Endoscopic Submucosal Dissection for Adenocarcinomas of the Esophagogastric Junction Free

The incidence of adenocarcinoma of the esophagus and esophagogastric junction (EGJ), particularly Barrett’s adenocarcinoma, has been dramatically increasing in Western countries [1‒3]. On the other hand, esophageal adenocarcinoma is rare, accounting for 1–4% of esophageal cancer in Eastern countries including Japan [2, 4, 5]. However, it is believed that the incidence of adenocarcinoma of the EGJ will increase in Asia in the future due to a decreasing incidence of Helicobacter pylori infection [2, 3].

Endoscopic submucosal dissection (ESD) has been accepted as a minimally invasive and curative treatment for superficial gastrointestinal cancers, including esophageal, gastric, and colonic cancers [6]. ESD allows higher en bloc and complete resection rates than conventional endoscopic mucosal resection (EMR), regardless of tumor size. This results in precise pathological assessment, such as that of invasion depth, lymphovascular involvement, or cancer margin. A few reports have described the efficacy of ESD for EGJ cancers, but some problems remain in this area [7‒10].

Since most cases of Barrett’s adenocarcinoma arise from short segment Barrett’s esophagus in Japan, it is difficult to distinguish Barrett’s adenocarcinoma from gastric cardiac adenocarcinoma endoscopically and histologically in patients with type II EGJ cancers (i.e., tumors invading the EGJ, in which the center is located between 1 cm above and 2 cm below the EGJ; Fig. 1a–c) [3, 5, 11]. Although both types exhibit different potentials for lymph-node metastasis, most studies on EGJ adenocarcinoma have conflated the two [10]. In addition, the curative resection criteria of ESD differ between esophageal cancer and gastric cancer, since the incidence of lymph node metastasis is different [12, 13]. It is unknown which criteria is better for EGJ adenocarcinoma, since the incidence of metastasis of EGJ adenocarcinoma is unclear.

According to the Nishi classification in the guidelines of Japanese Classification of Esophageal Cancer, EGJ cancer was defined as a cancer located 2 cm above and 2 cm below the EGJ. This classification includes not only adenocarcinoma but also squamous cell carcinoma [9, 13]. On the other hand, Siewert type II tumors, located 1 cm above and 2 cm below the EGJ, are considered true cardiac adenocarcinomas arising from the EGJ epithelium [2, 9, 11, 13]. Barrett’s esophagus is defined as columnar epithelium that is replaced from stratified squamous epithelium of distal esophagus [1, 3, 13]. Two endoscopic findings of EGJ were reported: upper limit of the gastric fold and the end of the lower esophageal palisade vessels [9, 13]. Barrett’s adenocarcinoma is diagnosed when cancer is located on the oral side of EGJ because Barrett’s adenocarcinoma is defined as cancer arising from Barrett’s esophagus (Fig. 2). However, it is sometimes difficult to detect these findings in preoperative observation endoscopically due to reflux esophagitis or atrophic gastritis. Therefore, the following histological criteria are used to diagnose Barrett’s esophagus and Barrett’s adenocarcinoma after endoscopic resection, the presence of esophageal glands, duplication of the muscularis mucosae under the lesions, or a squamous island within the lesions according to the Japanese Classification of Esophageal Cancer (Fig. 1d–f) [7, 9, 11, 13]. Gastric cardiac adenocarcinoma was defined as cancer that did not have these characteristics (Fig. 2). However, these histological criteria are sometimes missing. In addition, when cancer is located on just the EGJ, it is difficult to diagnose whether it is Barrett’s adenocarcinoma or gastric cardiac adenocarcinoma (Fig. 2). Therefore, most studies on EGJ adenocarcinoma have conflated the 2 adenocarcinomas [10].

Detecting early EGJ adenocarcinoma is sometimes difficult because of the location of the growth and the subtle differences between the affected tissue and the surrounding mucosa [2, 5]. Patients are advised to inspire deeply, which reduces intrathoracic pressure and facilitates observation of the EGJ by distention of the lumen (Fig. 1a–c) [14]. Therefore, deep sedation is not recommended for observation of EGJ. In addition, because most EGJ adenocarcinoma types, including Barrett’s adenocarcinoma, develop between the 12 and 3 o’clock position from the esophageal side, it is important to pay attention to such a region [8, 14, 15]. While narrow-band imaging (NBI) is useful for detecting esophageal squamous cell carcinoma [16], its effectiveness for EGJ adenocarcinoma is unknown. However, chromoendoscopy with indigocarmine, acetic acid splaying, and NBI may sometimes be useful for detecting such lesions [5, 14].

