Recurrence following successful eradication of neoplasia with endoscopic mucosal resection compared with endoscopic submucosal dissection in Barrett’s esophagus: a retrospective comparison

Abstract

Background Endoscopic mucosal resection (EMR) and endoscopic submucosal dissection (ESD) are effective treatments for Barrett’s neoplasia. However, little is known about recurrence rates following these techniques. We compared long-term neoplasia recurrence rates following EMR and ESD.

Methods This study included patients with Barrett’s neoplasia (high grade dysplasia/adenocarcinoma) treated between July 2019 and December 2023 at a tertiary referral center in Canada. Outcomes were residual neoplasia at first follow-up, complete remission of neoplasia (CRN), and neoplasia recurrence following CRN.

Results 157 patients were included (87 EMR, 70 ESD). Compared with EMR, the ESD group had larger lesions (median 2 vs. 3 cm, P<0.05), more adenocarcinoma (85.1% vs. 94.3%, P = 0.07), and deeper submucosal invasion (T1a: 71.6% vs. 75.8%; T1b-SM1: 25.7% vs. 6.1%; T1b≥SM2: 2.7% vs. 18.2%; P<0.05). Among 124 patients with follow-up (71 EMR, 53 ESD), 84.9% of ESD-treated patients had curative resections (i.e. R0 resection with low risk for lymph node metastasis), whereas 94.4% of EMR-treated patients had deep margin R0 resection of low risk lesions. At first follow-up, residual neoplasia (14.1% vs. 11.3%) and CRN (97.2% vs. 100%) were similar in the EMR and ESD groups, but neoplasia recurrence following CRN was significantly higher with EMR (13% vs. 1.9%, P<0.05), with cumulative probability of recurrence at 3 years of 18.3% vs. 4.2%, respectively.

Conclusions Neoplasia recurrence following CRN was significantly higher following EMR compared with ESD, suggesting that ESD may be superior to EMR in preventing neoplasia recurrence in Barrett’s esophagus.

Introduction

Barrett’s esophagus is a precursor condition that may lead to the development of esophageal adenocarcinoma (EAC) [1] [2]. Visible lesions within Barrett’s esophagus may possess high grade dysplasia (HGD) or early-stage EAC and should be treated by endoscopic resection [3] [4] [5].

Endoscopic mucosal resection (EMR) has traditionally been the standard technique for resecting Barrett’s neoplasia, and involves a cap-assisted or band ligation technique to remove pieces of esophageal mucosa. This method typically can only resect small areas en bloc and requires piecemeal resection for larger lesions [3] [4]. Owing to its high efficiency and safety profile, EMR has been widely accepted as the standard treatment for Barrett’s neoplasia. However, endoscopic submucosal dissection (ESD) has emerged as a viable alternative. ESD enables the dissection of the submucosal layer using an endoscopic knife, allowing for en bloc resection of lesions, irrespective of size. Although ESD is more technically demanding and resource intensive than EMR, this technique potentially offers tumor-free margins (R0 resection) and curative resection of larger lesions [5].

Both EMR and ESD are effective techniques for achieving complete remission of neoplasia (CRN), although EMR may require multiple procedures to achieve the same outcome. Recent guidelines recommend selection of ESD for lesions with suspected submucosal invasion or for larger lesions exceeding 20 mm in size [4] [5]. However, there is limited evidence regarding recurrence rates following CRN achieved by EMR compared with ESD. Therefore, this study sought to compare the efficacy of EMR and ESD, focusing on neoplasia recurrence rates post-CRN [4] [5].

Methods

Study design and setting

We conducted a retrospective cohort study at a single tertiary referral center in Toronto, Canada. Eligibility criteria for inclusion required patients to have histologically confirmed HGD or EAC based on endoscopically resected specimens between July 2019 and December 2023. Endoscopic treatment modalities encompassed both EMR and ESD, with a mandatory follow-up endoscopy after the initial treatment. Exclusions included low grade dysplasia histology and patients lacking follow-up. Data collection included patient demographic details, procedure information, histopathological findings, and follow-up data. Research Ethics Board approval was obtained from University of Toronto (REB #08–265). We reported our findings following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [6].

