Diagnostic Approach to Early Barrett's Neoplasia: Japanese Perspective.

Introduction
The incidence of esophageal adenocarcinoma (EAC), which arises from Barrett’s esophagus (BE), has increased substantially over recent decades. The histological distribution of esophageal cancer now varies markedly by region. In the USA, EAC accounts for approximately 66.5% of all cases, exceeding squamous cell carcinoma (SCC), according to the SEER Cancer Statistics Review [1]. Similarly, GLOBOCAN 2020 data analyzed by Morgan et al. show that EAC predominates in Northern Europe (65%), with lower proportions in Western (42%), Southern (28%), and Central-Eastern Europe (14%) [2]. SCC remains the leading subtype in most of Asia, Africa, and Eastern Europe. In Japan, although SCC still predominates, the proportion of EAC has risen from 3% to around 8% over the past decade [3, 4], likely reflecting lifestyle westernization and improved recognition through surveillance. In addition, the declining prevalence of Helicobacter pylori infection is considered a contributing factor, as it may lead to increased gastric acid secretion and a higher incidence of gastroesophageal reflux disease, Barrett’s esophagus, and subsequent neoplastic progression [5].
When confined to the early stage, EAC is suitable for curative endoscopic resection, achieving excellent long-term outcomes with minimal invasiveness [6, 7]. Accordingly, surveillance plays a central role in the early detection and management of superficial BE-related neoplasia (SBERN). This review summarizes contemporary strategies for endoscopic surveillance in BE, emphasizing the Japanese “target biopsy” approach, qualitative endoscopic assessment, and their integration with advanced imaging technologies.

Cancer Risk in Barrett’s Esophagus
The extent of BE, measured as the maximum (M) extent according to the Prague C & M classification [8], is a major determinant of progression to EAC. In a large multicenter Japanese cohort study, the annual incidence of EAC was 0.0032% for ultrashort-segment BE (USSBE; M <1 cm), 0.026% for short-segment BE (SSBE; M ≥1 cm to <3 cm), and 0.58% for long-segment BE (LSBE; M ≥3 cm), showing a marked length-dependent increase in cancer risk [5]. These findings are consistent with Western data; meta-analyses and longitudinal studies report annual EAC incidence rates of approximately 0.01% for USSBE, 0.03–0.07% for SSBE, and 0.22–0.31% for LSBE [9–11]. Moreover, long-term follow-up studies have shown cumulative cancer incidence of 6–10% over a decade in LSBE [12]. Both Eastern and Western data underscore BE length as a strong, clinically relevant predictor of EAC, supporting surveillance strategies tailored to segment extent of BE.

