Characterization of alveolar soft part sarcoma using a large national database

Abstract

Background:

Alveolar soft part sarcoma is a rare, histologic subtype of soft tissue sarcoma that remains poorly defined. We aimed to describe patient characteristics and treatment patterns and to examine factors associated with survival for patients with alveolar soft part sarcoma.

Methods:

After identifying patients with alveolar soft part sarcoma in the National Cancer Database, we recorded their clinicopathologic characteristics. Univariable log-rank survival analysis and Cox proportional hazards model were employed. For context, survival comparisons were included for patients with other sarcoma subtypes.

Results:

Overall, 293 patients with alveolar soft part sarcoma were identified. Interestingly, patients with head and neck tumors were least likely to present with distant disease (40%, P = .025). The majority of patients underwent resection (n = 183, 63%). Among those, no predictors of lesser survival were identified other than the presence of metastases (hazard ratio 6.04, P ≤ .001). Patients with stage IV alveolar soft part sarcoma who underwent resections experienced improved survival relative to similar patients with more common subtypes of soft tissue sarcomas (P ≤ .001).

Conclusion:

Alveolar soft part sarcoma is exceedingly rare, and patients often present with metastases. Primary tumors can occur anywhere in the body, and location impacts the rates of metastases at presentation. Resection is associated with a favorable survival advantage when compared to other, more common histologic subtypes of soft tissue sarcomas.

Introduction

Alveolar soft part sarcoma (ASPS) is a rare histologic subtype of soft tissue sarcoma representing less than 1% of all diagnosed soft tissue sarcomas.¹ First described in 1952, ASPS presents most commonly as a painless mass in young female patients.^1,2 Diagnosis can be challenging and is made by histopathologic analysis using a combination of morphology and immunohistochemistry.³ ASPS is further characterized by the frequent presence of the chromosomal rearrangement der(17)t(X;17)(p11;q25).^4,5 ASPS carries a high rate of distant metastasis, with a high propensity for brain metastases, but remains a relatively indolent disease.⁶

The literature on ASPS is rather robust for a rare disease and includes multiple, large, single-institution case series from referral centers for sarcomas.^6–9 A series of 102 patients collected over 63 years demonstrated a 5-year overall survival (OS) of 60% for those presenting with localized disease.⁸ In a more recent series of 70 patients, the 5-year OS of 80% in stage II/III patients was reported, with a median OS of 40 months even in patients with M1 disease.⁶ Resection remains the standard of care when feasible. The inactivity of traditional cytotoxic chemotherapy has led to investigation of targeted therapies. In prospective trials of the vascular endothelial growth factor receptor inhibitor cediranib, investigators observed an objective response rate of 35%¹⁰ and an improvement in tumor response over placebo, but there was no difference in progression-free survival or OS.¹¹ Furthermore, a recent single-arm, Phase II trial of axitinib and pembrolizumab for advanced sarcomas demonstrated particular activity in ASPS with 6 of 11 patients achieving partial response.¹²

Despite such series at individual referral centers, little is known about patient characteristics, prognosis, or treatment patterns on a national level. We aimed to further characterize ASPS using the National Cancer Database (NCDB) and to examine the factors associated with prognosis and survival for patients undergoing resection. To provide context, we compared the OS of patients with ASPS to those with more common histologic subtypes of soft tissue sarcomas.

Methods

Data source

We performed a retrospective cohort study using the Participant User Files (PUFs) of the NCDB. The NCDB is a joint program of the American College of Surgeons Commission on Cancer (CoC) and the American Cancer Society. The NCDB is a national oncology outcomes database for more than 1,500 Commission-accredited cancer programs in the United States. About 70% of all newly diagnosed cases of cancer in the United States are captured at the institutional level and reported to the NCDB. This database contains information on patient demographics, tumor characteristics, initial treatment, and outcomes. The CoC and the American Cancer Society have not verified and are not responsible for the analytic or statistical methodology used or for the conclusions drawn from these data. This study was deemed exempt from review by our Institutional Review Board.

