Diagnostic Value of Serum CA72-4,CA242,CA19-9 and CEA for Gastric Cancer Based on PCA-Decision Tree Model / 现代检验医学杂志

ABSTRACT

Objective To evaluate the diagnostic value of serum tumors CA72-4,CA242,CA19-9 and carcino-embryonic anti-gen (CEA)for patients with gastric cancer based on principle component analysis (PCA)-decision tree analysis.Methods Serum levels of CA72-4,CA242,CA19-9 and CEA in 193 patients with gastric cancer,106 patients with benign gastric disea-ses and 86 nornal controls were measured by electrochemiluminescence assay,and data were analyzed by the receiver operat-ing characteristic (ROC)curve,PCA and PCA-decision tree analysis.Results The area under the ROC curve of CA72-4, CA242,CA19-9 and CEA was 0.741[95% confidence interval (95%CI),0.692~0.791],0.863 (95%CI,0.827~0.898), 0.783 (95%CI,0.737~0.828)and 0.827 (95%CI,0.785~0.869),respectively.The combined four serum tumor markers in the PCA-AUC model was 0.935 (95%CI,0.912~0.958)at the cutoff value (PC score)of 44.13 with 78.2% of sensi-tivity and 94.8% of specificity.The accuracy of serum CA72-4,CA242,CA19-9 and CEA for the diagnosis of gastric cancer group and nongastric cancer group (benign gastric diseases and nornal controls)in the decision tree model were 76.2% and 94.8%,56.5% and 96.5% for prediction,respectively.The combined four serum tumors for the diagnosis of gastric cancer group and nongastric cancer group in PCA-decision tree model were 90.3% and 100%,72.4% and 92.2% for prediction,re-spectively.Conclusion The PCA-decision tree model based on serum CA72-4,CA242,CA19-9 and CEA were helpful for the diagnosis of gastric cancer.