A clinical model to predict brain metastases in resected early-stage non-small cell lung cancer.

Abstract

Background: Despite the rising diagnosis of early-stage non-small cell lung cancer (NSCLC), there remains a limited understanding of the risk factors associated with postoperative brain metastases in early-stage NSCLC. Our goal was to identify the risk factors and construct a predictive model for postoperative brain metastases in this population.

Methods: This study retrospectively enrolled patients with resected stage I-II NSCLC at the Department of Thoracic Surgery, West China Hospital from January 2015 to January 2021. Risk factors were identified through univariable and multivariable Cox regression analyses, followed by the construction of a nomogram. Evaluation of the model involved metrics such as the area under the curve (AUC), C-index, and calibration curves. To ensure reliability, internal validation was performed through bootstrap resampling.

Results: This study included 2106 patients, among whom 67 (3.18%) patients were diagnosed with postoperative brain metastases. Multivariable Cox regression analysis revealed that higher pT and pN stages, along with specific histological subtypes, particularly solid/micropapillary predominant adenocarcinoma, were identified as independent risk factors for brain metastases. The performance of the nomogram in the training set exhibited AUC values of 0.759, 0.788, and 0.782 for predicting 1-year, 2-year, and 3-year occurrences, respectively. Bootstrap resampling validated its reliability, with C-index values of 0.758, 0.799, and 0.792 for the respective timeframes. Calibration curves affirmed consistency of the model.

Conclusions: A nomogram was developed to predict the likelihood of postoperative brain metastases in individuals with early-stage NSCLC. The tool aids in identifying high-risk patients and facilitating timely interventions.