Annals of Surgical Oncology

, Volume 16, Issue 3, pp 667–675

Establishing Prognosis in Retroperitoneal Sarcoma: A New Histology-Based Paradigm

Authors

  • Daniel A. Anaya
    • Department of Surgical Oncology, Unit 444The University of Texas M. D. Anderson Cancer Center
  • Guy Lahat
    • Department of Surgical Oncology, Unit 444The University of Texas M. D. Anderson Cancer Center
  • Xuemei Wang
    • Department of Biostatistics and Applied MathematicsThe University of Texas M. D. Anderson Cancer Center
  • Lianchun Xiao
    • Department of Biostatistics and Applied MathematicsThe University of Texas M. D. Anderson Cancer Center
  • Daniel Tuvin
    • Department of Surgical Oncology, Unit 444The University of Texas M. D. Anderson Cancer Center
  • Peter W. Pisters
    • Department of Surgical Oncology, Unit 444The University of Texas M. D. Anderson Cancer Center
  • Dina C. Lev
    • Department of Cancer BiologyThe University of Texas M. D. Anderson Cancer Center
    • Department of Surgical Oncology, Unit 444The University of Texas M. D. Anderson Cancer Center
Bone and Soft Tissue Sarcomas

DOI: 10.1245/s10434-008-0250-2

Cite this article as:
Anaya, D.A., Lahat, G., Wang, X. et al. Ann Surg Oncol (2009) 16: 667. doi:10.1245/s10434-008-0250-2

Abstract

Background

Retroperitoneal sarcoma (RPS) American Joint Committee on Cancer (AJCC) staging applies to primary tumors only; due to specific RPS disease characteristics, staging is driven primarily by grade, stratifying patients into only two distinct prognostic subsets. The objective of this study was to help improve currently available staging for RPS by establishing a new, more robust histology-based prognostic system.

Methods

A RPS database of 1,118 patients seen at our institution (1996–2006) identified 343 patients treated for resectable primary or recurrent disease; a histologic subtype-based RPS prognostic system was designed and evaluated for prognostic accuracy in comparison with the current AJCC staging system.

Results

Histology stratified patients into three groups by prognosis (P < 0.0002): atypical lipomatous tumor (ALT), non-ALT liposarcoma (LPS), and “other,” an improvement compared with AJCC staging which could only identify two distinct prognostic groups. In contrast to AJCC staging, this prognostic stratification was reproducible for both primary and recurrent RPS (P < 0.0001). After multivariate analysis, LPS (P = 0.0004) and “other” histologies (P < 0.0001) were found to be independent predictors of worse survival. The concordance ratio of this model was 0.74, equivalent to that of the model using the AJCC staging system.

Conclusions

A histology-based RPS prognostic system has two advantages over AJCC staging: it can stratify into three versus two distinct prognostic groups, and it can be used for both primary and recurrent RPS. Distinct risk stratification is critical for specific assessment of prognosis as well as decisions regarding individualized adjuvant therapies, hence the advantage of a three-tiered histology-based system applicable in both primary and recurrent RPS.

Retroperitoneal sarcomas (RPS) constitute approximately 10–15% of all soft tissue sarcomas (STS) with an annual estimated incidence of 1,000 new cases in the USA.1,2 The only curative treatment for RPS is surgical resection. However, despite resections with curative intent, local recurrence is common (approximately 70% 5-year local recurrence) and accounts for close to 90% of disease-related mortality with 5-year overall survival (OS) rates of approximately 50%.35 In contrast to patients with extremity STS, these relatively poor outcomes have remained stagnant over the last few decades despite the introduction of more effective chemotherapeutic agents and safer radiation therapy techniques.6 The rarity of RPS, its remarkable array of histologic subtypes and tumor heterogeneity, and the typically small and retrospective published series have limited the risk stratification of these patients, preventing adequate interinstitutional comparisons and design of risk-stratified treatment strategies.

