Article PDF first page preview

Article PDF first page preview

Introduction: Superficial non-ampullary duodenal epithelial tumors (SNADETs) include low-grade adenoma (LGA) and high-grade adenoma or carcinoma (HGA/Ca) and are classified into two different epithelial subtypes, gastric-type (G-type) and intestinal-type (I-type). We attempted to distinguish them by endoscopic characteristics including magnifying endoscopy with narrow-band imaging (M-NBI). Methods: Various endoscopic and M-NBI findings of 286 SNADETs were retrospectively reviewed and compared between G- and I-types and histological grades. M-NBI findings were divided into four patterns based on the following vascular patterns; absent, network, intrastructural vascular (ISV), and unclassified. Lesions displaying a single pattern were classified as mono-pattern and those displaying multiple patterns as mixed-pattern. Lesions showing CDX2 positivity were categorized as I-types and those showing MUC5AC or MUC6 positivity were categorized as G-types based on immunohistochemistry. Results: Among 286 lesions, 23 (8%) were G-type and 243 (85%) were I-type. More G-type lesions were located oral to papilla (91.3 vs. 45.6%, p < 0.001), and had protruding morphology compared to those of I-types (65.2 vs. 14.4%, p < 0.001). The major M-NBI pattern was ISV in G-type (78.2 vs. 26.3%, p < 0.001), and absent for I-type (0 vs. 34.5%, p = 0.003). Three endoscopic characteristics; location oral to papilla, protruding morphology, and major M-NBI pattern (ISV) were independent predictors for G-type. Mixed-pattern was more common in HGA/Ca than LGA for I-type (77.0 vs. 58.8%, p = 0.01); however, there was no difference for those in G-type. Conclusion: Endoscopic findings including M-NBI are useful to differentiate epithelial subtypes.

Superficial non-ampullary duodenal epithelial tumors (SNADETs) are classified into low-grade adenoma (LGA), high-grade adenoma (HGA), and adenocarcinoma (Ca). Although SNADETs are rare tumors, the incidence has been gradually increasing due to widespread esophagogastroduodenoscopy screening, recent advances in endoscopic imaging, and aging population [1, 2]. Patients with submucosal or deeper invasion require invasive treatment such as pancreatoduodenectomy and the prognoses depend on the depth of tumor invasion [3, 4]. Therefore, early detection and diagnosis of SNADETs are essential. Recently, several endoscopic diagnostic algorithms have been reported in the diagnosis of SNADETs [5‒9].

The epithelium of SNADETs are composed of two subtypes, gastric-type (G-type) and intestinal-type (I-type), based on morphologic and immunophenotypic features [10]. Previous studies suggested that the localization and malignant potential are different between G-type and I-type [10‒13]. Patients with G-type advanced Ca had a poorer prognosis than patients with I-type Ca [10], and a higher rate of submucosal invasion was observed among G-type SNADETs [12, 13]. However, the incidence of G-type among SNADETs is rather low; thus, endoscopic characteristics are not fully understood.

Magnifying endoscopy with narrow-band imaging (M-NBI) is widely used for the diagnosis of superficial gastrointestinal tumors including SNADETs. Ishii et al. [6] reported a high accuracy in the differential diagnosis between LGA and HGA/Ca using a scoring system consisted of tumor diameter, color, and M-NBI findings. Another study reported that mixed-type of M-NBI patterns observed within a lesion could predict HGA/Ca of SNADETs [7]. However, these studies did not distinguish I- and G-type lesions. Considering that the majority of SNADETs are I-type lesions, it is possible that the differential diagnosis using M-NBI between LGA and HGA/Ca reported in those studies may not be applicable for G-type lesions. So far, there are few studies that focus on endoscopic diagnosis of epithelial subtypes of SNADETs. If we can endoscopically predict the epithelial subtypes of SNADETs, it would be useful in predicting the possibility of malignancy to select appropriate treatment strategies. We therefore conducted an analysis of SNADETs using endoscopic characteristics including M-NBI and investigated the correlation with epithelial subtypes.

Patients

A total of 293 consecutive patients with superficial 309 SNADETs which were diagnosed and treated by surgery or endoscopic resection at Nagoya University Hospital between August 2007 and May 2021 were reviewed. Subsequently, 23 lesions were excluded based on the following criteria: (i) final histology of non-neoplastic tissue (9 lesions); (ii) final histology of neuroendocrine tumor (1 lesion); (iii) familial adenomatous polyposis (7 lesions); and (iv) no images of M-NBI (6 lesions). Finally, 286 lesions were included in this study. Clinical data such as patient age, sex, location were obtained from the institutional database. This study complied with the Declaration of Helsinki and was approved by the ethics committee of Nagoya University Hospital (No. 2021-152).

