Analysis and validation of probabilistic models for predicting malignancy in solitary pulmonary nodules in a population in Brazil.

ABSTRACT

OBJECTIVE: To analyze clinical and radiographic findings that influence the pathological diagnosis of solitary pulmonary nodule (SPN) and to compare/validate two probabilistic models for predicting SPN malignancy in patients with SPN in Brazil. METHODS: This was a retrospective study involving 110 patients diagnosed with SPN and submitted to resection of SPN at a tertiary hospital between 2000 and 2009. The clinical characteristics studied were gender, age, presence of systemic comorbidities, history of malignancy prior to the diagnosis of SPN, histopathological diagnosis of SPN, smoking status, smoking history, and time since smoking cessation. The radiological characteristics studied, in relation to the SPN, were presence of spiculated margins, maximum transverse diameter, and anatomical location. Two mathematical models, created in 1997 and 2007, respectively, were used in order to determine the probability of SPN malignancy. RESULTS: We found that SPN malignancy was significantly associated with age (p = 0.006; OR = 5.70 for age > 70 years), spiculated margins (p = 0.001), and maximum diameter of SPN (p = 0.001; OR = 2.62 for diameters > 20 mm). The probabilistic model created in 1997 proved to be superior to that created in 2007-area under the ROC curve (AUC), 0.79 ± 0.44 (95% CI 0.70-0.88) vs. 0.69 ± 0.50 (95% CI 0.59-0.79). CONCLUSIONS: Advanced age, greater maximum SPN diameter, and spiculated margins were significantly associated with the diagnosis of SPN malignancy. Our analysis shows that, although both mathematical models were effective in determining SPN malignancy in our population, the 1997 model was superior.