A nomogram prediction model for brain metastases in patients with lung adenocarcinoma

Patients with lung adenocarcinoma (LUAD) with brain metastasis (BM) frequently present without central nervous system (CNS) symptoms at initial diagnosis, underscoring the clinical need for early predictive tools. This study aimed to evaluate the role of epidermal growth factor receptor (EGFR) mutation subtypes in BM risk and to develop a nomogram prediction model combining EGFR mutation status with serum biomarkers in patients with LUAD. A retrospective cohort of 615 patients with LUAD-401 with BM and 214 without BM, matched by age and sex-was recruited from a single center in China between June 2021 and June 2024. Logistic regression, receiver operating characteristic (ROC) curve analysis and nomogram modeling were performed, with internal validation via 10-fold cross-validation. Although the overall distribution of EGFR mutations did not differ significantly between BM and non-BM groups, the proportion of exon 19 deletion (19del) mutations was significantly higher in the BM group (P=0.013), with an odds ratio of 2.10 in the univariate analysis, and 19del was confirmed as an independent risk factor in the multivariate analysis. Serum concentrations of neuron-specific enolase, CEA, CA125, CA153 and cytokeratin 19 fragment were markedly elevated in patients with BM (all P<0.0001); ROC analyses stratified by 19del mutation status demonstrated further improvement in predictive performance for selected biomarkers. A nomogram integrating 19del mutation status with the five serum biomarkers achieved an area under the ROC curve (AUC) of 0.835 in the training cohort and a cross-validated AUC of 0.814, with good calibration confirmed by the Hosmer-Lemeshow test (P=0.151). These findings indicate that LUAD patients harboring EGFR 19del mutations have an elevated BM risk and the combined nomogram provides an effective tool for BM risk stratification in clinical practice.