Endoscopic Evaluation

Preoperative endoscopic observation was done using GIF-H260Z or 290Z gastrointestinal videoscopes (Olympus, Tokyo, Japan). The structure enhancement function was set at B8 for M-NBI, with the NBI color mode fixed at level 1. Four board-certified endoscopists of the Japan Gastroenterological Endoscopy Society (Y.K., H.T., K.H., and N.K.) retrospectively examined all endoscopic images without referring to patient information and pathological diagnosis. Macroscopic type, tumor diameter, color (red, isochromatic or white), and M-NBI findings were noted.

Macroscopic types were assessed according to the Paris classification [14] and were classified into protruding types (0-I) and non-protruding/nonexcavated types (0-IIa, 0-IIb, 0-IIc) based on the major type of morphology. Findings of M-NBI were evaluated by reference to the Kikuchi classification [7]. Vascular patterns were divided into four patterns: (i) absent; no microvessels visible due to the white opaque substance (WOS), (ii) network; vasculature of regular networks, (iii) intrastructural vessels (ISV); dilated or tortuous vessels within the mucosal structure, and (iv) unclassified; patterns other than the three patterns above (Fig. 1). Unclassified pattern included obscure or disappeared structure with vessels of irregular caliber and arrangements. Lesions were classified as mono-pattern or mixed-pattern according to the number of vascular patterns observed within the lesion. When discrepancies in individual interpretation were present, the findings were discussed by multiple endoscopists and a consensus was reached.

Fig. 1.

Typical M-NBI images of each vascular patterns of SNADETs. a Lesion displaying absent pattern. b Lesion displaying network pattern. c Lesion displaying ISV pattern. d Lesion displaying unclassified pattern. e Lesion displaying mixed pattern. e2 Magnified image of red square in (e1). Network pattern was observed in the central part and absent was observed in the marginal part. M-NBI, magnifying endoscopy with narrow-band imaging; ISV, intrastructural vascular.

Fig. 1.

Typical M-NBI images of each vascular patterns of SNADETs. a Lesion displaying absent pattern. b Lesion displaying network pattern. c Lesion displaying ISV pattern. d Lesion displaying unclassified pattern. e Lesion displaying mixed pattern. e2 Magnified image of red square in (e1). Network pattern was observed in the central part and absent was observed in the marginal part. M-NBI, magnifying endoscopy with narrow-band imaging; ISV, intrastructural vascular.

Close modal

Pathological Examination

Resected specimens were re-evaluated for this study and diagnosed by a board-certificated pathologist of the Japanese Society of Pathology, who was blinded to the clinical information and endoscopic diagnosis. Tumors were classified as LGA or HGA/Ca, in accordance with the WHO classification [15]. Epithelial subtypes were diagnosed based on the histomorphology and the result of immunohistochemical staining. Immunohistochemistry for CDX2, MUC5AC and MUC6 were performed. The tumor was judged as positive when >10% of tumor cells were stained. G-type SNADETs were MUC5AC- and/or MUC6-positive and CDX2-negative. On the other hand, I-type was CDX2-positive and negative for MUC5AC and MUC6 (Fig. 2). GI-type was CDX2-positive and positive for MUC5AC and/or MUC6. The result of histomorphology and immunohistochemical staining were concordant for all cases.

Fig. 2.

Histologic images of four types of SNADETs. a Gastric-type low-grade adenoma. Hematoxylin and eosin (HE)-stained image. b Gastric-type high-grade adenoma, HE-stained image. c Immunohistochemical staining for gastric-type tumor, CDX2 (left), MUC5AC (middle), and MUC6 (right). d Intestinal-type low-grade adenoma, HE-stained image. e Intestinal-type high-grade adenoma, HE-stained image. f Immunohistochemical staining for intestinal-type tumor, CDX2 (left), MUC5AC (middle), and MUC6 (right).

Fig. 2.

Histologic images of four types of SNADETs. a Gastric-type low-grade adenoma. Hematoxylin and eosin (HE)-stained image. b Gastric-type high-grade adenoma, HE-stained image. c Immunohistochemical staining for gastric-type tumor, CDX2 (left), MUC5AC (middle), and MUC6 (right). d Intestinal-type low-grade adenoma, HE-stained image. e Intestinal-type high-grade adenoma, HE-stained image. f Immunohistochemical staining for intestinal-type tumor, CDX2 (left), MUC5AC (middle), and MUC6 (right).

Close modal

Statistical Analysis

All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan). Continuous data were compared with Mann-Whitney U test. Categorical analysis of variables was performed using the χ2 and Fisher’s exact test. A multivariate logistic regression analysis was performed by adjusting the variables with a value <0.001 by univariate analysis. Significance was defined as p value <0.05.

