Endoscopic submucosal dissection of gastrointestinal lesions on an outpatient basis

Abstract

Background

Endoscopic submucosal dissection (ESD) is usually associated with hospital admission.

Objectives

To evaluate, prospectively, the feasibility, safety and efficacy of outpatient gastrointestinal ESD.

Methods

Patients with suitable lesions were invited to participate. Those that dwelt more than 1 hour from the hospital, lived alone, had severe co-morbidities, were <18 years old, had duodenal lesions, or that had ESD-related complications were admitted. The remaining patients were discharged if no complications were detected. A patients’ inquiry was performed.

Results

Of the 164 ESD patients, 122 were outpatient-based, corresponding to 115 patients, 47% male and mean age 63 ± 12 years-old. Outpatients tended to be younger, female, to have gastric lesions, less advanced lesions, and shorter and less complicated ESDs (all p < 0.05). Outpatients’ mean tumour size was 38 mm, en bloc and R0 resection rates were 88 and 78%, respectively. Seven ESD outpatients (5.7%) had complications: delayed bleeding (n = 4), pneumonitis (n = 2) or emphysema (n = 1), all managed conservatively. Colorectal location of the lesions was predictive of hospital admission (p = 0.03). In total, 97% of patients were satisfied with the outpatient strategy.

Conclusion

Risks of ambulatory ESD are low and complications can be successfully managed. This strategy has high patient satisfaction. More studies are needed to evaluate its implications on costs and patients’ management.

Key summary

• 1. What is the established knowledge?

  • Endoscopic submucosal dissection (ESD) usually requires patient admission.
  • There is a global trend to perform gastrointestinal surgical and endoscopic procedures on an outpatient basis.
  • Advance endoscopic procedures such as wide-field mucosal resection are performed on an outpatient basis.
• 2. What are the new findings of this study?

  • Outpatient ESD has a high safety profile.
  • All patients that needed medical assistance after outpatient ESD were treated conservatively.
  • This strategy is associated with high patient satisfaction.

Introduction

There is a global trend to perform several medical and surgical interventions on an outpatient basis. This is related to the increase in procedures’ safety, and the rise in the costs and complications associated with hospitalization.¹ Currently in the US, most gastrointestinal (GI) surgical procedures are performed in an ambulatory setting, which is the case in 55% of cholecystectomies, 60% of laparoscopies, and 89% of inguinal and femoral repairs. This approach is safe, benefits patient’s well-being and has significant economic implications.²

A few decades ago, endoscopic resection of large polyps was performed on an inpatient basis.³ Nowadays, even large endoscopic mucosal resections (EMRs) are performed in an outpatient setting, irrespective of lesions’ location and size. Endoscopic submucosal dissection (ESD) is used for the removal of GI superficial neoplasias.⁴ Contrary to EMR, ESD allows en bloc resection of lesions larger than 2 cm, enabling proper pathological assessment. The latter is associated with increased curative rates and reduced recurrence rates, but is more often complicated with perforation. Both techniques have similar bleeding rates.⁵

ESD in Japan is associated with hospital admission for up to 5–7 days, mainly due to reimbursement policies.⁶ Our centre was the first to report the safety and efficacy of outpatient ESD in 2012.⁷ Since then, others in the East and West have described similar results.^8–10 However, those studies were retrospective, small or related to ESD in one specific GI organ.

The aim of this study was to evaluate the feasibility, safety and efficacy of outpatient GI ESD. The primary outcomes were dedicated ambulatory endpoints such as adverse events or unplanned medical assistance after discharge. The secondary outcomes were patient satisfaction and the rates of en bloc, R0 and curative resections.

Methods

Study design

This was a prospective, longitudinal cohort study approved by our institution’s Ethical Committee in 2012 (reference 15/2012). The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki, as reflected in a priori approval by the institution's human research committee.

Patients

All participants were managed by a GI multidisciplinary team (MDT). Patients with GI neoplasia larger than 2 cm, with risk of submucosal invasion but no criteria for deep submucosal invasion according to international recommendations,^11–13 and that were proposed by the MDT for en bloc resection, were invited to participate in the study. Those with anaemia due to large gastric hyperplastic lesions or with subepithelial lesions causing symptoms, or in which investigations with endoscopy and endoscopic ultrasound were inconclusive, were also invited to participate.

Lesions with suspected deep submucosal invasion or with poor differentiated histology were not resected by ESD. Patients that dwelt more than 1 hour from the hospital, lived alone, had severe co-morbidities (American Society of Anesthesiologists (ASA) score ≥3), were <18 years old or that had duodenal lesions were admitted after ESD (not included in this outpatient ESD study).

