Prognostic Value of Relevant Clinicopathologic Variables in Epithelioid Sarcoma: A Multi-Institutional Retrospective Study of 44 Patients

Background Epithelioid sarcoma (ES) is an extremely rare soft tissue sarcoma. Recently, the proximal variant has been reported to be a more aggressive subtype; however, as most reports of ES have involved small case series, the actual prognostic implications remain unclear. We investigated the clinicopathological features of patients with ES to identify the prognostic factors that influence survival. Methods We retrospectively analyzed the clinicopathological features of 44 patients with ES who had been treated at our institutions between 1991 and 2011. Among these patients, 26 were diagnosed histologically as having classic-type ES, whereas the remaining 18 had proximal-type ES. Thirty-three of the patients, all without distant metastases, underwent curative surgery, and the remaining 11 with distant metastases (M1) received palliative treatment. Results The proximal subtype was significantly correlated with a proximal tumor location, distant metastases at presentation, presence of rhabdoid cells, a higher tumor grade, and vascular invasion. The overall survival (OS) rate at 5 years for the 44 patients was 45 %. A superficial tumor location and lymph node metastases (N1) at presentation were independently predictive of local recurrence-free survival (LRFS), and N1 and M1 tumors were independently predictive of distant metastasis-free survival and OS, respectively. The proximal subtype was associated with unfavorable LRFS and OS, although not to a statistically significant degree. Conclusions Proximal-type ES has significantly more aggressive clinicopathological features than classic-type ES, and lymph node or distant metastasis has the most critical impact on prognosis.

ABSTRACT Background. Epithelioid sarcoma (ES) is an extremely rare soft tissue sarcoma. Recently, the proximal variant has been reported to be a more aggressive subtype; however, as most reports of ES have involved small case series, the actual prognostic implications remain unclear. We investigated the clinicopathological features of patients with ES to identify the prognostic factors that influence survival. Methods. We retrospectively analyzed the clinicopathological features of 44 patients with ES who had been treated at our institutions between 1991 and 2011. Among these patients, 26 were diagnosed histologically as having classic-type ES, whereas the remaining 18 had proximaltype ES. Thirty-three of the patients, all without distant metastases, underwent curative surgery, and the remaining 11 with distant metastases (M1) received palliative treatment.
Results. The proximal subtype was significantly correlated with a proximal tumor location, distant metastases at presentation, presence of rhabdoid cells, a higher tumor grade, and vascular invasion. The overall survival (OS) rate at 5 years for the 44 patients was 45 %. A superficial tumor location and lymph node metastases (N1) at presentation were independently predictive of local recurrence-free survival (LRFS), and N1 and M1 tumors were independently predictive of distant metastasis-free survival and OS, respectively. The proximal subtype was associated with unfavorable LRFS and OS, although not to a statistically significant degree. Conclusions. Proximal-type ES has significantly more aggressive clinicopathological features than classic-type ES, and lymph node or distant metastasis has the most critical impact on prognosis.
Epithelioid sarcoma (ES) is an uncommon malignant soft tissue tumor comprising approximately 1 % of all soft tissue sarcomas. It is a slow-growing tumor occurring mainly in young adult males, predominantly affecting subcutaneous tissues, fasciae, or tendon sheaths of the extremities, with a predilection for the hands and forearms. [1][2][3] Histologically, the classic type shows a distinctive nodular, granuloma-like pattern, with spindle and epithelioid cells circumscribing areas of central degeneration and necrosis. 4,5 However, a subtype of ES, the proximal type, occurs mostly in deep-seated soft tissues in the truncal region and is characterized histologically by sheets of large, atypical, epithelioid cells with vesicular nuclei and prominent nucleoli, showing a rhabdoid phenotype. 6,7 Unfavorable prognostic factors for overall survival (OS) have been reported to include a proximal location. 2,8,9 Recently, several authors have reported that the proximal subtype has a poor prognosis. 6,7,[9][10][11] However, as most previous reports have involved only small case series, especially with regard to proximal-type ES, the true prognostic impact of the ES subtype remains unclear. 9,12 In the present study, we therefore investigated the prognostic value of relevant clinicopathologic variables in 44 patients with ES.

Patients
We reviewed a prospective database for four institutions (Higashi-nihon Orthopaedic and Pediatric Sarcoma Group; HOPES) covering the period from 1991 to 2011. A total of 44 patients with a confirmed diagnosis of ES made by a specialized pathologist (AY) were analyzed. The median follow-up period was 26.5 (range 1-168) months overall. All patients or their guardians provided informed consent, according to the rules approved by the respective Institutional Review Boards.

