Long-term outcomes of endoscopic submucosal dissection for gastric dysplasia and early neoplasia in a United Kingdom Caucasian population.

INTRODUCTION
Gastric cancer poses a major public health challenge, ranking fifth for incidence and fourth for cancer-related deaths worldwide[1-3]. In the West, prognosis is poor compared to Eastern countries like Korea or Japan, with five-year survival rate of 20% and ten-year around 17%[4-6]. Endoscopic resection, consisting endoscopic mucosal resection (EMR) and endoscopic submucosal dissection (ESD), is effective at treating gastric dysplasia and early neoplasia[7,8]. Endoscopic resection offers advantages like shorter hospital stays, fewer complications and most importantly, stomach preservation[9].
ESD has higher en-bloc and histologically complete resection rates, lower local recurrence and longer survival than EMR[10-14]. However, ESD has increased risks of complications like bleeding, perforation, and stricture formation[15,16]. Hybrid ESD, an intermediate technique between EMR and ESD, is sometimes performed to improve curative resection (CR) rates compared to EMR, and technical difficulties associated with ESD[17]. Despite its proven efficacy, only a few tertiary centres practise upper gastrointestinal ESD in the United Kingdom due to its highly specialist and technically demanding nature. Data concerning the United Kingdom Caucasian population is limited. This is concerning because there are genetic and morphological differences between early gastric cancers (EGCs) from Western and Eastern populations[18-20]. The aim of this study is to evaluate the efficacy and long-term outcomes of ESD in treating gastric dysplasia and neoplasia in a Caucasian population at a United Kingdom tertiary referral centre over a ten-year period. This expands on our previously published 2-year and 5-year data[21,22]. This will add significant clinical value to patient care, particularly around long-term survival for patients in the non-CR group.

MATERIALS AND METHODS
This is a retrospective study of Caucasian patients undergoing gastric ESD between May 2012 and July 2023 at Salford Royal University Hospital (Northern Care Alliance) under the hospital’s usual audit and service improvement process. Hence, no ethics application is required after institutional review. Inclusion criteria: Only Caucasian patients with gastric dysplasia and early neoplasm on histology and tumour stage (T1N0M0 or below) on computed tomography (CT) were included. Exclusion criteria: Lesions in the gastro-oesophageal junction and patients staged T2N0M0 and above were excluded.

Pre-ESD
Mapping oesophagogastroduodenoscopy was carried out to determine if lesions were amenable to ESD, using the Kato classification which is how liftable the lesion[23]. Kato 1 indicates that the lesion lifts without any resistance, Kato 2 suggests some resistance to lifting, while Kato 3 means lesion is not liftable[23]. Only Kato 1 or 2 Lesions were included. Lesion features (size, location, ulceration) were recorded. The locations were coded as upper (cardia and fundus), mid (body) and lower (antrum, pylorus and incisura) stomach[22]. Biopsies were taken for histological analysis. Only lesions with definite cancer were staged using iodine-based contrast CT.

ESD procedure
All procedures involved at least two experienced endoscopists using Olympus Double Channel Double-Headed Scope or Olympus single channel therapeutic Endoscope. Argon plasma was used to mark out the circumference around the lesion, with minimum 5 mm margins. Afterwards, EMR solution (indigo carmine and 0.1% lidocaine) was injected outside the marked dots and a mucosal incision was done. To improve safety, the EMR solution is injected into the submucosal layer to lift the lesion. Hyaluronic acid was injected into the submucosa if lifting was inadequate. The submucosal layer was then dissected and image enhancing techniques were used if necessary. For hybrid ESD cases, after partial submucosal layer dissection, the lesion was snared wholly. This is done when we need to finish the procedure faster (i.e. due to deterioration in clinical conditions from anaesthetic complications).

Post-ESD
All specimens were retrieved, pinned onto a specimen board, fixed with formalin, and sprayed with hyoscine before histological analysis. Histological reports were used to assess if the lesions achieved CR, non-CR or were indeterminate for CR. Resection is considered curative if the resected lesion meets all the following criteria on histology: (1) Peripheral and deep margins are clear; (2) No lympho-vascular invasion; and (3) No submucosal involvement[21,22] Resection is non-CR (high-risk) if it fails to meet at least 1 of the criteria, and indeterminate if data is insufficient to conclude CR or non-CR.
All patients were given oral proton pump inhibitors twice daily for at least three months after the procedure and scheduled for endoscopic surveillance. Data was collected using endoscopic and histological reports from electronic patient records. This was subsequently recorded and analyzed using Microsoft Excel. The Kaplan-Meier survival curve was created using R software (version 4.4.0, R Foundation for Statistical Computing).

RESULTS
Between May 2012 and July 2023, ESD was attempted on 111 Lesions in 93 Caucasian patients. Patient demographics and features of lesions with ESD attempted are presented in Table 1. The lesions were mainly in the lower stomach (58.6%) and had a mean diameter of 30.4 mm (Table 1). The largest 100 mm lesion spanned across the superior antrum, extending into the distal body and angularis, occupying 70% circumference. ESD was attempted but not feasible due to its positioning and lesion size. About 20% of the lesion were partially resected using EMR for tissue acquisition and histology to guide future management. Pre-ESD histological analysis demonstrated that most lesions were either low-grade dysplasia (38.7%) or high-grade dysplasia (36.9%). 22.5% of lesions were intramucosal adenocarcinoma (Table 1).

