国际肿瘤学杂志 ›› 2023, Vol. 50 ›› Issue (8): 463-469.doi: 10.3760/cma.j.cn371439-20230426-00089

• 论著 • 上一篇    下一篇

基于列线图模型探讨血清学指标对鼻咽癌预后的判断价值

王潇1, 李盈2, 罗玉杰1, 晋舒1()   

  1. 1 四川省资阳市第一人民医院耳鼻咽喉头颈外科,资阳 641300
    2 四川省资阳市第一人民医院眼科,资阳 641300
  • 收稿日期:2023-04-26 修回日期:2023-06-16 出版日期:2023-08-08 发布日期:2023-10-24
  • 通讯作者: 晋舒 E-mail:udhdudh674@163.com

Study on the prognostic value of serological indicators for nasopharyngeal carcinoma based on nomogram model

Wang Xiao1, Li Ying2, Luo Yujie1, Jin Shu1()   

  1. 1 Department of Otolaryngology Head and Neck Surgery, First People's Hospital of Ziyang of Sichuan Province, Ziyang 641300, China
    2 Department of Ophthalmology, First People's Hospital of Ziyang of Sichuan Province, Ziyang 641300, China
  • Received:2023-04-26 Revised:2023-06-16 Online:2023-08-08 Published:2023-10-24
  • Contact: Jin Shu E-mail:udhdudh674@163.com

摘要: 目的 基于列线图模型探讨血清学指标对鼻咽癌预后的判断价值。方法 回顾性分析2020年1月至2021年12月在四川省资阳市第一人民医院首次接受治疗的231例鼻咽癌患者的临床资料,按照3∶1的比例采用随机数字表法将入组患者随机分为训练队列(n=173)和验证队列(n=58)。在训练队列中使用最小绝对收缩和选择算子(Lasso)Cox回归模型生成一个预后模型,使用一致性指数(C指数)和决策曲线分析(DCA)比较新预后模型与TNM分期和EB病毒(EBV) DNA拷贝数的预测准确性。在此基础上,构建包含预后模型、TNM分期和EBV DNA拷贝数的总生存期(OS)列线图,用于预测鼻咽癌预后。结果 训练队列和验证队列的中位随访时间分别为31.4个月、30.4个月。训练队列12个月、24个月和36个月OS率分别为97.4%、83.8%和63.3%;验证队列分别为94.2%、84.4%和72.8%。两组间临床特征和血清学指标的分布差异均无统计学意义(均P>0.05)。建立的预后模型包括如下变量:年龄、体重指数、血小板计数、淋巴细胞-单核细胞比值、C-反应蛋白(CRP)、CRP-白蛋白比值、球蛋白值、白蛋白-球蛋白比值和预后营养指数。在训练队列中,预后模型的C指数为0.790(95%CI为0.730~0.852),显著高于EBV DNA拷贝数(0.693,95%CI为0.625~0.760)(χ2=5.74,P=0.028),但与TNM分期(0.738,95%CI为0.688~0.786)差异无统计学意义(χ2=2.32,P=0.074)。在验证队列中,预后模型的C指数(0.797,95%CI为0.742~0.874)显著高于EBV DNA拷贝数(0.636,95%CI为0.559~0.713)(χ2=6.03,P=0.024),但与TNM分期(0.755,95%CI为0.705~0.815)差异无统计学意义(χ2=1.74,P=0.121)。DCA显示,在训练队列和验证队列中,预后模型比EBV DNA拷贝数具有更好的总体净效益,并与TNM分期相当。校准曲线显示,列线图预测训练队列和验证队列的12个月、24个月和36个月OS率与实际观察结果一致。 结论 本研究建立的包括血清学指标的预后模型比传统的TNM分期和单独的EBV DNA拷贝数显示出更好的鼻咽癌预后预测准确性。基于预后模型、TNM分期和EBV DNA拷贝数构建的列线图可以增强预后模型的预测能力。

关键词: 鼻咽癌, 列线图, 预后, 模型

Abstract: Objective To explore the prognostic value of serological indicators for nasopharyngeal carcinoma based on nomogram model. Methods The clinal data of 231 patients diagnosed as nasopharyngeal carcinoma who were treated for the first time in the First People's Hospital of Ziyang of Sichuan Province from January 2020 to December 2021 were studied retrospectively. The patients were randomly divided into training group (n=173) and verification group (n=58) according to the ratio of 3∶1 by random number table method. Lasso Cox regression model was used to generate a prognosis model in the training cohort. Consistency index (C index) and decision curve analysis (DCA) were used to compare the prediction accuracy between the new prognosis model and TNM staging and Epstein-Barr virus (EBV) DNA copy number. Subsequently, an overall survival (OS) nomogram including prognosis model, TNM staging and EBV DNA copy number was established to predict the prognosis of nasopharyngeal carcinoma. Results The median follow-up time of the training cohort was 31.4 months, and that of the verification cohort was 30.4 months. In the training cohort, the 12-month, 24-month and 36-month OS rates were 97.4%, 83.8% and 63.3% respectively. In the verification cohort, the 12-month, 24-month and 36-month OS rates were 94.2%, 84.4% and 72.8% respectively. There was no significant difference in the distribution of clinical features and serological indicators between the training cohort and the verification cohort(all P>0.05). The established prognostic model included the following variables: age, body mass index, platelet, lymphocyte/monocyte ratio, C-reactive protein(CRP), CRP/albumin ratio, globulin, albumin/globulin ratio and prognostic nutrition index. In the training cohort, the C index of the prognostic model was 0.790 (95%CI: 0.730-0.852), which was significantly higher than that of EBV DNA copy number (0.693, 95%CI: 0.625-0.760) (χ2=5.74, P=0.028), but there was no significant difference compared to TNM staging (0.738, 95%CI: 0.688-0.786)(χ2=2.32, P=0.074). In the validation cohort, the C index of the prognostic model (0.797, 95%CI: 0.742-0.874) was significantly higher than the DNA copy number of EBV (0.636, 95%CI: 0.559-0.713) (χ2=6.03, P=0.024), but there was no significant difference compared to TNM staging (0.755, 95%CI: 0.705-0.815)(χ2=1.74, P=0.121). DCA showed that the prognostic model had a better overall net benefit than EBV DNA copy number and was comparable to TNM staging in both the training and validation cohorts. The calibration curve of the training cohort and the verification cohort showed that the 12-month, 24-month and 36-month OS rates predicted by the nomograms were consistent with actual observations. Conclusion The prognosis model established in this study, including serological indicators, shows better accuracy in predicting the prognosis of nasopharyngeal carcinoma than traditional TNM staging and EBV DNA copy number alone. The nomogram based on prognosis model, TNM staging and EBV DNA can enhance the predictive ability of prognosis model.

Key words: Nasopharyngeal carcinoma, Nomograms, Prognosis, Model