Introduction

Endoscopic submucosal dissection (ESD) is widely accepted for early gastric cancer (EGC), and it can be performed regardless of tumor size, location, or fibrosis [1–11]. According to the Japanese Gastric Cancer Association (JGCA), [12] the absolute indication for curative ESD are well- or moderately differentiated intramucosal adenocarcinoma ≤2 cm in diameter without ulceration and absence of lymph node (LN) metastasis and lymphatic vessel invasion. Expanded indications include well- or moderately differentiated intramucosal adenocarcinoma >2 cm in diameter without ulceration, well- or moderately differentiated adenocarcinoma ≤3 cm in diameter with submucosa invasion <500 μm without ulceration, well- or moderately differentiated adenocarcinoma with ulceration, and undifferentiated intramucosal adenocarcinoma ≤2 cm in diameter without ulceration.

We have possibly seen an increase in the number of ESD for EGC and a corresponding increase in the number of en bloc specimens with positive horizontal margin (HM). However, there is no consensus regarding the management of positive HM. In a retrospective study, we evaluated the specimens after en bloc ESD for intramucosal EGC to identify risk factors and to determine the best management strategy for positive HM in cases of EGC resected by en bloc ESD.

Patients and methods

Patients

The study group comprised 890 consecutive patients with 1,053 intramucosal EGCs resected by en bloc ESD at Hiroshima University Hospital between April 2005 and June 2011 (Fig. 1). We classified the lesions by the HM, whether positive (n = 21) or negative (n = 1,032). The indications for ESD were based on the expectation that the procedure would be curative, as set out in the Japanese classification of gastric carcinoma issued by the JGCA [12]. This study was carried out with approval from the institutional review board of Hiroshima University (No. 658), and all patients provided written informed consent for the ESD procedure and follow-up assessment.

Fig. 1
figure 1

Flowchart showing the 1,053 ESD cases used in the study. ESD endoscopic submucosal dissection, pM pathologic intramucosal carcinoma, pSM pathologic submucosal invasive carcinoma, HM horizontal margin

Full size image

ESD procedure

ESD was performed with a conventional single-channel endoscope (GIF-450RD5, FUJIFILM Medical, Tokyo, Japan; GIF-H260, -H260Z, or -Q260J, Olympus, Tokyo, Japan) or a two-channel endoscope (GIF-450D5, FUJIFILM Medical; or 2TQ260M; Olympus), as reported previously [5, 6, 8, 9, 13]. Between April 2005 and September 2009, we identified the tumor demarcation line (DL) by white light endoscopy and chromoendoscopy. And, starting in October 2009, we used magnifying endoscopy with narrow band imaging (NBI) together with white light endoscopy and chromoendoscopy. When the demarcation line was unclear with endoscopy examination, we had negative biopsy around the lesion. After recognizing the DL, marking dots were placed around the lesion using an argon plasma coagulation probe. When the tumor were well- or moderately differentiated adenocarcinoma, marking dots were made 5 mm from the DL, and when the tumor were undifferentiated adenocarcinoma the marking dots were at least 10 mm from DL. The ESD procedure was performed mainly with the IT knife/IT knife 2 and an electrosurgical generator (ICC 200 or VIO 300D, Erbe, Tubingen, Germany; or ESG 100, Olympus). We used a needle knife for a 1- to 2-mm precut, and we performed a circumferential mucosal incision 5 mm outside the marking with the IT knife/IT Knife2. We performed preventative coagulation for all exposed vessels on the artificial ulcer with hemostatic forceps (FD-410LR, Olympus; or HDB2418W-W, Pentax, Tokyo, Japan) upon completion of ESD [13]. Poor control of bleeding during ESD was defined as bleeding that required multiple coagulations (10 times), as reported previously [13]. In all patients, the presence or absence of submucosal fibrosis was assessed under white light conventional endoscopy and chromoendoscopy. Endoscopic findings were classified according to the system previously reported [8, 14], and based on observations at the time of injection of sodium hyaluronate with indigo carmine: F0, no fibrosis, which manifests as a blue transparent layer; F1, mild fibrosis, which appears as a white web-like structure in the blue submucosal layers; and F2, severe fibrosis, which appears as a white muscular structure without a blue transparent layer in the submucosal layers. The white muscular structure is similar to the muscular layer; therefore, this structure cannot be easily separated from the muscular layer.

