Background
Total and subtotal gastrectomies have been performed for the treatment of gastric cancer and benign peptic ulcers [
1]. In addition, esophagectomy with reconstruction using a gastric tube has been used for esophageal cancer [
2,
3]. However, there is a risk of gastric cancer in the postoperative stomach after surgery if total gastrectomy is not performed [
2‐
4]. Gastric cancer in the postoperative stomach and non-operative stomachs can be detected by follow-up surveillance endoscopy [
5]. Endoscopic mucosal resection and endoscopic submucosal dissection (ESD) were performed for early gastric cancer (EGC) with a negligible risk of lymph node metastasis (LNM). The indication for endoscopic resection (ER) was defined according to the current Japanese Gastric Cancer Treatment Guidelines [
6]. ESD is the currently accepted treatment for lesions with absolute or expanded indications for ER. Long-term outcomes have been favorable for patients who underwent curative resection [
7,
8]. Similar to ESD of the non-operative stomach (ESD-N) for EGC, ESD of the postoperative stomach (ESD-P) for EGC has been applied to and performed. Several studies reported that ESD-P is an effective and safe treatment modality for EGC and ESD-N [
9‐
12]. Furthermore, it was reported that long-term outcomes of ESD-P for EGC were favorable when curative resection was obtained [
11,
12].
However, ESD-P for EGC is known as a technically difficult procedure because of the presence of submucosal fibrosis and staples around the suture line or anastomotic site [
9‐
13]. Some reports have shown high resection rates but with a higher risk of complications with ESD-P than with ESD-N [
14,
15]. However, there were some differences in the baseline characteristics for both groups, which might have affected the outcomes. The rates of tumors in the upper third of the stomach were higher than those of the tumors in the other parts of the stomach after distal gastrectomy, which is also technically difficult [
16‐
19]. Recently, a propensity score-matching analysis has been used to compensate for the differences in the baseline characteristics between the two groups [
20‐
22]. Therefore, the present study aimed to examine the difficulty of ESD-P for EGC compared with ESD-N for EGC by using a propensity score-matching analysis.
Discussion
To the best of our knowledge, this is the first study comparing the clinical outcomes between ESD-P and ESD-N for the treatment of EGC using a propensity score-matching analysis. Additionally, various types of post-operative stomachs, including not only distal gastrectomy but also proximal gastrectomy, pylorus preserving gastrectomy and esophagectomy, were included and assessed in the present study. Although a high rate of curative resection with a low rate of adverse event for ESD was achieved regardless of the type of post-operative stomach, ESD-P for EGC required significantly longer procedure times than ESD-N for EGC. Moreover, one patient who underwent ESD-P for EGC only on the suture line required emergency surgery.
In this study, 65.9% (27/41) of the patients in the ESD-P group who underwent distal gastrectomy required resection of the middle and lower thirds of the stomach, indicating that EGC of ESD-P frequently originally developed in the upper third of the stomach. It is generally accepted that it is more difficult to complete ESD in the upper third than in the middle third or lower third of the stomach [
16‐
19]. Therefore, it is not appropriate to directly compare the clinical outcomes between ESD-P and ESD-N, because the lesion locations in both procedures significantly differ; thus, we performed a propensity score-matching analysis and included this factor as a covariate-matching factor to compensate for this bias. Therefore, we were able to compare the treatment outcomes for both groups after controlling for potential sources of bias according to the covariates. The tumor location and other covariates did not affect the treatment outcomes of ESD for both groups.
Lesions on the suture line or anastomotic site involve the submucosal fibrosis [
9,
13,
15]; therefore, performing ESD for these lesions was difficult to complete. In the present study, 17 lesions had developed on the suture line of the postoperative stomach, whereas 2 lesions on the anastomotic site. Even though ESD of these lesions was performed only by experts, the median ESD time was significantly longer for these lesions than for lesions not involving the suture line or anastomotic site (120.0 min vs 60.5 min;
P < 0.001). There was no significant difference in the procedure time between lesions on the suture line and those on the anastomotic site (120.0 min vs 130 min;
P = 0.69). Lesion development on the suture line or anastomotic site was the only independent factor associated with prolonged procedure times (Table
4). In addition, it has been reported that ESD of lesions on the suture line is associated with a high rate of perforation [
9,
15]. In this study, there was only one perforation case during ESD of a lesion located on the suture line in the postoperative stomach, even though experts performed the ESD procedure. Despite the non-negligible risk of perforation and prolonged procedure time, a high rate of curative resection has been achieved, even for lesions that developed on the suture line or anastomotic site of the postoperative stomach. The complication rate of ESD-P in the present study was lower than those of previous reports [
9‐
12]. This might be because of the fact that 87.8% (36/41) of ESD-P procedures were performed by experts. ESD-P should be performed cautiously by highly experienced experts, especially when the lesions are located on the suture line or anastomotic site.
