Survival predictive nomograms for non-surgical lung squamous-cell carcinoma patients with brain metastases: a SEER-based retrospective cohort study with external validation

Abstract Objectives Brain metastasis (BM) is a devastating complication in patients with lung squamous cell carcinoma (LUSC), often precluding surgical intervention. The majority of LUSC patients with BM are managed non-surgically and face a poor prognosis. There is a lack of validated prognostic tools for this significant patient population. This study aims to fulfill this gap among this cohort. Methods A retrospective cohort of 1,581 non-surgical LUSC patients with BM was identified from the SEER database (2010–2015) and randomly split into training (n=1,106) and internal validation (n=475) cohorts. An external validation cohort consisted of 102 patients from the real world. Univariable and multivariable Cox regression analyses were performed to identify independent prognostic factors for overall survival (OS) and cancer-specific survival (CSS). Nomograms were constructed and validated internally and externally using calibration curves, time-dependent receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and Kaplan-Meier analysis. Results Multivariable analyses identified independent factors for OS and CSS. The constructed nomograms demonstrated relatively high area under the curve (AUC) valuesfor predicting 1-, 2-, and 3-year OS and CSS in the training (>0.74), internal validation (>0.69), and external validation cohorts (>0.76). Calibration curves showed satisfactory agreement between predicted and observed outcomes. DCA confirmed the clinical utility of the nomograms across a range of threshold probabilities. Risk stratification based on nomogram scores effectively distinguished between low-risk and high-risk groups. Conclusion This study developed and validated nomograms for predicting OS and CSS in non-surgical LUSC patients with BM. These tools demonstrate relatively good performance and clinical utility.