Developing a predictive model and uncovering immune influences on prognosis for brain metastasis from lung carcinomas.

OBJECTIVE

Primary lung carcinomas (LCs) often metastasize to the brain, resulting in a grim prognosis for affected individuals. This population-based study aimed to investigate their survival period and immune status, while also establishing a predictive model.

METHODS

The records of 86,763 primary LCs from the Surveillance, Epidemiology, and End Results (SEER) database were extracted, including 15,180 cases with brain metastasis (BM) and 71,583 without BM. Univariate and multivariate Cox regression were employed to construct a prediction model. Multiple machine learning methods were applied to validate the model. Flow cytometry and ELISA were used to explore the immune status in a real-world cohort.

RESULTS

The research findings revealed a 17.49% prevalence of BM from LCs, with a median survival of 8 months, compared with 16 months for their counterparts (0.001). A nomogram was developed to predict survival at 1, 3, and 5 years on the basis of these variables, with the time-dependent area under the curve (AUC) of 0.857, 0.814, and 0.786, respectively. Moreover, several machine learning approaches have further verified the reliability of this model's performance. Flow cytometry and ELISA analysis suggested the prediction model was related the immune status.

CONCLUSIONS

BM from LCs have an inferior prognosis. Considering the substantial impact of these factors, the nomogram model is a valuable tool for guiding clinical decision-making in managing patients with this condition.