Risk Factors Of Malignant Solitary Pulmonary Nodules And Comparison Of Relevant Clinical Prediction Model

Background:With the spread of CT and popularization of people's awareness of health,more and more lung nodules or masses have been detected.For tumors larger than 3cm,the degree of malignancy is significantly higher than that of less than 3cm,which doctors and patients will take more active treatment.However,for solitary pulmonary nodules(SPNs)less than 3cm,the judgment of benign and malignant has brought us many difficulties,such as excessive psychological burden,over-medical treatment and high medical expenses.Currently dealing with solitary pulmonary nodules depends largely on the doctor's experience combined with laboratory tests to judge benign and malignant.Different doctors may treat the same solitary pulmonary nodules differently.In the time of precision medicine,we need more objective guidelines and mathematical prediction models came into being.Objective:To analyze the independent risk factors of solitary pulmonary nodules,and compare the diagnostic efficeiency of Mayo model and PKUPH model,and finally help doctors to better deal with the SPNs.Methods:245 SPNs with definite pathological diagnosis who had an operation were collected in the first affiliated hospital of NANCHANG UNIVERSITY from January2017 to December 2018,we retrospectively collected the data of clinical information(such as age,gender,symptom,smoking history,previous history of tumor,diabetes and family history of cancer),imaging features(such as maximum nodule diameter,density,lesion position,clear border,lobulation,spiculation,pleural retraction sign,vacuole sign,calcification,etc.),serum tumor marker(CEA,CA125,CA199,NSE,CYFRA21-1)and pathological type(NSCLC,SCLC,tuberculosis,cryptococcosis,etc.).We divided them into two groups according to the benign SPNs and malignant SPNs,then used SPSS 19.0 software to analyze those data and compared Mayo model with PKUPH model by calculating the area under the ROC curve.Results:Univariate analysis showed that there were significant differences between the benign and malignant SPNs in age,CEA,nodule size,previous history of tumor more than 5 years,symptom,diabetes,density,clear border,lobulation,spiculation,pleural retraction sign,vacuole sign,bronchotomy sign,calcification and enhancement.Multivariate logistic regression analysis showed that age,CEA level,diameter,pulmonary symptoms,diabetes,lobulation,calcification,pleural retraction sign,enhancement and density were independent predictors of benign and malignant solitary pulmonary nodules.The SPN malignant probability prediction model established in this study was P=e~x/(1+e~x),where e is the natural logarithm,X=-10.583-5.388*density+3.67*enhancement-9.164*calcification+3.317*pleural retraction sign+2.489*lobulation+2.735*diameter+0.794*CEA-3.806*diabetes-5.507*symptoms+0.103*age,the area under the ROC curve of this study was 0.977(95%CI0.962-0.993),the cut-off value was 0.222,the sensitivity and specificity were 0.971and specificity 0.887,respectively.The area under the ROC curve for PKUPH model was 0.825(95%CI0.772-0.878),for Mayo model was 0.745(95%CI 0.683-0.807),there were significant difference between the two models(P<0.05).Conclusion:(1)Age,CEA level,diameter,pulmonary symptoms,diabetes,lobulation,calcification,pleural retraction sign,enhancement and density are independent risk factors of malignant SPNs;(2)The diagnostic efficeiency of PKUPH model is better than Mayo model.