Ecacy of Neoadjuvant Chemotherapy for Synovial Sarcoma: Retrospective Analysis of a Nationwide Database in Japan

Synovial sarcoma is an aggressive but chemosensitive soft-tissue tumor. We retrospectively analyzed the ecacy of neoadjuvant chemotherapy for synovial sarcoma with data from the nationwide database, Bone and Soft Tissue Tumor Registry in Japan. This study included 316 patients diagnosed with synovial sarcoma between 2006 and 2012. Oncologic outcomes were analyzed using a Cox-hazard regression model. The effects of neoadjuvant chemotherapy on outcomes were evaluated using a matched-pair analysis. The oncologic outcomes of patients who did or did not receive neoadjuvant chemotherapy were compared (cx+ and cx-). Multivariate analysis revealed signicant correlations of distant postoperative metastasis (hazard ratio [HR] = 0.01, p<0.001) with overall survival; surgical margin type (marginal resection, HR=0.12, p=0.011 and intralesional resection, HR=0.08, p=0.022 versus wide resection) with local recurrence; and postoperative local recurrence (HR=0.30, p=0.027) and surgical margin (marginal resection, HR=0.31, p=0.023 versus wide resection) with distant relapse-free survival.

adolescents and young adults, and the median patient age at diagnosis is 35 years [1][2][3][4][5][6]. These tumors can be divided into three histologic subtypes: monophasic tumors, which are composed of spindle cells; biphasic tumors, which are composed of spindle cells and epithelial cells; and poorly differentiated tumors, which are composed of small round cells [4]. SS is considered to be chemosensitive [4,7], and wide excision with a negative margin is necessary for effective treatment [8][9][10]. Therefore, the administration of neoadjuvant chemotherapy might be a rational approach to reduce micro-invasion from the primary site. However, chemotherapy for SS remains controversial because it is di cult to conduct a prospective study on the e cacy of neoadjuvant therapy, speci cally for this tumor type. Moreover, several pretreatment characteristics, including the tumor size, age, histologic grade, and tumor depth [6,9,11,[14][15][16], in uence the prognosis of a patient with SS and may have affected the results of previous studies.
We designed this study based on a matched-pair analysis (MPA) to clarify the role of neoadjuvant chemotherapy in the prognosis of SS patients. For a secondary analysis, we conducted subgroup analyses to detect populations that might bene t from neoadjuvant chemotherapy.

Patient selection
We extracted patient data from the Bone and Soft Tissue Tumor (BSTT) Registry of Japan, a nationwide organ-speci c cancer registry for bone and soft-tissue tumors. Eighty-nine Japanese Orthopedic Association (JOA)-certi ed hospitals that specialize in musculoskeletal oncology participated obligatorily in this registry, and other hospitals participated voluntarily. The annual reports published by the BSTT include patient characteristics, such as basic data ( , and information about additional treatments (chemotherapy, radiotherapy, and hyperthermia) [17].Follow-up surveys were conducted to collect information after 2, 5, and 10 years following the initial registration.
These surveys included items about several types of outcomes, such as local recurrence, distant metastasis, and oncologic outcomes, at the time of the latest follow-up. This study was approved by the Institutional Review Board of the JOA.
From the BSTT registry, we identi ed 579 patients who were diagnosed with SS between 2006 and 2012. Of these, we excluded 133 patients who did not undergo primary tumor resection, 53 patients with missing data, and 77 patients with metastatic lesions. The nal analysis dataset included 316 patients (Fig. 1).

Statistical analysis
The primary objective of this study was to investigate the following oncologic outcomes: overall survival (OS), de ned as the time from diagnosis to death from any cause; distant relapse-free survival (D-RFS), de ned as the time from surgery to distant progression or death; and local control (LC), de ned as the time from surgery to local recurrence. Standardized intergroup differences were calculated using Kaplan-Meier and log-rank analyses. Potential risk factors for oncologic outcomes were analyzed with a stepwise Cox proportional hazards model, and hazard ratios (HRs) were calculated from these data.
We divided patients into two groups based on treatment with or without neoadjuvant chemotherapy (cx + versus cx-group). For the MPA, data on statistical variables, including age, sex, tumor location, tumor size, tumor histology, tumor depth, and administration of neoadjuvant chemotherapy, were obtained from the BSTT registry. A multivariate logistic regression analysis was conducted to determine associations between these factors and the administration of neoadjuvant chemotherapy. Propensity scores were calculated using a logistic regression model that included the weights of the contributions of each patient's demographic data. After calculating these scores, we propensity score-matched patients in a 1:1 ratio by using a nearest-neighbor algorithm, allowing a maximum tolerated difference of ≤ 30% between propensity scores [18].  Before adjustment with the propensity score, we observed some differences between patients who did (n = 151) or did not (n = 165) receive neoadjuvant chemotherapy; particularly, the former group tended to be younger and to have deeper tumor locations, larger tumors, more advanced-stage disease, and monophasic-type disease (  Fig. 2). year D-RFS rates were 76.9% (± 5.3%) in the cx + group and 78.7% (± 5.1%) in the cx-group (HR = 1.01 [0.53-1.92], p = 0.982, Fig. 2).

