Abstract
Background:
Superficial non-ampullary duodenal epithelial tumors (SNADETs) are increasingly detected with advances in endoscopic imaging. However, the preoperative diagnosis remains challenging. Indigo carmine-assisted texture and color enhancement imaging (IC-TXI) is a novel modality that enhances mucosal contrast and topography.
Objectives:
To evaluate the diagnostic performance of IC-TXI for identifying Vienna classification (VCL) C4/5 SNADETs based on surface morphology.
Design:
Single-center retrospective study.
Methods:
This study analyzed 111 consecutive SNADETs resected between March 2021 and August 2024. IC-TXI images were assessed for abnormal surface structure, component size difference (>3-fold), and invisible surface pattern. Significant findings were defined as the coexistence of abnormal surface structure and component size or invisibility. Diagnostic performance was evaluated using histopathological VCL classification as the reference.
Results:
Ten (9.0%) lesions were classified as VCL C4/5. IC-TXI demonstrated a sensitivity of 50.0%, specificity of 97.0%, positive predictive value of 62.5%, negative predictive value of 95.1%, and accuracy of 92.8% for identifying VCL C4/5 lesions. Abnormal surface structure alone was observed in 5/10 (50.0%) VCL C4/5 lesions, abnormal size difference alone in 6/29 (20.7%), and invisible surface structure in 1/3 (33.3%). These findings suggest that abnormal surface structure may be diagnostically relevant than component size variation, and invisibility may have malignant potential.
Conclusion:
IC-TXI may serve as a practical method for the real-time exclusion of VCL C4/5 SNADETs. Given its high specificity and negative predictive value (NPV), this method may have the potential to reduce unnecessary biopsies and shorten observation time. However, rather than being considered a standard modality, IC-TXI should be regarded as an adjunctive tool to support clinical decision-making in combination with other clinical and endoscopic factors, particularly in light of its relatively low sensitivity (50.0%) observed in this study.
Clinical trial registration:
UMIN000041436.
Introduction
Superficial non-ampullary duodenal epithelial tumors (SNADETs) are increasingly being detected owing to advances in endoscopic technology.1,2 Although most SNADETs are adenomas, they carry a well-recognized malignant potential, and early identification of high-grade dysplasia or carcinoma is crucial for determining the appropriate therapeutic strategy.3,4 Recent clinical guidelines have underscored the importance of accurate endoscopic diagnosis to appropriately stratify lesions for endoscopic versus surgical resection. 5 Although biopsy is frequently performed, its diagnostic accuracy for SNADETs remains limited, with reported rates of only 68%–71.6%.6,7
In recent years, image-enhanced endoscopy (IEE) has emerged as a valuable tool in the evaluation of SNADETs. In particular, magnifying endoscopy combined with narrow-band imaging (ME-NBI) has demonstrated good diagnostic capability.8 –10 Reported accuracy rates for lesions classified as Vienna Classification (VCL) C3/C4–511,12 have been relatively favorable, suggesting that endoscopic assessment is at least comparable to biopsy. Nevertheless, no consensus has yet been reached regarding the optimal diagnostic strategy.
In 2020, Olympus Medical Systems Corporation (Tokyo, Japan) introduced a novel IEE technology known as texture and color enhancement imaging (TXI). TXI processes white-light images by separating them into texture and base components, which are then recombined after texture enhancement and adjustments to color tone and brightness. 13 Indigo carmine-assisted TXI magnified imaging (IC-TXI) is an innovative technique that integrates chromoendoscopy with TXI, providing enhanced topographic visualization of mucosal patterns in SNADETs. 14 Building on this, the present study aimed to evaluate the diagnostic performance of IC-TXI in the assessment of SNADETs.
Materials and methods
Study design
This was a single-center study conducted at Chiba University Hospital, designed as a retrospective analysis of prospectively collected cases. Consecutive patients with SNADETs who underwent endoscopic resection (ER) following preoperative observation with IC-TXI between March 2021 and August 2024 were included. Written informed consent for endoscopic observation was obtained from all patients prior to enrollment. The Institutional Review Board of Chiba University School of Medicine (approval numbers: 3740, 3951) and registered in the University Hospital Medical Information Network (UMIN000041436) and was conducted in accordance with the Declaration of Helsinki. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement (Supplemental Material). 15
Inclusion and exclusion criteria
All consecutive patients with endoscopically diagnosed SNADETs who were scheduled for ER and provided informed consent were considered eligible. The inclusion criteria were: (1) lesions located in the non-ampullary duodenum; (2) histologically confirmed neoplasia (adenoma or adenocarcinoma) after resection; and (3) observation using IC-TXI performed after obtaining informed consent. Patients diagnosed as non-neoplastic on histopathological examination after resection were excluded.
