A diagnostic model to detect silent brain metastases in patients with non-small cell lung cancer.

We aimed to discriminate subgroups according to the risk of brain metastases in patients with non-small cell lung cancer (NSCLC) lacking neurological symptoms. We performed a retrospective review of 433 patients with NSCLC who underwent chest computed tomography (CT), brain magnetic resonance imaging (MRI) and bone scans at an initial staging work-up between April 2003 and April 2007. Brain metastases were determined by MRI. Patients were stratified into groups according to the number of risk factors (0-3) identified by multivariate analysis. In multivariate analysis, histopathology with non-squamous cell carcinoma, nodal stage 2 on CT and presence of bone metastases were three risk factors for brain metastases. Patients were divided into four groups according to the number (0-3) of these predictive factors. The proportions of patients with brain metastases in the four groups were 2%, 3%, 17% and 35%, respectively, and these differences were significant (P<0.001). When analysis was performed in patients with localised disease, the number of risk factors was correlated with the prevalence of brain metastases (P=0.013) but stage was not (P=0.153). Although this diagnostic model should be validated through further studies, our data suggest that the number of risk factors might be a useful tool to identify silent brain metastases in patients with NSCLC.