- Email: [email protected]
Long-term outcome comparison of endoscopic resection and surgery in early gastric cancer meeting the absolute indication for endoscopic resection
ORIGINAL ARTICLE: Clinical Endoscopy
Long-term outcome comparison of endoscopic resection and surgery in early gastric cancer meeting the absolute indication for endoscopic resection Il Ju Choi, MD, PhD,1,* Jun Ho Lee, MD, PhD,1,* Young-Il Kim, MD,1 Chan Gyoo Kim, MD, PhD,1 Soo-Jeong Cho, MD, PhD,1 Jong Yeul Lee, MD,1 Keun Won Ryu, MD, PhD,1 Byung-Ho Nam, PhD,2 Myeong-Cherl Kook, MD, PhD,1 Young-Woo Kim, MD, PhD1 Goyang, South Korea
Background: Endoscopic resection (ER) of early gastric cancer (EGC) meeting the absolute indication has excellent long-term outcomes. Objective: To compare long-term outcomes of ER with those of surgery in patients with EGC who met the absolute indication for ER. Design: Retrospective cohort study. Setting: A specialized center for the treatment of cancer. Patients and Interventions: We retrospectively reviewed data from patients who underwent gastrectomy or ER for EGC between 2002 and 2007. Gastric cancers were differentiated-type adenocarcinoma without ulceration confined to the mucosal layer and 2 cm or smaller in size. Main Outcome Measurements: The primary outcome was overall survival (OS). Metachronous cancer rates and adverse event rates were compared. Kaplan-Meier plots and Cox proportional hazard regression analyses were applied for comparisons. Differences in baseline characteristics were adjusted by propensity score. Results: Among 375 patients, 261 underwent ER and 114 underwent surgery. The median follow-up duration was 76.4 months. The 5-year OS rates did not significantly differ between the ER and surgery groups (95.7% vs 93.6%, respectively; P Z .725 by log-rank test). There were no gastric cancer–related deaths in either group. Metachronous gastric cancer developed more frequently in the ER group (6.1%, 16/261) than in the surgery group (0.9%, 1/114) (P Z .024). However, most patients (93.8%, 15/16) in the ER group were curatively treated with repeat ER. Adverse event rates were higher in the surgery group than those in the ER group (7.9% vs 2.7%, P Z .028). Limitations: Retrospective, single-center study. Conclusions: The OS rate after ER for mucosal gastric cancer that met the absolute indication was comparable to that achieved with surgery. Although metachronous cancers were more common after ER, they were usually treatable and did not affect survival. (Gastrointest Endosc 2015;81:333-41.)
Abbreviations: DFS, disease-free survival; DSS, disease-specific survival; EGC, early gastric cancer; ER, endoscopic resection; ESD, endoscopic submucosal dissection; OS, overall survival; RFS, recurrence-free survival; SD, standard deviation. DISCLOSURE: All authors disclosed no financial relationships relevant to this publication. This work was supported by grant 1310280 from the National Cancer Center, South Korea. See CME section; p. 439. Current affiliation of Dr Jun Ho Lee is the Department of Surgery, Sungkyunkwan University School of Medicine, Samsung Medical Center, Seoul, South Korea.
www.giejournal.org
*Drs Il Ju Choi and Jun Ho Lee contributed equally to this work. Copyright ª 2015 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2014.07.047 Received March 12, 2014. Accepted July 17, 2014. Current affiliations: Center for Gastric Cancer (1), Center for Clinical Trials (2), National Cancer Center, Goyang, Korea. Reprint requests: Il Ju Choi, MD, PhD, Center for Gastric Cancer, National Cancer Center, 323 Ilsan-ro Ilsandong-gu, Goyang Gyeonggi, 410-769 South Korea.
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 333
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
Early gastric cancer (EGC) is defined as gastric cancer in which tumor invasion is limited to the mucosa or submucosa. Early diagnosis of gastric cancer has increased in South Korea due to screening programs and recent improvements in diagnostic techniques.1,2 Gastrectomy with lymph node dissection is a standard treatment for gastric cancer, including EGC.3 Long-term outcomes of gastrectomy are excellent, with 5-year overall survival (OS) rates of more than 92%.4,5 Endoscopic resection (ER), including EMR and endoscopic submucosal dissection (ESD), is widely accepted as a curative treatment for EGC without the risk of lymph node metastasis.6-8 Currently, ER is considered a standard treatment for EGC that meets the absolute indication suggested in the Japanese gastric cancer treatment guidelines for ER.3 The 5-year OS rates after ER are between 84% and 97%,6,9-11 which is acceptable for standard treatment. Recent studies show that ER for EGC has long-term outcomes that are comparable to those of gastrectomy.12,13 However, past studies had limitations, including significant differences in tumor characteristics for the ER and gastrectomy groups12 and inclusion of tumors that did not meet the criteria for absolute indication.13 Thus, the aim of this study was to investigate whether long-term outcomes of ER are comparable to those of surgery for EGC that meets the criteria for absolute indication.
from our database of prospectively collected data based on medical records. Tumor stage was based on the sixth edition of the International Union Against Cancer/ American Joint Committee on Cancer classification system.14 The data related to treatments were ER methods, rate of complete resection, surgical method, extent of lymph node dissection, multiple tumors, and adverse events of each treatment. Early adverse events were defined as those that occurred within 30 days of treatment.
METHODS
Pathological evaluation
Study population This study was a retrospective cohort study of 665 patients who underwent gastrectomy or ER for EGC at the Center for Gastric Cancer in the National Cancer Center between January 2002 and December 2007. We included patients who met the following inclusion criteria: (1) 20 years of age and older, (2) newly diagnosed EGC without previous treatment, (3) histologically confirmed well- or moderately differentiated adenocarcinoma, (4) intramucosal tumor without ulceration and 2 cm or less in diameter, (5) received gastrectomy with lymph node dissection or ER with curative intent. Choice of treatment modality was usually decided after full consultation between the attending physician and the patient. Many patients chose surgery as opposed to EMR before April 2004 because ESD was not routinely covered by insurance during that time and because of the limitations in achieving en bloc resection with EMR. A detailed flow chart of this study is summarized in Figure 1. This study was approved by the institutional review board of the National Cancer Center, Korea (NCCNCS-13-798).
ER and surgery All of the patients included in this study underwent EGD and abdominal CT for clinical staging before treatment. EMR was performed until April 2004, and all ER methods were ESD after that time. Detailed ER methods were described previously.15,16 Patients treated by surgery underwent an open or laparoscopic gastrectomy with D1 þ b or more lymph node dissection. A distal subtotal gastrectomy or total gastrectomy was performed depending on the tumor location. Reconstruction methods included gastroduodenostomy or gastrojejunostomy after distal gastrectomy and Roux-en-Y esophagojejunostomy after total gastrectomy. The extent of lymph node dissection was based on the recommendations of the Japanese Gastric Cancer Association.17 All resected specimens were flattened and fixed in 10% formalin for pathological evaluation. Specimens obtained from ER were sliced serially at 2-mm intervals,18,19 embedded in paraffin blocks, and stained with hematoxylin and eosin. Resected surgical specimens were prepared for pathological evaluation in a similar manner except that serial sectioning was performed at 4-mm intervals. Although the Japanese guidelines recommended that the section intervals of surgically resected specimens be 5 to 7 mm, we made sections of our surgical specimens at 4-mm intervals to achieve a more accurate evaluation.19 Histological subtypes were classified according to the World Health Organization classification of gastric cancer.20 Histologically differentiated types included well- or moderately differentiated tubular adenocarcinoma and papillary adenocarcinoma. Undifferentiated types included poorly differentiated tubular adenocarcinoma, signet ring cell carcinoma, and mucinous adenocarcinoma.18
Follow-up surveillance after treatment
Data related to baseline characteristics, endoscopic and pathological findings, and treatments were obtained
EGD was performed 3 and 12 months after ER or surgery and then annually thereafter. In cases of ER, EGD was also performed at 6 months. Abdominal CT was performed annually. A biopsy of the ER scar tissue was performed at each EGD examination to evaluate the presence of local recurrence in patients who underwent ER.
334 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
www.giejournal.org
Data collection
Choi et al
Long-term outcomes of endoscopic resection for gastric cancer
665 Patients underwent gastrectomy or ER for EGC from Jan 2002 to Dec 2007
Excluded (n= 255): Not included in absolute indication of ER
410 Patients were included in the primary cohort
Excluded (n= 35): 24 Previous ER history 7 Gastric surgery history 4 Other organ cancer
375 Patients meeting absolute indication of ER were included in the the final analyses Figure 1. Flow chart of this study. EGC, early gastric cancer; ER, endoscopic resection.
