A prospective study of the accuracy of preoperative computed tomographic staging of patients with biopsy-proven rectal carcinoma

From June 1983 to January 1986, 91 patients with biopsy-proven adenocarcinoma of the rectum had computed tomographic scans of the pelvis performed before treatment as part of a "sandwich" radiotherapy-surgery regimen. Two experienced diagnostic radiologists performed locoregional staging of all scans according to the University of California at San Francisco criteria; one of these radiologists repeated this staging at a later time to test the reproducibility of a single observer. Staging was performed without the use of any other radiographic studies or of any clinical information except the patients' age, sex, and the diagnosis of rectal carcinoma, to test the value of computed tomographic scans alone for staging. Agreement between the two stagings performed by the first observer was 51 percent, and interobserver agreement was only 37 percent. Agreement with Dukes' staging was only 33 percent. Therefore, preoperative pelvic computed tomographic scanning of primary rectal adenocarcinoma should not be relied upon for staging or for the selection of patients for treatment options.