No population-based studies of retroperitoneal sarcoma (RPS) have been conducted, and the use and timing of adjuvant radiotherapy for RPS is controversial. The objective of this study was to examine the incidence and treatment of RPS, specifically regarding the use of adjuvant radiotherapy.
The Surveillance, Epidemiology, and End Results (SEER) database was used to evaluate the incidence of RPS over a 29-year period (1973-2001). The rate of surgery, the rate and timing of adjuvant radiotherapy, and the influence of demographic factors on treatment were evaluated.
A total of 2348 cases of RPS were identified. The mean annual incidence of RPS was 2.7 cases per 106 persons and did not change significantly over time (2.6 in 1973 vs. 2.8 in 2001; P = .92). Most patients (1654; 70.4%) underwent surgical resection. Radiotherapy was used in 428 patients (25.9%) who underwent surgery; radiation was given postoperatively in 366 (85.5%), preoperatively in 20 (4.7%), and intraoperatively or unknown in 42 (9.8%). Patients who received any adjuvant radiotherapy were on average 5 years younger than those who underwent surgery alone (P < .0001). Radiotherapy was more commonly used among whites than African Americans (25.8% vs. 16.7%; P = .02) and there was significant variation in the use of adjuvant radiotherapy by geographic location (P = .003). On multivariate analysis, race (P = 0.004), age (P < .0001), and geographic location (P = .006) were independently associated with the use of adjuvant radiotherapy.
Soft-tissue sarcomas represent a heterogeneous group of rare tumors that arise predominantly from the embryonic mesoderm. It is estimated that there will be 9420 new cases of soft-tissue sarcoma in 2005 in the US; approximately 15% of such cases will arise in the retroperitoneum.1 The prognosis for patients with retroperitoneal sarcoma (RPS) is relatively poor, with a 36% to 58% overall 5-year survival rate and a natural history characterized by late recurrence.2–7 Locoregional recurrence remains a frequent cause of death; only 28% of patients do not experience such a recurrence within 5 years. The only known potentially curative treatment is macroscopically complete, margin-negative surgical resection.2, 8, 9
Although surgical resection remains the mainstay of RPS treatment, the size and complexity of RPS tumors often results in microscopic residual disease after surgery; thus, the use of adjuvant radiotherapy has been proposed. To date, only one randomized trial, performed more than 2 decades ago, has been performed examining the role of intraoperative radiation for RPS.10 That trial, which randomized 35 patients to receive postoperative external-beam radiotherapy with or without intraoperative radiotherapy, demonstrated a significant reduction in local recurrence in the group randomized to the treatment arm that included intraoperative radiotherapy. More recently, 2 prospective, nonrandomized, single-institution trials have demonstrated the feasibility of preoperative radiotherapy for resectable RPS, and the results suggest improved outcomes compared with historical surgery-alone data.11, 12 This body of evidence prompted the American College of Surgeons Oncology Group to initiate a large, randomized clinical trial comparing surgery alone with preoperative radiotherapy plus surgery (ACOSOG Z9031).
Although there have been several, primarily single-institution, cohort studies of this rare disease,2–9, 13, 14 it is unclear whether their findings can be generalized to the majority of RPS patients. Given the lack of population-based studies of RPS, as well as the controversy surrounding the use and timing of adjuvant radiotherapy, the goal of the present study was to examine the incidence and treatment of RPS in a population-based cohort, specifically regarding the use of adjuvant radiotherapy.
