Establishing the diagnostic model of SCC in cervical cancer by using Logistic regression combined with CHAID analysis of decision tree / 中华检验医学杂志

ABSTRACT

Objective To explore the relationship between serum tumor markers and cervical cancer by using Logistic regression, and to further establish the diagnosis model of squamous cell carcinoma antigen (SCC) in cervical cancer by using chi-squared automatic interaction detector (CHAID) analysis of decision tree.Methods Total of 581 cases of cervical cancer,342 cases of cervical benign diseases and 341 cases of healthy controls who detected tumor markers in Taizhou Hospital of Zhejiang during 2010-2013, were retrospectively studied.The test results of carbohydrate antigen 199 (CA199), carbohydrate antigen 125 (CA125), carcinoembryonic antigen (CEA), SCC, and alpha fetoprotein (AFP) were reviewed.The Logistic regression were firstly used in screening the related tumor markers of cervical cancer, and then the CHAID method of decision tree was used in determining the values of the related tumor markers on the diagnosis of cervical cancer.The SCC elevated cases of uterine disease from January 2014 to December 2014 were collected to verify the diagnostic value of SCC in cervical cancer patients.Results Logistic regression analysis showed that among the 5 tumor markers which might be associated with cervical cancer, SCC was the only one which had statistical significance between cervical cancer and cervical benign diseases (Wald x2 =22.120,P =0.000), the OR and its 95％ CI was 1.900 (1.454-2.483).With the SCC numerical increases, the proportion of patients with cervical cancer also gradually increased, when SCC ＞ 2.20 μg/L, the positive predictive value was 94.7％.In 284 cases of SCC higher than 2.20 μg/L who considered as uterus related diseases, there were 270 cases (95.1％) who were eventually confirmed as cervical cancer.Conclusion There was a good diagnostic value of SCC for cervical cancer patients.