In recent years, the indications for endoscopic submucosal dissection (ESD) have been expanded to include undifferentiated-type early gastric cancers (UD-type EGCs; intramucosal, diameter ≤2 cm without ulceration).^1-4 However, demarcating UD-type EGCs is difficult because of the presence of proliferative zone extensions.^5,6 The horizontal margin-positive rate can range from 1.7% to 52.6%, and the proportion of curability by ESD can range from 63.9% to 82.5%; hence, both of these parameters suggest noticeable inter-institutional differences.^7-14 Moreover, the curative resection rate of ESD reportedly decreases as the tumor diameter increases in UD-type EGCs.¹⁵ Accordingly, improvements in the preoperative diagnostic ability are expected. Recently, the usefulness of magnifying endoscopy with narrow-band imaging (ME-NBI) in delineating the demarcation of differentiated-type EGCs has been reported;^16-19 however, its usefulness for UD-type EGC remains unclear.

Helicobacter pylori is observed in most patients with gastric cancer.^20,21 Eradication of H. pylori reportedly reduces the risk of developing gastric cancer;²² therefore, eradication is widely considered in daily clinical practice. However, H. pylori eradication in differentiated-type EGCs makes it difficult to demarcate cancer lesions because the luminal surface of the stomach is covered with gastritis-like tissue.^23,24 Additionally, its effect on UD-type EGCs remains unclear.

Cases of H. pylori-uninfected EGC have also been reported, but with a prevalence as low as 0.66%.²⁵ In addition, UD-type EGC is predominant in H. pylori-uninfected EGCs.²⁵ However, the difference in endoscopic findings between H. pylori-uninfected and H. pylori-infected cases is unclear.

We previously examined the diagnosis of demarcation of UD-type EGCs, the features of UD-type EGCs following H. pylori eradication, and the features of H. pylori-uninfected UD-type EGCs.^26-29 Accordingly, this study aimed to introduce the findings of our previous studies on the subject and discuss the diagnosis of demarcation in UD-type EGCs, the features of UD-type EGCs following H. pylori eradication, and the features of H. pylori-uninfected UD-type EGCs.

We previously examined the utility of ME-NBI for the diagnosis of demarcation in UD-type EGCs in ESD.²⁶ Specifically, we compared the diagnostic accuracy of demarcation between white-light imaging (WLI) alone and WLI+ME-NBI in 76 patients (76 lesions). UD-type EGC invasion collated with ME-NBI findings as shown in Fig. 1.^26,30,31 If the UD-type EGC remains in the proliferative zone, it is considered an extended intervening part. If UD-type EGC is found spanning the proliferative zone up to the superficial layer, it can be observed as wavy microvessels. If UD-type EGC is found in the entire layer, it is observed as a cork-screw pattern. Based on these findings, the demarcation of UD-type EGC was delineated, and argon plasma coagulation (APC) markings were made on the utmost oral and the utmost anal sides of each lesion diagnosed using WLI or ME-NBI. The APC markings are shown in Fig. 2. A correct diagnosis was defined as a diagnosis extent that coincided with the extent determined on the pathological tissue slides. We subsequently compared the accuracies of WLI and WLI+ME-NBI.

Figure 1. Undifferentiated-type early gastric cancer (UD-type EGC) invasion collated with magnifying endoscopy with narrow-band imaging (ME-NBI) findings. (A) If the UD-type EGC remains in the proliferative zone, it can be observed as an extended intervening part of ME-NBI. (B) If UD-type EGC is found to extend from the proliferative zone to the superficial layer, wavy microvessels are observed on ME-NBI. (C) If UD-type EGC extends across the whole layer, it is observed as a cork-screw pattern in ME-NBI. Taken from Horiuchi Y et al. Gastric Cancer 2016;19:515-523, with permission from Springer Nature.²⁶