MATERIALS AND METHODS
We used data from the Surveillance, Epidemiology, and End Results (SEER) cancer registry to conduct this study. SEER is a population-based cancer registry sponsored by the National Cancer Institute that was created in 1973. More than 3.5 million cancer cases are included in the SEER database, with approximately 170,000 new cases added annually. It collects information on cancer incidence and survival from 11 population-based cancer registries and 3 supplemental areas; these 11 registries include approximately 14% of the US population.15 Epidemiologically distinct subgroups focusing on race, socioeconomic status, and education are incorporated in SEER to enhance its generalizability to the US population. For example, 12% of SEER compared with 13% of the US population is below the poverty level.15 Of the 11 registries, 2 were added in 1992. The information collected by SEER includes patient characteristics, county of residence, primary tumor site, tumor type, first course of treatment (through completion of the initial treatment plan, including treatment within the first year after diagnosis or until there is evidence either of disease progression or of treatment failure within the first year), timing of radiation, and follow-up for vital status.15
Included in our study were all patients ≥ 18 years old in the SEER database who were diagnosed with RPS from January 1, 1973, through December 31, 2001. For the purposes of this study, the patients included were those reported to have a primary tumor of the retroperitoneum (ICD-0-2 topography site code C480), as well as histologic morphology consistent with RPS; these tumor histologies included soft tissue (ICD-0-2 880), fibromatous (ICD-0-2 881-883), myxomatous (ICD-0-2 884, 889-892), lipomatous (ICD-0-2 885-888), complex mixed/stromal (ICD-0-2 893-899, excluding endometrial stromal, mullerian mixed, and nephroblastoma), fibroepithelial (ICD-0-2 900-903), synovial (ICD-0-2 904), angiosarcoma (ICD-0-2 9120), hemangiopericytoma (ICD-0-2 9150), neurofibrosarcoma (ICD-0-2 9540), and alveolar soft part sarcoma (ICD-0-2 9581).
SEER routinely collects data on the first course of treatment, including surgery and/or radiation. For our study, we examined the use of adjuvant radiotherapy among patients undergoing surgical resection. Only patients undergoing surgical resection were categorized as having surgery; patients undergoing open biopsy or palliative procedures such as ostomy creation were not categorized as having surgery. The use of adjuvant radiotherapy was categorized as none, preoperative, postoperative, intraoperative, or other.
The crude annual incidence of RPS for the 29-year period was calculated by dividing the total number of new RPS cases each year by the population included in the registries (1973-1992, 9 registries; 1993-2001, 11 registries). We then calculated age-adjusted annual incidence rates, standardized to the 2001 US population (10 year age groups were used). To test for significant trends in age-adjusted RPS incidence from 1973 to 2001, we performed a nonparametric test for trends using the Cochran-Armitage test based on 1 degree of freedom.16, 17
We calculated the rate of surgical resection as a proportion of all new cases of RPS; the use and timing of adjuvant radiotherapy was examined among patients undergoing surgical resection. We then assessed the association of the following factors on treatment: patient age (evaluated in 10-year increments), gender, race (white vs. African American), marital status (married vs. unmarried), year of diagnosis, and geographic location (SEER registry) on treatment. Univariate analysis was done using the chi-square test. Multivariate analysis, by logistic regression, was performed to evaluate factors associated with the use of adjuvant radiotherapy using a forward stepwise approach incorporating all clinicodemographic variables examined as well as all 2-way interaction terms; acceptance in the final model required significance at P<0.10. No interaction terms were found to be statistically significant; however, we kept the interaction between registry and year of diagnosis in the final model given that 2 of the registries contributed data only during 1993-2001. All analyses were performed by using SPSS for Windows 11.0 (Chicago, IL). As this analysis used publicly available data with no personal identifiers, the Research Ethics Board at Dalhousie University determined that it was exempt from review.
We identified 2348 patients with newly diagnosed RPS registered in the SEER database during the specified study period. The median age of the study cohort was 64 years. Other sociodemographic characteristics are summarized in Table 1. The annual incidence of RPS, both crude and age-adjusted using rolling 3-year averages, is displayed in Figure 1. No significant trends in incidence of RPS over time were identified.
Table 1. Demographic and Patient Characteristics of the Study Cohort (N = 2348)
No. of Patients
Nine registries were available until 1991; San Jose-Monterey and Los Angeles were added in 1992.
