This article presents the current understanding of prognostic factors and outcomes for retroperitoneal sarcoma. The discussion focuses on the literature published since 2000, including studies with cohorts of at least 30 patients. An interpretation of factors that have contributed to an improvement in outcomes has been performed. With demonstration of the potential for good long-term outcomes and a better appreciation for the importance of minimizing treatment-related morbidity, the international sarcoma community has increasingly recognized and emphasized the importance of expert multidisciplinary assessment and care for this rare malignancy.
Retroperitoneal sarcoma (RPS) represents 10% to 15% of all soft tissue sarcoma (STS), with a stable annual incidence of approximately 2.7 cases per 1 million persons in the United States. A similar incidence has been reported for the population of southern Sweden. The disease has no apparent geographic variation in incidence and no predisposition by gender or race. In patients with hereditary or sporadic genetic conditions that predispose to STS, the proportion of these tumors that arise in the retroperitoneum is not appreciably different from that in the general population. However, the distribution of histologic subtypes in RPS differs from that of STS arising at other anatomic sites. Approximately 50% of RPSs are liposarcoma. The second most frequent histology is leiomyosarcoma; a small proportion will arise from the smooth muscle of the inferior vena cava (IVC). Ongoing debate surrounds whether other histologic subtypes exist (eg, malignant fibrous histiocytoma [MFH], pleomorphic sarcoma); some authors suggest that many of these are, in fact, dedifferentiated liposarcoma.
Currently, no consensus exists in the literature regarding the need for a pretreatment percutaneous biopsy of a suspected RPS. Given that the differential diagnosis is actually fairly broad (eg, lymphoma, testicular/germ cell tumors, benign or malignant adrenal/renal tumors, benign peripheral nerve sheath tumors, ganglioneuroma, lymphangioma, retroperitoneal fibrosis, metastatic carcinoma) with widely differing treatments, the benefits of securing a histologic diagnosis before formulating a treatment plan outweigh the theoretical risk of tumor seeding along the biopsy track. Furthermore, if neoadjuvant treatment is being contemplated, confirming the RPS diagnosis is important. Fine needle aspiration biopsy is of little use because a pathologic diagnosis commonly relies on tissue architecture; therefore, core biopsy is strongly preferred. With modern techniques, including withdrawal of the core needle and specimen into the hub of the biopsy gun, and sophisticated image guidance, the potential for tumor contamination of surrounding tissues is minimized. In contrast, open surgical biopsy of a suspected RPS is much more likely than a percutaneous biopsy to expose virgin tissue planes to tumor and miss viable tumor material, and requires a sizable incision with attendant risks. Therefore, open biopsy should be used only if repeated percutaneous biopsies have failed and the management would change based on a definitive RPS diagnosis.
Surgical resection remains the cornerstone of RPS treatment. In the United States from 1973 to 2001, the resection rate has increased; it is currently approximately 80%. Several reports indicate that resectability may be optimized by referral to specialized centers. However, an attempt at resection may not be appropriate in all cases, because the presence of distant metastatic disease or poor patient performance status can portend abbreviated survival regardless of treatment. Resection as a sole therapeutic modality has resulted in 5-year local recurrence rates of 50% to 70% and 5-year overall survival rates of 40% to 50%. Local recurrence and death from sarcoma continue at a significant rate beyond 10 years, and even up to 20 years after resection.
Given the failure of surgery alone to control most RPS, considerable interest has been shown in adjuvant therapies. Except for particularly chemosensitive histologies, such as extraosseous Ewing’s, chemotherapy in general has not been systematically used. Because of the success of radiation therapy (RT) at reducing local recurrence of extremity STS, its potential benefit in RPS (in which margins are typically compromised) has been explored. Before 2001, RT was used in approximately 25% of patients with RPS undergoing resection (postoperatively in 75% of cases). Postoperative RT is limited by toxicity and does not seem to prevent recurrence; it has been largely abandoned at many experienced centers. The role of preoperative RT remains unclear, as discussed later in this article.