Selection of the study population

The evaluated PUFs were composed of patients aged 18 and over who were diagnosed between 2004 and 2015. Cases were identified by searching data files of soft tissue and retroperitoneal tumors for the International Classification of Diseases for Oncology, Third Edition morphology code specific to ASPS (9581/3). Frequencies of sex, race, Charlson-Deyo score, primary site, stage, and sites of metastases were recorded in addition to median age at diagnosis. Patients who underwent resection of their primary tumor were identified for survival analysis. Cases with procedure codes indicating at least partial removal of the lesion were included in this operative cohort.

Patient and tumor characteristics

Stage was assigned based on the NCDB analytic stage variable which categorizes patients based on American Joint Committee on Cancer (AJCC) pathologic stage when available. Patients without documented metastases were grouped together as “localized disease.” Of note, the data set contained little information on tumor grade. To examine treatment patterns according to facility type, we broke down the cohort into those treated at high- and low-volume centers. High-volume centers were considered those in the upper third of total cases seen in the NCDB Soft Tissue PUF.¹³ For the purposes of multivariate analysis, certain variables were stratified into more generalized groups. Race was stratified according to white, African American, and other, which includes Asian, Asian Indian, and Pacific Islander populations. Margin status was divided into 2 categories (positive and negative) to include those cases that were coded as “Residual Tumor, NOS.” Tumor size was stratified into >7.4 cm vs ≤7.4 cm based on the mean tumor size of the operative cohort. Patients who underwent any systemic or radiation therapy as an adjunct to their operation were categorized as having undergone multimodal therapy. The extent of operation was stratified as “local” versus “radical” based on surgical codes for each case.

Survival analyses

Only patients who underwent a resection were included in the survival analysis. Patients with missing information on each variable of interest pertaining to treatment regimen, tumor characteristics, and relevant demographic information were excluded from that individual analysis. Univariate analysis was performed for OS. Multivariate analysis was done using covariates approaching statistical significance on univariate analysis (P < .1). Patients with missing data on any of the relevant variables were excluded from the multivariate analysis.

Comparison with other sarcoma histologic subtypes

Given the rarity of ASPS, we selected 4, common, histologic subtypes of soft tissue sarcomas, including those with similar age of presentation, to perform survival comparisons in order to better frame the outcomes of the study cohort.^14–16 Patients with metastatic disease who underwent an operation were selected for comparison, because this was the most robust ASPS cohort and prolonged survival with advanced stage disease had been described previously.⁶ The PUF for soft tissue tumors was searched for International Classification of Diseases for Oncology, Third Edition morphology codes for synovial sarcoma (9040/3–9044/3), liposarcoma (8850/3–8858/3), desmoplastic small round cell tumor (8806/3), and rhabdomyosarcoma (8900/3–8902/3). Patients with those pathologies who underwent resection for metastatic disease were then selected in order to compare similar populations.

Statistical analysis

Statistical analysis was done using SPSS software version 25.0 (IBM Corporation, Armonk, NY). Categorical variables were compared using a χ² or Fisher exact test. Univariable analysis for OS was performed using Kaplan-Meier analysis and log-rank tests. Multivariable survival analysis was done using a Cox regression model. Factors included in the model were those with P ≤ .1 on univariate analysis. A 2-tailed P value of <.05 was considered statistically significant. Continuous data are presented as medians with interquartile range (IQR).