The sixth edition of the American Joint Committee on Cancer (AJCC) staging system for STS is currently the most widely used such system for RPS.7 It classifies patients into stages I, II, III, and IV, based on tumor size, depth, grade, lymph node involvement, and distant metastatic disease, with higher mortality rates accompanying higher stages. Although this is a well-designed system for patients with STS, it is not representative and less useful in patients with RPS, particularly those with nonmetastatic disease. The AJCC staging system was derived from a large single-institution experience of patients with STS and, although it has been externally validated with data from other institutions, the initial cohort did not include patients with RPS nor has it been validated for this specific subset of patients.8,9 Further, given the high local recurrence rate in patients with RPS it becomes extremely important to be able to predict survival in the subgroup of patients presenting with recurrent disease, an additional feature not provided by the AJCC staging system.

Furthermore, specific disease characteristics in patients with RPS practically eliminate the role of some variables included in the AJCC staging system.7 Specifically, depth, which was designed to classify extremity STS based on penetration beyond the deep muscular fascia, is irrelevant in patients with RPS, all of which are considered deep by definition. A similar effect is seen with tumor size. Primary RPS only rarely present as tumors measuring less than 5 cm, limiting the applicability of the AJCC system variable size, which distinguishes between tumors measuring less than 5 cm and those measuring ≥ 5 cm. This leaves grade as the only discriminatory and applicable variable for AJCC staging of patients with nonmetastatic primary RPS. As per the AJCC guidelines and regardless of the grading system used (three- versus four-tiered grading system), grade stratifies patients into essentially two distinct groups: those with low-grade tumors and those with intermediate- and high-grade tumors. Therefore, when recording AJCC variables in nonmetastatic primary RPS, the vast majority of patients are classified only into two distinct risk categories: stage I (low-grade tumors regardless of size) and stage III (intermediate- or high-grade tumors measuring ≥ 5 cm).

Although RPS include a large number of different histologies and histologic subtypes, previous studies have shown that liposarcomas (LPS) account for the majority of RPS.4,1015 Similarly, based on our work and work from other institutions, it is now commonly accepted that well-differentiated liposarcoma, otherwise known as atypical lipomatous tumor (ALT), behaves and exhibits different outcomes compared with non-ALT LPS.1622 Other non-LPS histologies are less common so establishing patterns of clinical behavior and/or outcome for each histologic subtype has been more challenging. Based on those considerations, we conducted a retrospective analysis of our experience with nonmetastatic resectable primary and recurrent RPS, with the primary objective of improving currently available staging for RPS by establishing a more robust histology-based prognostic system.

Methods

Patient Selection

This study was approved by the institutional review board of the University of Texas M.D. Anderson Cancer Center (MDACC). We identified 1,118 patients from a prospectively collected database of patients evaluated for retroperitoneal sarcoma at MDACC between 1996 and 2006, exclusive of desmoid tumors (low-grade fibrosarcoma) or gastrointestinal stromal tumors (GIST) patients. From this group we selected patients treated with surgery at MDACC and excluded those with metastatic or unresectable disease, yielding 343 patients with primary or recurrent RPS, who constituted the cohort of this study. Medical records were reviewed and demographic and clinicopathologic characteristics and treatment and outcome variables were retrieved. All 343 patients were evaluated by a multidisciplinary group and treated surgically with curative intent. Adjuvant therapies, including chemotherapy (primarily doxorubicin based) and radiation therapy, were employed selectively as deemed appropriate by the treating physicians. Patients were followed every 3–4 months for the first 5 years after resection and annually thereafter.

Variable Definition and Risk Stratification

Patients were classified by stage using the AJCC staging system (6th edition).7 This system uses tumor size, depth, and grade to classify patients with primary nonmetastatic sarcoma into stages I, II or III. Size was defined by the postresection measurement of the surgical specimen (in centimeters) as determined during gross pathological evaluation. When incomplete resection was performed, preoperative cross-sectional imaging together with intraoperative findings were used to assess the size of the tumor. Retroperitoneal sarcomas are considered deep by definition and were interpreted as such for staging purposes. Grade was determined using the French Federation of Cancer Centers Sarcoma Group Grading System (FNCLCC).2326 This is a three-tiered system that uses histologic type/subtype, extent of necrosis, and mitotic figures to determine grade. Grade was recorded as low, intermediate or high, and the latter two categories were grouped together, in keeping with the AJCC sarcoma staging guidelines. Using these variables, patients with primary tumors were classified as stage I, II or III. Patients with stage I disease included all those with low-grade tumors, regardless of size. Patients with stage II disease were those with intermediate- or high-grade tumors measuring less than 5 cm and stage III disease patients were those with intermediate- or high-grade tumors measuring more than 5 cm. Patients with recurrent disease were defined as those presenting with tumors at least 3 months after surgical resection; AJCC staging was not applicable to this subset.