Statistical analysis Continuous variables were expressed as mean with standard deviation or median with interquartile range. For comparisons of baseline and clinicopathological characteristics between the ER group and the surgery group, continuous variables were analyzed by the Student t test or MannWhitney U test, and categorical variables were analyzed by the chi-square or the Fisher exact test. The primary endpoint of this study was OS, and the secondary endpoints were disease-specific survival (DSS), disease-free survival (DFS), recurrence-free survival (RFS), and occurrence of metachronous gastric cancer. OS and DSS were determined from the time of ER or surgery to death. OS included deaths of any cause, but DSS included only gastric cancer related–deaths. DFS was determined from the time of ER or surgery to gastric cancer recurrence, occurrence of a new gastric cancer, or death of any cause. Finally, RFS was determined from the time of ER or surgery to tumor recurrence, death with evidence of recurrence, or occurrence of a metachronous gastric cancer. For RFS, patients who died without known tumor recurrences were censored at the last follow-up time point. In addition to the study endpoints, short-term treatment outcomes, including endoscopic and pathological en bloc resection rates and histologically complete resection rate, were compared between EMR and ESD, and the occurrence of a metachronous gastric cancer was
investigated after ER or surgery. A histologically complete resection was defined when both of the following conditions were fulfilled: (1) en bloc resection and (2) no tumor cells in both horizontal and vertical margins on the final pathological examination.21 A metachronous gastric cancer was defined as a new gastric cancer at a previously uninvolved site in the remnant stomach that was detected at least 1 cm from the muscularis mucosa disruption or from fibrosis at previous ER or anastomosis sites. A new cancer occurring less than 1 year after the initial treatment was defined as a synchronous gastric cancer and was excluded from the analysis. Follow-up data on deaths and recurrences were obtained until December 2011. Survival curves were constructed with the Kaplan-Meier method, and comparisons were performed with the log-rank test. Multivariate analyses for OS, DFS, and RFS were performed using the Cox proportional hazards model. To reduce the effect of selection bias, we used the propensityadjustment method in the multivariate analyses.22,23 The estimated propensity score was obtained from the logistic regression model, which included those factors that were significantly different between the 2 treatment groups. Then the estimated propensity scores were included as a covariate along with age and sex in the multivariate analyses. All statistical analyses were performed with SAS version 8 (SAS Institute, Cary, NC). A P value !.05 was considered statistically significant.
www.giejournal.org
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 335
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
TABLE 1. Baseline and clinicopathological characteristics of the study population
Median age, y (IQR)
ER (n [ 261)
Surgery (n [ 114)
P value
62 (54–68)
62 (54–66)
.541*
Sex, no. (%)
.608
Male
195 (74.7)
88 (77.2)
Female
66 (25.3)
26 (22.8)
Comorbid disease, no. (%)
.083
Diabetes mellitus
10 (3.8)
7 (6.1)
Cardiovascular disease
41 (15.7)
19 (16.7)
Chronic liver disease
11 (4.2)
11 (9.6)
Chronic pulmonary disease
13 (5.0)
1 (0.9)
Location of tumor, no. (%)
.923
Upper third
14 (5.4)
7 (6.1)
Middle third
75 (28.7)
31 (27.2)
Lower third
172 (65.9)
76 (66.7)
1.2 0.4
1.5 0.5
Size of tumor, cm, mean SD Histology of tumor, no. (%)
.029
Well-differentiated
200 (76.6)
75 (65.8)
Moderately differentiated
61 (23.4)
39 (34.2)
Morphology of tumor, no. (%)
.281
Elevated
46 (17.6)
15 (13.2)
Flat or depressed
215 (82.4)
99 (86.8)
74.9 (59.5–92.2)
78.1 (64.2–94.0)
Median follow-up duration, mo (IQR)
!.001
.180*
ER, Endoscopic resection; IQR, interquartile range; SD, standard deviation. * P values were calculated by using the Mann-Whitney U test.
RESULTS
Comparisons of long-term outcomes of ER and surgery
Baseline and clinicopathological characteristics Baseline and clinicopathological characteristics are summarized in Table 1. Of all included patients, 261 patients underwent ER (ER group), and 114 underwent surgery (surgery group). The median age of the 375 patients was 62 years, and there were 283 (75.5%) men. Most tumors (94.4%, 354/375) were located in the middle third or lower third of the stomach. Patients in the ER group had significantly smaller tumors and fewer moderately differentiated adenocarcinomas than those in the surgery group. In the surgery group, the median number of dissected lymph nodes was 36 (interquartile range 27–46). The median follow-up duration of all patients was 76.4 months.
The 5-year OS, DFS, and RFS rates in the ER group were 95.7%, 90.7%, and 94.8%, respectively. The 5-year OS, DFS, and RFS rates in the surgery group were 93.6%, 92.8%, and 99.1%, respectively. In the KaplanMeier analysis, there were no significant differences in OS (Fig. 2A) or DFS (Fig. 2B) between the 2 treatment groups. In contrast, RFS was significantly better in the surgery group than in the ER group (P Z .022 by the log-rank test) (Fig. 2C). Because there were no gastric cancer– related deaths during the follow-up period, the 5-year DSS rates were 100% in both treatment groups. The characteristics of survival estimates in both treatment groups are summarized in the Supplementary Table, available online at www.giejournal.org.
336 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
www.giejournal.org
Long-term outcomes of endoscopic resection for gastric cancer
100
100
80
80
Cumulative DFS rate (%)
Cumulative OS rate (%)
Choi et al
ER group 60 Surgery group 40
20
ER group 60
Surgery group
40
20
P = .725 by log-rank test
P = .200 by log-rank test
0
0 0
A
No at risk ER group Surgery group
2
4
6
8
10
0
Time, year 261 114
258 113
251 110
139 74
52 27
B
2
4
6
8
131 74
52 27
10
Time, year
No at risk ER group 261 Surgery group 114
255 113
241 110
Cumulative RFS rate (%)
100
80 ER group 60
Surgery group
40
20 P = .022 by log-rank test 0 0
C
No at risk ER group Surgery group
2
4
6
8
10
Time, year 261 114
255 113
241 110
131 74
52 27
Figure 2. Comparisons of long-term outcomes between endoscopic resection (ER) and surgery. A, Overall survival (OS). B, Disease-free survival (DFS). C, Recurrence-free survival (RFS) curves.
There were no local recurrences of primary gastric cancer after ER or surgery. Metachronous gastric cancer developed in 17 patients (4.5%, 17/375). Patients in the ER group had a higher rate of metachronous gastric cancer than those in the surgery group (6.1% [16/261] vs 0.9% [1/114], P Z .024). Fifteen of 16 patients (93.8%) in whom metachronous gastric cancer developed were successfully treated with repeat ER. The remaining 1 patient underwent a total gastrectomy because of metachronous gastric cancer that invaded the submucosal layer and was located in the upper third of the stomach. In the surgery group, ER was performed to treat metachronous gastric cancer, which developed in 1 patient. The propensity score was calculated by the multivariate logistic model that included tumor size and histological type and showed significant differences between the ER group and the surgery group. Multivariate analyses with the Cox proportional hazards model adjusted for age, sex, and propensity score showed no significant differences in OS or DFS in the ER group compared with the surgery group (Table 2). However, the ER group showed a worse RFS
rate than the surgery group (adjusted hazard ratio, 8.77; 95% confidence interval, 1.13–67.97; P Z .038) (Table 2).
www.giejournal.org
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 337
Comparisons of short-term and long-term outcomes in EMR and ESD Of the patients in the ER group, 86 patients (33.0%) underwent EMR and 175 (67.0%) underwent ESD. Tumor location and size did not differ between the 2 ER groups. The ESD group had significantly higher endoscopic and pathological en bloc resection rates (Table 3). In addition, the ESD group had a significantly lower rate of tumor involvement in the resection margin and a higher histologically complete resection rate than the EMR group. Of the 14 patients with a positive resection margin, 10 patients underwent argon plasma coagulation and 3 underwent a repeat ER (Table 3). In the ER group, there were no significant differences in OS (P Z .896, by the log-rank test) (Fig. 3A), DFS (P Z .259 by the log-rank test) (Fig. 3B) and RFS (P Z .101 by the log-rank test) (Fig. 3C) between the EMR and ESD groups. In the ER group, metachronous gastric cancer developed in 16 patients, in 9 patients in the EMR group,
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
TABLE 2. Hazard ratios for death and recurrence for patients with early gastric cancer according to the treatment modality Overall survival cHR
95% CI
aHR*
Disease-free survival 95% CI
cHR
95% CI
aHR*
95% CI
Recurrence-free survival cHR
95% CI
aHR*
95% CI
Treatment modality Surgery
1.00
ER
0.86
1.00 0.36-2.02
Propensity scorey
1.00
0.76
0.32-1.84
1.25
0.04-41.34
1.62
1.00 0.77-3.40
1.00
1.63
0.75-3.53
0.64
0.04-9.47
7.41
1.00 0.98-55.88
8.77
1.13-67.97
0.20
0.01-8.67
cHR, Crude hazard ratio; CI, confidence interval; aHR, adjusted hazard ratio; ER, endoscopic resection. * Hazard ratio adjusted for age, sex, and propensity score. y Propensity score was adjusted by size and histology of tumor.