Surgical resection was performed in 1654 (70.1%) patients. The proportion of patients undergoing surgical resection increased significantly over the study period (Fig 2). The average resection rate was 54.8% over the first 6 years of the study, compared with 78.5% over the last 6 years (P < 0.0001).
Adjuvant radiotherapy was used in 428 of the 1654 patients who underwent surgery (25.9%); most such patients received radiotherapy postoperatively (366, 85.5%), whereas 20 patients (4.7%) received radiotherapy preoperatively (Table 2). Factors associated with the use of adjuvant radiotherapy are shown in Table 3. Patients who received adjuvant radiotherapy were on average 5 years younger than those who underwent surgery alone (median age = 59 vs. 64 years; P < .0001). Adjuvant radiotherapy was used more often among whites than African Americans (25.8% vs. 16.7%; P = .02) and there was significant variation in the use of adjuvant radiotherapy by geographic location (P = 0.003). No significant associations were identified between the use of adjuvant radiotherapy and sex, marital status, or year of diagnosis.
Table 2. Type of Adjuvant Radiotherapy Used Among Patients Undergoing Surgical Resection (N = 1654)
To further investigate the variation in use of adjuvant radiotherapy by geographic location, we categorized the 11 registries into high- and low-volume locations, based on the average number of RPS cases per year, dichotomized around the 50th percentile of cases per year. Three registries were in the high-volume group (9.1-21.8 cases/yr) and eight registries in the low-volume group (1.9-7.1 cases/yr). No differences in the use of adjuvant radiotherapy were identified between low- and high-volume registries (26.6% vs. 25.1%, respectively; P = 0.5).
On multivariate analysis using logistic regression, race (P = .01), age (P < .0001), and geographic location (P = .02) were found to be independently associated with the use of adjuvant radiotherapy (Table 3).
This study demonstrated, at a population level, a stable incidence of RPS over a 29-year period. Although no published incidence data specific to RPS exist, stable incidence of all cases of soft-tissue sarcoma has been reported within the National Cancer Database.18 In SEER, the incidence of extremity soft-tissue sarcoma did not change from 1973 to 1998.19
The overall resection rate of 70.4% identified in this study falls within the wide range of 50% to 95% reported in single-institution cohort studies.2, 4, 6, 14 To the authors' knowledge, no population-based data on RPS resection rates exist. It is possible that improvements in pretreatment radiologic staging, particularly the widespread introduction and use of computed tomography (CT), and secondary progressive improvements in CT image quality, might have significantly contributed to the increased resection rate over time in our analysis. It is also possible that changes in surgical technique over time, particularly the increased use of extensive resection with en-bloc removal of adjacent organs, may have contributed to the increase in resection rate. In a review of 192 RPS patients undergoing surgery at the Mayo Clinic between 1960 and 1995, complete resection rates increased from 49% (during 1960-1982) to 78% (during 1983-1995), with a concomitant reduction in biopsy-only procedures.20 The authors attributed the increase in resections to a more liberal use of multivisceral resection but did not consider the possible role of improved pretreatment staging.
