Ligation-assisted endoscopic mucosal resection for esophageal granular cell tumors is safe and effective

SUMMARY

Given their malignant potential, resection of esophageal granular cell tumors (GCTs) is often undertaken, yet the optimal technique is unknown. We present a large series of dedicated endoscopic resection using band ligation (EMR-B) of esophageal GCTs. Patients diagnosed with esophageal GCTs between 2002 and 2019 were identified using a prospectively collected pathology database. Endoscopic reports were reviewed, and patients who underwent dedicated EMR-B of esophageal GCTs were included. Medical records were queried for demographics, findings, adverse events, and follow-up. We identified 21 patients who underwent dedicated EMR-B for previously identified esophageal GCT. Median age was 39 years; 16 (76%) were female. Eight (38%) had preceding signs or symptoms, potentially attributable to the GCT. Upon endoscopic evaluation, 12 (57%) were found in the distal esophagus. Endoscopic ultrasound was used in 15 cases (71%). Median lesion size was 7 mm, interquartile range 4 mm—8 mm. The largest lesion was 12 mm. A total of 20 (95%) had en bloc resection confirmed with pathologic examination. The only patient with tumor extending to the resection margin underwent surveillance endoscopy that showed no residual tumor. No patients experienced bleeding, perforation, or stricturing in our series. No patients have had known recurrence of their esophageal GCT. EMR-B of esophageal GCT achieves complete histopathologic resection with minimal adverse events. EMR-B is safe and effective and seems prudent compared with observation for what could be an aggressive and malignant tumor. EMR-B should be considered first-line therapy when resecting esophageal GCT up to 12 mm in diameter.

INTRODUCTION

Granular cell tumors (GCTs) are rare soft tissue tumors that can occur anywhere in the body and are thought to arise from Schwann cells.^1–3 Limited literature suggests that <2% of GCTs are malignant, but given that they can be associated with a poor prognosis, resection or close follow-up are often undertaken.^4–6 Malignant features are described both clinically (rate of growth, size, invasion) and histopathologically, via features initially posed by Fanburg-Smith et al.⁷ where a tumor has >3 of the following 6 features: necrosis, spindling, vesicular nuclei with large nucleoli, increased mitotic activity, high nuclear-to-cytoplasmic ratio, and pleomorphism. Malignant GCTs act as high-grade sarcomas and are associated with short survival.

While GCTs are rarely encountered in the gastrointestinal tract, most are found in the esophagus and are restricted to the submucosal layer.^8,9 The majority of esophageal GCTs are asymptomatic, often found incidentally, precluding a true incidence rate, though when large, they can cause symptoms.^10,11 As noted, due to malignant or symptomatic potential, GCTs are often removed. Surgical or endoscopic removal is frequently performed for large lesions or those with high suspicion for malignancy, whereas smaller or asymptomatic lesions are often followed clinically.^12,13 With advances in endoscopic resection, endoscopy is preferred to surgical resection when feasible, depending on size and depth of invasion, especially when combined with endoscopic ultrasound (EUS) to determine safety of endoscopic resection.⁸ Various resection methods have been described, including endoscopic mucosal resection (EMR), both classic and by band ligation, endoscopic submucosal dissection (ESD), submucosal tunnel endoscopic resection (STER).^14–16 In 1995, Yasuda et al.¹⁷ proposed criteria for safe and effective use of EMR for esophageal GCTs, including lesions <20 mm without invasion of the muscularis propria. Newer techniques, such as STER, can manage larger lesions, including those between 20 mm and 30 mm, further advancing the role of endoscopy over surgery in the management of these lesions.^9,12

Yet for the majority of lesions, EMR remains the standard; though, because the vast majority of esophageal GCTs are presented as case reports, the ability to conclusively recommend further management options is limited.^11,18 There have been reports that endoscopic resection using band ligation (EMR-B) using a mucosectomy kit is an effective technique, associated with ease of use and complete pathologic resection.^16,19,20 Submucosal injection has been used at time of resection, with assumption that this may help procure clean margins.^16,19,21 Here, we set out to describe our experience in order to address optimal EMR technique for esophageal GCTs.

