Adjuvant therapy for retroperitoneal sarcoma: a meta-analysis

Retroperitoneal sarcomas (RPS) are the second common malignancy after soft tissue sarcoma (STS) of the extremities, accounting for 10–15% of all STS and 30% of all malignant retroperitoneal tumors [1]. R0 surgical resection is the most potential treatment to cure patients with localized disease, which means the adjacent organs invaded are often not preserved. According to current studies, the rate of complete resection ranges from 41.8 to 76% [2]. However, local recurrence remains high even if negative margins is being achieved as far as possible, and leads to poor prognosis, with 5-year overall survival (OS) ranging from 39 to 65% and a mortality rate of 20–75% [3‐7].

To improve local control and overall survival, adjuvant therapy (AT) such as adjuvant radiotherapy (ART) has been investigated. However, there are insufficient evidences to compile treatment guidelines due to the different conclusions based on limited retrospective clinical studies (RCSs) [8‐22]. For example, multiple analyses of the Surveillance, Epidemiology, and End Results (SEER) database and a retrospective analysis from French have shown that ART could not improve OS in RPS patients [5, 9, 12, 23], but Trovik et al. [8] and others [10, 18] demonstrated a significant improvement in OS as well as recurrence-free survival (RFS) in patients undergoing ART, which are obviously confusing. In addition, most trials were undertaken in the setting of advanced extremity sarcomas and the generalizability of these data is limited [24, 25], resulting chemotherapy cannot be as a standard approach to treat RPS. Therefore, it is greatly clinical significance to discuss whether advanced RPS patients can benefit from adjuvant chemotherapy (ACT).

The aim of this meta-analysis is to review the latest body of literature comparing AT with surgery alone in RPS patients, and to clarify the role of ART and ACT in the prognostic outcome of these patients.

The unique biological behavior of RPS brings great challenges to clinicians in the treatment of this disease. R0 resection is the only potentially curative therapy, but the risk of LR is substantial because large size and anatomical structure of the tumor frequently preclude resection with widely clear margins. Five-year LR rates range from 28 to 60% [16, 26‐28]. With long-term follow-up, almost all patients are likely to recur [29]. Therefore, it is urgent to clarify the comprehensive treatment of RPS. ART, remains a controversial component of treatment, unlike STS in extremity, there is no rigorous level I data to support ART significantly reduce LR in RPS patients, and lessons from ART trials in extremity STS are difficult to translate directly to the RPS due to the potential for significant toxicity at equivalent dose [26]. Here our pooled analysis revealed that LR of RPS patients is distinctly reduce in ART group as compared to surgery alone, it is a key parameter to evaluate the prognosis of patients under different treatments. Besides, whether ART is beneficial to OS is also controversial. Several RCSs indicated consistently that ART improve local control (LC) and RFS but failed to demonstrate a statistically significant association with OS [5, 9, 12, 21]. Others obtained opposite results that ART not only improved the OS, but also has obvious curative effect in RFS and LC with a premise that the margin of surgical resection reaches R0 or R1 [3, 10, 11, 13, 16, 18]. In our meta-analysis, the results support the latter.

[In addition to therapeutic effects, radiation toxicities are primary consideration in selecting an appropriate radiation dose. Among the included studies, detailed radiation doses were given in seven studies (Table 1), and the median radiation doses were all 50 Gy, with the highest not exceeding 65 Gy. However, radiotherapy toxicities and target volumes were not mentioned in any of these studies, and the main reason was the lack of records on the aspect in the databases selected by these retrospective studies. Besides, the location occupied by the original sarcomas would be filled with adjacent normal tissues after resection, which also led to uncertainty in the radiotherapy target volumes. Previous two trials (phase1 and phase 2) have been reported that 45-54 Gy were optimal doses according to the potential benefits and risks assessed by them, to avoid bowel complications reported with higher doses and the potential negative impact on surgery [30, 31]. Recently, a phase 3 trials (multicenter, open-label, randomized) analysed the radiation toxicities in the ART, radiotherapy was delivered as 50.4 Gy (in 28 daily fractions of 1.8 Gy). The resulted showed the most common grade 3–4 adverse events (lymphopenia, anaemia, and hypoalbuminaemia) were reported in 127/127 patients in ART group, and in 16/128 patients in surgery alone. Serious adverse events were reported in 30/127 patients in ART group, and in 13/128 patients in surgery alone. One patient was died in ART group due to gastropleural fistula [32]. Complications rates were lower in this trial than have been reported with ART, ranging from20-40% in retrospective series [12, 33, 34]. In our study, the radiation doses performed were basically consistent with the optimal doses Therefore, we considered that the radiation toxicities are acceptable, but the specific radiation toxicities need to be further investigated.]

