Diagnostic performance of narrowed spectrum endoscopy, autofluorescence imaging, and confocal laser endomicroscopy for optical diagnosis of colonic polyps: a meta-analysis

Background: Novel endoscopic technologies could allow optical diagnosis and resection of colonic polyps without histopathological testing. Our aim was to establish the sensitivity, specificity, and real-time negative predictive value of three types of narrowed spectrum endoscopy (narrow-band imaging [NBI], image-enhanced endoscopy [i-scan], and Fujinon intelligent chromoendoscopy [FICE]), confocal laser endomicroscopy (CLE), and autofluorescence imaging for differentiation between neoplastic and non-neoplastic colonic lesions.

Methods: We identified relevant studies through a search of Medline, Embase, PubMed, and the Cochrane Library. Clinical trials and observational studies were eligible for inclusion when the diagnostic performance of NBI, i-scan, FICE, autofluorescence imaging, or CLE had been assessed for differentiation, with histopathology as the reference standard, and for which a 2 × 2 contingency table of lesion diagnosis could be constructed. We did a random-effects bivariate meta-analysis using a non-linear mixed model approach to calculate summary estimates of sensitivity and specificity, and plotted estimates in a summary receiver-operating characteristic curve.

Findings: We included 91 studies in our analysis: 56 were of NBI, ten of i-scan, 14 of FICE, 11 of CLE, and 11 of autofluorescence imaging (more than one of the investigated modalities assessed in eight studies). For NBI, overall sensitivity was 91·0% (95% CI 88·6-93·0), specificity 85·6% (81·3-89·0), and real-time negative predictive value 82·5% (75·4-87·9). For i-scan, overall sensitivity was 89·3% (83·3-93·3), specificity 88·2% (80·3-93·2), and real-time negative predictive value 86·5% (78·0-92·1). For FICE, overall sensitivity was 91·8% (87·1-94·9), specificity 83·5% (77·2-88·3), and real-time negative predictive value 83·7% (77·5-88·4). For autofluorescence imaging, overall sensitivity was 86·7% (79·5-91·6), specificity 65·9% (50·9-78·2), and real-time negative predictive value 81·5% (54·0-94·3). For CLE, overall sensitivity was 93·3% (88·4-96·2), specificity 89·9% (81·8-94·6), and real-time negative predictive value 94·8% (86·6-98·1).

Interpretation: All endoscopic imaging techniques other than autofluorescence imaging could be used by appropriately trained endoscopists to make a reliable optical diagnosis for colonic lesions in daily practice. Further research should be focused on whether training could help to improve negative predictive values.

Funding: None.