RPS clinical outcome measures
Local Recurrence
RPS has a marked propensity to recur locoregionally within the abdominal cavity, as opposed to at distant sites from hematogenous dissemination. In particular, patients with liposarcoma rarely develop distant metastases. The mechanism of death from recurrent RPS is most frequently related to progression of locoregional recurrence or complications of its treatment. Thus, RPS local progression-free survival closely mimics overall cancer-specific survival. Although complete gross resection of a locoregional recurrence is often technically feasible and some patients undergo multiple resections, these are highly selected cases. Therefore, reports on the results of resection of recurrent disease are prone to significant selection bias. Patients who experience recurrence with an aggressive, technically unresectable tumor are usually not included in surgical series. Those who experience recurrence with more indolent, resectable disease are noted to have a surprisingly favorable survival, probably reflecting tumor biology as much as a beneficial treatment effect.
Baseline postresection imaging, typically a CT scan with both oral and intravenous contrast, is critical to act as a comparator for future surveillance studies. Radiologic evidence of RPS recurrence often predates the development of symptoms. For low-grade tumors, this interval can sometimes be measured in years. Survival after resection of an asymptomatic local recurrence might be superior to resection of a symptomatic recurrence, but this may simply reflect lead-time bias. Another issue related to assessing RPS treatment outcomes is the confirmation of a local recurrence; sometimes it requires serial imaging to definitively document progressive disease. Whether the date of recurrence should be reported as that of the first suspicious scan or the ensuing confirmatory scans is unclear. In addition, the interval between scans during long-term follow-up can often be 1 to 2 years, further confounding this outcome measure.
The risk for local recurrence of RPS occurs at a significant rate even after 5 years following curative intent surgery. For LPS, local recurrence rates 5 years after complete gross resection are approximately 50% for well-differentiated and 80% for dedifferentiated tumors. In studies that include longer follow-up, local recurrence risk does not plateau at 5 years, but instead continues at a near-linear pace. Therefore, the follow-up of patients with RPS assumes a lifelong aspect, highlighting the importance of institutional commitment to the follow-up of these patients and also the difficulties of designing clinical trials that can be concluded within a reasonable time-frame.
Overall Survival
RPS is distinct from STS at other sites in that cancer-specific survival continues to decline significantly beyond 5 and even 10 years after resection. Most older published series (single-institution series collected over 20 to 30 years using surgery alone) note 5-year overall survival rates of 50%. More recent population-based estimates are similar to these older series, also showing 5-year overall survival rates of approximately 50%. In studies that include longer follow-up, 10-year actuarial disease-related survival decreases to 20% to 30%, again highlighting the importance of extended follow-up in assessing RPS treatment outcomes.
Prognostic variables for RPS outcome
The failing of the American Joint Committee on Cancer (AJCC) STS staging system as applied to RPS outcomes has been emphasized by several authors. Very few of these tumors are smaller than 5 cm, with the median maximal dimension in most series being 25 to 30 cm. By definition, all RPS are deep. Distant metastases are uncommon at diagnosis but do predict poor survival when present. The element of the AJCC system most valuable in predicting outcome in RPS is grade. Grade and resection status are the two prognostic factors recognized as useful for comparing outcomes between series and in counseling individual patients. Although resection status and grade are interrelated, with high-grade tumors less likely to be completely resected, the two variables are still independent predictors of prognosis in most larger series.
Grade
Grade is consistently found to be an important prognostic factor in most RPS series, including an analysis of 1535 patients from the Surveillance, Epidemiology, and End Results (SEER) database. Using the French three-tier grading system, Bonvalot and colleagues showed that grade was the most significant independent predictor of abdominal recurrence-free and overall survival in 364 patients with resected primary RPS. An analysis of 261 cases of primary RPS from merged tumor registries from more than 150 United States centers yielded similar results for high versus low grade. Grobmyer and colleagues analyzed 78 cases of recurrent RPS and also found that high-grade tumor was an independent predictor of worse overall survival.
Resection (R) status
R0 resection (histologically confirmed circumferentially negative margins) is the universally upheld goal of RPS surgery. In practice, its achievement is challenging and, some would argue, illusory. With large tumors, it is not technically feasible to pathologically examine all margins of resection. Areas of concern identified by the surgeon or pathologist should be specifically assessed. Pathologic analysis may reveal tumor within a particular distance of the margin of resection, such as within 0.1 mm; this resection would then be labeled R1. Variations in the sampling intervals of the circumference of the tumor and the strictness with which measurements are made likely account for the wide variations in the R0 versus R1 rates reported in the literature ( Table 1 ).