Discussion
De nitive treatment strategies for STSs, which vary based on histological characteristics, have not been fully determined. Moreover, the various subtypes of STS are rare, and it is di cult to deduce meaningful results. Therefore, we analyzed data from the largest soft-tissue tumor-speci c database in Japan to determine the risk factors associated with SS outcomes. Notably, we identi ed distant metastasis after primary resection, surgical margins, and local recurrence after primary resection and surgical margins as risk factors that affect the oncologic outcomes of OS, LC, and D-RFS, respectively. In contrast, previous studies had identi ed distant metastasis at diagnosis, SS subtype (poorly differentiated < biphasic type and monophasic < biphasic type), tumor depth (deep < super cial), tumor size (> 5 cm), and local recurrence as independent prognostic factors [6,7,9,[14][15][16]. These ndings indicated the importance of complete surgical resection to avoid micro/macro-residues of the tumor in the post-resection margins. To verify improvements in the surgical outcome, we further attempted to demonstrate the e cacy of neoadjuvant chemotherapy for SS by comparing the oncologic outcomes between the cx + and cxgroups.
Theoretically, neoadjuvant therapy for locally advanced and high-risk STS has several advantages over adjuvant chemotherapy. First, this modality could potentially reduce possible invasion around the tumor and thus prevent micro-residual resection. In cases where a tumor arises near neurovascular bundles, neoadjuvant chemotherapy improves the likelihood that resection will spare the neurovascular bundles and may allow the patient to forgo amputation, which allows the preservation of muscle function [22]. Second, neoadjuvant chemotherapy can potentially improve patient survival by eradicating micrometastatic disease without delaying adjuvant chemotherapy. Finally, neoadjuvant therapy can provide guidance for postoperative therapeutic strategies.
Currently, the role of neoadjuvant chemotherapy in SS remains controversial [19] because of the challenges associated with prospective studies and the potential for various selection biases in retrospective studies. Therefore, to reduce the possible bias of the retrospective analysis, we examined the oncologic outcomes of SS using an MPA of a relatively large population and thus present a novel report. We found that the cx + population was more likely to have monophasic-type disease, a large tumor size, and distant metastasis at diagnosis, suggesting that selection bias might affect the oncologic outcomes. However, we did not observe a signi cant difference in the oncologic outcomes of patients in the cx + and cx-groups, despite propensity matching to reduce intergroup differences. Possibly, similar numbers of patients in both groups derived bene ts from chemotherapy. Although we did not identify signi cant intergroup differences, our data suggested that patients with large tumors (> 5 cm) and those with monophasic-type disease would bene t from neoadjuvant chemotherapy (Appendices A and B). These characteristics might affect oncologic outcomes and may explain the controversial results of previous reports.
In a previous study of the analysis of resectable SS, the 5-year OS rate was 64%, and age > 35 years, Grade 3 tumor, and margins other than R0 were identi ed as signi cant independent predictive factors for improved survival. Despite adjustments based on these factors, we did not observe any signi cant impact of chemotherapy on survival in the present analysis. The HRs for OS with neoadjuvant chemotherapy and adjuvant therapy were 1.01 (p = 0.358) and 1.62 (p = 0.099), respectively [12]. Similarly, 52% of patients with localized SS in an Italian study were treated with a combination of ifosfamide and doxorubicin or epirubicin, and the 5-year OSs of those who did or did not receive chemotherapy were 69% and 82%, respectively (p = 0.20). In that study, the negative impact of chemotherapy was explained by the exclusive administration of this treatment modality to patients with larger tumors (> 5 cm) and re-excision cases. These preconditions may have in uenced the outcomes [9].
