Original Article
Nomogram predicting cancer-specific mortality in patients with esophageal adenocarcinoma: a competing risk analysis
Abstract
Background: Many factors are reported to be related to the prognosis of patients with esophageal adenocarcinoma (EAC), but few reliable and straightforward tools for clinicians to estimate individual mortalities have been developed. This study aimed to evaluate the probability of cancer-specific death for patients with EAC and to build nomograms for predicting long-term cancer-specific mortality and overall mortality for EAC patients.
Methods: Between 2004 and 2013, a total of 20,623 patients were identified from the surveillance, epidemiology, and end results (SEER) database and randomly divided into training (N=14,436) and validation (N=6,187) cohorts. The cumulative incidence functions (CIFs) of EAC-specific death and other causes were evaluated at the 1st, 3rd, and 5th year after diagnosis. We integrated the significant prognostic factors to construct nomograms and subjected them to internal and external validation.
Results: The CIFs of EAC-specific survival at 1, 3, and 5 years after diagnosis were 60.9%, 37.1%, and 31.3%, respectively. Predictors for cancer-specific mortality for EAC comprised tumor grade, tumor extension, the involvement of lymph nodes, distant metastasis, surgery of primary site, insurance recode, and marital status. For overall mortality, it also included the predictor of age at diagnosis. The nomograms were well-calibrated and had good discriminative ability with concordance indexes (c-indexes) of 0.733, 0.728, and 0.728 for 1-, 3- and 5-year prognosis prediction of EAC-specific mortality respectively, and 0.726, 0.720, 0.719 for 1-, 3-, and 5-year prognosis prediction of overall mortality respectively.
Conclusions: We proposed and validated the effective and convenient nomograms to predict cancer-specific mortality and the overall mortality for patients with EAC, which only require the basic information available in clinical practice.
Methods: Between 2004 and 2013, a total of 20,623 patients were identified from the surveillance, epidemiology, and end results (SEER) database and randomly divided into training (N=14,436) and validation (N=6,187) cohorts. The cumulative incidence functions (CIFs) of EAC-specific death and other causes were evaluated at the 1st, 3rd, and 5th year after diagnosis. We integrated the significant prognostic factors to construct nomograms and subjected them to internal and external validation.
Results: The CIFs of EAC-specific survival at 1, 3, and 5 years after diagnosis were 60.9%, 37.1%, and 31.3%, respectively. Predictors for cancer-specific mortality for EAC comprised tumor grade, tumor extension, the involvement of lymph nodes, distant metastasis, surgery of primary site, insurance recode, and marital status. For overall mortality, it also included the predictor of age at diagnosis. The nomograms were well-calibrated and had good discriminative ability with concordance indexes (c-indexes) of 0.733, 0.728, and 0.728 for 1-, 3- and 5-year prognosis prediction of EAC-specific mortality respectively, and 0.726, 0.720, 0.719 for 1-, 3-, and 5-year prognosis prediction of overall mortality respectively.
Conclusions: We proposed and validated the effective and convenient nomograms to predict cancer-specific mortality and the overall mortality for patients with EAC, which only require the basic information available in clinical practice.