Histologic type was recorded for all patients. Table 1 summarizes the histologies identified in this cohort as classified by histologic risk group and their distribution among the whole cohort. This group categorization was used based on our own experience as well as that of others and is consistent with the high prevalence of liposarcoma among RPS as well as the differing behavior and clinical outcomes among these different histological subtypes.1622
Table 1

Histological subtypes included in each risk group and their distribution among patients with retroperitoneal sarcoma (n = 343)

Histologic group and subtypes

N

%

Atypical lipomatous tumor (ALT) 16,17

57

17

    Conventional ALT

    Cellular ALT

Non-ALT liposarcoma (LPS)16,17

90

26

    Dedifferentiated LPS

    Pleomorphic LPS

Other

196

57

    Leiomyosarcoma

62

18

    Unclassified sarcoma

57

17

    Malignant fibrous histiocytoma

43

12

    Myxoid LPS

7

2

    Malignant hemangiopericytoma/angiosarcoma

6

1.7

    Malignant peripheral nerve sheath tumor

5

1.5

    Primitive neuroectodermal tumor/extraskeletal Ewing’s sarcoma

4

1.2

    Desmoplastic small round cell tumor

4

1.2

    Rhabdomyosarcoma

3

0.9

    Synovial sarcoma

2

0.6

    Fibrosarcoma (nondesmoid)

2

0.6

    Extraskeletal osteosarcoma

1

0.3

Statistical Analysis

Overall survival was the primary outcome of interest and was calculated from date of diagnosis of the current disease (primary or recurrent) until date of death. Patients alive at the time of the analysis were censored using the date of last follow-up. Kaplan–Meier method was used to estimate overall survival in all patients. Log-rank test was used to evaluate differences between the risk categories within each staging system (AJCC and histology grouping). Univariate and multivariate regression analysis were performed to identify independent predictors of worse survival. Two models were used: one including the AJCC stages and another including the histology groups. The concordance ratio between predicted and actual overall survival was calculated for each of these models. A P value of ≤0.05 was considered statistically significant. All computations were performed using Splus 7.0 (Insightful Corporation, Seattle, WA, USA) and SAS 9.1 (SAS Institute Inc. Cary, NC, USA).

Results

Descriptive Analysis

A total of 343 patients were included in the analysis. Demographic, clinicopathologic, and treatment variables are listed in Table 2. All patients were at least 18 years of age with a mean age of 56.1 years. Almost one-third of patients (30%) had recurrent tumors, and nearly 40% of these patients presented either with or after a second recurrence. Mean tumor size was 15.7 cm; only 7% of patients had tumors smaller than 5 cm and nearly half (46%) had tumors greater than 15 cm in largest dimension. Liposarcoma was the most common histology (43%), followed by leiomyosarcoma (18%), unclassified sarcoma (17%), and malignant fibrous histiocytoma (12%); the vast majority of tumors were classified as intermediate or high grade (80%). Gross total resection was achieved in 308 patients (90%) and associated organ resections were performed in 69%. Adjuvant chemotherapy or radiation therapy was employed in 45% and 32% of patients, respectively.
Table 2

Demographic, clinicopathological, and treatment characteristics of 343 patients with retroperitoneal sarcoma

Variable

Category

Number of patients (%)

Age (years)

≥65

95 (28%)

<65

248 (72%)

Gender

Female

163 (47%)

Male

180 (53%)

Presentation

Primary

241 (70%)

Recurrent

102 (30%)

Number of recurrences

0

241 (70%)

1

60 (18%)

>1

42 (12%)

Location

Pelvis

60 (17%)

