Related Experiment Video
Updated: Jul 12, 2026
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
A recurrence risk prediction model and recurrence patterns in pulmonary lymphoepithelial carcinoma based on clinical
Haiwen Chen1, Gengda Huang1, Hong He2
1Department of Respiratory and Critical Care Medicine, State Key Laboratory of Respiratory Disease, National Clinical Research Center for Respiratory Disease, National Center for Respiratory Medicine, Guangzhou Institute of Respiratory Health, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, China.
Background:
Pulmonary lymphoepithelial carcinoma (pLELC) is a rare Epstein-Barr virus (EBV)-associated non-small cell lung cancer (NSCLC). In patients with resected pLELC, postoperative recurrence is the major barrier to long-term survival. However, predictors of recurrence remain unclear, and reliable risk prediction models are lacking. This study aimed to develop a recurrence risk prediction model for patients with resected pLELC based on clinical and dynamic hematologic parameters and to characterize postoperative recurrence patterns.
Methods:
We retrospectively analyzed patients who underwent radical resection for pLELC. Recurrence-related variables were screened by univariable and stepwise multivariable Cox regression to construct a predictive model. The model's performance was assessed via receiver operating characteristic (ROC) analysis, the concordance index (C-index), and bootstrap internal validation, with a comparison against the tumor-node-metastasis (TNM) staging system. Recurrence sites and their prognostic implications were also evaluated.
Results:
Of 238 patients, 57 developed recurrences. Independent predictors of recurrence were node (N) stage, central tumor location, elevated preoperative D-dimer, postoperative platelet count increase, and elevated postoperative neutrophil-to-lymphocyte ratio (NLR). The model demonstrated strong discriminative performance in the training cohort, with 1-, 3-, and 5-year of area under the curves (AUCs) of 0.822, 0.792, and 0.744, respectively, outperforming the conventional TNM staging system. Internal validation yielded a C-index of 0.752. Among patients with recurrence, the median disease-free survival (DFS) was 21.8 months. Most recurrences were locoregional (76.0%), and those with distant metastasis experienced significantly poorer post-recurrence survival (PRS) (P=0.04).
Conclusions:
Our recurrence prediction model for resected pLELC, which integrates clinical and dynamic hematologic factors, outperformed TNM staging. It provides a basis for stratifying postoperative surveillance and guiding early intervention in high-risk patients. Additionally, patients with distant metastasis had worse PRS than those with locoregional relapse, indicating distinct prognostic implications based on recurrence pattern.