TABLE 3. Comparisons of short-term outcomes according to the ER modality
Median age, y (IQR)
TABLE 3. Continued
EMR (n [ 86)
ESD (n [ 175) P value
61 (53–67)
62 (54–69)
Sex, no. (%)
.329 .495
Male
62 (72.1)
133 (76.0)
Female
24 (27.9)
42 (24.0)
Location of tumor, no. (%) 7 (4.0)
Middle third
24 (27.9)
51 (29.1)
Lower third
55 (64.0)
117 (66.9)
Histology of tumor, no. (%)
7
3
–
Endoscopic resection
2
1
–
Complete resection, no. (%)
67 (77.9)
157 (89.7)
.010
5 (5.8)
2 (1.1)
.032
Bleeding
3 (3.5)
0 (0)
Perforation
2 (2.3)
2 (1.1)
.002
ESD, Endoscopic submucosal dissection; IQR, interquartile range; SD, standard deviation. * One patient did not receive an additional treatment, but local recurrence did not occur.
and 7 patients in the ESD group. There was no difference in metachronous gastric cancer rates between the 2 ER methods (10.5% in EMR and 4.0% in ESD; P Z .054). The 5-year metachronous gastric cancer rates were 7.1% in the EMR group and 4.1% in the ESD group, and the rates did not different between the 2 ER methods in the KaplanMeier analysis (P Z .101 by the log-rank test).
Well differentiated
56 (65.1)
144 (82.3)
Moderately differentiated
30 (34.9)
31 (17.7)
Size of tumor, cm, mean SD
1.1 0.5
1.2 0.5
.280
Endoscopic
65 (75.6)
160 (91.4)
!.001
Pathological
76 (88.4)
171 (97.7)
.003
En bloc resection, no. (%)
Treatment-related adverse events and hospital stay
Positive resection margin, no. (%)
Additional treatment*
Argon plasma coagulation
Adverse event, no. (%) 7 (8.1)
Vertical margin
ESD (n [ 175) P value
.409
Upper third
Horizontal margin
EMR (n [ 86)
10 (11.6)
4 (2.3)
.002
0 (0)
0 (0)
–
9 (10.5)
4 (2.3)
–
338 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
Detailed descriptions of the treatment-related adverse events are shown in Table 4. The ER group had significantly lower rates for all adverse events (2.7% [7/261] vs 7.9% [9/114], P Z .028) and late adverse events (0% vs 5.3%, P Z .001) compared with the surgery group. All of the patients with perforation or postprocedural bleeding www.giejournal.org
Long-term outcomes of endoscopic resection for gastric cancer
100
100
80
80
60
Cumulative DFS rate (%)
Cumulative OS rate (%)
Choi et al
EMR group ESD group
40
20
60
EMR group ESD group
40
20 P = .259 by log-rank test
P = .896 by log-rank test 0
0
0 2
4
6
8
10
0
2
4
Time, year
A
No at risk EMR group ESD group
86 175
85 173
84 167
6
8
74 57
52 0
10
Time, year
81 58
52 0
B
No at risk EMR group 86 ESD group 175
84 171
79 162
Cumulative RFS rate (%)
100
80
60 EMR group ESD group
40
20 P = .101 by log-rank test 0 0
C
No at risk EMR group ESD group
2
4
6
8
73 58
52 0
10
Time, year 86 175
84 171
77 164
Figure 3. Comparisons of long-term outcomes between EMR and endoscopic submucosal dissection (ESD). A, Overall survival (OS). B, Disease-free survival (DFS). C, Recurrence-free survival (RFS).
adverse events after ER were successfully managed by endoscopic treatment. The most common adverse event in the surgery group was postoperative adhesion. A total of 4 patients needed surgical treatment because of adverse events in the surgery group. One patient underwent surgical repair for wound dehiscence 1 month after surgery, and 3 other patients underwent surgery for intestinal obstructions because of postoperative adhesions that occurred more than 3 years after surgery. The mean duration of hospital stay was shorter in the ER group than in the surgery group (2.5 days vs 9.8 days, P ! .001). There were no treatmentrelated or in-hospital mortalities in either treatment group.
This study was conducted to compare the long-term outcomes of ER with those of surgery as curative treatments of EGC lesions that met the criteria for absolute indication. During the median follow-up period of 76.4
months, we found that the OS rate after ER was comparable to that achieved by surgery. Although metachronous gastric cancers after treatment occurred more commonly, adverse events were less frequent in the ER group than in the surgery group. There are several reasons why long-term outcomes after ER are comparable to those after surgery for EGCs that fall under the absolute indication for ER by Japanese criteria. First, the major difference between the 2 treatment modalities is that dissection of lymph nodes can be performed only in the surgery group. However, there was almost no lymph node metastasis reported in EGC patients meeting absolute indications for ER in 1 study.24 Second, the earlier ER method of EMR was replaced with ESD, which has higher rates of en bloc resection and complete resection.21,25 Moreover, additional endoscopic treatments such as argon plasma coagulation, which we used, seem to decrease recurrence, even in patients undergoing EMR with a positive resection margin or piecemeal resection.26 Third, although the metachronous gastric cancer rate was higher in the ER group, surveillance at 1-year intervals or
www.giejournal.org
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 339
DISCUSSION
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
at shorter intervals can detect new lesions at the EGC stage that can be curatively treated mostly by ER or less commonly by surgery. Favorable long-term outcomes have been reported in the limited number of single-arm studies testing ER for EGC that meet the absolute indication. After ER, the 5year OS rates were between 84% and 97%,6,9-11 and the 5-year DSS rates were almost 100%.6,11 The 5-year OS rate for the ER group was 96%, and there was no gastric cancer–related deaths in our study, which was consistent with past studies. Two recent studies that compared long-term outcomes of ER with those of surgery showed that OS rates were not significantly different between groups.12,13 Chiu et al12 reported the long-term outcomes of the ESD and surgery groups (3-year OS rates: 95% in the ESD group and 90% in the surgery group). Limitations of this study were the significantly higher proportion of submucosal lesions (53% vs 11%) and lymph node metastases (10% vs 3%) in the surgery group compared with the ER group. Another study by Choi et al13 also showed that 5-year OS rates were not significantly different between the EMR group (93.6%) and the surgery group (94.2%). However, this study included poorly differentiated or signet ring cell carcinoma
up to 9.3% in the EMR group. Because submucosal layer invasion and undifferentiated type histology were associated with the risk of lymph node metastasis and subsequently might affect survival rates, only EGCs that strictly corresponded to absolute indications were included in our study. ESD is currently the preferred method for EGC treatment because of better short-term treatment outcomes.21,25 The proportions of tumor involvement in the resection margin and additional treatments for the positive resection margin were higher with EMR than ESD. Furthermore, the rates of en bloc resection, and the rate of complete resection was significantly lower with EMR than ESD. However, in this study, there was no local recurrence in patients who underwent EMR and ESD, and the longterm outcomes after EMR were comparable to those after ESD. Therefore, EMR might also be an acceptable method for the treatment of EGC lesions with the absolute indication as previously suggested.11 The major advantage of this study is that we compared the outcomes of EGCs that met an absolute indication for ER suggested by the Japanese guidelines. Recent studies reported no significant differences in survival in patients undergoing ESD for EGCs falling under expanded versus absolute indication.27,28 Therefore, it is possible that at least some of our results are similar for EGCs meeting expanded indications for ER as well. However, further studies are needed to confirm this similar comparability in expanded indication. Another advantage of this study was that we demonstrated the lower adverse event rate associated with ER (2.7%) compared with surgery (7.9%). In contrast, there was no difference in the adverse event rate between EMR and surgery in a previous study.13 However, in the study, 2 patients who underwent gastrectomy died of panperitonitis and small-bowel strangulation. In our study, long-term adverse events occurred only after surgery (5.3%), and surgical treatments were performed for long-term adverse events in 4 patients (3.5%). ER might be a good treatment option that can decrease late adverse events and possibly prevent the need for further surgical intervention. The high rate of metachronous gastric cancer development may be a major negative aspect of ER. The rate of metachronous gastric cancer ranges from 2.9% to 14.0%6,10,13,29 after ER and from 1.8% to 2.4% after surgery.30,31 Similar to previous studies cite, the ER group in our study had significantly higher rates of metachronous gastric cancer and shorter RFS compared with the surgery group. The higher rate of metachronous gastric cancer in the ER group may be related to the larger area of remaining gastric mucosa in the distal part of the stomach, which usually has more severe glandular atrophy or intestinal metaplasia.32 These changes are associated with a higher risk of gastric cancer development. However, most metachronous gastric cancers were detected at an early stage by surveillance endoscopy. As a result, 94% of
340 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
www.giejournal.org
TABLE 4. Adverse events associated with treatments ER Surgery (n [ 261) (n [ 114) P value Early adverse event, no. (%)
7 (2.7)
5 (4.4)*
Postprocedural bleeding
3 (1.1)
0 (0)
Perforation
4 (1.5)
0 (0)
Wound discharge
0 (0)
2 (1.8)
Ileus
0 (0)
2 (1.8)
Intra-abdominal abscess
0 (0)
1 (0.9)
Late adverse event, no. (%)
0 (0)
6 (5.3)*
Postoperative adhesion
0 (0)
3 (2.6)
Intra-abdominal abscess
0 (0)
1 (0.9)
Anastomosis site stricture
0 (0)
1 (0.9)
Wound dehiscence
0 (0)
1 (0.9)
.524
.001
ER, Endoscopic resection. * Two patients in the surgery group had both early and late adverse events.
