Patterns of Chemotherapy Administration in High-Risk Soft Tissue Sarcoma and Impact on Overall Survival

View More View Less
  • 1 From the Departments of Medical Oncology and Biostatistics, Fox Chase Cancer Center, Philadelphia, Pennsylvania.

Background: Conflicting data exist on the benefit of chemotherapy in the management of high-risk soft tissue sarcoma (STS). Use of chemotherapy may be dependent on patient, tumor, and facility characteristics. This study sought to identify these factors and compare survival between treatment groups. Patients and Methods: Patients with stage III STS were identified from the National Cancer Data Base (NCDB) from 1998 to 2012. Multiple logistic regression analysis was used to determine factors that influenced the probability of receiving chemotherapy. In a subset of patients, we determined the relationship between chemotherapy use and overall survival, using Kaplan-Meier curves and Cox regression analysis with propensity score adjustment. We also examined the effect of chemotherapy by histologic subgroup using interaction models. Results: A total of 16,370 patients were included (N=5,377 for survival analysis). Patients who were younger than 40 years; male; privately insured; earned a higher income; had no comorbidities; had synovial sarcoma, angiosarcoma or “other” histology; and whose tumors were high-grade, greater than 10 cm, or from the lower extremity were significantly more likely to receive chemotherapy. Median unadjusted overall survival (OS) in the nonchemotherapy and chemotherapy groups was 51.3 and 82.7 months, respectively (P<.001). On adjusted analysis, the survival benefit remained significant (hazard ratio [HR], 0.85; P=.004). The benefit was particularly strong in the undifferentiated pleomorphic sarcoma (UPS) group on adjustment, with a median OS of 49.1 and 77.8 months for nonchemotherapy versus chemotherapy, respectively (HR, 0.77; P=.02). Conclusions: In addition to expected tumor and patient factors, histology, location of primary tumor, and socioeconomic status are associated with receipt/nonreceipt of chemotherapy in stage III STS. Chemotherapy use was associated with improved OS in the overall population, and specifically in the UPS subgroup.


Soft tissue sarcomas (STS) are uncommon tumors of the connective tissues. According to the current AJCC staging system, patients with stage III disease (large, deep, high-grade) have an expected overall survival (OS) of approximately 50% at 5 years.1 However, even within this group of patients, there is a wide range of outcomes, owing to disease heterogeneity. Although adequate surgical resection remains the mainstay of therapy for patients with localized disease, chemotherapy can be administered in the adjuvant setting with the goal of decreasing the risk of recurrence and improving OS. These goals must be balanced with the potential long-term toxicity.

In 1997, the Sarcoma Meta-Analysis Collaboration (SMAC) performed a meta-analysis of 14 randomized trials of doxorubicin-based adjuvant chemotherapy for STS from a variety of primary sites. There was a benefit for local and distant recurrence-free intervals and overall recurrence-free survival favoring the chemotherapy arm. However, there was no statistically significant benefit for OS in the total population. In a preplanned analysis limited to patients with extremity tumors, there was an absolute benefit in OS of 7% favoring the chemotherapy group.2 In 2007, a separate meta-analysis was conducted including 4 additional trials of combination therapy with doxorubicin and ifosfamide. The benefit of chemotherapy remained for local and distant recurrence and there was an absolute benefit in OS of 6% for the entire population. In the group of patients who received doxorubicin and ifosfamide, this benefit was 11%.3

Despite these results, patterns of care and opinions with regard to adjuvant chemotherapy vary. The meta-analyses have been criticized for the inclusion of patients with wide tumor heterogeneity. We used the National Cancer Data Base (NCDB) to identify factors associated with receipt/nonreceipt of chemotherapy specifically in patients with high-risk sarcomas. We also analyzed the effect of chemotherapy on OS in these patients. Finally, given the large size of our study population, we examined the effect of chemotherapy on OS for individual STS histologies.

Patients and Methods

Study Population

Individuals diagnosed with a stage III STS between January 1, 1998, and December 31, 2012, were identified from the NCDB. The NCDB, which is a joint program of the American College of Surgeons Commission on Cancer (CoC) and the American Cancer Society, is a nationwide oncology outcomes database for more than 1,500 Commission-accredited cancer programs in the United States and Puerto Rico. Approximately 70% of all newly diagnosed cancer cases in the United States are captured at the institutional level and reported to the NCDB.4 Cases are staged in the NCDB according to AJCC staging at time of diagnosis. Therefore, stage III STS in this analysis includes T2a/bN0M0, G3 or G4 or TanyN1M0, G1–3 (2.3% of cases).

