Establishment of a mathematical prediction model to evaluate the probability of malignancy or benign in patients with solitary pulmonary nodules

OBJECTIVE To evaluate the clinical factors affecting the definite pathological diagnosis of solitary pulmonary nodules (SPN) with multivariate Logistic regression analysis, and to build the clinical prediction model to estimate the probability of malignancy. METHODS A retrospective cohort study in our institution included 371 patients (197 males and 174 females) with definite pathological diagnosis of solitary pulmonary nodules from Jan 2000 to Sep 2009 (group A). Clinical data included age, gender, course of disease, symptoms, history and quantity of smoking history, history of tumor, family history of tumor, site, diameter, calcification, speculation, border, lobulation, traction of pleural, vascular convergence sign, and cavity. The independent predictors of malignancy were estimated with multivariate analysis, then the clinical prediction model was built. Other 62 SPN patients (group B) with definite pathological diagnosis in our institute from Oct 2009 to Mar 2010, were used to validate value of this clinical prediction model. RESULTS 53.1% of the nodules were malignant, and 46.9% were benign in goup A. Logistic regression analysis showed that seven clinical characteristics [age of patient (OR: 1.073), diameter (OR: 1.966), border (OR: 0.245), calcification (OR: 0.199), spiculation (OR: 2.088) and the family history of tumor (OR: 3.550)] were independent predictors of malignancy in patients with SPN (P<0.05). The cut-off value was 0.463. The sensitivity in group B was 92.5%, specificity 81.8%, positive predictive value 90.2%, and negative predictive value 85.7%. The area under the ROC curve for our model was 0.888±0.054. CONCLUSION Age of patient, diameter, border, calcification, spiculation and the family history of tumor are independent predictors of malignancy in patients with SPN. Our prediction model is accurate and sufficient to estimate the malignancy of patients with SPN.