Endoscopic Detection and Classification of SBERNs
SBERNs are typically detected using white-light endoscopy (WLE), appearing as subtle reddish, rough-surfaced, flat, elevated, or depressed lesions [13]. Excessive insufflation can obscure these subtle changes, underscoring the importance of proper air control (Fig. 1). Previous studies from Japanese and Western groups have consistently reported that SBERNs predominantly occur on the anterior or right wall of the Barrett’s segment, including both short- and long-segment Barrett’s esophagus [14–16]. However, more recent data indicate that this tendency is mainly observed in SSBE and becomes less evident in LSBE. Yamasaki et al. [17] reported that about half of LSBE-related lesions arise on the left anterior wall, and Ikenoyama et al. [18] described LSBE-related neoplasia as having no characteristic locational preference.
When such findings are observed, narrow-band imaging with magnifying endoscopy (NBIME) is employed for further evaluation. Several NBIME-based classification systems have been proposed for the SBERNs, demonstrating diagnostic utility [19–22], but none have gained widespread international adoption, primarily due to their complexity and limited interobserver reproducibility. To address these issues, the Japan Esophageal Society Barrett’s Esophagus Working Group (JES-BE) developed a novel endoscopic classification system incorporating a diagnostic flowchart [23]. Mucosal and vascular patterns are assessed on NBIME, and dysplasia is defined as either an irregular mucosal pattern or an invisible mucosal pattern with irregular vascular pattern. Control of luminal insufflation is also essential during NBIME. Excessive distension flattens the mucosa and obscures pit structures, while deflation may render invisible mucosal patterns visible. When the mucosal pattern is invisible despite optimal insufflation, the vascular pattern should be assessed next. If a vascular network is present, its regularity is evaluated; an irregular vascular pattern usually indicates tubular adenocarcinoma. When no vascular network is observed, acetic acid enhancement provides a more reliable assessment of the surface pattern (Fig. 2). However, in rare cases, poorly differentiated components that usually coexist with well-differentiated glands may be present in the deeper layer. In such lesions, the superficial layer shows differentiated features, whereas the poorly differentiated component is confined to the deeper invasive front, indicating dedifferentiation during tumor progression [24, 25]. Most of these cases correspond to T1a-DMM or T1b carcinomas, and this possibility should be considered in lesions with substantial thickness or obvious deep invasion.
In a nationwide multicenter study, 156 NBIME images were evaluated using the JES-BE classification by both expert and non-expert reviewers. Using histopathology as the reference standard, sensitivity, specificity, and overall accuracy for dysplasia detection were 87%, 97%, and 91%, respectively. Interobserver agreement for classification was substantial (κ = 0.75 for experts, 0.79 for non-experts), showing good reproducibility even among non-experts. Notably, within the refined version of the JES-BE algorithm [26], the presence of a green thick vessel – a cyan-colored, dilated vessel in the deeper layer beneath thinned epithelium – is regarded as a reliable marker of non-neoplastic mucosa, helping to avoid false-positive diagnoses associated with invisible mucosal patterns. Although structurally simple, the JES-BE classification demonstrates high diagnostic accuracy and reproducibility, supporting its potential for widespread clinical implementation.

Acetic Acid-Enhanced Endoscopic Techniques in SBERNs
Acetic acid-enhanced endoscopy highlights focal and rapid loss of acetowhitening, aiding the detection of SBERNs. Targeted biopsy after acetic acid enhancement has been reported to increase diagnostic yield compared to random biopsy while reducing the number of biopsies required [27]. However, a large Western prospective study found that acetic acid-enhanced endoscopy, even when combined with NBI, did not significantly improve dysplasia detection compared to WLE in routine surveillance [28].
In contrast, acetic acid-enhanced magnifying endoscopy enables detailed three-dimensional visualization of glandular architecture in Barrett’s mucosa. Regular ridge, gyrus, or villous patterns correspond to specialized intestinal metaplasia [21, 29, 30], whereas irregular or disrupted structures are associated with high-grade dysplasia or early adenocarcinoma [31–33]. In lesions with an indistinct mucosal pattern on NBIME, acetic acid can reveal underlying glandular pits otherwise difficult to visualize. This technique has also been applied to detect subepithelial tumor spread beneath squamous epithelium (Fig. 3). Yamagata et al. [34] reported characteristic findings – termed the “small white sign,” including small pits with white margins, white spots, and sulciform structures – as useful for delineating tumor extent. Nevertheless, no large multicenter prospective studies have confirmed the additional value of acetic acid-enhanced magnifying endoscopy over WLE or NBIME. Its diagnostic superiority should be evaluated in well-designed randomized controlled trials.

Molecular Imaging Techniques in SBERNs: CLE and Endocytoscopy
Confocal laser endomicroscopy (CLE) and endocytoscopy allow real-time, ultra-high-magnification visualization of cellular morphology for in vivo histologic assessment. CLE is not available for clinical use in Japan, whereas endocytoscopy, although developed domestically, is rarely used in routine practice because of its technical complexity, narrow field of view, and long procedure time.
CLE provides high diagnostic accuracy for SBERNs. The endoscope-integrated platform demonstrates excellent performance, with 98.1% sensitivity, 94.1% specificity, and high interobserver agreement [35]. The probe-based system, compatible with a 2.8-mm working channel, also shows acceptable accuracy, with 75–89% sensitivity and 75–91% specificity [36, 37]. Endocytoscopy has potential diagnostic value, but in vivo studies have reported suboptimal image quality and low interobserver agreement, with neoplastic features identified in only 4–41% of images [38]. Recently, van der Laan et al. [39] applied artificial intelligence to endocytoscopic images, achieving 89.7% diagnostic accuracy with an area under the curve of 0.93. These findings indicate that, despite ongoing technical advances such as artificial intelligence, the clinical applicability of molecular imaging remains limited at present.