Abdomen

123 (36%)

Pelvis + abdomen

160 (47%)

Size (cm)

0–5

23 (7%)

5–10

86 (25%)

10–15

77 (22%)

>15

157 (46%)

Multifocal disease

No

279 (81%)

Yes

64 (19%)

Histology

ALT

57 (17%)

Non-ALT LPS

90 (26%)

Other

196 (57%)

Grade

Low grade

69 (20%)

Intermediate grade

25 (7%)

High grade

249 (73%)

Treatment variables

Chemotherapy

No

188 (55%)

Yes

155 (45%)

Radiation therapy

No

232 (68%)

Yes

111 (32%)

Gross total resection

No

35 (10%)

Yes

308 (90%)

Number of organs resected

0

107 (31%)

1

102 (30%)

>1

134 (39%)

ALT atypical lipomatous tumor; LPS liposarcoma

Median follow-up for the whole cohort was 50 months. Median overall survival was 59 months [95% confidence interval (CI), 48–75 months] and 3- and 5-year survival rates were 64% (95% CI, 59–70%) and 50% (95% CI, 44–57%), respectively.

Patient Distribution in AJCC and Histology Prognostic Systems

Considering only patients with primary tumors (n = 241) the AJCC staging system established that the majority of these patients (75%) presented with stage III disease. Stage I patients were also relatively common (21%), whereas only 4% of patients were classified as having stage II disease. This finding was driven by the small number of patients with intermediate- or high-grade tumors measuring less than 5 cm (n = 9). When using the histology grouping system, a more even distribution among the three categories was demonstrable, with the majority of tumors having “other” histology (61%) and ALT and non-ALT LPS being evenly distributed (18% and 21%, respectively). Similarly, when evaluating the histology distribution in recurrent disease patients, although non-ALT LPS was more prevalent, approximately even distribution among the three histology prognostic groups was observed (Table 3).
Table 3

Distribution of patients according to AJCC staging and histology group systems, and 5-year overall survival for each category

Staging system

No. of patients

% of patients

5-year OS (95% CI)

Primary tumors

AJCC staging system (6th edition)

    Stage I

51

21

96% (90–100%)

    Stage II

9

4

58% (30–100%)

    Stage III

181

75

42% (34–52%)

Histology system (current study)

    Atypical lipomatous tumor (ALT)

43

18

95% (88–100%)

    Non-ALT liposarcoma

50

21

40% (26–59%)

    Other

148

61

48% (39–59%)

Recurrent tumors

AJCC staging system (6th edition)

Not applicable

Histology system (current study)

    Atypical lipomatous tumor (ALT)

14

14

86% (63–100%)

    Non-ALT liposarcoma

40

39

25% (12–54%)

    Other

48

47

43% (30–63%)

OS overall survival; CI confidence interval

Survival Analysis in Patients with Primary RPS

A total of 94 patients with primary RPS had died by the time of this analysis. Figure 1a illustrates the survival curves for patients with primary RPS according to AJCC staging. Median overall survival for patients with stage I disease was not reached at the time of this analysis and 5-year overall survival was 96% (95% CI, 90–100%). Survival times for patients with stage II and III disease were significantly lower with median OS times of 67 months (95% CI, 53 months to not reached) and 47 months (95% CI, 41–61 months), respectively. Three- and 5-year OS rates were 78% (95% CI, 55–100%) and 58% (95% CI, 30–100%) for stage II patients and 62% (95% CI, 54–70%) and 42% (95% CI, 34–52%) for stage III patients, respectively. Although the difference in survival between the three groups was found to be statistically significant (P < 0.0001), survival times between patients with stage II and III overlapped (Fig. 1a). Furthermore, due to the small number of patients with stage II disease (n = 9), interpretation of these survival times is problematic. These statistical limitations were addressed by grouping patients with stage II and III into one risk category that can better depict outcomes for this cohort of patients. Median, 3-, and 5-year OS for this combined group were 51 months (95% CI, 44–64 months), 62% (95% CI, 55–71%), and 43% (95% CI, 35–53%), respectively. Figure 1b highlights the difference in survival between patients with stage I versus those with stage II and III combined (< 0.0001).
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-008-0250-2/MediaObjects/10434_2008_250_Fig1_HTML.gif
Fig. 1

