Dynamic assessment of long-term survival in survivors with stage III non-small cell lung cancer: a novel conditional survival model with a web-based calculator

Background: Conditional survival (CS) analysis can estimate further survival probabilities based on the time already survived, providing dynamic updates for prognostic information. This study aimed to develop a CS-nomogram to promote individualized disease management for stage III non-small cell lung cancer (NSCLC).

Methods: This study included patients diagnosed with stage III NSCLC in the Surveillance, Epidemiology, and End Results database from 2010 to 2017 (N = 3,512). The CS was calculated as CS(y|x) = OS(y + x)/OS(x), where OS(y + x) and OS(x) were the overall survival (OS) in the year (y + x) and year x, respectively, calculated by the Kaplan-Meier method. We used the least absolute shrinkage and selection operator (LASSO) regression to identify predictors and developed the CS-nomogram based on these predictors and the CS formula.

Results: The CS analysis provided real-time updates on survival, with 5-year OS improving dynamically from 14.4 to 29.9%, 47.9, 66.0, and 80.8% (after 1-4 years of survival). Six independent predictors (age, tumor size, N status, surgery, radiotherapy and chemotherapy) were identified for the development of the CS-nomogram and its web version (https://dynapp.shinyapps.io/NSCLC/). The model performed with an excellent concordance index (C-index) of 0.71 (95% CI: 0.70-0.72), and a median time-dependent AUC of 0.71-0.73 from 200 iterations 5-fold cross-validation.

Conclusion: The study demonstrated the improvement in real-time OS over time in stage III NSCLC survivors and developed the novel CS-nomogram to provide patients with updated survival data. It provided novel insights into clinical decisions in follow-up and treatment for survivors, offering a convenient tool for optimize resource allocation.