Management Strategy of Non-curative ESD in Gastric Cancer: Curative Criteria, and the Critical Building Block for Determining Beyond It

Gastric cancer remains the fourth leading cause of cancer-related death worldwide. However, its mortality has significantly declined because of aggressive screening programs and the eradication of Helicobacter pylori infection. In addition to such a downward trend in mortality, the diagnosis of gastric cancer at an early stage has enabled the significantly more frequent use of endoscopic resection, especially endoscopic submucosal dissection (ESD), when appropriate indications are met. Therefore, ESD is now acknowledged as a common treatment modality for early gastric cancer (EGC) [1, 2, 3, 4, 5, 6, 7, 8]. However, ESD should be limited to cases in which the risk of lymph node metastasis (LNM) is negligible, necessitating a thorough post-ESD pathological evaluation to estimate the risk and consider other rescue therapies if the pathological curability criteria are not met [9]. Thus, the pathological curability criteria to identify non-curative resection following ESD are important, and they are crucial to guide further treatment depending on the specific pathological parameters of the criteria for non-curative resection. In addition, patient-related factors must be considered when making treatment decisions [10]. Interest in therapeutic strategies for managing non-curative ESD has been growing, particularly owing to the high incidence of upfront ESD in older adults or patients with severe comorbidities related to perioperative risks associated with gastrectomy [11]. This review aimed to summarize the key pathological criteria for determining non-curative resection after ESD, explore appropriate treatment modalities for non-curative resection, and review the implementation of treatment strategies considering patient-related factors and lesion characteristics.

Curative resection following ESD is defined as the absence of tumor involvement in the horizontal or vertical resection margins, lymphovascular invasion, or complete resection of lesions that meet the absolute or expanded indications. Lesions that meet the absolute or expanded indications include the following: 1) differentiated intramucosal carcinoma without ulceration, regardless of tumor diameter; 2) differentiated intramucosal carcinoma with ulceration, with a tumor diameter of ≤3 cm; 3) differentiated submucosal carcinoma with an invasion depth of <500 µm, with a tumor diameter of ≤3 cm; and 4) undifferentiated intramucosal carcinoma without ulceration, with a tumor diameter of ≤3 cm. Lesions that do not meet these criteria are considered non-curative [2, 12].

The criteria for non-curative resection proposed by traditional guidelines do not account for the differences between the various factors contributing to non-curative status. Recently, the Japanese guidelines introduced the e-Cura system, which categorizes the non-curative resection group (e-CuraC group) into e-CuraC-1 and e-CuraC-2. Tailored recommendations for additional treatment were provided for the 2 groups in response to different long-term outcome risks [13, 14]. In practice, the rate of non-curative resection following ESD ranges from 12% to 21% across facilities, depending on their preference for performing upfront ESD or adopting a more conservative approach for case selection. For example, a multicenter study in Korea reported a non-curative resection rate of 21.4% (661 cases) in 3,094 ESD cases. Similarly, a prospective cohort study analyzing outcomes from 41 facilities in Japan using a web registry system reported a non-curative resection rate of 20.0% [15, 16].

In cases where additional treatment is required following non-curative endoscopic resection but high-risk perioperative morbidity is anticipated due to comorbidities, observation may be chosen over additional gastrectomy with lymph node dissection. However, long-term follow-up data on the natural course of untreated non-curative resection remain limited. Some studies have explored the clinical outcomes of patients with differentiated EGC who underwent non-curative resection and were followed up without additional rescue treatment. One study with a median follow-up period of 33 months reported a 25.2% mortality rate, with 85% of deaths attributed to causes unrelated to gastric cancer, and underlying disease was identified as a key factor linked to mortality [17]. Another study of patients with submucosal invasive cancer who underwent non-curative resection and were placed on long-term follow-up without additional treatment reported a cancer recurrence rate of 9.6%. In this study, mortality due to non-cancer-related causes was significantly higher than that due to gastric-cancer-specific causes (13.5% vs. 1.9%) [18]. In contrast, a study analyzing the long-term outcomes of 512 patients who underwent non-curative resection found that among 198 patients who were observed without additional treatment, cancer-specific survival and disease-free survival were significantly worse than those who underwent gastrectomy [19]. A long-term follow-up study from Japan reported that while gastric cancer-related deaths did not significantly increase in patients who did not receive additional treatment, the mortality rate due to other underlying diseases was significantly higher. This increase is attributed to the nature of these patients, who were unable to undergo surgical treatment because of their underlying conditions [20].