Figure 2. Actual argon plasma coagulation (APC)-marking images. The dotted lines indicate the tumor border region, as viewed under white-light imaging (WLI). The arrow indicates a tumor border region as viewed using magnifying endoscopy with narrow-band imaging (ME-NBI). (A) Tumor border at the utmost oral site under WLI. (B) Demarcation line as viewed using ME-NBI (coinciding with A). (C) A higher magnification view of B. (D) An APC marking placed on the demarcation line. (E) Tumor border at the utmost anal site under WLI. (F) Demarcation line as viewed using ME-NBI (inconsistent with E). (G) A higher magnification view of F. (H) An APC marking placed on the demarcation line. Taken from Horiuchi Y et al. Gastric Cancer 2016;19:515-523, with permission from Springer Nature.²⁶

As a result (in our previous study²⁶, from among 76 lesions, a diagnosis using WLI was feasible in 41 lesions (53.9%), while it was difficult or infeasible in 35 lesions (46.1%). With ME-NBI, the diagnosis became feasible for 21 (27.6%) of these 35 difficult or infeasible lesions, and the accurate diagnosis rate increased to 81.5% because of this additional effect. When comparing the accurate demarcation diagnosis rate according to the degree of background mucosal status based on the Updated Sydney System,³² no significant differences in atrophy or intestinal epithelial metaplasia were observed between the normal-to-mild and moderate-to-marked groups (87.2% vs 76.7%, 85.5% vs 64.3%, respectively). However, patients with milder neutrophil and mononuclear cell infiltration had a significantly higher accurate diagnosis rate (87.1% vs 57.1%, p<0.05; 93.5% vs 73.3%, p<0.05, respectively).

Subsequently, we examined the utility of ME-NBI for the diagnosis and demarcation of UD-type EGCs in surgical resection.²⁷ APC marking in these cases was performed in the same manner as in the ESD cases. The evaluation method was also the same as that for the ESD cases. We subsequently compared the accuracies of WLI and WLI+ME-NBI.

As a result (in our previous study²⁷, 35 lesions (47.3%) could be diagnosed using WLI alone, and 39 lesions (52.7%) could not be diagnosed or were difficult to diagnose using WLI alone. ME-NBI allowed for diagnosis in 27 (36.5%) of these 39 lesions, and this additional effect increased the rate of correct diagnosis to 83.8%. Comparison of the accurate demarcation diagnosis rate according to the degree of background mucosal status based on the Updated Sydney System³² revealed no significant difference in intestinal metaplasia between the normal-to-mild and moderate-to-marked groups (87.2% vs 77.8%). However, cases with milder atrophy, neutrophil, and mononuclear cell infiltration had a significantly higher accurate diagnosis rate (96.0% vs 77.6%, p<0.05; 94.0% vs 62.5%, p<0.05; 100% vs 78.6%, p<0.05, respectively).

To summarize, the diagnostic accuracy of demarcation with ME-NBI was approximately 80% in ESD and surgical cases, indicating that ME-NBI has an add-on effect of approximately 30% compared to conventional WLI. Therefore, in UD-type EGCs, ME-NBI is useful for delineating demarcation, regardless of tumor size. Additionally, inflammatory cell infiltration appeared to be a cause of misdiagnosis, suggesting that the resolution of inflammation could contribute to the accurate diagnosis of demarcation.

H. pylori is a major cause of inflammation in the stomach.^20,21 Therefore, eradication of H. pylori may improve the rate of accurate demarcation diagnosis. However, eradication of H. pylori with differentiated-type EGCs makes it difficult to demarcate cancer lesions as the luminal surface of the stomach is covered with gastritis-like tissue.^23,24 To the best of our knowledge, there have been no reports of UD-type EGCs to date. Therefore, we compared accurate diagnosis rates for APC marking between patients with and without eradication of H. pylori to determine whether eradication was as beneficial as believed.²⁸

The inclusion criteria for the eradication group were as follows: (1) negative H. pylori antibody or C urea breath test results if the patient underwent H. pylori eradication at our hospital or another hospital, or (2) confirmed negative urea breath test result ≥4 weeks after initiating H. pylori eradication if the patient was positive for H. pylori antibodies or had a positive urea breath test result at the first examination at our hospital. As such, the H. pylori-infected without eradication group did not meet these criteria and was labeled as the “non-eradication group.”

