Recurrence rate of premalignant and early malignant lesions of the gastrointestinal tract following endoscopic submucosal dissection: a single-centre cohort

Endoscopic submucosal dissection (ESD) is an advanced endoscopic technique achieving en bloc resection of premalignant and early malignant lesions within the gastrointestinal tract, which previously had necessitated surgical removal. Originally pioneered in Japan, ESD has gained increasing traction in Western countries, including New Zealand, as institutional expertise has expanded and endoscopic technology has advanced.^1,2

ESD has many advantages compared to endoscopic mucosal resection as it enables more comprehensive histopathological evaluation of lesions, which facilitates precise staging and subsequent risk stratification and management.^3,4 The technique is also particularly advantageous in significantly reducing the need for surgical intervention, and offers organ preservation, reduced post-operative morbidity and a curative alternative for a selected group of patients who may otherwise require extensive surgery.^3,4

However, despite its technical advantages, the risk of recurrence remains a concern, particularly in cases with incomplete resection, deep submucosal invasion, positive resection margins or other high-risk histological features.^5,6

The recurrence rate following ESD varies considerably across studies, with influencing factors including lesion location and size, histological subtype and resection margin status, as well as a centre’s yearly case volume.^5–7 While expert centres report high rates of en bloc and R0 resection, local recurrence remains a challenge, particularly for oesophageal lesions, which exhibit a greater propensity for residual neoplasia or metachronous recurrence due to field cancerisation effects.^8–11 Understanding these recurrence patterns is paramount in refining patient selection criteria, optimising procedural techniques and enhancing post-procedural surveillance to ensure better long-term clinical outcomes.

This study aimed to evaluate the recurrence rate of premalignant and early malignant lesions of the gastrointestinal tract following ESD in a single-centre cohort at Middlemore Hospital (MMH), New Zealand. We sought to stratify recurrence rates according to lesion characteristics and resection status, with an emphasis on benchmarking our outcomes against internationally recognised recurrence rates, with a target rate of ≤5%.^3,7

Methods

Study design and patients

This was a retrospective cohort study conducted in a single centre of MMH in Auckland, New Zealand. MMH provides interventional endoscopy services for the Health New Zealand – Te Whatu Ora Counties Manukau catchment area, caring for a population of approximately 600,000.

In our centre, AS is one of two interventional endoscopists performing ESD. All consecutive adult patients (age ≥18 years) who underwent ESD for premalignant or early malignant lesions by a single endoscopist (AS) at MMH between 11 February 2019 and 6 October 2023 were eligible for the study. Patients were excluded if they had not had their surveillance endoscopy following the ESD. Patient notes were reviewed for surveillance endoscopy between October 2023 and October 2024. The primary outcome was recurrence rate of premalignant and early malignant lesions of the gastrointestinal tract following ESD.

Recurrence was defined as confirmed neoplasm on histopathology on first follow-up surveillance endoscopy. The target recurrence rate was less than or equal to 5% based on comparative international data. Secondary outcome was to determine recurrence rate stratified by location of the lesion (oesophagus, gastric, right colon, left colon/rectum), size of the lesion (≥20mm or <20mm), en bloc resection status, R0 resection status and histopathological type of lesion.

Prior to proceeding to ESD, all cases were discussed at a multidisciplinary meeting (MDM). The majority of cases were referred with intent of curative resection; however, some cases were referred for staging. Procedures were undertaken using either conscious sedation with fentanyl and midazolam or under general anaesthesia. ESD was performed using DualKnife J (Olympus America). Each ESD specimen was reviewed by a specialised gastrointestinal pathologist, with all histology results reviewed in an MDM.

For carcinomas, R0 resection was defined as a resection specimen with radial and deep margins clear of dysplasia or cancer, while R1 was defined as a specimen with presence of dysplasia or cancer at the margin.

For patients who had surveillance endoscopy performed, the resection scar site was identified and examined using magnification endoscopy with white light and narrow band imaging, with any abnormalities sampled with biopsy forceps.

