Competing Risk Nomogram to Predict Cancer-Specific Survival in Esophageal Cancer Patients During the Intensity-Modulated Radiation Therapy Era: A Single Institute Analysis

Abstract

Recently, competing risk method based on the subdistribution hazard model is recommended to accurately determine the effects of risk factors on specific events, because it takes into account the informative nature of censoring. The aim of present study was to investigate the probability of cancer-specific mortality of esophageal cancer (ESC) patients undergoing intensity-modulated radiation therapy (IMRT) and establish a competing risk nomogram to predict esophageal cancer-specific survival (ESC-SS) of these patients. A total of 213 ESC patients undergoing IMRT were identified to establish nomograms based on Fine and Gray’s competing risk analysis. Predictive accuracy and discriminative ability of the model were performed using concordance index(C-index),calibration curve and the area under receiver operating characteristic (ROC) curve. Decision tree analysis was performed for patients grouping. With a median 19 month (range:3-50) follow-up, the 2-year ESC-specific mortality (ESC-SM) and non-cancer specific mortality (NC-SM) of the cohorts were 35.4% and 3.51% respectively, and elevated 2-year ESC-SM were observed in patients with tumor length≥4.5cm (45.8% vs. 21.4%, p<0.001), non-squamous cell carcinoma(non-SCC, 49.9% vs. 33.7%, p=0.025), and N3 stage (43.2%, p=0.005). The 2-year NC-SM was observed in tumor length≥4.5cm cohorts (6% vs. 0%, p=0.016). All validated factors including tumor length, histology type and N stage, were integrated into the competing nomograms into the ES-SCC model [concordance (c)-index=0.72, 95%CI: 0.66-0.77]. In addition, the nomograms displayed better discrimination power than 7th edition Tumor-Node-Metastasis (TNM) stage systems for predicting ESC-SS (the AUC value 0.707 vs. 0.634).The classification tree analysis also showed that N stage is the initial node, tumor length is also determinants for ESC-SM of this patients population. NC-SM represents a significant competing event for esophageal cancer with tumor length ≥4.5cm. The competing risk nomograms could be considered as convenient individualized predictive tools for cancer specific survival in esophageal cancer undergoing IMRT treatment.Abstract 2449; Table 1VariableUnivariateMultivariateSHR95%CIpSHR95%CIpGenderMale1---Female0.680.34-1.370.29---Age<601-≥600.840.54-1.330.47---Histologic TypesSCC11Non-SCC2.031.06-3.910.0342.711.48-4.930.001GradeHigh11Moderate/poor1.441.01-2.080.0461.140.75-1.740.61SurgeryYes11No0.620.40-0.960.0331.520.85-2.720.15Tumor length≤4.511＞4.5cm2.431.48-3.99《0.0011.801.03-3.160.04MetastasisYes11No2.121.08-4.130.0281.410.66-3.000.37 Open table in a new tab