|Home | About | Journals | Submit | Contact Us | Français|
Endoluminal treatment of Barrett's esophagus has become the preferred option for initial intervention for advanced neoplasia without invasive carcinoma. Data from abstracts presented at Digestive Disease Week 2008 provide greater insight into optimal use of existing techniques and an early look at potential next generation therapies. Results from the AIM Dysplasia trial describe a larger study with longer post-treatment surveillance highlighting the efficacy and tolerability of radiofrequency ablation, while early results from liquid nitrogen cryotherapy studies suggest a potential to obtain similar eradication results with very high tolerability. Endoscopic resection, despite its risks, remains a popular option for focal as well as more widespread resection of Barrett's mucosa. Additional abstracts highlight novel approaches to ablation and resection. Enhanced imaging techniques and molecular marker analysis also appear to improve treatment outcomes. However, time and further studies of combined approaches to diagnosis and eradication are necessary to optimize treatment algorithms.
The rapid rise in the incidence of Barrett's esophagus over the past three decades has generated a great deal of interest regarding the optimal approach to management of this condition, where specialized intestinal metaplasia (SIM) with malignant potential replaces the normal esophageal squamous epithelium. With discovery of and subsequent improvements to endoluminal techniques for the removal of SIM, attention has focused on how and when to select an endolum-inal therapy to eradicate the Barrett's esophagus. Previously [Smith and Lightdale, 2008], we described the variety of endoluminal alternatives available for treatment of Barrett's esophagus. In this article, and in advance of Digestive Disease Week 2009, we discuss some of the over 200 related abstracts presented at DDW 2008, with a particular focus on those which supplement our previous review of the use of endoscopic-based therapies for SIM.
As with our previous article, we have separated our discussion by type of endoluminal therapy. It is important to remember, however, that for many Barrett's specialists, selection of a treatment often does not involve utilization of a single modality, rather a combined approach which is optimized given individual patient findings on endoscopy and, potentially, other imaging. In addition, other factors such as comorbidities and the ability to tolerate certain procedures may also affect choice of therapy. Therefore, what follows is a discussion of key abstracts across the spectrum of endoluminal treatment options, not just an update on those which have received the most attention.
The use of multipolar electrocautery (MPEC), argon plasma coagulation (APC), lasers and other early thermal therapies has, to a large degree, fallen out of favor. To effectively treat anything other than a small, focal area of Barrett's esophagus is a labor-intensive process, and there is still a risk of leaving residual Barrett's mucosa after treatment. While higher power APC decreases the risk of residual and subsquamous intestinal metaplasia (SSIM) persisting after treatment, it increases the risk of stricture formation due to a greater depth of tissue destruction. However, these techniques are well known to many gastroenterologists, whose endoscopy rooms already contain the required equipment.
One particular study by Wani et al.  followed 39 patients treated for eradication of Barrett's esophagus using MPEC, APC or a combination of both therapies. This cohort, whose median Barrett's length was 3 cm, consisted of 85% nondysplastic Barrett's esophagus (NDBE) and 15% low-grade dysplasia (LGD). All patients followed for at least 4 years after treatment with surveillance endoscopy and biopsies whose timing was appropriate to histologic Barrett's grade. Findings from the last endoscopy for each patient demonstrated that 31 of these patients achieved complete endoscopic replacement of Barrett's epithelium with squamous mucosa, while thirty (77%) of these patients demonstrated both histologic and endoscopic resolution of SIM. Of the nine patients who displayed persistent SIM, all biopsies showed NDBE, though three patients had SSIM on biopsies. No patient developed dysplasia or carcinoma during this period, leading the authors to conclude that reversal of Barrett's esophagus can persist in about three-quarters of patients who are treated with APC and MPEC. This degree of success does not reach the level of recent studies using radiofrequency ablation (RFA), but the longer follow-up period of older technologies increases confidence in the durability of the SIM eradication.
Another abstract suggests that use of a carbon dioxide (CO2) laser could improve success rates while minimizing the undesired consequences of these thermal therapies [Anandasabapathy et al. 2008]. The authors note that CO2 lasers have a predictable penetration depth and minimal collateral damage due to efficient absorption of laser energy by the water already present in esophageal cells. Using a porcine model, the safety and efficacy of a photonic band gap CO2 laser (OmniGuide, Inc., Cambridge, MA) were tested. Results showed complete transepithelial ablation in 96% of the treated area, with 100% coverage if a second treatment was applied. Only one pig showed focal esophageal injury to the superficial muscularis propria, thought to be from wedging of the laser against the esophageal wall. However, no strictures or other injuries, including transmural damage, were identified. While use of this device is clearly in the early stages of exploration, the high rates of successful ablation are comparable to currently existing devices. Further studies will be needed to determine whether this technique can translate to human use with preservation of efficacy and tolerability, and without being time and labor-intensive like its predecessors.