Pathologic evaluation of resected specimens

The resected specimen was fixed in formalin solution and cut into 2-mm-thick sections. Histopathologic diagnosis was based on the Japanese Classification of Gastric Carcinoma [15]. Positive HM was defined as a margin directly infiltrated by tumor cells (type A) or the presence of cancerous cells on either side of the 2-mm-thick sections (type B), or tumor margin that could not be evaluated because of crush or burn damage (type C) (Fig. 2).

Fig. 2
figure 2

Types of positive horizontal margins. Type A positive horizontal margin was defined as a margin directly infiltrated by tumor cells. Type B the presence of cancerous cells on either side of the 2-mm-thick sections. Type C tumor margin that could not be evaluated because of crush or burn damage

Full size image

According to the Japanese Gastric Cancer Association (JGCA), [12] the absolute indication for curative ESD are well- or moderately differentiated intramucosal adenocarcinoma ≤2 cm in diameter, no ulceration, and absence of lymph node metastasis and lymphatic vessel invasion. The expanded indications add well- or moderately differentiated intramucosal adenocarcinoma >2 cm in diameter without ulceration, well- or moderately differentiated adenocarcinoma ≤3 cm in diameter and invading the submucosa <500 μm without ulceration, and well- or moderately differentiated adenocarcinoma with ulceration.

Patient follow-up

Follow-up endoscopy after ESD was performed immediately when positive HM was found by pathologic diagnosis, and, if there was no local recurrence, surveillance endoscopy was performed at 3–6 months, at 12 months, and every 12 months thereafter. When the HM was negative, follow-up endoscopy was performed every 12 months. Biopsy of the ESD scar site was always performed, whether or not local recurrent tumor was observed endoscopically.

Outcome measures

Clinicopathological characteristics of the positive HMs, the incidence of positive HM, incidence of local recurrence, risk factors for positive HM, characteristics of local recurrent tumors, and outcomes of treatment for local recurrent tumors were evaluated.

Statistical analysis

Differences between lesions with positive HM and those with a negative HM or between the various types of positive margins were analyzed by χ 2 test with Yates’ correction, by Fisher’s exact test, or Student’s t test. P < 0.05 was considered significant.

Results

Clinical characteristics of the 21 HM-positive lesions are shown in Table 1. Tumor location in the upper third of the stomach, the presence of severe fibrosis, and the ESD specimen not meeting the absolute indication were significantly more frequent among lesions with positive HM than among lesions with a negative HM. Six of the lesions were >2 cm in diameter, and three had ulceration ≤3 cm diameter. All three lesions with non-curative indication were undifferentiated adenocarcinomas >2 cm in diameter. These three lesions were considered as the expanded indication cases prior to ESD (within 2 cm in size); however, pathological findings showed the resected lesion as over 2 cm (non-curative indication). All cases were intramucosal undifferentiated carcinoma without ulceration and vessel involvement.

Table 1 Characteristics of EGCs (pM) treated by ESD
Full size table

Multivariate analysis with logistic regression showed tumor location in the upper third of the stomach and tumor characteristics not meeting the absolute indication to be independent risk factors for positive HM (Table 2).

Table 2 Results of multivariate logistic regression analysis of risk factors for positive horizontal margin
Full size table

EGCs with positive HM are shown per margin type in Table 3. Five lesions had a type A margin, 9 had a type B margin, and 7 had a type C margin. Tumor location in the upper third of the stomach was significantly more frequent in association with type A and C margins than in association with type B margins. Magnifying endoscopy with NBI identified the demarcation line significantly more frequently in lesions with a type A or type B margin than in lesions with a type C margin. No local recurrent tumor was found in association with type C margins, but the difference between types was not significant.

Table 3 Characteristics of EGCs with positive horizontal margin (n = 21), shown per margin type
Full size table

Characteristics of the three HM-positive cases, which occurred the local recurrent tumor, are shown in Table 4. All local recurrent tumors were resected curatively by additional ESD. The times to discovery of local recurrent tumor were 11, 8, and 34 months.