Results

Patient and tumor characteristics

Overall, 293 patients with soft tissue and retroperitoneal ASPS were identified. The median age of the cohort was 27 years. Patients were divided evenly between males and females. The majority of patients were white (n = 152, 52%) and had a Charslon-Deyo score of 0 (n = 272, 92.8%). The median tumor size of the total cohort was 8.0 cm (IQR 5.0 cm, 11.0 cm) and over half of the tumors were located in the extremities (n = 175, 60%); of those, 152 were in the lower extremity (52%). Of the 173 patients with metastatic disease, nearly all (n = 169; 98%) had distant disease at presentation. The majority of patients underwent treatment at a high-volume sarcoma referral center (n = 244, 83%) (Table 1). Interestingly, metastatic disease was diagnosed in 73% of patients with primary disease of the extremities compared with 62% of those in the trunk, 43% in the retroperitoneum, and 40% in the head and neck. Compared with extremity-based ASPS, patients with head and neck ASPS had a lesser rate of metastases at diagnosis (P = .025) (Fig 1).

Table I.

Patient demographics and disease patterns (number of patients 293)

Variable	n (%)
Female sex	147 (50.2)
Median age (IQR)	27 (23–36)
Race and ethnicity
White	152 (51.9)
White, Hispanic	33 (11.3)
African American	76 (25.9)
Asian/Pacific Islander	18 (6.1)
NOS	14 (4.8)
Charlson-Deyo score
0	272 (92.8)
≥1	21 (7.2)
Median tumor size (IQR)	8.0 cm (5–11 cm)
Primary site
Upper extremity	23 (7.8)
Lower extremity	152 (51.9)
Trunk	80 (27.3)
Head and neck	10 (3.4)
Retroperitoneum	7 (2.4)
NOS	21 (7.2)
Stage
Localized (I–III)	83 (28.3)
Metastatic (IV)	172 (59.0)
NOS	37 (13.6)
Site of metastasis
Lung	64 (37.0)
Liver	7 (4.0)
Brain	13 (7.5)
Bone	16 (9.2)
NOS	73 (42.2)
Facility type
High volume	244 (83.3)
Low volume	49 (16.7)

NOS, not otherwise specified.

Fig 1.

Disease stage at presentation by primary site. *Indicates statistical significance. RP, retroperitoneum.

Treatment patterns

In total, 183 patients had their primary tumor at least partially resected, the majority of whom (n = 107; 58%) had localized disease. Most resections achieved a negative margin of their primary tumor (n = 128; 70%). Table II outlines treatment patterns according to facility type and stage. Patients with localized disease were more likely to undergo a resection at a high-volume center (n = 96, 92%) than at a low-volume center (n = 11, 73%) (P = .013). In the setting of metastatic disease, 47% of patients were selected for resection at high-volume centers as opposed to 32% at low-volume centers (P = .129). There were no differences between facility type with respect to extent of resection or perioperative treatment. Of the 76 patients with metastatic disease who underwent resection, 42 (55%) underwent perioperative radiation, while 31 (41%) received perioperative systemic therapy. Both modalities were given most commonly in the adjuvant setting.

Table II.

Treatment patterns at high- and low-volume centers by stage of disease

Variable (n, %)	High volume (n = 244)	Low volume (n = 49)	P
Localized
Total patients	104	15
Operative	96 (92.3)	11 (73)	.013
Nonoperative	8 (7.7)	4 (27)
Extent of operation			.137
Local/partial excision	39 (37.5)	8 (53)
Radical resection, limb salvage	52 (50.0)	3 (20)
Partial/total amputation	3 (2.8)	0
Resection + chemo	10 (9.6)	2 (13)	.334
adjuvant	4 (3.8)	1 (7)
Resection + XRT	56 (53.8)	6 (40)	.810
Adjuvant	40 (38.5)	5 (33)
Metastatic
Total patients	139	34
Operative	65 (46.8)	11 (32)	.129
Nonoperative	74 (53.2)	23 (68)
Extent of operation			.298
Local/partial excision	23 (16.5)	6 (17.8)
Radical resection, limb salvage	35 (25.1)	3 (8.3)
Partial/total amputation	4 (2.9)	1 (2.9)
Resection + chemo	27 (19.4)	4 (126)	.745
adjuvant	20 (14.4)
Resection + XRT	35 (25.1)	7 (2051)	.746
adjuvant	26 (18.7)	5 (14.7)

XRT, radiotherapy.

