基于筛查队列和高维代谢组数据的食管癌前病变动态进展的风险预测研究

Esophageal precancerous lesions are reversible, predictable, and have a good prognosis. They are the focus of esophageal cancer screening and early diagnosis and treatment. The current endoscopic screening program is limited by low detection rate and poor compliance, while the risk prediction model and metabolomics biomarkers provide an effective way to expand the beneficiary population and improve screening efficiency. This study is based on the established screening cohort for esophageal cancer in high-risk areas. It focuses on precancerous lesions and include the following research: 1) Construct a risk prediction model for esophageal precancerous lesions based on the first screening cohort, to identify high risk individuals and concentrate screening subjects; 2) Based on the longitudinal cohort of multiple screening, the dynamic Bayesian network model is used to construct a risk prediction model for the dynamic progression of esophageal precancerous lesions, providing guidance for follow-up, diversion and treatment of patients diagnosed with precancerous lesions; 3) Based on classic case-control study and nested case-control study, LC-MS platform is used to detect metabolites, and metabolomics biomarkers are screened and validated for the diagnosis and progression of precancerous lesions to explore the underlying mechanisms of precancerous lesions; 4) Based on high-dimensional metabolomics data, the metabolic risk scores and related risk prediction models are constructed to evaluate of the role of metabolomics biomarkers in predicting precancerous lesions and their progression. The results of this study will provide a scientific basis for early diagnosis and early warning of precancerous lesions, and provide a reference for exploring individualized screening and precision prevention of esophageal cancer.

食管癌前病变可逆、可预测、预后良好，是食管癌筛查与早诊早治的重点。现行内镜筛查方案检出率低、依从性差，风险预测模型和代谢组标志物是扩大受益人群、提高筛查效果的有效途径。本项目计划以课题组在食管癌高发区建立的人群筛查队列为基础，针对癌前病变开展以下研究：1）基于首次筛查队列，构建癌前病变风险评估模型，识别高危人群，浓缩筛查对象；2）基于多次筛查的纵向队列，采用动态贝叶斯网络模型探索癌前病变动态进展的风险预测，指导患者的随访、分流与治疗；3）基于经典病例-对照研究和巢式病例-对照研究，采用LC-MS平台检测代谢产物，筛选并验证癌前病变诊断与进展的代谢组标志物，探讨癌前病变发生发展的潜在机制；4）基于高维代谢组数据构建代谢风险评分和预测模型，评价代谢组标志物在预测癌前病变及其进展中的作用。本研究的结果将为癌前病变的早期诊断和预警预测提供科学依据，为探索食管癌的个体化筛查和精准预防提供参考。

筛查和早诊早治是降低食管癌发病率和死亡率的重要措施。现行筛查方案存在检出率低、依从性差、难以大规模推广等局限性。风险预测模型和代谢组分子标志物为提高筛查效率、优化筛查方案提供了有效途径。本项目以课题组前期建立的食管癌高发区人群筛查队列为基础，基于筛查队列构建食管癌前病变风险评估模型，低级别上皮内瘤变模型AUC为0.706，高级别上皮内瘤变模型AUC为0.778。基于多次内镜筛查，整合饮食和生活习惯因素构建危险评分，结合性别、年龄、治疗和预后情况，以队列数据作为参数，构建食管癌前病变进展的动态贝叶斯网络模型，实现筛查患者的个体化预测。基于液相色谱-质谱联用的代谢组学检测技术，对癌前病变的代谢组标志物进行检测、筛选和验证。本研究针对高级别上皮内瘤变和低级别上皮内瘤变分别筛选了7种和9种代谢标志物，主要为脂质和脂质样分子、有机氧化合物、有机酸及其衍生物等，富集在氨基酸的生物合成、蛋白质消化和吸收、癌症的胆碱代谢、癌症的中心碳代谢等通路。结合传统危险因素和代谢组标志物构建癌前病变的预测模型，高级别癌前病变和低级别癌前病变的AUC分别为0.923和0.875，提示代谢组标志物对食管癌的早期预防具有重要研究价值。本研究成果将为食管癌前病变诊断和进展的分子标志物研究提供参考，为食管癌的初筛和随访提供科学依据。