In general, presentations during DDW 2008 continued to demonstrate a paradigm shift away from photodynamic therapy (PDT) as a primary ablative option for Barrett's esophagus in favor of other techniques such as RFA. However, one PDT-based study searched for factors associated with recurrence of neoplasia after ablation [Badreddine et al. 2008a]. Among 261 patients included in the analysis, prior to ablation 29% had intramucosal carcinoma (IMC), 52% had high-grade dysplasia (HGD), 14% had LGD, and 5% had NDBE as their worst Barrett's grade. Following treatment, 175 had complete eradication of SIM, while 86 had persistent NDBE present on postablation biopsies. A total of 45 patients had recurrence of neoplasia, 30 with LGD, 13 with HGD and 2 with carcinoma. Multivariate analysis showed that advanced age, a history of tobacco use and the inability to completely eradicate SIM were all associated with an increased risk of recurrence. Further studies are needed using other treatment options to determine if they carry the same risk factors for recurrence, and if so how we should incorporate this information into patient counseling and treatment modality selection.
Perhaps the most excitement regarding endoluminal treatment of Barrett's esophagus arose from the multiple abstracts presented that came out of the AIM Dysplasia trial. This prospective randomized sham-controlled study evaluated the efficacy and safety of RFA for use in treatment of Barrett's esophagus with documented LGD or HGD. Nineteen different centers provided up to four total ablative treatments with the HALO® system (BARRX Medical, Sunnyvale, CA), proceeding stepwise starting with a circumferential ablation (CA) followed by focal ablation (FA) as needed. Post-treatment biopsies were obtained either every 3 or 6 months, depending on the grade of dysplasia present when the patient enrolled. Shaheen et al. [2008c] presented interim data from 127 patients, 63 treated for HGD and 64 for LGD. Of these, 58 patients (15 with HGD, 24 with LGD, 19 sham) could be evaluated for the primary endpoint, which was histological clearance of dysplasia and/or SIM at 12 months. These patients had a mean Barrett's length of 4.6 cm (LGD) and 5.2 cm (HGD), and those in the study arm received a mean of 3.5 ablations.
Results of the RFA group showed clear benefit over sham, with efficacy similar to trials from other ablative therapies [Shaheen et al. 2008d]. Overall, using an intention-to-treat (ITT) analysis, 77% of RFA patients had clearance of all SIM at 12 months (83% using per protocol (PP) analysis). Results were even better with clearance of dysplasia, where 83% (ITT) or 94% (PP) of RFA patients had absence of dysplasia at 12 months. For HGD RFA patients, 80% (ITT) or 91% (PP) of HGD patients had no dysplasia after a year, where 60% (ITT) or 75% (PP) of patients lacked any Barrett's metaplasia on biopsy. Not a single HGD patient receiving sham treatment showed either clearance of dysplasia or absence of SIM after 12 months. Similar results were seen with LGD patients, where 90% (ITT) or 95% (PP) of RFA patients had clearance of dysplasia, and 83% (ITT) or 87% (PP) of patients had complete Barrett's eradication after 1 year. Again, these results were significantly better than LGD patients receiving sham therapy, with only about one-third having cleared dysplasia and none having cleared SIM at the time of surveillance biopsies. Additionally, while only 4.7% of the patients who received RFA treatment had progression to more severe dysplasia, significantly more (18.9%) sham patients advanced in their dysplasia grade during the trial.
A separate analysis of this trial focused on identifying the factors responsible for an incomplete response to RFA treatment [Lightdale et al. 2008]. Demographic and endoscopic data were evaluated for the 35 patients who had completed RFA and had biopsies obtained 12 months after ablation. While the sample size was small (29 patients who were eradicated completely and 6 who had incomplete response), a longer pretreatment period with dysplasia present (average of 2.2 versus 0.6 years) was a significant factor in predicting incomplete response (p < 0.05). Other risk factors such as advanced age, higher body mass index, duration of SIM, increased Barrett's segment length and the presence of multifocal dysplasia tended toward statistical significance. However, the sample size was too small to achieve the requisite power. Hiatal hernia size, though, was found to be similar in both groups. It is important to note that all patients with incomplete response to RFA did demonstrate a downgrade in their respective degrees of dysplasia, suggesting at least a partial benefit to undergoing RFA.