Assessment of Invasion Depth in SBERNs
In Japan, current guidelines do not define clear criteria for endoscopic treatment of SBERNs [40, 41]. Western guidelines recommend endoscopic resection for lesions confined to the lamina propria mucosae (T1a-LPM), while those invading the deep muscularis mucosae or beyond (≥T1a-DMM) generally require surgery [42, 43]. A Japanese multicenter study analyzed adenocarcinomas of the esophagogastric junction, which included both Barrett’s adenocarcinomas and gastric cardia cancers [44]. In that cohort, no lymph node metastasis was observed in T1a-LPM or in T1a-DMM and T1b-SM (≤500 µm) lesions without risk factors such as tumor size >3 cm, lymphovascular invasion, or poor differentiation, whereas metastasis was seen in 58.3% of T1a-DMM lesions with risk factors, 15.4% of ≤500 µm SM lesions with risk factors, and 10.5% of ≥501 µm SM lesions even without such features. Although these findings provide useful reference data for SBERNs, evidence focusing exclusively on Barrett’s adenocarcinoma remains scarce in Japan, and further dedicated studies are warranted. These results align with Western data suggesting that endoscopic resection may be appropriate for selected T1b cases with favorable features – well or moderately differentiated histology, shallow invasion depth, and no lymphovascular invasion – or in patients unfit for surgery [42, 45]. In clinical practice in Japan, lesions diagnosed endoscopically as clinical DMM/T1b-SM slight invasion are often treated by endoscopic resection. If pathological assessment confirms complete resection and T1b-SM invasion ≤500 µm without any risk factors, careful follow-up with periodic CT is considered an acceptable management option. However, SM invasion often exhibits poor differentiation and lymphovascular invasion – with reported rates of 26–29% and 18–23%, respectively – compared to 4–6% and 0–2% in intramucosal cancers [24, 46], making truly favorable T1b cases infrequent. Such findings help guide endoscopic treatment decisions in clinical practice.
These findings highlight the importance of accurate assessment of invasion depth. Unlike esophageal SCC, SBERN lacks a standardized endoscopic depth classification; as in early gastric cancer, evaluation mainly relies on WLE findings, and occasionally endoscopic ultrasonography. On WLE, sessile elevated form, rigidity upon deflation, or deep depression may indicate T1b-SM lesion. Based on the Japanese classification of esophageal cancer [47], most of type 0-I lesions invade the submucosal layer, although those with a constricted base are often limited to T1a-DMM or T1b-SM1 (≤500 µm) (Fig. 4). On the other hand, flat lesions are generally confined to the mucosa (SMM/LPM) and rarely invasive [14, 48].
NBIME data in SBERN are limited, but Yoshizawa et al. [49] reported large-caliber vessels associated with submucosal invasion. In clinical practice, loss of surface pattern and absent or dilated microvessels are often interpreted as signs of submucosal invasion, extrapolating criteria from early gastric and colorectal cancers. However, the diagnostic value of NBIME for assessing invasion depth has not yet been fully elucidated, and further studies are needed to establish its role in endoscopic diagnosis of invasion depth.
The diagnostic performance of EUS for invasion-depth assessment in SBERNs is also limited. A meta-analysis from the USA reported frequent over- and under-staging with an overall accuracy of about 75% [50]. A recent expert image-based study from Germany also showed modest predictive values (PPVs 40%) and moderate interobserver agreement (κ = 0.41), indicating that ancillary imaging including EUS is not particularly reliable for distinguishing T1b from T1a lesions [51]. In Japan, EUS is not routinely used for this reason but is applied selectively as an adjunct when WLE suggests possible submucosal invasion.

