自身免疫性胃炎内镜诊断模型的建立与验证

Diagnosis model for autoimmune gastritis based on endoscopic featuresLuo Hao1, Xu Xue1,Li Wenbin1, Xi Wu1, Rui Gang1, Du Jun2, Qiang Wang3, Xi Wu3, Luo Qingfeng1, Shi Jihua1

摘要:

【摘要】目的 总结自身免疫性胃炎（autoimmune gastritis，AIG）的内镜特征，建立一个基于内镜的AIG诊断模型（endoscopy-based model for AIG，EBM-AIG）并评估该模型的诊断效力。方法 回顾性纳入2022年1月至2023年11月于北京医院及北京协和医院诊断为慢性胃炎的患者共448例。按2:1的比例随机分为建模队列（302例）与验证队列（146例）。在建模队列中，通过单因素及多因素logistic回归分析筛选与AIG独立相关的内镜特征，并依据回归系数为其赋分，构建EBM-AIG评分模型。采用受试者工作特征（receiver operating characteristic, ROC）曲线分析评估模型的诊断效能，并在验证队列中进行验证。结果 多因素分析显示，逆萎缩（P＜0.001，OR=42.59，95%CI：41.66~43.52）、黏稠黏液（P＜0.001，OR=14.25，95%CI：13.47~15.03）、残存泌酸腺黏膜（P=0.031，OR=4.63，95%CI：3.57~5.69）以及胃体黏膜肿胀发红（P＜0.001，OR=21.52，95%CI：20.72~22.32）是AIG的独立预测因素。基于此四项因素构建EBM-AIG评分模型，总分0~10分。在建模队列中，该模型的ROC曲线下面积为0.985，以评分≥5分为最佳截断值，诊断准确率、敏感度、特异度分别为95.4%（288/302）、92.6%（112/121）、97.2%（176/181）。在验证队列中，EMB-AIG评分显示了良好的AIG诊断能力，ROC曲线下面积为0.994（95%CI 0.983~1.000）。结论 基于逆萎缩、黏稠黏液、残存泌酸腺黏膜及胃体黏膜肿胀发红这四项内镜特征构建的EBM-AIG评分模型，在内部验证中显示出优异的AIG诊断效能。评分≥5分对AIG具有很高的诊断价值，可作为内镜下提示AIG的实用工具。

Abstract:

【Abstract】Objectives Autoimmune gastritis (AIG) is a chronic and progressive damage to gastric corpus and fundus mucosa caused by autoimmune mediated inflammation. Endoscopy is an important opportunity to discover and diagnose AIG. However, there is often a substantial diagnostic delay in clinical practice. The aim of this study was to characterize the endoscopic appearances of AIG, and develop an endoscopy-based model for AIG (EBM-AIG) diagnosis. Methods A total of 448 patients diagnosed with chronic gastritis were retrospectively enrolled from two tertiary hospitals in China between January 2022 and November 2023. They were randomly divided into a derivation cohort (n=302) and a validation cohort (n=146) at a ratio of 2:1. In the derivation cohort, endoscopic features were evaluated and an EBM-AIG scoring model was developed using multivariate logistic regression analysis, with points assigned based on the regression coefficients. The diagnostic performance of the model was assessed using receiver operating characteristic (ROC) curve analysis and validated in the validation cohort. Results Reverse atrophy (OR 42.59; 95% CI 41.66-43.52), sticky adherent dense mucus (OR14.25; 95% CI 13.47-15.03), remnant oxyntic mucosa (OR 4.63; 95% CI 3.57-5.69) and diffuse reddened and edematous gastric fundus gland mucosa (OR 21.52; 95% CI 20.72-22.32) were significantly associated with AIG. An EBM-AIG scoring model (total score 0-10) was constructed based on these four features. In the derivation cohort, the model showed excellent discrimination with an area under the ROC curve (AUC) of 0.985. Using a cut-off score of ≥5, the diagnostic accuracy, sensitivity, and specificity were 95.4%, 92.6%, and 97.2%, respectively. In the validation cohort, the EBM-AIG score showed good diagnostic ability of AIG with the AUROC 0.994 (95%CI 0.983-1.000). Conclusions The EBM-AIG scoring model, based on the four endoscopic features of reverse atrophy, sticky mucus, remnant oxyntic mucosa, and diffuse reddened and edematous gastric fundus gland mucosa, demonstrated excellent diagnostic performance for AIG in internal validation. A score of ≥5 holds high diagnostic value and can serve as a practical tool for endoscopically suggesting AIG, guiding targeted biopsies for pathological confirmation.