Clinical, Histologic, and Endoscopic Characteristics of SNADETs

Among 286 lesions, 205 (71.7%) were LGA and 81 (28.3%) were HGA/Ca, including four submucosal invasive cancer (two G-type and two GI-type). Both G-type and I-type epithelium were present in 20 lesions (GI-type:7%), 23 (8%) were composed of G-type epithelium, and 243 (85%) were composed of I-type epithelium (Table 1). Patients with G-type were significantly older (72 vs. 64 years-old, p < 0.001), and G-type lesions were mostly located in the oral side of papilla (91.3 vs. 45.6%, p < 0.001), compared to those of I-type. There was no significant difference in color between G- and I-types. Macroscopically, G-type included more protruding types than I-type lesions (65.2 vs. 14.4%, p < 0.001). G-types tended to be larger and consisted a higher proportion of HGA/Ca compared to those of I-type, although the difference was not significant. GI-type lesions showed characteristics that were intermediate between G- and I-types in terms of age, tumor diameter, location, and morphology.

Table 1.

Characteristics between different epithelial subtypes of SNADETs

G-type (N = 23)GI-type (N = 20)I-type (N = 243)G versus I p value
Age, years, median (range) 72 (54–84) 71 (45–80) 64 (27–89) <0.001 
Male, n (%) 16 (69.5) 14 (70.0) 168 (69.1) 0.966 
Diameter, mm, median (range) 20.0 (2–50) 15.5 (7–60) 13 (2–68) 0.131 
Location, oral of papilla, n (%) 21 (91.3) 12 (60.0) 111 (45.6) <0.001 
Color, reddish, n (%) 10 (43.4) 11 (55.0) 103 (42.3) 0.919 
Morphology, n (%) 
 Protruding 15 (65.2) 8 (40.0) 35 (14.4) <0.001 
 Nonprotruding/nonexcavated 8 (34.8) 12 (60.0) 208 (85.6)  
Histological grade, n (%) 
 LGA 14 (60.8) 9 (60.0) 182 (74.9) 0.144 
 HGA/Ca 9 (39.1) 11 (40.0) 61 (25.1)  
Submucosal cancer, n (%) 2 (8.7) 2 (10.0) 
M-NBI finding, n (%) 
 Mono-type 17 (73.9) 7 (35.0) 89 (36.6)  
 Mixed-type 6 (26.0) 13 (65.0) 154 (63.4) <0.001 
Treatment method: endoscopic resection, n (%) 18 (78.2) 17 (85.0) 211 (86.8) 
G-type (N = 23)GI-type (N = 20)I-type (N = 243)G versus I p value
Age, years, median (range) 72 (54–84) 71 (45–80) 64 (27–89) <0.001 
Male, n (%) 16 (69.5) 14 (70.0) 168 (69.1) 0.966 
Diameter, mm, median (range) 20.0 (2–50) 15.5 (7–60) 13 (2–68) 0.131 
Location, oral of papilla, n (%) 21 (91.3) 12 (60.0) 111 (45.6) <0.001 
Color, reddish, n (%) 10 (43.4) 11 (55.0) 103 (42.3) 0.919 
Morphology, n (%) 
 Protruding 15 (65.2) 8 (40.0) 35 (14.4) <0.001 
 Nonprotruding/nonexcavated 8 (34.8) 12 (60.0) 208 (85.6)  
Histological grade, n (%) 
 LGA 14 (60.8) 9 (60.0) 182 (74.9) 0.144 
 HGA/Ca 9 (39.1) 11 (40.0) 61 (25.1)  
Submucosal cancer, n (%) 2 (8.7) 2 (10.0) 
M-NBI finding, n (%) 
 Mono-type 17 (73.9) 7 (35.0) 89 (36.6)  
 Mixed-type 6 (26.0) 13 (65.0) 154 (63.4) <0.001 
Treatment method: endoscopic resection, n (%) 18 (78.2) 17 (85.0) 211 (86.8) 

G-type, gastric type; GI-type, gastrointestinal mixed type; I-type, intestinal type; LGA, low grade adenoma; M-NBI, narrow-band imaging with magnification; SNADET, superficial non-ampullary duodenal epithelial tumors.

M-NBI Patterns among Epithelial Subtypes and Histologic Grades

To compare the characteristic features of M-NBI between G- and I-type lesions, analysis was performed in 266 lesions after excluding GI-type lesions (Table 2). The predominant pattern in G-type was ISV, and the ratio of lesions with ISV was significantly higher than that of I-type (78.2 vs. 26.3%, p < 0.001). For I-type lesions, network and absent were frequently observed, and the ratio of lesions with absent was significantly higher than that of G-type (34.5 vs. 0%, p < 0.001). A higher ratio of lesions in G-type was diagnosed as mono-pattern (73.9 vs. 36.6%, p < 0.001). Table 3 shows the distribution of M-NBI patterns according to histological grade in G- and I-type lesions. The ratio of lesions with mono-pattern between LGA and HGA/Ca did not have significant difference among G-type lesions (p = 0.525). For both LGA and HGA/Ca in G-type, ISV was the predominant pattern observed. For I-type lesions, the ratio of lesions with mixed-pattern in HGA/Ca was significantly higher than that of LGA (77.0 vs. 58.8%, p = 0.01).

