Oncotarget

Research Papers:

A novel classifier based on three preoperative tumor markers predicting the cancer-specific survival of gastric cancer (CEA, CA19-9 and CA72-4)

Jing Guo, Shangxiang Chen, Shun Li, Xiaowei Sun, Wei Li, Zhiwei Zhou, Yingbo Chen and Dazhi Xu _

PDF  |  HTML  |  How to cite  |  Order a Reprint

Oncotarget. 2018; 9:4814-4822. https://doi.org/10.18632/oncotarget.23307

Metrics: PDF 1535 views  |   HTML 1840 views  |   ?  


Abstract

Jing Guo1,2,*, Shangxiang Chen1,2,*, Shun Li1,2,*, Xiaowei Sun1,2, Wei Li1,2, Zhiwei Zhou1,2, Yingbo Chen1,2 and Dazhi Xu1,2

1State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangzhou, China

2Department of Gastric Surgery, Sun Yat-sen University Cancer Center, Guangzhou, China

*These authors have contributed equally to this work

Correspondence to:

Dazhi Xu, email: xudzh@sysucc.org.cn

Keywords: tumor marker; CEA; CA19-9; CA72-4; gastric cancer

Received: July 07, 2017     Accepted: December 01, 2017     Published: December 14, 2017

ABSTRACT

Several studies have highlighted the prognostic value of the individual and the various combinations of the tumor markers for gastric cancer (GC). Our study was designed to assess establish a new novel model incorporating carcino-embryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9), carbohydrate antigen 72-4 (CA72-4). A total of 1,566 GC patients (Primary cohort) between Jan 2000 and July 2013 were analyzed. The Primary cohort was randomly divided into Training set (n=783) and Validation set (n=783). A three-tumor marker classifier was developed in the Training set and validated in the Validation set by multivariate regression and risk-score analysis. We have identified a three-tumor marker classifier (including CEA, CA19-9 and CA72-4) for the cancer specific survival (CSS) of GC (p<0.001). Consistent results were obtained in the both Training set and Validation set. Multivariate analysis showed that the classifier was an independent predictor of GC (All p value <0.001 in the Training set, Validation set and Primary cohort). Furthermore, when the leave-one-out approach was performed, the classifier showed superior predictive value to the individual or two of them (with the highest AUC (Area Under Curve); 0.618 for the Training set, and 0.625 for the Validation set), which ascertained its predictive value. Our three-tumor marker classifier is closely associated with the CSS of GC and may serve as a novel model for future decisions concerning treatments.


Creative Commons License All site content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 License.
PII: 23307