Association between body fat composition and disease duration, clinical activity, and intravenous corticosteroid-induced response in inflammatory bowel disease

Abstract

Background Body fat composition is believed to be associated with the progression, medical response, and prognosis of inflammatory bowel disease (IBD). Hence, we conducted this study to explore if fat metrics were associated with the disease activity of severe IBD and the response to intravenous corticosteroids (IVCS). Methods We included 69 patients with ulcerative colitis (UC) and 72 patients with Crohn's disease (CD) who had previously received IVCS during hospitalization. We quantified individual fat distribution using abdominal computed tomography slices. The correlations between fat parameters and disease activity were available with Spearman correlation analysis. The prediction model was developed using independent risk factors derived from multivariable logistic regression analysis. Model discrimination was evaluated leveraging the receiver operating characteristic curve. 1000 bootstrap resamples internally validated the model's prediction performance. Results Notable differences in age, nutritional status, serum cytomegalovirus replication, stool condition, and extraintestinal involvement between UC and CD patients were observed. UC subjects who responded to IVCS had higher subcutaneous adipose tissue index (SATI), visceral adipose tissue index (VATI), and mesorectal adipose tissue index (MATI) than non-responders. IVCS-responding CD individuals had lower VATI and mesenteric fat index (MFI) than non-responders. CD patients with a prolonged disease duration had a decreased SATI and an elevated MFI. VATI and MATI were reduced as UC clinically progressed, while more prominent clinical activity in CD correlated with increased VATI, MATI, and MFI. A high SATI indicated that patients with UC were more prone to be IVCS responders. For patients with CD, levels of VATI and MFI were negatively associated with effective IVCS treatment. The established models showed a discriminative accuracy of 0.834 [95% confidence interval (CI) 0.740–0.928] in the UC cohort and 0.871 (95% CI 0.793–0.949) in the CD cohort. Repeated samples supported the reliability of the developed models (AUCUC = 0.836, 95% CI 0.735–0.919; AUCCD = 0.876, 95% CI 0.785–0.946). Conclusion Human fat indexes represent novel imaging biomarkers for identifying IBD patients who respond to IVCS, thus building accelerated therapy regimens and avoiding the adverse effects of ineffective IVCS.