Establishment and validation of a prediction model for estimating the risk of brain metastasis in patients with small-cell lung cancer

Abstract

Abstract We aimed to identify the predictors of brain metastases (BM) and develop a model to estimate the binary risk of BM in patients with small-cell lung cancer (SCLC). Patients diagnosed with SCLC between 2010 and 2017 were identified from the SEER Database and a logistic regression model was applied to identify the risk factors for BM. Independent predictors were used to establish a BM prediction model, which was evaluated in terms of discrimination, calibration, and clinical usefulness using the area under the receiver operating characteristic curve (AUC), calibration plot, and decision curves. The results were validated using an independent cohort. A total of 39,271 patients with SCLC were randomly assigned to the development and validation cohorts. Multivariate logistic regression analysis revealed age, race, number of malignancies, primary site, laterality, chemotherapy, radiotherapy, surgery, liver metastasis, lung metastasis, and bone metastasis as independent risk factors for BM. These factors were used to establish the BM risk prediction model, which was then visualized as a nomogram that showed good predictive accuracy, calibration, and clinical usefulness (development cohort: AUC, 0.715; 95% confidence interval (CI), 0.705–0.725; validation cohort: AUC, 0.706; 95% CI, 0.696–0.716). The new prediction model can better evaluate the risk of BM in patients with SCLC, thus providing a clinical reference value when making decisions regarding prophylactic cranial irradiation.