The use of adjuvant radiotherapy in RPS has been debated for more than 40 years. Although the only randomized clinical trial of therapy for this disease suggests a benefit of intraoperative radiotherapy with postoperative radiotherapy,10 the small study size (35 patients), significant radiation-related toxicity, lack of widespread availability of intraoperative radiotherapy, and high rates of recurrence in both arms of this trial have prevented the widespread acceptance of its results. Several retrospective studies have suggested that postoperative radiotherapy yield better outcomes than surgery alone,7, 21 although other similarly designed studies have shown no advantage to radiotherapy.2, 20, 22 Moreover, the potential advantages of preoperative radiotherapy, and demonstration of its feasibility, have provided sufficient clinical equipoise to launch a large Phase III trial of surgery versus preoperative radiotherapy plus surgery (ACOSOG Z9031).11, 12
Our data demonstrate that surgery alone is the most common treatment administered to patients with localized RPS; 74.1% of patients in this study were treated in this fashion. Among the 25.9% of patients in whom adjuvant radiotherapy was used, the most common approach was postoperative (85.5%). The more common use of postoperative radiation was likely related to the common use of diagnostic and therapeutic primary tumor resection as the initial treatment for most patients with radiologically resectable retroperitoneal neoplasms. In the absence of a pretreatment diagnosis or clear evidence suggesting that radiotherapy would improve outcome, most patients are offered surgical resection as the initial treatment for localized disease. Once the pathologic diagnosis has been made and margin assessment done, selected patients (e.g., microscopic margin-positive) are subsequently referred for consideration of radiotherapy. Unfortunately, the SEER data do not reliably distinguish R0 from R1 resections, making it impossible to determine whether this practice was followed in this study. It is important to note that no conclusive evidence supports an outcome difference between patients undergoing an R0 versus R1 resection, and similarly, no evidence points to a preferential benefit of radiotherapy in patients undergoing R1 resections.2, 4, 23
Variation in the use of cancer therapies among different populations has been documented in many cancers, such as those of the breast, cervix, colorectal tract, lung, and prostate.24–31 As with any medical therapy, factors involved in such nonrandom variation can be categorized as patient, structural barriers, and physician/clinical.32 The racial disparity identified in this study, specifically that African Americans were significantly less likely than whites to receive adjuvant radiotherapy, has been demonstrated for rectal carcinoma and breast carcinoma.25, 28, 32 In an analysis of SEER data from 1973 to 1999, Alderman et al.19 found that African Americans were half as likely as Caucasians to receive adjuvant radiotherapy for upper extremity sarcomas, with a magnitude of difference virtually identical to that found in our study. Our data add to the vast body of literature demonstrating that African Americans receive less medical care and have poorer outcomes than whites.32 Further research is required to better delineate possible explanations for this pattern, which may include physician/provider bias, differences in patient acceptance/compliance, differences in access to care, or race serving as a surrogate for socioeconomic status.
Variations in cancer treatment have also been noted between institutions and geographic areas.19, 30, 33 Such variations have been attributed to clinical volume of the area or institution.33–37 Interestingly, we found that the variation in use of adjuvant radiotherapy for RPS across the 11 SEER registries did not appear to be related to the volume of RPS cases within individual registries. It is plausible that RPS in the high-volume registries was still so uncommon that simple individual institution practice patterns, rather than ‘experience,’ explains the treatment differences.
Limitations of this study include the potential for unmeasured factors that could confound results. The benefits of a large database such as SEER in terms of generalizability can come at the cost of a lack of the detailed sociodemographic and clinical data that are characteristic of smaller clinical cohort studies. For example, data on RPS grade are not reliably available through SEER for the entire 29-year study period. Additionally, coding of race for such groups as Native Americans, Asians, and Hispanics was not consistent over the SEER data collection periods, making more detailed race analyses impossible. Our approach of categorizing race as “White,” “African American,” and “Other” is congruent with other publications using SEER data from 1973 to 2001.19, 28 Detailed data on radiation dose or field was also not available. However, the data for use of both surgery and radiotherapy collected by SEER are accurately recorded for breast, endometrial, colorectal, lung, pancreatic, and prostate carcinomas.38, 39 There is, therefore, no apparent reason to believe that these data would not be the case for RPS. Finally, this study did not include any measurement or analysis of population-based outcomes, making it impossible to determine the impact of temporal changes in resectability rates on survival.
In conclusion, the incidence of RPS, a rare disease, appears stable and resection rates have increased over time. Most patients who undergo surgery do not receive any adjuvant radiotherapy, and very few receive preoperative radiotherapy. Differences in adjuvant radiotherapy use that seem to be based on demographic and geographic factors may, at least in part, reflect differences in individual and institutional practice patterns. These data, representing current practice, will be important in considering how to implement future results of an ongoing randomized trial (ACOSOG Z9031) comparing preoperative radiotherapy and surgery with surgery alone. A demonstrated benefit to preoperative radiotherapy would require a significant change to the current practice patterns.