Choi et al
Long-term outcomes of endoscopic resection for gastric cancer
1. Yoo KY. Cancer control activities in the Republic of Korea. Jpn J Clin Oncol 2008;38:327-33. 2. Park B, Choi KS, Lee YY, et al. Trends in cancer screening rates among Korean men and women: results from the Korean national cancer screening survey (KNCSS), 2004-2011. Cancer Res Treat 2012;44:113-20. 3. Japanese gastric cancer treatment guidelines 2010 (ver. 3). Gastric Cancer 2011;14:113-23. 4. Nashimoto A, Akazawa K, Isobe Y, et al. Gastric cancer treated in 2002 in Japan: 2009 annual report of the JGCA nationwide registry. Gastric Cancer 2013;16:1-27. 5. An JY, Heo GU, Cheong JH, et al. Assessment of open versus laparoscopy-assisted gastrectomy in lymph node-positive early gastric cancer: a retrospective cohort analysis. J Surg Oncol 2010;102:77-81. 6. Isomoto H, Shikuwa S, Yamaguchi N, et al. Endoscopic submucosal dissection for early gastric cancer: a large-scale feasibility study. Gut 2009;58:331-6. 7. Soetikno R, Kaltenbach T, Yeh R, et al. Endoscopic mucosal resection for early cancers of the upper gastrointestinal tract. J Clin Oncol 2005;23:4490-8. 8. Kim SG. Endoscopic treatment for early gastric cancer. J Gastric Cancer 2011;11:146-54. 9. Gotoda T, Iwasaki M, Kusano C, et al. Endoscopic resection of early gastric cancer treated by guideline and expanded National Cancer Centre criteria. Br J Surg 2010;97:868-71. 10. Ahn JY, Jung HY, Choi KD, et al. Endoscopic and oncologic outcomes after endoscopic resection for early gastric cancer: 1370 cases of absolute and extended indications. Gastrointest Endosc 2011;74:485-93.
11. Uedo N, Iishi H, Tatsuta M, et al. Longterm outcomes after endoscopic mucosal resection for early gastric cancer. Gastric Cancer 2006;9:88-92. 12. Chiu PW, Teoh AY, To KF, et al. Endoscopic submucosal dissection (ESD) compared with gastrectomy for treatment of early gastric neoplasia: a retrospective cohort study. Surg Endosc 2012;26: 3584-91. 13. Choi KS, Jung HY, Choi KD, et al. EMR versus gastrectomy for intramucosal gastric cancer: comparison of long-term outcomes. Gastrointest Endosc 2011;73:942-8. 14. Sobin LH, Wittekind C. TNM classification of malignant tumours. 6th ed. New York (NY): Wiley-Liss; 2002. 15. Choi IJ, Kim CG, Chang HJ, et al. The learning curve for EMR with circumferential mucosal incision in treating intramucosal gastric neoplasm. Gastrointest Endosc 2005;62:860-5. 16. Lee JY, Choi IJ, Cho SJ, et al. Routine follow-up biopsies after complete endoscopic resection for early gastric cancer may be unnecessary. J Gastric Cancer 2012;12:88-98. 17. Nakajima T. Gastric cancer treatment guidelines in Japan. Gastric Cancer 2002;5:1-5. 18. Japanese Gastric Cancer Association. Japanese classification of gastric carcinoma – 2nd English edition. Gastric Cancer 1998;1:10-24. 19. Japanese Gastric Cancer Association. Japanese classification of gastric carcinoma: 3rd English edition. Gastric Cancer 2011;14:101-12. 20. Watanabe H, Jass JR, Sobin LH. Histological typing of esophageal and gastric tumors: WHO international histological classification of tumors, 2nd ed. Berlin (Germany): Springer; 1990. 21. Lian J, Chen S, Zhang Y, et al. A meta-analysis of endoscopic submucosal dissection and EMR for early gastric cancer. Gastrointest Endosc 2012;76:763-70. 22. D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265-81. 23. Kurth T, Walker AM, Glynn RJ, et al. Results of multivariable logistic regression, propensity matching, propensity adjustment, and propensity-based weighting under conditions of nonuniform effect. Am J Epidemiol 2006;163:262-70. 24. Gotoda T, Yanagisawa A, Sasako M, et al. Incidence of lymph node metastasis from early gastric cancer: estimation with a large number of cases at two large centers. Gastric Cancer 2000;3:219-25. 25. Park YM, Cho E, Kang HY, et al. The effectiveness and safety of endoscopic submucosal dissection compared with endoscopic mucosal resection for early gastric cancer: a systematic review and metaanalysis. Surg Endosc 2011;25:2666-77. 26. Murakami M, Nishino K, Inoue A, et al. Argon plasma coagulation for the treatment of early gastric cancer. Hepatogastroenterology 2004;51:1658-61. 27. Lee H, Yun WK, Min BH, et al. A feasibility study on the expanded indication for endoscopic submucosal dissection of early gastric cancer. Surg Endosc 2011;25:1985-93. 28. Park CH, Shin S, Park JC, et al. Long-term outcome of early gastric cancer after endoscopic submucosal dissection: expanded indication is comparable to absolute indication. Dig Liver Dis 2013;45:651-6. 29. Nasu J, Doi T, Endo H, et al. Characteristics of metachronous multiple early gastric cancers after endoscopic mucosal resection. Endoscopy 2005;37:990-3. 30. Takeda J, Toyonaga A, Koufuji K, et al. Early gastric cancer in the remnant stomach. Hepatogastroenterology 1998;45:1907-11. 31. Hosokawa O, Kaizaki Y, Watanabe K, et al. Endoscopic surveillance for gastric remnant cancer after early cancer surgery. Endoscopy 2002;34: 469-73. 32. Kim CG, Choi IJ, Lee JY, et al. Biopsy site for detecting Helicobacter pylori infection in patients with gastric cancer. J Gastroenterol Hepatol 2009;24:469-74. 33. Shimamura Y, Ishii N, Nakano K, et al. Repeat endoscopic submucosal dissection for recurrent gastric cancers after endoscopic submucosal dissection. World J Gastrointest Endosc 2013;5:600-4.
www.giejournal.org
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 341
metachronous gastric cancers in our study were successfully treated by repeat ER and did not affect OS as previous studies have reported.33 This study had several limitations. First, because of the retrospective nature of this study, potential selection bias cannot be ignored, especially in terms of tumor size and histological differentiation. There was a greater chance of surgery being selected for patients having relatively larger EGCs on endoscopic evaluation. However, we speculate that this might not affect long-term outcomes because tumor size might not be a major factor, as suggested in the expanded criteria. Moreover, we applied the propensity score–adjustment method to minimize the effects of biases on the clinical outcome.22,23 Second, we used final pathology after ER or surgery rather than the criteria of pretreatment evaluation. This might introduce some bias into our study because we cannot perform an intention-totreatment analysis. Our data suggest that if the final pathology meets guideline criteria, we can then expect a favorable outcome after ER. However, caution should be used in extrapolating our results to preprocedural criteria. In conclusion, we confirmed that long-term OS after ER was comparable to that of surgery with fewer adverse events for EGC lesions that met the absolute indication for endoscopic treatment. Although metachronous gastric cancer occurred more frequently after ER than after surgery, it did not affect long-term OS. Careful endoscopic surveillance is needed to detect metachronous cancer at an early stage. REFERENCES
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
SUPPLEMENTARY TABLE. The characteristics of survival estimates in ER and surgery groups Treatment group
Censored, no. (%) Death, no. (%) Estimated mean survival, year (95% CI)
ER group (n [ 261)
Surgery group (n [ 114)
246 (94.3)
106 (93.0)
15 (5.7)
8 (7.0)
9.5 (9.3–9.7)
9.4 (9.1–9.7)
ER, endoscopic resection; CI, confidence interval.