Data from Europe, where RFA followed as-needed endoscopic mucosal resection (EMR), provided even better results than the AIM Dysplasia trial. One study involved patients with less than 10 cm of Barrett's esophagus who had either HGD or IMC on biopsy and no sub-mucosal involvement or lymphadenopathy (LAD) on endoscopic ultrasound (EUS) [Pouw et al. 2008b,d]. After visible lesions were removed by EMR, CA was performed followed by repeat CA or FA every 2 months until there was no visible evidence of Barrett's mucosa. Narrow band imaging (NBI) guided biopsies were then taken during a median of 12 months of surveillance endoscopy following the last ablation. Forty-four patients, with a median Prague Barrett's classification of C5M7 (5 cm of circumferential SIM, 7 total cm of esophagus from the top of the gastric folds to the top of the Barrett's mucosa), received a maximum of four RFA treatments. When three patients who required one additional treatment of multiband mucosectomy (MBM) resection each are included, a total of 98% of these patients had complete clearance of SIM. Adverse effects of treatment were uncommon but important, including three non-transmural lacerations at sites of prior EMR when a larger diameter CA balloon catheter was utilized. Four patients developed dysphagia, which resolved after endoscopic dilation. All of these patients had a history of widespread EMR and/or an already narrow esophagus preablation. No stenoses or lacerations were identified in patients who did not require preablation EMR. As with the AIM Dysplasia trial, there was no recurrence of dysplasia seen on follow-up biopsies. Like their American counterparts, investigators from this trial found RFA to be an effective ablative option which compared favorably to other approaches such as PDT, surgery and radical EMR.
Interim analysis from a community-based trial suggested similar eradication results to the trials from major academic centers [Hernandez et al. 2008]. After a single CA in 38 patients, 44% demonstrated complete remission of SIM, while 48% of patients had 50–99% of their biopsies free of Barrett's mucosa. For the ten patients who completed additional therapy as needed for eradication (including RFA and EMR), 70% were in complete remission. No serious complications of therapy were reported, but the trial is ongoing and further reports are expected. The data indicate that gastroenterologists other than early adopters can be successful in using RFA to treat Barrett's; however, broader and longer-term studies are needed to more fully assess this.
As discussed in our prior article [Smith and Lightdale, 2008], we appear to have reached a paradigm shift in the approach to ablation of Barrett's esophagus. While never a truly desirable option, until recently PDT stood as the ablative therapy of choice for any Barrett's esophagus other than small focal lesions. With the initial and now interim returns from early studies, however, RFA has become the ablative therapy of choice for most experienced practitioners who treat this difficult patient population. New data continue to support this trend, such as the abstract comparing RFA with PDT efficacy and complications [Bumgarner et al. 2008]. Charts were reviewed for 225 total patients, 103 who underwent RFA and 122 who received PDT. While not statistically significant, data suggest that RFA was more successful in eradicating dysplasia (odds ratio 0.69), and left a smaller median percent of Barrett's mucosa after the first treatment application (15% versus 30% for PDT, p = 0.40). However, there were significantly fewer strictures seen if RFA was used as the initial ablative modality (p = 0.05), and the risk of developing a stricture was lower for RFA versus PDT use (odds ratio 0.29, 95% confidence interval 0.09–0.95). These data suggest that, when compared to PDT, RFA is at least equally effective with a significantly lower risk of developing a common PDT complication.
Other abstracts also addressed potential adverse effects of RFA use. In one study, the diameter, compliance and motility of 12 esophagi in patients undergoing stepwise CA and FA for Barrett's esophagus with HGD or IMC were evaluated both prior to and two months after RFA was performed [Beaumont et al. 2008]. Ten healthy volunteers also were evaluated to obtain a normal baseline. Data showed that there was no significant change in esophageal diameter (31.5 mm to 31.3 mm), motility or compliance, though interestingly compliance in the pre-RFA cohort was higher than that of healthy volunteers. These data suggest there is no undesired long-term effect on the anatomy or function of the esophagus following RFA.
Attention also was paid to the issue of SSIM and its presence following RFA, due to the small but significant frequency of discovering buried Barrett's following other ablative therapies. Utilizing data from the AIM Dysplasia trial, Shaheen et al. [2008a] evaluated biopsies from both before and after ablation to address the effect of RFA on the prevalence of SSIM. Analysis of 2151 pretreatment biopsies from 127 patients demonstrated SSIM in 25.2% of ablation-nä?ve patients (3.1% of biopsies). SSIM was found to be more common in LGD than HGD patients. Following RFA, however, only one of 1223 biopsy fragments demonstrated SSIM (0.1% of total biopsies), which was a significant decrease from pre-RFA findings in the same cohort (p<0.001). The sham cohort, however, showed a 4.1% prevalence rate for SSIM, unchanged from pretherapy totals. These findings were similar to the previously mentioned European trial [Pouw et al. 2008d], where one in 1,475 biopsies demonstrated SSIM. These results provide further support of RFA over other ablative modalities, which tend to have much higher rates of post-therapy SSIM. They also cast doubt on SSIM being a byproduct of ablation instead of being a not uncommon finding in untreated Barrett's esophagus.
