背景
随着免疫检查点抑制剂（immune checkpoint inhibitors, ICIs）在结直肠癌新辅助治疗中的广泛应用，出现了一些影响后续治疗甚至危及生命的免疫相关不良反应（immune-related adverse effects, irAEs），逐渐成为研究者们关注的焦点。既往研究大多基于小样本量、单中心回顾性数据，或仅将irAEs作为次要研究终点，缺乏系统性分析，限制了对其发生特征和预测因子的深入理解。
目的
本研究旨在系统评估直肠癌新辅助免疫治疗中irAEs的发病谱、临床特征及其潜在预测标志物，探索多中心数据中irAEs的发生特点以及治疗方案的影响，为个体化管理提供循证依据。
方法
本研究采用三阶段逐步深入的方法设计：（1）系统综述：检索PubMed、Embase和Cochrane等数据库，纳入17项涉及直肠癌新辅助ICIs治疗的临床研究，评估irAEs的整体发生率、等级分布、受累器官类型及起始时间；（2）单中心回顾性队列分析： 纳入2022年7月至2024年9月于北京协和医院接受替雷利珠单抗联合放化疗的82例直肠癌患者，提取临床资料，采用单因素和多因素Logistic回归分析irAEs的独立预测因子，并绘制ROC曲线评估预测模型效能；（3）多中心登记研究：联合全国8家三级医院建立统一登记系统，收集患者临床特征、免疫治疗方案、irAEs发生情况、干预处理及围手术期结局，设立标准化随访机制，评估irAEs的发生发展特点及与病理缓解程度之间的关系。
结果
第一部分系统综述显示，irAEs总体发生率为52.6%，以皮肤、肝脏及内分泌系统最常见，不同研究中发生率存在较大差异。严重irAEs主要集中在肝功能异常与肠炎，发生时间主要集中在治疗后前8周内。第二部分的回顾性研究共纳入本中心82例接受ICIs治疗的患者，其中有30例（39.0%）患者发生37例次irAEs，以内分泌不良反应最为多见，发生I-II级irAEs共32人（86.5%）。进一步分析发现低体质指数（body mass index, BMI）[P=0.004，OR: 13.81 (2.37-80.55)]、低中性粒淋巴细胞比值（neutrophil to lymphocyte ratio, NLR）[P=0.024，OR: 0.81 (0.68-0.97)]及有自身免疫病史者[P=0.008，OR: 0.36（0.17-0.77）]是irAEs发生的预测标志物。第三部分多中心登记性研究共纳入148例发生irAEs的患者，I–II级毒性占多数（83.7%），以肝毒性和内分泌毒性最为常见，其次为皮肤毒性。通过多因素分析发现，SCRT联合治疗是肝毒性的独立危险因素[P=0.015，OR: 7.92 (1.50-41.84), Ref: LCRT]，而既往自身免疫性疾病病史的患者内分泌毒性的发生风险显著增加。我们未发现严重irAEs对手术时间、术中失血量或术后并发症发生率产生显著影响。
结论
直肠癌新辅助免疫治疗中irAEs具有发生率高、类型多样、以轻中度为主等特征。通过系统分析和多中心验证，本研究明确了多项高风险因素并探索了预测标志物，为临床早期识别与干预提供理论基础，未来可以结合多种生物标志物建立更加精准的预测模型。irAEs在合理管理下并不显著影响围手术期安全性，未来可进一步研究其与免疫治疗长期疗效之间的关联。

Background:

With the widespread application of immune checkpoint inhibitors (ICIs) in the neoadjuvant treatment of colorectal cancer, increasing attention has been paid to immune-related adverse events (irAEs), some of which may impact subsequent treatment or even be life-threatening. Most previous studies were based on small sample sizes, single-center retrospective data, or treated irAEs as secondary endpoints, lacking systematic analysis and limiting the understanding of their characteristics and predictive factors.

Objective:

This study aims to systematically evaluate the spectrum, clinical characteristics, and potential predictive biomarkers of irAEs in neoadjuvant immunotherapy for rectal cancer, and to explore the patterns of irAE occurrence and the impact of treatment regimens based on multi-center data, thereby providing evidence for individualized management.

Methods:

A three-stage progressive research design was adopted: (1) Systematic Review: A comprehensive search was conducted in PubMed, Embase, and Cochrane databases, and 17 clinical studies involving neoadjuvant ICIs for rectal cancer were included to assess the overall incidence, grade distribution, affected organs, and onset timing of irAEs; (2) Single-center Retrospective Cohort Study: A total of 82 rectal cancer patients treated with tislelizumab combined with chemoradiotherapy at Peking Union Medical College Hospital between July 2022 and September 2024 were included. Clinical data were extracted, and univariate and multivariate logistic regression analyses were performed to identify independent predictors of irAEs. The predictive model's performance was evaluated using ROC curves; (3) Multi-center Registry Study: A standardized registry system was established across eight tertiary hospitals in China. Baseline characteristics, immunotherapy regimens, irAE occurrence, interventions, and perioperative outcomes were collected. A uniform follow-up protocol was implemented to analyze the evolution of irAEs and their relationship with pathological response.

Results:

The systematic review revealed an overall irAE incidence of 52.6%, with skin, hepatic, and endocrine toxicities being the most common. Severe irAEs mainly included liver dysfunction and colitis, typically occurring within the first 8 weeks of treatment. In the retrospective cohort, a total of 82 patients receiving ICIs treatment were enrolled. 30 patients (39.0%) experienced 37 irAE episodes, with endocrine toxicities being the most frequently observed. Of these, 32 patients (86.5%) developed Grade I-II irAEs. Low body mass index (BMI) [P=0.004, OR: 13.81 (2.37–80.55)], low neutrophil to lymphocyte ratio (NLR) [P=0.024, OR: 0.81 (0.68–0.97)], and a history of autoimmune disease [P=0.008, OR: 0.36 (0.17–0.77)] were identified as significant predictive factors. In the multi-center registry, 148 patients with irAEs were included. The majority experienced grade I–II toxicities (83.7%), while grade III or higher events occurred in 16.3%. Hepatic and endocrine toxicities were most prevalent, followed by skin-related events. Multivariate analysis identified short-course radiotherapy (SCRT) as an independent risk factor for hepatic toxicity [P=0.015, OR: 7.92 (1.50–41.84), Ref: LCRT], while a history of autoimmune disease significantly increased the risk of endocrine toxicity. Notably, severe irAEs did not significantly impact operative time, intraoperative blood loss, or postoperative complication rates.

Conclusion:

irAEs in neoadjuvant immunotherapy for rectal cancer are common, diverse in type, and predominantly mild to moderate in severity. This study, through systematic analysis and multi-center validation, identified several high-risk factors and explored potential biomarkers for irAE prediction, providing a theoretical basis for early clinical identification and intervention. Under proper management, irAEs do not significantly affect perioperative safety. Further research is warranted to explore the relationship between irAEs and long-term immunotherapy outcomes.