Outcomes and definitions

Outcomes measured were the residual neoplasia rate at the first follow-up, the CRN rate, and the neoplasia recurrence rate following CRN.

En bloc resection was defined as the resection of the entire lesion in a single piece. R0 resection was defined as tumor-free deep (vertical) and lateral resection margins. For EMR, deep margin R0 resection was defined as tumor-free deep resection margins, irrespective of lateral resection margin. For ESD, curative resection (i.e. R0 resection of low risk lesions) was defined as en bloc and R0 resection with low risk of lymph node metastasis (i.e. submucosal invasion ≤500 μm, no poor differentiation, and no lymphovascular invasion) [4] [5]. For EMR, deep margin R0 resection of low risk lesions was defined as tumor-free deep resection margins with low risk of lymph node metastasis.

CRN was defined as complete absence of endoscopic and histological evidence of neoplasia (HGD/EAC) on follow-up. Complete remission of intestinal metaplasia (CRIM) was defined as complete absence of histological intestinal metaplasia on follow-up. Recurrence of neoplasia was defined as reappearance of HGD or EAC after CRN.

Indications for EMR and ESD

As per guideline recommendations [4] [5], EMR was indicated for visible lesions ≤20 mm with a low likelihood of submucosal invasion. ESD was indicated for lesions suspected of submucosal invasion (Paris classification 0-Is, 0-IIc, 0-IIa+IIc, and 0-IIa+Is) and lesions exceeding 20 mm.

Endoscopic resection protocol

EMR was carried out using the multiband ligation technique, utilizing a therapeutic gastroscope (GIF-1T190; Olympus Co., Tokyo, Japan) and Duette multiband mucosectomy device (Cook Medical Inc., Bloomington, Indiana, USA) ([Fig. 1]). ESD was carried out utilizing a gastroscope (GIF-HQ190; Olympus Co.) fitted with a soft plastic cap. Voluven (Fresenius Kabi AG, Bad Homburg, Germany) with methylene blue was injected submucosally. Incision, dissection, and hemostasis were performed with DualKnifeJ and Coagrasper (Olympus Co.), and electrosurgical generator (VIO-300D, Erbe USA Inc., Marietta, Georgia, USA) ([Fig. 2]).

Fig. 1 Endoscopic images of endoscopic mucosal resection of a 20-mm 0-IIb lesion. a Marking around the lesion. b Suctioning the lesion into the cap. c Attaching the band and performing resection with a snare. d Resection defect.

Fig. 2 Endoscopic images of endoscopic submucosal dissection of an 80% circumferential 0-IIb lesion. a Marking around the lesion. b,c Mucosal incision and submucosal dissection with an endoscopic knife. d Resection defect.

Histopathological analysis

Histopathological assessments were performed by two expert gastrointestinal pathologists. EMR/ESD specimens were evaluated based on the grade of differentiation, depth of tumor invasion, presence of lymphovascular invasion, and the completeness of resection at the deep and lateral margins (lateral margins for EMR specimens were not investigated) [7]. The depth of invasion was classified into T1a, T1b SM1, and T1b SM2 or deeper (>500 μm) [8].

Post‑EMR/ESD protocol

Patients underwent follow-up endoscopy approximately 3 months after EMR/ESD. In cases where residual neoplasia was detected, repeat endoscopic resection was performed until CRN was achieved. Radiofrequency ablation (RFA) (Medtronic, Minneapolis, Minnesota, USA) was then performed on the remaining Barrett’s esophagus without visible lesions, aiming to achieve CRIM. Patients then underwent surveillance endoscopy every 3 months during the first year, every 6 months during the second year, and subsequently on an annual basis. Surveillance biopsies were systematically collected in four quadrants at 1-cm intervals along the initial length of the Barrett’s esophagus segment, based on the Seattle protocol [9]. Targeted biopsies were taken from any visually detected abnormalities.

Statistical analysis

Descriptive statistics, encompassing mean with SD, median with interquartile range (IQR), and percentages were utilized to characterize variables and outcomes. The Kaplan–Meier method was used to estimate the cumulative probability of neoplasia recurrence over time. The log-rank test and the generalized Wilcoxon test were used to compare the probabilities between groups. Recurrence rates were presented as percentages with 95%CIs. Statistical analyses were executed using JMP Pro17 software (SAS Institute Inc., Cary, North Carolina, USA), with a two-sided significance level of P<0.05.