Endoscopic procedures and image acquisition
The mechanism of TXI has been described previously. 13 In summary, TXI separates the WLI image into base and texture components, enhances texture and brightness adjustment to create mode 2, and applies color tone enhancement to generate mode 1. Using a high-definition endoscope (GIF-XZ1200, Olympus Medical Systems, Tokyo, Japan) equipped with the EVIS X1 system and TXI function, we performed standard white-light and TXI (mode 1, structure enhancement set to “strong”) observation. Subsequently, 0.2% indigo carmine was sprayed to enhance the mucosal structure of SNADETs, followed by IC-TXI observation (Figure 1). Still images were obtained for subsequent review. Resection was performed by cold snare polypectomy (CSP), underwater endoscopic mucosal resection (UEMR), or endoscopic submucosal dissection (ESD), as appropriate. ESD was performed exclusively by expert endoscopists (defined as those with more than 10 years of endoscopic experience). In contrast, both CSP and UEMR were performed by both expert and non-expert endoscopists. For ESD and UEMR, post-procedural closure was routinely performed using clips. In contrast, post-procedural closure was generally not performed after CSP. In general, patients with unclear or positive horizontal and/or vertical margins underwent follow-up endoscopy 3 months after resection, followed by annual surveillance. In contrast, patients with negative horizontal and/or vertical margins underwent follow-up endoscopy 1 year after resection. Follow-up was conducted at our institution or at the referring institutions.
The flow of observation. TXI separates the WLI image into base and texture components, enhances texture and brightness adjustment to create mode 2, and applies color tone enhancement to generate mode 1. Subsequently, 0.2% indigo carmine was sprayed to enhance mucosal structure of SNADETs, followed by IC-TXI observation.
Definition of abnormality significant findings
The typical cases of WLI were shown in Figure 2(a). Surface structures were categorized as round, oval, others, and invisible (Figure 2(b)). A mixture of components was defined as the coexistence of different morphological types (round, oval, or others) and was regarded as an abnormal finding. On the other hand, a difference in component size exceeding threefold was defined as abnormal (Figure 2(c)). Significant findings were defined as the coexistence of abnormal surface structure and component size or invisible structure (Figure 3).
Classification of surface structure and a difference in component size. (a.a) 8 mm elevated lesion located at bulb. (a.b) 10 mm elevated lesion located in the 2nd portion. (a.c) 20 mm elevated lesion located in the 2nd portion. (a.d) 7 mm elevated lesion located in the 2nd portion. (b) Classification of surface structure (The yellow circles each represent a component). (b.a) Round. (b.b) Oval. (b.c) Others. (b.d) Invisible. (c) A difference in component size. Each yellow bar represents a single component of the lesion. (The surface structure was classified as “others.”) The difference in component size exceeded threefold, which was regarded as an abnormal difference.
Definition of significant findings.
Diagnostic performance of IC-TXI
The diagnostic performance of significant findings detected by IC-TXI and VCL C3/C4–5 was evaluated by calculating sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy.
Histopathological evaluation
All resected specimens were fixed in 10% neutral-buffered formalin and sectioned for histopathological examination. To ensure diagnostic consistency and minimize inter-observer variability, all lesions were reviewed and confirmed by a senior gastrointestinal pathologist (K.M.) who was blinded to the endoscopic findings. Each lesion was independently reviewed by experienced gastrointestinal pathologists who were blinded to the endoscopic findings. Histological grading was determined according to the Vienna Classification (VCL): category 3 (low-grade adenoma), category 4 (high-grade adenoma/dysplasia), and category 5 (carcinoma). For further subclassification, category 4 and 5 lesions were defined in detail as follows: high-grade adenoma (C4.1), non-invasive carcinoma (C4.2), suspicious for invasive carcinoma (C4.3), intramucosal carcinoma (C5.1) and submucosal carcinoma (C5.2). In cases where the diagnosis was challenging, a final diagnosis was reached through consensus discussion between two senior pathologists (K.M. and J.I.). In addition, mucin phenotype (gastric, intestinal, and mixed) was determined using immunohistochemistry for MUC2, MUC5AC, MUC6, and CD10. Tumors positive for MUC5AC or MUC6 were classified as the gastric type, those positive for MUC2 and/or CD10 as the intestinal type, and those expressing both as the mixed type. Pathological diagnosis was performed according to VCL classifying into VCL C3 and C4/5.