In a related abstract, AIM Dysplasia trial data was evaluated in consideration of whether post-RFA biopsies were ‘missing’ SSIM by failing to achieve adequate depth [Shaheen et al. 2008b]. Biopsies from 45 patients whose endoscopies were performed 12 months after the last ablation were evaluated for the presence of basement membrane (signifying a full epithelium) and sub-epithelial structures including lamina propria, muscularis mucosa and submucosa. While only one biopsy with SSIM was seen in the RFA group, 16 fragments (3.6% of biopsies) in the sham group showed buried Barrett's glands. A significantly smaller percentage of post-RFA fragments contained subepithelial structures, raising concern that the biopsies were not deep enough to detect SSIM due to postablation scarring. However, untreated squamous mucosa had a lower rate of finding subepithelial structures than did untreated Barrett's mucosa. Given the high success rate of RFA in converting SIM to neosquamous epithelium, this finding may explain the discrepancy in results between the RFA and sham groups.
Several abstracts focused on early clinical trials using a low pressure catheter-based liquid nitrogen spray system (CSA Medical, Inc., Baltimore, MD). The central abstract discussed a study evaluating the safety and tolerability of using this technique for Barrett's esophagus with varying degrees of dysplasia, T1 or T2 stage esophageal adenocarcinoma (EAC), and severe squamous dysplasia (SSD) [Greenwald et al. 2008]. Patients were treated with liquid nitrogen application every 4–6 weeks in one of four trial sites, and post-treatment telephone interviews were performed up to 10 days post-ablation. Seventy-seven patients with a median Barrett's length of 4 cm received a total of 323 treatments. Of these patients, 9.1% had NDBE, 58.4% had HGD, 16.9% had IMC, 13% had EAC and 2.6% had SSD. Patients denied any side effects after 48% of the procedures. The most serious adverse event was a gastric perforation in the setting of known Marfan's disease, where inadequate ventilation of warming nitrogen gas was felt to be the most likely etiology. Another patient developed a lip ulcer from contact with the cold endoscope, and three patients developed postablation strictures which all responded to a single dilation. Other complaints noted included chest pain (17% of cases), dysphagia (13.3%) and odynophagia (12.1%), with mild nausea, abdominal pain, sore throat, irregular heartbeat and fever, reported after no more than 5% of the treatments.
A separate study aimed to compare the relative frequency of common postablative symptoms in three key ablative therapies: cryoablative therapy, RFA and PDT [Gross et al. 2008]. Data was collected prospectively from each study arm of 10 patients, with the entire group having a mean Barrett's length of 5 cm. Assessment for symptoms was performed between 1 and 8 weeks after the treatment was administered, when patients were asked to note whether they experienced any of the following: dysphagia, odynophagia, chest pain, weight loss, photosensitivity, or development of a stricture. While patients undergoing PDT reported all six sequelae of treatment, and RFA patients reported all but photosensitivity, cryotherapy patients denied any weight loss, photosensitivity or stricture formation. Additionally, cryoablation led to fewer reports of odynophagia (1/10) and chest pain (4/10) than either RFA (4/10, 5/10) or PDT (5/10, 6/10). Only for dysphagia (3/10) was cryotherapy not the technique producing the fewest complaints (2/10 for RFA, 5/10 for PDT). While the sample size was small, the data supports earlier, more anecdotal evidence that cryoablation appears to be the best tolerated therapy after completion of the treatment.
But can cryoablation effectively eradicate Barrett's esophagus? A separate abstract evaluated results of a pilot study which enrolled 32 patients, 20 with HGD and 12 with IMC (one being squamous cell carcinoma) [Dumot et al. 2008]. Eight of these patients had received prior ablative therapy or EMR. Cryotherapy was administered every 6 weeks until there was no visual endoscopic evidence of residual Barrett's esophagus, and then four quadrant biopsies were taken every centimeter throughout the original Barrett's length. A total of 27 patients completed the postablation biopsies, and were characterized as having either complete remission (CR–no SIM on biopsies), partial remission (PR–focal Barrett's mucosa without dysplasia), incomplete remission (IR–focal SIM with downgrade to LGD), or failure of treatment. Of 18 HGD patients treated, 89% showed at least incomplete remission (10 CR, 5 PR, 1 IR), whereas 67% of IMC patients had some form of regression of disease (3 CR, 1 PR, 2 IR). While these numbers are from a small, single center pilot study, they have fueled optimism that this well-tolerated treatment may be similarly efficacious to RFA and therefore play a role in the future of Barrett's ablation.