Overall survival in patients with primary retroperitoneal sarcoma (n = 241) stratified by (a) AJCC staging system (6th edition) (b) AJCC staging system (6th edition) with stage II and III patients combined, and (c) histology grouping. ALT, atypical lipomatous tumor; LPS, liposarcoma

Overall survival curves for the three histology groups are shown in Fig. 1c. Median overall survival has not been reached for patients with ALT histology and 5-year overall survival rate was 95% (95% CI, 88–100%). For patients with “other” histology median, 3-, and 5-year OS were 57 months (95% CI, 44 months to not reached), 64% (95% CI, 56–73%), and 48% (95% CI, 39–59%), respectively. Although survival times were close in patients with “other” and non-ALT LPS histology, the latter group experienced lower median OS time (47 months; 95% CI, 41–70 months), and lower 5-year OS rate (40%; 95% CI, 26–59%). As shown on Fig. 1c, survival times were initially close between these two groups but clearly diverge later, resulting in differences in median overall survival and long-term survival. The difference in survival between these groups was statistically significant (P = 0.0002).

Due to the close relation between ALT histology with stage I, and non-ALT LPS and “other” histology with stages II and III combined, two separate multivariate models were created to identify independent predictors of outcome using each of these prognostic systems. In the AJCC model, age ≥ 65 years, size ≥ 15 cm, multifocal disease, and incomplete resection together with the group including stages II and III combined were all independent predictors of worse survival (Table 4). Patients with stage II or III had almost three times higher risk of dying [hazard ratio (HR) 2.8; 95%CI, 1.8–4.5] as compared with those with stage I disease. For the histology model, the same independent predictors of overall survival were identified; however both non-ALT LPS and “other” histology groups were each found to be independent predictors of worse survival. Those with non-ALT LPS had a sixfold risk (HR 6.1; 95% CI, 2.1–17.8) of dying as compared with patients with ALT histology; patients with “other’ histology had even worse prognosis with a HR 10.3 (95% CI, 3.6–29.2). The concordance ratio was 0.74 for each of these models, emphasizing equivalent performance despite different variables.
Table 4

Multivariate analysis of overall survival in patients with primary retroperitoneal sarcoma using the AJCC and histology models

Variable

AJCC model

Histology model

Hazard ratio (95% CI)

P

Hazard ratio (95% CI)

P

Age ≥65 years

1.6 (1.3–2)

<0.0001

2.7 (1.8–4.3)

<0.0001

Size ≥15 cm

1.9 (1.3–2.9)

0.002

2.3 (1.5–3.6)

0.0002

Multifocal disease

1.5 (1.1–1.9)

0.006

2.6 (1.5–4.6)

0.0008

Complete resection

0.7 (0.5–1)

0.04

0.5 (0.3–0.9)

0.02

AJCC stage II/III

2.8 (1.8–4.5)

<0.0001

 

Non-ALT LPS versus ALT

 

6.1 (2.1–17.8)

0.0009

Other histology versus ALT

 

10.3 (3.6–29.2)

<0.0001

ALT atypical lipomatous tumor; LPS liposarcoma

Survival Analysis in Patients with Primary and Recurrent RPS

A similar analysis using the histology grouping was conducted for the whole cohort of patients (n = 343) which included primary and recurrent RPS. Figure 2 illustrates the survival curves for the three histology risk groups; the differences in survival were statistically significant (P < 0.0001). Univariate and multivariate analysis was performed, and the same independent predictors of survival were identified for this cohort as was demonstrated for the primary RPS subset (Table 5). Additionally, non-ALT LPS and “other” histology were each independent predictors of survival in this cohort as well with an association similar to that seen in patients with primary tumors only (HR 6.1; 95% CI, 2.6–14.3 and 7.5; 95% CI, 3.2–17.5, respectively). The concordance ratio for this model was 0.74, suggesting equivalent performance for patients with primary tumors and for those with primary and recurrent tumors combined.
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-008-0250-2/MediaObjects/10434_2008_250_Fig2_HTML.gif
Fig. 2

