Abstract
BACKGROUND: Sarcopenia is prevalent in Crohn's disease (CD) patients, but the prognostic value of skeletal muscle-to-intermuscular adipose tissue (IMAT) balance in penetrating CD remains unexplored. OBJECTIVE: To determine whether skeletal muscle-IMAT imbalance predicts non-surgical treatment failure in CD. METHODS: This retrospective study included CD patients undergoing computed tomography enterography (CTE, 2013-2022), stratified by disease behavior (penetrating vs. non-penetrating). Automated CTE segmentation algorithm quantified skeletal muscle and IMAT volumes at baseline. Skeletal muscle ratio was calculated as skeletal muscle/(skeletal muscle + IMAT). Sarcopenia was defined by the skeletal muscle area at the third lumbar vertebra. Patients received ≥1 year of non-surgical therapy, with outcomes categorized as 'maintenance therapy' or 'treatment escalation'. Cox proportional hazards analysis identified predictors of escalation; mediation analysis evaluated inflammatory-nutritional pathways. RESULTS: Among 157 patients (penetrating: n = 42; non-penetrating: n = 115), treatment escalation rates were 64.3% (27/42) and 53.0% (61/115) respectively, without significant intergroup difference (p = 0.21). Skeletal muscle ratio predicted escalation in both cohorts (penetrating: AUC = 0.822 [0.673, 0.923], non-penetrating: AUC = 0.922 [0.857, 0.964]), outperforming conventional sarcopenia metrics (p = 0.002 and p < 0.001). Cox regression confirmed skeletal muscle ratio as an independent protective factor (penetrating: HR = 0.098 [0.014 - 0.680], p = 0.02; non-penetrating: HR = 0.597 [0.442 - 0.804], p = 0.001; combined: HR = 0.637 [0.493 - 0.823], p = 0.001), while penetrating disease as risk factor (HR = 3.778 [1.281 - 11.14], p = 0.02). Body mass index mediated 6.9% of the skeletal muscle ratio-treatment escalation relationship in non-penetrating CD. CONCLUSIONS: Skeletal muscle-IMAT imbalance independently predicts treatment escalation in CD, offering superior prognostic utility to traditional sarcopenia measures.