From 1998 to 2012 there were 21,098 newly diagnosed cases of analytic stage III STS in the NCDB. We excluded cases in which receipt of chemotherapy was unknown (N=643; 3.0%) and without a confirmed surgical resection of the primary tumor (N=2,345; 11.1%). Also excluded were STS histologies with a limited role for chemotherapy (eg, alveolar soft part sarcoma) and those for which chemotherapy is standard of care or commonly used (eg, alveolar rhabdomyosarcoma) (N=1,542; 7.3%). Cases in which receipt of chemotherapy occurred more than 180 days after surgery (N=198; <1%) were also excluded.

Definition of Variables

Tumor Characteristics: Tumor histology was categorized as liposarcoma (myxoid/round cell, pleomorphic, dedifferentiated, and other), leiomyosarcoma, fibrosarcoma, undifferentiated pleomorphic sarcoma (UPS), synovial sarcoma, malignant peripheral nerve sheath tumor (MPNST), angiosarcoma, and other. The other histology category included spindle cell sarcoma, sarcoma not otherwise specified (NOS), giant cell sarcoma, small cell sarcoma, epithelioid sarcoma, myosarcoma, myxosarcoma, pleomorphic rhabdomyosarcoma, embryonal sarcoma, and angiomyosarcoma. Tumor grade was categorized as low (grade 1) and intermediate/high (grade 2–4). Size was grouped as 10 cm or less, between greater than 10 cm and 20 cm or less, and greater than 20 cm. Tumor primary sites were defined according to the third edition of the International Classification of Diseases for Oncology (ICD-O-3): head and neck, lower extremity, retro-intra-abdominal, pelvis, thoracic or trunk, upper extremity, visceral, nerve, or NOS. With the exception of cardiac sarcoma, visceral sarcomas are coded in the NCDB according to their site of origin. Because of the use of nonsarcoma staging, other visceral sites were not included in the analysis.

Patient, Demographic, and Facility Characteristics: Age at diagnosis was classified as younger than 40 years, 40 to 65 years, and older than 65 years. The number of comorbidities was available for cases diagnosed after January 1, 2003, using the Charlson/Deyo score, wherein “0” indicates no comorbidities. Both median income and education level were determined by linking the patient's zip code with the 2000 US census data. Insurance status was recorded at time of primary diagnosis or treatment. Institutions were categorized as community (including comprehensive community cancer programs), academic, or other.

Treatment and Survival: Treatment was documented as no chemotherapy or chemotherapy administered. Patients may have received single-agent or multiagent chemotherapy, but specific drug regimens are unknown. Data on whether chemotherapy was given neoadjuvantly or adjuvantly were reported in known cases. OS time was defined as months from diagnosis until death or last follow-up. Patients who remained alive at the end of follow-up were considered censored.

Statistical Analysis

We compared sociodemographic and tumor characteristics by receipt of chemotherapy using Chi-square tests. We then evaluated these factors with a multiple logistic regression model, using generalized estimating equations5 to account within-hospital effects.

The survival analysis cohort was limited to patients diagnosed during 2003 through 2007 based on availability of comorbidity data and at least 5 years of follow-up. We constructed survival curves using Kaplan-Meier methods, testing for significance with the log-rank test.

We examined the association between chemotherapy administration and survival using propensity score methods.6 In large, observational data sets, propensity scores have several advantages over traditional regression adjustment.7 We used variance-stabilized inverse probability of treatment weights8 to balance covariates across the 2 treatment groups. We estimated the propensity score using all covariates described earlier, and confirmed balance between treatment groups after propensity score weighting. We then created weighted Kaplan-Meier curves9 to estimate adjusted median survival, and used proportional hazards regression to determine the effect of chemotherapy in the study population. We used clustered bootstrap resampling at the hospital level10 with propensity score reestimation to determine P values and confidence intervals. We also used standard proportional hazards regression (unweighted model) to examine the relationship between each covariate and survival.

