Endoscopic submucosal dissection is a minimally invasive and effective procedure for early gastric cancer with a negligible risk of LNM. When the pathological findings do not meet the curative criteria, an additional gastrectomy with lymph node dissection is recommended [1—3]. However, certain patients, such as older adults, may have additional risks with this surgery [6, 7]. Furthermore, an additional surgical resection may amount to overtreatment for approximately 90% or more of the patients identified in ESD with noncurative treatment . Therefore, whether additional surgery should be performed after noncurative ESD for early gastric cancer is controversial [4, 5, 9]. This retrospective analysis revealed that the incidence rate of LNM in patient receiving additional gastrectomy was 14% (14/100), which is slightly higher but comparable to incidence rates previously reported by others ( ranging from 6.3–12.7% ) in studies on additional gastrectomy after ESD [2, 10–14].
Previous reports identified several predictive factors for LNM in cases of early gastric cancer, including a submucosal invasion, a positive lymphatic invasion, a positive vascular invasion, an undifferentiated tumor type, and a tumor size > 30 mm [14–16].
These factors are critical in determining whether to perform an additional gastrectomy with lymph node dissection in patients not meeting the curative criteria after ESD. In this study, lymphatic invasion and undifferentiated type were independent risk factors for LNM in patients not meeting the curative criteria after ESD.
Furthermore, having a vascular invasion was correlated with LNM in a univariate analysis. LNM was observed in 41.2% (7/17) of patients with both lymphatic invasion and vascular invasion and in 19.2% (5/26) of those with a lymphatic invasion but without a vascular invasion. In contrast, LNM was only observed in 3.8% (2/52) of patients who had neither a lymphatic nor vascular invasion. Of patients not meeting the curative criteria after ESD, those without both lymphatic and vascular invasions are considered to be at low risk of LNM. Therefore, both factors being positive are thought to be a markedly high-risk factor for LNM. Neither the tumor size > 30 mm nor deeper submucosal invasion (SM2) was correlated with LNM in this study. Hatta et al  established the eCura scoring system to predict the risk of LNM in patients after a noncurative ESD. This system evaluates patients on 5 factors, and patients are categorized into 3 risk groups. Salvage surgery is expected to benefit those in the high-risk group, while follow-ups alone are sufficient for those in the low risk group . When evaluating the eCura scores in our study, the incidence of LNM was 31.3%(10/32)in the high-risk group and 2.9% (1/34) in the low risk group (data not shown). Therefore, follow-ups without additional surgery may be sufficient not only for patients of advanced age, or those with severe comorbidities, but also for patients in whom lymphovascular invasion is absent.
In this study, local RC after an additional gastrectomy was observed in 7% (7/100) of patients. Multivariate analysis identified as an independent risk factor for local RC after ESD included having a positive horizontal margin. Our result is consistent with previous studies that reported having a positive horizontal margin is a risk factor for local RC, but a positive vertical margin is not [14, 17]. This may be explained by the weaker cautery effect in the horizontal compared to vertical direction [14, 17]. Furthermore, having an undifferentiated tumor type was not significantly related to the risk of local RC in this study. After noncurative ESD, treatment options include additional surgery and additional endoscopic procedures. Because of the low risk of LNM in cases with a positive horizontal margin and a differentiated tumor type without positive lymphatic invasion, local treatment, such as a secondary ESD, can be considered [2, 14, 18]. Confirming the utility of this approach, one of 2 patients in the observation group who had local recurrence underwent secondary ESD and is alive without recurrence.
The local recurrence rate after an en bloc resection with a negative margin by ESD has been reported to range between 0% and 0.7% [19–21]. For comparison, there were two patients in our study that had local RC observed after en bloc resection with negative margin (One patient was in the additional gastrectomy group, the other in the observation group). Both of these lesions were large tumors with lymphovascular invasion. Previous reports stated that in these cases the cancer cells might have spilled out from the vessels, accumulated within the vessels due to vascular stasis, or remained after resection because the coagulated portion of the resection margin could not be assessed accurately [14, 19].
We have found that having SM2 or a deeper invasion was not associated with the risk of LNM, but the reverse was not true: all patients with LNM had SM2 or a deeper invasion. Significantly, all three patients who experienced a recurrence after the additional gastrectomy had more than 2 LNM. There is no consensus or ideal recommendation for the extent of lymph node dissection in the additional surgery . A D1 + lymph node dissection was performed as standard treatment in our study. However, D2 dissection was selected, when the lesion depth was SM2 with a positive vertical margin or when LNM was clinically suspected. In our study, all LNM instances were within the D1 + level. Therefore, performing a D1 + dissection appears reasonable after noncurative ESD for early gastric cancer.
Surgery related complications classified as CD grade Ⅱ or worse occurred in 14 patients (14%) in this study. Of those who underwent additional surgery, 1 patient (1%) died due to a probable perioperative adverse event. During the follow-up period, 3 patients(3%)had recurrences: 1 each in the peritoneum, bone, and lymph nodes༎Two patients (2%) died of gastric cancer, and 6 (6%) died of other diseases. In the observation group, 1 patient (2.0%) died of gastric cancer and 10 patients (19.6%) died from other diseases. Clinicopathological characteristics of the patients did not differ significantly between the two groups. Although, age and ASA-PS were significantly higher in the observation group. Therefore, careful follow-ups without additional surgery may be acceptable for patients with differentiated type without lymphovascular invasion, those at an advanced age, or those with a severe comorbidity. On the other hand, an additional gastrectomy with a lymphadenectomy may be more appropriate than simple follow-ups in noncurative ESD patients without concomitant disease who are young enough to undergo a surgical intervention [1, 23].
This study has some limitations. First, we used the relatively small sample size from a single institute. Second, the criteria of ESD differed slightly depending on doctors who performed the treatment. Third, there was resection bias in the treatment path (gastrectomy contraindicated due to patients’ old age, comorbidities, or their own decision). In addition, the average age of the observation group was significantly higher than that of the additional surgery group. To address these limitations, a prospective, multicenter, large-scale further analysis should be considered.