Diagnostic Criteria for Each Histological Subtype
The diagnostic criteria for ES were the same as those documented previously. 6 Histologically, classic-type ES commonly presents as a multinodular proliferation of eosinophilic, epithelioid, and spindle-shaped cells. Usually, the lesion shows minimal cytologic atypia, with vesicular nuclei and often single and central nucleoli. Occasionally, the epithelioid cells may have a rhabdoid appearance, and tumor nodules may frequently undergo central necrosis, resulting in a pattern resembling a benign necrobiotic granulomatous process. Proximal-type ES is characterized by a predominantly large epithelioid cytomorphology, marked cytologic atypia, frequent occurrence of rhabdoid features, and lack of a granuloma-like pattern in most cases. ES with hybrid features of both classic and proximal-type ES is grouped as proximal-type ES.

Study Parameters
At the time of diagnosis, local tumor extent was assessed using computed tomography (CT) and/or magnetic resonance imaging (MRI). Lymph node and/or distant metastasis before treatment was assessed using enhanced CT and/or whole-body positron emission tomography (PET)-CT. Tumor location in the distal extremity was defined as 'distal', and localization in the proximal extremity or axial trunk was defined as 'proximal'. Disease stage was classified using the American Joint Committee on Cancer/International Union Against Cancer TNM staging system version 7 (AJCC stage). 13 Histological grade was assessed using the French Federation of Cancer Centers (FNCLCC) grading system. 14

Treatment
Thirty-three patients without distant metastasis at presentation (M0) underwent surgical treatment of the primary tumor with curative intent, which comprised wide resection in 14 patients, additional wide resection in 9 patients, and amputation in 10 patients. Regional lymphadenectomy was performed in 12 patients who had regional lymph node swelling. Six of these 12 patients were found to have regional lymph node metastases, and the remaining 6 were found to be negative after histological examination. The surgical margins were microscopically negative in 24 (73 %) patients, positive in 8 (24 %) patients, and not available (NA) in one FNCLCC French Federation of Cancer Centers patient. Adjuvant therapy included preoperative radiotherapy in 2 patients (6 %), postoperative radiotherapy in 4 (12 %) patients, pre-/postoperative systemic anthracyclinebased chemotherapy in 1 patients (3 %), and postoperative chemotherapy in 2 patients (6 %). Eleven patients (25 %) with distant metastasis at presentation (M1) received palliative treatment with various types of chemotherapy consisting of doxorubicin and/or ifosfamide, radiotherapy, and/or surgery.

Statistical Analysis
In order to compare the differences in clinicopathological features between classic-and proximal-type ES, the v 2 test was employed. Local recurrence-free survival (LRFS), distant metastasis-free survival (DMFS), and OS were calculated from the clinical databases using the Kaplan-Meier method. Patients with M0 tumors were evaluated for LRFS, DMFS, and OS from the date of definitive surgery  Cox regression analyses were performed to identify prognostic factors that were significant. Prognostic factors with statistical significance (p \ 0.05) in the univariate analysis were included in the multivariate analysis. Statistical analysis was performed using the PASW statistics 18 package (SPSS Inc., Chicago, IL, USA).

Local Recurrence
Among the 33 patients with M0 tumors who underwent curative surgery, local recurrences were noted in 10 (30 %). The LRFS rates at 2 and 5 years were 82 and 62 %, respectively (Fig. 1a). The median interval from surgery until the first local recurrence was 25 (range 4-165) months. Three of these ten patients developed both local recurrence and distant metastasis simultaneously.
Univariate analysis showed that a proximal tumor location (p = 0.020), superficial localization (p = 0.026), lymph node metastasis at presentation (N1) (p = 0.005), and a positive surgical margin (p = 0.036) were all correlated with LRFS. There was a tendency (p = 0.052) for an increased incidence of local recurrence in proximal-type ES relative to classic-type ES. Multivariate analysis showed that tumor depth (p = 0.040) and lymph node metastasis (p = 0.013) remained independent prognostic factors for LRFS (Table 2).

Distant Metastasis
Distant metastases occurred in 32 (73 %) patients. The sites of metastasis at initial presentation included distant lymph nodes in 13 patients, the lung in 12 patients, lymph nodes and lungs in 4 patients, lung and muscle in 1 patient, the lung, liver and scalp in 1 patient, and bone in 1 patient.
Twenty-one (64 %) of the 33 patients with M0 tumors developed metastasis after a median period of 20 (range 6-122) months. The DMFS rates for these patients after 2 and 5 years were 56 and 35 %, respectively (Fig. 1b). Univariate analysis showed that only lymph node metastasis at presentation (p = 0.003) was a significant prognostic factor for DMFS, and this remained significant after multivariate analysis (p = 0.005) ( Table 3).
At 2 and 5 years, the OS of the 18 patients with proximal-type ES was 53 and 33 %, respectively. This was worse than the OS of the 26 patients with classic-type ES, i.e. 81 and 53 %, respectively (Fig. 1e).