341.e1 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
www.giejournal.org
Long-term outcome comparison of endoscopic resection and surgery in early gastric cancer meeting the absolute indication for endoscopic resection Il Ju Choi, MD, PhD,1,* Jun Ho Lee, MD, PhD,1,* Young-Il Kim, MD,1 Chan Gyoo Kim, MD, PhD,1 Soo-Jeong Cho, MD, PhD,1 Jong Yeul Lee, MD,1 Keun Won Ryu, MD, PhD,1 Byung-Ho Nam, PhD,2 Myeong-Cherl Kook, MD, PhD,1 Young-Woo Kim, MD, PhD1 Goyang, South Korea
Background: Endoscopic resection (ER) of early gastric cancer (EGC) meeting the absolute indication has excellent long-term outcomes. Objective: To compare long-term outcomes of ER with those of surgery in patients with EGC who met the absolute indication for ER. Design: Retrospective cohort study. Setting: A specialized center for the treatment of cancer. Patients and Interventions: We retrospectively reviewed data from patients who underwent gastrectomy or ER for EGC between 2002 and 2007. Gastric cancers were differentiated-type adenocarcinoma without ulceration confined to the mucosal layer and 2 cm or smaller in size. Main Outcome Measurements: The primary outcome was overall survival (OS). Metachronous cancer rates and adverse event rates were compared. Kaplan-Meier plots and Cox proportional hazard regression analyses were applied for comparisons. Differences in baseline characteristics were adjusted by propensity score. Results: Among 375 patients, 261 underwent ER and 114 underwent surgery. The median follow-up duration was 76.4 months. The 5-year OS rates did not significantly differ between the ER and surgery groups (95.7% vs 93.6%, respectively; P Z .725 by log-rank test). There were no gastric cancer–related deaths in either group. Metachronous gastric cancer developed more frequently in the ER group (6.1%, 16/261) than in the surgery group (0.9%, 1/114) (P Z .024). However, most patients (93.8%, 15/16) in the ER group were curatively treated with repeat ER. Adverse event rates were higher in the surgery group than those in the ER group (7.9% vs 2.7%, P Z .028). Limitations: Retrospective, single-center study. Conclusions: The OS rate after ER for mucosal gastric cancer that met the absolute indication was comparable to that achieved with surgery. Although metachronous cancers were more common after ER, they were usually treatable and did not affect survival. (Gastrointest Endosc 2015;81:333-41.)
Abbreviations: DFS, disease-free survival; DSS, disease-specific survival; EGC, early gastric cancer; ER, endoscopic resection; ESD, endoscopic submucosal dissection; OS, overall survival; RFS, recurrence-free survival; SD, standard deviation. DISCLOSURE: All authors disclosed no financial relationships relevant to this publication. This work was supported by grant 1310280 from the National Cancer Center, South Korea. See CME section; p. 439. Current affiliation of Dr Jun Ho Lee is the Department of Surgery, Sungkyunkwan University School of Medicine, Samsung Medical Center, Seoul, South Korea.
www.giejournal.org
*Drs Il Ju Choi and Jun Ho Lee contributed equally to this work. Copyright ª 2015 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2014.07.047 Received March 12, 2014. Accepted July 17, 2014. Current affiliations: Center for Gastric Cancer (1), Center for Clinical Trials (2), National Cancer Center, Goyang, Korea. Reprint requests: Il Ju Choi, MD, PhD, Center for Gastric Cancer, National Cancer Center, 323 Ilsan-ro Ilsandong-gu, Goyang Gyeonggi, 410-769 South Korea.
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 333
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
Early gastric cancer (EGC) is defined as gastric cancer in which tumor invasion is limited to the mucosa or submucosa. Early diagnosis of gastric cancer has increased in South Korea due to screening programs and recent improvements in diagnostic techniques.1,2 Gastrectomy with lymph node dissection is a standard treatment for gastric cancer, including EGC.3 Long-term outcomes of gastrectomy are excellent, with 5-year overall survival (OS) rates of more than 92%.4,5 Endoscopic resection (ER), including EMR and endoscopic submucosal dissection (ESD), is widely accepted as a curative treatment for EGC without the risk of lymph node metastasis.6-8 Currently, ER is considered a standard treatment for EGC that meets the absolute indication suggested in the Japanese gastric cancer treatment guidelines for ER.3 The 5-year OS rates after ER are between 84% and 97%,6,9-11 which is acceptable for standard treatment. Recent studies show that ER for EGC has long-term outcomes that are comparable to those of gastrectomy.12,13 However, past studies had limitations, including significant differences in tumor characteristics for the ER and gastrectomy groups12 and inclusion of tumors that did not meet the criteria for absolute indication.13 Thus, the aim of this study was to investigate whether long-term outcomes of ER are comparable to those of surgery for EGC that meets the criteria for absolute indication.
from our database of prospectively collected data based on medical records. Tumor stage was based on the sixth edition of the International Union Against Cancer/ American Joint Committee on Cancer classification system.14 The data related to treatments were ER methods, rate of complete resection, surgical method, extent of lymph node dissection, multiple tumors, and adverse events of each treatment. Early adverse events were defined as those that occurred within 30 days of treatment.
METHODS
Pathological evaluation
Study population This study was a retrospective cohort study of 665 patients who underwent gastrectomy or ER for EGC at the Center for Gastric Cancer in the National Cancer Center between January 2002 and December 2007. We included patients who met the following inclusion criteria: (1) 20 years of age and older, (2) newly diagnosed EGC without previous treatment, (3) histologically confirmed well- or moderately differentiated adenocarcinoma, (4) intramucosal tumor without ulceration and 2 cm or less in diameter, (5) received gastrectomy with lymph node dissection or ER with curative intent. Choice of treatment modality was usually decided after full consultation between the attending physician and the patient. Many patients chose surgery as opposed to EMR before April 2004 because ESD was not routinely covered by insurance during that time and because of the limitations in achieving en bloc resection with EMR. A detailed flow chart of this study is summarized in Figure 1. This study was approved by the institutional review board of the National Cancer Center, Korea (NCCNCS-13-798).
ER and surgery All of the patients included in this study underwent EGD and abdominal CT for clinical staging before treatment. EMR was performed until April 2004, and all ER methods were ESD after that time. Detailed ER methods were described previously.15,16 Patients treated by surgery underwent an open or laparoscopic gastrectomy with D1 þ b or more lymph node dissection. A distal subtotal gastrectomy or total gastrectomy was performed depending on the tumor location. Reconstruction methods included gastroduodenostomy or gastrojejunostomy after distal gastrectomy and Roux-en-Y esophagojejunostomy after total gastrectomy. The extent of lymph node dissection was based on the recommendations of the Japanese Gastric Cancer Association.17 All resected specimens were flattened and fixed in 10% formalin for pathological evaluation. Specimens obtained from ER were sliced serially at 2-mm intervals,18,19 embedded in paraffin blocks, and stained with hematoxylin and eosin. Resected surgical specimens were prepared for pathological evaluation in a similar manner except that serial sectioning was performed at 4-mm intervals. Although the Japanese guidelines recommended that the section intervals of surgically resected specimens be 5 to 7 mm, we made sections of our surgical specimens at 4-mm intervals to achieve a more accurate evaluation.19 Histological subtypes were classified according to the World Health Organization classification of gastric cancer.20 Histologically differentiated types included well- or moderately differentiated tubular adenocarcinoma and papillary adenocarcinoma. Undifferentiated types included poorly differentiated tubular adenocarcinoma, signet ring cell carcinoma, and mucinous adenocarcinoma.18
Follow-up surveillance after treatment
Data related to baseline characteristics, endoscopic and pathological findings, and treatments were obtained
EGD was performed 3 and 12 months after ER or surgery and then annually thereafter. In cases of ER, EGD was also performed at 6 months. Abdominal CT was performed annually. A biopsy of the ER scar tissue was performed at each EGD examination to evaluate the presence of local recurrence in patients who underwent ER.
334 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
www.giejournal.org
Data collection
Choi et al
Long-term outcomes of endoscopic resection for gastric cancer
665 Patients underwent gastrectomy or ER for EGC from Jan 2002 to Dec 2007
Excluded (n= 255): Not included in absolute indication of ER
410 Patients were included in the primary cohort
Excluded (n= 35): 24 Previous ER history 7 Gastric surgery history 4 Other organ cancer
375 Patients meeting absolute indication of ER were included in the the final analyses Figure 1. Flow chart of this study. EGC, early gastric cancer; ER, endoscopic resection.
