Several EMR techniques have been developed for excision of mucosal based lesions; the most commonly used are the cap-assisted technique (ER-Cap) and the multi-band ligation assisted technique (MBM). No significant differences in the safety and efficacy profiles have been reported between these two approaches. The only observed difference was the maximum diameter of the resected specimens, where the ER-cap method was favoured[34-36]. In both modalities, after marking 2 mm away from the lesion margins and lifting with saline solution injection, the targeted area is suctioned into the cap and grasped by the snare or by releasing the rubber band to create a pseudopolyp. The lesion is then cut using a snare with blended-current electrocautery. If the MBM technique is performed, the procedure can be carried out safely with no prior submucosal injection and lifting[37-40].

There is extensive experience of performing focal EMR for treatment of macroscopically visible lesions arising in BE. The available data show complete regression of neoplasia in 97%-100% of cases and 5-year survival rates of 98%-100%[41-54] (Table 1). EMR is the only endoscopic technique that has proved increasing the 5-year survival rate in Barrett’s patients in uncontrolled trials[41]. In addition, endoscopic resection has been demonstrated to be a highly safe technique. Alvarez-Herrero and colleagues reporting the outcome of more than 1000 EMR procedures performed in 243 patients, observed an acute bleeding rate was 2.9% and delayed bleeding rate of 2.1%, with no perforations and successfully management of all adverse events by an endoscopic approach[38].

The radical differences between treatments for T1m and T1sm tumors make a definitive histopathological staging essential, in order to identify the patients amenable for curative endoscopic therapy. There are several concerns about the ability of conventional biopsy specimens to provide an accurate histological diagnosis. The sampling error associated with the random biopsy protocol is well known and there are also important doubts about the adequacy of the depth of specimen obtained with conventional biopsy forceps. Published studies have reported a limited reproducibility, particularly for dysplasia, as well as low inter-observer agreement rates. Rates are between 61% and 75% when three categories are evaluated (no dysplasia, indefinite for dysplasia/LGD and HGD/carcinoma), but go down to κ value of 0.49 when HGD is diagnosed separately from carcinoma[55].

A recent study performed in two tertiary referral centers, has demonstrated a higher inter-observer agreement for diagnosis of dysplasia from the analysis of EMR specimens than from conventional biopsies (κ 0.33 vs 0.22, P < 0.001 for LGD; 0.43 vs 0.35, P = 0.018 for HGD). Submucosa was present in up to 88% of EMR specimens but only in 1% of biopsy samples and the presence of muscularis mucosae was observed only in 58% of biopsy specimens[56]. EMR samples permit an accurate evaluation of depth and lateral resection margins and also provide information about the presence of submucosal involvement. The histological examination of EMR pieces can also asses the degree of lymph and blood vessel invasion, important risk factors for the presence of LN metastasis[57-59]. Different studies have shown that final staging by EMR modifies the previous diagnosis in up to 48% of cases[54,59,60] and dramatically changes the clinical management of these patients (Table 2). Similar discrepancy rates have been reported for gastrointestinal neoplasia from other locations[61]. Finally, EMR staging has shown to be consistent with surgical pathology staging. The presence of free of disease margins in EMR samples, directly correlates with the absence of residual tumor at esophagectomy[62].

The major drawback of using focal EMR as the only treatment for Barrett’s neoplasia is the possible development of synchronous, metachronous and recurrent lesions, arising in the residual Barrett’s epithelium. After a mean follow-up period of 3 years, the reported incidence rates range between 11% and 47% and are even higher with longer follow-up. Because of this, complete Barrett’s resection has been proposed as an alternative treatment[41-54].

