Background & aims: Endoscopic trimodal imaging (ETMI) incorporates high-resolution endoscopy (HRE) and autofluorescence imaging (AFI) for adenoma detection, and narrow-band imaging (NBI) for differentiation of adenomas from nonneoplastic polyps. The aim of this study was to compare AFI with HRE for adenoma detection and to assess the diagnostic accuracy of NBI for differentiation of polyps. This was a randomized trial of tandem colonoscopies. The study was performed at the Academic Medical Center in Amsterdam.
Methods: One hundred patients underwent colonoscopy with ETMI. Each colonic segment was examined twice for polyps, once with HRE and once with AFI, in random order per patient. All detected polyps were assessed with NBI for pit pattern and with AFI for color, and subsequently removed. Histopathology served as the gold standard for diagnosis. The main outcome measures of this study were adenoma miss-rates of AFI and HRE, and diagnostic accuracy of NBI and AFI for differentiating adenomas from nonneoplastic polyps.
Results: Among 50 patients examined with AFI first, 32 adenomas were detected initially. Subsequent inspection with HRE identified 8 additional adenomas. Among 50 patients examined with HRE first, 35 adenomas were detected initially. Successive AFI yielded 14 additional adenomas. The adenoma miss-rates of AFI and HRE therefore were 20% and 29%, respectively (P = .351). The sensitivity, specificity, and overall accuracy of NBI for differentiation were 90%, 70%, and 79%, respectively; corresponding figures for AFI were 99%, 35%, and 63%, respectively.
Conclusions: The overall adenoma miss-rate was 25%; AFI did not significantly reduce the adenoma miss-rate compared with HRE. Both NBI and AFI had a disappointing diagnostic accuracy for polyp differentiation, although AFI had a high sensitivity.