Validity of additional surgical resection by comparing the operative risk with the stratified lymph node metastatic risk in patients with early gastric cancer after endoscopic submucosal dissection

Recently, endoscopic treatment for early gastric cancer (EGC) has been widely performed due to the improvement of diagnostic ability and progress of surgical procedures and medical devices [1]. The advent of endoscopic submucosal dissection (ESD) not only allowed accurate histopathological diagnosis but also reduced the local recurrence rate for even lesions larger than 2 cm and/or with ulcer scar which were difficult with conventional endoscopic mucosal resection (EMR) [2‐4].

According to the guidelines for the treatment of gastric cancer, an absolute indication of ESD against EGC is defined as being intra-mucosal carcinoma and differentiated type without ulcer formation regardless of tumor diameter and also 3 cm or less in intra-mucosal carcinoma and differentiated type with ulcer formation. More recently, the indication has been expanded to lesions 2 cm or less of intra-mucosal undifferentiated carcinoma without ulcer formation following the results of clinical trials [5‐7]. On the other hand, lesions, which do not meet the criteria for curative resection in the histopathological examination after ESD, are deemed to have a potential risk of lymph node metastasis (LNM), and additional gastrectomy with lymph node dissection (LND) is principally recommended only in terms of metastatic risk [7]. In recent years, however, the number of elderly patients has increased, having severe comorbidities, and an additional gastrectomy itself can be a considerable risk in such patients. In fact, it must also be recognized that there are many cases without metastasis in the retrieved lymph nodes even when additional gastrectomy with LND is performed.

Shoda et al. previously reported the usefulness of stratified LNM risk factors (RFs) for patients with EGC who did not meet absolute endoscopic resection in a single institution [8]. In the first step of this study, we reanalyzed the stratified LNM RFs by adding our case series. In the second step, we calculated a 30-day surgical death and hospital death risks of cases in which additional gastrectomy with regional LND was actually performed after ESD and assessed the validity of additional gastrectomy by comparing the operative risks with the stratified LNM risk in each case.

Previous reports demonstrated that the LNM frequency of gastrectomy with LND for cases diagnosed as non-curative resection by pathological examination after endoscopic resection was 3.8–8.2% [12‐14]. Some endoscopists and surgeons examined whether the LNM frequency could be stratified according to several clinicopathological metastasis-related factors. In previous reports, Shoda et al. demonstrated that the LNM frequency well correlated with the number of RFs defined by them, and LNM frequency was extremely low as 0.58% in the 0/1 RF group [8]. We reanalyzed the LNM frequency with the criteria reported in a total of 861 patients treated at two different institutions and confirmed similar results, and the LNM frequency of the lowest risk group with 0/1 RF was found to be as low as 0.76%. Hatta et al. also reported in a multicenter collaborative study showing the risk-scoring system for LNM after ESD that does not meet the current curative criteria. In their study, they counted lymphatic invasion, which was reported to be a frequent LNM factor [15‐17], as 3 points and counted each of other factors associated LNM (tumor size > 30 mm, positive vertical margin, venous invasion, and submucosal invasion ≥ 500 μm) as 1 point, stratified according to the total number, and the frequency of LNM was examined. The results of this study also showed that the LNM rate of the “low group” with the lowest risk of LNM was 2.5%, extremely low [18].

In Japan, the gastrectomy-related mortality has been considered to be very low, so additional gastrectomy with LND is recommended without much consideration for cases with potential risk after non-curative ESD. Recently, a so-called real-world Japanese data based on NCD clarified that in-hospital mortality by gastrectomy was not so low, 1.2% in distal gastrectomy [10] and 2.3% in total gastrectomy [11]. The mortality rate would not be negligible although the frequency might be relatively lower only in cases with EGC than reported. In recent years, elderly patients with EGC have been increasing due to recent aging society and popularization of medical examination in Japan. The physiological functions of various organs are generally deteriorated and comorbidities as potential surgical risk often present in elderly patients [19, 20]. These findings prompted us to examine whether the surgical risk of additional gastrectomy may be higher than the LNM frequency in some patients with EGC after ESD. In this study, we calculated the surgical risk of each cases using the NCD risk calculator and compared the predicted mortality risk with the LNM potential risk stratified according to the number of RFs. As a result, the predicted surgical mortality was higher than the potential frequency of LNM in 25% of the case in the 0/1 RF group, which indicates the additional gastrectomy might be an over-indication. All of our case series after ESD were also applied to the “eCura system” [18] to verify the categories of LNM. As a result, 4% of patients were classified as “low group,” and the LNM risk was lower than the predicted surgery-related mortality in 2.5% (one patient) in our series. The difference of LNM risks between our results and eCura system may be due to the facts, firstly, that the evaluated risk factors of the eCura system included vertical positive margin and, secondly, that it did not include histological undifferentiated adenocarcinoma. Further nation-wide large trials should enable more accurate LNM risk diagnosis after ESD.

In this study, patients who underwent additional gastrectomy after ESD and whose surgical risk was assessed using the NCD risk calculator had no in-hospital death. However, 36.2% of patients have postoperative complications. Postoperative complications are known as recurrence and poor prognosis factors after surgery for gastric cancer [21, 22].In addition, postoperative complications were noted in 5 of 10 cases, and their surgical risk was higher than the frequency of LNM. We think that if they had not received additional gastrectomy, their prognosis might have been better.

On the other hand, 35 patients with EGC were observed which were treated by ESD without surgery under the diagnosis of non-curative resection from 2005 to 2017 at the University of Yamanashi Hospital. Among them, only one case had a recurrence in regional lymph nodes (2.9%), and he was in the group with 2RFs. There was no recurrence in the 0/1RF group with low risk of LNM.

There are some limitations in our study. First is that our stratified LNM risk model was constructed based on clinicopathological data of EGC resected surgically. Second, the review of the comparison of LNM frequency and the surgical risk was a small-scale retrospective examination at a single facility. In the future, we might compare the prognosis of patients with higher surgical risk than LNM risk between follow-up without additional treatment group and additional gastrectomy group with regional lymph node dissection.

In conclusion, the present study clearly demonstrated that the LNM frequency could be stratified by the total number of the LNM RFs for the patients who do not meet the curative criteria after ESD. Furthermore, the risk comparison study suggested that there is a considerable number of cases in which their surgical risk was higher than the LNM risk among cases which additional gastrectomy was recommended for potential LNM risk. These findings indicated, at least, that we should discuss individually the indication of additional gastrectomy after ESD for each patient from both perspectives of LNM and surgical risks.