Results

Study population

A total of 157 patients with Barrett’s esophagus with HGD/EAC were included in the study. Of these, 87 patients were treated with EMR, while 70 patients were treated with ESD (see Fig. 1s in the online-only Supplementary material). In the EMR group, the mean age was 70.1 (SD 10.6) years with 78.2% male. In the ESD group, the mean age was 68.8 (SD 8.6) years with 90% male ([Table 1]). Regarding the Prague classification, median circumferential (C) and maximal (M) extent of Barrett’s esophagus was C1M3 in the EMR group and C2M4 in the ESD group. The prevalence of long-segment Barrett’s esophagus was 52.9% in the EMR group and 65.7% in the ESD group. The median lesion size in the EMR group was 2 cm (IQR 1–2 cm), whereas the ESD group had a larger median lesion size of 3 cm (IQR 2–5.5 cm; P<0.001). Regarding the Paris classification, the ESD group had more protruded, depressed, and complex types (P<0.001). The distribution of the Paris classification in the EMR group was as follows: 0-Is (4.6%), 0-IIa (64.4%), 0-IIa+IIc (2.3%), 0-IIb (25.3%), 0-IIc (3.4%). In the ESD group, the distribution was 0-Is (12.9%), 0-IIa (41.4%), 0-IIa+IIc (18.6%), 0-IIa+Is (12.9%), 0-IIb (7.1%), 0-IIc (7.1%).

Table 1 Study population.

	EMR (N = 87)	ESD (N = 70)	P value
C, circumferential; EMR, endoscopic mucosal resection; ESD, endoscopic submucosal dissection; IQR, interquartile range; LSBE, long-segment Barrett’s esophagus; M, maximal; SSBE, short-segment Barrett’s esophagus.
Age, mean (SD), years	70.1 (10.6)	68.8 (8.6)	0.41
Male sex, n (%)	68 (78.2)	63 (90.0)	0.05
Prague C, median (IQR), cm	1 (0–4)	2 (0–4.25)	0.57
Prague M, median (IQR), cm	3 (1–6)	4 (2–7)	0.24
Barrett’s length, n (%)			0.14
SSBE	41 (47.1)	24 (34.3)
LSBE	46 (52.9)	46 (65.7)
Size of lesion, median (IQR), cm	2 (1–2)	3 (2–5.5)	<0.001
Circumferential extent of lesion, median, (IQR), %	50 (25–63)	50 (36.3–70)	0.39
Paris classification, n (%)			<0.001
0-Is	4 (4.6)	9 (12.9)
0-IIa	56 (64.4)	29 (41.4)
0-IIa+IIc	2 (2.3)	13 (18.6)
0-IIa+Is	0 (0)	9 (12.9)
0-IIb	22 (25.3)	5 (7.1)
0-IIc	3 (3.4)	5 (7.1)

Histopathological outcomes

In the EMR group, there were 13 cases (14.9%) of HGD and 74 cases (85.1%) of EAC, of which 73% were well differentiated, 23% were moderately differentiated, and 4.1% were poorly differentiated ([Table 2]). The ESD group had 4 cases (5.7%) of HGD and 66 cases (94.3%) of EAC, comprising 56.1% well differentiated, 33.3% moderately differentiated, and 10.6% poorly differentiated adenocarcinomas. Regarding invasion depth, the EMR group had 71.6% T1a, 25.7% T1b SM1, and 2.7% T1b SM2 or deeper, whereas the ESD group had 75.8% T1a, 6.1% T1b SM1, and 18.2% T1b SM2 or deeper (P<0.001). Lymphovascular invasion was less prevalent in the EMR group (2.3%) compared with the ESD group (21.4%; P <0.001). Overall, more advanced lesions were selected for ESD with a greater likelihood of having deep submucosal invasion or positive lymphovascular invasion.

Table 2 Histopathological and procedure outcomes.