Outcomes of completed ESD
ESD was completed for 91.0% of all attempted lesions (Table 2). Out of the 101 successfully completed ESD procedures, 74.3% achieved en-bloc resections while the other 25.7% were hybrid ESD with piecemeal resections (Table 2). Reasons for adopting a piecemeal approach were mainly due to large lesion size, difficult positioning around the pylorus, excessive fibrosis and active bleeding. 95.0% of completed procedures achieved endoscopic clearance (Table 2). Reasons for the five cases of incomplete resections on endoscopy include: Extensive fibrosis of the base due to possible neoplasia infiltration, irregular polypoid surface requiring snare resections, active bleeding and prolonged procedure from extensive submucosal fat and large polyp size. 34.7% of completed procedures achieved histological CR (Table 2). Non-CR and indeterminate resections were 33.7% and 28.7% respectively (Table 2). Reasons for the 29 Indeterminate cases include end pieces affected by dysplasia hampering assessment, poor tissue preservation and piecemeal resections (Table 2). Many resections were considered indeterminate due to the piecemeal nature of hybrid ESD specimens and diathermy changes at the peripheries hampering assessment. A minority of indeterminate samples were due to inflammatory changes and poor sample preservation.

Five-yearly comparisons of ESD outcomes
The number of completed ESD procedures in 2018-2023 was twice that of 2012-2017. (Table 3). Additionally, en-bloc resection rates rose from 57.6% to 82.4%, while piecemeal resection rates decreased from 42.4% to 17.6% (Figure 1A, Table 3). Endoscopic clearance increased from 87.9% in 2012-2017 to 98.5% in 2018-2023 (Figure 1B, Table 3). This was accompanied by nearly five-fold increase in histological CR (Table 3).

Immediate and late complications
Immediate complications related to the procedure (< 48 hours) were observed in 9.9% of procedures, with bleeding and perforation accounting for 3.6% and 6.3% of the cases respectively. Late complications (> 48 hours post procedure) were seen in two patients. One patient experienced delayed gastric perforation confirmed by CT three days later, possibly due to muscle injury from diathermy effect. There was no evidence of perforation upon review of endoscopic photos. A patient with aborted ESD due to intraprocedural perforation was transferred to a high-dependency unit and managed with nasogastric tube and antibiotics. A contrast CT later revealed an ongoing upper gastrointestinal bleed, and the patient was subsequently sent for laparotomy and polyp excision.

Pre-ESD and ESD histological changes
Histology of 101 lesions before and after ESD completion was analyzed (Figure 2A, Supplementary Table 1). ESD changed the histology of 44.5% of lesions – 28.7% of lesions were upstaged and 15.8% were downstaged (Figure 2B). 55.4% had no changes in histology (Figure 2B). Dysplasia was absent or indefinite in three lesions on index histology. Upon review, one of the external pre-ESD biopsies undertaken did not show any dysplasia or early neoplasia, which was concordant with the subsequent ESD histological sample. The other two lesions had genuine downstaging from high-grade dysplasia and intramucosal adenocarcinoma respectively; the reasons remain unclear.

Disease recurrence
Overall disease recurrence was 10.9% at latest follow-up (63 months, median follow-up) (Table 4). 100% of lesions in the CR group and 47.8% in the Indeterminate group had no disease recurrence on future follow-up (Table 4). 43.2% of lesions in the Indeterminate and Non-CR group showed disease recurrence at subsequent follow-ups (Table 4).

Survival
The median and mean survival duration was 69 months and 48.9 months post-ESD attempt, with a 95% confidence interval between 60 and 111 months (Figure 3). The mean age at death was 82.8 years. No patient died within two months of ESD. Overall, 32 out of 93 patients who had ESD attempted died within our ten-year study – 14 people died from non-disease-related causes, 11 patients died from disease-related causes and 7 patients did not have a known cause of death. Of the 11 patients who had known disease-related causes, six had metastatic gastric cancer and one had primary synchronous gastric and oesophageal lesions.

Surgery
Overall, surgery was considered in 21 patients and completed in 9 patients (1 indeterminate, 5 non-CR, 3 aborted). The remaining 12 patients were listed for endoscopic surveillance or conservative management because of medical co-morbidities. The outcomes and complications from surgery in patients from the Indeterminate, Non-CR and Aborted ESD group are listed in Supplementary Table 1.

Aborted ESD cases
ESD was aborted in nine patients (ten lesions) for reasons such as excessive scarring and fibrosis, concern for perforation, difficult positioning and profuse bleeding (Supplementary Table 2). Six patients were not offered surgery due to medical unfitness. Two patients were managed conservatively and one patient underwent endoscopic surveillance only. 2 patients received argon therapy for residual low-grade dysplasia. Unfortunately, one patient died before endoscopic follow-up. Three patients underwent surgery and all had complete disease resolution at time of data collection.