Table 4 Characteristics of the HM-positive cases which occurred the local recurrent tumor
Full size table

Discussion

In the present study, the incidence of positive HM in en bloc specimens of intramucosal adenocarcinoma resected by ESD was 2.0 %, and the incidence of local recurrence in HM-positive cases was 14.3 %. Kakushima et al. [16] reported that the diameter of the tumor, recurrent-type cancer, submucosal cancer, and undifferentiated-type cancer were risk factors for positive HM in cases of EGC, including submucosal carcinoma, resected by ESD. Asada-Hirayama et al. [17] reported that a flat component, a large lesion, and moderately differentiated adenocarcinoma are factors predictive of inaccurate endoscopic assessment of the horizontal extent of intestinal-type EGC. In the present study, we limited the lesions to intramucosal adenocarcinomas resected by en bloc ESD so that we could accurately determine the number of lesions for which curative resection was achieved.

Tumor location in the upper third of the stomach and a lesion not meeting the absolute indication were shown to be risk factors for positive HM. Identification of the demarcation line was more difficult in these lesions. In addition, crush or burn damage occurred most often in these lesions, which can result in an unclear HM. Lesions with direct HM infiltration (type A) or with cancerous cells on either sides of the specimen (type B) were prone to local recurrence. Thus, a type A or type B pathologic diagnosis requires strict follow-up. The majority (71.4 %, 5/7) of lesions with an unclear tumor margin attributed to crush or burn damage (type C) existed in the upper third of the stomach. We think that the technical difficulty of ESD in the upper third of the stomach explains the frequency of such crush or burn damage. We had been practicing ESD approximately 7 years before we began using magnifying endoscopy with NBI, so our ESD technique was advanced, thus reducing the potential for crush and burn. This might explain our relatively low number of type C cases with magnifying endoscopy with NBI. No local recurrent tumor was found in association with a type C HM, so we think that our type C cases were misdiagnosed as HM-positive because of the crush or burn effect. To reduce the number of type C assessments, it will be necessary to devise a marking system that accounts for the crush or burn effect and to further improve the ESD procedure. Crush or burn damage occurred most often in the upper third of the stomach, therefore we think it is better to make marking dots enough away from DL of the lesions in the upper third of the stomach compared with the lesions in the other locations to reduce the type C HM cases.

There are various types of HM positivity so specimens that are HM-positive should be further evaluated according to the type-based risk of local recurrence.

ESD was developed for the purpose of reducing the incidence of local recurrence associated with endoscopic mucosal resection (EMR), [18–21], but local recurrence remains problematic. Takenaka et al. [22] identified tumor size >30 mm, tumor location in the upper third of the stomach, and incomplete resection as factors significantly associated with local recurrence. In the present study, the incidence of local recurrence was significantly higher in cases of positive HM than cases of a negative HM.

In cases of local recurrence, severe fibrosis after endoscopic resection makes the depth of invasion difficult to determine, [23, 24] and there is a high risk of perforation in such cases [3, 25, 26], so the usefulness of laparoscopic surgery for local recurrence has been reported [27–29]. However, we have reported that ESD, which is far less invasive than surgical resection, can be performed safely for local recurrence after EMR [6] or ESD [8], and we believe that even recurrent intramucosal carcinoma after ESD for intramucosal EGC is an indication for re-ESD. However, when a local recurrent tumor proves to be a submucosal invasive carcinoma or undifferentiated adenocarcinoma, surgical resection should be considered.

In conclusion, to reduce the incidence of positive HMs, it will be important to improve (1) the diagnostic performance of magnifying endoscopy in identifying the demarcation line, (2) the ESD procedure itself, and (3) the specimen handling technology. When positive HM is found, it is important to consider the margin type and to conduct appropriate follow-up. Instead of performing surgical resection as soon as positive HM is found, we should perform strict surveillance and consider whether re-ESD is possible if the local recurrence proves to be a differentiated intramucosal carcinoma. In addition, a prospective randomized controlled trial that involves many patients is needed so that enough data can be obtained to set appropriate surveillance intervals.