	EMR (N = 87)	ESD (N = 70)	P value
Data are n (%). EMR, endoscopic mucosal resection; ESD, endoscopic submucosal dissection; NA, not applicable. 1Out of the 10 esophagectomy patients, only one patient had residual cancer in the surgical specimen (pT1aN0). The remaining nine patients had no residual cancer and no lymph node metastasis.
Histopathological outcomes
Histology			0.07
High grade dysplasia	13 (14.9)	4 (5.7)
Adenocarcinoma	74 (85.1)	66 (94.3)
Differentiation of adenocarcinoma			0.08
Well differentiated	54 (73.0)	37 (56.1)
Moderately differentiated	17 (23.0)	22 (33.3)
Poorly differentiated	3 (4.1)	7 (10.6)
Invasion depth of adenocarcinoma			<0.001
T1a	53 (71.6)	50 (75.8)
T1b SM1	19 (25.7)	4 (6.1)
T1b SM2 or deeper	2 (2.7)	12 (18.2)
Lymphovascular invasion	2 (2.3)	15 (21.4)	<0.001
Procedure outcomes
Adverse events			0.19
Delayed bleeding	0 (0)	2 (2.9)
Perforation	1 (1.2)	0 (0)
En bloc resection	32 (36.8)	68 (97.1)	<0.001
R0 resection	NA	58 (82.9)	–
Low risk lesion (i.e. curative resection)	NA	45 (64.3)
High risk lesion	NA	13 (18.6)
R1 resection	NA	12 (17.1)
Deep margin R0 resection	85 (97.7)	NA
Low risk lesion	80 (92.0)	NA
High risk lesion	5 (5.7)	NA
Deep margin R1 resection	2 (2.3)	NA
Additional treatment			<0.001
Esophagectomy	0 (0)	10 (14.3)1
Chemoradiotherapy	3 (3.4)	7 (10.0)

Procedure outcomes

En bloc resection rate was higher with ESD than with EMR (97.1% vs. 36.8%; P<0.001) ([Table 2]). In the ESD group, the R0 resection rate was 82.9% and curative resection (i.e. R0 resection with low risk pathology) was achieved in 64.3% of patients. Meanwhile, 18.6% of patients in the ESD group had R0 resection of lesions with high risk pathology and 17.1% had R1 resection with positive margins. In comparison, 97.7% of patients in the EMR group had deep margin R0 resection with 92.0% having low risk pathology considered curative and 5.7% with high risk pathology considered noncurative. Finally, 2.3% of patients in the EMR group had R1 resection with positive deep margin.

Three patients (3.4%) in the EMR group underwent subsequent chemoradiotherapy due to high risk pathology and none underwent esophagectomy ([Table 2], Fig. 1s). In contrast, 10 patients (14.3%) underwent esophagectomy in the ESD group and 7 patients (10.0%) received chemoradiotherapy following potentially noncurative pathology. Interestingly, only one of the 10 patients who underwent esophagectomy because of high risk pathological findings following ESD resection had residual cancer in the surgical specimen (pT1aN0). Patients who received these additional treatments were excluded from the follow-up outcome analysis, whereas eligible patients who declined subsequent surgery or chemoradiotherapy were enrolled into surveillance and included in the follow-up outcomes. Among the EMR group, these included two patients with R0 high risk pathology and two patients with positive deep margin R1 pathology. Within the ESD group, these included three patients with R0 high risk pathology and five patients with R1 resections.

Follow-up outcomes

All patients were enrolled in the follow-up cohort except those who underwent esophagectomy or chemoradiotherapy or were lost to follow-up. Follow-up was completed for 124 patients, including 71 from the EMR group and 53 from the ESD group. The EMR group comprised 94.4% with negative deep margins and low risk pathology considered curative, 2.8% with deep margin R0 resection but high risk pathology considered insufficiently curative, and 2.8% with positive deep margins ([Table 3]). The ESD group comprised 84.9% with R0 resection of lesions with low risk pathology considered curative, 5.7% with R0 resection of lesions with high risk pathology considered noncurative, and 9.4% with R1 resections.

Table 3 Patient outcomes at follow-up.