Overall survival in patients with primary and recurrent retroperitoneal sarcoma (n = 343) stratified by histology grouping. ALT, atypical lipomatous tumor; LPS, liposarcoma

Table 5

Multivariate analysis of overall survival in patients with primary and recurrent retroperitoneal sarcoma including histology as a variable

Variables

Hazard ratio

95% Confidence interval

P value

Age ≥ 65 years

2.3

1.6–3.4

 <0.0001

Size ≥ 15 cm

1.8

1.2–2.5

0.002

Multifocal disease

2.3

1.6–3.5

 <0.0001

Complete resection

0.4

0.3–0.7

0.0006

Non-ALT LPS versus ALT

6.1

2.6–14.3

0.0004

Other histology versus ALT

7.5

3.2–17.5

 <0.0001

ALT atypical lipomatous tumor; LPS liposarcoma

Discussion

In this study we have established a histology-based prognostic system to predict overall survival in patients with primary and recurrent RPS. This system stratifies patients into three risk categories according to tumor histology, with ALT having the best prognosis, “other” histology (non-LPS) having an intermediate prognosis, and non-ALT LPS histology having the worst overall survival (P < 0.0001). The distribution of patients among these three categories was relatively even, rendering this system representative of patients presenting with RPS. Multivariate analysis demonstrated that both non-ALT LPS (HR 6.1; 95% CI, 2.6–14.3; = 0.0004) and “other” histology (HR 7.5; 95% CI 3.2–17.5; < 0.0001) were independent predictors of OS; these results were reproducible for patients with primary tumors as well as those with primary and recurrent tumors combined. The accuracy of this multivariate model, as measured by concordance ratio, was equivalent to that of the AJCC staging system. These findings constitute an improvement over current AJCC staging for RPS, in which only primary patients can be staged, and only two distinct groups are identified.

Histological subtype was used for risk stratification based on the distribution of histologies among RPS patients and the different outcomes observed in comparing these different histologic subgroups. One of the main drawbacks of the current AJCC staging system is that it is derived from patients with extremity STS, hence the variable size has almost no role in staging patients with RPS and patients with stage II disease are therefore extremely rare, limiting the discriminative utility of this risk category.8,9 This has been well described in previous studies in which patients with RPS had tumors measuring <5 cm in fewer than 6% of cases, which corroborates well with our results (7%).4,10,1315,27 In contrast, LPS is the most common histology in RPS patients; it is commonly accepted that within this larger histologic group there are at least two different histologic subtypes that have different clinical characteristics, resulting in markedly different post-treatment outcomes.1015,18

Histologic subgroup classification allowed stratifying patients into three different risk categories. Further risk stratification can be determined within each of these groups. This is particularly applicable to “other” histology RPS where the outcome can vary widely depending on specific histopathological (histology or grade, for example) or clinical features (primary versus recurrent tumors, for example); our group is currently working on clarifying the specific prognostic factors for this group of patients. The influence of these additional predictors on OS is exemplified by our own results in which patients with non-ALT LPS appeared to fair worse than patients with “other” histology, as per univariate analysis. However, and in contrast to this finding, after multivariate analysis, although both histology groups were each found to be independent predictors of worse OS, patients with “other” histology had a higher hazard ratio of dying than those with non-ALT LPS. After evaluating the demographic, clinicopathologic, and treatment features of the different histologic risk groups, it became apparent that non-ALT LPS seemingly do worse when no other factors are evaluated because these tumors are more commonly associated with other independent predictors of OS as compared with those with other histology (Table 6). When adjusting for these important predictors, “other” histology per se has a greater impact on decreased survival than non-ALT LPS histology. Similarly, we also performed univariate and multivariate analysis for OS in each of the histology risk groups (results not shown), demonstrating that in ALT and non-ALT LPS patients the specific histologic subtype appeared to be the dominant feature driving prognosis, whereas in patients with “other” histology additional independent predictors (including histologic type and tumor grade) helped in further risk stratification for this group of patients.
Table 6

