Performance of computed tomography and its reliability for the diagnosis of transmural gastrointestional necrosis in a setting of acute ingestion of predominantly strong acid substances in adults

Abstract

Objectives Computed tomography has become a critical component in evaluating adult patients with acute caustic ingestions and an alternative to endoscopy for detecting transmural gastrointestinal necrosis. This study assessed the performance and reliability of computed tomography findings of transmural gastrointestinal necrosis, given that the presence of the disease potentially signifies the need for surgery. Methods A retrospective database search was performed to identify consecutive adult patients with acute caustic ingestions who had computed tomography with endoscopy or surgery within 72 h of admission. Eight physicians reinterpreted computed tomography in two separate rounds. Diagnostic performance utilized eight rounds of radiologists' reinterpretations against reference endoscopic or surgical grades. Intra- and interobserver agreements were calculated. Results Seventeen patients (mean age, 45.6 years; 9 men; 46 esophageal and 34 gastric segments; 16 ingested strong acid substances) met the inclusion criteria. Eight patients (10 esophageal and 13 gastric segments) had transmural gastrointestinal necrosis. The highly differentiating findings between those with and without transmural gastrointestinal necrosis were esophageal wall thickening (100% vs. 42%, P = 0.001; 100% sensitive), gastric abnormal wall enhancement and fat stranding (100% vs. 57%, P = 0.006; 100% sensitive), and gastric absent wall enhancement (46% vs. 5%, P = 0.007; 100% specific). The intra- and interobserver percentage agreements were 47–100%, and 54–100%, which increased to 53–100%, and 60–100%, respectively, when considering only radiologists’ reinterpretations. Conclusions In a very small sample of adults who primarily ingested acid, contrast-enhanced computed tomography performed well when interpreted by a panel of radiologists.