DISCUSSION
Despite growing evidence towards the use of EMR and other endoscopic esophagus-preserving techniques for treatment of intramucosal carcinoma, endoscopic resection as an alternative to esophagectomy is not yet seen as a consensus treatment (3). Our findings support the use of EMR in the setting of IMC with invasion to T1sm1 depth or less. We successfully eradicated IMC in 10/11 (91%) of patients using the EMR technique. This is similar to described success rates in the literature (15, 16). This was accomplished in most cases with one or two endoscopy sessions on an outpatient basis with practically no morbidity; only one patient was admitted for monitoring for bleeding but no patient actually had rebleeding or required intervention.
Comparatively, esophagectomy is fraught with significant morbidity. Two-thirds of patients undergoing esophagectomy had some form of complication, including two who required reintervention or intensive care unit stays. Moreover, esophagectomy carries with it all the disadvantages of a surgical intervention, including post-operative pain, a decrease in quality of life, and a median hospital stay of over two weeks.
In the EMR group, one patient failed to have eradication even with an EMR specimen showing two foci of IMC with margins negative for malignancy. He was found 8 months later to have persistent disease in the esophagus and nodal disease, but was not a surgical candidate. In the esophagectomy group, despite surgical resection, one patient was found to have recurrent nodal disease a year and a half after esophagectomy. Both of these patients died of their disease, showing an estimated similar long-term mortality rate between both groups.
Another patient had successful EMR eradication of IMC with biopsies showing no persistence of adenocarcinoma but presence of high-grade dysplasia. He continued to have regular endoscopic treatment with the goal of removing his high-grade dysplasia but was found 4 years later to have another adenocarcinoma which was not amenable to endoscopic treatment. Given the difference in time between the initial IMC and the recurrence, this patient may have actually had metachronous development of a new lesion. Nonetheless, he was able to be treated successfully with a salvage esophagectomy, from which he unfortunately suffered an anastomotic leak.
EMR also serves as a method for obtaining a larger biopsy, therefore allowing better determination of depth of invasion (19). Standard endoscopic biopsies yielding pathology compatible with intramucosal carcinoma are sometimes unable to detect a nearby invasive component due to sampling error or lack of depth on the biopsy specimen. Although excluded from our analysis, two patients were initially found to have pathology consistent with intramucosal carcinoma and were taken for EMR. These EMRs showed deeper disease beyond the submucosa with positive margins confirming a stage greater than T1 and therefore requiring esophageal resection.
Limitations of our study include the retrospective nature of the data, which can limit the interpretation of our long-term results. The relatively small number of patients also allows for potential type II error but is compatible with the relative rarity of the disease. The limited number of patients may not capture more rare, but potentially important, complications such as post-EMR bleeding. There is also a length-time selection bias inherent to transitioning from esophagectomy to EMR, with overall in-hospital care changing from 2005 to 2013.
In conclusion, EMR was successful in eradicating IMC in 10/11 patients with lower morbidity, and similar long-term recurrence and mortality compared with esophagectomy patients. Patients with IMC may benefit from EMR as initial therapy by obviating the need for a complex and morbid operation.