Am Fam Physician. 1999;60(1):274-277
Computed tomographic (CT) scans are often used to establish a diagnosis of suspected colon cancer or colonic diverticulitis. Although CT images are generally effective in identifying these conditions, the imaging appearance overlaps in about 10 percent of patients. Chintapalli and associates sought to identify specific signs on CT scan to help distinguish diverticulitis from colon cancer. They conducted a retrospective review of CT images from patients with confirmed diverticulitis or colon cancer to establish diagnostic criteria. These criteria were then compared prospectively with CT images from another series of patients with the same suspected diagnoses.
To establish the criteria, 58 CT images were obtained from a series of patients diagnosed with diverticulitis or colon cancer and evaluated retrospectively by five radiologists for the presence or absence of previously reported CT findings. The findings were correlated with the final clinical or histologic diagnosis. These initial CT findings were then compared prospectively with a series of 72 CT scans from different patients and evaluated by the same five radiologists. Each new scan was evaluated on a confidence scale of one to five, ranging from “definitely diverticulitis” to “definitely malignant.” Individual CT diagnoses were correlated with final clinical diagnoses. The combinations of readings from the five radiologists were tabulated to identify the overlap in each of the categories on the confidence scale. CT findings for which the accuracy of diagnosis was the highest were determined.
In the retrospective part of the study, 27 patients were diagnosed with diverticulitis and 31 were diagnosed with colon cancer. The findings most specific for diverticulitis were pericolonic inflammation and a length of segment longer than 10 cm. The presence of lymph nodes and a luminal mass were most specific for colon cancer. The presence of pericolonic edema suggested diverticulitis, but the results were not statistically significant in this portion of the study.
In the prospective part of the study, 40 patients were diagnosed with diverticulitis and 32 were diagnosed with colon cancer. Eighty-three readings were rated definite diverticulitis and, of these, 80 were diverticulitis and three were colon cancer. Ninety-six readings were rated definite colon cancer and, of these, 92 were cancer and four were diverticulitis. The individual readings of the CT images were 96 percent accurate in making a correct, unequivocal diagnosis, suggesting a very low rate of interpretive mistakes.
Additional analysis showed that fluid at the root of the mesentery, pericolonic edema and pericolonic inflammation had the highest specificity for diverticulitis. The presence of pericolonic lymph nodes, shoulder formation and a luminal mass had the highest specificity for colon cancer. Although the presence of pericolonic edema was not significant in the retrospective review, it became highly significant in the prospective study. The presence of lymph nodes with or without pericolonic edema resulted in a sensitivity of 90 percent and a diagnostic accuracy of 79 percent for colon cancer.
The authors conclude that the specificity of the CT findings in identifying diverticulitis and colon cancer is so high that, depending on the radiologist's experience, an unequivocal diagnosis carries an accuracy of up to 100 percent for both diverticulitis and colon cancer. In this situation, no additional diagnostic work-up is necessary, and further evaluation for surgical planning may be initiated.