Table 2.

Comparison of major vascular patterns observed with M-NBI between G- and I-type lesions

G-type 23I-type 243p value
Vascular pattern 
 Absent, n (%) 0 (0) 84 (34.5) <0.001 
 ISV, n (%) 18 (78.2) 64 (26.3) <0.001 
 Network, n (%) 5 (21.7) 85 (34.9) 0.199 
 Unclassified, n (%) 0 (0) 10 (4.1) 0.321 
G-type 23I-type 243p value
Vascular pattern 
 Absent, n (%) 0 (0) 84 (34.5) <0.001 
 ISV, n (%) 18 (78.2) 64 (26.3) <0.001 
 Network, n (%) 5 (21.7) 85 (34.9) 0.199 
 Unclassified, n (%) 0 (0) 10 (4.1) 0.321 

M-NBI, narrow-band imaging with magnification; G-type, gastric type; I-type, intestinal type; ISV, intrastructural vascular.

Table 3.

Distribution of M-NBI patterns according to histological grade in (a) gastric-type and (b) intestinal-type SNADETs

Low-grade adenoma (N = 14)High-grade adenoma/carcinoma (N = 9)p value
a) Gastric-type 
 Mono-pattern, n (%) 11 (78.5) 6 (66.7) 0.525 
 Mixed-pattern, n (%) 3 (21.4) 3 (33.3)  
 M-NBI pattern in mono-pattern lesions, n (%) N = 11 N = 6  
  Absent 
  Network 1 (9.1) 
  ISV 10 (90.9) 6 (100) 0.751 
  Unclassified 
 M-NBI pattern in mixed-pattern lesions, n N = 3 (total 6 patterns) N = 3 (total 6 patterns)  
  Absent 3/3 2/3 
  Network 3/3 1/3 
  ISV 0/3 2/3 
  Unclassified 0/3 1/3 
Low-grade adenoma (N = 14)High-grade adenoma/carcinoma (N = 9)p value
a) Gastric-type 
 Mono-pattern, n (%) 11 (78.5) 6 (66.7) 0.525 
 Mixed-pattern, n (%) 3 (21.4) 3 (33.3)  
 M-NBI pattern in mono-pattern lesions, n (%) N = 11 N = 6  
  Absent 
  Network 1 (9.1) 
  ISV 10 (90.9) 6 (100) 0.751 
  Unclassified 
 M-NBI pattern in mixed-pattern lesions, n N = 3 (total 6 patterns) N = 3 (total 6 patterns)  
  Absent 3/3 2/3 
  Network 3/3 1/3 
  ISV 0/3 2/3 
  Unclassified 0/3 1/3 
Low-grade adenoma (N = 182)High-grade adenoma/carcinoma (N = 61)p value
b) Intestinal-type 
 Mono-pattern, n (%) 75 (41.2) 14 (23.0) 0.01 
 Mixed-pattern, n (%) 107 (58.8) 47 (77.0)  
 M-NBI pattern in mono-pattern lesions, n (%) N = 75 N = 14  
  Absent 28 (37.3) 7 (50.0) 0.388 
  Network 16 (21.3) 4 (28.6) 0.508 
  ISV 30 (40.0) 2 (14.3) 0.077 
  Unclassified 1 (1.3) 1 (7.1) 0.291 
 M-NBI pattern in mixed-pattern lesions, n N = 107 (total 217 patterns) N = 47 (total 98 patterns)  
  Absent 86/107 35/47 0.299 
  Network 73/107 32/47 0.847 
  ISV 51/107 24/47 0.787 
  Unclassified 7/107 7/47 0.106 
Low-grade adenoma (N = 182)High-grade adenoma/carcinoma (N = 61)p value
b) Intestinal-type 
 Mono-pattern, n (%) 75 (41.2) 14 (23.0) 0.01 
 Mixed-pattern, n (%) 107 (58.8) 47 (77.0)  
 M-NBI pattern in mono-pattern lesions, n (%) N = 75 N = 14  
  Absent 28 (37.3) 7 (50.0) 0.388 
  Network 16 (21.3) 4 (28.6) 0.508 
  ISV 30 (40.0) 2 (14.3) 0.077 
  Unclassified 1 (1.3) 1 (7.1) 0.291 
 M-NBI pattern in mixed-pattern lesions, n N = 107 (total 217 patterns) N = 47 (total 98 patterns)  
  Absent 86/107 35/47 0.299 
  Network 73/107 32/47 0.847 
  ISV 51/107 24/47 0.787 
  Unclassified 7/107 7/47 0.106 

M-NBI, magnifying endoscopy with narrow-band imaging; SNADET, superficial non-ampullary duodenal epithelial tumor; ISV, intrastructural vascular.

