重庆医学2017，Vol.46Issue(8)：1038-1041,4.DOI:10.3969/j.issn.1671-8348.2017.08.010
Logistic回归和PLS-DA模型评价肿瘤标志物对结肠癌的诊断价值
Diagnostic of serum tumor markers for colon cancer assessed with logistic regression analysis and PLS-DA model
摘要
Abstract
Objective To investigate the diagnostic value of serum tumors M2 pyruvate kinase (TuM2-PK),carcino-embryonic antigen (CEA),CA199,CA724,CA125 and CA242 in colon cancer.Methods Serum levels of TuM2-PK,CEA,CA199,CA724,CA125 and CA242 in 231 patients with colon cancer,105 patients with colon benign lesion diseases and 166 normal controls were measured by the ELISA and electrochemiluminescence assays.The operation working characteristic curve (ROC) and the logistic regression-ROC were used to evaluate the diagnostic value of tumor markers for colon cancer individually and in combination.The models of logistic regression analysis and partial least-squares discriminant analysis (PLS-DA) were established to diagnose patients with colon cancer based on the optimal panel of serum tumor markers.Resalts Concentrations of serum TuM2-PK,CEA,CA199,CA125 and CA724 in the colon cancer group are higher than those in the colon benign diseases group and the normal controls (P＜0.05).The area under the operation working characteristic curve (AUC) of CA199 is the highest 0.79 [95％ confidence interval (95％CI),0.75-0.83] at the cutoff value of 69.5 U/L with 64.1％ of sensitivity and 89.7％ of specificity.The AUC of combined serum tumor markers based on logistic regression analysis is higher than those in individuals,of which serum (CEA+ CA199+TuM2-PK) is the optimal [AUC=0.89,95％ confidence interval(95％CI),0.86-0.92].The diagnostic accuracy of logistic analysis and PLS-DA model for colon cancer is 82.7％ and 77.5％,and for colon cancer is 93.7％ and 95.6％,respectively.Conclusion The combination of serum tumor markers CEA,CA199 and TuM2-PK is more suitable as a diagnostic model for the screening of colon cancer.关键词
结肠肿瘤/肿瘤M2型丙酮酸激酶/ROC曲线/Logistic回归分析Key words
colon cancer/tumor M2 pyruvate kinase/receiver operating characteristic curve/logistic regression analysis分类
医药卫生引用本文复制引用
曾龙飞,田刚,姚健..Logistic回归和PLS-DA模型评价肿瘤标志物对结肠癌的诊断价值[J].重庆医学,2017,46(8):1038-1041,4.基金项目
四川省教育厅重点项目(2005A076) （2005A076）
四川省泸州市科技计划项目[2015-S-51(1/3)] （1/3）
西南医科大学自然科学基金(30543) （30543）
西南医科大学附属医院人才基金资助项目(12277). （12277）