The long-term outcomes of patients placed under observation without additional treatment following non-curative resection are subject to selection bias, as they were obtained from retrospective cohort studies. These mixed results reflect the varying impacts of different causes of non-curative resection on long-term prognosis. For instance, the prognosis of patients who undergo non-curative resection due to lateral margin involvement differs significantly from that of patients with lymphovascular invasion. Hence, outcomes differed according to the composition of the patient cohort. To address these differences, the eCura system was introduced to stratify the risk of non-curative ESD cases based on the eCura score. The low- and intermediate-risk groups did not show better overall survival rates after observation than after surgery. However, in high-risk groups, the 5-year overall survival and recurrence-free survival rates were significantly lower in the observation group [21]. A study analyzing the timing of recurrence in patients under observation after non-curative resection reported that early recurrence (within 2 years) was associated with lymphatic invasion, whereas late recurrence (after 2 years) was linked to venous invasion. This suggests that treatment planning should consider life expectancy and the likelihood of recurrence in each patient [22].

Among patients who showed positive horizontal margin involvement and were under long-term follow-up without additional treatment, the local recurrence rate ranged from 11.9% to 25.9%, and local recurrence was associated with cancer-positive lateral margin length [23, 24, 25, 26].

Figs. 1 and 2 illustrate cases of gastric cancer-related mortality, whereas Fig. 3 shows a case of non-cancer-related mortality during long-term follow-up without additional treatment following non-curative ESD for EGC. Fig. 4 presents a case of mortality due to non-gastric cancer in a patient who underwent gastrectomy after non-curative ESD.

Fig. 1
Representative case of mortality in a patient with only follow-up after non-curative ESD. A 73-year-old female with a Charlson Comorbidity Index score of 0 underwent ESD for an early gastric cancer IIb + IIc lesion at an angle detected during esophagogastroduodenoscopy as part of a routine health checkup (A, B). Final pathology confirmed a 4.8 cm papillary adenocarcinoma with 200 μm submucosal and lymphatic invasion (C). Although the patient was in the intermediate-risk group for lymph node metastasis, she refused additional surgery. During follow-up, recurrent gastric cancer was detected at the ESD site after 40 months (D). Abdominal computed tomography showing peritoneal nodules and omental-mesenteric infiltrations compatible with peritoneal carcinomatosis (E). The patient received palliative chemotherapy but died 11 months later because of disease progression.
ESD = endoscopic submucosal dissection.

• Click for larger image

Fig. 2
Representative case of mortality in a patient with only follow-up after non-curative ESD. A 74-year-old female with a Charlson Comorbidity Index score of 0 underwent ESD for an early gastric cancer IIc + IIa lesion in the posterior antrum, which was discovered during esophagogastroduodenoscopy as part of a routine health checkup (A, B). Final pathology confirmed a 2.6 cm moderately differentiated tubular adenocarcinoma with 1,000 μm of submucosal invasion, lymphatic invasion, and positive vertical margin invasion (C). Despite being in the high-risk group for LNM, the patient refused additional surgery and was lost to follow-up. The patient returned 30 months post-ESD. Abdominal computed tomography revealed extensive LNM in the abdominopelvic cavity and metastatic soft tissue density in the iliac bifurcation and prevertebral body areas (D). Chest computed tomography showed metastatic lung nodules (E). The patient died 2 months later.
ESD = endoscopic submucosal dissection; LNM = lymph node metastasis.

• Click for larger image

Fig. 3
Representative case of mortality in a patient with only follow-up after non-curative ESD. An 86-year-old male with a Charlson Comorbidity Index score of 3 underwent ESD for an early gastric cancer IIa + IIb lesion in the posterior lower body, which was discovered during esophagogastroduodenoscopy as part of a routine health checkup (A, B). Final pathology revealed a 6.4 cm moderately differentiated tubular adenocarcinoma with 2,400 μm submucosal and lymphatic invasion (C). The patient was in the high-risk group for lymph node metastasis but did not undergo additional surgery. The patient died 7 months after experiencing pneumonia with sepsis.
ESD = endoscopic submucosal dissection.