All patients provided written informed consent. Those under antiplatelet drugs or anticoagulants were handled according to published guidelines. Included patients arrived in the morning of the procedure and were handled as ambulatory surgery, being discharged after careful monitoring.

Endoscopic protocol

All ESDs were performed under anaesthesiologist-administered sedation. Intravenous proton pump inhibitors (PPIs) were administered in upper GI ESDs. All procedures were carried out by a senior experienced endoscopist (FBS or MM) with extensive expertise in endoscopic diagnosis and therapeutics,^14–16 namely on ESD. All ESDs were performed under carbon dioxide insufflation. Endoscopy was performed using high-definition endoscopes (GIF-180/190, PCF/CF-180/190; Olympus Corp.) with white light, followed by narrow band imaging for lesions’ characterization and delineation. Chromoendoscopy with lugol, acetic acid, methylene blue or indigo carmine was used according to the respective organ and lesion. Lesions were macroscopically classified according to the Paris classification. In upper GI cases, marking dots were applied 5 mm outside the lesion using the tip of the Dual-knife (Olympus Corp). A saline solution with diluted epinephrine, methylcellulose and methylene blue was used for submucosal injection and lifting. An incision was then made outside the rim of dots (5 mm outside the lesion in colorectal cases). After incision, Dual-knife, IT-knife 2 or IT-nano (Olympus Corp.) were used for submucosal dissection according to the endoscopist’s preference. When access to the submucosa was hampered, auxiliary techniques were used.¹⁶ Bleeding during the procedure was treated with the tip of the knife or with a coagrasper (Olympus Corp.). After resection, preventive coagulation of the visible vessels was performed. In cases of perforation, clips were used for closure (Figures 1 to 3).

Figure 1.

(a) Endoscopic imaging of a squamous cell neoplasia in the middle oesophagus (Paris IIa–b) after lugol chromoscopy. (b) “Tatami-no-me” sign. (c) Well demarcated lesion with dots. (d) Oesophageal ulcer after endoscopic submucosal dissection. (e) En bloc resection. (f) Resected specimen after lugol instillation, highlighting free margins.

Figure 2.

(a) White light endoscopic imaging of a gastric lesion with high-grade dysplasia in the antrum’s lesser curvature (Paris IIa–b). (b) Same lesion with narrow band imaging. (c) Proximal incision. (d) Submucosal dissection with IT Knife 2. (e) Gastric ulcer after endoscopic submucosal dissection. (f) Resected specimen with free margins.

Figure 3.

(a) Lateral spreading tumour (granular mixed type) in the colon ascendens (retroflexed view). (b) Proximal subepithelial injection showing good lifting. (c) Proximal incision. (d) Presence of fibrosis in the submucosa. (e) Submucosal dissection with Dualknife. (f) Resected specimen with free margins.

Assessment after ESD

Patients were under continuous monitoring for a minimum of 6 hours after the procedure. If there were no signs of complications, participants were discharged. To exclude post-ESD bleeding, endoscopy or digital rectal examination were performed after upper and lower GI ESD, respectively. In upper GI ESDs, patients had twice-daily PPIs for 8 weeks. All participants had 24/7 phone contact to the endoscopists, and were advised to call anytime or to go to the emergency room in case of complications. Adverse events were checked through a telephone interview within 48 hours and during an outpatient clinic visit at 4–6 weeks. Signs or symptoms of complications, readmission, transfusion requirements and interventions were checked at both assessments. Afterwards, a telephone interview was conducted by a person not related to our team (NS or PA) to assess participants’ satisfaction.

Follow-up endoscopy was performed at 3 months, and further surveillance was performed according to histological results and published literature.⁵ All histological specimens were reviewed by two dedicated GI pathologists.

Outcome parameters

Primary outcome parameters

Adverse events were defined as any complication in which ESD or ESD-related procedures (such as anaesthesia) were a contributing factor. Perforation was defined as a full thickness breach in the GI wall with or without symptoms. Post-polypectomy syndrome was considered as pain after ESD that resulted in hospital admission. Intraprocedural bleeding was considered significant when it could only be managed with endoclips, when blood transfusion was needed or when it caused premature termination of ESD. Delayed bleeding was defined as occurrence of haematemesis, melena or haematochezia up to 1 month after the procedure that demanded medical assistance. Aspiration pneumonitis was defined as acute lung injury after the inhalation of regurgitated gastric content. ESD-related mortality was considered as any death in which ESDs or related procedures were a contributing factor.