Anthracyclines were the first systemic chemotherapeutic agents to demonstrate activity in STS, and doxorubicin was main representative. According currently studies, using ACT in STS could not benefit patients, neither patients whose tumors remained resectable or patients who had metastasized at an advanced stage. A multicenter phase III RCT (EORTC) [35] randomly assigned 351 patients with non-metastatic macroscopically resected II-III tumors to the postoperative chemotherapy group (175 patients with ifosfamide/doxorubicin) or to the control group (176 patients). The results demonstrated OS did not differ significantly between groups (HR = 0.94, 95% CI 0.68–1.31; P = 0.72) nor did RFS (HR = 0.91, 95% CI 0.67–1.22; P = 0.51). A retrospective study of efficiency of ACT in resected RPS [11], published in 2017, showed that utilization of ACT was associated with worse long-term survival (HR = 1.30, 95% CI 1.05–1.61; P = 0.017). However, the trend of OS improvement with ACT were found in spindle cell (HR = 0.37, 95% CI 0.10–1.38), giant cell (HR = 0.82, 95% CI 0.32–2.13) and synovial sarcoma (HR = 0.26, 95% CI 0.05–1.33). Here our findings show that ACT cannot benefit RPS patients from OS, RFS and MFS, and even as previously reported study, it may cause worse OS (sensitivity analysis). Although whether metastasis occurs for patients is not the main reason to cause the increase in mortality, MFS was also included in the study as a prognostic parameter for PRS. In the studies we reviewed, only two studies included MFS [16, 17], and the two studies were conducted by the same author in different years.

In our study, notable heterogeneity was seen in LR of ART and OS of ACT, and only OS can perform sensitivity analysis due to lack of studies in LR. The sensitivity analysis showed that ACT was associated with worse OS, which was different from previous results of pooled analysis. In any case, the two results support that ACT cannot improve OS in patients with RPS. From another perspective, we have to admit that the result is not robust. Meanwhile, some limitations should be considered before appraising the results of this study. First, interventions in the treatment group in some studies were not limited to postoperative radiotherapy and chemotherapy, and in order to minimize the interference factors, we extracted HR form the multivariate COX regression analysis. Second, all included studies were RCSs, no RCT was found in databases we search. Finally, studies were insufficient in the sub-analysis of LR of ART and RFS of ACT duo to only two studies were included, limiting the validity of the comparisons between studies and conclusions drawn.

In this study, the quality of evidence was moderate but sufficient to establish the efficacy of ART for RPS. The relationship between ACT and the prognosis of RPS needs to be further studied, especially for patients with resectable RPS. Since there was a trend that ART is more likely to improve OS and LC of RPS patients, while ACT is for MFS, distinguishing the different efficiency between ART and ACT was also urgent. In addition, further studies could significantly change the results that ACT was associated with a wore OS, and more prognostic factors, such as pathological type, surgical resection method, dose, and related toxic complications, need to be included for analysis. Of course, all results will eventually need to be verified by multicentered RCTs.

Overall, our Meta-analysis showed RPS patients who underwent ART had better prognostic outcome than those who underwent surgery alone, including a longer OS, a longer RFS, and a lower LR. However, Those positive therapeutic effects have not been demonstrated in ACT, either in OS, RFS or MFS.