Among the variables with a value <0.001 by univariate analysis, location oral to papilla, protruding morphology, and major M-NBI pattern (ISV) were selected for multivariate logistic regression analysis. All three variables were independent predictors for G-type (Table 4).

Table 4.

Multivariate analysis of the association between endoscopic characteristics and gastric-type SNADETs

Odds ratio95% CIp value
Location 
 Oral to papilla 8.860 1.862–42.158 0.006 
 Anal to papilla   
Morphology 
 Protruding 6.119 2.101–17.816 0.0001 
 Non-protruding   
M-NBI pattern 
 ISV 3.306 1.031–10.604 0.044 
 Network   
Odds ratio95% CIp value
Location 
 Oral to papilla 8.860 1.862–42.158 0.006 
 Anal to papilla   
Morphology 
 Protruding 6.119 2.101–17.816 0.0001 
 Non-protruding   
M-NBI pattern 
 ISV 3.306 1.031–10.604 0.044 
 Network   

SNADET, superficial non-ampullary duodenal epithelial tumor; M-NBI, magnifying endoscopy with narrow-band imaging; ISV, Intrastructural vascular.

Based on our finding, we propose a diagnostic algorithm to diagnose the epithelial subtype of SNADETs (Fig. 3). Any lesion anal to the papilla or with a dominant absent pattern could be an I-type. Lesions with a dominant ISV, located on the oral side of papilla, G-type should be considered. Based on this algorithm, the sensitivity was 84.7%, specificity was 89.5%, and accuracy was 85%.

Fig. 3.

Algorithm of diagnosing epithelial subtypes of SNADETs. G-type, gastric-type; I-type, intestinal-type; ISV, intrastructural vascular; M-NBI, magnifying endoscopy with narrow-band imaging.

Fig. 3.

Algorithm of diagnosing epithelial subtypes of SNADETs. G-type, gastric-type; I-type, intestinal-type; ISV, intrastructural vascular; M-NBI, magnifying endoscopy with narrow-band imaging.

Close modal

This study focused on epithelial subtypes using a large cohort of 286 lesions and revealed that 8% of SNADETs were G-type. Previous studies included smaller cases and reported that the frequency of G-type was 3.5–17.7% [9, 11, 16, 17]. As far as we know, we analyzed the largest number of SNADETs and confirmed that the ratio of G-type was low among SNADETs. By accumulating the largest number of G-type, a more detailed comparison of endoscopic characteristics including M-NBI with those of I-type was performed. Three endoscopic characteristics; location oral to papilla, protruding morphology, and major M-NBI pattern (ISV) were independent predictors for G-type.

Patients with G-type SNADETs were significantly older than patients with I-type. This finding was in line with previous studies [10, 16, 18, 19]. In addition, G-types were mostly located in the oral side of the papilla, had larger size than I-types, and were protruding types. These characteristics were consistent with previous studies [11, 12, 16, 18, 19]. Interestingly, the features of mixed-type lesions were intermediate between G-type and I-type. A previous study reported that the reddish color was a characteristic of G-type [12, 16], while another reported that there was no significant association between color and phenotypes [20]. In our study, there was no significant difference in terms of color between G- and I-type lesions. Reddish color has also been reported as a characteristic of HGA/Ca [5, 6, 8]. However, there was no significant correlation between color and histological grade in our study [red in LGA versus HGA/Ca; 42.9% (88/205) versus 43.9% (36/82), p = 1]. Since color is a subjective evaluation and changes even with inflammation, further research is needed on the relationship between lesion color and histologic subtypes.

Analysis of M-NBI revealed that ISV was the predominant pattern of G-type. ISV pattern is a unique vascular form in which dilated or tortuous vessels are observed within the mucosal structure [7]. Based on an analysis of 11 G-type SNADETs, oval-shaped marginal epithelium, dense pattern, and dilatation of the intervening part were reported as characteristic M-NBI findings [16]. Another study reported that oval structures were more frequently observed in G-type SNADETs using acetic acid with M-NBI [21]. Histological analysis revealed that villous structure and low tumor gland density were observed in G-type SNADETs [12]. Thus, we consider that the dilatation of the intervening part construct a roundish mucosal structure with dilated or tortuous vessels within the mucosal structure, that is ISV. Although the terminology varies among previous studies, ISV, oval-shape or structure, dense pattern, and dilatation of the intervening part are considered to capture the same M-NBI findings that reflect the histology of G-type.