	EMR (N = 71)	ESD (N = 53)	P value
Data are n (%) unless otherwise stated. CRIM, complete remission of intestinal metaplasia; CRN, complete remission of neoplasia; EMR, endoscopic mucosal resection; ESD, endoscopic submucosal dissection; IQR, interquartile range; NA, not applicable; RFA, radiofrequency ablation. 1The pathological results of the esophagectomy showed no residual cancer and no lymph node metastasis. 2The pathological result of the esophagectomy was pT1aN1. 3None of the patients under surveillance following ESD required an esophagectomy. The number of esophagectomies for patients with noncurative resections based on initial ESD pathology is shown in [Table 2].
R0 resection
Low risk lesion (i.e. curative resection)	NA	45 (84.9)	–
High risk lesion	NA	3 (5.7)
R1 resection	NA	5 (9.4)
Deep margin R0 resection
Low risk lesion	67 (94.4)	NA	–
High risk lesion	2 (2.8)	NA
Deep margin R1 resection	2 (2.8)	NA
Residual neoplasia at first follow-up	10 (14.1)	6 (11.3)	0.78
Total number of endoscopic resections until CRN, median (IQR)	1 (1–1)	1 (1–1)	0.92
CRN	69 (97.2)	53 (100)	0.21
Non-CRN	2 (2.8)	0 (0)
Treatment for non-CRN
Esophagectomy	1 (1.4)1	0
Palliative care	1 (1.4)	0
RFA	50 (70.4)	29 (54.7)	0.090
CRIM	24 (33.8)	21 (39.6)	0.50
Recurrence of neoplasia after CRN	9/69 (13.0)	1/53 (1.9)	0.042
Treatment for recurrence
Endoscopic treatment	8	1
Esophagectomy	12	03
Total follow-up, mean (SD), days	437 (301)	362 (304)	0.28

During the first follow-up, residual neoplasia was observed in 14.1% (10 cases) of the EMR group and 11.3% (6 cases) of the ESD group (P = 0.78) ([Table 3]). Residual neoplastic lesions (HGD or EAC) were treated with additional endoscopic resection, which ultimately resulted in CRN in all 53 patients (100%) in the ESD group and 69 patients (97.2%) in the EMR group. Two patients in the EMR group had residual neoplasia and did not achieve CRN, of whom one underwent esophagectomy (pathology: no residual cancer, negative lymph nodes) and the other received palliative care.

Following resections of all visible lesions, 70.4% of the EMR group and 54.7% of the ESD group underwent RFA, with 7.0% of EMR patients and 20.8% of ESD patients already achieving CRIM at the completion of the resection procedures without needing subsequent ablation. A further 22.5% of EMR patients and 24.5% of ESD patients did not undergo RFA therapy due to decompensation of medical comorbidities, need to manage complex esophageal strictures following near-circumferential endoscopic resection, or because of “noncurative” endoscopic resection with pathological findings considered high risk for lymph node metastasis, in which case endoscopic follow-up was often performed without initiation of RFA. In the subgroup of patients targeted for CRIM, CRIM was achieved in 43.6% of the EMR group and in 52.5% of the ESD group, with the remainder continuing with ongoing RFA therapy at the time of this analysis. Overall, CRIM was achieved in 33.8% of the EMR group and 39.6% of the ESD group (P = 0.50) ([Table 3]).

Among patients who achieved CRN, there was a significantly higher rate of recurrent neoplasia in the EMR group compared with the ESD group (13.0% vs. 1.9%; P = 0.04) ([Table 3]). The mean follow-up period was similar: 437 (SD 301) days for the EMR group and 362 (SD 304) days for the ESD group (P = 0.28). In the EMR group, eight cases of neoplastic recurrence were successfully managed with endoscopic treatment (seven by repeat EMR and one by hot avulsion), whereas one case failed repeat EMR due to recurrence in severe fibrosis and subsequently underwent esophagectomy (pathology: pT1aN1). There was only one case of recurrent neoplasia in the ESD group, which was successfully treated by EMR ([Table 3], Fig. 1s)

A Kaplan–Meier curve showing the cumulative probability of neoplasia recurrence following CRN is presented in [Fig. 3]. The probability of neoplasia recurrence between the EMR group and the ESD group was notably different (log-rank test P = 0.049 and generalized Wilcoxon test P = 0.04). In the EMR group, the 1-year neoplasia recurrence rate was 14.4% (95%CI 6.4%–22.4%), which increased to 18.3% (95%CI 6.6%–30.0%) at both the 2-year and 3-year marks. In contrast, the ESD group maintained a consistent recurrence rate of 4.2% (95%CI 0.0–12.2%) at the 1-year, 2-year, and 3-year time points ([Table 4]).

