Lars J Grimm MD, Presenter: Nothing to Disclose

Sebastian Feuerlein MD, Abstract Co-Author: Nothing to Disclose

Kelan James Brown MD, Abstract Co-Author: Nothing to Disclose

Matthew Scott Davenport MD, Abstract Co-Author: Nothing to Disclose

Clare M. Haystead MD, Abstract Co-Author: Nothing to Disclose

Chad Michael Miller MD, Abstract Co-Author: Nothing to Disclose

Amy Marie Neville MD, Abstract Co-Author: Nothing to Disclose

Tracy Anne Jaffe MD, Abstract Co-Author: Research Consultant, Osiris Therapeutics, Inc

To investigate the ability of computed tomography (CT) to detect and accurately localize colonic tumors and to compare localization to that of endoscopic colonoscopy.

The local institutional review board approved this retrospective and HIPAA-compliant study. Forty-six patients with colonic adenocarcinomas and both preoperative colonoscopy and staging CT within 60 days of resection were included in the study. All patients underwent preoperative contrast enhanced CT imaging without bowel preparation or oral contrast. The colon was divided into 4 segments (sigmoid, descending, transverse and ascending). Operative reports were used as a gold standard for tumor localization. CT scans were reviewed by 5 readers in a segmental binary fashion using a 5-point confidence scale in two reading sessions (read 1, without colonoscopy report; read 2, with report). A colonic segment was considered positive for tumor if it was rated positive by 3 or more readers.

At surgery 49 tumors were found in 46 patients. CT correctly localized more lesions 46/49 (93.9%) than colonoscopy 39/49 (79.6%). Readers confidently identified 38 of 49 tumors (77.6%) based on a confidence score of ≥4with 94.7% accuracy. For a confidence score cutoff of 5 the accuracy remained very high (96.6%) while more tumors were missed (reduced sensitivity of 59.2%). The overall accuracy of colonoscopic localization was lower with 39 of 49 (79.6%) correctly localized. Accuracy was particularly diminished in the descending colon (42.9%) and the transverse colon (55.6%). CT interobserver agreement was almost perfect (κ=0.82) for the unblinded and moderate (κ=0.60) for the blinded read. There was a significant positive correlation between tumor size and confidence score with a Pearson correlation coefficient of 0.523.

CT is more accurate than colonoscopy in the localization of colonic tumors.

If a tumor is seen on CT with high confidence in a different colonic segment than given in the colonoscopy report, CT is more accurate and should be used for planning of the operative procedure.

Grimm, L, Feuerlein, S, Brown, K, Davenport, M, Haystead, C, Miller, C, Neville, A, Jaffe, T, Localization of Primary Colonic Tumors: A Comparison of CT and Colonoscopy.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11034428.html