DISCUSSION
Using parameters recommended by the National Institute for Health and Care Excellence[24], we evaluated the efficacy of ESD procedures on gastric lesions using en-bloc and CR, local recurrence and survival rates. The diagnostic efficacy of ESD was also demonstrated by comparing pre- and post-ESD histology. In this study, we demonstrated that 74.3% of completed ESD procedures achieved en-bloc resections, which is higher than our previously published five-year study[22]. However, this figure is lower than that of a recent meta-analysis comparing the outcomes between Eastern and Western countries and a recent ten-year multi-centre study in the West[25,26]. This could be attributed to lesion size, anatomical location, extensive fibrosis and bleeding. Moreover, 39.6% of lesions were resected using Hybrid ESD, which is associated with higher rates of piecemeal resections[26]. As a whole, 95% of completed resections achieved endoscopic clearance, which is also higher than our previously published rate of 86%[22].
Currently, ESD is considered for selected patients on a case-by-case basis through multi-disciplinary team discussions based on its absolute and expanded indications laid out in the European Society of Gastrointestinal Endoscopy guidelines 2022: ESD for superficial gastrointestinal lesions[27]. As our data was from ten years ago, we did not use the European Society of Gastrointestinal Endoscopy 2022 guidelines for categorizing lesions as CR or non-CR[27]. Histological CR was achieved in 34.7% of all lesions, which is higher than our previously published 19% histological CR rate[22]. Although 33.7% of resections were histologically non-CR, this high rate could be due to 14 lesions being pT1b lesions which are not resectable endoscopically.
Nonetheless, our ten-year analysis showed improvements in efficacy outcomes over the two time periods which could be due to accumulated experience and improvements in the technical skills of endoscopists over the years. This upward trend is likely to continue over time, with many studies highlighting improvements in endoscopic clearance and en-bloc rates over ten years[26]. The number of completed ESD procedures also more than doubled during this time due to the completion of more referrals from reduced procedural durations.
Overall disease recurrence was 10.9% at latest follow-up (63 months, median follow-up). There was no disease recurrence from the CR group, demonstrating the effectiveness of ESD when histological CR is achieved, mainly a result of en-bloc resections. As shown in the indeterminate group, even when there is uncertainty surrounding CR or possible involved margins, many patients are still cured with 69.6% showing no recurrence at latest follow-up. Immediate complications were seen in 9.9% of patients - bleeding (3.6%) and gastric perforation (6.3%). These figures are comparable to the reported range of 0%-15.6% and 1.2%-5.2% for bleeding and perforation ESD-related complications respectively[28]. All immediate and late complications were resolved.
ESD could also have diagnostic properties[29]. The histology of 44.5% lesions was changed. Important therapeutic interventions could have been missed or delayed if clinicians were unaware of some lesions upstaging to adenocarcinoma with submucosal invasion, also demonstrated in our previous study. Surgery was performed in 9 patients although it was considered in 21 patients due to personal choice and medical comorbidities.
In total, 32 out of 93 patients died within our ten-year study. The mean age at death is 82.8 years old which is well within the average life expectancy in the United Kingdom (79.0 years old for males and 82.9 years old for females)[30]. The median survival time was 69 months, almost 7 years, post-ESD attempt. These patients would have likely died from old age or other health co-morbidities. The average duration between the latest ESD and death was 48.9 months, which is a long duration given that the mean age of our patients was 75 years old. With limited treatment options, ESD is a safe procedure that can help to prolong their lifespan without debilitating effects on their quality of life. No patient died within two months of ESD so it is highly unlikely their death was a direct consequence of ESD. Moreover, CR is dependent on many histological factors, but none of which is necessarily predictive of the 10-year survival as our whole ESD cohort has comparable life expectancy to the general population.
As demonstrated in our study, ESD results in shorter hospital stays (1.60 days) and fewer complications, as compared to subtotal and total gastrectomies with reported hospital stays of 13.8 days to 15.4 days and higher morbidity[31,32]. The main limitation of our study was the limited patient pool from a single tertiary institution. More data should be collected from multiple tertiary institutions across the United Kingdom to gain a more holistic viewpoint in the country as the effectiveness of ESDs are known to be highly variable and multifactorial[33]. Despite this limitation, given the small number of ESD procedures carried out in the United Kingdom, our sample size is relatively large. To our knowledge, this is the first single-center study in the United Kingdom that looks at ten-year outcomes of ESD in the treatment of gastric dysplasia and neoplasia in the United Kingdom.

CONCLUSION
Overall, there are clear improvements in efficacy outcomes of gastric ESD over the past ten years with acceptable rates of immediate and late complications. Our long-term follow-up study adds to the growing database of Caucasian patients undergoing gastric ESD in the United Kingdom. Overall, ESD is demonstrated to be safe and effective, and plays a pivotal role in treating early gastric neoplasia and dysplasia.