Bold indicates statistical significance.

Notably, selection for operative treatment in stage IV patients appeared to depend primarily on tumor size and site of disease. Median tumor size in those resected was 9.0 cm (IQR 5.5 cm,11.0 cm) as opposed to 10.0 cm (IQR 8.1 cm, 13.6 cm) those not resected (P = .003). Furthermore, 50% of those extremity ASPSs underwent resection compared with just 29% of those in the trunk (P =.014). Age, race, sex, insurance status, and facility type were not associated with resection of primary tumor in stage IV disease (all P ≥ .05).

Factors associated with survival

The 1-, 2-, and 5-year survival rates for patients undergoing resections for ASPS were 95%, 86%, and 73%, respectively. For patients undergoing resections for stage IV disease, the 1-, 2-, and 5-year survival rates were 89%, 76%, and 46%. Survival analyses did not reveal a difference between groups stratified according to sex, race, primary tumor site, or extent of operative resection. Positive resection margins (P = .03), the presence of metastatic disease (P < .001), primary tumor size greater than 7.4 cm (P = .002), the receipt of multimodal therapy (P = .005), and treatment at a low-volume center (P = .021) were associated with a lesser OS (Table III). On multivariate analysis, the only independent predictor of lesser OS was the presence of metastatic disease (P ≤ .001, hazard ratio 6.04) (Table III).

Table III.

Univariate and multivariate survival analysis for ASPS patients after resection

Variable (n, %)	Operative patients (n = 171)	Median OS (mon) (IQR)	95% CI	P univariate	HR	95% CI	P multivariate
Median age (IQR)	27 (23, 37)				0.98	0.95–1.0	.341
Sex				.067			.884
Male	79 (46.2)	75.4 (42, NR)	54–96		Ref	-
Female	92 (53.8)	127.4 (74, NR)	78–176		1.05	0.55–1.99
Race				.075
White	109 (63.7)	96.5 (48, NR)	71–122		Ref	-	-
AA	41 (23.9)	74.3 (30, NR)	60–89		0.858	0.41–1.80	.687
Other	21 (12.3)	NR	-		0.540	0.16–1.82	.321
Margin status				.003			.413
Negative	119 (69.6)	127.4 (69, NR)	-		Ref	-
Positive	34 (19.8)	66.4 (21, NR)	29–104		1.37	0.64–2.9
Stage				<.001			≤.001
Localized	99 (57.9)	NR	-		Ref	-
Metastatic	72 (42.1)	48.0 (22, 75)	28–68		6.04	2.79–13.0
Location				.894
Trunk	42 (24.6)	NR	-
Extremity	106 (62)	96.5 (48, NR)	66–127
H & N	10 (5.8)	77.7 (-)	14–141
RP	5 (2.9)	76.8 (-)	-
Tumor size (cm)				.007			.536
>7.4	62 (36.3)	76.8 (30, 127)	58–95		Ref	-
≤7.4	90 (52.6)	NR	-		1.23	0.64–2.35
Extent of surgery				.211
Local	70 (40.9)	77.7 (48, NR)	56–99
Radical	96 (56.1)	107.4 (66, NR)	69–146
Multimodal Rx				.008			.340
No	51 (29.8)	NR	-		Ref	-
Yes	110 (64.3)	75.4 (42, NR)	71–80		1.47	0.66–3.28
Facility type				.021			.659
High volume	150 (87.7)	127.4 (47, NR)	82–173		Ref	-
Low volume	21 (12.3)	67.2 (54, 87)	56–78		1.19	0.55–2.56

Patients with missing survival information are excluded. Bold indicates statistical significance.

AA, African American; H & N, head and neck; NR, not reached; RP, retroperitoneum.

