The construction and validation of the model for predicting the incidence and prognosis of brain metastasis in lung cancer patients

BACKGROUND: Brain metastasis (BM) causes high morbidity and mortality rates in lung cancer (LC) patients. The present study aims to develop models for predicting the development and prognosis of BM using a large LC cohort. METHODS: A total of 266,522 LC cases diagnosed between 2010 and 2016 were selected from the Surveillance, Epidemiology, and End Results (SEER) Program cohort. Risk factors for developing BM and prognosis were calculated by univariable and multivariable logistic and Cox regression analysis, respectively, and nomograms were constructed based on risk factors. Nomogram performance was evaluated with receiver operating characteristics (ROC) curve, or C-index and calibration curve. RESULTS: The prevalence of BM was 13.33%. Associated factors for developing BM include: advanced age; Asian or Pacific Islander race; uninsured status; primary tumor site; higher T stage; higher N stage; poorly differentiated grade; the presence of lung, liver, and bone metastases; and adenocarcinoma histology. Median overall survival (OS) was 4 months; associated prognosis factors were similar to risk factors plus female gender, unmarried status, and surgery. The calibration curve showed good agreement between predicted and actual probability, and the AUC/C-index was 73.1% (95% CI: 72.6–73.6%) and 0.88 (95% CI: 0.87–0.89) for risk and prognosis predictive models, respectively. CONCLUSIONS: BM was highly developed in LC patients, and homogeneous and heterogeneous factors were found between risk and prognosis for BM. The nomogram showed good performance in predicting BM development and prognosis.