Risk Prediction Models for Recurrence After Curative Treatment of Early‐Stage or Locally Advanced Lung Cancer: A Systematic Review

Introduction Lung cancer remains the leading cause of cancer‐related mortality worldwide. Despite advances in treatment, postoperative recurrence rates in early‐stage and locally advanced non‐small cell lung cancer (NSCLC) remain substantial. Prognostic models may support risk stratification and guide posttreatment management. Objective To systematically identify and compare prognostic models predicting survival and recurrence outcomes in patients with NSCLC treated with curative intent. Methods This systematic review was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines. MEDLINE, EMBASE, and the Cochrane Library were searched from inception to November 29, 2022, with an updated search to capture newly published studies. Eligible studies reported quantitative performance measures of prognostic models predicting recurrence‐free survival (RFS), cancer‐specific survival (CSS), or overall survival (OS) in patients with curatively treated NSCLC. Data extraction followed the CHARMS checklist, and risk of bias was assessed using the PROBAST tool. The review protocol was registered with PROSPERO (CRD42021226091). Results Of 6369 screened records, 40 studies met inclusion criteria. Most models employed Cox proportional hazards regression and varied widely in predictor selection. Common predictors included age ( n = 40), sex ( n = 40), and tumor characteristics ( n = 36). Radiomic features were incorporated in 11 models and were associated with improved discrimination. Reported model performance ranged from poor to excellent (AUROC 0.5–0.9), with most models demonstrating acceptable to good discrimination (AUROC > 0.7). The highest reported 3‐year OS AUROC was 0.90 (95% CI: 0.85–0.95). However, only four models underwent external validation, and reporting of calibration and other performance metrics was inconsistent. Conclusion Prognostic models for survival and recurrence in resected NSCLC frequently demonstrate good discrimination but are limited by inadequate external validation and variable reporting. Future research should prioritize transparent model development, robust validation, and clinical usability to improve posttreatment decision‐making.