Survival comparison with other histologic subtypes of soft tissue sarcomas

Patients with Stage IV ASPS who underwent a resection experienced prolonged OS as compared with an identically selected population of patients with more common histologic sarcoma subtypes. The median OS for patients undergoing resection for Stage IV soft tissue sarcomas was 48 months for ASPS (95% confidence interval [CI] 29–67), 11 months for rhabdomyosarcoma (95% CI 9–14), 28 months for desmoplastic small round cell tumor (95% CI 25–32), 21 months for synovial sarcoma (95% CI 18–24), and 18 months for liposarcoma (95% CI 14–22) (P ≤.001, Fig 2).

Fig 2.

Kaplan-Meier curves for patients with various soft tissue sarcoma histologic subtypes who underwent resection for stage IV disease (P ≤ .001). DSRCT, desmoplastic small round cell tumor.

Discussion

Evaluating patients with a rare cancer such as ASPS is difficult given the paucity of evidence to guide surgical decision-making. Although retrospective reviews of large national databases do not provide level 1 evidence, they do inform about treatment patterns and provide power to determine outcome associations with clinicopathologic parameters. Using the NCDB, we report on the largest cohort of patients with ASPS to date. In doing so, this study has identified several previously unreported results. Namely, the site of the primary tumor seems to influence the rates of metastases at presentation. We also report on a preponderance of ASPS treatment at high-volume sarcoma referral centers. Furthermore, although the only independent predictor of survival for patients with ASPS was the presence of metastatic disease, our cohort experienced prolonged survival after resection of stage IV disease as compared with more commonly encountered soft tissue sarcomas. Such findings have not been published from a heterogeneous, national cohort of ASPS patients treated in a variety of settings. These results provide rationale for an aggressive approach to the management of advanced disease when feasible and safe, especially considering the response rates to the systemic agents which could be reserved for disease recurrences.

We identified a lesser rate of distant metastases associated with a primary tumor site in the head and neck. This finding may be a result of increased visibility of early lesions leading to prompt presentation to medical attention and a lead-time bias. There may, however, be other contributing factors, such as tumor biology. In 2 retrospective reviews examining a total of 236 patients with head and neck sarcomas, the authors reported a combined total of only 52 distant recurrences after resection compared with 123 local recurrences, a finding that also argues against a lead-time bias theory and suggests that other factors may play a role.^17,18 Notably, in a study by Torosian et al of 565 patients with soft tissue sarcomas, 52/237 (18%) of patients with extremity soft tissue sarcomas had metastatic disease compared with 4/21 (19%) of those with head and neck tumors.¹⁹

The association between use of multimodal therapy and worse OS that we identified for patients with ASPS is more likely a reflection of the intrinsic behavior of those specific ASPSs than of the treatment itself. There is currently no data to support adjuvant or neoadjuvant standard chemotherapy for ASPS, leading to the assumption that most adjuvant/neoadjuvant treatment regimens are based on experience with more common soft tissue sarcomas, or on the response rates observed with targeted therapies. Despite considerable ambiguity of the literature regarding radiation therapy for soft tissue sarcomas, most patients received adjuvant rather than neoadjuvant therapy.^20–22 Given the limited sample of patients who received multimodal therapy in addition to resection, it is difficult to draw conclusions regarding the role of perioperative treatment. It is, however, clear that the only independent predictor of worse OS after resection is the presence of distant metastases. Nonetheless, there is a substantial survival advantage for patients with ASPS compared with well-matched patients with other subtypes of soft tissue sarcomas. These findings suggest that aggressive operative treatment of advanced ASPS should be considered in appropriate patients.