Results
Baseline characteristics
A total of 111 SNADETs were analyzed, with a median size of 10 mm (range, 3–50 mm). Most lesions were located in the second portion of the duodenum (82.0%) and presented as elevated morphology (85.6%). ER was performed primarily by UEMR (55.0%) or CSP (44.1%). The majority of lesions were classified as VCL C3 (91.0%), with VCL C4/5 lesions accounting for 9.0%. Regarding mucin phenotype, intestinal type tumors predominated (82.9%), followed by mixed (9.0%) and gastric types (8.1%; Table 1). Of those 111 lesions, 74 lesions were taken prior to biopsy. The proportion of non-tumor, VCL C3, C4/5 in prior biopsy and resected specimen was 4 (5.4%), 56 (75.7%), 14 (18.9%). On the other hand, among the 56 cases diagnosed as VCL C3 on biopsy, 3 were reclassified as VCL C4/5. In addition, among the 14 cases diagnosed as VCL C4/5 on biopsy, 7 were reclassified as VCL C3. Abnormal surface structure alone was observed in 5/10 (50.0%) VCL C4/5 lesions, abnormal size difference alone in 6/29 (20.7%), and invisible surface structure in 1/3 (33.3%). As for adverse events (AEs), there were 2 delayed bleeding out of 61 UMER cases (3.2%). No other AEs were observed.
Baseline characteristics of SNADETs.
CSP, cold snare polypectomy; ESD, endoscopic submucosal dissection; SNADETs, superficial non-ampullary duodenal epithelial tumors; UEMR, underwater endoscopic mucosal resection; VCL, Vienna classification.
Diagnostic performance for high-grade lesions
When VCL C4/5 lesions were used as the reference standard, IC-TXI demonstrated a sensitivity of 50.0%, specificity 97.0%, PPV 62.5%, NPV 95.1%, and overall accuracy 92.8% (Table 2).
Diagnostic performance of IC-TXI for VCL 4/5.
Defined as the coexistence of abnormal surface structure and component size or invisible structure.
IC-TXI, indigo carmine-assisted texture and color enhanced magnified imaging; NPV, negative predictive value; PPV, positive predictive value; VCL, Vienna classification.
Correlation between surface pattern and mucin phenotype
Surface pattern distribution differed according to mucin phenotype (Table 3). Gastric-type lesions most frequently exhibited a round pattern (55.6%), whereas intestinal-type lesions were predominantly classified as others (90.2%). From the opposite perspective, half of all round pattern lesions were gastric type and 30% were mixed type, while oval and invisible patterns occurred exclusively in intestinal-type tumors.
Correlation between surface structure and mucin phenotype.
Case presentation of VCL C4/5 and C3
Representative cases of VCL C4/5 and C3 are shown in Figures 4 and 5. In the VCL C4/5 case (Figure 4(a) and (b)), IC-TXI revealed a 10-mm elevated lesion located in the bulb, displaying abnormal surface structure and abnormal size difference. Histologically (Figure 4(c)), the lesion consisted of atypical glandular epithelium with proliferative activity extending throughout the glands, consistent with a high-grade adenomatous lesion. Immunohistochemically (Figure 4(d)), the tumor cells were strongly positive for MUC6, indicating a gastric phenotype with differentiation toward pyloric or Brunner’s glands. Although not shown in the figure, the tumor cells were negative for MUC5AC, CD10, and MUC2. The p53 staining demonstrated a wild-type pattern, characterized by scattered weak to moderate nuclear positivity. In the VCL C3 case (Figure 5(a) and (b)), IC-TXI depicted a 7-mm elevated lesion in the 2nd portion. A surface structure classified as 'others' was observed without apparent size differences. Histologically (Figure 5(c)), the lesion was composed of tall columnar atypical glandular epithelium with elongated, spindle-shaped nuclei forming densely packed and irregular glandular structures. The proliferative zone was mainly distributed near the superficial mucosa, while toward the surface the cells exhibited differentiation and maturation with nuclear shrinkage, resulting in a flat mucosal surface. The tumor cells had eosinophilic cytoplasm consistent with intestinal absorptive epithelium, without evidence of a gastric phenotype.