Statistical analysis Continuous variables were expressed as mean with standard deviation or median with interquartile range. For comparisons of baseline and clinicopathological characteristics between the ER group and the surgery group, continuous variables were analyzed by the Student t test or MannWhitney U test, and categorical variables were analyzed by the chi-square or the Fisher exact test. The primary endpoint of this study was OS, and the secondary endpoints were disease-specific survival (DSS), disease-free survival (DFS), recurrence-free survival (RFS), and occurrence of metachronous gastric cancer. OS and DSS were determined from the time of ER or surgery to death. OS included deaths of any cause, but DSS included only gastric cancer related–deaths. DFS was determined from the time of ER or surgery to gastric cancer recurrence, occurrence of a new gastric cancer, or death of any cause. Finally, RFS was determined from the time of ER or surgery to tumor recurrence, death with evidence of recurrence, or occurrence of a metachronous gastric cancer. For RFS, patients who died without known tumor recurrences were censored at the last follow-up time point. In addition to the study endpoints, short-term treatment outcomes, including endoscopic and pathological en bloc resection rates and histologically complete resection rate, were compared between EMR and ESD, and the occurrence of a metachronous gastric cancer was
investigated after ER or surgery. A histologically complete resection was defined when both of the following conditions were fulfilled: (1) en bloc resection and (2) no tumor cells in both horizontal and vertical margins on the final pathological examination.21 A metachronous gastric cancer was defined as a new gastric cancer at a previously uninvolved site in the remnant stomach that was detected at least 1 cm from the muscularis mucosa disruption or from fibrosis at previous ER or anastomosis sites. A new cancer occurring less than 1 year after the initial treatment was defined as a synchronous gastric cancer and was excluded from the analysis. Follow-up data on deaths and recurrences were obtained until December 2011. Survival curves were constructed with the Kaplan-Meier method, and comparisons were performed with the log-rank test. Multivariate analyses for OS, DFS, and RFS were performed using the Cox proportional hazards model. To reduce the effect of selection bias, we used the propensityadjustment method in the multivariate analyses.22,23 The estimated propensity score was obtained from the logistic regression model, which included those factors that were significantly different between the 2 treatment groups. Then the estimated propensity scores were included as a covariate along with age and sex in the multivariate analyses. All statistical analyses were performed with SAS version 8 (SAS Institute, Cary, NC). A P value !.05 was considered statistically significant.
www.giejournal.org
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 335
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
TABLE 1. Baseline and clinicopathological characteristics of the study population
Median age, y (IQR)
ER (n [ 261)
Surgery (n [ 114)
P value
62 (54–68)
62 (54–66)
.541*
Sex, no. (%)
.608
Male
195 (74.7)
88 (77.2)
Female
66 (25.3)
26 (22.8)
Comorbid disease, no. (%)
.083
Diabetes mellitus
10 (3.8)
7 (6.1)
Cardiovascular disease
41 (15.7)
19 (16.7)
Chronic liver disease
11 (4.2)
11 (9.6)
Chronic pulmonary disease
13 (5.0)
1 (0.9)
Location of tumor, no. (%)
.923
Upper third
14 (5.4)
7 (6.1)
Middle third
75 (28.7)
31 (27.2)
Lower third
172 (65.9)
76 (66.7)
1.2 0.4
1.5 0.5
Size of tumor, cm, mean SD Histology of tumor, no. (%)
.029
Well-differentiated
200 (76.6)
75 (65.8)
Moderately differentiated
61 (23.4)
39 (34.2)
Morphology of tumor, no. (%)
.281
Elevated
46 (17.6)
15 (13.2)
Flat or depressed
215 (82.4)
99 (86.8)
74.9 (59.5–92.2)
78.1 (64.2–94.0)
Median follow-up duration, mo (IQR)
!.001
.180*
ER, Endoscopic resection; IQR, interquartile range; SD, standard deviation. * P values were calculated by using the Mann-Whitney U test.
RESULTS
Comparisons of long-term outcomes of ER and surgery
Baseline and clinicopathological characteristics Baseline and clinicopathological characteristics are summarized in Table 1. Of all included patients, 261 patients underwent ER (ER group), and 114 underwent surgery (surgery group). The median age of the 375 patients was 62 years, and there were 283 (75.5%) men. Most tumors (94.4%, 354/375) were located in the middle third or lower third of the stomach. Patients in the ER group had significantly smaller tumors and fewer moderately differentiated adenocarcinomas than those in the surgery group. In the surgery group, the median number of dissected lymph nodes was 36 (interquartile range 27–46). The median follow-up duration of all patients was 76.4 months.
The 5-year OS, DFS, and RFS rates in the ER group were 95.7%, 90.7%, and 94.8%, respectively. The 5-year OS, DFS, and RFS rates in the surgery group were 93.6%, 92.8%, and 99.1%, respectively. In the KaplanMeier analysis, there were no significant differences in OS (Fig. 2A) or DFS (Fig. 2B) between the 2 treatment groups. In contrast, RFS was significantly better in the surgery group than in the ER group (P Z .022 by the log-rank test) (Fig. 2C). Because there were no gastric cancer– related deaths during the follow-up period, the 5-year DSS rates were 100% in both treatment groups. The characteristics of survival estimates in both treatment groups are summarized in the Supplementary Table, available online at www.giejournal.org.
336 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
www.giejournal.org
Long-term outcomes of endoscopic resection for gastric cancer
100
100
80
80
Cumulative DFS rate (%)
Cumulative OS rate (%)
Choi et al
ER group 60 Surgery group 40
20
ER group 60
Surgery group
40
20
P = .725 by log-rank test
P = .200 by log-rank test
0
0 0
A
No at risk ER group Surgery group
2
4
6
8
10
0
Time, year 261 114
258 113
251 110
139 74
52 27
B
2
4
6
8
131 74
52 27
10
Time, year
No at risk ER group 261 Surgery group 114
255 113
241 110
Cumulative RFS rate (%)
100
80 ER group 60
Surgery group
40
20 P = .022 by log-rank test 0 0
C
No at risk ER group Surgery group
2
4
6
8
10
Time, year 261 114
255 113
241 110
131 74
52 27
Figure 2. Comparisons of long-term outcomes between endoscopic resection (ER) and surgery. A, Overall survival (OS). B, Disease-free survival (DFS). C, Recurrence-free survival (RFS) curves.
There were no local recurrences of primary gastric cancer after ER or surgery. Metachronous gastric cancer developed in 17 patients (4.5%, 17/375). Patients in the ER group had a higher rate of metachronous gastric cancer than those in the surgery group (6.1% [16/261] vs 0.9% [1/114], P Z .024). Fifteen of 16 patients (93.8%) in whom metachronous gastric cancer developed were successfully treated with repeat ER. The remaining 1 patient underwent a total gastrectomy because of metachronous gastric cancer that invaded the submucosal layer and was located in the upper third of the stomach. In the surgery group, ER was performed to treat metachronous gastric cancer, which developed in 1 patient. The propensity score was calculated by the multivariate logistic model that included tumor size and histological type and showed significant differences between the ER group and the surgery group. Multivariate analyses with the Cox proportional hazards model adjusted for age, sex, and propensity score showed no significant differences in OS or DFS in the ER group compared with the surgery group (Table 2). However, the ER group showed a worse RFS
rate than the surgery group (adjusted hazard ratio, 8.77; 95% confidence interval, 1.13–67.97; P Z .038) (Table 2).
www.giejournal.org
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 337
Comparisons of short-term and long-term outcomes in EMR and ESD Of the patients in the ER group, 86 patients (33.0%) underwent EMR and 175 (67.0%) underwent ESD. Tumor location and size did not differ between the 2 ER groups. The ESD group had significantly higher endoscopic and pathological en bloc resection rates (Table 3). In addition, the ESD group had a significantly lower rate of tumor involvement in the resection margin and a higher histologically complete resection rate than the EMR group. Of the 14 patients with a positive resection margin, 10 patients underwent argon plasma coagulation and 3 underwent a repeat ER (Table 3). In the ER group, there were no significant differences in OS (P Z .896, by the log-rank test) (Fig. 3A), DFS (P Z .259 by the log-rank test) (Fig. 3B) and RFS (P Z .101 by the log-rank test) (Fig. 3C) between the EMR and ESD groups. In the ER group, metachronous gastric cancer developed in 16 patients, in 9 patients in the EMR group,
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
TABLE 2. Hazard ratios for death and recurrence for patients with early gastric cancer according to the treatment modality Overall survival cHR
95% CI
aHR*
Disease-free survival 95% CI
cHR
95% CI
aHR*
95% CI
Recurrence-free survival cHR
95% CI
aHR*
95% CI
Treatment modality Surgery
1.00
ER
0.86
1.00 0.36-2.02
Propensity scorey
1.00
0.76
0.32-1.84
1.25
0.04-41.34
1.62
1.00 0.77-3.40
1.00
1.63
0.75-3.53
0.64
0.04-9.47
7.41
1.00 0.98-55.88
8.77
1.13-67.97
0.20
0.01-8.67
cHR, Crude hazard ratio; CI, confidence interval; aHR, adjusted hazard ratio; ER, endoscopic resection. * Hazard ratio adjusted for age, sex, and propensity score. y Propensity score was adjusted by size and histology of tumor.
