported), while there were no data regarding systemic treatment.

Nevertheless, the majority of patients (90%) had primary site-directed surgery, which was done more frequently in children. External beam radiotherapy was delivered in half of the patients. Only a small percentage of patients (5.8%) did not receive any local treatment, and most of them were adults (71 of 73).

Survival and Prognostic Factors

The 5- and 10-year cancer-specific survival estimates for the whole group (n = 1268) were 66% ± 1.6% and 56% ± 1.9%, respectively (median follow-up 4.1 years). During the last decade of the studied period (1996 to 2005), the outcome improved in comparison with the previous period (1983-1995; HR for the more recent period, .79; 95% CI .64 to .99). Males (P = .012) and black patients (P = .009) had worse survival when compared with others (Fig. 3). There was no difference in outcome among different histologic subtypes, with the exception of the rare monophasic epithelioid-cell subtype (n = 11, 5 year cancer-specific survival 27% ± 15%, P = .032). When tumors were classified according to primary site, patients with tumors located in the extremities fared best, whereas the worst survival was seen for tumors located in the trunk or in other sites (including retroperitoneum and NOS cases) (P < .001).

Figure 3.

Kaplan-Meier survival curves (endpoint = cancer-specific mortality) of patients with synovial sarcoma compared according to (A) sex, (B) race, (C) histologic subtype, (D) primary tumor site, (E) SEER stage, and (F) tumor size. Log-rank test was used to compare curves, and significance (p) is shown on each panel. Curves of patients with missing data (eg, unstaged, tumor size unknown) were not shown.

Regional and distant spread were associated with significantly worse outcome (Fig. 3): the 10-year cancer-specific survival for localized, regional and, distant stages were 69% ± 2.4%, 43% ± 4.7% and 8.9% ± 3.1%, respectively (P < .001). Also, the size of tumor at diagnosis affected survival significantly (P < .001).

Comparing Different Age Groups

The estimated 5- and 10-year cancer-specific survival for children/adolescents were 83% ± 3.1% and 75% ± 4.2%, respectively. The corresponding estimates for adults were 62% ± 1.8% and 52% ± 2.1%, respectively (HR for adults vs children/adolescents = 2.52, 95% CI = 1.56 to 2.56; P < .001). The Kaplan Meier survival curves of different age groups are shown in Figure 4.

Figure 4.

(A) Kaplan-Meier survival curves of patients with synovial sarcoma stratified according to age groups. Log-rank test was used to compare survival estimates, followed by curves showing 10-year cancer-specific survival estimates, when available for patients compared according to (B) primary tumor location, (C) histologic subtype, (D) SEER stage, and (E) primary tumor size. Survival estimates in different age groups were tested using log-rank test to compare all curves shown in each panel; an asterisk (*) is used to indicate a significant difference (P < .05).

The survival rates also decreased with increasing age when data were analyzed according to tumor site (extremity vs trunk tumors), histology, stage (the survival of patients with distant metastases was dismal across all ages), and tumor size (the survival differences between smaller (<5 cm), and larger (≥5 cm) tumors significantly increased with increasing age (Figure 4).

Cox Proportional Hazards Regression

Multivariate analysis was performed using a Cox proportional hazards regression model. As shown in Table 2, adult patients had significantly higher mortality rates than children after adjusting for other variables. Middle-aged patients ranging from 40 to 49 years had the highest HR when compared with other age groups.

Table 2. Cox Proportional Hazards Regression for Cancer-Specific Mortality Conducted on 950 Patients (170 Children) Diagnosed From 1983 to 2005, With Tumor Size Available
VariableAll PatientsChildren/AdolescentsAdults
HR (95% CI)PHR (95% CI)PHR (95% CI)P
  1. HR indicates hazard ratio; CI, confidence interval; —, not enough numbers to conduct a multivariate analysis.

Age category, y      
  0 to 181.0Reference
 19 to 291.94 (1.20-3.14).0074
 30 to 392.22 (1.39-3.55)<.001
 40 to 493.36 (2.09-5.39)<.001
 ≥503.28 (2.06-5.21)<.001
 Male1.55 (1.22-1.98)<.001.74 (.33-1.63).451.70 (1.32-2.20)<.001
 Black1.65 (1.18-2.30)<0.001.41 (.054-3.07).391.80 (1.28-2.52)<.001
 Others.90 (.55-1.45).66.54 (.073-4.06).55.96 (0.58-1.59).88
Primary site      
 Head & neck1.27 (.76-2.10).364.20 (.87-20.30).0761.18 (.69-2.02).55
 Lung & pleura2.79 (1.78-4.39)<.001 2.67 (1.68-4.23)<.001
 Trunk1.26 (.92-1.73).15.85 (.25-2.95).801.33 (.96-1.85).091
 Others1.74 (.96-3.15).067 1.50 (.78-2.86).22
 Regional/distant2.70 (2.10-3.47)<.0011.78 (.79-4.03).172.87 (2.20-3.74)<.001
 <5 cm1.0Reference1.0Reference1.0Reference
 ≥5 cm3.21 (2.35-4.38)<.0012.58 (1.02-6.53).0483.26 (2.34-4.55)q<.001

When analyzed for all patients, male sex and black race remained important factors. The cancer-specific mortality rate was higher among patients with tumors in the lung and pleura but not in other sites when compared with tumors located in the extremities. Likewise, stage at diagnosis was a significant prognostic factor, and regional/distant stages were associated with a high HR (2.70). The small number of patients with distant stage in the model did not permit a separate analysis for regional and distant stage. Tumor size (≥5 cm) was associated with the highest HR (3.21).