Case presentation of VCL C4/5. (a) Image of IC-TXI (10-mm elevated lesion located in the bulb). (b) Both abnormal surface structures (round: yellow circle, others: green circle) and abnormal size differences were observed. (c) Histopathological findings. The lesion consists of tall columnar atypical glandular epithelium showing enlarged, rounded nuclei with increased chromatin content and a greater short-axis diameter. These atypical cells proliferate while forming intricately branched and dilated glandular structures. The proliferative zone extends throughout the entire length of the atypical glands, indicating enhanced proliferative activity. The nuclear-to-cytoplasmic ratio exceeds 50%. The proliferative zone is exposed at the surface of the lesion, which exhibits papillary undulations. Scale bar = 100 μm. (d) Immunohistochemistry. The tumor cells show strong cytoplasmic positivity for MUC6, indicating a gastric phenotype with differentiation toward pyloric glands or Brunner’s glands. Scale bar = 100 μm.
Case presentation of C3. (a) Image of IC-TXI (7-mm elevated lesion located in the 2nd portion). (b) A surface structure classified as “others” was observed (yellow circle), whereas no abnormal size difference was observed. (c) Histopathological findings. Tall columnar atypical glandular epithelium with elongated, spindle-shaped nuclei proliferates while forming densely packed and irregular glandular structures. The proliferative zone is mainly distributed near the superficial portion of the mucosa; however, toward the very surface, the cells show differentiation and maturation accompanied by nuclear shrinkage, forming a flat mucosal surface. The nuclear-to-cytoplasmic ratio is less than 50%. Scale bar = 100 μm.
Discussion
This study represents the first report evaluating the diagnostic performance of IC-TXI for SNADETs. Tsuji et al. demonstrated that prior biopsy may lead to overdiagnosis of C3 lesions as C4/5. 16 Likewise, our study also supports discrepancies between biopsy and resected specimens, suggesting that accurate endoscopic diagnosis of SNADETs is crucial. IC-TXI demonstrated its ability to be a supportive tool for exclusion. For example, IC-TXI may be helpful in selecting appropriate lesions for CSP (as its target is mainly VCL C3 lesions). On the other hand, given the sensitivity of 50.0%, when SNADETs are suspected to have malignant potential based on IC-TXI findings, the diagnosis should be made in conjunction with other clinical factors such as lesion size. Mixture of surface component and surface pattern irregularity has been discussed in previous reports,17 –19 and our evaluation items share conceptual similarities with those studies; however, combining heterogeneity and size irregularity (>3-fold variation) as a composite definition of “significant findings” represents a novel approach. To ensure adequate detection of VCL C4/5 lesions, we defined a size difference exceeding threefold as abnormal, which was considered a sufficiently large margin to ensure reliability.
In daily clinical practice, distinguishing neoplastic from non-neoplastic duodenal lesions can be challenging, and non-neoplastic lesions are not uncommon among SNADET-like lesions. A relatively simple diagnostic approach for non-neoplastic lesions, emphasizing enlargement of the marginal epithelium, has already been reported. 20 The present study did not aim to further evaluate or redefine the diagnostic features of non-neoplastic lesions. Instead, we focused specifically on histopathologically confirmed neoplastic lesions.
Previous studies using ME-NBI reported diagnostic accuracies ranging from 65% to 87% for differentiating VCL C3 from C4/5 lesions.10 –12 In our study, IC-TXI achieved comparable overall accuracy (92.8%) with the advantage of simplicity, as it relies solely on surface morphology rather than vascular pattern assessment. As described in our previous report, TXI enhances luminance and color contrast by algorithmically amplifying micro-textural information. The addition of indigo carmine further augments the contrast between depressed and elevated mucosal areas, thereby highlighting subtle irregularities that correspond to histological atypia. In fact, IC-TXI demonstrated higher visibility of surface patterns compared with magnified TXI alone and ME-NBI. 14 Focusing exclusively on surface structures without evaluating vascular patterns makes IC-TXI easier and faster to apply in clinical settings, as the findings are more easily recognizable.
