A Prognostic Nomogram via Competing Risk Model with Patterns of Extracranial Metastasis in Elderly NSCLC Patients with Synchronous Brain-metastasis: A Large Population-based Study

Abstract

Introduction: To determine whether the patterns of extracranial metastasis (ECMs) provide supplementary prognositc information to DS-GPA in elderly NSCLC patients with synchronous BM. Methods: This study included 4974 NSCLC patients with initial BM diagnosed from 2010 to 2015 using the Surveillance Epidemiology and End Results (SEER) program. Patients were divided randomly into training and hold-out test sets. Patterns of ECMs were established based on the difference of survival via competing risk analysis in the training set. A nomogram prediction of 6-month, 12-month, and 18-month disease-specific survival (DSS) was built using independent prognostic factors. Results: Three patterns of ECM were recognized: MA (neither liver, bone, nor lung involvement), MB (without liver involvement), and MC (with liver involvement). Comparing MA, MB and MC showed significant correlation to survival (SHR, 1.126, 95% CI, 1.053-1.205, P<0.001; SHR, 1.46, 95% CI, 1.339-1.592, P<0.001, respectively). In the hold-out test set, the AUC of the ROC curve for the 6-month DSS prediction reached 0.778, whereas reaching 0.757 in the training set. The calibration curves did not deviate from the reference line. Decision curve analyses revealed the net benefit of the nomogram for clinical utility. Conclusions: These results help clinicians make decisions for brain-metastatic NSCLC in the era of precision therapy. The risk stratification of extracranial involvements indicates differential treatment for elderly NSCLC patients with synchronous brain-metastasis.