Canto et al.  presented another low-flow cryotherapy system utilizing a different chemical — liquid carbon dioxide (Polar Wand GI Supply, Camp Hill, PA). As with liquid nitrogen, the ablative effect is based on a rapid drop in temperature at the luminal surface associated with a pressure drop due to the release of compressed gas, called the Joule–Thompson effect. While the system already is used for treatment of gastric arteriove-nous malformations and radiation proctitis, this study aimed to evaluate the safety and potential efficacy of a new use for the device. As with the liquid nitrogen system, the active agent is applied via a catheter placed through the endoscope's working channel, and a suction catheter is also placed into the upper gastrointestinal tract to remove excess gases. A total of 33 patients were enrolled, all of whom had either HGD or IMC and were not surgical candidates (due to age or comorbidities) who failed a primary therapeutic approach. After placement of a tattoo at the proximal end of long segment Barrett's esophagus, a mean of 2.9 sessions of multiple treatments were performed for each patient. Biopsies showed complete resolution of endoscopic and histologic Barrett's esophagus in 21% and a partial response in the remaining 79%, with a mean reduction in SIM area of 72%. Within this group were three patients who had a positive deep margin on EMR for cancer, all of whom were disease-free on post-cryotherapy biopsies. No procedural complications or adverse effects, other than mild ‘heartburn’ in one patient, were reported. These encouraging pilot study results have led to planning for a phase II trial.
Abstracts presented evaluated the use of endoscopic mucosal resection in all aspects of care for Barrett's esophagus, from initial evaluation to removal. Several of these addressed the shortcomings of diagnostic biopsies in surveillance endoscopy, and whether EMR should play a role or even replace standing biopsies in monitoring of Barrett's esophagus with advanced dysplasia. In one study, 14 patients with HGD and 14 with at least IMC on biopsy were referred for a planned two-stage endoscopic resection of Barrett's esophagus [Fukami et al. 2008]. EMR led to an upgrade in dysplasia grade in 50% of HGD patients, with all four multi-focal HGD (MFHGD) patients and three of ten single-level HGD patients having at least IMC. Of the 14 patients referred with adenocarcinoma (IMC), 21% were downgraded after EMR as no dysplasia more advanced than HGD was identified. Similarly, a retrospective chart review of 14 patients with either HGD or IMC on surveillance biopsies who underwent EMR of the dysplastic site or level showed similar discordance between biopsy and EMR results [Chennat et al. 2008a]. Only 29% of patients demonstrated agreement, while five patients had an upgrade in severity of dysplasia and another five had a downgrade following EMR. Another abstract, which focused on the use of the Duette® multiband mucosectomy system (Cook Medical, Inc., Bloomington, IN), demonstrated a histologic grade change in 37.5% (p< 0.0001) and identification or confirmation of submucosal invasion in 16.9% of 65 patients who underwent EMR [Bhat et al. 2008]. A systematic review of the available literature, covering 15 original articles and 4 abstracts, concluded that 23% of patients had a change in dysplasia grade when EMR was used after biopsy [Sayana et al. 2008]. While EMR clearly carries a greater risk of complications than performing biopsies, these data suggest that greater use of EMR in the diagnosis and surveillance of Barrett's esophagus has the potential to significantly improve accuracy.
Several other abstracts advocated expanded use of endoscopic mucosal resection in Barrett's esophagus. In one study, Manner et al.  demonstrated a 95% complete remission rate following endoscopic mucosal resection of early invasive EAC. Twenty-one patients with ‘low risk’ lesions, defined as invasion only into the outermost ‘SM1’ layer, a lack of lymphatic or vascular infiltration, a histologic grade of G1 or G2 and a gross type of I or II were considered for endoscopic resection for palliative purposes only. Of these patients, 19 were able to complete the therapeutic intervention, and 95% remained in complete remission after a mean follow-up of 5.3 months. Further surveillance of these patients showed that, after a mean of 69 months, 28% developed recurrent or meta-chronous lesions, all of which were treated successfully with repeat resection or APC. The only patient not to achieve tumor freedom died pre-operatively from a myocardial infarction, and no patients died from complications of Barrett's esophagus or their tumors. The overall calculated 5-year survival rate was 66%.
As EMR is the only therapeutic technique that allows for deeper sampling of tissue containing neoplastic tissue, several abstracts advocated widespread use of endoscopic mucosal resection for removal of Barrett's mucosa. A chart review of 21 cases where EMR was performed for HGD or carcinoma found invasive cancer in 7% of non-nodular flat specimens, including one patient whose initial diagnosis was only HGD [Piraka et al. 2008]. The authors concluded that, with a significant risk of flat lesions harboring invasive cancer, focal EMR of nodularity followed by an ablative therapy may be insufficient. They advocated consideration of an approach like that reviewed by Siddiqui and Gerke , where 27 patients with a median Barrett's length of 3cm underwent complete EMR of their Barrett's segments containing either HGD or IMC. Band ligator resection alone was used in 22 patients, while five also underwent cap-based EMR and three received APC treatment. After a median of two sessions, 70% of patients had complete eradication of Barrett's mucosa, while three patients had incomplete resection after being lost to follow up and one died of unrelated causes. EMR failed to completely remove the Barrett's esophagus in only 15% of patients, with two patients having residual HGD or IMC and another two showing only NDBE. The complication rate, however, was high, as 52% developed strictures, 7% had perforations, and 19% had minor postprocedural bleeding. All complications were managed successfully with conservative measures.