Predominant M-NBI types for I-type lesions were absent and network in this study. Absent pattern is similar to WOS, also known as milk-white mucosa, which was first described as a characteristic finding of duodenal adenomas [22]. WOS has been shown to be intraepithelial lipid droplets, which can be also observed in nontumoral duodenal mucosa after high-fat meal [23, 24]. Previous studies reported that WOS was characteristic for I-type SNADETs [16, 19], suggesting that WOS-positive tumors have originated from the intestinal epithelium. On the other hand, network has been observed in 36% of I-type in this study, which ratio was similar to absent. A tubular architecture has been reported as the predominant histological architecture of I-type SNADETs [12]. In the stomach, a previous study revealed that fine network pattern of M-NBI was correlated with I-type early gastric cancer [25]. Thus, the regular arrangement of ducts and blood vessels is a characteristic M-NBI, that is network, and we consider that network could be predictable of I-type lesions.

Previous SNADET studies based on M-NBI reported a diagnostic algorism for HGA/Ca by the presence of mixed-pattern or a mono-pattern of irregular/unclassified pattern [7, 9, 26‒28]. Our study also showed that the proportion of mixed-pattern was significantly higher among HGA/Ca than LGA (71.6 vs. 56%, p = 0.016) without considering the epithelial subtypes. However, one study reported that the accuracy of this diagnostic algorism was lower among G-type compared to that of I-type lesions [9]. In our study, the presence of mixed-pattern was significantly higher in I-type than in G-type and was significantly correlated with HGA/Ca in I-type but not with HGA/Ca of G-type. In G-type, lesions with ISV pattern accounted for the majority in both LGA and HGA/Ca. In a previous study, ISV was not useful to differentiate LGA and HGA/Ca SNADETs [7]. We also found that there was no difference in the ratio of lesions with ISV between LGA and HGA/Ca among G-type. Therefore, it is still difficult to distinguish between LGA and HGA/Ca in G-type and further investigation is required.

Recently, a randomized controlled study reported that cold snare polypectomy (CSP) is preferable for most small SNADETs, but underwater endoscopic mucosal resection (UEMR) should be selected for lesions that cannot be definitively diagnosed as LGA [29]. Based on our diagnostic algorithm, if the lesion is small with absent M-NBI mono-type, it can be diagnosed as I-type LGA and CSP would be the resection method. On the other hand, UEMR should be considered for lesions with dominant ISV or M-NBI mixed-type since the lesion could be G-type or I-type HGA/Ca.

This study has some limitations. First, it was a single-center retrospective study. Although we evaluated endoscopic images from consecutive lesions, there might be possible selection bias. Pathological diagnosis of histologic grades of SNADETs is not well established and may differ depending on each pathologists and classification. All endoscopic images and histological findings were reviewed for this study; thus, both endoscopic diagnosis and pathological handling could be standardized for this study. Re-evaluation of pathological diagnosis was standardized according to the WHO classification. Second, 251 lesions (88%) including all 23 G-type lesions had received biopsy at previous facilities, which might have affected the endoscopic findings including M-NBI. We performed a sub-analysis between lesions with or without biopsy among I-type lesions and did not find significant differences in terms of lesion color and M-NBI findings (online suppl. Table; for all online suppl. material, see https://doi.org/10.1159/000539308). Third, there may have been differences in magnification, which may have influenced the M-NBI findings. Thus, it is necessary to evaluate M-NBI in a prospective study. Fourth, there is no consensus on thresholds for immunostaining positivity so far. We defined the threshold as >10% in this study, although the population of epithelial subtypes can change depending on the thresholds. In conclusion, we found that not only absent but also network is often observed in I-type, and ISV is predictive of G-type based on M-NBI findings. In addition, mixed-pattern of M-NBI findings can be used for predicting HGA/Ca among I-type SNADETs but not for G-types. Together with tumor location and macroscopic type, M-NBI is useful to predict epithelial subtype and malignant potential.

We thank Yuko Katayama for her great effort in preparing the slides.

This study protocol was reviewed and approved by the Ethics Committee of Nagoya University Hospital (approval No. 2021-152). Opt-out informed consent protocol was used for use of participant data for research purposes. This consent procedure was reviewed and approved by the Ethics Committee of Nagoya University Hospital (approval No. 2021-152), July 12, 2021.

The authors have no conflicts of interest to declare.

This study was not supported by any sponsor or funder.

Y.K., N.K., and T.H. contributed to the conception. Y.K., N.K., T.H., M.N., O.Y., and H.T. contributed to acquisition of data for the work. Y.K. and N.K. drafted the manuscript. T.H., M.N., O.Y., H.T., H.K., and M.F. revised it critically for important intellectual content. All authors agreed to be accountable for all aspects of the work and gave final approval of the manuscript.

All data generated or analyzed during this study are included in this article. Further enquiries can be directed to the corresponding author.