• Click for larger image

Fig. 4
Representative case of mortality in a patient with additional surgery after non-curative ESD. A 75-year-old male with a Charlson Comorbidity Index score of 3 underwent ESD for an early gastric cancer IIa + IIb lesion on the lesser curvature of the lower body, which was discovered during esophagogastroduodenoscopy as part of a routine health checkup (A, B). Final pathology revealed a 2.8 cm moderately differentiated intramucosal tubular adenocarcinoma with multiple positive lateral margin invasions (C). The patient was in the low-risk group; however, subtotal gastrectomies B-I were performed (D). A 2.5 cm residual intramucosal tumor was noted, but there was no evidence of lymph node metastasis. The patient died of hemorrhagic stroke 26 months postoperatively.
ESD = endoscopic submucosal dissection.

• Click for larger image

Non-curative resection results from a single or combination of various pathological factors, such as tumor size, differentiated or undifferentiated cancer, submucosal invasion, presence of an ulcer, invasion of the horizontal or vertical resection margin, and presence of lymphovascular invasion. Thus, LNM rates in cases involving different factors are clinically important for treatment stratification. In a multicenter study, lymphatic invasion was identified as the most significant risk factor for LNM, followed by tumor size >30 mm, positive vertical margins, venous invasion, and deep submucosal invasion. Risk stratification was performed based on these risk factors, and the LNM rate ranged from 2.5% to 22.7% [13]. A meta-analysis of 24 studies involving 3,877 patients who underwent gastrectomy following non-curative ESD reported LNM rates ranging from 0% to 12%, with a pooled prevalence of 8.1%. These rates were similar across the countries in which the studies were conducted. One key conclusion of this meta-analysis was that the significant risk factors for LNM included lymphatic invasion, vascular invasion, deep submucosal invasion, positive vertical margins, and tumor size >30 mm. In contrast, patients with only positive horizontal margins showed negligible LNM risk. By demonstrating that LNM risk varies markedly between the e-CuraC-1 and e-CuraC-2 groups, this study substantiates the utility of the newly introduced e-Cura system [10]. A retrospective study in Korea that analyzed cases determined as non-curative resection based on factors other than lymphatic invasion, the most well-established risk factor for LNM, found an overall LNM rate of 7.1%, with tumor size greater than 3 cm, submucosal invasion, and undifferentiated cancer types being strongly associated with LNM [27]. In particular, the LNM risk associated with the undifferentiated type varies widely depending on various pathological criteria. For instance, while tumors <2 cm, without ulceration, and limited to the mucosa are considered indications for endoscopic resection, further stratification is required based on emerging pathological evidence [28, 29]. Recent meta-analyses have also demonstrated that the LNM rate of mixed-type intramucosal undifferentiated cancers is higher (7.4%–7.8%) than that of pure undifferentiated EGC, suggesting that individualized approaches are needed to select additional treatments for mixed-type cases [30, 31, 32, 33, 34]. In a recent multicenter retrospective cohort study of undifferentiated-type EGC, analysis using the e-Cura system showed LNM rates of 2.6%, 10.9%, and 14.8% in the low-, intermediate-, and high-risk categories, respectively. Using a modified e-Cura system, the LNM rates were 1.1%, 5.4%, and 13.3%, respectively [35, 36]. Table 1 shows the LNM rates among patients who underwent gastrectomy following non-curative ESD across various studies.

Table 1
LNM and local residual positive rates among patients who underwent gastrectomy following non-curative ESD for EGC

• Click for larger image
• Click for full table
• Download as Excel file

Treatment plans can be established based on various results regarding LNM rates in EGC, whether primary surgery or additional surgery after non-curative ESD. Generally, gastrectomy with lymph node dissection is the standard rescue therapy [64]. In a propensity score matching analysis comparing patients who underwent ESD without an indication for ESD and were placed only on follow-up without additional surgery to those who underwent initial standard surgery, the former group showed significantly higher 5-year mortality (26.0% vs. 14.5%) and cancer recurrence rates (17.0% vs. 0), demonstrating that additional gastrectomy offers survival benefits in non-curative ESD [76]. Another study also showed that additional surgery after non-curative ESD for EGC significantly reduced gastric cancer-related death (hazard ratio, 0.33; 95% confidence interval, 0.12–0.79) [77]. Similarly, in an analysis of 341 non-curative ESD cases, the status of rescue surgery was an independent predictor of 5-year overall survival (hazard ratio, 0.43; 95% confidence interval, 0.18–1.00) [64]. Table 2 lists studies comparing the effectiveness of additional treatment in non-curative ESD for EGC with observation groups.