| Author, Year | Number Resected | Primary Present n (%) | R0 n (%) | R1 n (%) | R2 n (%) | R Status Unknown n (%) | N.B. |
|---|---|---|---|---|---|---|---|
| Bonvalot et al, 2010 | 249 | 100 | R0/1:232 (93) | 17 (7) | |||
| Strauss et al, 2010 | 200 | 100 | 55 (28) | 85 (42) | 30 (15) | 30 (15) | 170 known to have total gross resection (R0/1) |
| Sampath et al, 2010 | 261 | 100 | 109 (42) | R1/2: 30 (11) | 122 (47) | ||
| Lehnert et al, 2009 | 99 | 63 | 38 (38) | 36 (36) | 25 (25) | 11 explored or biopsied only | |
| Bonvalot et al, 2009 | 374 | 100 | 176 (47) | 103 (28) | 38 (10) | 57 (15) | 8 biopsied only |
| van Dalen et al, 2007 | 115 | 100 | R0/1:78 (68) | 37 (32) | 8 explored only | ||
| Ballo et al, 2007 | 83 | 72 | 44 (53) | R1/uncertain: 39 (47) | |||
| Chiappa et al, 2006 | 47 | 49 | 28 (60) | 3 (7) | 16 (33) | ||
| Pierie et al, 2006 | 103 | 100 | R0/1:62 (60) | 41 (40) | |||
| Alldinger et al, 2006 | 117 | 100 | 20 (17) | 54 (46) | 41 (35) | 2 (2) | 47 resected at outside institution |
| Erzen et al, 2005 | 100 | 55 | 55 (55) | 42 (42) | 3 (3) | 2 biopsied only | |
| Zlotecki et al, 2005 | 39 | 88 | 25 (64) | 9 (23) | 5 (13) | 1 died preoperatively | |
| Hassan et al, 2004 | 89 | 100 | R0/1:76 (85) | 13 (15) | 8 biopsied only | ||
| Gronchi et al, 2004 | 167 | 49 | R0/1:147 (88) | 20 (12) | |||
| Gilbeau et al, 2002 | 45 | 100 | 17 (38) | 26 (58) | 2 (4) | ||
| Jones et al, 2002 | 46 | 65 | R0/1:46 (100) | 0 | 9 progressed preoperatively | ||
| Stoeckle et al, 2001 | 145 M0 | 100 | R0/1:94 (65) | Not available | Not available | 20 had M1 | |
| Gieschen et al, 2001 | 33 | 78 | R0/1:29 (88) | 4 (12) | 4 no attempt to resect | ||
| Alektiar et al, 2000 | 32 | 38 | R0/1:30 (94) | 2 (6) |
Despite the noted limitations, R status has been identified as a significant predictor of local recurrence and overall survival for resected RPS. Singer and colleagues were among the first to emphasize that R status, which is not a component of AJCC/International Union Against Cancer (UICC) staging, trumps every other staging element in patients with no obvious distant metastatic disease. Patients who have undergone an R2 resection have inferior overall survival compared with R0 or R1 resection. In some reports, R2 resection rates are markedly lower at specialized centers versus community hospitals. Although this may reflect a more aggressive surgical approach, it may also suggest better patient selection based on review of preoperative imaging. Because the goal of surgery should be complete gross tumor resection, the potential benefits of incomplete resection remain controversial. Some series have shown that overall survival is longer in patients who have undergone R2 resection than in those who did not undergo surgical exploration. In this regard, a significant source of bias and also heterogeneity between series is in patient selection. For example, in most series in which patients had an R2 resection, that was not the intent of the planned surgery. Rather, they were taken to the operating room with the plan of performing a total gross resection. Therefore, this group would be different a priori from patients in whom the imaging suggested an inability to perform a complete resection, thus precluding surgical exploration.
Primary Versus Recurrent Tumor
For the most part, large population-based analyses and multi-institutional series have focused on primary RPS. Most available information on patients with recurrent RPS comes from small retrospective surgical series. These series typically are composed of patients who underwent resection or were at least subjected to laparotomy with the intent of a curative resection. There is almost certainly a major selection bias by excluding from these analyses patients with “bad” recurrent disease who were deemed inoperable. The experienced RPS surgeon is likely to be very selective in choosing patients with recurrent RPS for attempted resection. Patient, treatment, and tumor factors (eg, obesity, debility, difficulty of the first operation, multifocal tumor, bone or major vascular involvement) that may not be readily captured in a published multivariable analysis will nonetheless be considered by the expert surgeon. These different factors may at least partially explain the discrepancies in the literature with respect to the prognosis of patients with recurrent disease, which is generally inferior to that of those with primary RPS.