Adverse events were classified as acute (during procedure), early (<48 h) or late (>48 h). In terms of severity they were characterized as mild (unplanned medical assistance, hospital admission, hospitalization <3 days, haemoglobin drop <3 g or no transfusion), moderate (4–10 days hospitalization, <4 units blood transfusion, repeat endoscopic intervention or radiological intervention), severe (hospitalization >10 days, intensive care unit admission, need for surgery or >4 units blood transfusion) or fatal (death attributable to procedure).¹⁷

Secondary outcome parameters

En bloc resection was defined as resection of the aimed lesion in one fragment. R0 resection corresponded to en bloc resection with histologically free horizontal and vertical margins. Curative resections were those intramucosal and R0, or those with superficial submucosal invasion and low histological risk criteria (according to the respective organ) and R0.^11–13 All cases with invasive neoplasia or non-curative resection were discussed again at the MDT conference. Recurrence was defined by the presence of suspicious neoplastic tissue in the post-ESD scar under high-definition endoscopy confirmed by histology.

Statistical analysis

The sample size was calculated for the primary endpoint of complications that led to medical observation after outpatient ESD. It was based on available data of complications of GI ESD that ranged between 5–30%.^18,19 As we wanted to have enough power to detect complications if they occurred, the sample size was calculated based on a lower complications rate than previously described. The calculation assumed a complication rate of 4%, a confidence interval of 95%, a power of 80% and a margin of error of 3%. From this, it was determined that 112 ESD procedures were required and that as many as 122 ESD procedures would be needed to account for possible losses during follow-up.

Demographic and baseline characteristics were summarized using descriptive statistics. Mean (±SD) and Student’s t-tests were used in case of normal distribution, or presented as median (interquartile range) and compared with the Mann–Whitney U-test for skewed distribution, in continuous variables. Categorical data were presented as percentages and were compared with the χ²-test. Multiple logistic regression was used to identify the independent predictors of outcomes of interest. Statistical analyses were performed using specific software (SPSS 22; SPSS Inc).

Results

Patients

A total of 157 patients, corresponding to 164 GI lesions, were invited to participate in the study. Forty-two patients, corresponding to 42 lesions, were excluded and treated on an inpatient basis (Figure 4). Of these, only two needed surgical treatment of complications (one gastric perforation and one severe gastric bleed the day after ESD, not possible to be managed endoscopically), with both patients being discharged within 7 days. One-hundred and fifteen patients, corresponding to 122 lesions, 47% male and mean age 63 ± 12 years, fulfilled the inclusion criteria and were included in the study, i.e. ESD was performed on an outpatient basis. The corresponding lesions were located in the stomach (60%), rectum (18%), colon (17%) and oesophagus (4%). Patients managed in an outpatient setting tended to be younger, female, to have gastric lesions, less advanced lesions, shorter ESDs and fewer complications (all p < 0.05). There were no significant differences between inpatient and outpatient ESD groups in terms of anticoagulation or antithrombotic use, Paris classification and lesion size (Table 1).

Figure 4.

Study flow diagram.

ESD: endoscopic submucosal dissection.

Table 1.

Patients demographics, lesions and endoscopic submucosal dissection characteristics between the inpatient and outpatient endoscopic submucosal dissection groups.

	Inpatient ESD	Outpatient ESD	p-value
Gastrointestinal lesions (n)	42	122	–
Age (years ± SD)	68 ± 12	63 ± 12	0.04*
Gender (male/female)	31/11	57/65	0.02**
Use of antithrombotic drugs (anticoagulants/antiplatelet drugs)	2	3	0.2**
Tumour location (oesophagus/stomach/duodenum/colon/rectum)	4/24/2/7/5	5/73/0/20/22	0.04**
Lesion size (mm: mean ± SD)	32.8 ± 21	38 ± 11	0.12*
Histology(non-carcinoma/carcinoma)	27/15	99/23	0.04**
Procedure time (min: mean ± SD)	190 ± 103	103 ± 64	<0.001**

ESD: endoscopic submucosal dissection.

*Student’s t-test.

**Pearson’s χ²-test.

Primary outcomes

Complications occurred in 5.7% (7/122) of cases: four delayed bleeds, two aspiration pneumonitis and one subcutaneous emphysema in an ESD in the distal rectum. Five cases were early and two were late complications. Considering the severity of complications, one was mild, five moderate and one severe. All were admitted and endoscopy was performed in all bleeding cases: three of four needed endoscopic therapy and three of four patients needed transfusion of red cell concentrate (Table 2). The bleeding was caused by ulceration after ESD in the antrum in the four cases. No patients needed emergency surgery due to complications and no mortality occurred. Multiple logistic regression analysis showed that colorectal location of the lesions (p = 0.03) was predictive of a need for hospital return. Tumour size (p = 0.08), histology (p = 0.33), type of resection (en bloc/piecemeal, p = 0.34), procedure length (p = 0.22) and anticoagulation (p > 0.05) could not predict the occurrence of complications needing hospital admission.