METHODS

This is a descriptive analysis of patients seen at our tertiary care referral center at the University of Pennsylvania (Philadelphia, PA). A prospectively collected pathology database was used to identify all patients diagnosed with esophageal GCT since 2002 at our medical center. Endoscopic reports were reviewed, and patients who underwent ligation-assisted resection were included. EMR-B was performed using the Duette Multi-Band Mucosectomy system (Cook Medical, Bloomington, IN). Resections performed without ligation were excluded.

The electronic medical record (EPIC as electronic medical record) was reviewed to collect information regarding patient demographics, procedural findings, technical details of resection, pathologic findings, adverse events, and follow-up. This study was approved after expedited review by the Institutional Review Board at the University of Pennsylvania. Endoscopies were performed by therapeutic endoscopists, who elected to perform EUS in addition to the resection at their discretion. Statistical analysis was performed using Stata/IC 15.1 (College Park, TX).

RESULTS

We identified 21 patients who underwent EMR-B for esophageal GCT between 2002 and 2019. Median age was 39 years, range 22 years—64 years. Sixteen patients (76%) were female. Of the 21 patients, 8 (38%) had described preceding signs or symptoms that could have been related to the esophageal GCT. These included dysphagia, acid reflux, and abnormal imaging that had detected the GCT. The others were incidentally found to have esophageal GCT upon endoscopic evaluation for nonesophageal indications prior to the resection procedure.

Upon endoscopic evaluation, 5 (24%) GCTs were found in the proximal esophagus, 4 (19%) in mid esophagus, and 12 (57%) in the distal esophagus. The lesions were seen on forward-viewing endoscope and ranged from umbilicated to smooth contoured (Fig. 1A–C). Narrow band imaging did not demonstrate features of epithelial neoplasia (Fig. 1D). EUS was used to interrogate the lesion in 15 cases (71%). Figure 2 demonstrates the classic EUS appearance of a GCT ideal for resection, with a well-circumscribed lesion lying within the deep mucosa.

Fig. 1.

(A–D): Endoscopic images of esophageal granular cell tumors, including narrow band imaging (1D).

Fig. 2.

Classic endoscopic ultrasound appearance of a granular cell tumor (indicated by red arrow), with surrounding submucosal layer (blue dashed line) and muscularis layer (yellow arrow).

Resection was performed in all 21 cases. Submucosal lift with normal saline was used in 9 (43%) cases at the operators’ discretion. Figure 3 demonstrates a typical postresection image.

Fig. 3.

Endoscopic image after resection.

Median size of the GCT as defined by pathologic specimen evaluation was 7 mm, interquartile range 4 mm—8 mm. The largest lesion was 12 mm. Upon pathologic evaluation, 20/21 (95%) had complete, en bloc resection. None were deemed malignant upon pathologists’ review. The only case with positive histologic margins was a 7-mm distal esophageal GCT, resected using saline lift followed by EMR-B without preceding EUS. Repeat endoscopy with biopsies 6 months later showed no residual tumor.

No patient in our series experienced bleeding, perforation, or subsequent stricturing. No recurrence of esophageal GCT was noted in this series.

DISCUSSION

Esophageal GCTs are uncommon lesions, and clinical experience has become concentrated at referral centers.^22,23 Options for management include observation, operative resection, or endoscopic resection.^15,24 Due to their malignant potential, or potential to cause symptoms, resection is often undertaken, though strict guidelines do not exist. Endoscopic resection has been described using freehand inject and resect techniques, EMR with cap-assist, STER, and ESD.^14–16 Here, we demonstrate that EMR-B is a safe, effective, and efficacious technique to curatively resect to small esophageal GCTs.

In our series of 21 patients, the largest report of dedicated esophageal GCT resection in the US, we demonstrate that ligation-assisted esophageal GCT resection is safe and effective. The combination of typical endoscopic features, EUS findings, and (when available) preexisting pathology confirmed the nature and resectability of lesions for our endoscopists. Histopathologic resection was complete for lesions up to 12 mm. Typical endoscopic appearance of a GCT is a raised, firm, yellow–white, with negative pillow sign.²² Many lesions had central umbilication, possibly a sequela of prior forceps biopsy. Our experience suggests that EUS proves valuable in affirming the lesion’s origin in the deep mucosa and ensuring preservation of the submucosal and muscularis propria layers. In our setting, with existing histopathologic diagnoses, EUS was useful, but not essential. In other series, EUS findings have been helpful, especially in differentiating from other nongranular cell lesions, but GCTs have notably variable findings.^18,23,25 In general, they present as homogenous lesions <2 cm, though have been described as both hyperechoic and hypoechoic.^26,27 As such, pathologic confirmation can be important prior to resection to affirm the characteristics of the lesion.¹⁸ Given the referral nature of our center, we were able to have preexisting histopathology confirming the GCT and its associated features prior to the resection procedure.