An international multicenter trial also evaluated radical endoscopic resection for complete removal of HGD or IMC [Gondrie et al. 2008]. In this study, 149 patients with up to 5 cm of Barrett's mucosa completed the treatment protocol with EMR sessions every 4–8 weeks until complete eradication was obtained. After a median of two treatment sessions, only one patient showed persistent dysplasia (IMC), and was referred for surgery. There was complete removal of all Barrett's mucosa in 97% of patients, with significant complications (two perforations and two delayed bleedings) after only 1% of the procedures. However, 79 strictures formed and two patients experienced perforations during dilation. The stricture rate significantly increased with greater length of the Barrett's segment (p = 0.04). After a median of 18 months of follow-up, 5% of patients had visible Barrett's mucosa during endoscopy with 6% of patients showing SIM and 3% neoplastic Barrett's mucosa on biopsy.
While the previous studies evaluated use of EMR during sequential treatment sessions, another abstract presented safety data for complete removal of Barrett's esophagus during a single EMR session [Chennat et al. 2008b]. Chart review of a total of 44 patients with HGD or IMC was performed, 28 with short segment Barrett's esophagus (SSBE) and 16 with long-segment Barrett's esophagus (LSBE). While there were no perforations or uncontrolled bleeding noted, 12 patients (27.3%) did develop a stricture. Seven strictures resolved after a single dilation, but others required multiple dilations and two patients required steroid injections. One patient even required a temporary esophageal stent. Three postprocedure hospitalizations were required for abdominal pain, chest pain, and a deep mucosal tear that did not show evidence of perforation on imaging. There was no correlation between developing an adverse event and the Barrett's segment length, number of resections performed or number of prior EMR sessions. The authors concluded that the benefit of deeper treatment and resection, along with a majority of complications easily fixed by a single intervention, made this technique a viable and even preferable option for removal of neoplastic Barrett's mucosa.
There has been some reluctance to use EMR for endoluminal treatment of Barrett's esophagus due to possible undesired effects on the physiology of the esophagus that could adversely affect surgical outcomes should the patient proceed to esophagectomy. Badreddine et al. [2008b] assessed whether use of preoperative EMR increased surgical or postoperative complications through a retrospective cohort study of patients who had up to three EMR for at least HGD, then underwent esophagectomy without receiving neoadjuvant chemotherapy. Their results were compared to gender and age-matched patients from the same study period, and after a mean of 35 days from final EMR to surgery, there was no significant difference in operative time, length of stay or postoperative complications (p < 0.81). Especially with consideration of expanded use of EMR for diagnosis or treatment of neoplastic Barrett's mucosa, this study suggests that EMR is not a contraindication to surgery. However, further studies are required to determine if results are similar when attempts at widespread or complete endoscopic resection are made.
With both cap-based and MBM systems available for use in EMR, a prospective trial was performed to compare the two methods for piecemeal resection of Barrett's esophagus with either HGD or IMC [Pouw et al. 2008a]. Forty-five patients with a median Prague classification of C3M5 were randomized to one of the methods, and data was collected including the number of resections performed, procedure time, complications, maximum specimen diameter and equipment cost. The results demonstrated significant time and cost savings with the MBM system, but MBM resections produced smaller specimens (p < 0.05 for all). Two perforations, however, occurred with the cap system, and none were seen with MBM. These data led the authors to conclude that the MBM system obtains smaller and less deep specimens, making it better for flat lesions with low risk of submucosal invasion, whereas they recommend use of cap-based EMR for elevated and nodular lesions that may extend deeper into the esophageal wall.
Attempting to improve on currently available techniques, several abstracts described new technologies for and approaches to endoscopic mucosal resection. Stadlhuber et al.  discussed use of a flexible cold blade-based device that, without cautery, allows for exact and accurate histologic analysis of resection margins (SafeStitch Medical Inc., Miami, FL). After dilute epinephrine injection, a total of ten specimens were excised from ex vivo fresh cadaveric esophagi and ranged in size from 3.0 × 2.5 cm to 2.5 × 2.2 cm. Random examination of a total of 150 areas within these specimens, averaging 0.297 mm in depth, showed that only 2% lacked muscularis mucosa while others contained deep submucosal glands. Another abstract explored the use of directed resection after creating a submucosal working space, allowing for complete resection of any intramucosal process [Gostout, 2008]. In this early study, resections were performed in less than 10 minutes using porcine stomachs. A third study evaluated targeting EMR with use of confocal laser microscopy (CLM) [Badreddine et al. 2008c]. In this prospective cohort study of 62 patients, 40 underwent EMR with 22 of these sites being flat lesions with dysplasia on CLM. The detection rate of advanced dysplasia in EMR specimens identified with use of CLM in flat mucosa was similar to that of specimens resected at nodular sites. The above findings suggest that all three of these technologies have benefits deserving of further investigation.
