Prognostic And Predictive Significance Of Pathological Tumor Length In Esophageal Squamous Cell Carcinoma

Purpose:There has been a lot of controversy over the prognostic impact of tumor length in esophageal cancer. The aim of the present study was to evaluate the prognostic and predictive significance of pathological tumor length in esophageal squamous cell carcinoma and to determine the optimal cutoff values of pathological tumor length for accurate risk stratification.Methods:From2003to2010,1613patients with esophageal cancer were surgically treated at our department. Clinicopathological data were collected from electronic medical recording system. Martingale residuals from a Cox Proportional hazards regression model were used to detect an optimal cutoff value for pathological tumor length. The prognostic value of pathological tumor length and the correlation between pathological tumor length and other clinicopathological factors were investigated. A Cox regression model based on Akaike Information Criterion was used to identify independent clincopathological variables. Predictive accuracy of each clinicopathological variable was assessed by Harrell Concordance Index (C-index). Cox proportional hazards regression models were fitted to model the relationship between clinicopathological variables and prognosis. Each model was internally validated for discrimination and calibration by using bootstrap resampling. Model-based nomograms were constructed to predict prognosis. Time-dependent receiver operating characteristic (ROC) was also used to compare predictive accuracy among models. Enhanced predictive accuracy of TNM staging system model with and without pathological tumor length was evaluated by net reclassification index (NRI) and integrated discrimination improvement (IDI) for censored survival data.Results:According to inclusion criteria,1435patients with esophageal squamous cell carcinoma who underwent radical esophagectomy were enrolled in this study. Martingale residuals analysis revealed a cutoff value of4cm for pathological tumor length was appropriate in these patients. Pathological tumor length was correlated with age, sex, tumor location, T stage, N stage and the number of resected lymph nodes. The overall survival of patients with pathological tumor length≤4cm was longer than that of patients with pathological tumor length>4cm (median survival time,48months vs27months, P<0.001). Tumor grade, T Stage, N stage, age, number of resected lymph nodes and pathological tumor length were identified as independent prognostic factors. The predictive accuracy of pathological tumor length (C-index=58.1%) was inferior to that of N stage (C-index=67.1%) and T stage (C-index=60.5%). The bootstrap-corrected C-index of TNM staging base model is higher than that of the model combining TNM staging and pathological tumor length in predicting5-year survival (69.4%vs69.8%,P <0.001). Calibration of plots of each model showed that the predicted survival reasonably approximated observed outcomes. Similar results of predictive accuracy of models were shown in time-dependent ROC analysis. Compared with TNM staging base model, the model combining TNM staging and pathological tumor length did not show the enhanced predictive accuracy with regard to NRI (0.001, P=0.637) and IDI (0.046, P=0.498).Conclusion:Pathological tumor length is an independent prognostic factor in patients with esophageal squamous cell carcinoma. The optimal cutoff value of4cm of pathological tumor length is appropriate for risk stratification. But pathological tumor length does not increase predictive accuracy of the current TNM staging system for esophageal squamous cell carcinoma.