Endoscopic submucosal dissection (ESD) has been widely accepted as a curative procedure for gastric neoplasm such as early gastric cancer (EGC), gastric adenoma, and gastric subepithelial tumor. With the development and widespread of the screening system in South Korea, the diagnosis of gastric adenoma and EGC has increased. The importance of ESD technique is also increasing. As ESD technique developed, there was no difference in the 5-year overall survival rate in treating gastric neoplasm between ESD and surgery. Thus, ESD is currently considered a standard treatment of EGC if cases are corresponding to absolute and expanded indications.^1-4

However, it is technically difficult to perform successful ESD if one of the following reasons is present: (1) the lesion is located at the esophago-gastric junction, cardia, high-body, or pyloro-duodenum; (2) large tumor size; (3) ulceration or fibrotic scar coexists; and (4) demarcation line is not clear.^5-10 Among them, ESD of gastric neoplasia involving the pyloric channel (GNPC) is technically challenging due to narrow lumen of channel and difficulty in precise assessment of the distal margin.^7,11-13 However, few studies have shown the efficacy and safety of endoscopic resection of GNPC.

The method of ESD through retro-flexion of the endoscope for GNPC was first reported in 2002. Several authors later have reported clinical outcomes of ESD through retro-flexion in case series studies.^6,7,12-15 The retrograde resection method has been studied and used a lot in endoscopic resection of rectal lesions. However, in terms of GNPC, the number of cases is relatively small and specific procedure methods are not demonstrated.^16,17

In this study, we proposed a novel technique to treat GNPC more effectively and safely. We evaluated the feasibility and effectiveness of the ESD through retro-flexion of the endoscope for GNPC and compared with the conventional anterograde resection method. We also investigated predictive factors for postprocedural stenosis during follow-up.

1. Patients

We retrospectively reviewed medical records of a total of 97 patients with GNPC who underwent ESD from January 2007 to October 2017 in a tertiary hospital. We obtained informed consent from all patients before the procedure who agreed to undergo endoscopic resection for the treatment of GNPC and the possibility of occurrence of postprocedural complications including pyloric stenosis. This study was approved by the Ethics Committee of Korea University Guro Hospital (IRB number: K2019-2354-001).

2. ESD procedure by retrograde method

All patients underwent ESD procedure under sedation with an intravenous administration of propofol (1.0 mg/kg) or midazolam (0.035 mg/kg) with close cardio-pulmonary function monitoring. We used conventional flexible endoscopy (GIF-Q260J; Olympus, Shinjuku, Tokyo, Japan) for performing ESD. All procedures were performed by an expert (J.J.P.) who had experience of performing more than 2,000 cases of gastric ESD.

Initially, we checked the margin of the lesion and marked it using needle coagulation. Injection of diluted epinephrine saline solution (dilution rate 1:10,000) that was mixed with indigo carmine dye was followed. Precutting and limited incision on both sides of the distal margin around the pyloric ring were done using a needle knife (KD-IL-1; Olympus), serving as a starting point for incision with retro-flexed endoscope and retrograde dissection from the bulb to antrum. The small incision pocket enhanced mucosal contraction and bulging of the dissecting flap during dissection within the bulb. Retrograde dissection using an insulated-tipped electrosurgical knife (KD-610L; Olympus) crossing the pyloric channel was done as far as possible in order to lift up the dissected bulbar portion of the lesion over the pyloric ring to make a “floating tongue-like” flap toward antrum. Then, dissecting with anterograde approach from antrum to pyloric channel was done for complete resection. The more distance is dissected by retrograde approach beyond pyloric channel, the more meeting point moves upwards towards the antrum, making it easier to finalize resection of the lesion with anterograde approach (Fig. 1). After removing the lesion, electro-coagulation or endo-clipping of visible vessel was done to prevent major bleeding.

Figure 1. Techniques of retrograde transpyloric antral dissection. (A) A flat lesion is found on the anterior wall of the prepylorus. (B) Precutting and limited incision on both sides of the distal margin around the pyloric ring are done, serving as a starting point for incision. (C) With a retro-flexed endoscope, retrograde dissection is done from the bulb to antrum. (D) Retrograde dissection crossing the pyloric ring is done as far as possible in order to lift up the dissected bulbar portion of the lesion over the pyloric ring to make a “floating tongue-like flap.” (E) Dissection is performed in an antegrade manner from antrum to pyloric channel, where the antegrade dissection (red arrow) and retrograde dissection (yellow arrow) points meet.