We should take the difference in cost of ESD between the groups into account in addition to the procedure time. The proportion of patients who underwent ESD-P using multiple kinds of endo-knives seems to be higher than that of patients who underwent ESD-N using multiple kinds of endo-knives, although it did not reach to a statistical significance (36.6% in ESD-P vs 17.1% in ESD-N;
P = 0.080, shown in Table
3). Therefore, it seemed that ESD-P was costlier than ESD-N in terms of the number of devices used during the procedure.
In this study, consistent with previous reports [
9‐
12], high rates of en bloc and complete resections and resulting high rates of curative resection were achieved not only in the ESD-N group but also in the ESD-P group. All patients with curative resection in both groups had no recurrence. In contrast, there were four patients with non-curative resection in each group. Regarding ESD-N, one of the four patients with non-curative resection underwent additional surgery, which resulted in finding a remnant tumor of the resected specimen. The remaining patients did not undergo any additional treatments, and all had no recurrence. According to the current guidelines [
6], additional surgery is indicated for all patients with non-curative ESD resection and significant risk for LNM. However, it has been reported that almost half of the EGC patients with non-curative ESD resection did not undergo additional surgery due to their advanced age or comorbidities [
34]. Furthermore, it has been reported that LNM was found in only 5–10% of patients with such lesions in the non-operative stomach [
35]. Regarding ESD-P, all four patients who underwent non-curative ESD resection were followed up without any additional treatments. The risk of LNM in T1 remnant cancer after distal gastrectomy resection was reported to be only 2.4% (1/42) [
36]. Patients who underwent non-curative ESD resection in the remnant stomach refused radical completion gastrectomy of the remnant stomach due to the high risk of complications and postoperative mortality rates (from 13 to 41%) [
37‐
40]. Completion gastrectomy for EGC of the remnant stomach is much riskier during the perioperative period compared to that of the non-remnant stomach [
37‐
40]. Given the few data regarding the prognosis of patients with non-curative ESD resection for the postoperative stomach, it should be considered whether additional surgery should be performed based on the current guidelines due to the non-negligible risk of postoperative mortality for radical completion gastrectomy of the remnant stomach. As a result, all four patients with non-curative ESD resection in the ESD-P group had no recurrence. A recent report suggested a useful scoring system for evaluating the risk of LNM for non-curative patients [
41]. Using this scoring system, 3 patients who underwent non-curative ESD-P were classified as low-risk patients, with an estimated LNM risk of 2–3%. The other patient was classified as having an intermediate risk, with estimated LNM risk of 6–7%. In this study, there were no high-risk patients. This system may be useful for deciding additional surgery even for patients who underwent non-curative ESD-P. However, some lymph nodes might be resected in patients who underwent the previous surgery for a malignant lesion. This system may not be applicable to non-curative patients with post-operative stomach. In the future, further expanded criteria or more appropriate risk stratifications may be required for ESD-P for EGC.
This study had some limitations. First, this was a single-center retrospective study with a relatively small sample size. Second, we did not consider the use of traction method, such as a traction method using dental floss and a hemoclip, which was recently reported to shorten the procedure time of ESD [
42]. Third, we did not consider medications, especially anti-thrombotic drugs, in this study, since they were discontinued during the perioperative period, because the patients enrolled in this study had undergone ESD treatments before the current Japanese guidelines for the use of anti-thrombotic agents were established [
43]. Hence, there are no data regarding the influence of anti-thrombotic agents on ESD-P. Fourth, we could not deny the possibility that the learning curve might affect the technical outcome of this present study, which could not be fully compensated by the propensity score-matching. Fifth, ESD-P was performed by only 9 experts, specifically 8 experts who performed ESD-P involving the suture line or anastomotic site, whereas, 8 experts performed ESD-N after propensity score-matching. Seven experts were in common among ESD-P and ESD-N after matching Furthermore, 6 experts were in common among ESD-P involving suture line or anastomic site and ESD-N after matching. Therefore, there seems to be some bias of operators. Sixth, we did not consider the difference of the electrosurgical units (VIO300D or ICC200) used in the ESD procedures, which might have affected the study outcome. A larger prospective, randomized, controlled trial will be required to determine further treatment guidelines, including the influence of anti-thrombotic agents, for ESD-P.