Use of magnifying endoscopy with narrow‐band imaging can change the clinical practice of screening endoscopy for early upper gastrointestinal neoplasia

Endoscopic diagnosis of early upper gastrointestinal (GI) neoplasia comprises two steps: detection and characterization.1 Magnifying endoscopy with narrow-band imaging (M-NBI) is more useful than conventional white-light imaging (C-WLI) for detecting and/or characterizing early GI neoplasia in the oropharynx, esophagus, stomach, and duodenum.2-5 The key findings reported and the methods for detection and characterization differ depending on the organ. In the oropharynx and esophagus, a well-demarcated brownish area on nonmagnifying endoscopy with NBI is a marker for detecting early squamous cell carcinoma, and an irregular microvascular pattern is a marker of characterization on M-NBI.2 Since the usefulness of nonmagnifying endoscopy with NBI remains unclear, C-WLI is used for detecting early gastric cancer (EGC).3 Although the endoscopic diagnostic criteria for EGC using C-WLI has not been established, we proposed criteria for characterization termed color plus surface (CS) classification system.4 Based on this system, the criteria for EGC using C-WLI are as follows: (i) the presence of irregularity in color within a well-demarcated area; and/or (ii) the presence of irregularity in the surface within a well-demarcated area.1, 4 When these criteria were applied to screening endoscopy for high-risk patients, the sensitivity and specificity for diagnosing EGC were 80.0% and 88.0%, respectively.3 When the vessel plus surface (VS) classification system was applied to M-NBI for diagnosing EGC after detecting suspicious lesions, the sensitivity and specificity increased up to 99.4% and 100%, respectively, with high confidence prediction if a proper algorithm considering the limitations of M-NBI is used.6 No specific reports have described the diagnostic performance of C-WLI in detecting superficial nonampullary duodenal epithelial tumors (SNADETs). However, M-NBI can be a powerful tool for the characterization of SNADET, with a sensitivity and specificity of 88.4% and 98.4%, respectively, for differentiating between neoplasia and nonneoplasia according to an algorithm.5 According to the VS classification system for nonbiopsied lesions, the sensitivity and specificity of M-NBI were 95% and 70%, respectively, for differentiating SNADET from low-grade adenoma and high-grade adenoma/cancer.7 In contrast, the specificity was reported to be remarkably low at 14% if M-NBI was performed after SNADET was biopsied because of overdiagnosis since the lesion was distorted owing to biopsy scarring.7 Accordingly, M-NBI can be a powerful tool for the detection and/or characterization of early upper GI neoplasia if a proper algorithm is used because there are limitations and pitfalls in each organ.6, 7 In this issue of Digestive Endoscopy, Takinami et al. retrospectively investigated the diagnostic performance of magnifying endoscopy (ME) for upper GI screening in a population with low prevalence of upper GI cancers.8 They determined the epithelial neoplasm detection rate, biopsy rate, and positive predictive value (PPV) between ME and non-ME groups using propensity score-matched analysis. The following results were obtained. There were no significant differences in the epithelial neoplasm detection rate (0.8% vs. 0.3%; P = 0.14) between the ME and non-ME groups. The biopsy rate was significantly lower in the ME group than in the non-ME group (12% vs. 15%; P = 0.003). PPV for biopsy was significantly higher in the ME group than in the non-ME group (6.6% vs. 2.8%; P = 0.048). Takinami et al. concluded that ME could reduce unnecessary biopsies in screening for upper early GI neoplasia in a healthy, asymptomatic population by improving the PPV for biopsy without compromising the epithelial neoplasm detection rate.8 Although similar findings were described in previous studies,6, 9 this was the first study to compare the diagnostic performance of ME with that of non-ME in screening with upper GI endoscopy. A limitation of this study was that it was a retrospective study. However, their conclusion coincides with what is followed in our daily clinical practice. Except for the systematic screening protocol used for upper GI endoscopy, Takinami et al. did not specify the standardized endoscopy technique that was employed, such as whether a black soft hood attachment for magnifying observation was used, which can provide endoscopists with images of the highest resolution to obtain precise endoscopic diagnoses. They also did not report on the endoscopic criteria that were employed in the screening for endoscopic detection and characterization of early upper GI neoplasia. If ME was to be performed using the best standardized technique for screening while following the criteria that have been proven to produce a consistently high diagnostic performance for detection and characterization, the number of unnecessary biopsies are predicted to decrease with an increase in the PPV. In this study, the PPV in the ME group was higher than that in the non-ME group. However, a PPV of 6.6% is not excellent. In fact, when the diagnostic performance of C-WLI and M-NBI for endoscopic diagnosis of EGC was tested using standardized techniques and criteria in the multicenter prospective trials, their PPV of C-WLI and M-NBI was 13.5%3 and 50.0% (recalculated data from the reference),6 respectively. Furthermore, the limitations and pitfalls of ME should be considered. One of the limitations of M-NBI for the diagnosis of EGC is a flat pale mucosal lesion (histologically characterized as signet-ring cell carcinoma)6 and a subepithelial neoplastic lesion (adenocarcinoma of fundic gland type).10 Another pitfall is biopsied SNADET whose appearance is distorted owing to a biopsy scar, resulting in overdiagnosis using M-NBI. Based on this perspective, well-designed prospective clinical trials are pivotal. After completing the trials, M-NBI could be optical biopsy that is quick, safe, and accurate for the endoscopic diagnosis of early upper GI neoplasia. I thank Dr Hisashi Doyama (Department of Gastroenterology, Ishikawa Central Hospital, Kanazawa, Japan) for proofreading. Author K.Y. has received consulting fees from Olympus Co. None.