Fig. 3 Kaplan–Meier curve. Cumulative probability of neoplasia recurrence after complete remission of neoplasia.

Table 4 Cumulative probability of neoplasia recurrence after complete remission of neoplasia.

	EMR	ESD
EMR, endoscopic mucosal resection; ESD, endoscopic submucosal dissection.
Recurrence rate (95%CI), %
1 year	14.4 (6.4–22.4)	4.2 (0.0–12.2)
2 years	18.3 (6.6–30.0)	4.2 (0.0–12.2)
3 years	18.3 (6.6–30.0)	4.2 (0.0–12.2)

Discussion

This study compared neoplasia recurrence rates of EMR and ESD following complete remission of neoplasia in Barrett’s esophagus. Notably, the neoplasia recurrence rate after CRN was significantly higher with EMR, occurring in 13% of patients following EMR but in only 1.9% of patients following ESD. These findings suggest that ESD may be superior to EMR in completely eradicating microscopic dysplasia, thereby reducing the risk of neoplasia recurrence in follow-up.

The findings from our study demonstrate a notable increase in recurrent HGD/EAC over time among patients in whom all apparent neoplasia was successfully resected by EMR. Specifically, the recurrence rates in the EMR group at 1, 2, and 3 years post-procedure were 14.4%, 18.3%, and 18.3%, respectively. In contrast, there was a 4.2% recurrence rate in the ESD group at 1 year with no further recurrences over time at 2 and 3 years. These recurrences of HGD/EAC are described among a cohort of patients in whom there had been no apparent residual dysplasia on detailed endoscopic examination or on biopsies, and CRN was believed to have been achieved. Furthermore, these recurrences occurred despite a higher rate of subsequent treatment with RFA in the EMR group.

The key difference between EMR and ESD is the anticipated extent and depth of resection. EMR typically involves piecemeal resection while ESD enables en bloc excision of larger lesions with likely deeper submucosal planes, ideally achieving higher rates of R0 resection (negative vertical and lateral margins). Piecemeal EMR requires lining up each cut so that there are no bridges of intervening tissue and may lead to indistinct lateral margins, increasing the risk of leaving residual neoplastic cells. A possible explanation for our findings is that residual neoplastic cells may be left behind at the lateral margins of piecemeal EMR resections, which are more definitively removed by en bloc resection with ESD. This effect persists despite the fact that a greater proportion of patients underwent RFA in the EMR group than in the ESD group (70.4% vs. 54.7%; P = 0.09), which would be expected to eradicate dysplasia if it was arising from the remaining Barrett’s segment. Furthermore, despite the rates of CRN being similar in both groups, our results demonstrate significantly higher neoplasia recurrence rate in the EMR group. This suggests that even after complete endoscopic resection of all visible neoplasia followed by RFA with the achievement of CRN, the likelihood of neoplasia recurrence is higher with EMR than with ESD. This is likely because residual metaplastic and dysplastic cells are left behind following piecemeal EMR and RFA. Therefore, even when short-term follow-up esophagogastroduodenoscopy shows no visible dysplasia and negative Seattle protocol biopsies, there may be a higher chance of persistent dysplastic cells following piecemeal EMR than after ESD. In addition, previous literature has shown that achieving CRIM is a factor in reducing neoplasia recurrence [10]. Our results showed a low rate of CRIM (likely due to ongoing RFA therapy that has not yet been completed), which may have contributed to the recurrence rate observed in both groups, but would not account for the observed difference in recurrence rate between EMR and ESD.

