机器人食管切除术的学习曲线：一项系统综述及拟议的培训与能力评估框架

文献信息

中文摘要

背景： 机器人辅助微创食管切除术（RAMIE）是食管癌手术的重大进展，可提供更好的可视化、灵活性和手术精准度。然而，RAMIE技术要求高，学习曲线陡峭。本系统综述的目的是评估关于机器人食管切除术学习曲线的现有证据，并制定RAMIE的结构化培训和评估框架。
方法： 我们根据PRISMA 2020声明，在PubMed/MEDLINE、Scopus和Cochrane Library中进行了从建库至2026年6月的系统文献检索。纳入评估成人患者机器人食管切除术学习曲线或实施情况的研究。结局包括熟练度阈值、手术时间、喉返神经（RLN）损伤、并发症、淋巴结清扫数量以及影响学习进展的因素。使用纽卡斯尔-渥太华量表（NOS）评估偏倚风险。
结果： 我们纳入了2013年至2026年间发表的25项研究。大多数研究是来自东亚、欧洲和北美高容量中心的回顾性观察性队列研究。最常见的学习曲线评估方法基于累积和（CUSUM）分析。通常，各研究中初始熟练度在约20-50例后达到，但并发症的稳定和肿瘤学质量的优化往往需要更高的手术量。外科医生经验的增加与手术时间、RLN损伤、淋巴结清扫数量、中转率和术后并发症的改善相关。既往微创食管切除术经验、结构化督导路径、高机构手术量以及基于团队的实施模式，始终能够促进更快速地获得熟练度并更安全地实施RAMIE。
结论： 机器人食管切除术的学习曲线陡峭，但可以达到熟练。结构化培训项目、专门的机器人团队和标准化的实施路径可能缩短学习曲线并提高患者安全性。学习曲线的进展似乎受到病例量以外因素的影响，支持在RAMIE实施过程中制定结构化培训和评估框架。基于现有证据，我们提出了一个基于证据的专家框架，该框架应被视为产生假设，并需要前瞻性验证。需要前瞻性多中心研究和学习曲线的标准化报告，以优化机器人食管手术的培训并建立普遍接受的熟练度阈值。
系统综述注册： https://www.crd.york.ac.uk/PROSPERO/，标识符CRD420261417234。

英文摘要

BACKGROUND: Robotic-assisted minimally invasive esophagectomy (RAMIE) is a significant evolution in esophageal cancer surgery, providing improved visualization, dexterity and operative precision. However, RAMIE is technically demanding and has a steep learning curve. The goal of this systematic review was to assess the existing evidence on the learning curve of robotic esophagectomy and to develop a structured training and assessment framework for RAMIE.
METHODS: We performed a systematic literature search in PubMed/MEDLINE, Scopus and the Cochrane Library from inception to June 2026 according to the PRISMA 2020 statement. Studies that assessed the learning curve or implementation of robotic esophagectomy in adult patients were included. The outcomes included proficiency thresholds, operative time, recurrent laryngeal nerve (RLN) injury, complications, lymph node harvest, and factors affecting learning progression. The risk of bias was assessed with the Newcastle-Ottawa Scale (NOS).
RESULTS: We included 25 studies published between 2013 and 2026. Most studies were retrospective observational cohorts from high-volume centers in East Asia, Europe and North America. The most common assessment methodologies of the learning curve were based on cumulative sum (CUSUM) analysis. Typically, initial proficiency was attained after approximately 20-50 cases across studies, but stabilization of complications and optimization of oncologic quality often required higher procedural volumes. Increasing surgeon's experience was associated with improvements in operative time, RLN injury, lymph node harvest, conversion rates, and postoperative complications. Prior experience in minimally invasive esophagectomy, structured proctoring pathways, high institutional volume, and team-based implementation models consistently enabled more rapid proficiency acquisition and safer RAMIE implementation.
CONCLUSIONS: The learning curve for robotic esophagectomy is steep, but attainable. Structured training programs, dedicated robotic teams, and standardized implementation pathways may shorten the learning curve and enhance patient safety. Learning curve progression appears to be influenced by factors beyond case volume alone, supporting the development of structured training and assessment frameworks during RAMIE implementation. Based on the available evidence, we propose an evidence-informed expert framework that should be considered hypothesis-generating and requires prospective validation. Prospective multicenter studies and standardized reporting of learning curves are required to optimize training in robotic esophageal surgery and to establish universally accepted thresholds of proficiency.
SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/, identifier CRD420261417234.