Table 2.

Characteristics of lesions, procedures and outcomes of outpatient endoscopic submucosal dissection.

	Oesophagus	Stomach	Rectum	Colon	Total
Lesions (n)	5	74	22	21	122
Lesion size median (range), mm	32.5 (15–45)	33 (15–85)	40 (20–70)	35 (15–80)	38 (15–85)
En bloc resection n (%)	5 (100)	71 (97.1)	17 (77.3)	18 (94.1)	108 (88)
Procedure time, median (range), min	120 (40–255)	70 (15–310)	112 (15–240)	145 (20–345)	90 (15–345)
Lateral free margins n (%)	5 (100)	70 (95)	16 (73)	16 (76)	104 (85)
Deep free margins n (%)	4 (80)	72 (97)	19 (86)	18 (86)	108 (88)
R0 n (%)	4 (80)	70 (95)	16 (73)	16 (76)	95 (78)

Secondary outcomes

In the included patients, the mean tumour size was 38 mm (range 15–85 mm) and the median procedure time was 90 minutes (range 15–346 minutes). En bloc resection rate was 88% and the R0 rate was 78% (Table 3). The pathology result showed presence of adenoma with low-grade dysplasia in 29% of cases (stomach n = 20, colon n = 9 and rectum n = 6), 26% had high-grade dysplasia (stomach n = 18, colon n = 7 and rectum n = 8), 11% intramucosal adenocarcinoma (oesophagus n = 1, stomach n = 7, colon n = 2 and rectum n = 4) and 8% corresponded to invasive adenocarcinoma (oesophagus n = 2, stomach n = 4, colon n = 3 and rectum n = 1). In gastric ESD specimens, there were eight cases indeterminate for dysplasia, eight hyperplastic polyps, five inflammatory polyps, two heterotopic pancreas and two lipomas. Additionally, there were two cases of hyperplastic lesions in the rectum, and one schwannoma and one squamous cell neoplasia in the oesophagus.

Table 3.

Characteristics of patients and corresponding lesions in the complicated outpatient endoscopic submucosal dissection.

	Gender	Age	Location of lesion	Size (mm)	Onset (days)	Complication	OAC	Admission (days)	RCC transfusion (units)	Endoscopic treatment
1	Male	41	Rectum	47	1	Emphysema	No	8	–	–
2	Female	68	Stomach	35	1	Bleeding	No	6	2	Adrenalin + coagrasper
3	Male	69	Stomach	44	23	Bleeding	No	3	2	Adrenalin + haemoclips
4	Female	61	Colon	50	1	pneumonitis	No	5	–	–
5	Male	78	Stomach	55	15	Bleeding	No	9	4	Adrenalin + coagrasper
6	Female	67	Stomach	40	1	Bleeding	No	3	–	No treatment
7	Male	64	Colon	38	1	Pneumonitis	No	12	–	–

OAC: oral anticoagulants; RCC: red cells concentrate.

Seven patients went through surgery due to invasive carcinoma without curative criteria, of which three had carcinoma on the surgical specimen and four had no neoplasia. During the 39 months of follow-up, three patients had adenoma with low-grade dysplasia in the scar, all of which were treated successfully with endoscopic resection.

When asked at the survey, 97% (111/115) of patients considered that going home on the same day was the right decision and would choose outpatient ESD in case of need for new ESD.

Discussion

The acceptable risk of adverse events in outpatient surgery is considered to be around 5%, depending on the population and procedure.²⁰ The present study addressed the question of whether ESD may be performed in an outpatient setting and if the associated complications may be safely handled with good patient satisfaction. Our results are encouraging based on the fact that only 5.7% of patients needed unplanned medical assistance after ESD. Of seven re-admitted patients, six were discharged within 10 days and all were managed conservatively.

Endoscopic practice is influenced by local expertise and reimbursement policies.²¹ That is the case of radiofrequency ablation for Barrett’s oesophagus treatment, which is usually performed in an ambulatory setting but requires admission in Germany due to suboptimal reimbursement.²² In Japan, ESD patients are usually admitted for 5–7 days and in Europe for 48–96 hours after ESD.²³

A recent nationwide review from Japan shows that total complication rates of GI ESD are low, ranging from 3.3% in the oesophagus to 4.6% in the colon.⁴ Currently, wide-field EMR is performed in an outpatient setting, irrespective of a lesion’s size or location. The main reasons that would hamper outpatient ESD would be the same that would hamper outpatient wide-field EMR, namely a significant rate of complications after discharged.