We used a similar propensity score weighting procedure to examine the effects of chemotherapy according to histology by reestimating the propensity score without histology. We created adjusted Kaplan-Meier curves for each histology, and used proportional hazards regression with an interaction between histology and chemotherapy to test significance. For all models, we checked the proportional hazards assumption using Cox-Snell residuals and accounted for nonproportionality in separate models with time-varying covariates.


Chemotherapy Administration

A total of 16,370 individuals with stage III STS who met inclusion criteria were identified, with 4,428 (27.0%) receiving chemotherapy. Median age of the population was 62 years. Patients were more commonly male (53.7%) and of white race (85.3%). The most common histologies were UPS (N=3,534; 21.6%), liposarcoma (N=3,320; 20.3%), and leiomyosarcoma (N=2,815; 17.2%). The “other” histology group constituted 24.3% of the population. Most tumors arose from the lower extremity (40.0%), followed by the retro-intra-abdominal area (21.9%), thoracic/trunk area (10.5%), and upper extremity (10.0%) (Table 1).

Median age at diagnosis was 53 years in the chemotherapy group and 66 years in the no chemotherapy group. Patients who were younger than 40 years, male, privately insured, earned a higher income, had no comorbidities, had synovial sarcoma, angiosarcoma or “other” histology, and whose tumors were high grade, greater than 10 cm, or were from the lower extremity were significantly more likely to receive chemotherapy on both univariate and multivariate analysis. Visceral sarcomas, which were pre-dominantly cardiac, approached significance as more likely to receive chemotherapy (P=.057). Race, education level, and type of treating facility (academic vs community) were not significantly associated with receipt of chemotherapy (Table 2).

Multiagent chemotherapy was used in 80.0% of patients; 12.5% received single-agent therapy and the remainder received an unknown number of agents. Patients who received neoadjuvant chemotherapy were more likely to have extremity tumors greater than 10 cm (P<.0001; data not shown).

Survival Analysis

A total of 5,377 patients (N=1,494 chemotherapy) were included in the survival analysis. The median follow-up was 47.7 months. A higher risk of death was seen in patients with the following characteristics: age older than 65 years (hazard ratio [HR], 1.35; P<.0001), Charlson/Deyo score of 1 or greater (HR, 1.23; P<.0001), high-grade tumors (HR, 2.72; P<.0001), large tumors (10–20 cm: HR, 1.43; P<.0001; >20 cm: HR, 1.84; P<.0001), male sex (HR, 1.17; P=.0001), and with Medicare (HR, 1.39; P<.0001) or Medicaid insurance (HR, 1.33; P<.0028). Compared with lower extremity tumors, the risk was also higher for retro-intra-abdominal, pelvic, thoracic/trunk, visceral, head and neck, and neural tumors. Compared with UPS histology, MPNSTs were associated with a higher risk of death (HR, 1.44; P=.001) and synovial sarcoma approaching significance (HR, 1.20; P=.068). Patients with fibrosarcoma or liposarcoma histologies were less likely to die (see supplemental eTable 1, available with this article at

Median unadjusted OS in the nonchemotherapy and chemotherapy groups was 51.3 and 82.7 months, respectively (P<0.001). The benefit of chemotherapy was also seen on adjusted analysis (HR, 0.85;

Table 1

Demographic and Clinical Characteristics of Patients with Stage III Soft Tissue Sarcoma Stratified by Chemotherapy Administration

Table 1Table 1
95% CI, 0.76–0.96; P=.004), when controlling for sociodemographic and tumor factors using propensity score weighting. Median adjusted OS was 55.1 months in the no chemotherapy group and 67.1 months in the chemotherapy group (Figure 1). After weighting, all covariates were balanced across treatment group (P>.05), and there were no significant violations of the proportional hazards assumption (P=.2). The results of the propensity score analysis were confirmed using a multivariate regression framework.

Survival by Sarcoma Histology

On univariate analysis, OS was improved in the chemotherapy group for UPS, angiosarcoma, dedifferentiated liposarcoma, leiomyosarcoma, and other histologies (all P≤.05). On adjustment using propensity score weighting, UPS histology remained significant (median OS, 49.1 and 77.8 months for the no chemotherapy and chemotherapy groups, respectively; HR, 0.77; 95% CI, 0.62–0.96; P=.02) (Figure 1 and Table 3).