Another technique which already has gained popularity in Japan is endoscopic submucosal dissection (ESD). Currently an accepted treatment for early gastric cancer, ESD provides the benefit of a single en bloc resection where there is no confusion stemming from lateral margins of multiple specimens. Ishii et al.  treated 28 superficial esophageal cancers (25 squamous cell and 3 adenocarcinoma) with ESD, and were able to perform single en bloc resections in all patients with a mean resection size of 34 mm. Lateral and vertical margins were completely free of invasion on 96% of the specimens. While there were no immediate complications, six patients required subsequent dilation of residual strictures. A prospective randomized study, however, cast doubt on the benefits of ESD over EMR [Deprez et al. 2008]. A total of 23 patients with biopsies showing either superficial squamous cell carcinoma or Barrett's esophagus with HGD or IMC underwent either EMR or ESD. As expected, ESD procedure time and costs were significantly higher than for EMR. However, there was no statistical difference in neoplasia-free deep margins (83% versus 82%) or complete R0 resection (67% versus 73%) between the EMR and ESD specimens. These data suggest that further improvements in technique are required before changing from EMR to the more expensive and time-consuming technique of ESD.
While not an endoluminal therapy itself, EUS has played an important role in selecting a Barrett's treatment modality for many gastroen-terologists. Three abstracts were presented that evaluated the utility of EUS when used as a staging tool prior to endoluminal intervention. Data from the AIM Dysplasia trial was analyzed by Muthusamy et al. , where 62 of 84 patients with HGD had EUS results available. Only nine patients (14%) had any abnormal findings on EUS, including two patients with nodularity (already seen on endoscopy) and seven with LAD. Fine needle aspiration (FNA) of two of the enlarged lymph nodes revealed one benign node, and another malignant node containing metastatic lung cancer. A benign 7mm diameter submucosal nodule also was identified and resected prior to initiation of RFA therapy. The only EUS complication was a cervical esophageal perforation by the echoendoscope. With the frequency of EUS findings actually altering management about the same as the frequency of major complications, the authors questioned the need for routine EUS in patients with HGD. It is not clear, however, whether a radial echoendoscope was used for all cases, or if an endoscopic miniprobe was used in any patients.
A similar evaluation was performed by Whang et al. , who performed a retrospective review of EUS findings from 33 patients with HGD. Two of these patients ultimately were diagnosed with EAC, with both cases suspected on endoscopy. One patient had a friable, nodular mass showing only HGD on biopsy, but on EUS a stage T2 lesion was seen with malignant LAD confirmed by FNA. A second patient had a flat erythematous patch, though EUS was normal, and repeat biopsies showed IMC. While LAD was a common finding, occurring in 70% of patients, only 4% of patients with LAD had malignant nodes, and no patients who underwent FNA demonstrated LAD from EAC. In this study, there were no immediate complications for either endoscopy or EUS. Similar results were seen in another abstract by Rastogi et al. , where 20 total patients (16 HGD, 4 IMC) with a mean Barrett's length of 6.3 cm underwent EUS. Only two patients demonstrated LAD, one with a negative FNA and the other FNA positive for presumed lung cancer given CT findings of a lung mass. All patients except for the one with lung cancer underwent endoluminal therapy, eight by EMR, nine by laser, and two by PDT. After a mean follow-up of 25 months, none of the patients showed any residual HGD or progression to more advanced neoplasia. While all of these studies cast doubt on the clinical utility of EUS for pretreatment evaluation of advanced dysplasia in Barrett's esophagus, there was apparent consensus that further, larger studies are needed before making a decision to remove this generally well-tolerated procedure from the diagnostic algorithm.
The study of molecular markers in Barrett's esophagus is making steady progress toward becoming part of the evaluation of endoluminal treatment. Studies are evaluating whether markers can help predict resistance to treatment and propensity for recurrence of dysplastic SIM. An abstract by Wang et al.  demonstrated in vivo that, as seen in vitro, expression of important cell cycle regulatory genes which play a key role in stress-induced senescence were significantly reduced in nonresponders to PDT. Measuring expression of these genes may be helpful in determining if senescence, in addition to necrosis and apoptosis, is a necessary component of successful esophageal response to ablative therapy. In a different study, none of the patients who had complete resolution of prior abnormalities on fluorescent in situ hybridization (FISH) analysis after ablation for HGD developed recurrent advanced dysplasia [Prasad et al. 2008]. However, two of six patients who retained FISH abnormalities had recurrent HGD after a median of 22 months of follow-up.
A third abstract incorporated mutational analysis into microscopic evaluation in the setting of RFA for Barrett's esophagus for LGD [Finkelstein and Lyday, 2008]. For the 16 patients who underwent a single RFA session, all but one had complete eradication of Barrett's mucosa. The previously seen mutations present in these 15 patients were no longer detectable in any of the postablation specimens. For the one refractory patient, however, while low clonally expanded mutations had disappeared after ablation, the high clonally expanded mutations remained. Similarly, for the patients who required two sessions to ablate all Barrett's mucosa, the high clonally expanded mutations persisted after the first treatment but resolved after the repeat ablation. The authors also noted that mutational regression was not always immediate, but could take up to 12 months to occur. While these findings are promising for future utility of molecular markers in algorithms, at present there is more work to be done to make their use logistically and economically feasible.