Demographic, clinicopathologic and treatment variables by histologic risk-group in patients with retroperitoneal sarcoma (n = 343)

Variables

ALT (n = 57)

Non-ALT LPS (n = 90)

Other (n = 196)

P

Age ≥65 years

17 (29%)

38 (42%)

40 (20%)

 <0.0001

Male gender

28 (49%)

58 (64%)

94 (48%)

0.002

Recurrent tumors

14 (25%)

40 (39%)

48 (25%)

 <0.0001

Size ≥15 cm

42 (74%)

52 (58%)

63 (32%)

 <0.0001

Multifocal disease

5 (9%)

26 (29%)

33 (17%)

 <0.0001

Adjuvant chemotherapy

1 (2%)

34 (38%)

120 (61%)

 <0.0001

Adjuvant radiation therapy

9 (16%)

19 (21%)

83 (42%)

 <0.0001

Gross total resection

52 (91%)

78 (87%)

178 (91%)

0.02

ALT atypical lipomatous tumor; LPS liposarcoma

An additional major concern is that, while local recurrence is the most common pattern of treatment failure in patients with RPS, there are currently no staging systems to predict overall survival in patients presenting with recurrent RPS. However, it is well known that patients with recurrent disease exhibit worse outcomes and that number of recurrences is associated with a higher rate of incomplete resections.10 Although we have observed this same pattern in patients with recurrent RPS treated at MDACC, the multivariate analysis of our series did not identify recurrent presentation as an independent predictor of outcome. This may be explained by the inclusion of all known and new prognostic factors in our analysis, and their association with recurrent tumors. Specifically, factors such as incomplete resection and multifocality are both features intimately related to RPS recurrence and found to be independently associated with worse OS.28

The superiority of the histology system over the current AJCC staging also relates to the limitations of light-microscopy tumor grading in RPS. From a practical standpoint, when using grade, one can reliably discriminate between low grade and intermediate/high grade as two distinct groups, limiting the identification of intermediate-risk patients. Similarly, the role of grade as a predictor of OS in RPS may not be as robust as in extremity STS, where high-grade tumors are associated with a higher risk of metastasis, the most common cause of disease-related mortality.8,2931 This is in contrast to patients with RPS, in whom mortality is primarily associated with locally advanced disease concerns.10,12,15 Additionally, tumor grade has also been advocated as useful in predicting potential response to chemotherapy, which is more commonly indicated in high grade tumors.32 However, histology is even more helpful in this regard by identifying specific histologic groups that either exhibit greater overall chemoresponsiveness (e.g., extraskeletal osteosarcoma, Ewing’s sarcoma, rhabdomyosarcoma or synovial sarcoma) or other high-grade histologies that have shown less meaningful chemoresponsiveness (e.g., non-ALT LPS).32 These major drawbacks are exacerbated by the reality that high-grade assignments are less accurate when interobserver discordance rates are demonstrable even among experienced sarcoma pathologists.3335 Accuracy of grade assignment is also limited when using tissues derived from fine-needle aspiration or core-needle biopsies or when scoring grade in the presence of presurgical treatment effects.26,36

The results in this study may be limited by its retrospective nature. However, well-defined variables that included all known prognostic factors coupled to strict surveillance and long follow-up may overcome some of these limitations. This histology-based system broadly classifies patients into three different prognostic groups; however, additional variables such as molecular markers can be incorporated within each well-defined risk category while maintaining the main histologic-based stratification of this new staging system.

In summary, we established a histologic-based prognostic system for predicting OS in patients with RPS. This system is applicable to all RPS patients, whether presenting with primary or recurrent disease. This change in RPS is an improvement over AJCC staging, which does not account for significant RPS disease characteristics. This histology-based RPS prognostic system is practical, easily reproducible, and constitutes a more robust baseline framework within which to compare patient characteristics, treatment strategies, and outcomes within and between institutions, and can also be adapted in the future for inclusion of novel molecular markers of prognosis or when developing new RPS-specific staging systems.

Acknowledgements

This study was partially supported by an AJCC grant awarded to Raphael E. Pollock towards the soft tissue sarcoma staging.

Copyright information

© Society of Surgical Oncology 2008