We found it interesting that while similar findings were observed when the analysis was restricted to adults, quite different results were noted when the analysis was restricted to children/adolescents. In young patients, male sex and race had no prognostic value; tumor size was the only significant prognostic factor (P = .048). Small number of children tested in the model may have contributed to the difference.


This study has analyzed the largest series to date of patients with SS, and has compared the clinical characteristics and outcomes of 213 children/adolescents and 1055 adults treated contemporarily, showing that adults may have a similar clinical presentation but a worse outcome than pediatric patients. As for other retrospective population-based studies, this analysis comes with major limitations that make the interpretation of the results difficult: ie, the lack of pathological review (errors in diagnosis of STS are well-known to occur, particularly in nonspecialized centers29-31), and the inadequate data on administered treatment modalities, in particular, the lack of details on systemic therapy.

However, interesting findings emerged. First of all, we found that the most relevant clinical characteristics, notably the disease stage, were similar between children and adults. Remarkably, a somewhat different distribution (“better clinical features” including more extremity primaries, smaller tumors, mostly localized) was observed in the small group of patients younger than 10 years of age (Fig. 2), a group that also seemed to have the best outcome. Whether SS has unique clinicobiological findings in prepubertal patients, in which this tumor rarely occurs (2.5% of the cases in the SEER series), remains to be proven.

Pediatric patients were more likely to have “better clinical features”, including more extremity and smaller primaries, which might have contributed to better outcome in univariate models. Despite the similarity in presentation between age groups, cancer-specific mortality was higher in adults (34% vs 16%). This result is consistent with previously reported data; 5-year overall survival rates of around 80% are reported in the pediatric series,11-15 higher values than those usually seen for adults (in the range of 50%-60%).16-20 Importantly, increasing age as an adverse prognostic factor was also seen within the pediatric age group; young children (<10 years) had better outcomes than adolescents.

Our study also confirmed that sites other than limbs, regional and distant spread, and large tumor size are the main adverse prognostic factors. However, the outcome for adults remained consistently worse than for children when the analysis was controlled for the individual clinical characteristics, suggesting that factors other than an unfavorable clinical presentation might be involved in the unsatisfactory outcome of adult SS patients.

Although older age consistently correlates with a worse prognosis, age as an independent prognostic factor in SS remains to be tested as treatment variables may play a significant role. As the SEER database provides little information on treatment modalities, our study can only report the small differences in local control methods between children and adults: the majority of patients who did not have local treatment were adults (97%), a difference that may be explained by comorbidities likely to be more prevalent in the elderly. Unfortunately, our analysis cannot capture the different approaches that pediatric and adult oncologists have adopted over the years concerning the use of chemotherapy.9, 21 In fact, because relatively high response rates to chemotherapy have been well documented in pediatric series,11-15 pediatric oncologists approached SS as a chemosensitive tumor, and they designed treatments around this concept. This has been particularly the case in Europe, where children with SS have often been included in the rhabdomyosarcoma protocols (SS was defined a “rhabdomyosarcoma-like” tumor), thus always receiving systemic treatment, regardless of stage.21 By contrast, in common with other adult STSs, SSs were generally regarded as poorly chemosensitive tumors, and the standard therapeutic approach was focused on local control. In this context, the use of adjuvant chemotherapy was considered investigational.22, 23

On this matter, a retrospective analysis on 271 patients of all ages performed at the Istituto Nazionale Tumori (National Cancer Institute) in Milan reported better outcomes for children (particularly among patients with grossly resected disease); in that series, adjuvant chemotherapy was administered to most of the children and to a minority of the older patients, and a strong correlation of survival with age and use of chemotherapy was noted.9

Recently, adult oncologists have recognized characteristics that suggest SS should be managed differently from other adult STS, such as younger age and increased metastatic potential, and evidence that this sarcoma is a chemosensitive histiotype.24-28 In fact, SS chemosensitivity probably represents a midpoint in the continuum between the most typical adult STS and the pediatric small round cell tumors, such as rhabdomyosarcoma.

Despite this historical difference in treatment approaches between pediatric and adult oncologists, it is unclear whether the survival gap observed in the SEER data is related to the different use of chemotherapy. Although this might be the case, it is noteworthy that different age cutoffs are used to define pediatric and adult cases across centers in the United States, and we do not have any evidence to suggest that the SEER “pediatric” cohort was treated homogeneously. Moreover, the widest survival gap was observed between prepubertal and postpubertal patients, rather than across the 18 years age cutoff, suggesting a role of “age-dependent” differences in tumor biology in determining tumor behavior and patient outcome (although, in SS, such differences have not been documented).

In summary, our analysis highlighted a similar clinical presentation but a dissimilar outcome for children/adolescents and adults with SS, suggesting a prognostic role for variables other than the clinical factors. It remains to be ascertained whether this difference may be related to biological variables or to historically different treatment approaches adopted in pediatric versus adult patients.

The uniqueness of SS, a tumor that encompasses both the pediatric and the adult age groups, should expedite the development of cooperative trials that would integrate uniform treatment concepts regardless of age and provide adequate numbers to answer relevant questions in a randomized manner. Furthermore, such cooperative studies would facilitate to the procurement of tumor samples and further our understanding of the biology of this malignancy. Such collaboration may speed up the development of novel therapeutic approaches based on the identification of molecular targets.32-34


The authors acknowledge the King Hussein Cancer Foundation and the American Lebanese Syrian Associated Charities for their support to King Hussein Cancer Center and St. Jude Children's Research Hospital, respectively.

Conflict of Interest Disclosures

The authors made no disclosures.