Chemotherapy improves distant control in localized high-grade soft tissue sarcoma of the extremity/trunk

Background Soft tissue sarcomas (STS) are rare and heterogeneous tumors making chemotherapy use controversial. Our goal was to identify a subset of patients with primary STS that benefit with the addition of chemotherapy. Methods A retrospective chart review included intermediate to high-grade localized primary STS of the extremity/trunk, and tumor size > 5 cm. The effect of chemotherapy was evaluated for local control (LC), distant control (DC), progression free survival (PFS), and overall survival (OS). Results In this cohort (n = 273), patients were treated with surgery (98%), radiation (81%), and chemotherapy (24.5%). With a median follow-up of 51 months, the entire cohort’s 5-year LC, DC, PFS, and OS are 79.1%, 59.9%, 43.8%, and 68.7%, respectively. The addition of chemotherapy did not provide a DC benefit (p = 0.238) for the entire cohort. High-grade disease (n = 210) experienced a 5-year benefit in DC (68% vs. 54.4%, p = 0.04), which was more pronounced with MAI (Mesna, Adriamycin, Ifosfamide) based regimens (74.2%, p = 0.016), and a 5-year PFS (50.8% vs 45%, p = 0.025) and OS benefit (76.2% vs 70%, p = 0.067) vs. no chemotherapy. On multivariate analysis of the high-grade subset, chemotherapy independently predicted for a DC benefit (HR 0.48 95% CI 0.26–89, p = 0.019). The benefit of chemotherapy was more pronounced with MAI, showing a significant benefit in DC (HR 0.333 95% CI 0.145–0.767, p = 0.01) and PFS (HR 0.52 95% CI 0.28–0.99, p = 0.047). Conclusion In patients with localized STS > 5 cm, the high-grade subset had a distant control benefit with the addition of chemotherapy, leading to improved progression free survival. This is more pronounced with the use of MAI and should be considered in patients eligible for this regimen.


Background
Sarcomas are a heterogeneous group of mesenchymal neoplasms that account for 1% of all adult malignancies with more than 50 histologic subtypes identified and a reported incidence of 13,040 cases in 2018 [1,2]. Approximately 60% of soft tissue sarcomas (STS) are diagnosed in the extremity or trunk [3]. The primary treatment modalities include surgery, radiation, and chemotherapy. Despite this trimodal approach, patients frequently experience local and distant recurrence. The risk of distant recurrence has been associated with various tumor characteristics, including: high-grade, size > 5 cm, anatomic site, histology, depth, and local recurrence [4,5].
Although low histologic grade and tumors measuring < 5 cm in size may be adequately treated with surgery alone, disease with unfavorable features are generally considered for adjuvant treatment [6,7]. Due to

Open Access
Clinical Sarcoma Research *Correspondence: victoria.rizk@gmail.com 1 Department of Hematology and Oncology, Moffitt Cancer Center and Research Institute, 12902 Magnolia Drive, Tampa, FL 33612, USA Full list of author information is available at the end of the article the heterogeneity and rarity of the disease, clinical studies evaluating the role of chemotherapy for patients at a high-risk of distant metastasis (e.g., high grade, > 5 cm, and often tumor depth) have led to inconsistent results, and its use is therefore not well defined in current guidelines [8,9].
Given the ongoing controversy regarding the role of systemic therapy for high-risk localized soft tissue sarcoma, we performed a retrospective review of the experience of our large, high-volume sarcoma referral center. The goal of this study was to review and contrast the treatment response of patients seen at the Moffitt Cancer Center (MCC) with intermediate to high-grade, tumor size ≥ 5 cm, STS of the extremities and trunk that received surgery and/or radiation with the response of those that received surgery and/or radiation combined with chemotherapy.