1.
Yoshida
M
,
Yabuuchi
Y
,
Kakushima
N
,
Kato
M
,
Iguchi
M
,
Yamamoto
Y
, et al
.
The incidence of non‐ampullary duodenal cancer in Japan: the first analysis of a national cancer registry
.
J Gastroenterol Hepatol
.
2021
;
36
(
5
):
1216
21
.
2.
Lu
Y
,
Fröbom
R
,
Lagergren
J
.
Incidence patterns of small bowel cancer in a population-based study in Sweden: increase in duodenal adenocarcinoma
.
Cancer Epidemiol
.
2012
;
36
(
3
):
e158
63
.
3.
Cloyd
JM
,
George
E
,
Visser
BC
.
Duodenal adenocarcinoma: advances in diagnosis and surgical management
.
World J Gastrointest Surg
.
2016
;
8
(
3
):
212
21
.
4.
Nakagawa
K
,
Sho
M
,
Okada
K
,
Akahori
T
,
Aoyama
T
,
Eguchi
H
, et al
.
Surgical results of non-ampullary duodenal cancer: a nationwide survey in Japan
.
J Gastroenterol
.
2022
;
57
(
2
):
70
81
.
5.
Kakushima
N
,
Yoshida
M
,
Iwai
T
,
Kawata
N
,
Tanaka
M
,
Takizawa
K
, et al
.
A simple endoscopic scoring system to differentiate between duodenal adenoma and carcinoma
.
Endosc Int Open
.
2017
;
5
:
E763
8
.
6.
Ishii
R
,
Ohata
K
,
Sakai
E
,
Takita
M
,
Minato
Y
,
Muramoto
T
, et al
.
Simple scoring system for the diagnosis of superficial non‐ampullary duodenal epithelial tumors
.
Dig Endosc
.
2021
;
33
(
3
):
399
407
.
7.
Kikuchi
D
,
Hoteya
S
,
Iizuka
T
,
Kimura
R
,
Kaise
M
.
Diagnostic algorithm of magnifying endoscopy with narrow band imaging for superficial non-ampullary duodenal epithelial tumors: NBI magnification for duodenal tumors
.
Dig Endosc
.
2014
;
26
(
Suppl 2
):
16
22
.
8.
Goda
K
,
Kikuchi
D
,
Yamamoto
Y
,
Takimoto
K
,
Kakushima
N
,
Morita
Y
, et al
.
Endoscopic diagnosis of superficial non-ampullary duodenal epithelial tumors in Japan: multicenter case series
.
Dig Endosc
.
2014
;
26
(
Suppl 2
):
23
9
.
9.
Toya
Y
,
Endo
M
,
Oizumi
T
,
Akasaka
R
,
Yanai
S
,
Kawasaki
K
, et al
.
Diagnostic algorithm of magnifying endoscopy with crystal violet staining for non-ampullary duodenal epithelial tumors
.
Dig Endosc
.
2020
;
32
(
7
):
1066
73
.
10.
Ushiku
T
,
Arnason
T
,
Fukayama
M
,
Lauwers
GY
.
Extra-ampullary duodenal adenocarcinoma
.
Am J Surg Pathol
.
2014
;
38
(
11
):
1484
93
.
11.
Toba
T
,
Inoshita
N
,
Kaise
M
,
Nomura
K
,
Kuribayashi
Y
,
Tanaka
M
, et al
.
Clinicopathological features of superficial non-ampurally duodenal epithelial tumor; gastric phenotype of histology correlates to higher malignant potency
.
J Gastroenterol
.
2018
;
53
(
1
):
64
70
.
12.
Yoshida
M
,
Shimoda
T
,
Abe
M
,
Kakushima
N
,
Kawata
N
,
Takizawa
K
, et al
.
Clinicopathological characteristics of non-ampullary duodenal tumors and their phenotypic classification
.
Pathol Int
.
2019
;
69
(
7
):
398
406
.
13.
Minatsuki
C
,
Yamamichi
N
,
Inada
K
,
Takahashi
Y
,
Sakurai
K
,
Shimamoto
T
, et al
.
Expression of gastric markers is associated with malignant potential of nonampullary duodenal adenocarcinoma
.
Dig Dis Sci
.
2018
;
63
(
10
):
2617
25
.
14.
Participants in the Paris Workshop
.
The Paris endoscopic classification of superficial neoplastic lesions: esophagus, stomach, and colon
.
Gastrointest Endosc
.
2003
;
58
(
6
):
S3
43
.
15.
IARC. WHO Classification of Tumours Editorial Board
.
WHO classification of tumours of the digestive system
. 5th ed.
Lyon
;
2019
.
16.
Akazawa
Y
,
Ueyama
H
,
Tsuyama
S
,
Ikeda
A
,
Yatagai
N
,
Komori
H
, et al
.
