Advertisement

Near-focus narrow-band imaging for endoscopic assessment of duodenal villi: Making the case more than ever?

Published:October 19, 2021DOI:https://doi.org/10.1016/j.gie.2021.09.006

      Abbreviations:

      CD (celiac disease), IOA (interobserver agreement), NBI (narrow-band imaging), NF (near focus), NPV (negative predictive value), VA (villous atrophy), WLE (white-light endoscopy.)
      Examination of the duodenal mucosa is necessary to rule out, rule in, and monitor various digestive disorders and is a common indication for EGD. Celiac disease (CD), a complex, polygenic, gluten-induced inflammatory enteropathy, is a hallmark example of a disorder wherein duodenal mucosal examination plays a fundamental role in clinical management. EGD has the benefit of facilitating direct and expedient assessment in this regard, and in CD, several classic endoscopic manifestations may be observed, including mucosal fissuring (ie, “cracked mud appearance”), scalloping or loss of the valvulae conniventes, and of course villous atrophy (VA).
      • Jabbari M.
      • Wild G.
      • Goresky C.A.
      • et al.
      Scalloped valvulae conniventes: an endoscopic marker of celiac sprue.
      Nevertheless, conventional white-light endoscopy (WLE) is insufficiently sensitive to enable confident assessment of the mucosa for certain conditions and findings, particularly when they are mild and/or subtle
      • Barada K.
      • Habib R.H.
      • Malli A.
      • et al.
      Prediction of celiac disease at endoscopy.
      ; for this reason, biopsy specimens are often obtained from the duodenum, often in an untargeted fashion, for histopathologic evaluation. However, this is a time- and resource-intensive approach and, albeit a generally safe undertaking, carries a small risk of overt postbiopsy bleeding. Therefore, considering factors such as how commonplace it is to perform a biopsy of the duodenum, the possibility of unrepresentative biopsy specimens resulting from patchy disease, the fact that a large proportion of duodenal biopsy specimens result as histopathologically normal, and the lifelong nature of CD (which may imply multiple EGDs in symptomatic patients over time), a reliable method to avoid unnecessary biopsies would be timely and useful from both a public health and an individual patient care perspective.
      Narrow-band imaging (NBI) is a form of electronic chromoendoscopy that offers enhanced capability to delineate mucosal surfaces and underlying vasculature compared with WLE.
      Olympus
      Technology.
      With the push of a button, NBI mode filters white light into 2 narrow bands at specific wavelengths: 415-nm blue and 540-nm green light components, respectively. Fundamentally underpinning the enhanced capability of NBI for examining the duodenal mucosa is the principle that depth of light penetration is proportionate to wavelength; ie, the longer the wavelength of light, the deeper the tissue penetration. Compared with biopsy-based techniques, NBI offers advantages in that it (1) is noninvasive (beyond the risks inherent to EGD); (2) does not require additional setup, (3) can be readily performed by the endoscopist and in real time; (4) may avoid a need for biopsies and associated histopathology charges; and (5) is theoretically less prone to sampling error. For these and other reasons, several groups have explored the utility of NBI-based examination of the duodenum in patients undergoing EGD, including all comers as well as those with suspected and known CD.
      • Tabibian J.H.
      • Perrault J.F.
      • Murray J.A.
      • et al.
      Narrow band imaging evaluation of duodenal villi in patients with and without celiac disease: a prospective study.
      • Singh R.
      • Nind G.
      • Tucker G.
      • et al.
      Narrow-band imaging in the evaluation of villous morphology: a feasibility study assessing a simplified classification and observer agreement.
      • De Luca L.
      • Ricciardiello L.
      • Rocchi M.B.
      • et al.
      Narrow band imaging with magnification endoscopy for celiac disease: results from a prospective, single-center study.
      • Banerjee R.
      • Reddy D.N.
      High-resolution narrow-band imaging can identify patchy atrophy in celiac disease: targeted biopsy can increase diagnostic yield.
      • Cammarota G.
      • Cesaro P.
      • La Mura R.
      • et al.
      Role of the "immersion technique" in diagnosing celiac disease with villous atrophy limited to the duodenal bulb.
      • Koay D.S.C.
      • Ghumman A.
      • Pu L.
      • et al.
      Narrow-band imaging with magnification and the water immersion technique: a case-finding, cost-effective approach to diagnose villous atrophy.
      • Tchekmedyian A.J.
      • Coronel E.