Diagnostic Challenges of LSBE Neoplasias
LSBE neoplasias are often flat with minimal or no surface irregularities and lack distinct endoscopic features, frequently arising within inflamed or regenerative mucosa where reduced contrast makes them difficult to distinguish from the background [52, 53]. Reflecting these challenges, studies from Europe and Japan have reported lower endoscopic R0 resection rates for LSBE neoplasias (48–70%) compared with SSBE neoplasias (85–91%) [6, 48]. This means diagnostic strategies effective in SSBE – such as gastric cancer-based criteria and the JES-BE classification – are often inadequate for LSBE, illustrating the limitations of current endoscopic diagnosis (Fig. 5).
These diagnostic limitations have prompted the development of distinct surveillance strategies in Western countries and Japan, as the higher prevalence of LSBE in Western populations favors systematic surveillance protocols, whereas the lower prevalence in Japan supports a targeted, endoscopy-based approach. Western guidelines advocate the Seattle protocol, combining four-quadrant random biopsies at 1–2 cm intervals with targeted sampling of visible lesions [54]. In contrast, Japanese endoscopists primarily rely on image-enhanced magnification techniques such as NBIME with or without acetic acid enhancement to identify mucosal and vascular abnormalities and perform targeted biopsies accordingly – an approach adapted from gastric cancer surveillance [55, 56]. If the prevalence of LSBE increases in the future, systematic biopsy protocols such as the Seattle protocol may gain partial relevance; however, given the widespread use of high-resolution magnifying endoscopy in Japan, a targeted, image-based approach is likely to remain the principal strategy.
Further complicating diagnosis, histological interpretation in LSBE lesions is hindered by substantial interobserver variability [57], largely due to overlapping features between inflammation-associated atypia and true dysplasia, and the inherent subjectivity of histologic criteria – even among expert gastrointestinal pathologists [58]. As a result, biopsy-based diagnosis alone may be insufficient for determining the presence or extent of neoplasia. Immunohistochemical markers such as p53, Ki-67, cyclin D1, and MCM2 have been shown to correlate with dysplasia grade and invasive carcinoma, thereby aiding distinction from reactive atypia and improving diagnostic reliability [59–63]. In Japan, p53 and Ki-67 immunostaining are particularly incorporated into routine practice for lesions indefinite for dysplasia.

Therapeutic Approaches for LSBE Neoplasias with Diagnostic Limitations
Because of the diagnostic complexity of LSBE-related neoplasias, therapeutic strategies differ between Western countries and Japan (Fig. 6). Western guidelines (ACG, ESGE) recommend focal endoscopic resection of visible lesions followed by radiofrequency ablation (RFA) of the residual Barrett’s mucosa to eradicate both the neoplasia and the at-risk epithelium [51, 64]. This combined approach carries a 6–12% stricture rate and a risk of recurrent intestinal metaplasia or dysplasia [65–68].
In contrast, Japanese practice favors en bloc ESD to achieve complete histological resection, despite its technical difficulty and risk of post-ESD stricture [50]. When the lesion is multifocal or has indistinct margins, circumferential or stepwise ESD is performed, usually with steroid-based therapy to prevent post-ESD stricture. A stepwise ESD technique, consisting of staged resections for circumferential LSBE, has recently been introduced [69]; although it appears effective in reducing strictures, it entails piecemeal resection and lacks long-term oncological validation. When the lesion margins are clearly demarcated, local en bloc ESD is generally performed, followed by endoscopic surveillance of the residual Barrett’s mucosa. As RFA is not available for clinical use in Japan, Takahashi and Oyama [70] reported that stepwise ESD may also be applied to completely remove the residual Barrett segment due to the risk of metachronous neoplasia. Although long-term evidence remains limited, this approach may serve as a practical alternative for managing residual Barrett’s mucosa in Japan.

Conclusion
Despite advancements in imaging techniques and classification systems, early detection and histologic assessment of SBERNs remain challenging, particularly in LSBE. Further refinement of endoscopic strategies and diagnostic criteria is essential to improve accuracy and optimize surveillance.

Conflict of Interest Statement
The authors have no conflicts of interest to declare.

Funding Sources
This study was not supported by any sponsor or funder.

Author Contributions
Kotaro Shibagaki contributed equally to the conception and design of the review article and performed the literature search and data collection. Yusuke Takahashi, Satoshi Kotani, Shinsuke Suemitsu, Shigeru Kawabata, and Daisuke Niino contributed to the selection of cases whose images were used in the figures and to the preparation of the figures. Norihisa Ishimura critically revised the manuscript for important intellectual content. Shunji Ishihara supervised the project. All authors reviewed and approved the final version of the manuscript.