The rationale for radical endoscopic resection of BE is the proven coexistence of multifocal HGD in Barrett’s mucosa, the aforementioned high rate of synchronous and metachronous lesions when focal EMR is performed as single treatment and the lack of histological correlation of the non-tissue acquiring ablative techniques[29,30,63]. With this approach, the whole Barrett’s segment is eradicated by endoscopic resection in a single or multiple sessions, achieving the treatment of any occult neoplasia and preventing the development of any new lesion during follow-up[60]. It was firstly described by Satodate et al[64], and since then, several studies have been conducted involving a total of 390 patients with HGD or IMC[54,60,63,65-69] and achieving complete eradication of IM in 86% to 100% of cases and eradication of any neoplasia from 75% to 100% of patients (Table 3). The global recurrence rate of neoplasia after a follow-up period of up to 32 mo was 3% (12/390), much lower than the previously reported with focal EMR[54,60,63,65-69].

Only one case of disease progression was observed (0.25%) with this approach. In the largest published series, Pouw et al[63] reported one case of progression to T1sm1 tumor after complete removal of a T1m2 cancer, the subsequent surgery showed neither residual tumor nor LN involvement. In the same study, all cases with recurrence of neoplasia [3 patients (1.8%), two HGD and one of T1sm1 tumor] were found distally to the neo-esophagogastric junction. This finding highlights the recommendation of a careful inspection of this area[63]. The complete Barrett’s eradication EMR (CBE-EMR) is a safe procedure when performed by expert endoscopists and complications are successfully treated by an endoscopic approach with no need of additional surgery in most of cases (Table 4).

The major limitation for CBE-EMR is the high incidence of symptomatic stenosis, with rates reaching 50% in some reports. The rate of esophageal stricture was related to the length of Barrett’s resected segment[63] and significant statistical differences were found with regard to the number of EMR procedures between patients who did and did not develop strictures[60]. New strategies to prevent the development of strictures should be evaluated[63]. A recent study reports a decrease in the incidence and severity of stricture after EMR/endoscopic submucosal dissection (ESD) involving more than 75% of circumference when preventive dilation is performed. Endoscopic balloon dilation was carried out 1 wk after treatment and once a week thereafter, until the mucosal defect was healed. No complications related to endoscopic dilation were observed[70]. Despite the relative low number of patients enrolled in these studies and the short follow-up period, CBE-EMR has shown excellent endoscopic and histological short-term results and could be considered as an alternative to esophagectomy in high volume centers for selected patients with short Barrett’s segment (≤ 5 cm)[60,63].

ESD is regarded in Japan as the treatment of choice for intramucosal gastric neoplasias, and when performed by experts the results for esophageal and colonic lesions are encouraging and superior to conventional EMR in terms of curative resection rate and recurrence[71-73]. With this approach, en bloc resection can be achieved regardless of the size of the lesions but it is a challenging technique, time consuming and is associated with a higher rate of adverse events[72,74,75].

The first step is marking the targeted lesion 5 mm away from its limits and perform submucosal injection using any of the available solutions (saline solution, hyaluronic acid, glycerine). The addition of epinephrine (1:100 000-1:300 000) is used for vasoconstriction of small submucosal vessels and indigo-carmine for a better visualization of the stained submucosal layer. Incision at the proximal and distal margins and then circumferential cutting of the surrounding mucosa is performed. Finally, dissection of the tissue beneath the isolated mucosa is carried out to achieve the removal of the lesion in one piece. Many different knives have been designed and developed.

Because of the low incidence of BE and adenocarcinoma in Japan and other eastern countries, the reported experience with early esophageal neoplasia is mainly limited to squamous cancer[75-81] (Table 5). Yoshinaga et al[82] reported a 100% of en bloc resection and a curative resection rate of up to 72% in adenocarcinoma located at the esophagogastric junction. When compared to EMR, ESD shows a better en bloc resection rate and a better curative resection rate (free of disease resection margins) for treatment of superficial tumors in the gastrointestinal tract, leading to a dramatically reduced local recurrence rate[75,83,84]. Perforation and bleeding were significant higher in the ESD group, although most of them were successfully managed by endoscopic intervention. There were no studies from Western countries and no randomized controlled trials included in this analysis[83-87] (Table 6).

