The Study On The Value Using Six Different Types Of Tumor Markers For The Diagnosis Of Colon Cancer

Objective:Using statistical methods to determine the value of the diagnosis of colon cancer by detecting the concentration of CEA,CA199,CA724,CA125,CA242,and Tum2-pk in the serum of cancer patients,and eventully to identify the optimal individual and joint tumor marker(s)for diagnosis.And to establish the principal component analysis(PCA),classic discriminant analysis,Logistic regression analysis and partial least squares discriminant analysis(PLS-DA)model using the tactics of metabonomics based on these six markers,and to figure out the most favorable model for the diagnosis of early colon cancer by assessing the value of diagnosis and prediction of this malignancy in different models.Methods:To collect the peripheral blood from patients with colon cancer,benign lesions in the colon and the healthy control group,and detecting the concentration of CEA,CA199,CA724,CA242 and CA125 in the serum with the electrochemical luminescence method.Detecting the concentration of Tu M2-PK with enzyme-linked immunosorbent assay(ELISA).Group comparison using single factor analysis of variance,comparison between the(further significant Difference),LSD test and area under the receiver-operating characteristic curve(ROC)(AUC)is a single index of diagnosis efficiency of Logistic regression-ROC to assess the diagnostic performance of jointindicators.Using the PCA model to analyze the metabolism spectrum of tumor markers in the serum of colon cancer,colon benign disease and healthy controls.Using classic discriminant analysis,Logistic regression analysis and partial least squares discriminant analysis(PLS-DA)model to assess the value of diagnosis and prediction of colon cancer.Results:The serum levels of CEA,CA199,CA125,CA242 and Tu M2-PK in colon cancer patients were significantly higher than that in benign disease and healthy controls,the difference was significant(P < 0.05).The serum levels of CEA,CA199,CA724 and CA242 and Tu M2-PK in benign tumor group were greatly higher than healthy controls(P < 0.05).The diagnosis efficiency of CEA,CA199,CA724 and Tu M2-PK among the six tumor markers is higher(AUC 0.70 or higher).And CA199,with the highest diagnostic performance,possessed the 64.1% sensitivity and 89.7% of the specific degree while having the critical value of 69.5 U/L.The single marker Tu M2-PK in the diagnosis of sensitivity was the same as CEA and CA199 in colon cancer.Among the Joint detection with two markers,the diagnosis efficiency of CA199 and Tu M2-PK was the highest(AUC =0.87),with 77.5% sensitivity and 90.0% of the specific degrees.Among three combined detection of serum tumor markers,CEA,CA199 and Tu M2-PK had the highest efficiency of diagnosis,with 82.3%sensitivity and 91.5% specificity.Among combined detection of four markers,CEA + CA72 4 + Tu M2-PK + CA19-9 possessed the specific degree of the sensitivity of 87.5%,87.5% and 84.5% accuracy.PCA model showed that thelevels of six serum tumor markers in patients with colon cancer significantly disorder,colon cancer scattered individual patients,normal and colon benign patients individual variation was small,three groups of individuals had a tendency to separate.With the highest AUC three tumor markers(CEA,CA199 and Tu M2-PK)classic Fisher discriminant analysis model was established,the model of colon cancer and controls(colon benign diseases and normal control group)in the diagnosis of accuracy respectively 70.1%(162/231)and 81.9%(227/271).To the highest AUC three tumor markers(CEA,CA199 and Tu M2-PK)PLS-DA diagnosis model was established,the model of colon cancer and controls(colon benign disease group and the control group)in the diagnosis of the accuracy of respectively 77.5%(179/231)and 95.6%(259/271).Conclusion:Among the six markers of a single serum tumor,the CA199 was the most valuable early diagnosis of colon cancer.And the combination of the serum tumer markers CEA,CA199 and tum2-pk is more economical and effective combination of markers that can help early diagnosis of colon cancer.Based on the serum tumor markers CEA,CA199 and Tu M2-PK,the establishment of logistic regression and partial least squares discriminant analysis(PLS-DA)diagnosis model was demonstrated as a feasiblely and effectively new model for the diagnosis of early colon cancer.