关键词: 列线图模型, Logistic回归分析, 肿瘤标志物, 肺癌

Abstract:

Objective Based on clinical characteristics and serum tumor markers，to construct a nomogram model to assist in the differential diagnosis of benign and malignant lung lesions. Methods Totally，1 335 patients with lung cancer（lung cancer group） and 234 patients with benign nodule（benign nodule group） who received surgical treatment in the Third Affiliated Hospital of Kunming Medical University from January 2018 to December 2019 were enrolled and randomly divided into training set and validation set according to 7:3. The clinical data of all patients were collected. Carcinoembryonic antigen（CEA），carbohydrate antigen（CA） 125，CA15-3，CA19-9，CA242，CA72-4，cytokeratin 19 fragment（CYFRA21-1），ferritin（FER），squamous cell carcinoma antigen（SCC-Ag） and neuron-specific enolase（NSE） were determined. Logistic regression analysis was used to screen the indicators with statistical significance，and the nomogram model was constructed. C-index，receiver operating characteristic（ROC） curve，calibration curve and decision curve were used to evaluate the performance of nomogram model. Results In the training set，there was statistical significance in age，CEA，CA125，CA15-3，CA19-9，CYFRA21-1，SCC-Ag and NSE levels between benign nodule group and lung cancer group（P<0.05）. Age，family history of cancer，CEA，CYFRA21-1 and NSE were all independent risk factors for lung cancer [odds ratios（OR） were 1.019，3.243，1.374，1.262 and 1.073，95% confidence intervals（CI） were 1.001-1.037，1.357-7.749，1.225-1.540，1.127-1.412 and 1.024-1.125，respectively，P<0.05]. ROC curve was used to establish the optimal cut-off values of CEA，CYFRA21-1 and NSE for the differential diagnosis of benign nodule and lung cancer. According to the optimal cut-off values，the 3 indicators were transformed into binary categorical variables，and a nomogram model was established combining age and family history of cancer. In the training set and validation set，the C-indexes of nomogram model were 0.816 and 0.843，respectively. The areas under curves（AUC） for differential diagnosis of benign nodule and lung cancer were 0.822 and 0.861，the sensitivities were 67.5% and 65.0%，the specificities were 81.7% and 91.8%，respectively. The nomogram model had a high net benefit rate，and the maximum net benefit rate was 0.78. In the validation set，the AUC of Logistic regression model in the diagnosis of TNM stage Ⅰ，Ⅱ and Ⅲ lung cancer were 0.775，0.843 and 0.911，the sensitivities were 58.2%，78.3% and 83.0%，and the specificities were 87.3%，76.1% and 88.7%，respectively. Conclusions The nomogram model based on clinical characteristics and serum tumor markers has high efficiency in the differential diagnosis of benign and malignant lung lesions.

Key words: Nomogram model, Logistic regression analysis, Tumor marker, Lung cancer