Several studies have compared the outcomes of EMR and ESD in Barrett’s esophagus, but ours is the first report to explore true neoplasia recurrence rates following the achievement of CRN. Terheggen et al. conducted a prospective randomized controlled trial with a small patient cohort, consisting of 20 patients with EMR and 20 patients with ESD [11]. At the 3-month follow-up, there was no significant difference in the rate of complete remission of dysplasia between the two groups (ESD 15/16 vs. EMR 16/17, P > 0.99). Similarly, the recurrence rates did not differ significantly (ESD 1/16 vs. EMR 0/17, P>0.99). However, the limited sample size and short follow-up period of this study are likely to be insufficient to evaluate the recurrence rate. Mejia Perez et al. conducted a retrospective study comparing EMR and ESD in 243 patients at 8 centers in the USA and Brazil [12]. The main outcome was the incidence of residual/recurrent lesions following a single session of either EMR or ESD for the treatment of HGD or T1a EAC. This study showed a significantly lower rate of residual/recurrent lesions of 3.5% (3/85) with ESD compared with 31.4% (44/140) for EMR. However, the methodology of this study did not isolate the true recurrence rate, as the outcomes included residual HGD found on first follow-up. This is not surprising given that EMR was performed to resect HGD without EAC in 55% of cases in the EMR group and 25% of cases in the ESD group, and such lesions often have poorly defined margins in Barrett’s esophagus. In contrast, the vast majority of patients in our study had EAC, comprising 85% and 94% of cases in our EMR and ESD groups, respectively. Furthermore, it is unlikely that the need for repeat EMR sessions for residual HGD should be considered a treatment failure of EMR relative to ESD, as such iterative treatments by EMR followed by RFA are an established standard of care in Barrett’s esophagus. Another study that compared EMR with ESD in Barrett’s esophagus was performed by Doumbe-Mandengue et al., who conducted a single-center retrospective study of 28 cases of EMR and 57 cases of ESD [13]. Over a median follow-up period of 27.5 months, the rate of neoplasia recurrence was 23% in the ESD group vs. 18% in the EMR group (P = 0.63). However, this study likely confounded cases of residual lesions as “recurrence,” as the recurrence assessment was not conducted after the confirmation of CRN, meaning that lesions that were reported as recurrences may have in fact been residual dysplasia. In our study, 14.1% of patients in the EMR group and 11.3% in the ESD group were found to have residual neoplasia at first follow-up and repeat endoscopic resections were performed before achieving CRN. Our analysis was then focused on patients who ultimately achieved CRN, establishing a follow-up cohort to assess for recurrence. By this approach, we were able to assess the true recurrence rate, explicitly excluding residual dysplasia. To the best of our knowledge, this study is the first to report the true recurrence rates after EMR and ESD in Barrett’s esophagus. Finally, a recently completed meta-analysis confirms that ESD is associated with increased rates of curative resection and reduced recurrence, seemingly in support of our findings [14].

This study has several limitations. First, it is a single-center, retrospective analysis, which may limit the generalizability of the results. Second, current guidelines provide criteria guiding the choice between EMR and ESD for treating Barrett’s neoplasia, which introduces an inherent selection bias [4] [5]. Consequently, the ESD group tended to have larger lesions, a higher incidence of adenocarcinoma, and deeper invasion depths, similarly to other previous studies [12] [13]. Therefore, the ESD group comprised a higher proportion of patients at increased risk for local recurrence and lymph node metastasis. Despite this, the ESD group demonstrated a lower recurrence rate, confirming that this selection bias does not compromise the interpretation of our findings, and if anything, strengthens the observed difference between resection methods. In addition, the association between treatment choice (EMR/ESD) and recurrence could not be adjusted for confounding factors due to the small sample size of this study. The ESD group had larger lesions and more cases with suspected submucosal invasion, which are confounding factors associated with a higher risk of recurrence, potentially exaggerating the association between ESD and increased recurrence. Despite this, our findings showed a lower recurrence risk in the ESD group. Therefore, adjusting for these confounding factors would likely further reduce the relative risk of recurrence in the ESD group. Consequently, whether or not we adjust for these factors, the interpretation remains that ESD is associated with a lower risk of recurrence. To more accurately evaluate the recurrence rates after EMR and ESD in similar lesions, a large-scale, randomized trial would be necessary. Finally, metachronous lesions are common in Barrett’s esophagus and distinguishing between local recurrence of the resected neoplasia and new metachronous lesions is difficult. However, regardless of whether patients have recurrence from resected neoplasia or develop de novo lesions, the ESD group had significantly fewer cases over time despite having more advanced cancers to begin with, indicating that ESD should be beneficial in reducing recurrence.

