Endoscopy 2008; 40(10): 811-817
DOI: 10.1055/s-2008-1077586
Original article

© Georg Thieme Verlag KG Stuttgart · New York

Narrow band imaging with magnification for the characterization of small and diminutive colonic polyps: pit pattern and vascular pattern intensity

J.  E.  East1 , N.  Suzuki1 , P.  Bassett1 , M.  Stavrinidis1 , H.  J.  W.  Thomas2 , T.  Guenther3 , P.  P.  Tekkis4 , B.  P.  Saunders1
  • 1Wolfson Unit for Endoscopy, St Mark’s Hospital, Imperial College London, London, UK
  • 2Family Cancer Group, Cancer Research Colorectal Cancer Unit, St Mark’s Hospital, Imperial College London, London, UK
  • 3Academic Department of Cellular Pathology, St Mark’s Hospital, Imperial College London, London, UK
  • 4Department of Biosurgery and Surgical Technology, Imperial College London, London, UK
Further Information

Publication History

submitted 20 January 2008

accepted after revision 9 July 2008

Publication Date:
30 September 2008 (online)

Zoom Image

Background and study aims: Narrow band imaging (NBI) can accurately characterize colonic polyps using microvascular appearances. We aimed to assess whether the Kudo pit pattern classification is accurate when used with NBI (without dye-spray), and if microvascular appearances or NBI pit patterns maintain accuracy for polyp characterization at sizes < 10 mm.

Patients and methods: 116 polyps < 10 mm in size were detected in 62 patients undergoing surveillance colonoscopy. The polyps were prospectively assessed using NBI and magnification for Kudo pit pattern (III-V neoplastic, I-II non-neoplastic) and vascular pattern intensity (VPI), a measure of microvascular density (strong VPI, neoplastic; normal or weak VPI, non-neoplastic). Sensitivity, specificity, and accuracy were calculated and compared with results from histopathology.

Results: The mean polyp size was 3.4 mm (range 1 – 9 mm). Overall, NBI pit pattern sensitivity, specificity, and accuracy were 0.88, 0.91, and 89.6 %, respectively. Equivalent values for VPI were 0.94, 0.89, and 91.4 %. Results were similar when polyps were subdivided into diminutive polyps (size ≤ 5 mm) and flat polyps. Combining both pit pattern and VPI improved the sensitivity (0.98, P = 0.06 versus NBI pit pattern alone). There was very good agreement between NBI pit pattern and VPI for prediction of dysplasia (kappa = 0.83). No evidence of a learning curve for VPI was found. The NBI pit pattern was better than the VPI at subclassifying hyperplastic from other non-neoplastic polyps (sensitivity 0.79 versus 0.56, respectively, P = 0.02), but accuracy was poor.

Conclusion: The NBI pit pattern and VPI are both highly accurate in characterizing neoplastic colonic polyps of < 10 mm, with VPI appearing to be simple to learn. NBI has the potential to replace conventional histology for small polyps.