A Model To Identify Individuals at High Risk for Esophageal Squamous Cell Carcinoma and Precancerous Lesions in Regions of High Prevalence in China

BACKGROUND & AIMS: We aimed to develop a population-based model to identify individuals at high risk for esophageal squamous cell carcinoma (ESCC) in regions of China with a high prevalence of this cancer. METHODS: We collected findings from 15,073 permanent residents (45-69 years old) of 334 randomly selected villages in Hua County, Henan Province, China who underwent endoscopic screening (with iodine staining) for ESCC from January 2012 through September 2015. The entire esophagus and stomach were examined; biopsies were collected from all focal lesions (or from standard sites in the esophagus if no abnormalities were found) and analyzed histologically. Squamous dysplasia, carcinoma in situ, and ESCC were independently confirmed by 2 pathologists. Before endoscopy, subjects completed a questionnaire on ESCC risk factors. Variables were evaluated with unconditional univariate logistic regression analysis; variables found to be significantly associated with ESCC were then analyzed by multivariate logistic regression modeling. We used the Akaike information criterion to develop our final model structure and the coding form of variables with multiple measures. We developed 2 groups of models, separately defining severe dysplasia and above (SDA) (lesions including severe dysplasia and higher-grade lesions) and moderate dysplasia and above (lesions including moderate dysplasia and higher-grade lesions) as outcome events. Age-stratified and whole-age models were developed; their discriminative ability in the full multivariate model and the simple age model was compared. We performed area under the receiver operating characteristic curve (AUC) and the DeLong test to evaluate model performance. RESULTS: Our age-stratified prediction models identified individuals 60 years of age or younger with SDA with an AUC value of 0.795 (95% confidence interval, 0.736-0.854) and individuals older than 60 years with SDA with an AUC value of 0.681 (95% confidence interval, 0.618-0.743). Factors associated with SDA in individuals 60 years or younger included age closer to 60 years, use of coal or wood as a main source of cooking fuel, body mass index of 22 kg/m(2) or less, unexplained epigastric pain, and rapid ingestion of meals. In subjects older than 60 years, SDA associated with age, family history of ESCC, cigarette smoking, body mass index of 22 kg/m(2) or less, pesticide exposure, irregular eating habits, intake of high temperature foods, rapid ingestion of meals, and ingestion of leftover food in summer months. Use of our model in screening could have allowed 27% of subjects 60 years or younger and 9% of subjects older than 60 years to avoid endoscopy without missing SDAs. This means that approximately 2500 of endoscopies in total (16.6%) could have been avoided. CONCLUSIONS: We developed a low-cost, easy-to-use model to identify individuals at risk for severe dysplasia or cancer of the esophagus living in a region of China with a high risk of ESCC. This model might be used to select individuals and groups of persons who should undergo endoscopy analysis for esophageal cancer.