Methods
A retrospective chart review was performed evaluating patients with STS of the extremity and trunk treated at the MCC from 1998 to 2016. Patients included were ≥ 18 years of age and diagnosed with localized extremity or trunk STS that was ≥ 5 cm size based on pathologic size and intermediate to high-grade on histology (as defined by the Federation Nationale des Centres de Lutte Contre le Cancer or FNCLCC). Patients excluded held a diagnosis of bone or cartilage sarcomas (i.e., osteosarcoma, Ewing Sarcoma, etc.), rhabdomyosarcoma, site of disease located on head/neck or abdomen/ pelvis, metastatic disease at time of diagnosis, and welldifferentiated on pathology (e.g., FNCLCC Grade 1). This data was collected from the Cancer Registry Source System, a large database at Moffitt Cancer Center. Of the 522 patients in the original dataset, 273 patients met inclusion criteria.
Patient, tumor, and treatment characteristics evaluated included: gender, primary site(s) of disease, tumor size, performance status, histology, and initial treatment (i.e., chemotherapy, radiation, or surgery). In the chemotherapy cohort, variables such as agents, dosages, and number of cycles were recorded. Grade 3 toxicity data (as defined by Common Terminology Criteria for Adverse Events v5.0) were collected, including neurotoxicity, GI toxicity (i.e., nausea/vomiting), infectious disease complications (i.e., neutropenic fever), renal impairment, myelosuppression, and cardiotoxicity. Any reasons for dose reduction and/or discontinuation of chemotherapy were also recorded. Subset analysis was performed on clinical features that pose an even higher risk of distant metastasis (size ≥ 8 cm and/or FNCLCC Grade 3), to identify those who would most benefit from chemotherapy.

Statistical methods
Patient and tumor characteristics were compared in the chemotherapy and non-chemotherapy cohorts with the use of Pearson Chi square, Fisher's Exact Test, and Mann-U-Whitney as appropriate.
Time-to-event outcomes are defined from the date of diagnosis to the date of event, with censorship at last follow-up. These variables include time to local control (LC), distant control (DC), any recurrence or death (progression free survival or PFS), and overall survival (OS). Recurrence was an event in LC, DC, and PFS, whereas death was an event in PFS and OS. Patients that progressed distantly during neoadjuvant treatment and did not undergo surgery were considered an event. Univariate analyses were performed using a cox-regression analysis, with trending and significant variables defined as a 2-sided p value of 0.1 and 0.05, respectively. Timeto-event outcomes were illustrated with Kaplan-Meier based curves and comparisons were made with the use of log-rank, univariate analyses was also performed with cox regression analysis.
Multivariate cox regression analysis was performed including age (> 70 vs. ≤ 70), tumor size (≥ 8 cm vs. < 8 cm), ECOG performance status (2+ vs. 0-1), sex, chemotherapy, histology (reference: sarcoma not otherwise specified), and disease site. The multivariate analysis was performed for the entire cohort, grade 3 subset, and grade 3 patients treated with MAI/MAID (vs. no chemotherapy). The Cox regression analysis was performed for local control, distant control, progression free survival, and overall survival. As MAI based chemotherapy regimens are becoming standard practice in STS [6,[10][11][12], the use of this chemotherapy regimen in particular was compared to patients that received no chemotherapy. Statistical analysis was performed with SPSS (version 22; IBM Corporation, Armonk, NY). This review was approved by the MCC and the University of South Florida Institutional Review Boards.

Institutional outcomes Disease control and survival on univariate analysis
After a median follow-up of 51 months, the 5 Table 2).

Independent predictors of outcome (multivariate analysis)
On multivariate analysis, no factors were independently associated with LC. Factors associated with a DC

Subset analysis Identifying subsets associated with a distant control benefit
Kaplan-Meier curves evaluating subsets based on FNCLCC grade (2 vs. 3), tumor size (< 8 cm vs. ≥ 8 cm), and chemotherapy were used (all vs. MAI based) (Fig. 1). In high-grade (FNCLCC grade 3) disease, there was a significant improvement in distant control for those patients who received chemotherapy (Fig. 2). Although there is no significant distant control benefit with the use of chemotherapy for the entire cohort (p = 0.238), subset analysis showed that MAI/MAID chemotherapy had a non-significant improvement in the 5-year DC (67% vs. 59.4%, p = 0.111, n = 264). Specifically in patients with highgrade disease (n = 202), the use of MAI/MAID chemotherapy provided a significant 5-year DC benefit (74.2% vs. 54.4%, p = 0.016, n = 202), when compared to the non-chemotherapy group (Fig. 3).