We did not perform routine preventive method for possible stenosis after ESD for GNPCs. However, if significant stenosis was found during follow-up esophagogastroduodenoscopy, rescue procedure such as endoscopic balloon dilatation was performed whenever necessary.

3. Definition

We defined GNPC as tumors with the distal margin located within a half centimeter from the pyloric ring when the ring was fully open. We defined en bloc resection when the tumor was completely resected as single piece and complete resection as the lesion was resected en bloc with negative vertical and lateral margin without lympho-vascular invasion on microscopic examination. Recurrence was defined as the new lesion confirmed by endoscopic biopsy after the procedure showed local recurrences, synchronous lesions, metachronous lesions, and distant metastatic lesions.^18,19 We defined pyloric stenosis when the endoscope could not pass through the lumen of pyloric channel and patients complained symptoms associated with pyloric stenosis such as nausea and vomiting.

4. Post-ESD follow-up and recurrence assessment

After ESD, an esophagogastroduodenoscopy was performed at intervals of 3 to 6 months for the first year and annually thereafter. Biopsy was performed on ESD scars to confirm local recurrence during endoscopic examination. Random biopsy on antrum and corpus or rapid urease test were performed during follow-up esophagogastroduodenoscopy for evaluation of Helicobacter pylori status. We regarded as current infection of was defined H. pylori if either biopsy with special stain or rapid urease test were positive.

5. Study designs

We subdivided cases based on the location of GNPC, the resection method, and the circumference of resection. First, we subdivided the resection method as anterograde and retrograde resection. In anterograde resection, we performed precut incision from distal margin with forward viewing and dissected the submucosa as standard manner. In contrast, in retrograde resection, we performed precut incision and submucosal dissection of distal part with retroflection of the scope in the bulb and subsequently finished the proximal part with forward viewing. Second, we subdivided the location of GNPC as confined to channel, antrum to channel, channel to bulb, and antrum to bulb through the channel. Third, we subdivided the circumference of resection as <25%, 25% to 75%, and ≥75%. We compared clinical outcomes of anterograde resection to retrograde resection and retrospectively analyzed risk factors with odds ratio for postprocedural stenosis.

6. Statistical analysis

All statistical analyses were performed using SPSS software version 20.0 (IBM Corp., Armonk, NY, USA) with significance level set at p<0.05. The Student t-test was used for continuous data and chi-square test for categorical data to compare clinical outcomes of procedures. A multivariate logistic regression analysis was performed to analyze risk factors and odds ratio for postprocedural stenosis.

1. Baseline Characteristics and procedure outcomes

The mean age of subjects was 63.1±9.6 years (60.8% male). A total of 77 cases (79.4%) were confined to channel with or without involvement of the antrum or bulb. Among them, 23 cases (23.7%) were confined to the channel, 49 cases (50.5%) were confined from the antrum to channel, and 5 cases (5.2%) were confined from the channel to bulb. The other 20 cases (20.6%) were located from the antrum to bulb through the channel. Forty cases (41.2%) showed H. pylori infection. On histopathologic evaluation, 42 cases (43.3%) were low grade dysplasia, 16 cases (16.5%) were high grade dysplasia, and 34 cases (35.1%) were EGC. Mean tumor size was 14.6±9.2 mm (Table 1).

Table 1 . Clinicopathological Characteristics of Gastric Neoplasm Involving the Pyloric Channel.

Variable	Data (n=97)
Age, mean±SD, yr	63.1±9.6
Sex, No. (%)
Male	59 (60.8)
Female	38 (39.2)
Location, No. (%)
Confined to channel, antrum or bulb	77 (79.4)
Confined to channel	23 (23.7)
Antrum–channel	49 (50.5)
Channel–bulb	5 (5.2)
Antrum–bulb	20 (20.6)
Helicobacter pylori infection, No. (%)	40 (41.2)
Histopathology, No. (%)
Low grade dysplasia	42 (43.3)
High grade dysplasia	16 (16.5)
Early gastric cancer	34 (35.1)
Miscellaneous	5 (5.2)
Tumor size, mean±SD, mm	14.6±9.2

All cases were resected with the ESD technique. Circumference of resection was <25% in 16 cases (16.5%), 25% to 75% in 39 cases (40.2%), and ≥75% in 42 cases (43.3%). Mean procedure time was 38.2±19.6 minutes. Only four cases (4.2%) showed major complications (perforation and major bleeding in two cases, respectively). En bloc resection rate and complete resection rate were 87.6% (85/97) and 83.5% (81/97), respectively. Postprocedure stenosis occurred in 16 cases (16.5%). During 37 months (median, 1 to 118 months) of follow-up period, there was one (1.0%) case of recurrence that was done by anterograde resection. Three cases (3.1%) required additional surgery (Table 2).

