Screening for colon polyps using spiral CT with and without virtual colonoscopy

BACKGROUND: New colon imaging techniques called CT computed colography and virtual colonoscopy (VC) are being developed to detect colon neoplasia and perform colorectal cancer screening and are based on spiral CT. AIM: Evaluate the sensitivity of spiral CT without and with virtual endoscopy software in a blinded fashion in a screening population using colonoscopy (CS) as gold standard. METHOD: 50 asymptomatic subjects (SJ) with no prior polyps or cancer participated and were given bowel prep. SJ had either never had a colon examination or had a screening flex sig which showed a polyp. Glucagon was given IV and carbon dioxide was given per rectum. Spiral CT was performed with the CT Twin instrument (Elscint, Inc., Hackensack, NJ) with the patient supine using dual-slice technique, with the beam collimated to 10 mm across 2 parallel detector arrays using a table speed of 15 mm/sec (slice width = 5 mm). The patient was turned prone and the scan repeated. Axial images were reviewed blindly by KK. CT datasets were processed by DV using the Bowman Gray Virtual Endoscopy software (FreeFlight) running on a Silicon Graphics Onyx workstation (Silicon Graphics, Mountain View, CA). DV inspected intreractive multiplanar (axial, sagittal, coronal), 3D volume-rendered images of the colon. Colonoscopy was performed the same day as CT scan with videotaping of all examinations and removal of all polyps. RESULTS: Of 27 fully evaluated SJ (mean age 68 yr, 25 male), 17 had never been examined and 10 had polyp(s) on flex sig. 19 SJ had 51 adenomas (AD) at CS, including 9 ≥ 1 cm (all sizes confirmed by path measurement), 8 were 6-9 mm and 34 were ≤ 5 mm in size. Inspection of 2-D axial CT images detected 2 AD in each size range (sensitivity 22%, 25% and 6%, respectively). VC detected 3 of 9 AD ≥ 1 cm (33%), 3 of 8 AD 6-9 mm (38%) and 3 of 34 AD ≤ 5 mm (9%). VC was normal in 15 of the 19 SJ with AD, including 4 of 6 with one or more AD≥1 cm in size and 2 of 2 with an AD ≥ 20mm in size (both sessile). SUMMARY: Despite state of the art spiral CT and virtual endoscopy software, VC had low sensitivity for AD when evaluated blindly in an asymptomatic population, although the population was selected to increase prevalence. VC did not substantially improve the sensitivity of reading the 2-D CT images. CONCLUSION: Further blinded testing should be done in asymptomatic persons before VC is used for colorectal cancer screening.