Diagnostic performance of CT for the detection of transmural bowel necrosis in non-occlusive mesenteric ischemia

Objectives To evaluate the diagnostic performance of CT for transmural necrosis (TN) in non-occlusive mesenteric ischemia (NOMI) according to the bowel segment involved. Methods From January 2009 to December 2019, all patients admitted to the intensive care unit (ICU) and requiring laparotomy for NOMI were retrospectively studied. CT had to have been performed within 24 h prior to laparotomy and were reviewed by two abdominal radiologists, with a consensus reading in case of disagreement. A set of CT features of mesenteric ischemia were assessed, separating the stomach, jejunum, ileum, and right (RC) and left colon (LC). Univariate and multivariate analyses were performed to identify features associated with TN. Its influence on overall survival (OS) was assessed. Results Among 145 patients, 95 (66%) had ≥ 1 bowel segment with TN, including 7 (5%), 31 (21%), 43 (29%), 45 (31%), and 52 (35%) in the stomach, jejunum, ileum, RC, and LC, respectively. Overall inter-reader agreement of CT features was significantly lower in the colon than in the small bowel (0.59 [0.52–0.65] vs 0.74 [0.70–0.77] respectively). The absence of bowel wall enhancement was the only CT feature associated with TN by multivariate analysis, whatever the bowel segment involved. Proximal TN was associated with poorer OS ( p < 0.001). Conclusions The absence of bowel wall enhancement remains the most consistent CT feature of transmural necrosis, whatever the bowel segment involved in NOMI. Inter-reader agreement of CT features is lower in the colon than in the small bowel. Proximal TN seems to be associated with poorer OS. Key Points • The absence of bowel wall enhancement is the most consistent CT feature associated with transmural necrosis in NOMI, whatever is the bowel segment involved. • Inter-reader agreement is lower in the colon than in the small bowel in NOMI. • In NOMI, the more proximal the bowel necrosis, the worse the prognosis.