It is important to keep in mind that ESD is a time-consuming and technically demanding procedure. Learning methods should be standardized, with animal models playing a significant role[88-90] and the technique should be performed in an appropriate stepwise fashion. The minimum training requirements recommended by a panel of experts were recently published: enough previous experience with conventional EMR; knowledge of indications, instruments and complications management; visits to expert centers and observation of at least 15 live procedures performed by the experts; hands-on experience in isolated animal models and live pigs; starting with treatments on less challenging locations such as rectum and then moving to distal stomach, colon, proximal stomach and esophagus[73]. There is no consensus regarding of the minimum case load, but Japanese experts recommend at least 50 ESD procedures in distal stomach before performing the technique in the more challenging locations[91,92]. The role of ESD in the therapeutic algorithm of BE in the western countries is still not established[87,93]. Long-term results with EMR techniques are excellent, as previously shown, and ESD is a challenging technique with an increased risk of perforation compared to EMR and it does not provide a high R0 rate (lateral resection margins free of tumor) in Barrett’s lesions. In these patients, the entire Barrett’s segment must be eradicated after resection of any visible lesion regardless of the negative resection margins. Thus, the potential advantages of ESD compared to conventional EMR could be less relevant in treatment of early Barrett’s neoplasia[93].

Radiofrequency ablation (RFA) using the HALO^® system (BÂRRX Medical Inc., Sunnyvale, California, United States) uniformly delivers thermal energy with a prefixed density (12-15 J/cm²) and power (40 W/cm²). With these settings, the tissue penetration depth of the RF energy is limited to 500-1000 μm, which has been demonstrated as sufficient for the successful ablation of esophageal epithelium with no submucosal injury in animal models and humans[94]. The HALO360^® device is a balloon catheter with spindle-shaped electrodes on its surface that allows the ablation of 3 cm long segments in a circumferential fashion. In order to choose the appropriate balloon size (available diameters 18 mm, 22 mm, 25 mm, 28 mm, 31 mm and 34 mm) an inflatable sizing balloon is used to measure the esophageal inner diameter. The catheter is introduced into the esophagus over a guide-wire and the RFA is performed under endoscopic direct view. The HALO90^® is a square-shaped catheter with the same electrodes on its external surface, which is attached to the tip of the endoscope. It allows the focal ablation of small areas of residual Barrett’s epithelium[29,95,96].

The available data from prospective trials are summarized in Table 7; they show a complete eradication of dysplasia in 70%-100% of cases and the eradication of IM in 50% to 100%[97-111]. Several trials assessing the efficacy and safety of RFA in BE have been conducted. After the publication of several studies in non dysplastic BE[97-99] Ganz et colleagues in 2008 conducted the first study in patients with HGD. A complete regression of IM, any dysplasia and HGD was achieved respectively in 54%, 80% and 90% of the 142 enrolled patients[100]. In the only multicenter, randomized and sham-controlled trial conducted to date, 127 patients with prior diagnosis of dysplastic BE (63 HGD and 64 LGD) were randomized in a 2:1 ratio to receive either RFA or sham endoscopic procedure (control group). After 1-year follow-up, all measured primary and secondary outcomes showed significant differences favoring the treatment group: progression rate, progression rate to cancer, complete regression of IM, complete regression of LGD and complete regression of HGD[112] (Table 8). Only three relevant adverse events occurred in the treatment group and five patients (6%) developed esophageal stricture (with or without dysphagia), a rate markedly lower than reported with resection therapies[112].

A systematic review of nine observational studies, involving 429 patients, and at least 12 mo of follow-up was recently published[113]. After analysis, complete eradication of IM was achieved in 46%-100% of patients and complete regression of neoplasia in 46%-100%. There were only 6 cases of stenosis after treatment (1.4%) and no major complications were observed. RFA has proved to be a safe procedure. Of all 657 patients involved in the aforementioned trials, only one case of perforation has been reported (0.15%) and only 3 patients required hospitalization for any complication related to ablation. The global rate of stenosis is 2.3%, with all instances successfully treated by endoscopic dilation, and the most frequent adverse event is chest pain, usually controlled with conventional analgesics[97-111].