We found significant violations of the proportional hazards assumption in the model allowing interactions by histology; therefore, we also ran a time-varying survival model. Based on the shape of the survival curves, we allowed the effect of chemotherapy to vary at 24 months, and found that the difference between time periods was statistically significant (P<.001). We found no significant effect of chemotherapy in months 0 through 24 regardless of histology; however, after 24 months, chemotherapy

Table 2

Multivariate Predictors of Receipt of Chemotherapy in Patients With Stage III Soft Tissue Sarcoma

Table 2
was associated with improved survival in UPS (HR, 0.94; P=.81 vs HR, 0.64; P<.001), leiomyosarcoma (HR, 1.10; P=.41 vs HR, 0.75; P=.014), and other (HR, 1.01; P=.67 vs HR, 0.69; P<.002) histologies.


Our retrospective analysis of 16,370 patients with STS in the NCDB demonstrates that most patients (73%) with high-risk sarcoma in the United States do not receive neoadjuvant or adjuvant chemotherapy. As expected, use of chemotherapy is preferred in patients who are young, have fewer comorbidities, and have large tumors. Synovial sarcomas are considered chemosensitive, and this is reflected by the increased use of chemotherapy in this group. Interestingly, patients with myxoid/round cell liposarcoma were less likely to receive chemotherapy, although this subtype is also considered relatively chemosensitive.11 Patients with head and neck, retro-intra-abdominal, and upper extremity tumors were less likely to receive chemotherapy in our analysis. Socioeconomic status has not been previously examined as a factor for receipt of chemotherapy in high-risk STS. Our analysis demonstrated that patients who were privately insured or earned a higher income were more likely to receive chemotherapy, possibly related to access to care. Alternately, no significant differences

Figure 1
Figure 1

Propensity-score adjusted overall survival by receipt of chemotherapy (chemo) for (A) all histologies and (B) undifferentiated pleomorphic sarcoma.

Citation: Journal of the National Comprehensive Cancer Network J Natl Compr Canc Netw 13, 11; 10.6004/jnccn.2015.0165

were seen in chemotherapy administration based on patient race or type of treating facility.

Accepted factors contributing to OS include tumor size and grade, as well as patient-related factors, such as age and health status.12,13 In our study, the impact of these factors on OS was consistent with expectations. In addition, patients with nonprivate insurance had a higher risk of death and were the group least likely to receive chemotherapy. Previous work has suggested that patients with MPNSTs and nonextremity primary sites are at the highest risk for death from sarcoma, whereas patients with liposarcoma and fibrosarcoma tend to fare better.14 Our analysis confirmed these findings. Although patients with retro-intra-abdominal and head and neck sarcomas had a higher risk of death, they were also less likely to receive chemotherapy.

The SMAC meta-analysis and its update suggested an OS benefit for adjuvant chemotherapy, especially in patients receiving doxorubicin and ifosfamide. Included in the updated analysis was an Italian Sarcoma Group study of epirubicin and ifosfamide in high-risk patients (N=104). Patients were eligible if they had a high-grade extremity/girdle tumor that was 5 cm or greater or had recurrent disease. Initially, an OS benefit for the chemotherapy arm was noted (75 vs 46 months; P=.03), with an absolute benefit in OS of 19% at 4 years.15 However, in an updated analysis, after a median follow-up of 89.6 months, the OS difference was no longer significant. The 5-year OS estimate remained statistically significant (66% and 46.1%; P=.04), favoring chemotherapy.16 In contrast, the randomized trial EORTC 62931 (N=351) of

Table 3

Median Survival by Histology Group

Table 3
adjuvant doxorubicin and ifosfamide in patients with intermediate/high-grade STS of any site reported no benefit in relapse-free survival or OS. Included were 85 patients (24.2%) with tumors less than 5 cm. Our study, although not a randomized controlled trial, included more than 5,000 patients in the survival analysis. Furthermore, we defined high-risk STS as tumors that were both large and intermediate/high grade. In our sample, we saw an association between chemotherapy and increased survival, with a HR of 0.85 (95% CI, 0.76–0.96; P=.004). Although the details of chemotherapy are not reported in the NCDB, most patients did receive multiagent chemotherapy. In the United States, the combination of an anthracycline with ifosfamide is often used.

