Lemini Riccardo, Díaz Vico Tamara, Trumbull Denslow A, Attwood Kristopher, Spaulding Aaron C, Elli Enrique F, Colibaseanu Dorin T, Kukar Moshim, Gabriel Emmanuel
Department of Surgery, Mayo Clinic, Jacksonville, FL, USA.
University of Florida, College of Medicine, Gainesville, FL, USA.
J Gastrointest Oncol. 2021 Oct;12(5):1963-1972. doi: 10.21037/jgo-20-337.
Determining the best approach for esophageal cancer and predicting accurate prognosis are critical. Multiple studies evaluated characteristics associated with overall survival, and several prediction models have been developed. This study aimed to evaluate existing models and perform external validation of selected models.
A retrospective investigation of a multi-site institutional enterprise for patients with a diagnosis of esophageal cancer between 2013-2014 was performed. Selected survival prediction models included the Roswell Park Comprehensive Cancer Center (RPCCC) calculator, Oregon Health & Science University (OHSU) calculator, and two nomograms published by Shapiro and Sun One-year overall survival, level of agreement, and performance for each model were evaluated.
A total of 104 patients were included and used to assess the prediction models. One-year overall survival was 0.76. Different calculators tended to rank patients similarly; however, they did not agree on predicted overall survival. The least disparity in correlation was observed between OHSU and Shapiro calculators. Shapiro's model achieved the highest performance [area under the curve (AUC) =0.63].
Selected models showed fair results in estimating individual overall survival, although none achieved a high performance. While these tools may support the decision-making process for esophageal cancer patients, their implementation in clinical practice requires improved refinement to optimize their clinical utility.
确定食管癌的最佳治疗方法并准确预测预后至关重要。多项研究评估了与总生存期相关的特征，并开发了几种预测模型。本研究旨在评估现有模型并对选定模型进行外部验证。
对2013年至2014年间诊断为食管癌的患者进行了一项多机构企业的回顾性调查。选定的生存预测模型包括罗斯韦尔帕克综合癌症中心（RPCCC）计算器、俄勒冈健康与科学大学（OHSU）计算器，以及夏皮罗和孙发表的两个列线图。评估了每个模型的一年总生存期、一致性水平和性能。
共纳入104例患者用于评估预测模型。一年总生存期为0.76。不同的计算器对患者的排名往往相似；然而，它们在预测的总生存期上并不一致。在OHSU和夏皮罗计算器之间观察到的相关性差异最小。夏皮罗的模型表现最佳[曲线下面积（AUC）=0.63]。
选定的模型在估计个体总生存期方面显示出尚可的结果，尽管没有一个模型表现出色。虽然这些工具可能支持食管癌患者的决策过程，但在临床实践中实施它们需要进一步改进以优化其临床效用。