Locality authorisation approval was gained from the Counties Manukau Health Research Office prior to the commencement of this study.

Data collection

Patient cases for this study were recorded in a prospectively collected database of ESD procedures. Medical records of these cases were reviewed, and the variables collected included age, gender, ethnicity, location and size of index lesion, en bloc resection status, R0 resection status, histopathological type of lesion and date of surveillance endoscopy. If follow-up surveillance endoscopy had occurred, endoscopy and histology data were collected to determine recurrence.

Statistical analysis

Recurrence rate following ESD was calculated as the number of patients with premalignant and early malignant lesion recurrence divided by the total number of patients treated with ESD who had undergone a surveillance endoscopy. A 95% confidence interval (CI) for the recurrence rate was calculated using the exact binomial method. Categorical data were expressed as counts and percentages. Continuous data were expressed as median (interquartile range [IQR]).

Results

Study participants

A total of 119 ESD procedures were completed during the study timeframe, with 91 (76.5%) having had a surveillance endoscopy following the ESD. Twenty-eight cases did not have surveillance endoscopy completed for a variety of reasons (five deceased, six due to patient factors, seven had surgical resection of area, three pending follow-up in future, seven unknown) as shown in Figure 1.

The cause of death of the five patients who were deceased prior to follow-up surveillance endoscopy included two with aspiration pneumonia (following stroke and acute subdural haematoma), one with aortic graft infection, one with metastatic duodenal adenocarcinoma found after ESD of rectal lesion and one unknown.

Seven patients had surgical resection following ESD. Of these, three were to treat a separate synchronous lesion. The remaining four were due to positive deep histological ESD margins, three of which had no residual cancer found in the resected specimens. The one case that had residual cancer in the specimen at the ESD site had an oesophagectomy after the gastro-oesophageal junction lesion was found to invade the muscle layer during the ESD procedure.

Baseline characteristics for the study population are shown in Table 1. The median age of the population was 68 years; 36 (39.6%) were female and 49 (53.8%) were New Zealand European, which was the most common ethnicity group. Of the 91 cases included in the analysis, seven were oesophageal, 26 were gastric, four were right colonic (caecum to distal transverse) and 54 were left colonic (splenic flexure to rectum). Most (87.9%) were >20mm in size. Initial en bloc resection was achieved in 80 of the 91 cases (87.9%) with histopathological R0 resection confirmed in 64 cases (70.3%). Median time to surveillance endoscopy was 231 days.

View Figure 1, Table 1–2.

Recurrence rate

The recurrence rate was 3.3% (95% CI: 0.7–9.3%), with three cases identified on follow-up endoscopy. Two recurrences were oesophageal squamous cell carcinomas, while the other was a rectal sessile serrated adenoma with dysplasia. Of the oesophageal recurrences, one had a malignant stricture with local lymph node involvement, which was treated with palliative radiotherapy, and the second was treated with radiofrequency ablation therapy with no features of recurrence on subsequent surveillance. The index resection for the rectal lesion that had recurrence did not achieve en bloc or R0 resection status. The recurrence was treated with repeat ESD with no recurrence on subsequent surveillance endoscopy. The secondary outcome of recurrence rate stratified by age (≥70 or <70), gender, ethnicity, location of lesion (oesophagus, gastric, right colon, left colon/rectum), size of lesion (≥20mm or <20mm), en bloc resection status and R0 resection status are shown in Table 2, although due to low numbers these did not meet statistical significance.

Discussion

ESD is a technically advanced resection technique, and although there are comprehensive data on recurrence rates from Asian countries and more recently from Western countries, there remain limited data from New Zealand centres. Our study demonstrated a recurrence rate of 3.3% following ESD, within our target benchmark of ≤5%. A study from a registry in Germany revealed recurrence rates of 2.3% at 3 months and 2.1% at 12 months,⁷ which is comparable to our recurrence rate. This finding reinforces the role of ESD as an effective curative intervention for appropriately selected premalignant and early malignant gastrointestinal lesions in a New Zealand cohort of patients in a centre with appropriate endoscopic expertise.