Given the rarity of STS, it is not possible to determine the impact of chemotherapy on OS by histology from randomized trials. Retrospective series have suggested a disease-specific survival (DSS) benefit for adjuvant ifosfamide chemotherapy in patients with synovial sarcoma17,18 and liposarcoma.19 As expected, the most common histologies in our series were UPS, liposarcomas, and leiomyosarcoma. On univariate analysis, OS was improved in the chemotherapy group for UPS, angiosarcoma, dedifferentiated liposarcoma, leiomyosarcoma, and “other” histologies. On adjustment, the benefit for chemotherapy remained in the UPS group. Small sample sizes may have limited our ability to demonstrate statistically significant differences in the remaining histologies. Interestingly, the OS benefit in our analysis occurred mainly after 2 years. Because this is an observational study, the survival difference may be attributed to confounding by indication (ie, patients who receive chemotherapy are healthier and less likely to die of other causes); however, the analysis did carefully control for all known factors potentially affecting survival, and there is no reason to believe that the risk of death from other causes would change after 2 years. Therefore, our result seems to indicate that chemotherapy improves OS in high-risk patients by preventing later deaths from recurrence. This result is in contrast to the retrospective analysis by Cormier et al20 of doxorubicin-based chemotherapy in stage III STS. Here the benefit of chemotherapy on DSS was not maintained after 1 year, suggesting that chemotherapy simply delays time to recurrence. Data for this analysis were derived from 2 institutions between 1984 and 1999. Considerable changes in the diagnosis and classification of STS have occurred in recent years, which may account for the difference in results between the studies.

As with any retrospective database study, there is a potential for coding errors and missing data. Our rate of error appears low, however, with only 1% to 4% of tumors either low grade or less than 5 cm. Given the rarity of STS, classification of these tumors is prone to inherent error, and the use of a database for this analysis precludes verification of the pathologic diagnosis; however, institutions reporting to the NCDB are accredited by the CoC and audited every 3 years.4 Another limitation of the database is the lack of information on cancer-specific survival, time to recurrence, and patient preference regarding chemotherapy use. As in any observational study, our results may be confounded by many known and unknown factors affecting survival. Although our study rigorously controlled for all known factors using propensity score weighting, there may be residual bias from unknown confounders. Finally, we did see violations of the proportional hazards assumption in some survival models, which may result in biased conclusions. To account for this, we used time-varying models, and also calculated nonparametric Kaplan-Meier curves using propensity score weighting, which do not rely on the assumption of proportional hazards.

Our analysis of this large data set suggests that chemotherapy administration in patients with high-risk sarcomas may be related to socioeconomic factors, in addition to expected tumor and patient factors. We also showed that OS was higher in patients receiving chemotherapy in this high-risk population, specifically in patients with UPS. In conclusion, this study demonstrates an OS benefit for chemotherapy in stage III STS after adjusting for comorbidities and other known factors. To our knowledge, this is the first retrospective OS analysis restricted to patients with stage III STS that also adjusts for socioeconomic status.


The data used in the study are derived from a deidentified NCDB file. The American College of Surgeons and the Comission on Cancer have not verified and are not responsible for the analytic or statistical methodology used, or the conclusions drawn from these data by the investigators.

See for supplemental online content.

The authors have disclosed that they have no financial interests, arrangements, affiliations, or commercial interests with the manufacturers of any products discussed in this article or their competitors.


  • 1

    Edge SB BD, Compton CC, Fritz AG et al. , eds. American Joint Committee on Cancer Staging Manual, 7th ed. New York, NY: Springer; 2010.

  • 2

    Adjuvant chemotherapy for localised resectable soft-tissue sarcoma of adults: meta-analysis of individual data. Sarcoma Meta-analysis Collaboration. Lancet 1997;350:16471654.

    • Search Google Scholar
    • Export Citation
  • 3

    Pervaiz N, Colterjohn N, Farrokhyar F et al. . A systematic meta-analysis of randomized controlled trials of adjuvant chemotherapy for localized resectable soft-tissue sarcoma. Cancer 2008;113:573581.

