Differential prognostic significance of distant and locoregional recurrence on survival in surgically resected non-small cell lung cancer post-chemotherapy: Multicenter dynamic prediction with landmark model.

8062 Background: Recurrence remains a significant challenge in patients with surgically resected non-small cell lung cancer (NSCLC) following adjuvant chemotherapy. Recurrence status evolves over the course of follow-up, dynamically influencing survival outcomes. This study aimed to investigate the differential impact of distant metastasis (DM) and locoregional recurrence (LR) on survival, and to develop prognostic models to guide treatment strategies and individualize follow-up protocols. Methods: From four institutions, patients with pN2 NSCLC who underwent complete resection followed by four cycles of platinum-based doublet chemotherapy were included. A dynamic prediction landmark model was constructed to assess the impact of LR and DM on survival. The primary endpoint was overall survival (OS). Baseline factors included age, sex, smoking history, histology, tumor laterality, pT stage, and the number of positive lymph nodes, while DM and LR status were treated as time-dependent covariates. Results: A total of 2,120 patients were included in the study, with a median follow-up time of 55.80 months (IQR: 39.47–85.12). The landmark model identified older age, smoking history, advanced T stage, DM, and LR as significant factors associated with worse OS. DM had the most substantial impact on OS (odds ratio [OR], 3.85; 95% CI, 3.32–4.48; P < 0.01), while LR also significantly decreased OS (OR, 2.07; 95% CI, 1.73–2.47; P < 0.01). Multivariate Cox analysis revealed that the pT stage, number of positive lymph nodes, and histology were independently associated with DM. A nomogram was developed to predict the risk of DM for individual patients, categorizing them into three distinct risk subgroups. Postoperative radiotherapy did not significantly improve OS in the low- or high-risk subgroups but demonstrated a survival benefit in the medium-risk subgroup (hazard ratio [HR], 0.73; 95% CI, 0.63–0.87; P < 0.01). Intensified systemic therapy and closer monitoring would be required for patients in the high-risk subgroup. Conclusions: The dynamic prediction model estimated future survival probabilities based on individual recurrence status throughout follow-up in patients with surgically resected NSCLC post-chemotherapy. Both DM and LR significantly affected OS, with DM being more detrimental. The DM risk nomogram aids in assessing the benefits of additional treatments and guiding personalized follow-up strategies.