APC markings were made on the utmost oral and the utmost anal sides of each lesion diagnosed using ME-NBI. A correct diagnosis was defined as a diagnosis result matching the one determined from the pathological tissue slides.

Compared with the noncancerous region, the cancerous region shows an expansion of the distance between glands in UD-type EGCs.⁶ The intercrypt distance is defined as the distance between the center of a crypt and that of an adjacent crypt. Based on this, the intercrypt distance was measured at 5 to 10 points in each of the cancerous and noncancerous regions, and the mean distances were calculated from the measurements. The mean distance ratio of the noncancerous region to the cancerous region was calculated as the intercrypt distance ratio (Fig. 3). Based on this, the eradication and non-eradication groups consisted of 27 (27 lesions) and 33 (33 lesions) cases, respectively. In both groups, the lesions were discolored, the macroscopic type was mostly depressed, and the majority of lesions were located in the middle and lower regions of the stomach, with no difference between the groups.

Figure 3. Intercrypt distance expansion. The cancerous region shows an expansion of the distance between glands (intercrypt distance) compared with the noncancerous region. Taken from Horiuchi Y et al. Gastric Cancer 2018;21:988-997, with permission from Springer Nature.²⁹

The diagnostic accuracy was significantly higher in the eradication group than in the non-eradication group (92.6% vs 60.6%, p<0.05). The mean intercrypt distance ratio was significantly higher in the eradication group than in the non-eradication group (1.94, 95% confidence interval [CI] 1.69–2.2 vs 1.59, 95% CI 1.40–1.78; p<0.05). Comparison of the normal-mild inflammation of the background mucosa between the eradication and non-eradication groups revealed that neutrophil infiltration was significantly milder in the eradication group than in the non-eradication group (85.2% vs 57.6%, p<0.05). In contrast, no significant differences in atrophy, intestinal metaplasia, or mononuclear cells were observed between the eradication and non-eradication groups (44.4% vs 27.3%, 77.8% vs 66.7%, 40.7% vs 18.2%, respectively).

To summarize, the diagnostic accuracy rate was significantly higher in the eradication group than in the non-eradication group. This suggests that H. pylori eradication is beneficial for diagnostic demarcation and may improve the rate of curative resection with ESD. This is consistent with the findings of previous reports that inflammation disappears within 4 weeks following the eradication of H. pylori, in contrast to that in patients without H. pylori eradication.³³ Neutrophil infiltration has also been reported to disappear within 4 weeks.³⁴ These results indicate that when H. pylori is eradicated after cancer detection, the demarcation of the lesion is more defined after 4 weeks in patients who undergo H. pylori eradication than in those who do not.

However, it is not advisable to postpone ESD for UD-type EGCs for several months to confirm whether eradication is successful because UD-type EGCs may progress during this time. The current success rate of eradication is reported to be 92.6%.³⁵ Therefore, it is important to assume that most patients benefit from eradication and to perform eradication as soon as cancer is detected and then perform ESD promptly without waiting to confirm whether the eradication has been successful.

Most patients with gastric cancer are H. pylori positive,^20,21 and the prevalence of H. pylori-uninfected EGCs is low; however, the UD type is predominant in H. pylori-uninfected EGCs.²⁵ In general, the major characteristics of H. pylori-uninfected UD-type EGCs are as follows:^29,36 color, discolored; location, the middle and/or lower regions; macroscopic type, flat; ME-NBI, extended intervening part; depth, mucosa; histological type, sig; diameter, small (Fig. 4).²⁹ However, to the best of our knowledge, there have been no reports on the differences in endoscopic findings between H. pylori-uninfected and infected cases. Therefore, we compared the endoscopic findings between the H. pylori-uninfected and infected groups (i.e., the eradication and non-eradication groups).²⁹