Table 2 . Technical Outcomes of Endoscopic Submucosal Dissection in Gastric Neoplasm Involving the Pyloric Channel.

Variable	No. (%)
Procedure type
Endoscopic submucosal dissection	97 (100)
Circumference of resection
<25%	16 (16.5)
25% to <75%	39 (40.2)
≥75%	42 (43.3)
Procedure time, mean±SD, min	38.2±19.6
Immediate complication
Perforation	2 (2.1)
Major bleeding	2 (2.1)
Loss of hemoglobin, mean±SD, g/dL	0.5±0.7
En bloc resection	85 (87.6)
Complete resection	81 (83.5)
Postprocedural stenosis	16 (16.5)
Recurrence rate	1 (1.0)
Additional surgery	3 (3.1)

2. Comparison of anterograde resection and retrograde resection

Anterograde resection was performed in 47 patients (48.5%), and retrograde resection was performed in 50 patients (51.5%). In the retrograde resection group, lesions were more frequently located from the antrum to bulb through the channel (3 [6.4%] vs 17 [34.0%], p=0.001) and tumor size was significantly larger (11.7±7.2 mm vs 17.4±10.0 mm, p=0.002). Circumference of resection ≥75% was significantly frequent in the retrograde resection group (14 [29.8%] vs 28 [56.0%], p<0.001). Procedure time was significantly longer in the retrograde group (30.8±17.0 minutes vs 45.1±19.4 minutes, p<0.001). Other variables including H. pylori infection rate, histopathology, complication rate, en bloc and complete resection rate, recurrence rate, and percentage of additional surgery were not significantly different between the two groups. Postprocedure stenosis occurred in five (10.6%) cases of the anterograde group and 11 (22.0%) cases of the retrograde group, showing no significant difference between the two groups (p=0.132). Among three patients who needed additional surgery, one case (2.1%) was in the antegrade resection group with recurrence of tumor and two cases (4.0%) were in the retrograde resection group (one was due to postprocedure stenosis and the other was due to involvement of deep resection margin by carcinoma) (Table 3).

Table 3 . Comparison of Clinicopathological Characteristics of Patients in Anterograde Resection and Retrograde Resection Groups.

Variable	Anterograde resection (n=47)	Retrograde resection (n=50)	p-value
Age, yr	62.7±9.1	63.5±10.2	0.694
Male sex	27 (57.4)	32 (64.0)	0.509
Location			0.001
Confined to channel, antrum or bulb	44 (93.6)	33 (66.0)
Antrum–bulb	3 (6.4)	17 (34.0)
Helicobacter pylori infection	20 (42.6)	20 (40.0)	0.798
Histopathology			0.471
Low grade dysplasia	20 (42.6)	22 (44.0)
High grade dysplasia	7 (14.9)	9 (18.0)
Early gastric cancer	19 (40.4)	15 (30.0)
Miscellaneous	1 (2.1)	4 (8.0)
Tumor size, mm	11.7±7.2	17.4±10.0	0.002
Circumference of resection			<0.001
<25%	15 (31.9)	1 (2.0)
25% to <75%	18 (38.3)	21 (42.0)
≥75%	14 (29.8)	28 (56.0)
Procedure time, min	30.8±17.0	45.1±19.4	<0.001
Immediate complication
Perforation	1 (2.1)	1 (2.0)	0.965
Major bleeding	1 (2.1)	1 (2.0)	0.976
Loss of hemoglobin, g/dL	0.6±0.7	0.5±0.7	0.905
En bloc resection	41 (87.2)	44 (88.0)	0.909
Complete resection	39 (82.9)	42 (84.0)	0.892
Postprocedural stenosis	5 (10.6)	11 (22.0)	0.132
Recurrence rate	1 (2.1)	0	0.300
Additional surgery	1 (2.1)	2 (4.0)	0.228

Data are presented as mean±SD or number (%)..