Regarding the surface structure, the round pattern probably corresponds to the “closed-loop” morphology or oval-shaped marginal epithelium described in previous studies10,21 and appears to represent a gastric type phenotype. At the same time, in our analysis, we considered the oval surface structure to represent a distinct entity from the round pattern based on our predefined criteria. In fact, approximately half of the lesions exhibiting the gastric type showed a round pattern. In contrast, oval or other patterns are more frequently associated with intestinal phenotypes. Although crystal violet staining has been reported as useful for delineating mucosal architecture, 17 its application requires caution due to its potential carcinogenicity 22 and time-consuming preparation. In this regard, IC-TXI offers a practical advantage as a safer and simpler alternative for mucosal contrast enhancement. Among the lesions classified as VCL C4/5, 5 of 10 (50.0%) showed only an abnormal surface structure, 6 of 29 (20.7%) showed only an abnormal size difference, and 1 of 3 (33.3%) exhibited an invisible surface structure. These findings suggest that, in SNADETs, an abnormal surface pattern may be more diagnostically important than an abnormal size difference. Furthermore, the presence of an invisible surface pattern may be associated with a higher malignant potential and could warrant careful evaluation. On the other hand, due to the limited number of SNADETs, including cases classified as “others” as well as those with round, oval, and invisible surface structures, definitive differentiation between these categories requires further investigation with a larger cohort.
This study has several limitations. First, it was conducted at a single tertiary center with a relatively small number of VCL C4/5 lesions. Because the proportion of VCL C4/5 lesions in this study was relatively low (9.0%), the NPV value may have been overestimated. Therefore, the NPV should be interpreted with caution. Second, histopathological correlation was retrospective, and the diagnostic criteria for “significant findings” were based on qualitative assessment rather than quantitative image analysis. Ideally, a more simplified classification may facilitate a more objective classification. Furthermore, although the histopathological diagnoses were reviewed by a senior gastrointestinal pathologist to ensure consistency, the potential for inherent inter-observer variability in SNADETs diagnosis remains a limitation of any retrospective study. Third, we analyzed only still images. Endoscopic appearance may vary depending on the viewing angle, and assessment based solely on still images may limit reproducibility. Fourth, the exclusion of non-neoplastic lesions may lead to selection bias and may not fully reflect real-world diagnostic difficulty.
In conclusion, IC-TXI may serve as a practical method for the real-time exclusion of VCL C4/5 SNADETs. Given its high specificity and NPV, this method may have the potential to reduce unnecessary biopsies and shorten observation time. However, rather than being considered a standard modality, IC-TXI should be regarded as an adjunctive tool to support clinical decision-making in combination with other clinical and endoscopic factors, particularly in light of its relatively low sensitivity (50.0%) observed in this study. In the future, quantitative analysis of IC-TXI images may help refine diagnostic criteria. Building upon this, future studies could integrate IC-TXI with artificial intelligence-based image analysis to develop objective and reproducible diagnostic algorithms for SNADETs, ultimately advancing real-time endoscopic diagnosis.
Supplemental Material
sj-docx-1-tag-10.1177_17562848261447809 – Supplemental material for Diagnostic performance of indigo carmine-assisted texture and color enhanced magnified imaging for superficial non-ampullary duodenal epithelial tumors
Supplemental material, sj-docx-1-tag-10.1177_17562848261447809 for Diagnostic performance of indigo carmine-assisted texture and color enhanced magnified imaging for superficial non-ampullary duodenal epithelial tumors by Kenichiro Okimoto, Keisuke Matsusaka, Tomoaki Matsumura, Tsubasa Ishikawa, Takuya Ohashi, Chihiro Goto, Ryosuke Horio, Akane Kurosugi, Michiko Sonoda, Hirotaka Oura, Tatsuya Kaneko, Yuki Ohta, Takashi Taida, Jun-ichiro Ikeda and Jun Kato in Therapeutic Advances in Gastroenterology
Footnotes
Acknowledgements
None.
Declarations
Availability of data and materials
The data and materials that support the findings of this study are available from the corresponding author upon reasonable request. Permission to reproduce material from other sources: The material was created by the authors and does not require permission.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