In summary, this study examined the neoplasia recurrence rates in Barrett’s esophagus following EMR and ESD, and found significantly lower recurrence rates with ESD. This suggests that ESD may be superior in reducing the recurrence of neoplasia when treating Barrett’s neoplasia.

Conflict of Interest

J.D. Mosko is a speaker for Boston Scientific, Pendopharm, SCOPE rounds, Vantage, and Medtronic; medical advisory board member for Pendopharm, Boston Scientific, Janssen, Pentax, and Fuji; has received grants and research support from CAG. G.R. May is a consultant for Olympus, and a speaker for Pentax, Fuji, and Medtronic. C.W. Teshima is a speaker for Medtronic, Fujifilm, and Boston Scientific, and a consultant for Boston Scientific and Olympus. Y. Fujiyoshi, K. Khalaf, D. Tham, M.R.A. Fujiyoshi, and N.C. Calo declare that they have no conflict of interest.

• References

• 1 de Jonge PJF, van Blankenstein M, Looman CWN. et al. Risk of malignant progression in patients with Barrett’s oesophagus: a Dutch nationwide cohort study. Gut 2010; 59: 1030-1036
• 2 Yousef F, Cardwell C, Cantwell MM. et al. The incidence of esophageal cancer and high-grade dysplasia in Barrett’s esophagus: a systematic review and meta-analysis. Am J Epidemiol 2008; 168: 237-249
• 3 Shaheen NJ, Falk GW, Iyer PG. et al. ACG Clinical Guideline: Diagnosis and management of Barrett’s esophagus. Am J Gastroenterol 2016; 111: 30-51
• 4 Weusten BLAM, Bisschops R, Dinis-Ribeiro M. et al. Diagnosis and management of Barrett esophagus: European Society of Gastrointestinal Endoscopy (ESGE) Guideline. Endoscopy 2023; 55: 1124-1146
• 5 Forbes N, Elhanafi SE, Al-Haddad MA. et al. American Society for Gastrointestinal Endoscopy guideline on endoscopic submucosal dissection for the management of early esophageal and gastric cancers: summary and recommendations. Gastrointest Endosc 2023; 98: 271-284
• 6 von Elm E, Altman DG, Egger M. et al. STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol 2008; 61: 344-349
• 7 Vieth M, Stolte M. Pathology of early upper GI cancers. Best Pract Res Clin Gastroenterol 2005; 19: 857-869
• 8 Stolte M, Kirtil T, Oellig F. et al. The pattern of invasion of early carcinomas in Barrett’s esophagus is dependent on the depth of infiltration. Pathol Res Pract 2010; 206: 300-304
• 9 Sampliner RE. Practice guidelines on the diagnosis, surveillance, and therapy of Barrett’s esophagus. The Practice Parameters Committee of the American College of Gastroenterology. Am J Gastroenterol 1998; 93: 1028-1032
• 10 Kobayashi R, Calo NC, Marcon N. et al. Predictors of recurrence of dysplasia or cancer in patients with dysplastic Barrett’s esophagus following complete eradication of dysplasia: a single-center retrospective cohort study. Surg Endosc 2022; 36: 5041-5048
• 11 Terheggen G, Horn EM, Vieth M. et al. A randomised trial of endoscopic submucosal dissection versus endoscopic mucosal resection for early Barrett’s neoplasia. Gut 2017; 66: 783-793
• 12 Mejia Perez LK, Yang D, Draganov PV. et al. Endoscopic submucosal dissection vs. endoscopic mucosal resection for early Barrett’s neoplasia in the West: a retrospective study. Endoscopy 2022; 54: 439-446
• 13 Doumbe-Mandengue P, Pellat A, Belle A. et al. Endoscopic submucosal dissection versus endoscopic mucosal resection for early esophageal adenocarcinoma. Clin Res Hepatol Gastroenterol 2023; 47: 102138
• 14 Fujiyoshi Y, Khalaf K, He T. et al. Comparison of endoscopic mucosal resection versus endoscopic submucosal dissection for Barrett’s neoplasia and esophageal adenocarcinoma: a systematic review and meta-analysis. Gastrointest Endosc 2024; 2024

Publication History

Received: 04 April 2024

Accepted after revision: 15 July 2024

Article published online:
03 September 2024