Endoscopic and clinicopathological features of superficial non-ampullary duodenal tumor based on the mucin phenotypes
.
Digestion
.
2021
;
102
(
5
):
663
70
.
17.
Mitsuishi
T
,
Hamatani
S
,
Hirooka
S
,
Fukasawa
N
,
Aizawa
D
,
Hara
Y
, et al
.
Clinicopathological characteristics of duodenal epithelial neoplasms: focus on tumors with a gastric mucin phenotype (pyloric gland-type tumors)
.
Plos One
.
2017
;
12
(
4
):
e0174985
.
18.
Toya
Y
,
Endo
M
,
Akasaka
R
,
Urushikubo
J
,
Gonai
T
,
Asakura
K
, et al
.
Clinicopathological features and magnifying chromoendoscopic findings of non-ampullary duodenal epithelial tumors
.
Digestion
.
2018
;
97
(
3
):
219
27
.
19.
Fukusada
S
,
Shimura
T
,
Iwasaki
H
,
Okuda
Y
,
Katano
T
,
Nishigaki
R
, et al
.
Relationship between immunophenotype and clinicopathological findings for superficial nonampullary duodenal epithelial tumor
.
Digestion
.
2021
;
102
(
6
):
870
7
.
20.
Hijikata
K
,
Nemoto
T
,
Igarashi
Y
,
Shibuya
K
.
Extra-ampullary duodenal adenoma: a clinicopathological study
.
Histopathology
.
2017
;
71
(
2
):
200
7
.
21.
Tanaka
Y
,
Fujii
S
,
Okada
K
,
Aoki
K
,
Shibuya
S
,
Kusaka
T
.
Magnifying narrow-band imaging with acetic acid spray for superficial non-ampullary duodenal epithelial tumors based on the mucin phenotype
.
Digestion
.
2022
;
103
(
6
):
421
7
.
22.
Tanaka
M
,
Ushida
K
,
Ohkura
Y
.
Significance of magnifying endoscopy in diagnosis of duodenal elevated lesions
.
Stomach Intestine
.
2003
;
38
:
1709
20
. (Japanese with an English abstract).
23.
Yao
K
,
Iwashita
A
,
Nambu
M
,
Tanabe
H
,
Nagahama
T
,
Maki
S
, et al
.
Nature of white opaque substance in gastric epithelial neoplasia as visualized by magnifying endoscopy with narrow-band imaging
.
Dig Endosc
.
2012
;
24
(
6
):
419
25
.
24.
Takahashi
H
,
Yao
K
,
Nagahama
T
,
Miyaoka
M
,
Ohtsu
K
,
Kanemitsu
T
, et al
.
Visualization of absorbed lipid in the normal duodenal epithelium using magnifying endoscopy with narrow-band imaging
.
Dig Dis Sci
.
2022
;
67
(
12
):
5610
6
.
25.
Kobayashi
M
,
Takeuchi
M
,
Ajioka
Y
,
Hashimoto
S
,
Sato
A
,
Narisawa
R
, et al
.
Mucin phenotype and narrow-band imaging with magnifying endoscopy for differentiated-type mucosal gastric cancer
.
J Gastroenterol
.
2011
;
46
(
9
):
1064
70
.
26.
Kakushima
N
,
Yoshida
M
,
Yamaguchi
Y
,
Takizawa
K
,
Kawata
N
,
Tanaka
M
, et al
.
Magnified endoscopy with narrow-band imaging for the differential diagnosis of superficial non-ampullary duodenal epithelial tumors
.
Scand J Gastroenterol
.
2019
;
54
(
1
):
128
34
.
27.
Tanaka
Y
,
Fujii
S
,
Oiwa
Y
,
Kusaka
T
,
Shibuya
S
,
Kokuryu
H
.
Efficacy of magnifying narrow band imaging with acetic acid spray in diagnosing superficial non-ampullary duodenal epithelial tumors
.
Digestion
.
2021
;
102
(
4
):
572
9
.
28.
Kakushima
N
,
Yoshida
M
,
Takizawa
K
,
Yabuuchi
Y
,
Kawata
N
,
Kishida
Y
, et al
.
White light and/or magnifying endoscopy with narrow band imaging for superficial nonampullary duodenal epithelial tumors
.
Scand J Gastroenterol
.
2021
;
56
(
2
):
211
8
.
29.
Miyazaki
K
,
Nakayama
A
,
Sasaki
M
,
Minezaki
D
,
Morioka
K
,
Iwata
K
, et al
.
Resectability of small duodenal tumors: a randomized controlled trial comparing underwater endoscopic mucosal resection and cold snare polypectomy
.
Am J Gastroenterol
.
2024
;
119
(
5
):
856
63
Online ahead of print.