Figure 4. Endoscopic and pathologic findings of Helicobacter pylori-uninfected undifferentiated-type cancers. (A) Endoscopic findings with white-light imaging. The discolored area is relatively clearly recognizable. (B) Chromoendoscopic finding. (C) Findings of magnifying endoscopy with narrow-band imaging. An extended intervening part is recognizable. (D) Pathological findings following endoscopic therapy. Signet ring cell carcinoma is confined to the proliferative zone of the mucosa. Taken from Horiuchi Y et al. Gastric Cancer 2018;21:988-997, with permission from Springer Nature.²⁹

H. pylori-uninfected cases were defined as meeting all of the following criteria: (1) no prior H. pylori eradication; (2) negative urea breath test results (UBIT; Otsuka, Tokushima, Japan); (3) negative results for H. pylori antibodies (H. pylori antibody II: EIKEN Co. Ltd., Tokyo, Japan); (4) negative pepsinogen (PG) test results (positive cutoff value: PGI ≤70 ng/mL; PGI/II ratio ≤3); (5) endoscopically confirmed positive regular arrangement of collecting venules in the lower gastric body;³⁷ and (6) a histologically confirmed H. pylori-uninfected case and negative inflammatory cell infiltration result based on the updated Sydney system.³²

APC markings were made on utmost oral and the utmost anal sides of each lesion diagnosed using ME-NBI. The definitions of correct diagnosis and mean intercrypt distance ratio were the same as those described in the section (Features of UD-type EGCs following H. pylori eradication).

As a result, the uninfected group included 21 cases (21 lesions), the eradication group included 27 cases (27 lesions), and the non-eradication group included 33 cases (33 lesions). The common endoscopic features in all three groups were as follows: discoloration in the WLI and many sites were in the lower and middle regions of the stomach.

Compared to the non-eradication group, the uninfected group revealed a significantly greater number of cases of the flat type as macroscopic type (42.9% vs 3.0%, p=0.0004), with the extended intervening part noted in the ME-NBI findings (97.6% vs 63.6%, p<0.0001), with signet ring cell carcinoma as the histological type (100% vs 69.7%, p=0.0043), and significantly smaller tumor diameters (7.3±5.2 mm vs 14.6±9.0 mm, p=0.0002). Additionally, in the uninfected group, the accuracy of demarcation and mean intercrypt distance ratio were significantly higher (100% vs 60.6%, p=0.0007; 1.96 [95% CI, 1.70 to 2.21] vs 1.59 [95% CI, 1.40 to 1.78], p=0.0103, respectively) than those in the non-eradication group. Additionally, regarding atrophy, intestinal metaplasia, neutrophil infiltration, and mononuclear cell infiltration (indicative of inflammation of the background mucosa), the uninfected group had significantly milder cases than the non-eradication group (100% vs 27.3%, p<0.0001; 100% vs 66.7%, p=0.0038; 100% vs 57.6%, p=0.0003; 85.7% vs 18.2%, p<0.0001, respectively).

Compared to the eradication group, the uninfected group revealed a significantly greater number of cases of the flat rather than macroscopic type (42.9% vs 11.1%, p=0.0184), with the extended intervening part noted in the ME-NBI findings (97.6% vs 70.4%, p=0.0011), with signet ring cell carcinoma as the histological type (100% vs 63.0%, p=0.0064), and with significantly smaller tumor diameters (7.3±5.2 mm vs 13.6±7.3 mm, p=0.0008). Additionally, compared to the eradication group, the uninfected group had significantly milder cases of atrophy and mononuclear cell infiltration (indicative of inflammation of the background mucosa): 100% versus 44.4%, p<0.0001; 85.7% versus 40.7%, p=0.0026, respectively. Meanwhile, the accuracy of demarcation and the mean intercrypt distance ratio were not significantly different between the groups.

