Endoscopic submucosal dissection (ESD) has emerged as a highly promising approach for the treatment of early-stage gastrointestinal cancers due to its favorable safety and efficacy profile. With advancements in endoscopic technology and growing evidence supporting its outcomes, the indications for ESD have expanded to include undifferentiated (UD) early gastric cancers (EGC).^1,2 Numerous prospective and retrospective studies have demonstrated favorable outcomes, with 5-year overall survival rates ranging from 94.2% to 98.6% following curative endoscopic resection.³ Noncurative resections have also shown high 5-year overall survival rates ranging from 82.5% to 96.3%. However, the curative endoscopic resection rate for UD EGC remains disappointing, ranging from 55.0% to 82.5%.³ These findings indicate that incomplete resection is not uncommon in patients with UD EGC. A nonrandomized single-arm confirmatory trial conducted by the Japan Clinical Oncology Group reported noncurative resections after ESD for UD EGC in 29% of cases.⁴ While most guidelines recommend radical gastrectomy with lymph node dissections for noncurative resection, the incidence of lymph node metastasis (LNM) in these cases is relatively low.^5,6 Endoscopists often experience significant anxiety regarding the absence of additional treatment options for UD EGC after incomplete endoscopic resection, even in patients at low risk of recurrence. Moreover, patients who have undergone ESD for UD EGC often are at high risk for surgical complications due to advanced age or other comorbidities. Therefore, when encountered with incomplete resection after ESD, endoscopists and patients face challenges in deciding whether to proceed with additional surgery or opt for follow-up without surgery. This decision-making process becomes even more challenging due to the limited accumulation of evidence specifically related to UD EGC. Consequently, there is a growing need for risk stratification systems that can guide clinicians in determining the next stage of treatment after incomplete endoscopic resection, aiming to optimize patient outcomes and avoid overtreatment.

Hatta et al.⁷ proposed a risk-scoring system, the eCura system, for predicting LNM in patients undergoing noncurative ESD for EGC. This system assigns weighted points to five pathological characteristics known as risk factors for LNM: lymphatic invasion (3 points), venous invasion (1 point), tumor size >30 mm (1 point), positive vertical margin (1 point), and SM2 (1 point). Based on the cumulative scores, patients are categorized into three LNM risk groups: low (0-1), intermediate (2-4), and high (5-7). The LNM rates in the development cohort were 2.5%, 6.7%, and 22.7% for the low-, intermediate-, and high-risk groups, respectively. The eCura system has been well validated and is now recommended in the latest Japanese guidelines.^2,7 However, the eCura system is only recommended for use in differentiated EGC currently, as the development cohort included only a small percentage (14.8%) of UD EGC cases. Therefore, the validity of the eCura system for UD EGC needs to be confirmed. Recently, Shiotsuki et al.⁸ reviewed 146 patients with UD EGC who underwent additional surgery after noncurative resection to determine whether the eCura system could be applied in this context. A comparison of their data with the eCura cohort data showed a similar risk of LNM for all five items, although there were slight differences in odds ratios.⁸ Nevertheless, due to the small number of cases, further large-scale studies are necessary to determine the applicability of the eCura system for UD EGC in real-world practice.

In last issue of Gut and Liver, Yang et al.⁹ conducted an external validation of the eCura system for UD EGC using a large number of cases, and their study findings are reported herein. The authors collected data from 634 patients with noncurative endoscopic resection, of whom 270 underwent radical surgery and 364 received no additional treatment, from 18 institutions in Korea. Among the 270 patients who underwent additional surgery, 18 (6.7%) had LNM, and the LNM rates significantly increased with increasing eCura risk score. In the additional surgery group, the LNM rate was 2.6% in the low-risk category, while it was higher in the intermediate- and high-risk categories at 10.9% and 14.8%, respectively. Although the LNM rate of 2.5% in the low-risk category was almost identical to the 2.6% observed in the development cohort of the eCura system, there were some differences in the other two risk categories. Considering the critical impact of LNM rates on patient outcomes, this validation cohort may be considered suitable especially for the low-risk category.

Patients in the high-risk category demonstrated a 14.8% LNM rate, which was associated with higher risks of overall mortality, cancer-specific mortality, and cancer recurrence compared to the low-risk category. Additionally, patients in the intermediate-risk category who did not receive additional treatment also exhibited significantly higher risks of overall mortality, cancer-specific mortality, and cancer recurrence. In the combined intermediate- to high-risk category, the mortality was significantly higher for patients who did not undergo additional treatment compared to those who underwent radical surgery (hazard ratio, 3.18; 95% confidence interval, 1.41 to 7.17; p=0.005). Consequently, the results confirm the necessity of additional surgery for patients categorized as intermediate- to high-risk following noncurative endoscopic resection.

The high LNM rates in patients classified as intermediate- and high-risk in the radical surgery group and the poor overall survival and cancer-specific survival observed in patients classified as intermediate- and high-risk in the no additional treatment group are expected findings. In real-world practice, it is indisputable that such patients should be sent for surgery after noncurative endoscopic resection. However, greater attention should be given to the outcomes of patients categorized as low risk since more consideration is required to determine the necessity of additional gastrectomy in these cases. In particular, if the guidelines recommend surgery for these patients, deciding whether to preserve or remove the stomach becomes more challenging. In this study, the 5-year cancer-specific survival rate of patients in the low-risk category was excellent at 99.7%, even without additional gastrectomy. Moreover, there were no differences in overall mortality, cancer-specific mortality, and cancer recurrence between the low-risk category patients with and without additional treatment. Unfortunately, the number of patients in the low-risk category experiencing these events was too small to confirm hazard ratios and 95% confidence intervals.

Based on the high proportion (87.1%) of individuals categorized as low-risk and their favorable outcomes in this retrospective analysis, it can be extrapolated that endoscopists have already been employing the components of the eCura system, if not the system itself, to guide their decision-making process. As additional evidence, such as the current validation study, becomes available, the eCura system for UD EGC may assist in making systematic and evidence-based decisions regarding the next stage of treatment for patients with noncurative endoscopic resection. The prediction of LNM in UD EGC has been a critical but challenging issue in the era of ESD, and Yang and colleagues in this study may have provided the initial answers to this question.

No potential conflict of interest relevant to this article was reported.