Epidemiology of Risk Stratification, Machine Learning Early Prediction Model, and Tumor Suppressive Mechanism of...

Esophageal cancer imposes a considerable health burden in the high-risk areas of Northwest China, necessitating the development of effective biomarkers for its early detection to address this public health challenge.

We integrated multi-omics analysis, machine learning algorithms, and population epidemiology investigations in Gansu Province to screen and develop biomarkers for the early detection of esophageal cancer.

Epidemiological findings revealed distinct geographical variations in esophageal cancer incidence, with Yugu County being the sole high-risk area (incidence rate: 54.2/100,000). Cases were predominantly in individuals aged > 40 years; males were the main affected population except in Yugu County. Protective factors for the disease included a healthy diet, regular exercise, and positive emotions, while smoking, alcohol consumption, and high-salt intake were identified as risk factors. A random forest machine learning model exhibited excellent predictive performance (AUC = 0.995) and identified key predictive factors for esophageal cancer. Proteomic analysis further revealed that RHBDF2 was downregulated and could serve as a potential biomarker for the disease.

This study provides robust epidemiological and molecular evidence for esophageal cancer prevention and early intervention strategies, and the identified potential biomarker RHBDF2 and high-performance predictive model offer valuable tools for the early detection of esophageal cancer in high-risk regions of Northwest China.