Background
There is variability in the prognosis of stage III-N2 lung adenocarcinoma (LUAD) patients. The current tumor-node-metastasis (TNM) staging is not sufficient to precisely estimate the prognosis of stage III-N2 LUAD patients. The Surveillance, Epidemiology, and End Results (SEER) database collected first-hand information from a large number of LUAD patients. Based on the SEER database, this study aimed to determine the prognostic factors that affect overall survival (OS) in stage III-N2 LUAD patients and then establish a nomogram for predicting OS in this type of cancer to identify the high-risk population that may require more frequent surveillance or intensive care.
Methods
Data for 1,844 stage III-N2 primary LUAD patients who were registered between 2010 and 2015 were obtained from the SEER database. These patients were randomly assigned to either training (n=1,290) or validation (n=554) cohorts at a 7:3 ratio. The univariate and multivariate Cox regression (UCR and MCR) analyses were performed to find the relevant independent prognostic factors. To predict the OS based on these prognostic factors, a nomogram was then developed. The performance of the nomogram was examined based on the calibration curves, and receiver operating characteristic (ROC) curves. The ability of nomogram to stratify patient risk was validated by Kaplan-Meier survival analysis.
Results
Age, gender, tumor location, T-stage and treatment modality (chemotherapy, radiation therapy, surgery and scope of lymph node dissection) of stage III-N2 LUAD patients were significantly associated with prognosis. The area under the curve (AUC) values of OS predicted by the nomogram constructed with these factors at 12-, 36- and 60-month were 0.784, 0.762 and 0.763 in the training cohort, whereas 0.707, 0.685 and 0.705 in the validation cohort, respectively. Additionally, calibration curves demonstrated concordance between predicted and observed outcomes. Nomogram risk stratification provides a meaningful distinction between patients with various survival risks.
Conclusions
A survival prediction model that may be useful for risk stratification and decision-making is developed and validated for stage III-N2 LUAD patients. A high-risk patient predicted by the prediction model may require more frequent surveillance or intensive care.