胰腺癌患者预后预测动态在线列线图的构建及应用
作者:
作者单位:

1.兰州大学第二临床医学院,甘肃 兰州 730030;2.兰州大学第一临床医学院,甘肃 兰州 730000;3.兰州大学口腔 医学院,甘肃 兰州 730000;4.兰州大学第二医院 普通外科,甘肃 兰州 730030

作者简介:

施华清,兰州大学第二临床医学院博士研究生,主要从事肝胆胰肿瘤方面研究。

通信作者:

周文策,Email: zhouwc129@163.com

基金项目:

甘肃省中医药科研课题基金资助项目(GZKP-2020-28);甘肃省兰州市城关区科技计划基金资助项目(2020-2-11-4)。


Construction and application of online dynamic nomogram for predicting prognosis of pancreatic cancer patients
Author:
Affiliation:

1.The Second Clinical Medical College, Lanzhou University, Lanzhou 730030, China;2.The First Clinical Medical College, Lanzhou University, Lanzhou 730000, China;3.School of Stomatology, Lanzhou University, Lanzhou 730000, China;4.Department of General Surgery, Lanzhou University Second Hospital, Lanzhou 730030, China

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 音频文件
  • |
  • 视频文件
    摘要:

    背景与目的 胰腺癌具有高度侵袭性,患者的预后很差,与其他癌症不同,在过去的几年中,胰腺癌的发病率继续增加,存活率几乎没有提高。目前临床上使用的TNM分期系统来评估患者预后指标较为单一。因此,本研究的目的是构建一个动态的在线列线图,用于预测胰腺癌患者预后,为临床个体化治疗提供参考。方法 从SEER数据库中提取了2000—2018年病理确诊为胰腺癌的患者信息,并按7∶3的比例随机分为训练队列与验证队列。采用单因素和多因素的Cox回归分析来确定预后风险因素,并使用R软件构建动态在线列线图。使用C-指数、与时间相关ROC曲线的曲线下面积(AUC)、校准曲线和决策曲线分析(DCA)来评估列线图的临床效用。根据列线图再将患者分为高风险组和低风险组,通过Kaplan-Meier生存曲线比较两组患者的预后。结果 共在SEER数据库中筛选出12 175例胰腺癌患者,年龄、肿瘤分化程度、原发部位、T分期、N分期、M分期、手术、化疗和肿瘤大小是总生存期(OS)的独立影响因素(均P<0.05)。在训练队列中,与OS相关列线图的C-指数为0.759(95% CI=0.745~0.772),预测1、3、5年OS的AUC分别为0.828、0.842和0.849。在验证队列中,C-指数为0.756(95% CI=0.735~0.776),1、3、5年OS的AUC分别为0.820、0.831和0.842。校准图和DCA曲线显示了该模型在训练和验证队列中有良好预测性能。Kaplan-Meier生存曲线显示,在验证集和训练集中,低风险组患者的总OS优于高风险组(均P<0.05)。结论 建立的动态在线列线图有良好预测性能,有助于个性化结合临床患者实际情况综合预测胰腺癌患者的预后,并可能比TNM分期系统具有更好的临床应用价值。

    Abstract:

    Background and Aims Pancreatic cancer is a highly aggressive malignancy and patients with pancreatic cancer will face a dismal prognosis. Unlike other cancers, the incidence of pancreatic cancer has continued to increase over the past few years with little improvement in survival rates. The prognostic indicators of the TNM staging system currently used in clinical practice to assess the prognosis of patients are relatively limited. Therefore, this study was designed to construct a dynamic online nomogram for clinical prediction of prognosis of pancreatic cancer patients, so as to provide guidance for clinical individualized treatment.Methods Information of patients with pancreatic cancer from 2000 to 2018 was extracted from the SEER database, and patients were randomly allocated to the training cohort and validation cohort at a ratio of 7∶3. Univariate and multivariate Cox regression analyses were used to identify the prognostic risk factors, and dynamic online nomogram was constructed using R software. The C-index, area under the curve (AUC) of time-dependent ROC curves, calibration plot, and decision curve analysis (DCA) was used to assess the clinical utility of the nomogram. Finally, the pancreatic cancer patients were divided into high-risk and low-risk groups according to the nomogram, and the prognostic results of the two groups of patients were compared by Kaplan-Meier survival curves.Results A total of 12 175 patients with pancreatic cancer were screened. Age, degree of tumor differentiation, primary tumor site, T stage, N stage, M stage, surgery, chemotherapy, and tumor size were independent influencing factors for OS (all P<0.05). In the training cohort, the C-index for the OS nomogram was 0.759 (95% CI=0.745-0.772), and the AUC values for predicting the 1-, 3- and 5-year OS were 0.828, 0.842, and 0.849, respectively. In the validation cohort, the C-index was 0.756 (95% CI=0.735-0.776), and the AUC values for predicting the 1-, 3- and 5-year OS were 0.820, 0.831, and 0.842, respectively. The calibration plot and DCA curves demonstrated good prediction performance of the model in both the training and validation cohorts. Results of Kaplan-Meier survival curves showed that the overall OS of patients in the low-risk group was superior to that of patients in the high-risk group in either the validation cohort or training cohort (both P<0.05).Conclusion The established dynamic online nomogram has a good prediction efficiency and it is helpful for comprehensive prediction of the prognosis of pancreatic cancer patients by a personalized combination of the actual clinical situation of patients. Moreover, the nomogram may have a better clinical application value than the TNM staging system.