      • Czul F.
      "Leopard skin sign": the use of narrow-band imaging with magnification endoscopy in celiac disease.
      In one of the largest studies to date reporting on NBI for the assessment of duodenal villous morphology, we found a notably high (92% to 100%) negative predictive value (NPV) across a broad range of observer expertise.
      • Tabibian J.H.
      • Perrault J.F.
      • Murray J.A.
      • et al.
      Narrow band imaging evaluation of duodenal villi in patients with and without celiac disease: a prospective study.
      A few groups have also explored NBI in combination with other endoscopic techniques to further enhance diagnostic performance, including water immersion and optical magnification, with similarly promising findings.
      • Cammarota G.
      • Cesaro P.
      • La Mura R.
      • et al.
      Role of the "immersion technique" in diagnosing celiac disease with villous atrophy limited to the duodenal bulb.
      • Koay D.S.C.
      • Ghumman A.
      • Pu L.
      • et al.
      Narrow-band imaging with magnification and the water immersion technique: a case-finding, cost-effective approach to diagnose villous atrophy.
      • Tchekmedyian A.J.
      • Coronel E.
      • Czul F.
      "Leopard skin sign": the use of narrow-band imaging with magnification endoscopy in celiac disease.
      Collectively, these studies have generated increasing interest in refining methodologies to optimize optical diagnosis and, when necessary, selectively obtaining biopsy specimens.
      • Tabibian J.H.
      • Perrault J.F.
      • Murray J.A.
      • et al.
      Narrow band imaging evaluation of duodenal villi in patients with and without celiac disease: a prospective study.
      ,
      • Ianiro G.
      • Bibbo S.
      • Pecere S.
      • et al.
      Current technologies for the endoscopic assessment of duodenal villous pattern in celiac disease.
      In this issue of Gastrointestinal Endoscopy, Gulati et al
      • Gulati S.
      • Emmanuel A.
      • Ong M.
      • et al.
      Near-focus narrow-band imaging classification of villous atrophy in suspected celiac disease: development and international validation.
      report on the findings of their prospective, United Kingdom–based, international collaborative study wherein they developed and validated a classification schema to examine the duodenal mucosa using NBI in combination with near focus (NF-NBI). NF is a dual-focus 2-stage optical lens technology from Olympus (Olympus Corporation, Tokyo, Japan), which, as with NBI, can be selected with the push of a button to facilitate high-resolution close-up examination of the mucosa while maintaining a broad field of view.
      Olympus
      Technology.
      In the present study, the investigators enrolled 100 consecutive adult patients (97 completed the study) who were referred for EGD with a clinical indication for duodenal biopsy, including 21 patients with known CD. All EGDs were performed by a single experienced endoscopist using an Olympus GIF-HQ290 endoscope and a CV-290 processor. The image capture protocol included expansive examination of the duodenum under WLE followed by capturing of 6 paired NF-WLE/NF-NBI images before duodenal biopsy. The biopsy protocol consisted of 6 biopsy specimens (4 from D2, 2 from D1) per patient taken with large-capacity radial-jaw forceps. Histopathology was scored based on the Marsh-Oberhuber classification. All EGDs were videorecorded and later reviewed to ensure that the paired images corresponded with the biopsy site. After image quality review, NF-NBI images (n = 548) were grouped into 3 nonoverlapping datasets: 2 for modified Delphi rounds and 1 for validation. The first Delphi round consisted of 3 experienced endoscopists reviewing 312 paired images from 87 patients so as to define NF-NBI classifiers; this resulted in the development of a 3-descriptor (villous shape, vascular pattern, and crypt phenotype) classification schema. The second Delphi round consisted of 2 experienced endoscopists reviewing 186 images from 86 patients using the 3 aforementioned descriptors and accepting them if the interobserver agreement (IOA) threshold (κ ≥ 0.6) was satisfied. Validation consisted of 13 endoscopists (5 expert, 8 nonexpert) from around the world taking a short training module (consisting of 13 images) on the proposed classification schema, on the basis of which they reviewed 50 paired images. The investigators reported numerous diagnostic performance and IOA statistics using the proposed classification schema, with biopsy as the criterion standard. Among the key findings were the following: substantial agreement among all 3 NF-NBI descriptors and histology (weighted κ = 0.72 to 0.75) as compared with NF-WLE and histology (κ = 0.34); high overall sensitivity and NPV of the NF-NBI classification schema for diagnosis of VA (range, 91.67% to 100% and 92.59% to 100%, respectively); and no significant difference in the performance of expert versus nonexpert endoscopist reviewers.
