A Long-term Survival Risk Prediction Model for Patients with Superficial Esophageal Squamous Cell Carcinoma.

Abstract

Objectives: Given the data regarding the long-term prognosis of superficial esophageal squamous cell carcinoma (SESCC) is still lacking, we aimed to identify reliable prognostic factors and establish a high-precision prognosis model for patients with SESCC. Methods: A retrospective cohort study was conducted including patients with SESCC at a high-volume tertiary medical center. The primary outcome was disease-specific survival (DSS) at the end of follow-up (minimum of 29 months). Independent prognostic factors including innovative hematological and clinicopathological parameters were identified using comprehensive and novel statistical methods including best subset regression (BSR), the univariate and multivariate Cox analysis, lasso regression, and a dynamic nomogram model was established. Results: A total of 1,171 patients were finally enrolled. The median follow-up time is 83 months (range 29-149 months). Ten independent prognostic risk factors for a poor DSS were identified as follows: male (P=0.127), higher Charlson Comorbidity Index (CCI) (P=0.006), poorly differentiated tumor (P<0.001), lymphovascular invasion (LVI) (P<0.001), lymph node metastasis (LNM) (P<0.001), additional treatment (P=0.007), neutrophils over 32.2x10⁹/L (P=0.003), red blood cell (RBC) lower than 4.45x10¹²/L (P<0.001), hemoglobin (Hb) lower than or equal to 98 g/L (P=0.023), alpha-fetoprotein (AFP) higher than 3.24 ng/ml (P=0.034). Subsequently, an online dynamic nomogram was established (https://yryouzu-tools.shinyapps.io/DynNomapp/). This prediction model showed favourable discrimination ability (area under the curve (AUC) was 0.913 (95% CI: 88.0 - 94.6) and a well-fitted calibration curve. Conclusions: We successfully established a long-term prognosis model for SESCC, which can be applied to effectively predict survival risks for patients, thus strengthening follow-up strategies.