Due to our low number of recurrence cases we were unable to ascertain any statistically significant associations with regard to our secondary outcome variables.

The majority of lesions in our cohort were colorectal and gastric, and our study demonstrated very low rates of recurrence in these groups—1.85% and 0% respectively—comparable with internationally reported rates of 0.5% for colorectal lesions⁵ and 3.1% for gastric lesions.⁶ This reinforces the efficacy of ESD in these locations at our centre, after careful endoscopic lesion characterisation and MDM discussion.

In contrast, oesophageal lesions exhibited the highest recurrence rate, with two found in oesophageal squamous cell carcinoma. This is consistent with prior reports indicating an elevated risk of both local recurrence and metachronous neoplasia in oesophageal squamous cell carcinoma.^8,9,11 This heightened risk may be attributed to several factors, including the intrinsic biological aggressiveness of oesophageal squamous cell carcinoma and field cancerisation effects.^8,9,11 These results suggest that tailored surveillance strategies may be required to mitigate recurrence risks in high-risk lesions, particularly in the oesophagus. ESD still remains a preferred initial treatment option of superficial oesophageal squamous cell carcinoma compared to surgery, even for T1b lesions demonstrating submucosal invasion, with lower complication rates and shorter hospital stay.¹⁰

Our findings further corroborate the role of achieving R0 resection in reducing recurrence risk. None of the patients who achieved R0 resection experienced recurrence, whereas those with R1 resection had an 11.1% recurrence rate. This observation aligns with previous literature indicating that incomplete resection significantly increases the risk of local recurrence, often necessitating additional interventions such as repeat endoscopic resection or surgical excision.¹²

However, we do recognise that our sample size was relatively small, and other larger studies have demonstrated conflicting findings, with no association of recurrence rate with R0 resection status or positive horizontal margins for premalignant non-invasive colorectal lesions.^13,14

Regardless, meticulous endoscopic technique remains paramount, particularly in challenging anatomical locations where technical difficulties may compromise the likelihood of achieving complete resection. The role of R0 resection as a quality reporting measure is an area that warrants further exploration and research.

Due to our low number of recurrence cases we were unable to ascertain any statistically significant associations with respect to patient-related factors, including age, gender and ethnicity. Understanding these potential associations may allow for more personalised surveillance protocols tailored to high-risk patient sub-groups.

The strengths of our study include being a real-world study with a well-defined patient cohort, standardised procedural techniques and histopathological assessments, all of which enhance the reliability of our findings. However, several limitations must be acknowledged. As a single-centre retrospective study, our results may not be generalisable to other institutions, particularly those with differing patient demographics or endoscopist expertise. Additionally, only 76.5% of our cohort had follow-up surveillance endoscopy after ESD in our study timeframe, introducing a potential bias in recurrence estimation. Of the 28 patients who did not undergo follow-up, five were deceased and all cases were felt unrelated to the ESD procedure or recurrence, and seven underwent surgical resection of the tissue, of which only one case had residual tumour at the ESD site. Furthermore, our median time to surveillance endoscopy was 231 days, and as such later recurrence is a possibility and may not have been captured in our study. The total number of recurrences in our cohort was small, and as such was not powered to detect any statistically significant differences in other collected variables.

Conclusion

Our study supports the efficacy of ESD as a curative treatment modality for premalignant and early malignant gastrointestinal lesions in a real-world New Zealand setting, demonstrating a recurrence rate within the acceptable international benchmark. The findings highlight the necessity for rigorous surveillance, particularly for high-risk lesions such as oesophageal squamous cell carcinoma. Future research should focus on refining post-ESD surveillance protocols, identifying risk stratification tools and exploring adjunctive therapeutic strategies to further mitigate recurrence risks and optimise patient outcomes.