    表 2 胰腺癌患者OS的单因素和多因素Cox回归分析Table 2 Univariate and multivariate Cox regression model of OS in pancreatic cancer patients
    表 3 胰腺癌患者OS的单因素和多因素Cox回归分析(续)Table 3 Univariate and multivariate Cox regression model of OS in pancreatic cancer patients (continued)
    图1 预测胰腺癌患者OS概率的列线图 A:通过训练队列中结合9个变量构建的预测胰腺癌患者的1、3、5年OS的列线图;B:在线动态列线图截图Fig.1 Nomogram for predicting OS in patients with pancreatic cancer A: The nomogram developed in the training cohort by combining 9 variables for predicting the 1-, 3-, and 5-year OS in patients with pancreatic cancer; B: Screenshot of the online dynamic nomogram
    图2 时间依赖性ROC曲线 A-C:训练队列中列线图预测1、3、5年OS的ROC曲线;D-F:验证队列中列线图预测1、3、5年OS的ROC曲线Fig.2 Time-dependent ROC curves A-C: ROC curves of the nomogram for predicting 1-, 3- and 5-year OS in training cohort; D-F: ROC curves of the nomogram for predicting 1-, 3- and 5-year OS in validation cohort
    图3 训练队列和验证队列的校准曲线 A-C:训练队列中预测1、3、5年OS列线图的校准曲线;D-F:验证队列中预测1、3、5年OS列线图的校准曲线Fig.3 Calibration curves in the training cohort and validation cohort A-C: Calibration curves of the nomogram for predicting 1-, 3- and 5-year OS in training cohort; D-F: Calibration curves of the nomogram for predicting 1-, 3- and 5-year OS in validation cohort
    图4 列线图的DCA曲线,以及AJCC TNM分期系统 A-C:在训练队列中预测1、3、5年OS的DCA曲线;D-F:在验证队列中预测1、3、5年OS的DCA曲线Fig.4 DCA curves of the nomogram, and AJCC stage system A-C: DCA curves for predicting 1-, 3- and 5-year OS in training cohort; D-F: DCA curves for predicting 1-, 3- and 5-year OS in validation cohort
    图5 高风险组和低风险组的Kaplan-Meier生存曲线 A:训练队列中OS的Kaplan-Meier生存曲线;B:验证队列中OS的Kaplan-Meier生存曲线Fig.5 Kaplan-Meier survival curves for the high- and low-risk groups A: Kaplan-Meier curve for OS in training cohort; B: Kaplan-Meier curve for OS in the validation cohort
    图1 预测胰腺癌患者OS概率的列线图 A:通过训练队列中结合9个变量构建的预测胰腺癌患者的1、3、5年OS的列线图;B:在线动态列线图截图Fig.1 Nomogram for predicting OS in patients with pancreatic cancer A: The nomogram developed in the training cohort by combining 9 variables for predicting the 1-, 3-, and 5-year OS in patients with pancreatic cancer; B: Screenshot of the online dynamic nomogram
    图2 时间依赖性ROC曲线 A-C:训练队列中列线图预测1、3、5年OS的ROC曲线;D-F:验证队列中列线图预测1、3、5年OS的ROC曲线Fig.2 Time-dependent ROC curves A-C: ROC curves of the nomogram for predicting 1-, 3- and 5-year OS in training cohort; D-F: ROC curves of the nomogram for predicting 1-, 3- and 5-year OS in validation cohort
    图3 训练队列和验证队列的校准曲线 A-C:训练队列中预测1、3、5年OS列线图的校准曲线;D-F:验证队列中预测1、3、5年OS列线图的校准曲线Fig.3 Calibration curves in the training cohort and validation cohort A-C: Calibration curves of the nomogram for predicting 1-, 3- and 5-year OS in training cohort; D-F: Calibration curves of the nomogram for predicting 1-, 3- and 5-year OS in validation cohort
    图4 列线图的DCA曲线,以及AJCC TNM分期系统 A-C:在训练队列中预测1、3、5年OS的DCA曲线;D-F:在验证队列中预测1、3、5年OS的DCA曲线Fig.4 DCA curves of the nomogram, and AJCC stage system A-C: DCA curves for predicting 1-, 3- and 5-year OS in training cohort; D-F: DCA curves for predicting 1-, 3- and 5-year OS in validation cohort
    图5 高风险组和低风险组的Kaplan-Meier生存曲线 A:训练队列中OS的Kaplan-Meier生存曲线;B:验证队列中OS的Kaplan-Meier生存曲线Fig.5 Kaplan-Meier survival curves for the high- and low-risk groups A: Kaplan-Meier curve for OS in training cohort; B: Kaplan-Meier curve for OS in the validation cohort
    表 1 12 175例胰腺癌患者的人口统计学和临床数据[n(%)]Table 1 Demographic and clinical data of the patients with 12 175 pancreatic cancers [n (%)]
    参考文献
    相似文献
    引证文献
引用本文

施华清,柴长鹏,陈洲,董仕,何茹,秦子顺,周文策.胰腺癌患者预后预测动态在线列线图的构建及应用[J].中国普通外科杂志,2022,31(9):1162-1172.
DOI:10.7659/j. issn.1005-6947.2022.09.005

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
历史
  • 收稿日期:2022-04-13
  • 最后修改日期:2022-08-04
  • 录用日期:
  • 在线发布日期: 2022-09-30