A Mathematical Model For Predicting Malignancy Of Solitary Pulmonary Nodules

Part I Establish a mathematical model to predicting malignancy of Solitary Pulmonary Nodules[Abstract] Objective:To evulate the clinical factors affecting the pathological diagnosis of solitary pulmonary nodules(SPN) with multivariate Logistic regression analysis, and to establish a mathematical model to pretest the probability of malignancy.Methods:A retrospective study in Fudan University Cancer Hospital included200patients (91males and109females) with definite pathological diagnosis of solitary pulmonary nodules from Janaury2011to December2011(group A). Clinical data included9items:age, gender, course, symptom, smoking history, smoking index, number of years since quitting smoking, family history, previous cancer history(a diagnosis of cancer within5yeasrs prior to the detection of the nodule were excleded), radiological date included9items:diamter (mm), calcification, spiculation(long spiculation, short spiculation), lobution, border, cavity, air bronchogram, pleural retraction sign, postion. The independent predictors of malignant nodules were estimated with multivariate analysis, then the mathematical prediction model was established.Results:32%of the nodules were malignant, and68%were benign in group A. Logstic regression analysis showed that two clinical characteristics[age of patient(OR:1.060),gender(OR:0.299)] and six radiological characteristics[calcification(OR:0.116),spiculation(short spiculation,OR:1.656; long spiculation OR:5.632),lobulation(OR:5.944),border(OR:0.366)]were independent predictors of malignancy in patients with SPN (P<0.05), positive predictive value90.4%, and negative predictive value57.8%, overall prediction accuracy80.0%. The area under the ROC curve for our model was0.842±0.030,95%CI:0.784-0.900.Conclusion:Age of patient, gender, calcification, spiculation, lobulation, border are independent predictors of malignancy in patients with SPN. Our prediction model is sufficient and accurate to pretest the malignancy of patients with SPN. Part Ⅱ Part Ⅱ verification and comparison of the the mathematical prediction model to predicting malignancy of Solitary Pulmonary Nodules[Abstract] Objective:This part was designed from the clinical cases and explore the diagnostic value in differernt lung cancer mathematical prediction models, in order to guide the clinical diagnosis and treatment more widely.Methods:Other89SPN patients (group B) with definite pathological diagnosis in Fudan University Cancer Hospital from January2012to June2012, added with the clinical and imaging characteristics of the patients, were used to valiadate clinical value of our mathematical prediction model、Mayo Clinical model、VA model and the domestic model.Results:89patients with SPN (group B) were malignant in71, benign in18. The area under the curve of our mathematical prediction model was0.888, which is greater than another domestic mathematical prediction model (0.773)、VA model (0.729) and Mayo Clinical model (0.701). Our mathematical model specificity was94.4%>Mayo Clinical(88.9%)> VA model model(72.2%)> domestic model (66.7%), domestic model had the highest sensitivity (88.7%)> our mathematical model (83.1%)>VA model (78.9%)> Mayo Clinical model (45.1%), P<0.05, the difference was statistically significant. The mathematical model built in our study had colleted the most comprehensive clinical data and imaging data, and all information consisted of Chinese people, which is better than foreign formula applied mechanically.Conclusion:1.The pre-established mathematical prediction model in our study has a high clinical value for diagnosis.2.The accuracy of our prediction model is more exacter than the foreign mathematical prediction models.