Prognostic Value of Tumor Length for Cause-Specific Death in Resectable Esophageal Cancer

Background. The current esophageal cancer American Joint Committee on Cancer TNM staging system may not capture the full prognostic implications of the primary tumor. A study is needed to explore the prognostic value of tumor size on esophageal cancer-specific death. Methods. Patients who underwent surgical resection for non-metastatic esophageal cancer were selected from the Surveillance, Epidemiology and End Results Program database (United States, 1988 to 2014). With the use of statistics methods, maximally selected rank and two hazard models (Cox model and Fine-Gray model), the optimum cutoff point for tumor length in each T classification was estimated and the prognostic value of tumor size on esophageal cancer-specific death was analyzed. Results. A total of 4,447 patients were identified. The median tumor size was significantly correlated with T classification, with the correlation coefficient of 0.43 (p < 0.001). Patients in the T1 to T3 classifications who had larger tumor size showed a larger probability of cancer-specific death. The multivariate Cox model showed that tumor size was significantly associated with an increase in cancer-specific death in T1 (2.15, 95% confidence interval [CI]: 1.72 to 2.69) and T2 (1.31, 95% CI: 1.06 to 1.62) but was marginally significantly associated in T3 (1.12, 95% CI: 1.00 to 1.27) and insignificantly associated in T4 classification (p > 0.1). Similar results were found by using the multivariate Fine-Gray model. Conclusions. We have found that combining T classification with tumor size can increase the precision in identifying the high-risk groups in T1 to T2 classifications. On the basis of esophageal cancer-specific death our study explores the prognostic cutoff point of tumor size by T classification. (C) 2018 by The Society of Thoracic Surgeons