Table 2
Long-term outcome after additional treatment compared with observation groups in non-curative ESD for EGC

• Click for larger image
• Click for full table
• Download as Excel file

However, it has been confirmed that the LNM rate varies according to several pathological risk factors involved in non-curative resection, and based on the e-Cura system that considers these factors, patients requiring gastrectomy are defined as a high-risk group [83]. In particular, guidelines have suggested that local treatment should be considered over gastrectomy in patients showing only positive horizontal resection margin involvement, given the negligible LNM rate in this patient group. Recent guidelines state that in cases where only positive horizontal resection margins are observed without other factors of non-curative resection, repeated ESD can be an option in addition to gastrectomy, and careful follow-up without any special additional treatment has been proposed as a viable approach [2, 12, 14, 92].

A multicenter survey of clinical approaches for patients with positive horizontal or indeterminable margins following ESD showed that only 42.8% of the patients underwent early intervention, whereas the rest were placed under observation. Among those who underwent early intervention, the most common treatment was ablation therapy through coagulation, followed by surgical resection or re-ESD, although the rates varied significantly across institutions. This suggests that treatment guidelines for managing positive horizontal or indeterminable margins after ESD are not well established [93]. Generally, patients with positive horizontal resection margins are not at significant risk of LNM but have a substantially higher likelihood of residual cancer. Therefore, additional resection, preferably with minimal excision, such as without lymph node dissection, may be necessary, as opposed to simple follow-up. Although additional endoscopic resection is ideal, technical challenges may arise due to fibrosis or swelling at the previous treatment site, or the location of the lesion may limit additional resection. In such cases, ablation therapy with argon plasma coagulation is often employed. An analysis of 76 patients with positive lateral margins who underwent non-curative resection showed that additional endoscopic treatment was sufficient regardless of absolute or expanded indications; however, 70.6% had residual cancer. This suggests that endoscopic resection is more appropriate than ablation therapy [58]. Similarly, another study comparing the long-term outcomes of patients with positive lateral margins who underwent repeated ESD, argon plasma coagulation, or surgery found that both endoscopic treatment groups had low local recurrence rates with no reported cancer-related mortality. Based on these results, the authors reported that favorable outcomes could be achieved with endoscopic treatment instead of gastrectomy [42]. A meta-analysis of 7 retrospective studies showed that the recurrence rate was significantly lower with re-endoscopic treatment for positive horizontal resection margins compared to close observation (relative risk, 0.22; 95% confidence interval, 0.06–0.86) but higher compared to gastrectomy (relative risk, 6.45; 95% confidence interval, 1.17–35.52). However, all recurrences were managed with additional endoscopic treatment or surgery [2].

After non-curative resection is confirmed after ESD, pathological curability is a key factor in determining subsequent treatment; however, patient-related factors cannot be neglected. In particular, comorbidities or patient age must be considered in treatment decision-making to weigh the benefits and risks of additional surgery.

In an analysis of patients who were followed up without further treatment after confirmation of non-curative resection following ESD, factors associated with short-term survival of 3 years or less were evaluated. Severe comorbidity, defined as a Charlson Comorbidity Index (CCI) ≥3, was a key factor, with no gastric cancer-related deaths observed. Based on these findings, treatment without additional gastrectomy has been proposed for patients with severe comorbidities [94]. Similar results were found in another retrospective study, in which high-risk comorbidity with a CCI of ≥3 was a clear predictor of unfavorable overall survival, regardless of the additional treatment status in non-curative ESD [80]. Although comorbidities cannot be clearly addressed in the guidelines, they can serve as an important consideration when deciding whether to strictly adhere to the pathological criteria for gastrectomy following non-curative ESD [95]. Other factors, such as nutritional status based on the prognostic nutritional index or the presence of sarcopenia, have also been reported as predictors of unfavorable overall survival after ESD. Patients with these factors often have a poor prognosis, regardless of whether additional treatment is administered following non-curative ESD. Therefore, some argue that the current curability criteria should be expanded to accommodate these patient populations [96, 97, 98, 99].

