Background:Systemic therapy for soft tissue sarcoma (STS) provides modest survival benefit but carries clinically relevant toxicity. Published trials report adverse events (AEs) of varying quality and extension. Poor toxicity reporting hampers balanced risk–benefit appraisal.Methods:A PRISMA-2020 systematic review was registered in PROSPERO CRD420251087366. PubMed, CENTRAL, and Google Scholar were searched from 16 December 2024 to 16 April 2025 for randomized controlled trials (RCTs) evaluating chemotherapy, kinase inhibitors, or immune checkpoint inhibitors in STS. AE terms were harmonized to CTCAE v5.0; event rates were normalized to patients evaluable for safety. Pooled proportions used DerSimonian–Laird random-effects models; between-group comparisons employed unpairedt-tests. Risk of bias (RoB 2) was assessed with the Cochrane RoB 2 tool.Results:Ten RCTs (1079 treated, 979 control patients; 1994–2024) met the inclusion criteria, although two lacked sufficient presentation of toxicity data and seven failed to report parallel control-arm AEs. Pooled normalized incidences for treated patients were as follows: grade ≥ 3 hematological AEs, 17% (95% CI 14–20); severe gastrointestinal AEs, 9% (8–11); and grade 4 AEs, ≤6%. Anthracycline-based and kinase-inhibitor regimens displayed comparable composite grade ≥ 3 burdens (58% vs. 84%,p= 0.64). Between-study heterogeneity was considerable for gastrointestinal and hematological events (I2> 60%), driven by differing AE scales and denominators. Late-effect toxicities (cardiac, hepatic, neurological, and nephrological) were rarely reported, occurring in <1% of the patients. Across the three RCTs with control-arm data, experimental therapy increased common grade 3 AEs by 4–12 percentage points (p= 0.001). RoB 2 flagged serious concerns in 4/10 trials.Conclusions:Severe AEs in STS systemic therapy are moderately frequent; while the toxicity spectrum differs across drug classes (e.g., hematological for anthracyclines vs. neuropathic or fatigue-related for agents such as eribulin), the aggregate burden of severe AEs has not been lower for newer agents. Confidence in these estimates is limited by incomplete and non-standardized AE reporting. Future sarcoma trials must adopt CTCAE v5.0, specify explicit safety denominators, and publish full AE matrices to enable high-certainty risk–benefit assessment.
背景：软组织肉瘤（STS）的系统性治疗虽能带来有限的生存获益，但伴随具有临床相关性的毒性反应。已发表的临床试验对不良事件（AEs）的报告在质量和详尽程度上存在差异。毒性报告不充分阻碍了风险-获益的平衡评估。 方法：本研究在PROSPERO平台注册了一项遵循PRISMA-2020指南的系统性综述（注册号CRD420251087366）。于2024年12月16日至2025年4月16日期间检索了PubMed、CENTRAL和Google Scholar数据库，筛选评估化疗、激酶抑制剂或免疫检查点抑制剂治疗STS的随机对照试验（RCTs）。不良事件术语统一采用CTCAE v5.0标准；事件发生率均以可进行安全性评估的患者数为基数进行标准化。采用DerSimonian–Laird随机效应模型计算合并比例；组间比较采用非配对t检验。使用Cochrane RoB 2工具评估偏倚风险。 结果：共10项RCTs（治疗组1079例，对照组979例；研究时间1994–2024年）符合纳入标准，但其中两项缺乏充分的毒性数据呈现，七项未能报告平行对照组的AEs。治疗组患者的合并标准化发生率如下：≥3级血液学AEs为17%（95% CI 14–20）；严重胃肠道AEs为9%（8–11）；4级AEs ≤6%。基于蒽环类药物的方案与激酶抑制剂方案的复合≥3级毒性负担相当（58% vs. 84%，p=0.64）。由于采用的AE分级标准和分母不同，研究间在胃肠道和血液学事件上存在显著异质性（I² > 60%）。迟发性毒性（心脏、肝脏、神经和肾脏毒性）报告罕见，发生率<1%。在提供对照组数据的三项RCTs中，试验性治疗使常见3级AEs的发生率增加了4–12个百分点（p=0.001）。RoB 2评估显示，10项试验中有4项存在严重偏倚风险。 结论：STS系统性治疗中严重AEs的发生率中等；尽管不同药物类别的毒性谱存在差异（例如蒽环类药物以血液学毒性为主，而艾日布林等药物则以神经病变或疲劳相关毒性为主），但新型药物的总体严重AEs负担并未降低。这些估计值的可信度因AE报告不完整和非标准化而受限。未来的肉瘤临床试验必须采用CTCAE v5.0标准，明确安全性评估分母，并公布完整的AE矩阵，以实现高确定性的风险-获益评估。
肿瘤(癌症)患者之家