Contrary to EMR, there is a systematic assessment of submucosal vessels during ESD with the possibility of preventive coagulation before bleeding occurs. All meta-analyses comparing EMR with ESD for gastric or colorectal lesions have shown similar intraoperative and delayed bleeding rates when comparing both techniques.²⁴ Delayed bleeding rates after ESD have been reported to be less than 1% in the oesophagus, 5% in the stomach, and 2% in the colon and rectum.^25–27 Our results reinforce those data, as in total only 1.8% had significant intraprocedural bleeding and only 3% had delayed bleeding in the outpatient group, all of which were treated conservatively.

Perforation is a rare complication, occurring in 0.9–2.0  and 3–6 % of EMR and ESD cases, respectively.^28,29 Perforations during ESD are usually small and are detected during the procedure, being treated endoscopically in most cases. Delayed perforation usually demands surgical treatment, but it is a rare complication of ESD, occurring in up to 0.1–0.5% of cases. In fact, < 0.8% of patients are operated on due to perforation.³⁰ There were no perforations in the outpatient group, favouring this strategy.

It has been demonstrated that the majority of ESD-related complications may be detected and treated in the first hours after the procedure.³¹ In order to detect early signs of perforation, our patients had full monitoring and were evaluated before discharge. We applied different strategies to detect and decrease the risk of delayed bleeding, namely through coagulation of exposed vessels, PPI administration and second-look endoscopy in upper GI cases. Although the value of second-look endoscopy has been questioned, it is still carried out in many institutions to reduce delayed bleeding.³² In most centres, second-look endoscopy is performed 24–48 hours after ESD. We performed it some hours after ESD, and that lead to the detection of bleeding or recent bleeding in eight patients and change management in 9% of upper GI ESDs. It enabled successful treatment in all but one patient (operated on). As delayed bleeding is rare after colonic ESD and as colonoscopy is a rather invasive procedure, second-look endoscopy is not usually advocated in this context.²³ It is important to note that we only had four cases of delayed bleeding, that two occurred more than 2 weeks after the ESD and that all were treated conservatively.

It has been demonstrated that higher ESD experience is associated with better outcomes. Our center performs more than 100 ESDs per year, being a high-volume centre.⁴ Additionally, ESD is performed by a restricted number of endoscopists, which plays a pivotal role in outcomes.³³ That may explain complication rates similar the ones reported by eastern centers.

One of the strengths of this study is the large number of included patients with close and broad follow-up, which is reflected by the fact that pneumonitis, a known complication of anaesthesia-assisted endoscopy³⁴ and seldom reported in ESDs studies, was detected in 1.6% of our patients. We also performed a patient’s inquiry, which is also scarcely performed in ESD studies. The patients’ feedback strengthens the safety and confidence in this strategy. We also included ESDs performed along the GI tract, reflecting the reality of a high-volume center. It has been demonstrated that the best management of GI neoplasia is based on MDT decisions³⁵ and our patients were managed in that setting.

There are certain limitations to our study. It reflects the experience of a high-volume centre with specialized endoscopists and our results might not be translated to other contexts. In order to implement ESD on an outpatient basis, the presence of endoscopists experienced in ESD should be advised. This is a first prospective assessment in this setting and it would not be possible to have a multicentric study at this stage. Another limitation is that 28% of patients were not included in the study (not ambulatory treated), but that is a limitation of any advanced technique, where some patients are admitted due complications or due to risks of complications. This is not a randomized control trial (RCT), but that was not the purpose of this study, and the fact that there are no published RCTs on inpatient/outpatient EMR did not preclude its dissemination. We also did not estimate the cost savings of this strategy. Finally, some of the resected lesions were benign and ESD may be disputable in that context. That is explained by the downgrading of the first histological assessment in some patients, and reflects the practice of a big centre when symptomatic benign or subepithelial lesions without full characterization after diagnostic workout are resected.¹⁰

In conclusion, this first prospective study on outpatient ESD along the GI tract demonstrates that this strategy is associated with high patient satisfaction, that risks of ambulatory ESD are low and that complications can be successfully treated. More studies are needed to evaluate the implications of outpatient ESD on patient management and cost savings.

Funding

There were no study sponsors or funding for the study design, collection, analysis and interpretation of the data. Besides the authors, no other individuals provided writing assistance.

Declaration of conflicting interests

All authors deny potential conflicts of interest relevant to this manuscript.

Informed consent

All patients provided written informed consent.

Ethics approval

The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki, as reflected in a priori approval by the institution’s human research committee.