Colorectal tumors detected by routine CT colonography (CTC) screening are significantly smaller and less conspicuous to radiologists than the tumors of symptomatic patients detected by CTC. They also have different morphology, with screen-detected colorectal cancer being more frequently polypoid, according to a British study published in the December issue of European Radiology.
CTC is the diagnostic imaging exam of choice for patients with symptoms of colorectal cancer. CTC is also increasingly being used for routine screening in lieu of conventional colonoscopy. Radiologists need to be aware of their differing visual characteristics and also that computer-assisted detection (CAD) software may be significantly less likely to identify screen detected CRC.
Lead author Andrew A. Plumb, MD, of the Center for Medical Imaging of University College London, and co-authors compared the morphology, radiological stage, conspicuity and CAD characteristics of colorectal cancers in screening and symptomatic populations to determine any differences between these two patient groups. They obtained CTC images of 35 symptomatic patients depicting 36 cancers from 10 hospitals that participated in the Special Interest Group in Gastrointestinal and Abdominal Radiology (SIGGAR) randomized clinical trial. CTC images of 98 asymptomatic patients diagnosed with 100 cancers were obtained from the English Bowel Cancer Screening Program (EBCSP). These patients had undergone CTC screening at one of 25 different centers between 2006 and 2014.
Two radiologists independently reviewed both sets of images. They used prone and supine images to record tumor morphology, the subtype of polypoid lesions when identified and for non-polypoid lesions if they were angular or non-angular, presence/absence of luminal stenosis, tumor dimensions, radiological T stage, extramural depth of spread beyond the muscularis propria for T3 and T4 lesions, presence/absence of radiologically involved lymph nodes, and presence/absence of vascular invasion. The radiologists also were asked to subjectively rate image quality using a combined assessment of bowel cleansing and distension on a 3-point scale (good/moderate/poor).
The authors generated objective measures of tumor location, ease of detection, and tumor volume to minimize subjectivity by the readers. They also recorded the total number of CAD marks for each patient, and the presence/absence of at least one CAD mark within 5 mm of the tumor in any direction.
The authors reported that more than two-thirds of tumors in both asymptomatic and symptomatic patients were left-sided, and that there was no significant difference in mean distance along the colonic centerline to the tumor. However, screen-detected tumors were more likely to be polypoidal (34%) than symptomatic tumors (13.9%). There were significantly fewer annual cancers in the screening group compared to the systematic group when considering only non-polypoidal tumors (40.9% compared to 67.7%).
Screen-detected tumors were also significantly smaller, with a median long-axis dimension of 3.0 cm compared to 4.3 cm for the symptomatic tumors. They also had a significantly lower volume (median 9.1 cm3, IQR 3.5-20.1 cm3 compared to 23.2 cm3, IQR 9.5-43.6 cm3 respectively). Also, screen-detected cancers were significantly less likely to cause 50% or greater luminal stenosis, at 17% compared to symptomatic tumors at 38.9%.
Both radiologists stated that the screen-detected tumors were significantly less conspicuous, although the scores assigned by each were differed considerably. The proportion of cancers missed by CAD also was significantly greater for screen-detected tumors. CAD marked 72 out of 93 screen detected-tumors, for a sensitivity of 77.4% compared to 31 out of 35 symptomatic tumors for a sensitivity of 96.9%.
Screen-detected tumors were of significantly lower local stage and were significantly less likely to have radiologically involved lymph nodes (28% screening compared to 55.6% symptomatic). As expected, asymptomatic patients had better prognoses: 24% of screening-group patients had a poor prognosis compared to 58.3% of the symptomatic patient group.
The authors point out that radiologists need to be aware of these differences. They recommend that radiologists embarking on screening CTC may need specific guidance, training and quality assurance prior to reporting for a screening program. They recommend this because many radiologists gain their experience in interpreting CTC findings with symptomatic patients, and may overlook the subtleties of colon cancer characteristics in an asymptomatic screening population. They also caution that while CAD is useful, it does not have the sensitivity to identify tumors in an asymptomatic screening population as it does for symptomatic patients.
CT colonography screen-detected cancers differ from cancers in symptomatic patients. Appl Radiol.