TABLE 3. Comparisons of short-term outcomes according to the ER modality
Median age, y (IQR)
TABLE 3. Continued
EMR (n [ 86)
ESD (n [ 175) P value
61 (53–67)
62 (54–69)
Sex, no. (%)
.329 .495
Male
62 (72.1)
133 (76.0)
Female
24 (27.9)
42 (24.0)
Location of tumor, no. (%) 7 (4.0)
Middle third
24 (27.9)
51 (29.1)
Lower third
55 (64.0)
117 (66.9)
Histology of tumor, no. (%)
7
3
–
Endoscopic resection
2
1
–
Complete resection, no. (%)
67 (77.9)
157 (89.7)
.010
5 (5.8)
2 (1.1)
.032
Bleeding
3 (3.5)
0 (0)
Perforation
2 (2.3)
2 (1.1)
.002
ESD, Endoscopic submucosal dissection; IQR, interquartile range; SD, standard deviation. * One patient did not receive an additional treatment, but local recurrence did not occur.
and 7 patients in the ESD group. There was no difference in metachronous gastric cancer rates between the 2 ER methods (10.5% in EMR and 4.0% in ESD; P Z .054). The 5-year metachronous gastric cancer rates were 7.1% in the EMR group and 4.1% in the ESD group, and the rates did not different between the 2 ER methods in the KaplanMeier analysis (P Z .101 by the log-rank test).
Well differentiated
56 (65.1)
144 (82.3)
Moderately differentiated
30 (34.9)
31 (17.7)
Size of tumor, cm, mean SD
1.1 0.5
1.2 0.5
.280
Endoscopic
65 (75.6)
160 (91.4)
!.001
Pathological
76 (88.4)
171 (97.7)
.003
En bloc resection, no. (%)
Treatment-related adverse events and hospital stay
Positive resection margin, no. (%)
Additional treatment*
Argon plasma coagulation
Adverse event, no. (%) 7 (8.1)
Vertical margin
ESD (n [ 175) P value
.409
Upper third
Horizontal margin
EMR (n [ 86)
10 (11.6)
4 (2.3)
.002
0 (0)
0 (0)
–
9 (10.5)
4 (2.3)
–
338 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
Detailed descriptions of the treatment-related adverse events are shown in Table 4. The ER group had significantly lower rates for all adverse events (2.7% [7/261] vs 7.9% [9/114], P Z .028) and late adverse events (0% vs 5.3%, P Z .001) compared with the surgery group. All of the patients with perforation or postprocedural bleeding www.giejournal.org
Long-term outcomes of endoscopic resection for gastric cancer
100
100
80
80
60
Cumulative DFS rate (%)
Cumulative OS rate (%)
Choi et al
EMR group ESD group
40
20
60
EMR group ESD group
40
20 P = .259 by log-rank test
P = .896 by log-rank test 0
0
0 2
4
6
8
10
0
2
4
Time, year
A
No at risk EMR group ESD group
86 175
85 173
84 167
6
8
74 57
52 0
10
Time, year
81 58
52 0
B
No at risk EMR group 86 ESD group 175
84 171
79 162
Cumulative RFS rate (%)
100
80
60 EMR group ESD group
40
20 P = .101 by log-rank test 0 0
C
No at risk EMR group ESD group
2
4
6
8
73 58
52 0
10
Time, year 86 175
84 171
77 164
Figure 3. Comparisons of long-term outcomes between EMR and endoscopic submucosal dissection (ESD). A, Overall survival (OS). B, Disease-free survival (DFS). C, Recurrence-free survival (RFS).
adverse events after ER were successfully managed by endoscopic treatment. The most common adverse event in the surgery group was postoperative adhesion. A total of 4 patients needed surgical treatment because of adverse events in the surgery group. One patient underwent surgical repair for wound dehiscence 1 month after surgery, and 3 other patients underwent surgery for intestinal obstructions because of postoperative adhesions that occurred more than 3 years after surgery. The mean duration of hospital stay was shorter in the ER group than in the surgery group (2.5 days vs 9.8 days, P ! .001). There were no treatmentrelated or in-hospital mortalities in either treatment group.
This study was conducted to compare the long-term outcomes of ER with those of surgery as curative treatments of EGC lesions that met the criteria for absolute indication. During the median follow-up period of 76.4
months, we found that the OS rate after ER was comparable to that achieved by surgery. Although metachronous gastric cancers after treatment occurred more commonly, adverse events were less frequent in the ER group than in the surgery group. There are several reasons why long-term outcomes after ER are comparable to those after surgery for EGCs that fall under the absolute indication for ER by Japanese criteria. First, the major difference between the 2 treatment modalities is that dissection of lymph nodes can be performed only in the surgery group. However, there was almost no lymph node metastasis reported in EGC patients meeting absolute indications for ER in 1 study.24 Second, the earlier ER method of EMR was replaced with ESD, which has higher rates of en bloc resection and complete resection.21,25 Moreover, additional endoscopic treatments such as argon plasma coagulation, which we used, seem to decrease recurrence, even in patients undergoing EMR with a positive resection margin or piecemeal resection.26 Third, although the metachronous gastric cancer rate was higher in the ER group, surveillance at 1-year intervals or
www.giejournal.org
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 339
DISCUSSION
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
at shorter intervals can detect new lesions at the EGC stage that can be curatively treated mostly by ER or less commonly by surgery. Favorable long-term outcomes have been reported in the limited number of single-arm studies testing ER for EGC that meet the absolute indication. After ER, the 5year OS rates were between 84% and 97%,6,9-11 and the 5-year DSS rates were almost 100%.6,11 The 5-year OS rate for the ER group was 96%, and there was no gastric cancer–related deaths in our study, which was consistent with past studies. Two recent studies that compared long-term outcomes of ER with those of surgery showed that OS rates were not significantly different between groups.12,13 Chiu et al12 reported the long-term outcomes of the ESD and surgery groups (3-year OS rates: 95% in the ESD group and 90% in the surgery group). Limitations of this study were the significantly higher proportion of submucosal lesions (53% vs 11%) and lymph node metastases (10% vs 3%) in the surgery group compared with the ER group. Another study by Choi et al13 also showed that 5-year OS rates were not significantly different between the EMR group (93.6%) and the surgery group (94.2%). However, this study included poorly differentiated or signet ring cell carcinoma
up to 9.3% in the EMR group. Because submucosal layer invasion and undifferentiated type histology were associated with the risk of lymph node metastasis and subsequently might affect survival rates, only EGCs that strictly corresponded to absolute indications were included in our study. ESD is currently the preferred method for EGC treatment because of better short-term treatment outcomes.21,25 The proportions of tumor involvement in the resection margin and additional treatments for the positive resection margin were higher with EMR than ESD. Furthermore, the rates of en bloc resection, and the rate of complete resection was significantly lower with EMR than ESD. However, in this study, there was no local recurrence in patients who underwent EMR and ESD, and the longterm outcomes after EMR were comparable to those after ESD. Therefore, EMR might also be an acceptable method for the treatment of EGC lesions with the absolute indication as previously suggested.11 The major advantage of this study is that we compared the outcomes of EGCs that met an absolute indication for ER suggested by the Japanese guidelines. Recent studies reported no significant differences in survival in patients undergoing ESD for EGCs falling under expanded versus absolute indication.27,28 Therefore, it is possible that at least some of our results are similar for EGCs meeting expanded indications for ER as well. However, further studies are needed to confirm this similar comparability in expanded indication. Another advantage of this study was that we demonstrated the lower adverse event rate associated with ER (2.7%) compared with surgery (7.9%). In contrast, there was no difference in the adverse event rate between EMR and surgery in a previous study.13 However, in the study, 2 patients who underwent gastrectomy died of panperitonitis and small-bowel strangulation. In our study, long-term adverse events occurred only after surgery (5.3%), and surgical treatments were performed for long-term adverse events in 4 patients (3.5%). ER might be a good treatment option that can decrease late adverse events and possibly prevent the need for further surgical intervention. The high rate of metachronous gastric cancer development may be a major negative aspect of ER. The rate of metachronous gastric cancer ranges from 2.9% to 14.0%6,10,13,29 after ER and from 1.8% to 2.4% after surgery.30,31 Similar to previous studies cite, the ER group in our study had significantly higher rates of metachronous gastric cancer and shorter RFS compared with the surgery group. The higher rate of metachronous gastric cancer in the ER group may be related to the larger area of remaining gastric mucosa in the distal part of the stomach, which usually has more severe glandular atrophy or intestinal metaplasia.32 These changes are associated with a higher risk of gastric cancer development. However, most metachronous gastric cancers were detected at an early stage by surveillance endoscopy. As a result, 94% of
340 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
www.giejournal.org
TABLE 4. Adverse events associated with treatments ER Surgery (n [ 261) (n [ 114) P value Early adverse event, no. (%)
7 (2.7)
5 (4.4)*
Postprocedural bleeding
3 (1.1)
0 (0)
Perforation
4 (1.5)
0 (0)
Wound discharge
0 (0)
2 (1.8)
Ileus
0 (0)
2 (1.8)
Intra-abdominal abscess
0 (0)
1 (0.9)
Late adverse event, no. (%)
0 (0)
6 (5.3)*
Postoperative adhesion
0 (0)
3 (2.6)
Intra-abdominal abscess
0 (0)
1 (0.9)
Anastomosis site stricture
0 (0)
1 (0.9)
Wound dehiscence
0 (0)
1 (0.9)
.524
.001
ER, Endoscopic resection. * Two patients in the surgery group had both early and late adverse events.