In contrast, an analysis of 14 trials reported that doxorubicin-based chemotherapy signi cantly improved oncologic outcomes. Although the SS subgroup extracted from these trials was better oriented for chemotherapy, that analysis identi ed no signi cant improvement in OS (57.5% and 47.3% for the chemotherapy and control groups, respectively) [20]. Eilber et al. reported favorable outcomes with ifosfamide-based chemotherapy for SS in a dataset limited to patients with tumors > 5 cm, deep tumors, as well as primary and extremity tumors and were treated between 1990 and 2002. In that study, the 4year disease-speci c survival rates were 88% and 67% in the chemotherapy and no-chemotherapy groups, respectively (p = 0.01). Additionally, treatment with an ifosfamide-based regimen was reported to improve D-RFS (HR = 0.4, p = 0.03) [13]. Ferrari et al. suggested that younger patients and those with tumors larger than > 5 cm achieved better outcomes with chemotherapy [21]. These retrospective studies indicate that high-risk patients might bene t from neoadjuvant chemotherapy, which supports our ndings.
Few published reports have focused on neoadjuvant therapy for the treatment of SS. One randomized phase 2 trial of adult patients with high-risk STS (tumor size > 8 cm of any grade, tumor size < 8 cm of grade 2/3, or locally recurrent sarcoma/after inadequate surgery of grade 2/3) did not indicate that a regimen of three cycles of neoadjuvant chemotherapy was superior to surgery alone in patients (5-year disease-free survival rates of 56% and 52% for the neoadjuvant chemotherapy and surgery-alone arms, respectively; p = 0.354) [23]. With regard to the comparison between neoadjuvant and adjuvant chemotherapy, a retrospective analysis reported that neoadjuvant chemotherapy (doxorubicin + ifosfamide + dacarbazine) for SS had a signi cant advantage over adjuvant chemotherapy (5-year OS rates of 84.5% and 55.6% for the neoadjuvant chemotherapy and adjuvant chemotherapy groups, respectively) [3]. However, that study did not su ciently balance patients who received neoadjuvant chemotherapy with those who received adjuvant therapy. Therefore, it remains di cult to draw meaningful conclusions on the actual contribution of neoadjuvant chemotherapy to patient outcomes.
This study had several limitations. First, the design was retrospective and, therefore, many biases, including selection and recall bias, may have in uenced the results despite a propensity-score adjustment. Second, the BSTT database covers only patients treated at orthopedic departments; thus, our dataset did not include patients treated at other departments. Third, we could not evaluate the exact intensity of chemotherapy. Consequently, patients who received attenuated chemotherapy were included in the cx + group. Fourth, we did not analyze differences in genotype. As fusion proteins resulting from the SYT-SSY1 or SYT-SSY2 fusions have been associated with the histological subtype and clinical behavior, these biomarkers should be reviewed in a further analysis of this database [24]. Finally, in Japan, the standard treatment protocol for SS involves a doxorubicin-based chemotherapy regimen. However, the different participating institutions do not use identical protocols. Accordingly, although 60.1% of patients received neoadjuvant chemotherapy via the AI regimen, many patients were treated with AI-IE, VDC/IE, MAID, or other regimens. These differences might have affected the study outcomes. We further note that, currently, new drugs are being approved rapidly in Japan, and pazopanib, trabectedin, and eribulin have been proven to yield improved oncologic outcomes in patients. These newly approved drugs may in uence patient outcomes and, therefore, potential changes in treatment strategy should be considered when applying our ndings.

Conclusion
We analyzed a large population database in Japan to determine the factors that affect the oncologic outcomes of patients with non-metastatic SS. Notably, we found that the margin status and postoperative local control were associated directly or indirectly with improvements in oncologic outcomes. However, we did not nd a signi cant contribution of neoadjuvant therapy to survival outcomes in either the non-adjusted or propensity score-matched populations.   Kaplan-Meier analyses of the oncologic outcomes of patients who did (cx+) or did not (cx-) receive neoadjuvant chemotherapy (red curve: cx+ group, black curve: cx-group). a-c: outcomes before propensity-score matching (n=316); d-f: outcomes after propensity-score matching (n=172). Triangles indicate the censored cases.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. additional gure1and2.docx