Abstract
The C-reactive protein-albumin-lymphocyte (CALLY) index has been identified as a useful and sensitive predictive tool for stratification in cancers. This investigation aimed to validate the prognostic ability of CALLY in esophageal squamous cell carcinoma (ESCC). Clinical characteristics of 318 patients with ESCC who underwent radical excision were gathered and analyzed retrospectively. A restricted cubic spline (RCS) model was used to determine an ideal threshold of CALLY due to the non-linear relation. To investigate the predictors, Cox hazard regression analysis was used. The recursive partitioning analysis (RPA), a method of risk categorization, was also developed for prognostic prediction. The receiver operating characteristic (ROC) curves and decision curve analysis (DCA) curves were used to distinguish from the traditional TNM stage. Patients were compared by groups according to the optimal threshold of CALLY index, which was depicted by the non-linear relation between the cancer-specific survival (CSS) and CALLY index (P < 0.0001). Compared to those with high CALLY index, patients with low CALLY index experienced significantly worse 5-year CSS (21.8% vs. 62.6%, P < 0.001). At different TNM stages, patients with high CALLY index also had better 5-year CSS (I: P = 0.029; II: P < 0.001; III: P < 0.001) in subgroup analyses. The hazard ratio for CSS was 0.368 and CALLY index was an independent predictive factor (P < 0.001). Using TNM stage and CALLY-based RPA algorithms, a new staging was created. The RPA model considerably outperformed the TNM classification for prognostication using ROC (P < 0.001). The DCA also demonstrated that the new model outperformed the TNM stage with significantly improved accuracy for CSS. The prognostic value of CALLY in ESCC undergoing radical resection was initially determined in this study. CALLY was substantially related to prognosis and might be utilized in conjunction with TNM to evaluate ESCC prior to surgery.