Univariate predictors of outcome in high-grade and large tumor subsets
Subset analysis evaluating cohorts with high-grade disease (n = 210) or tumor ≥ 8 cm (n = 179) in size were performed. On univariate analysis of the high-grade cohort, chemotherapy predicted for a significant DC benefit (HR 0.55 95% CI 0.31-0.98, p = 0.043), but the impact on PFS  Further subset analysis was performed on the chemotherapy cohort to evaluate outcome differences for high-grade tumors. The high-grade non-chemotherapy cohort's 5-year LC, DC, PFS, and OS are 77.3%, 54.4%, 38.6%, and 62.6%, respectively. The high-grade chemotherapy cohort's 5-year LC, DC, PFS, and OS are 73%, 68%, 45%, and 70%, respectively (Fig. 3). This is more pronounced when evaluating patients treated with MAI/ MAID with a local control, distant control, PFS, and OS of 73.3%, 74.2%, 50.8%, and 76.2%, respectively (Fig. 4). Although a DC benefit was observed in the high-grade chemotherapy arm (p = 0.04), this did not translate into a statistically significant PFS benefit (p = 0.064).

Independent predictors of outcome in high-grade subset
On multivariate analysis of the high-grade cohort, chemotherapy independently predicted for a DC benefit (HR 0.475 95% CI 0.236-.954, p = 0.037); however there was no statistical benefit noted on PFS (HR 0.629 95% CI 0.361-1.095, p = 0.101).
In the high-grade cohort, multivariate analysis showed that specifically MAI/MAID chemotherapy independently associated with improvements in DC (HR 0.333 95% CI 0.145-0.767, p = 0.01) and PFS (HR 0.52 95% CI