DISCUSSION
ES is an aggressive but rare soft tissue tumor with severe consequences, even with currently available multimodal therapy. Risk factors for OS of ES patients have been reported to be large tumor size, 2,15-17 deep-seated occurrence, 2,15,18 proximal location, 2,8,9 local recurrence, 15,18 lymph node involvement, 2,10,18 mitosis, necrosis, hemorrhage and vascular invasion. 2,16,19 However, because of the rarity of ES, few data on its clinical behavior or the survival of affected patients are available. Recently, the proximal subtype of ES has been reported to have a poorer prognosis than the classical   6,7,[9][10][11] Rekhi et al. reported that proximal-type ES included a slightly higher proportion of M1 tumors (43 %) than classic-type ES (35 %), although the difference was not statistically significant (p = 0.61). 10 In our present study of ES, the proximal type included a significantly higher proportion of M1 tumors (44 %) than the classic type (12 %) (p = 0.015) and tended to have a higher proportion of N1 tumors (39 %) than the classic type (15 %), although the difference was not statistically significant (p = 0.093). It has been reported previously that 65-80 % of ES cases were classified as FNCLCC grade 2 tumors and 20-35 % as grade 3 tumors. 9,10 In addition, it has been reported that proximal-type ES included a higher proportion of grade 3 cases (64-73 %). 9,10 In addition, in our present study the proximal subtype included a significantly higher proportion of higher-grade tumors (67 %; p = 0.045). Histologically, proximal-type ES frequently shows rhabdoid features. 6,7 Hasegawa et al. revealed that 14 (70 %) of 20 cases of proximal-type ES contained rhabdoid cells. 7 We also observed a higher percentage of rhabdoid cells in proximal-type ES (83 %) than in classictype (35 %) ES (p = 0.002). Proximal-type ES also showed a significantly higher rate of vascular invasion (50 %) than classic-type ES (13 %) (p = 0.007). To our knowledge, no previous report has mentioned the incidence of vascular invasion associated with ES subtypes. These results indicate that proximal-type ES has a more histologically aggressive nature than classic-type ES.
Deep-seated tumor localization has been reported to be an independent risk factor influencing LRFS. 12 In our present study, a superficial location and lymph node metastasis were independent risk factors for local recurrence. Although this contrasts with previous reports, our finding could be explained by the higher number of superficial tumors in surgically more challenging sites, such as the perineal and inguinal regions, compared with other case series. 6,7 In-transit metastasis is defined as a type of metastasis in which cancer spreads through a lymph vessel and begins to grow more than 2 cm away from the primary tumor before it reaches the nearest lymph node. 20 Although in-transit metastasis has not been reported in ES, it may play a role in inducing local recurrence in soft tissue sarcomas, especially when there is a high potential for lymph node metastasis. This mechanism could explain the higher rate of local recurrence of N1 tumors in our present series.
ES has been reported to have a high rate of distant metastasis ranging from 40 to 57 %. 12,[15][16][17][18]21 The rate of 73 % recorded in the present study is higher than in previous reports; however, this could have been because a substantial number of cases in our series were either at a high tumor stage (N1 or M1 tumors, 39 %) or were recurrent cases (43 %) at initial presentation. 12,18,22 The most common initial sites of distant metastasis in our series were the lung (45 %) and distant lymph nodes (43 %). Independent risk factors for DMFS have been reported to be large tumor size, 22 deep-seated location, 18 local recurrence, and regional lymph node metastasis. 12,18 In our present series, the presence of lymph node metastasis was the only independent risk factor for DMFS.
The OS rates at 5 years for ES have been reported to be 32-78 %. 12,[15][16][17][18]21 In our series, the OS rate at 5 years (45 %) was relatively low but within the previously described range. We found that distant metastasis was the only independent factor predictive of a poor outcome. This would likely have been due to the limited number of ES patients, as well as the relatively short follow-up period (median 26.5 months).
It is still debatable whether prophylactic lymph node dissection is beneficial for ES patients. 12,15,17,[21][22][23] In the present study, lymph node dissection was indicated only when lymph node involvement was observed by diagnostic imaging. As lymphatic relapse was the most critical factor associated with poor outcome in both types of ES, it will be necessary to improve the detection and treatment of such lymphatic spread. Currently, we perform PET-CT examinations on all ES patients, and if lymph node involvement is suspected, we perform serial lymph node dissection and adjuvant radiation therapy if the resected node is histologically positive.

CONCLUSIONS
We found that proximal-type ES had significantly higher clinicopathological aggressiveness than classic-type ES, and was associated with a proximal tumor location, a higher tumor stage, presence of rhabdoid cells, a higher tumor grade, and vascular invasion. Lymph node and distant metastases had the most critical impact on prognosis. Taking the rarity of this tumor into consideration, the number of patients in our series might have been too small to allow any definitive conclusion to be drawn. Therefore, further validation using an independent cohort of patients with ES is needed. Despite its slow growth, ES can be extremely aggressive when disseminated, and currently no useful treatments are available. Novel strategies are therefore needed to improve the survival of patients with these highly aggressive sarcomas.