      Several factors should be considered in the interpretation of the findings of this study and its implications moving forward, including these: (1) Would the results have been different if the endoscopist obtaining the images (as may be the case in a real-world setting) did not have the same level of expertise as the single performing endoscopist in this study? (2) Why was the “expansive examination” of the duodenum performed with WLE, not NBI? The latter could be performed in normal mode and used to target near mode (ie, NF-NBI), which may enhance overall performance characteristics. (3) It is stated that “A short period of training of only 13 images was required to achieve a significant benefit in diagnostic performance using the proposed NF-NBI classification over NF-WLE”; however, it is unclear what this training consisted of (eg, what 13 images, and how was instruction provided?), and the benefit ostensibly was not from the training but rather the inherent advantages of NBI over WLE in examining the mucosae. Still, this raises the questions whether focused training could further improve the diagnostic performance of NF-NBI, at what point a diminishing return is met, and whether uptake of NF-NBI, be it via this classification schema or otherwise, could provide an avenue to detect mucosal disease that endoscopists may have otherwise missed (and consequently perhaps not even thought to biopsy). (4) The power calculation mentioned in the study states that 301 observations were needed to achieve the prespecified α level; inasmuch as duodenal images in a given patient are not truly independent of each other, 301 patients, not images, would seem to have been needed. That is, additional images per patient would likely not satisfy the sample size requirement. (5) In a related vein, the investigators state “This study evaluates the role of NF-NBI in assessing villous morphology in patients with suspected CD”; however, the sample consisted of 21 patients with known CD and only 11 additional patients referred with positive celiac serology (ie, the clinical suspicion of CD in the remaining 65 patients who completed the study seems uncertain and ostensibly low). Therefore, given this heterogeneity and the small subgroup sizes, it may be difficult to reliably draw inferences to a cohort of patients in whom the suspected diagnosis is veritably CD (eg, as opposed to patients who may have a symptom broadly compatible with CD). (6) The IOA for the diagnosis of subtotal VA using the proposed NF-NBI classification was lower than for normal villi or total VA, with a tendency to diagnose subtotal VA as total VA; this represents a clinically important pitfall. Similarly, sensitivity for villous blunting using the classification in the seronegative cohort (n = 55), albeit better than NF-WLE, was only 83.4%. Perhaps these represent scenarios in which a biopsy is needed rather than relying on an optical diagnosis.
      In summary, we commend Gulati et al
      • Gulati S.
      • Emmanuel A.
      • Ong M.
      • et al.
      Near-focus narrow-band imaging classification of villous atrophy in suspected celiac disease: development and international validation.
      for building on the findings of earlier studies and taking them to the next level by way of multicenter validation of a 3-descriptor NF-NBI classification schema that they have developed. From a practical perspective, it remains unknown which patients are those in whom NF-NBI alone (ie, supplanting biopsies) would assess for mucosal disease (or healing) with sufficient diagnostic performance. Future studies with larger sample sizes of various patient subsets (eg, seronegative for CD [with low as well as high clinical suspicion of CD], seropositive, and monitoring of known CD) will be needed to enable confident discernment of the appropriate clinical applications of NF-NBI. On the basis of the data thus far, we believe, given its high NPV, that NF-NBI can be used to rule out CD (and possibly other duodenopathies, eg, peptic) without requiring duodenal biopsies in patients with a low pretest clinical probability (seronegative); conversely, it can be used to target biopsies in those who have a high pretest probability or are in need of histologic surveillance of known CD or other duodenopathy. In the remainder of clinical scenarios, duodenal abnormalities may be too subtle or deep to enable a reliable diagnosis to be made optically, and passing up the chance to obtain duodenal biopsies while already in situ may prove to have too high an opportunity cost.

