Autofluorescence endoscopy - not much gain after all?

As endoscopy advances in the 21st century, the quest for identifying early gastrointestinal neoplasia is ever more extensive for the enthusiastic endoscopist. We are now moving beyond white-light endoscopy (WLE) towards newer imaging techniques that utilize the properties of light-tissue interaction [1]. When light is directed against a tissue there can be four different outcomes: reflection, scattering, absorption, and fluorescence. The phenomenon of autofluorescence occurs when a light of shorter wavelength interacts with a tissue that contains endogenous fluorophores that then emit light of longer wavelength [2]. The term “fluorophore” is applied to a number of biological substances (collagen, elastin, nicotinamide, flavin, adenine dinucleotide, and porphyrins) that are normally present in the gastrointestinal mucosa and can emit autofluorescence.

Initially, autofluorescence imaging systems used fiberoptic endoscopes that failed to provide sufficient image quality to allow clinical applicability. However it is now possible to identify autofluorescence in real-time video endoscopy where pseudocolor images are produced using a charge-coupled device (CCD) [3]. This high resolution video endoscopy system allows both WLE and autofluorescence endoscopy (AFE) to be performed interchangeably using a single endoscope.

Autofluorescence is abnormal in neoplastic tissue because of: (i) the increase in the nuclear-cytoplasmic ratio, which leads to reduction of autofluorescence in neoplasia as nuclei show no autofluorescence compared with cytoplasm; (ii) loss of collagen, since collagen in submucosa is the strongest fluorophore and thickening of the mucosa in neoplasia results in the elimination of submucosal collagen autofluorescence; and (iii) neovascularization, resulting in increased hemoglobin concentration that absorbs autofluorescence light.

In this issue of Endoscopy, Katoh et al. report a test of the latest AFE system from the Olympus Corporation (Tokyo, Japan), in a prospective blinded study where AFE was compared with WLE in the detection of early superficial gastric cancer in a high risk population [4]. They have used a robust methodology, carefully avoiding bias from either technique by using two different endoscopists performing the two techniques independently with a neutral observer collecting overall data. They studied 51 patients with 91 lesions (39 neoplastic), with histology providing the gold standard final diagnosis. Of the neoplastic lesions, 56 % (22/39) were diagnosed by both WLE and AFE, 18 % (7/39) by WLE alone and 13 % (5/39) by AFE alone. Of 7 neoplasms detected only by WLE, 6 (86 %) were depressed lesions. In contrast, 3 of 5 neoplasms (60 %) detected by AFE alone were elevated lesions. Sensitivities of WLE alone and AFE alone were 74 % and 64 % (difference not significant) with corresponding specificities of 83 % and 40 % (P = 0.0003). The authors have concluded that AFE is of limited value in detecting superficial gastric neoplasia because of its poor specificity. Its sensitivity is comparable to but not better than that of WLE, thus its added value is questionable. However, a few neoplasms (mainly elevated lesions) were detected by AFE alone; this might have some clinical relevance that needs to be tested in future studies.

From an endoscopist’s perspective, the attraction of using advanced imaging techniques is to improve accuracy when screening for early superficial neoplasia. Any new imaging technique should be better than WLE in detecting such lesions. The sensitivity of WLE alone in this study was 74 %, which set quite a high value for AFE to compete against and demonstrate a statistically significant difference. The relative sensitivities obtained could be influenced by: (i) the standard of WLE used in this study, which was high resolution imaging; (ii) the weakness of the autofluorescence signal, given that the current AFE system is limited by low signal-to-noise ratios; and (iii) the skill of experienced Japanese endoscopists in identifying early gastric cancers. The possibility of false negatives with AFE is also of concern, especially with depressed lesions that tend to harbor more sinister disease.

The other important aspect of AFE with regard to accuracy is the low specificity (high false-positive rates). This is not a surprise and has been reported in Barrett’s esophagus [5]. In a recent multicenter international study, using trimodal imaging that incorporated high resolution endoscopy, AFE, and narrow band imaging (NBI) to detect early neoplasia in Barrett’s esophagus, AFE identified 90 % of the patients with neoplastic lesions, significantly more than the 53 % identified by WLE [6]. However the false-positive rate was 81 %; this was overcome by using NBI for detailed examination, which reduced the false-positive rate to 26 %. In this scenario, AFE can be used as a ‘red flag’ technique, aiding the endoscopist to concentrate on suspicious areas while a technique such as NBI with magnification is used to explore minute details.

It is now apparent that the performance of AFE may vary in different parts of the gastrointestinal tract. This could be due to tissue characteristics such as atrophy and inflammation, light intensity, lumen size, and other factors such as luminal contents. Advanced endoscopic imaging is still in its infancy, albeit we have nearly reached the summit with WLE. Although we have not gained much with AFE in the stomach, further refinements in endoscope technology are expected that might overcome the limitations of the present AFE system. Until then, AFE in the stomach is not yet ready for “prime time” deployment.

Competing interests: None

• 1 Dacosta R S, Wilson B C, Marcon N E. New optical technologies for earlier endoscopic diagnosis of premalignant gastrointestinal lesions.  J Gastroenterol Hepatol. 2002;  17 Suppl S85-S104
• 2 Dacosta R S, Wilson B C, Marcon N E. Light-induced fluorescence endoscopy of the gastrointestinal tract.  Gastrointest Endosc Clin N Am. 2000;  10 37-69
• 3 Uedo N, Iishi H, Tatsuta M. et al . A novel videoendoscopy system by using autofluorescence and reflectance imaging for diagnosis of oesophagogastric cancers.  Gastrointest Endosc. 2006;  62 512-528
• 4 Kato M, Kaise M, Yonezawa J. et al . Autofluorescence endoscopy versus conventional white light endoscopy for the detection of superficial gastric neoplasia: a prospective comparative study.  Endoscopy. 2007;  39 937-941
• 5 Kara M A, Peters F P, Ten Kate F J. et al . Endoscopic video autofluorescence imaging may improve the detection of early neoplasia in patients with Barrett’s oesophagus.  Gastrointest Endosc. 2005;  61 679-685
• 6 Curvers W LR, Singh R, Wong-Kee Song L M. et al . Endoscopic tri-modal imaging for detection of early neoplasia in Barrett’s oesophagus: a multi-centre feasibility study using high-resolution endoscopy, autofluorescence imaging and narrow band imaging incorporated in one endoscopy system.  Gut. 2007;  in press