Experience with techniques and improved technologies have combined to solidify endoluminal treatment for Barrett's esophagus with advanced dysplasia or early adenocarcinoma as the preferred first option for therapy over direct surgical referral. While the abstracts discussed in this paper cannot and do not carry the same weight as peer reviewed studies, in aggregate they support emerging trends in treatment of Barrett's esophagus and their conclusions should be considered along with those of full publications.
Presentation of the results of the AIM Dysplasia trial has provided the larger scale data to entrench RFA as the new standard of care for diffuse ablative therapies, with efficacy, safety and tolerability data comparable or better than alternatives such as PDT. Combinatorial therapy utilizing both ablation and resection likely will remain the default approach to more complex Barrett's mucosal eradication, however, and so future trials must continue to explore the adjustments required to obtain an optimal ‘working relationship’ between paired techniques. For example, Pouw et al. [2008c] concluded that complications of CA using RFA can be minimized if the extent of prior EMR is limited to a maximum of 50% of the esophageal circumference and a length of 2cm, and if the diameter of the ablation catheter treatment balloon is selected conservatively. As we refine treatment algorithms, these studies will play a key role in maximizing eradication outcomes while minimizing adverse effects.
While RFA has established itself as the leader of the pack in ablative therapies, other techniques are making a play for more widespread use. We only have early results for cryotherapy, but the data is encouraging from both an efficacy and tolerability standpoint. Unlike RFA, cryoablation has the advantage of not requiring a flat mucosal surface to deliver therapy. This proves particularly helpful in patients with tortuous or dilated esophagi, which are not uncommon in the older Barrett's esophagus population and in those with large hiatal hernias. Continued positive results from further studies containing larger patient cohorts, along with improvements in catheter technology allowing for more controlled delivery of the cryogen, may give cryoablation the potential to become at least a solid niche player in the ablative arena.
It is hard to imagine a treatment algorithm without some form of endoscopic mucosal resection, given its ability to obtain tissue from the deep therapeutic margin. For irregular or nodular mucosa, there is widespread consensus that EMR is a logical first step. However, as noted by Wood et al. , there is little consensus on optimal techniques for the equipment available, even among expert users. This is particularly true when searching for opinions on radical endoscopic resection of entire Barrett's esophagus segments, currently far more accepted as an option in Europe than in the United States. Potential complications, especially the significant stricture rate, must be dealt with before this approach to removal of Barrett's mucosa gains more widespread acceptance.
Particularly when considering larger resections, the high risk of stricture formation underscores that EMR and, in our opinion, other Barrett's eradication techniques are not suited for occasional use by the general gastroenterologist. Centralizing endoluminal treatment of Barrett's esophagus within referral centers, where specialists both are trained in the multiple advanced procedures that comprise the Barrett's treatment toolkit and also treat sufficient patients on a regular basis to maintain a high level of endoscopic skill, provides the highest likelihood of achieving the greatest rates of eradication along with the lowest rates of complication. While most of these techniques can be learned with relative ease by any gastroenterologist, it is their continued and frequent use that is essential for maintaining excellent outcomes for patients. Just as important is the gastroenterologist being familiar with a variety of the endoluminal techniques, as it is not uncommon for planned use of a technique to be jettisoned in favor of an alternate approach, once endoscopic visualization of the Barrett's mucosa has revealed an unexpected finding.
We remain optimistic that refinement of existing technologies and further development of newer techniques, along with use of advanced imaging techniques and molecular marker analysis, will only continue to improve patient outcomes for treatment of advanced Barrett's esophagus. Unfortunately, durability of eradication remains a crucial metric in evaluating the ultimate success of a therapy. Therefore, while we treat with our current knowledge base, we must also wait while important additional clinical trials are performed on the promising but young techniques currently under evaluation. Peer-reviewed publication of these longer-term studies will be critical to establishing the proper place of each technique in the Barrett's treatment algorithm.
CL has received small amounts of remuneration from various sources (AstraZeneca, Barrx Medical, Boston Scientific, Cook, CSA, Ethicon, Olympus, Takeda). MS has received small amounts of remuneration from Barrx Medical.
Michael S. Smith, Assistant Professor of Medicine, Temple University School of Medicine, Section of Gastroenterology, 3401 North Broad Street, 8PP, Zone “C”, Philadelphia, PA 19140, USA ; Email: ude.elpmet.shut@3htimS.leahciM.
Charles J. Lightdale, Professor of Clinical Medicine, Columbia University Medical Center, 161 Fort Washington Avenue, New York, NY 10032, USA.