Personalized Prediction of prognosis in ebv-associated lymphoepithelioma-like carcinoma: insights from a Chinese cohort study

Lymphoepithelioma-like carcinoma (LELC) is a rare tumor. The therapeutic strategies and personalized prediction of prognosis for this tumor remain inadequately defined. Five hundred seventy-four patients from Sun Yat-sen University Cancer Center were recruited and split into a training cohort (402) and a validation cohort (172). Significant variables identified through Cox regression analyses were incorporated to develop nomograms. We assessed the predictive precision and discriminatory capacity of the nomograms by employing various metrics, including the time-dependent receiver operating characteristic (ROC) curve, concordance index (C-index), decision curve analysis (DCA) curve, calibration curve, and Kaplan-Meier curves. Multivariate Cox analysis identified that surgery, chemotherapy, radiotherapy, N stage, M stage, platelet-to-lymphocyte ratio (PLR), and lactate dehydrogenase-to-albumin ratio (LAR) were independently associated with overall survival (OS) in LELC patients. For progression-free survival (PFS), surgery, chemotherapy, T stage, N stage and M stage emerged as independent predictors. Nomograms were developed by integrating the factors to provide a comprehensive predictive tool. The C-indices of nomograms for forecasting OS and PFS were 0.786 (95%CI 0.734–0.838) and 0.693 (95%CI 0.651–0.735), correspondingly. ROC curves and calibration curves demonstrated consistency with actual observations. Moreover, DCA curves demonstrated the practicality of nomograms, emphasizing their exceptional capacity to distinguish high-risk individuals compared with the 8th AJCC staging system. The newly devised nomograms demonstrated considerable potential in predicting OS. Furthermore, this research is the first to suggest that markers related to inflammation and nutrition, such as PLR and LAR, could be significant predictors for forecasting the outcome of individuals with LELC.