Construction a new nomogram prognostic model for predicting overall survival after radical resection of esophageal squamous cancer.

Abstract

Esophageal cancer is one of the deadliest malignancies in the world, and 5-year overall survival (OS) of esophageal cancer ranges from 12% to 20%. Surgical resection remains the principal treatment. The American Joint Commission on Cancer (AJCC) TNM (tumor, node, and metastasis) staging system is a key guideline for prognosis and treatment decisions, but it cannot fully predict outcomes. Therefore, targeting the molecular and biological features of each patient's tumor, and identifying key prognostic biomarkers as effective survival predictors and therapeutic targets are highly important to clinicians and patients. In this study, three different methods, including Univariate Cox regression, Lasso regression, and Randomforest regression were used to screen the independent factors affecting the prognosis of esophageal squamous cell carcinoma and construct a nomogram prognostic model. The accuracy of the model was verified by comparing with TNM staging system and the reliability of the model was verified by internal cross validation. Preoperative neutrophil lymphocyte ratio(preNLR), N-stage, p53 level and tumor diameter were selected to construct the new prognostic model. Patients with higher preNLR level, higher N-stage, lower p53 level and larger tumor diameter had worse OS. The results of C-index, Decision Curve Analysis (DCA), and integrated discrimination improvement (IDI) showed that the new prognostic model has a better prediction than the TNM staging system. The accuracy and reliability of the nomogram prognostic model were higher than that of TNM staging system. It can effectively predict individual OS and provide theoretical basis for clinical decision making.