Analysis of influencing factors and establishing predictive model of mucosal healing under endoscopy in Crohn′s disease
10.3760/cma.j.cn311367-20241122-00458
- VernacularTitle:克罗恩病内镜下黏膜愈合的影响因素分析与预测模型构建
- Author:
Tiange LI
1
;
Suqi ZENG
1
;
Junhai ZHEN
1
;
Weiguo DONG
1
Author Information
1. 武汉大学人民医院消化内科，武汉 430060
- Publication Type:Journal Article
- Keywords:
Crohn′s disease;
Mucosal healing;
Predictive modeling;
Nomogram
- From:
Chinese Journal of Digestion
2025;45(3):169-176
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To investigate the influencing factors of mucosal healing under endoscopy in patients with Crohn′s disease (CD) and to establish a predictive model.Methods:From January 1, 2023 to August 31, 2024, 124 patients with CD were hospitalized at the Department of Gastroenterology, Renmin Hospital of Wuhan University were retrospectively enrolled as the modeling group. And from January 1, 2021 to December 31, 2022, 88 patients with CD were hospitalized at the Department of Gastroenterology in the same hospital were collected as the validation group. The data including simple Crohn′s disease activity index (CDAI) scores, serological markers such as fibrinogen (FIB), and medication regimens (including ustekinumab) of the patients in the modeling group were collected. Multivariate logistic regression analysis was used to screen the independent predictors of mucosal healing in CD patients, and the nomogram predictive model was established. The receiver operating characteristic curve (ROC) was plotted to evaluate the predictive performance, and calibration curve was drawn for validation. Mann-Whitney U test and Chi-square test were used for statistical analysis. Results:According to the simple endoscopic score for CD and endoscopic findings, among the 124 patients in the modeling group, 92 cases were diagnosed as mucosal healing, while 32 cases did not. The simple CDAI and FIB of patients with mucosal healing were lower than those of patients without mucosal healing (2.00(2.00, 3.00) vs. 3.00(2.25, 4.00), 2.37(2.03, 2.88) g/L vs.2.92(2.40, 4.40) g/L); the proportion of patients who used ustekinumab in mucosal healing patients was higher than that of patients without mucosal healing (62.0%, 57/92 vs. 31.2%, 10/32), and the differences were statistically significant ( Z=-2.98 and -3.57, χ2=9.01; all P<0.01).The results of multivariate logistic regression analysis showed that low simple CDAI score ( OR=0.560, 95% confidence interval (95% CI): 0.343 to 0.913), low FIB ( OR=0.475, 95% CI: 0.302 to 0.747), and ustekinumab usage ( OR=4.218, 95% CI: 1.621 to 10.977) were independent predictive factors of mucosal healing under endoscopy in CD patients (all P<0.05). The regression equation was derived as ln( p/(1- p)) mucosal healing=4.215-0.580×simple CDAI score -0.745×FIB(g/L)+ 1.439×ustekinumab usage(1 for use, 0 for unused), and the nomogram model was established. The results of ROC demonstrated that the area under the curve of the nomogram model in the modeling and validation group were 0.791(95% CI: 0.700 to 0.883) and 0.781 (95% CI: 0.666 to 0.895), with the sensitivity of 0.859 and 0.868, and with the specificity of 0.688 and 0.650, respectively. The results of calibration curve analysis showed that the average absolute errors of the nomogram model in the internal and external validation were 0.032 and 0.039, respectively, indicating a good consistency between the predicted and actual probability. Conclusions:Low simple CDAI score, low FIB, and ustekinumab usage are the independent predictive factors of mucosal healing under endoscopy in CD patients. The predictive model has certain reference value for CD management.