    • Search Google Scholar
    • Export Citation
  • 4

    Bilimoria KY, Stewart AK, Winchester DP, Ko CY. The National Cancer Data Base: a powerful initiative to improve cancer care in the United States. Ann Surg Oncol 2008;15:683690.

    • Search Google Scholar
    • Export Citation
  • 5

    Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986;73:1322.

  • 6

    Rosenbaum PR, Rubin DB. The Central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:4155.

  • 7

    Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399424.

    • Search Google Scholar
    • Export Citation
  • 8

    Xu S, Ross C, Raebel MA et al. . Use of stabilized inverse propensity scores as weights to directly estimate relative risk and its confidence intervals. Value Health 2010;13:273277.

    • Search Google Scholar
    • Export Citation
  • 9

    Xie J, Liu C. Adjusted Kaplan-Meier estimator and log-rank test with inverse probability of treatment weighting for survival data. Stat Med 2005;24:30893110.

    • Search Google Scholar
    • Export Citation
  • 10

    Davison A, Hinkley DV. Bootstrap Methods and Their Application. New York, NY: Cambridge University Press; 1997.

  • 11

    Jones RL, Fisher C, Al-Muderis O, Judson IR. Differential sensitivity of liposarcoma subtypes to chemotherapy. Eur J Cancer 2005;41:28532860.

  • 12

    Zagars GK, Ballo MT, Pisters PW et al. . Prognostic factors for patients with localized soft-tissue sarcoma treated with conservation surgery and radiation therapy: an analysis of 1225 patients. Cancer 2003;97:25302543.

    • Search Google Scholar
    • Export Citation
  • 13

    Coindre JM, Terrier P, Guillou L et al. . Predictive value of grade for metastasis development in the main histologic types of adult soft tissue sarcomas: a study of 1240 patients from the French Federation of Cancer Centers Sarcoma Group. Cancer 2001;91:19141926.

    • Search Google Scholar
    • Export Citation
  • 14

    Kattan MW, Leung DH, Brennan MF. Postoperative nomogram for 12-year sarcoma-specific death. J Clin Oncol 2002;20:791796.

  • 15

    Frustaci S, Gherlinzoni F, De Paoli A et al. . Adjuvant chemotherapy for adult soft tissue sarcomas of the extremities and girdles: results of the Italian randomized cooperative trial. J Clin Oncol 2001;19:12381247.

    • Search Google Scholar
    • Export Citation
  • 16

    Frustaci S, De Paoli A, Bidoli E et al. . Ifosfamide in the adjuvant therapy of soft tissue sarcomas. Oncology 2003;65(Suppl 2):8084.

  • 17

    Canter RJ, Qin LX, Maki RG et al. . A synovial sarcoma-specific preoperative nomogram supports a survival benefit to ifosfamide-based chemotherapy and improves risk stratification for patients. Clinical Can Res 2008;14:81918197.

    • Search Google Scholar
    • Export Citation
  • 18

    Eilber FC, Brennan MF, Eilber FR et al. . Chemotherapy is associated with improved survival in adult patients with primary extremity synovial sarcoma. Ann Surg 2007;246:105113.

    • Search Google Scholar
    • Export Citation
  • 19

    Eilber FC, Eilber FR, Eckardt J et al. . The impact of chemotherapy on the survival of patients with high-grade primary extremity liposarcoma. Ann Surg 2004;240:686695; discussion 695–697.

    • Search Google Scholar
    • Export Citation
  • 20

    Cormier JN, Huang X, Xing Y et al. . Cohort analysis of patients with localized, high-risk, extremity soft tissue sarcoma treated at two cancer centers: chemotherapy-associated outcomes. J Clin Oncol 2004;22:45674574.

    • Search Google Scholar
    • Export Citation

Correspondence: Sujana Movva, MD, Department of Medical Oncology, Fox Chase Cancer Center, 333 Cottman Avenue, Philadelphia, PA 19111. E-mail:

Supplementary Materials

  • View in gallery

    Propensity-score adjusted overall survival by receipt of chemotherapy (chemo) for (A) all histologies and (B) undifferentiated pleomorphic sarcoma.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 358 142 2
PDF Downloads 134 53 2
EPUB Downloads 0 0 0