P0464Altered neural phenotypes in Crohn’s disease patients with varying degrees of bowel fibrosis

Background The brain-gut axis has recently been acknowledged as crucial factor in the pathogenesis of Crohn’s disease (CD). Bowel fibrosis, a frequently encountered but complex complication of CD, necessitates further investigation. This study aimed to explore the correlation between the severity of bowel fibrosis and alterations in neural phenotype, utilizing Magnetic Resonance Enterography (MRE) and multi-sequence brain MRI. Methods A prospective recruitment was conducted, enrolling a total of 109 CD patients who underwent MRE and brain MRI examinations within one week. The normalized Magnetization Transfer Ratio (MTR) from Magnetization Transfer Imaging (MTI) in MRE accurately classified CD patients into two groups: a group with non-mild bowel fibrosis (BF1) and another with moderate-severe bowel fibrosis (BF2). 12 MRE features were analyzed. The Mann-Whitney U test identified MRE features related to the BF2 group. The brain MRI data of CD patients were processed to extract multiparameter-multibrain-region radiomics features, followed by the construction of a logistic regression (LR) model (Fig.A). The training and test cohort were divided in a ratio of 7:3. The potential correlation between brain and intestinal fibrosis was investigated by utilizing Spearman correlation analysis to probe the relationship between the radiomics score and the MRE features associated with intestinal fibrosis, along with the normalized MTR. Results The study comprised 46 patients in the BF1 group and 63 in the BF2 group. Utilizing the LR model, 13 radiomics features were identified that effectively distinguished the BF2 from the BF1 group, achieving an impressive AUC of 0.846 in the test cohort (Fig.B). Amongst these brain features, five were deemed characteristic of the BF2 group. Their significance in the model was visualized in the SHAP plot, highlighting differences in brain microstructures among CD patients with varying degrees of bowel fibrosis (Fig.C). Four MRE features were associated with the BF2 group: stricture, penetration, effusion, and comb sign. Correlation analysis demonstrated a significant association (P<0.01) between brain rad-scores and both normalized MTR and stricture, suggesting the brain LR model can establish a correlation with bowel fibrosis and MRE morphological changes related to bowel fibrosis (Fig.D). Conclusion Distinct neural phenotypes were observed among CD patients with varying degrees of bowel fibrosis. Furthermore, the brain rad-score can establish a significant correlation with bowel fibrosis, highlighting the brain-gut interaction. Emphasizing both the neurological and gastrointestinal manifestations should be considered simultaneously in the management of CD patients. Conflict of interest: Ms. Shen, Xiaodi: No conflict of interest Zhang, Ruonan: No conflict of interest Wang, Yangdi: No conflict of interest Zheng, Qingzhu: No conflict of interest Wu, Luyao: No conflict of interest Ke, Yaoqi: No conflict of interest Zheng, Weikai: No conflict of interest Li, Xuehua: No conflict of interest