Delineating the cancer margin is important for complete resection with a histologically cancer-free margin by ESD. Magnifying endoscopy with NBI is useful for differential diagnosis and delineating the cancer margin in early gastric cancer [17]. Assessing the demarcation line for the presence of an irregular microvascular pattern or irregular microsurface pattern using a systematic, but simple, classification system (the VS classification system) may also be useful for differential diagnosis and for delineating the cancer margin in EGJ adenocarcinoma. Some studies have reported a high complete resection rate in patients without a positively detected cancerous margin using magnifying endoscopy with NBI [7, 8, 10, 11].

Subsquamous carcinoma extension, which is EGJ adenocarcinoma invasion beneath the normal esophageal squamous epithelium, is a clinical complication in ESD for EGJ adenocarcinoma [2, 4, 5]. This is because cancer margins on the oral side are difficult to detect when subsquamous carcinoma extension is not visible by endoscopy. Using white light endoscopy, subsquamous carcinoma extension in 97 of 175 (55%) Barrett’s adenocarcinomas was accurately diagnosed, whereas NBI diagnosed subsquamous carcinoma in 42 of 70 (60%) cases [4]. Several attempts to detect subsquamous carcinoma extension in EGJ cancers have been reported. Acetic acid-spraying and magnifying endoscopy with NBI have been used to detect small white signs, which represent cancerous grand openings toward the surface [18]. Furthermore, all of the 6 patients could be diagnosed with subsquamous carcinoma extension using such findings. Optical coherence tomography might be useful even for detecting subsquamous carcinoma extension [19]. Conversely, since most cases of subsquamous carcinoma extension were reported to be less than 1 cm in size, the oral safety margin that is placed 1 cm from the squamocolumnar junction is useful for ESD of EGJ adenocarcinoma resection with the negative oral side horizontal margin including the subsquamous carcinoma extension [11].

The indication of endoscopic resection for superficial EGJ adenocarcinoma, including Barrett’s adenocarcinoma, has not been established because the incidence of lymph node metastasis of EGJ adenocarcinoma is unclear. Several retrospective studies on ESD for EGJ adenocarcinoma have suggested feasible long-term outcomes using curative criteria based on gastric cancer guidelines (Table 1) [10]. However, EGJ cancers, in particular Barrett’s adenocarcinoma, may have a higher potential of lymph node metastasis than gastric cancer. The incidence of lymph node metastasis in Barrett’s adenocarcinoma was reported to be 1.3% in patients with mucosal cancer and 22% in patients with submucosal invasion [20]. A systematic review of surgical resections showed that 26 of 1,350 patients (1.93%) with mucosal Barrett’s adenocarcinoma had lymph node metastasis [21]. It is well known that differentiated mucosal cancer in early gastric cancer without lymphovascular involvement and ulceration exhibits almost no metastasis [12, 22]. Osumi et al. [9] suggested that Barrett’s adenocarcinoma has a higher malignant potential but exhibits similar outcomes to gastric cardiac cancer. In addition, an improved 5-year survival rate was observed in ESD for Barrett’s adenocarcinoma than in ESD for squamous cell carcinoma (88.9 vs. 75.9%), though a statistical difference was not reported [23]. Recently, a multicenter retrospective study of esophageal adenocarcinoma, including EGJ adenocarcinoma, using surgical and endoscopic resection data demonstrated that mucosal and submucosal cancer invasion within 500 μm from the muscularis mucosa without lymphovascular involvement, a poorly differentiated component, and a size of over 3 cm were not associated with metastasis [24]. However, the sample size associated with submucosal lesions was not large enough. Therefore, a prospective multicenter large study would be helpful to confirm these criteria.

Compared to esophageal squamous cell carcinoma, modified neuroleptanalgesia with midazolam is useful for sedation during ESD for EGJ adenocarcinoma because there are fewer patients who had history of heavy alcohol intake [25]. Although it has been suggested that it is technically difficult to complete the ESD procedure, favorable en bloc resection rates of ESD for EGJ cancers have been reported in a meta-analysis of 6 retrospective studies (overall rate: 98.6%; 95% CI 95.9–99.6) [10].

We previously showed that ESD for Barrett’s adenocarcinoma was significantly faster than that for gastric cardiac adenocarcinoma [11]. That may depend on the tumor location, since gastric cardiac adenocarcinoma extends beyond the cardia, including the angle of His, which makes it difficult for the endoknife to attach to the mucosa and the submucosal layer, while Barrett’s adenocarcinoma does not.