We have shown that ASPS is treated most commonly at high-volume referral centers. This observation is not surprising given the rarity of this pathology, unfamiliarity of most practitioners, and greater prevalence reported from large referral centers relative to the general population.^1,8 This finding may be exaggerated by misdiagnosis at low-volume centers unfamiliar with the histopathologic findings of ASPS. The differences in resection rates in both localized and metastatic disease may be explained either by referral to other centers not captured by the NCDB or differences in the comfort level of practitioners. The latter seems more likely, because resection occurring at any CoC site is captured by the database. Interestingly, no other major differences in treatment patterns were seen. We suspect this is because treatment of ASPS may default to that of other more common soft tissue sarcomas as noted previously. When controlling for other variables, facility type did not have an association with OS after resection. This conflicts with prior reports that have shown improvement in survival of soft tissue sarcomas treated at high-volume centers.^13,23

Within our cohort, we confirmed many of the findings put forth in previous single-institution studies. Namely, the majority of patients studied here were under the age of 30 at diagnosis with a primary tumor in the lower extremity and most often presented with metastatic disease. This finding is consistent with the most recently reported series, wherein almost 60% of patients were younger than 30, almost 40% had a lower extremity primary tumor, and 65% had stage IV disease at presentation.⁶ In addition, we note that 64% of patients with known sites of metastasis have disease in the lung. This site is in range of previously reported rates of 63% to 93%.^6,8,9 Finally, in our study, use of multimodal therapy did not lead to improved OS on univariable analysis and, in fact, was associated with a lesser OS. This is consistent with results reported by Lieberman et al where a cohort of 22 patients treated with adjuvant therapy had no advantage in recurrence-free or OS compared with the remaining patients who received no adjuvant therapy.⁸

Our study is not without limitations. These limitations include missing data that may skew our results. Specifically, staging information is absent frequently and sometimes ambiguous. Nodal staging was very infrequently documented in our patient population, and different versions of AJCC staging were used in the study period. We feel that this does not bias our results substantially, however, given that almost all stage IV patients had distant metastases and only 1 of the stage III patients had documented nodal disease. Additionally, information on disease progression or recurrence is lacking in the NCDB, which could help to inform decision-making and lend insight into tumor biology. Finally, information on resection is limited to interventions on the primary tumor. It is unclear whether patients undergo synchronous or metachronous metastasectomy, though this may be inferred from patients with stage IV disease undergoing R0 resection.

Lack of information on pediatric patients in the NCDB is another limitation, because ASPS is diagnosed commonly in children. The lack of pediatric patients in our study may indeed bias our results not only owing to differences in baseline health characteristics but also potential differences in biology between early and late-onset disease. Prolonged survival has been reported in association with early-onset disease⁸; this difference was attributed by the authors to the increased frequency of metastatic disease at presentation in the older age groups, because they noted a weaker trend within stage IV patients, which suggests that metastatic disease, not age of onset, is the primary driver of prognosis, a finding supported by this study.

One important note is the lack of information on tumor grade. To date, no histopathologic features predictive of prognosis have been reported, and it is not recommended currently that these tumors be graded at all.¹ Given that the AJCC staging guidelines are largely dependent on grade to separate stages I to III, it is unclear how stage was assigned to each case of localized disease. We attempted to mitigate this by grouping all localized disease together in our cohort. This grouping is supported by our finding of no difference in OS between patients with stage I, II, or III disease. The largest single-institution papers also group patients based on localized or metastatic disease.^6,8 While neither of these reports specifically mentions staging methods for localized ASPS, we posit that AJCC staging may not necessarily apply to ASPS, which perhaps should be stratified only according to presence or absence of distant disease.

In summary, ASPS is a rare tumor most often found in the extremities of young patients and commonly presents at an advanced stage. The cancer tends to follow a more indolent course and is associated with a greater survival after resection of advanced disease compared with more common histologic subtypes of soft tissue sarcomas. We assert that aggressive operative treatment should be pursued when feasible and safe, with systemic targeted therapies reserved for (1) recurrences during follow-up surveillance, (2) unresectable disease, and (3) in cases where tumor shrinkage may facilitate resection.