External Validation of a Survival Nomogram for Non-Small Cell Lung Cancer Using the National Cancer Database

Abstract

Survival nomograms offer individualized predictions using a more diverse set of factors than traditional staging measures, including the American Joint Committee on Cancer Tumor Node Metastasis (AJCC TNM) Staging System. A nomogram predicting overall survival (OS) for resected, non-metastatic non-small cell lung cancer (NSCLC) has been previously derived from Asian patients. The present study aims to determine the nomogram's predictive capability in the US using the National Cancer Database (NCDB). This was a retrospective review of adults with resected, non-metastatic NSCLC entered into the NCDB between 2004 and 2012. Concordance indices and calibration plots analyzed discrimination and calibration, respectively. Multivariate analysis was also used. A total of 57,313 patients were included in this study. The predominant histologies were adenocarcinoma (48.2%) and squamous cell carcinoma (31.3%), and patients were diagnosed with stage I-A (38.3%), stage I-B (22.7%), stage II-A (14.2%), stage II-B (11.5%), and stage III-A (13.3%). Median OS was 74 months. 1-, 3- and 5-year OS rates were 89.8% [95% confidence interval (CI) 89.5-90.0%], 71.1% (95% CI 70.7-71.6%), and 55.7% (95% CI 54.7-56.6%), respectively. The nomogram's concordance index (C-index) was 0.804 (95% CI 0.792-0.817). AJCC TNM staging demonstrated higher discrimination (C-index 0.833, 95% CI 0.821-0.840). The nomogram's individualized estimates accurately predicted survival in this patient collective, demonstrating higher discrimination in this population than in the developer's cohorts. However, the generalized survival estimates provided by traditional staging demonstrated superior predictive capability; therefore, AJCC TNM staging should remain the gold standard for the prognostication of resected NSCLC in the US.