Discussion
Our institutional experience showed patients with highgrade localized STS > 5 cm treated with MAI/MAID chemotherapy had a distant control benefit on univariate and multivariate analysis. However, chemotherapy for all intermediate to high-grade lesions > 5 cm in size was not significantly beneficial in improving distant control in STS. Though the role of chemotherapy in STS has been controversial, our study emphasized grade as an important factor to consider when assessing patients with localized STS > 5 cm for chemotherapy. This finding is consistent with previously identified high risk features commonly associated with distant spread (e.g., highgrade, > 5 cm, increased depth, and location) [13]. This review also emphasizes the importance of good performance status (ECOG 0-1) when making the decision to administer chemotherapy. This factor remained statistically significant in all analyses performed, and overall chemotherapy was well tolerated with only 36% experiencing grade 3 adverse events. Due to the rarity of STS, studies have typically evaluated the role of chemotherapy in a heterogeneous group, yielding higher power but lacking the ability to identify cohorts that would most benefit from chemotherapy [10,12]. While the decision to administer chemotherapy can be influenced by patient age, there can be discrepancies when comparing chemotherapy vs. non-chemotherapy cohorts due to age distribution rendering it difficult to accurately calculate an OS benefit. In contrast, evaluating a difference in DC allows us to examine the efficacy of systemic therapy, and is a known predictor to survival detriment [14].
In our study, chemotherapy has an independent and statistically significant DC benefit in patients with high-grade STS of the extremity and trunk. The benefit of chemotherapy was no longer significant when lower risk patients (e.g., intermediate grade sarcomas) were included, but this is likely because our study is underpowered to detect the smaller benefit seen in these patients. When appropriately powered, meta-analyses have shown an OS benefit with the use of perioperative chemotherapy [10,12]. While a study from the European Organization for Research and Treatment of Cancer (EORTC) did not show an OS benefit with the addition of chemotherapy, this may be attributed to the inclusion of patients with low to intermediate grade sarcoma (55%), which is consistent with our study that the benefit of chemotherapy in the intermediate grade may be less pronounced [11].
Since there is some debate regarding appropriate tumor size cut off for the categorization of high-risk patients (> 5 cm vs. > 8 cm), we evaluated the highgrade cohort for larger tumor size (> 8 cm). Evaluating the smaller cohort of high-grade (n = 210) and size > 8 cm (n = 179) further decreased our power, but chemotherapy independently predicted for a relative 25% distant control and 17% PFS benefit at 5 years. Despite prior evidence of improved chemotherapy responses in patients with pathologic tumor size > 8 cm, our current study did not find a statistically significant benefit with the use of chemotherapy in this cohort (n = 179, p = 0.193) [15,16]. The first meta-analysis by the Sarcoma Meta-Analysis Collaboration published in 1997 pooled together and reviewed 14 trials (n = 1568) of doxorubicin-based adjuvant chemotherapy, which demonstrated a significant improvement in distant recurrence and progression free survival, with a trend towards improved overall survival. Although this study was adequately powered and illustrated a 5-10% overall survival benefit with doxorubicinbased chemotherapy, it did not clarify which patients would benefit the most from chemotherapy [10,17].
In 2008, four more studies were added to the Sarcoma Meta-Analysis Collaboration. These studies used Adriamycin with Ifosfamide as the preferred treatment and results showed marginal efficacy of chemotherapy with decreased distant recurrence and improved overall survival [12]. Our study showed that MAI based chemotherapy has a stronger association with a distant control benefit, which translated to a progression free survival benefit and a trend towards a survival benefit at 5 years. Similarly, recent studies evaluating the role of chemotherapy in the treatment of STS have resulted in marginal improvement in disease control and overall survival, with difficulty reaching statistical significance attributed to limitations in sample size, various histologies, and heterogeneous treatment regimens [18][19][20].

Limitations
Due to the nature of retrospective cohort studies, there is an inherent bias in this patient population, especially for the choice to deliver chemotherapy. Through multidisciplinary tumor board discussions and recently established treatment pathways, there is an understandable bias for patients to be selected that we believe may benefit more from the addition of chemotherapy to their treatment plan. In addition, the rarity of this disease makes statistical power difficult, especially when evaluating cohorts with lower distant recurrence risk or when evaluating sub-cohorts to identify candidates ideal for this treatment. This study is likely underpowered to show a significant association between chemotherapy use and overall survival, and this may be due to cohort size and early patient censorship. In this cohort, competing risks in survival (e.g., PFS and OS) may be present since we were unable to identify their cause of death.
As with all retrospective studies, subgroup analysis should be interpreted with caution, as analyzing them may raise the type I error. A large prospective multi-institutional study would be required to adequately answer this question, but this can become difficult in this scarce and diverse population.

Conclusion
To date, the role of chemotherapy in localized STS of the extremity and trunk remains controversial. Expert opinion and current literature review suggests that chemotherapy in this population should be case-based with a preference towards chemotherapy in those that are high-risk. If a patient is deemed high-risk for recurrence, then perioperative chemotherapy is recommended as it has been shown to have a meaningful benefit in outcomes [21]. Although there are many publications to suggest a risk stratification score, standard guidelines have not yet been established [22,23]. Based on this current study, patients with the highest risk of distant recurrence (i.e., high-grade and > 5 cm) [24] should be strongly considered for perioperative chemotherapy. Specifically, MAI/MAID chemotherapy regimen should be used for appropriate candidates. Although this study was limited in size, the use of chemotherapy showed a significant decrease in distant recurrence for high grade STS of the extremity/trunk > 5 cm, which may translate into an improved overall survival. This should be confirmed with future prospective studies including longer follow-up and more patients.