Establishment and evaluation of prediction model for second primary malignancy of esophageal adenocarcinoma and esophageal squamous cell carcinoma based on competitive risk
Objective To evaluate the predictive effectiveness and clinical utility of establishment of prediction model of second primary malignancy (SPM) for esophageal adenocarcinoma (AC) and esophageal squamous cell carcinoma (SCC). Method The data of the patients with esophageal AC and esophageal SCC diagnosed by pathology (1998-2014) were collected from Surveillance, Epidemiology and End Results(SEER) database. The independent risk factors for SPM in such patients were assessed by competitive risk model. Based on obtained independent risk factors after using the competitive risk model, the SPM models for AC and SCC were respectively established. The judgment capability, predictive effectiveness and clinical : practicability were analyzed and judged by C-index, calibration cure and decision curve. Results A total of 13 526 cases of esophageal cancers including 8 700 cases of AC (480 cases of SPM) and 4 826 cases of SCC (337 cases of SPM) were included in this study. Multivariate analysis showed that age, SEER staging, number of positive lymph nodes and distant metastasis were the independent risk factors influencing the occurrence of the SPM of esophageal AC, and age, SEER staging, marital status, tumor location and chemotherapy situation were the independent risk factors influencing the occurrence of the SPM of esophageal SCC. The C-index of SPM predictive models for esophageal AC and SCC were 0. 691 and 0. 662, respectively, so it showed an ideal model discrimination ability. Correction curve indicated that SPM incidence was in good agreement with the actual incidence for the prediction model. The probability thresholds of decision curve for esophageal esophageal AC and SCC were 0. 020 ― 0. 177 and 0. 021 ― 0. 133, respectively, and this suggested that the model had an ideal clinical benefit in the above range. Conclusion In this study, the prediction models of SPM onset risks for esophageal AC and SCC are established and evaluated by means of competitive risk. The models showed good prediction ability and clinical : practicability, and this will contribute to the screening and the clinical intervention for high risk population of SPM.