Our series supports that pre-resection injection is unnecessary and may even hinder ligation-assisted curative resection, in as pre-ligation injection was used in the only case with tumor extending to the resection margins. The rate of curative resection was not improved by preresection injection. EMR-B negates the need for injection, as the ligation band exudes and excludes the muscularis propria layer.¹⁹

Previous reports have suggested the use of prophylactic mucosal clip closure to prevent delayed bleeding or perforation.²⁸ We did not perform closure and had no adverse effects related to bleeding or perforation. We successfully performed ligation-assisted EMR with en bloc resection in 2 cases with lesion size of 12 mm. Larger lesions may require ESD or STER but for lesions ≤12 mm, EMR-B suffices and at a fraction of the time, risk, cost, and training required for the former techniques.¹⁵

A limitation of our study is the lack of standardized long-term follow-up. Given that the histopathology confirmed en bloc curative resection in all but one case, surveillance endoscopy was deemed unnecessary. However, a previously conducted study in Korea of 15 patients who underwent EMR-B for esophageal submucosal tumors (10 of which were esophageal GCTs) noted no endoscopic recurrence during an almost 4-year long mean follow-up period.¹⁹ In this study, endoscopic follow-up was performed at 6 months after resection, then annually thereafter. Further limitations include its single-center and retrospective nature and potential selection bias in those patients who were referred. Lastly, we did not have preresection histologic diagnosis for all patients, which, if available, could help inform future guidelines regarding optimal management of GCTs (i.e. consideration of close observation for small lesions without worrisome histology).

Various methods for resection of esophageal GCTs have been proposed, but we demonstrate that EMR-B is safe, effective, and efficient, achieving complete histopathologic resection with minimal adverse events. EUS remains valuable in confirming the origin of the lesion and determining safety of resection. In as this resection technique appears safe, EMR-B seems more prudent than either observation for what could be an aggressive and malignant tumor or higher-risk procedures, such as surgery, STER, or ESD, though future studies should evaluate the appropriate role of these methods for resection of esophageal GCT. Future studies should be conducted to identify endoscopic, histologic, and other parameters to guide resection of esophageal GCTs, in order to develop guidelines regarding management of these lesions. Future research should evaluate whether follow-up and surveillance are needed in the absence of concerning pathologic findings, and if indicated, identify optimal methods of surveillance (i.e. endoscopy with mucosal biopsy, EUS, etc.). It is possible that en bloc curative resection may be a cost-effective management strategy, particularly if it obviates the need for long-term follow-up with multiple endoscopic procedures; however, further studies are needed to assess the costs associated with either strategy. Our experience suggests that EMR-B should be considered first-line therapy when pursuing resection of esophageal GCT up to 12 mm in diameter. Preligation submucosal injection does not improve the safety or quality of EMR-B for GCT.

Financial support: Dr Shria Kumar is supported by the National Institutes of Health training grant (5 T32 DK 7740-22).

Potential competing interests: Dr Shria Kumar: Travel (Boston Scientific Corporation, Olympus); Dr Vinay Chandrasekhara: Medical Advisory Board (Interpace Diagnostics), Shareholder (Nevakar Corporation); Dr Michael L. Kochman: Consultant: BSC, Olympus, Pentax MAB/Shareholder, Dark Canyon Labs, Virgo; Dr Peter J. Lee: Consultant (Abbvie); Dr Gregory G. Ginsberg: Consulting (Boston Scientific Corporation), Consulting (Olympus Corporation); and Drs. Nuzhat Ahmad, Sara Attalla, Immanuel K. Ho, David L. Jaffe, Kashyap V. Panganamamula, Monica Saumoy, and Danielle Fortuna have no potential interests to disclose.