To summarize, the uninfected group was characterized by smaller tumor size, predominantly of a flat type, extended intervening parts in ME-NBI, and pure signet ring cell carcinoma in the histological type. In contrast, the eradication and non-eradication groups were characterized by larger tumor size, of a depressed type, and wavy microvessels in ME-NBI. In support of these characteristics in the uninfected group, the cancer proliferative potential was assessed using the MIB-1 labeling index to determine whether there was a difference in biological behavior between the uninfected and infected groups (Fig. 5).³⁸ The uninfected group had a significantly lower proliferative capacity than the infected group. The tumor size was small owing to low proliferative potential, and the lesions remained in the middle layer of the mucosa instead of being exposed on the surface, resulting in the flat type as macroscopic type and an extended intervening part in ME-NBI (Fig. 1A). Regarding the presence of pure signet ring cell carcinoma in the histology, this lesion type is reported to have a lower risk of metastasis and a lower biological grade than mixed, poorly differentiated carcinoma.^39,40 Further, its presence in uninfected patients may support its lower proliferative potential. There was no difference in the mean intercrypt distance ratio, accuracy, or neutrophil invasion in the background mucosa between the uninfected and eradication groups. Eradication might have resulted in neutrophil infiltration at a level similar to that of the uninfected group and the demarcation between cancerous and noncancerous regions was clearly defined, resulting in a similar diagnostic accuracy between the uninfected and eradicated groups. Meanwhile, in the non-eradication group, there was no difference in the intercrypt distance between the cancer and non-cancer groups due to neutrophil infiltration, and the demarcation between cancerous and noncancerous regions was unclear, resulting in a lower diagnostic accuracy (Fig. 6).

Figure 5. The MIB-1 labeling index. The vertical axis shows the number of patients, and the horizontal axis shows the MIB-1 labeling index. The MIB-1 labeling index was significantly higher in Helicobacter pylori-positive (HP+) patients than in H. pylori-negative (HP−) patients (p<0.0001). Taken from Horiuchi Y et al. Gastric Cancer 2016;19:160-165, with permission from Springer Nature.³⁸

Figure 6. Schema of pathological images arranged by Helicobacter pylori infection status. (A) In the non-eradication group, there is no difference in the intercrypt distance between the cancerous and noncancerous regions due to neutrophil infiltration, and the demarcation is unclear. (B) In the eradication group, the neutrophil infiltration is improved, and the difference in the intercrypt distance is clearly visible, making the demarcation clear. (C) In the uninfected group, there is no neutrophil infiltration; therefore, the difference in the intercrypt distance is clearly visible and the demarcation is clear.

The usefulness of ME-NBI in diagnosing the demarcation of UD-type EGCs and the endoscopic characteristics of eradicated and uninfected cases were reviewed. In UD-type EGCs, ME-NBI was found to be useful for delineating demarcation, regardless of tumor size. Additionally, inflammatory cell infiltration appears to be a major cause of misdiagnosis, suggesting that the resolution of inflammation could contribute to the accurate diagnosis of demarcation. Therefore, the accuracy of demarcation was higher in eradicated and uninfected cases than in non-eradicated cases. The common features of the endoscopic findings included discoloration under WLI and a predominance of lesions in the lower and middle regions across all three groups. The uninfected group was more common than the eradication and non-eradication groups and was characterized by smaller tumor size, flat type, extended intervening parts in ME-NBI, and pure signet ring cell carcinoma.

With the increasing prevalence of eradication therapy for chronic gastritis⁴¹ and the low proportion of H. pylori infection among young patients,^42,43 the number of eradicated and uninfected cases of UD-type EGCs is expected to increase. As indicated in this study, diagnosis with ME-NBI will be useful to improve the curative resection rate by ESD. However, ME-NBI requires skill and has a steep learning curve, although e-learning and artificial intelligence have been reported to assist in diagnosis.^44-46 In addition, the feasibility of further expansion of ESD indications for UD-type EGCs has been reported,^47,48 and the accuracy of pretreatment diagnosis of UD-type EGCs is expected to become more important. Therefore, further improvement in endoscopists’ diagnostic skills, and the environment for such improvement, are expected.

Y.H. reports receiving a Grant-in-Aid for Early-Career Scientists (21K15962), the Japanese Foundation for Research and Promotion of Endoscopy Grant, an academic grant from the Japanese Gastric Cancer Association Research Committee, and personal fees for specific speaking and teaching commitments involving honoraria from Olympus Corp. and Kaken Pharmaceutical Co., Ltd. The remaining authors declare that they have no competing interests.