      Disclosure

      Dr Tabibian is a consultant for Olympus Corporation and Pfizer Inc. Dr Murray is the recipient of study grants from Nexpep/ImmusanT, National Institutes of Health, Immunogenix, Takeda Pharmaceutical, Allakos, Oberkotter, and Cour; the recipient of consultancy fees from Bionix, Lilly Research Laboratory, Johnson & Johnson, Dr. Schar USA, UCB Biopharma, Celimmune, Intrexon Corporation, Chugai Pharma, Kanyos, and Boehringer Ingelheim; the holder of patents licensed to Evelo Biosciences; and the recipient of royalties from Torax Medical.

      References

        • Jabbari M.
        • Wild G.
        • Goresky C.A.
        • et al.
        Scalloped valvulae conniventes: an endoscopic marker of celiac sprue.
        Gastroenterology. 1988; 95: 1518-1522
        • Barada K.
        • Habib R.H.
        • Malli A.
        • et al.
        Prediction of celiac disease at endoscopy.
        Endoscopy. 2014; 46: 110-119
        • Olympus
        Technology.
        (Available at:)
        • Tabibian J.H.
        • Perrault J.F.
        • Murray J.A.
        • et al.
        Narrow band imaging evaluation of duodenal villi in patients with and without celiac disease: a prospective study.
        World J Gastrointest Endosc. 2019; 11: 145-154
        • Singh R.
        • Nind G.
        • Tucker G.
        • et al.
        Narrow-band imaging in the evaluation of villous morphology: a feasibility study assessing a simplified classification and observer agreement.
        Endoscopy. 2010; 42: 889-894
        • De Luca L.
        • Ricciardiello L.
        • Rocchi M.B.
        • et al.
        Narrow band imaging with magnification endoscopy for celiac disease: results from a prospective, single-center study.
        Diagn Ther Endosc. 2013; 2013: 580526
        • Banerjee R.
        • Reddy D.N.
        High-resolution narrow-band imaging can identify patchy atrophy in celiac disease: targeted biopsy can increase diagnostic yield.
        Gastrointest Endosc. 2009; 69: 984-985
        • Cammarota G.
        • Cesaro P.
        • La Mura R.
        • et al.
        Role of the "immersion technique" in diagnosing celiac disease with villous atrophy limited to the duodenal bulb.
        J Clin Gastroenterol. 2007; 41: 571-575
        • Koay D.S.C.
        • Ghumman A.
        • Pu L.
        • et al.
        Narrow-band imaging with magnification and the water immersion technique: a case-finding, cost-effective approach to diagnose villous atrophy.
        Singapore Med J. 2019; 60: 522-525
        • Tchekmedyian A.J.
        • Coronel E.
        • Czul F.
        "Leopard skin sign": the use of narrow-band imaging with magnification endoscopy in celiac disease.
        Rev Gastroenterol Peru. 2014; 34: 321-324
        • Ianiro G.
        • Bibbo S.
        • Pecere S.
        • et al.
        Current technologies for the endoscopic assessment of duodenal villous pattern in celiac disease.
        Comput Biol Med. 2015; 65: 308-314
        • Gulati S.
        • Emmanuel A.
        • Ong M.
        • et al.
        Near-focus narrow-band imaging classification of villous atrophy in suspected celiac disease: development and international validation.
        Gastrointest Endosc. 2021; 94: 1071-1081

      Linked Article