The Amsterdam group has reported excellent outcomes from the stepwise treatment in patients with HGD and any visible lesion in the index endoscopy exam. This approach consists of the resection of all macroscopic lesions and the subsequent ablation by RFA of the remaining Barrett’s epithelium. Initially, circumferential ablation with HALO360^® is performed with a maximum of three sessions; thereafter focal ablation with a HALO90^® device is performed in order to eradicate any residual IM, with the same three session limit[102,103,114-116]. Complete eradication of neoplasia is achieved in up to 100% of cases and complete regression of IM in 96%. Escape EMR is performed if any abnormality is seen during follow-up. No recurrence of neoplasia has been observed 22 mo after treatment[114-116] (Table 9).

One of the most relevant concerns about RFA and other ablative therapies is the incidence of buried IM after treatment. The aforementioned review revealed only one case of buried Barrett’s epithelium, after the assessment of more than 8500 biopsy samples obtained during follow up from the 429 patients enrolled in the 9 analyzed trials[113]. No randomized controlled trials comparing RFA vs CBE-EMR have been conducted to date. According to the published data, stepwise treatment should be the treatment of choice for patients with visible lesions arising on HGD[117] and CBE-EMR could be recommended, in high volume centers, for patients with short segment BE (SSBE).

In this technique, ablation is achieved by light activation of a photosensitizer drug, which leads to oxygen radicals formation and thereafter cell death. The photosensitizing agent, usually porfimer sodium, is administered before the procedure and it is selectively accumulated in the malignant esophageal mucosa. Cylindrical or balloon-based diffuser fibers are then placed over the targeted lesion under endoscopic view[20,29]. The published trials have proved the efficacy of photodynamic therapy (PDT) in eradicating Barrett’s dysplasia[118-123] (Table 10). The only randomized and controlled trial reported complete regression of IM in 52% of cases and complete regression of any dysplasia in 59% out of 138 patients with dysplastic BE[122].

The major drawback of PDT is the relatively high rate of reported adverse events, mainly photosensitivity and symptomatic strictures, which have been reported in up to 36% of patients. Number of PDT treatments per session, prior EMR and a previous history of esophageal stenosis are associated with development of strictures[124]. Buried glands under the neo-squamous epithelium after PDT have been described in up to 51% of patients[125] and cases of adenocarcinoma arising from buried Barrett’s glands have also been reported[120]. For all these reasons, PDT has been abandoned in recent years in favour of other ablation techniques. Further investigations aimed to identify biomarkers, which may stratify the patients more likely to respond to this treatment, and the development new photosensitizing agents could improve its safety profile[29].

This is the latest added option to the therapeutical armamentarium of BE and has shown promising results in the available reports. For ablation, a liquid cryogen is focally sprayed onto the targeted lesion and results in freezing of the epithelium, causing intracellular disruption and ischemia. Liquid nitrogen and carbon dioxide have been used as cryogenic agents. The depth of ablation is limited to

2 mm and treatment sessions are performed every 4-6 wk until complete remission of the IM is achieved[126]. Several trials have shown cryotherapy as a safe and effective tool[127-130]. Short-term results are promising with eradication rates of IM in 46%-78% and of dysplasia between 79% and 87% of cases (Table 11). No major complications have been reported except for a gastric perforation in one patient with Marfan syndrome. This therapy is now contraindicated in patients with limited distensibility of the stomach. Multi-center randomized trials are required to confirm these results and determine the long-term response. It is still necessary to establish the optimal treatment protocol, duration and number of cycles per session, and frequency of treatment sessions. Finally, it remains to be determined if there is any clinical relevant difference in safety or efficacy profiles between CO₂ and N₂[126].

Argon plasma coagulation has reported eradication rates of IM in non dysplastic Barrett’s patients of up to 100%[131-134] and about 75% in cases of HGD, although with significant long term recurrence rates[29,135]. Techniques such as multipolar electrocoagulation and laser therapies have been replaced by the ablation modalities discussed in this manuscript[20,29].