中国实用外科杂志

中国实用外科杂志

• 论蓍 • 上一篇    下一篇

基于SEER数据库构建胆囊癌病人术后生存列线图研究

汪    涛，徐军明   

  1. 上海交通大学附属第一人民医院普外科，上海 200080
  • 出版日期:2021-04-01 发布日期:2021-04-20

  • Online:2021-04-01 Published:2021-04-20

摘要： 目的    构建可预测胆囊癌病人术后生存的列线图。 方法    从美国国立癌症研究院“监测、流行病学和结果”（surveillance，epidemiology，and end results，SEER）数据库提取1975—2016年诊断为胆囊癌的3616例病人数据，随机分为训练集和验证集，使用机器学习和Cox模型构建列线图，并评估列线图的判别能力和预测精度。结果    机器学习方法和多因素Cox风险回归模型结果表明，年龄、肿瘤分级、T分期、M分期、肿瘤分期和淋巴结清扫数目是胆囊癌病人预后的影响因素（P<0.05），以此建立列线图。ROC曲线下面积为0.825（95%CI 0.810~0.841）。验证集ROC曲线下面积为0.844（95%CI 0.819~0.868）。验证校正曲线提示，观察值与预测值之间有良好的一致性。分层分析显示，在T3、T4期胆囊癌病人中，淋巴结清扫数目≥4枚与1~3枚对预后的影响差异无统计学意义（P=0.538）；在T1、T2期胆囊癌病人中，淋巴结清扫数目≥4枚的病人较清扫1~3枚者有更好的预后，差异有统计学意义（P<0.01）。结论    列线图预后模型有较高的准确性，可为评估胆囊癌病人术后生存提供参考；T1、T2期胆囊癌病人增加淋巴结清扫数目相较于T3、T4期病人获益更多。

关键词: 胆囊癌, 列线图, 淋巴结清扫, SEER数据库

Abstract: Construction of prognostic nomogram of gallbladder cancer based on SEER database        WANG Tao，XU Jun-ming. Department of General Surgery，Shanghai General Hospital，Shanghai Jiao Tong University School of Medicine，Shanghai 200080，China
Corresponding author：XU Jun-ming，E-mail：xujunmingsh@163.com
Abstract    Objective    To construct a nomogram which can predict the survival rate of postoperative patients with gallbladder cancer. Methods    The data of 3616 patients diagnosed with gallbladder cancer from 1975 to 2016 were extracted from SEER database of National Cancer Institute. They were randomly divided into training group and validation group. The nomogram was established by machine learning and Cox model. The discriminant ability and prediction accuracy of nomogram were evaluated by receiver operator characteristic （ROC） curve and calibration chart. Results    Machine learning and multivariate Cox risk regression model showed that age，tumor grade，T stage，M stage，tumor stage and the number of lymph node dissection were the prognostic factors of patients with gallbladder cancer（P<0.05）. Nomogram was established. The area under the ROC curve was 0.825（95%CI 0.810-0.841）.  In the validation group，the area under the ROC curve was 0.844（95%CI 0.819-0.868）. The C-index of internal verification and the calibration chart indicated that the prediction model was moderately accurate and the prediction results is reliable. In patients with stage T3 and T4 gallbladder carcinoma，there was no significant difference in the effect of dissection of ≥4 lymph nodes and 1-3 lymph nodes on prognosis (P=0.538). In patients with stage T1 and T2 gallbladder carcinoma，the prognosis of ≥4 lymph nodes dissected was better than that of 1-3 lymph nodes dissected，and the difference was statistically significant（P<0.01）. Conclusion    The nomogram prognostic model has been proved to have high accuracy and can provide a reference for evaluating the prognosis and survival rate of patients with gallbladder cancer. Patients with stage T1 and T2 gallbladder cancer have more benefits than patients with stage T3 and T4 by increasing the number of lymph node dissection.

Key words: gallbladder cancer, nomogram, lymph node dissection, surveillance, epidemiology, and eng results program