3. Multivariate analysis of risk factors for postprocedural stenosis

Multivariate analysis showed that circumference resection ≥75% was the only significant risk factor for postprocedure stenosis (odds ratio, 20.155; 95% confidence interval, 2.105 to 193.000; p=0.009). Location, histopathology, tumor size, procedure time, and piecemeal resection or incomplete resection were not significant risk factors (Table 4).

Table 4 . Multivariate Analysis of Risk Factors for Postprocedural Stenosis.

Variable	Odds ratio (95% CI)	p-value
Location
Confined to channel, antrum or bulb	Reference
Antrum–bulb	2.703 (0.472–15.479)	0.264
Histopathology
Low grade dysplasia	Reference
High grade dysplasia	8.029 (0.664–97.084)	0.101
Early gastric cancer	3.608 (0.377–34.526)	0.265
Tumor size	1.142 (0.973–1.341)	0.103
Circumference of resection
<75%	Reference
≥75%	20.155 (2.105–193.000)	0.009
Procedure time	0.968 (0.920–1.018)	0.204
Piecemeal resection	0.318 (0.011–9.420)	0.507
Incomplete resection	2.433 (0.144–41.018)	0.537

CI, confidence interval..

In this study, we tried to devise a more effective endoscopic method for performing endoscopic resection for EGC and gastric adenoma including pyloric channels and to find out which situations might be closely associated with postprocedural complications such as stenosis.

Conventional anterograde resection is an effective treatment for resection of lesions such as low to mid-body and proximal antrum. However, for GNPC, it is technically difficult to properly access the distal margin. As a result, the curative resection rate may decrease and postprocedural complications such as pyloric stenosis may occur.^20-22 Thus, we assessed GNPC with retrograde resection, a method of accessing the distal margin through the retro-flexion of endoscope in the duodenal bulb, and compared clinical outcomes with those of conventional anterograde resection.

As mentioned in results, the retrograde resection group had larger lesion size with lesions located through the antrum to bulb and resected circumference over 75%. However, there were no significant differences in en bloc resection rate, complete resection rate, recurrence rate, or frequency of postprocedural complication rate. These results suggest that in GNPC, retrograde resection can be an effective method if the lesion has a larger size or if it is located through the antrum to bulb. This might be caused by the easy access to the distal margin through the retro-flexion of the endoscope.^6,13,14 Meanwhile, procedure time was significantly longer in the retrograde resection group, which might be caused by the larger lesion size and difficult assess to the lesion when it is located through antrum and bulb.

Postprocedural stenosis, one of the most concerned postprocedural complications of GNPC, was likely to occur in the case of channel circumference of resection ≥75%. Previous Korean and Japanese studies demonstrated that circumference of resection over 75% were significant risk factors for postprocedural stenosis in ESD of GNPCs in common, which was consistent with our result.^23-26 More frequent follow-ups should be considered and endoscopic balloon dilatation might be required if stenosis occurs.^23,24

This study was meaningful in that it introduced an effective method of resection of GNPC and identified risk factors for predicting postprocedural stenosis, a major complication. However, this study had several limitations. First, in addition to postprocedural stenosis, this study did not identify risk factors for bleeding or perforation. Bleeding and perforation are severe complications that could occur after ESD. However, we had only two cases of perforation and bleeding, respectively, suggesting that ESD in GNPC might be a safe therapeutic modality. Second, this study was based on a retrospective, single-center study. Therefore, several variables such as range of the lesion, tumor size, circumference of resection and procedure time are not fairly matched. Thus, it is premature to generalize our results. Further studies including larger cases performed by a multicenter are needed in the future.

In conclusion, ESD is a feasible method for treating GNPC and retrograde resection method may be effective for larger tumor located throughout the antrum and bulb with circumference of resection ≥75%. For a successful ESD of GNPCs, a systematic therapeutic strategy and appropriate response to complications based on abundant clinical experiences would be necessary.

No potential conflict of interest relevant to this article was reported.

Study design and concept: S.M.K., J.J.P. Performance of endoscopic procedure: J.J.P. Data collection and statistical analysis: all authors. Writing of manuscript: S.M.K., J.J.P., M.K.J. Advice for study design and writing of manuscript: B.J.L., J.J.P., S.W.L., H.J.C. Reading of article and final approval: all authors.