The rate of ESD in the older adult population is increasing, and ESD is recommended, as it is both safe and effective in terms of long-term oncologic outcomes for EGC [100, 101, 102]. However, deciding on the treatment approach for older adult patients after non-curative resection is challenging. Current evidence does not provide sufficient information on the maximum age for considering additional gastrectomy and the extent of comorbidities that influence the decision to opt for strict follow-up instead of gastrectomy. Discussions regarding the age threshold at which observation might be preferable to gastrectomy following non-curative ESD are currently ongoing. In general, patient age did not appear to be a greater limiting factor than comorbidities when gastrectomy was considered. On the other hand, however, it has been reported that non-cancer-specific death after gastrectomy is not negligible in very elderly stage I gastric cancer patients aged ≥85 years.[103] A cohort study of patients aged ≥75 years comparing non-curative endoscopic resection with and without surgery found significantly higher overall survival rates (95% vs. 63%) and lower rates of death from gastric cancer recurrence (0% vs. 6.8%) in the surgery group, suggesting that aggressive rescue surgery may still be necessary for older adult patients [91]. Similarly, the 5-year disease-specific survival was 100% in the surgery group compared to 73% in the surveillance group among patients aged ≥70 years with non-curatively endoscopically resected EGC, indicating that gastrectomy should be actively pursued after non-curative resection, even in older adults [86]. However, these studies did not stratify the non-curative resection group according to risk. Therefore, these findings cannot be directly applied in clinical practice. To address this issue, a multicenter analysis using the e-Cura system was conducted in patients aged ≥85 years who underwent non-curative ESD. The study reported that forgoing additional treatment was acceptable in the low- and intermediate-risk groups, whereas additional gastrectomy improved overall survival in the high-risk group among the oldest-old patients, suggesting that gastrectomy should be considered [104]. In a recent prospective study in Japan, patients aged ≥75 years with LNM risk were further divided into high and low groups based on a 10% estimated metastatic risk within the e-CuraC-2 group. The results indicated that the curability criteria could be expanded to cases where the tumor size was <3 cm and to patients in the elderly-low group [105].

The current treatment algorithms in the guidelines are straightforward and well-defined, as they were established based on extensive data on LNM risk from various studies. However, in clinical practice, decisions often need to consider patient factors in addition to curability criteria. Although several studies have explored these factors, the evidence remains insufficient to draw definitive conclusions. Nevertheless, it is possible to begin applying new treatment algorithms based on available data, with the understanding that they will need to be refined as higher-quality evidence becomes available. Fig. 5 illustrates the current treatment algorithms following non-curative ESD recommended by evidence-based guidelines [2, 3] and algorithms that could be proposed considering the age and comorbidities of patients.

Fig. 5
Guideline-based treatment strategy for early gastric cancer following non-curative ESD (A, adapted from ‘Korean Practice Guidelines for Gastric Cancer’ [2, 3]) and proposed a treatment algorithm incorporating patient factors (B). In this algorithm, age ≥85 years was considered super-elderly, and ≥75 years was considered elderly. In addition, the high-risk group for LNM refers to patients with deep submucosal or lymphovascular invasion.
ESD = endoscopic submucosal dissection; Diff = well or moderately differentiated; UI = ulcer lesion; UnDiff = poorly differentiated/poorly cohesive (including signet ring cells); APC = argon plasma coagulation; MDT = multidisciplinary team; LNM = lymph node metastasis.

• Click for larger image

In cases of non-curative endoscopic resection for EGC, additional treatment should be based on LNM risk to maximize curability. For patients with high-risk pathological factors, such as lymphatic invasion or deep submucosal invasion, additional gastrectomy is a clear choice. However, in cases with positive horizontal margins, the negligible risk of LNM makes endoscopic treatment a reasonable option. Indeed, repeat endoscopic resection is more appropriate than argon plasma coagulation therapy for ensuring complete resection. Most importantly, patient factors, such as age and comorbidities, must be considered in addition to pathological curability when planning further treatment. Other factors, including sarcopenia and nutritional status, have also been shown to affect treatment decisions. Treatment strategies should aim to minimize not only GC-specific mortality but also all-cause mortality in consideration of these additional factors.