The rate of complete resection, which was defined as en bloc resection with histologically cancer-free margins, and curative resection of EGJ cancers according to the Japanese Classification of Gastric Carcinoma was reported to be 87.0% (95% CI 79.7–92.0) and 74.6% (95% CI 69.8–78.9) respectively [10]. The curative resection rate seems to be lower than that in ESD for gastric cancer. Several causes have been considered. Gong et al. [26] speculated that this outcome was dependent on the incomplete resection due to technical difficulties in performing ESD near the EGJ, such as narrow and curved space of the tubular esophagus flares. In addition, the difficulty in determining invasion depth also depends on this narrow space. Osumi et al. [9] suggested that Barrett’s adenocarcinoma exhibited a higher malignant potential, which contributed to a lower curative resection rate than for gastric cardiac adenocarcinoma, given the higher incidence of submucosal invasion and lymphovascular invasion in Barrett’s adenocarcinoma. A meta-analysis of EGJ adenocarcinoma demonstrated that over 20% of lesions have been reported to be deep submucosal invasions [10]. Hoteya et al. [7] also described that the incidence of submucosal invasion and lymphovascular invasion in EGJ adenocarcinoma was higher than that of gastric cancer, and those findings suggested that EGJ adenocarcinoma can be highly malignant. Furthermore, the curative resection rate was 48.0% for Barrett’s adenocarcinoma, and 36% of Barrett’s adenocarcinomas had positive lateral margins due to subsquamous carcinoma extension [7]. We previously reported that 1 cm safety margins were useful for the higher rates of curative resection in cases of both undetectable and detectable subsquamous carcinoma extension based on preoperative endoscopy because all cases of subsquamous carcinoma extension were less than 1 cm in size [11].

Favorable long-term results were also reported in a meta-analysis of 6 retrospective studies (Table 1) [10]. No local or distant metastases were observed in 269 patients who met the curative resection criteria according to the Japanese Classification of Gastric Carcinoma, but one study used curative resection criteria for esophageal carcinoma. On the other hand, of the 90 patients who underwent noncurative resection, 3 (3.3%) had local recurrence and 2 (2.2%) had distant metastasis. However, the study was limited by a small number of patients and a short observational period of 3 years. A larger study would be helpful in eliminating the long-term outcomes. One retrospective comparative study involving surgery demonstrated similar 5-year overall survival rates during the ESD and surgery during a long-term follow-up period (93.9 vs. 97.3%) [26]. No local recurrence and disease-specific death were observed.

No severe adverse events associated with ESD for EGJ cancers were reported [9‒11]. Although it was believed to be technically difficult, similar results were reported in patients with gastric and esophageal ESD. Post-ESD bleeding, perforation, and stricture occurred in 3.4% (95% CI 1.7–6.7), 2.5% (95% CI 1.3–5.0), and 6.9% (95% CI 3.2–14.0), respectively, in a meta-analysis of 6 retrospective studies [10]. Most cases were controlled with endoscopic hemostasis, clipping, or balloon dilation. In a comparative study with surgery, adverse events in the ESD group (10.0%) were similar to those in the surgery group (17.9%, p = 0.308) [26]. In that study, 1 patient required emergency surgery due to uncontrolled bleeding in the ESD group, whereas 1 patient required surgery for ileus caused by internal herniation in the surgery group. Hospital stay in the ESD group was significantly shorter than that in the surgery group.

Surgical resection, including total gastrectomy with transhiatal resection and esophagectomy, has been performed for EGJ cancers [2, 3, 27]. Long-term outcomes for superficial cancers have been reported to be favorable. Recently, an algorithm for the extent of lymphadenectomy for EGJ cancer ≤4 cm in size, including squamous cell carcinoma, was constructed based on tumor location, histology, and T-categories [2, 12]. Surgical resection, especially esophagectomy, has a high mortality rate that often exceeds 2%, with substantial morbidity and no guarantee of curing metastasis [28]. In addition, Fukunaga et al. [29] suggested that with or without comorbidities, the difference of surgical technique contributed to long-term survival in patients treated with surgical resection for early gastric cancer. Furthermore, more invasive surgery results in a poorer quality of life, which may influence mortality. Since endoscopic therapy is less invasive and generally improves the quality of life after surgery, both the incidence of metastasis and the will of the patient are critical for determining whether endoscopic therapy should be administered. Recently, perigastric lymph-nodes in the lower half of the stomach, that is, 4d-6 lymph-nodes, were reported to be of no benefit to dissect [2, 12, 27]. Therefore, we should avoid dissecting 4d-6 lymph-nodes and proximal gastrectomy might be recommended. Hopefully, this would improve the clinical outcomes compared with total gastrectomy.

EMR and a combination therapy of EMR and RFA are mainly applied for the treatment of adenocarcinomas in long segment Barrett’s esophagus, since Barrett’s esophagus is associated with high incidence rates of recurrent or metachronous cancers [1, 30]. On the other hand, metachronous cancer in Barrett’s esophagus is rare in Japan because most Barrett’s adenocarcinoma cases arise from the short segment of Barrett’s esophagus. In addition, ESD results in higher complete resection and allows precise pathological assessment, thereby resulting in a lower recurrence rate in Japan compared with western countries. It depends on high accurate diagnosis to delineate cancerous lesion and high quality of ESD technique without positive margin.

ESD may be a good first indication for superficial EGJ adenocarcinoma, including Barrett’s adenocarcinoma. Additional information on the incidence of metastasis would help confirm the indication of ESD for EGJ adenocarcinoma.

The authors declare no conflicts of interest regarding this review article.