Compared with a 69% 5-year disease-specific survival rate and a 57% 5-year overall survival rate for primary RPS resection, the reported 5-year overall survival after R0/1 resection for recurrent RPS is much lower, at approximately 30%. In the authors’ own prospective study of multimodality radiation and surgery, disease-free and overall survivals were also significantly lower in the recurrent versus primary RPS subgroups. In 70 patients with RPS (6 primary, 64 recurrent), most of whom were given intraoperative radiation, Dziewirski and colleagues found that the patients with recurrent RPS had a 60% 2 year local recurrence–free survival rate (2-year disease-free survival was 58% in the authors’ series) and a 75% 2-year overall survival rate (the authors’ was 74%). These results emphasize that, even with limited follow-up, patients with recurrent RPS are quickly failing and dying of disease. Echoing these data, Ballo and colleagues showed that 5-year local control rate was only 27% for resected recurrent RPS (n = 23) versus 58% for primary tumors (n = 60).
In contrast, two different studies showed that, although recurrent status was an independent predictor of worse local control and disease-free survival, it did not impact overall survival. A group from Heidelberg analyzed the outcomes of curative intent resection in 110 patients with RPS (71 primary, 39 recurrent cases) with a long median follow-up (89 months). Resectability rates, morbidity, and operative mortality were not different between the groups. However, 5-year local control was only 9% in the recurrent group compared with 59% in the primary group. Disease-specific survival rates after complete resection were comparable between the groups: 51% versus 43%, respectively. This phenomenon of multiple resections in highly selected patients with recurrent RPS (who eventually succumb to progressive disease or fatal operative complications, but only many years after their initial local recurrence) highlights selection bias as an important confounder of reported recurrent RPS outcomes. In a tri-institutional German series, Alldinger and colleagues refer to this as an explanation for their finding that patients who had a local recurrence at some point in their disease course often had a better survival than those who never had local recurrence.
Prognostic variables for RPS outcome
The failing of the American Joint Committee on Cancer (AJCC) STS staging system as applied to RPS outcomes has been emphasized by several authors. Very few of these tumors are smaller than 5 cm, with the median maximal dimension in most series being 25 to 30 cm. By definition, all RPS are deep. Distant metastases are uncommon at diagnosis but do predict poor survival when present. The element of the AJCC system most valuable in predicting outcome in RPS is grade. Grade and resection status are the two prognostic factors recognized as useful for comparing outcomes between series and in counseling individual patients. Although resection status and grade are interrelated, with high-grade tumors less likely to be completely resected, the two variables are still independent predictors of prognosis in most larger series.
Grade
Grade is consistently found to be an important prognostic factor in most RPS series, including an analysis of 1535 patients from the Surveillance, Epidemiology, and End Results (SEER) database. Using the French three-tier grading system, Bonvalot and colleagues showed that grade was the most significant independent predictor of abdominal recurrence-free and overall survival in 364 patients with resected primary RPS. An analysis of 261 cases of primary RPS from merged tumor registries from more than 150 United States centers yielded similar results for high versus low grade. Grobmyer and colleagues analyzed 78 cases of recurrent RPS and also found that high-grade tumor was an independent predictor of worse overall survival.
Resection (R) status
R0 resection (histologically confirmed circumferentially negative margins) is the universally upheld goal of RPS surgery. In practice, its achievement is challenging and, some would argue, illusory. With large tumors, it is not technically feasible to pathologically examine all margins of resection. Areas of concern identified by the surgeon or pathologist should be specifically assessed. Pathologic analysis may reveal tumor within a particular distance of the margin of resection, such as within 0.1 mm; this resection would then be labeled R1. Variations in the sampling intervals of the circumference of the tumor and the strictness with which measurements are made likely account for the wide variations in the R0 versus R1 rates reported in the literature ( Table 1 ).
Related posts:
Prognostic Factors and Staging for Soft Tissue Sarcomas: An Update
Margin Status, Local Recurrence, and Survival: Correlation or Causation?
Adjuvant Chemotherapy for Soft Tissue Sarcomas
Pulmonary Metastasectomy for Soft Tissue Sarcoma
Isolated Regional Therapy for Advanced Extremity Soft Tissue Sarcomas
Sarcomas and the Immune System: Implications for Therapeutic Strategies
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