Choi et al
Long-term outcomes of endoscopic resection for gastric cancer
1. Yoo KY. Cancer control activities in the Republic of Korea. Jpn J Clin Oncol 2008;38:327-33. 2. Park B, Choi KS, Lee YY, et al. Trends in cancer screening rates among Korean men and women: results from the Korean national cancer screening survey (KNCSS), 2004-2011. Cancer Res Treat 2012;44:113-20. 3. Japanese gastric cancer treatment guidelines 2010 (ver. 3). Gastric Cancer 2011;14:113-23. 4. Nashimoto A, Akazawa K, Isobe Y, et al. Gastric cancer treated in 2002 in Japan: 2009 annual report of the JGCA nationwide registry. Gastric Cancer 2013;16:1-27. 5. An JY, Heo GU, Cheong JH, et al. Assessment of open versus laparoscopy-assisted gastrectomy in lymph node-positive early gastric cancer: a retrospective cohort analysis. J Surg Oncol 2010;102:77-81. 6. Isomoto H, Shikuwa S, Yamaguchi N, et al. Endoscopic submucosal dissection for early gastric cancer: a large-scale feasibility study. Gut 2009;58:331-6. 7. Soetikno R, Kaltenbach T, Yeh R, et al. Endoscopic mucosal resection for early cancers of the upper gastrointestinal tract. J Clin Oncol 2005;23:4490-8. 8. Kim SG. Endoscopic treatment for early gastric cancer. J Gastric Cancer 2011;11:146-54. 9. Gotoda T, Iwasaki M, Kusano C, et al. Endoscopic resection of early gastric cancer treated by guideline and expanded National Cancer Centre criteria. Br J Surg 2010;97:868-71. 10. Ahn JY, Jung HY, Choi KD, et al. Endoscopic and oncologic outcomes after endoscopic resection for early gastric cancer: 1370 cases of absolute and extended indications. Gastrointest Endosc 2011;74:485-93.
11. Uedo N, Iishi H, Tatsuta M, et al. Longterm outcomes after endoscopic mucosal resection for early gastric cancer. Gastric Cancer 2006;9:88-92. 12. Chiu PW, Teoh AY, To KF, et al. Endoscopic submucosal dissection (ESD) compared with gastrectomy for treatment of early gastric neoplasia: a retrospective cohort study. Surg Endosc 2012;26: 3584-91. 13. Choi KS, Jung HY, Choi KD, et al. EMR versus gastrectomy for intramucosal gastric cancer: comparison of long-term outcomes. Gastrointest Endosc 2011;73:942-8. 14. Sobin LH, Wittekind C. TNM classification of malignant tumours. 6th ed. New York (NY): Wiley-Liss; 2002. 15. Choi IJ, Kim CG, Chang HJ, et al. The learning curve for EMR with circumferential mucosal incision in treating intramucosal gastric neoplasm. Gastrointest Endosc 2005;62:860-5. 16. Lee JY, Choi IJ, Cho SJ, et al. Routine follow-up biopsies after complete endoscopic resection for early gastric cancer may be unnecessary. J Gastric Cancer 2012;12:88-98. 17. Nakajima T. Gastric cancer treatment guidelines in Japan. Gastric Cancer 2002;5:1-5. 18. Japanese Gastric Cancer Association. Japanese classification of gastric carcinoma – 2nd English edition. Gastric Cancer 1998;1:10-24. 19. Japanese Gastric Cancer Association. Japanese classification of gastric carcinoma: 3rd English edition. Gastric Cancer 2011;14:101-12. 20. Watanabe H, Jass JR, Sobin LH. Histological typing of esophageal and gastric tumors: WHO international histological classification of tumors, 2nd ed. Berlin (Germany): Springer; 1990. 21. Lian J, Chen S, Zhang Y, et al. A meta-analysis of endoscopic submucosal dissection and EMR for early gastric cancer. Gastrointest Endosc 2012;76:763-70. 22. D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265-81. 23. Kurth T, Walker AM, Glynn RJ, et al. Results of multivariable logistic regression, propensity matching, propensity adjustment, and propensity-based weighting under conditions of nonuniform effect. Am J Epidemiol 2006;163:262-70. 24. Gotoda T, Yanagisawa A, Sasako M, et al. Incidence of lymph node metastasis from early gastric cancer: estimation with a large number of cases at two large centers. Gastric Cancer 2000;3:219-25. 25. Park YM, Cho E, Kang HY, et al. The effectiveness and safety of endoscopic submucosal dissection compared with endoscopic mucosal resection for early gastric cancer: a systematic review and metaanalysis. Surg Endosc 2011;25:2666-77. 26. Murakami M, Nishino K, Inoue A, et al. Argon plasma coagulation for the treatment of early gastric cancer. Hepatogastroenterology 2004;51:1658-61. 27. Lee H, Yun WK, Min BH, et al. A feasibility study on the expanded indication for endoscopic submucosal dissection of early gastric cancer. Surg Endosc 2011;25:1985-93. 28. Park CH, Shin S, Park JC, et al. Long-term outcome of early gastric cancer after endoscopic submucosal dissection: expanded indication is comparable to absolute indication. Dig Liver Dis 2013;45:651-6. 29. Nasu J, Doi T, Endo H, et al. Characteristics of metachronous multiple early gastric cancers after endoscopic mucosal resection. Endoscopy 2005;37:990-3. 30. Takeda J, Toyonaga A, Koufuji K, et al. Early gastric cancer in the remnant stomach. Hepatogastroenterology 1998;45:1907-11. 31. Hosokawa O, Kaizaki Y, Watanabe K, et al. Endoscopic surveillance for gastric remnant cancer after early cancer surgery. Endoscopy 2002;34: 469-73. 32. Kim CG, Choi IJ, Lee JY, et al. Biopsy site for detecting Helicobacter pylori infection in patients with gastric cancer. J Gastroenterol Hepatol 2009;24:469-74. 33. Shimamura Y, Ishii N, Nakano K, et al. Repeat endoscopic submucosal dissection for recurrent gastric cancers after endoscopic submucosal dissection. World J Gastrointest Endosc 2013;5:600-4.
www.giejournal.org
Volume 81, No. 2 : 2015 GASTROINTESTINAL ENDOSCOPY 341
metachronous gastric cancers in our study were successfully treated by repeat ER and did not affect OS as previous studies have reported.33 This study had several limitations. First, because of the retrospective nature of this study, potential selection bias cannot be ignored, especially in terms of tumor size and histological differentiation. There was a greater chance of surgery being selected for patients having relatively larger EGCs on endoscopic evaluation. However, we speculate that this might not affect long-term outcomes because tumor size might not be a major factor, as suggested in the expanded criteria. Moreover, we applied the propensity score–adjustment method to minimize the effects of biases on the clinical outcome.22,23 Second, we used final pathology after ER or surgery rather than the criteria of pretreatment evaluation. This might introduce some bias into our study because we cannot perform an intention-totreatment analysis. Our data suggest that if the final pathology meets guideline criteria, we can then expect a favorable outcome after ER. However, caution should be used in extrapolating our results to preprocedural criteria. In conclusion, we confirmed that long-term OS after ER was comparable to that of surgery with fewer adverse events for EGC lesions that met the absolute indication for endoscopic treatment. Although metachronous gastric cancer occurred more frequently after ER than after surgery, it did not affect long-term OS. Careful endoscopic surveillance is needed to detect metachronous cancer at an early stage. REFERENCES
Long-term outcomes of endoscopic resection for gastric cancer
Choi et al
SUPPLEMENTARY TABLE. The characteristics of survival estimates in ER and surgery groups Treatment group
Censored, no. (%) Death, no. (%) Estimated mean survival, year (95% CI)
ER group (n [ 261)
Surgery group (n [ 114)
246 (94.3)
106 (93.0)
15 (5.7)
8 (7.0)
9.5 (9.3–9.7)
9.4 (9.1–9.7)
ER, endoscopic resection; CI, confidence interval.
341.e1 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 2 : 2015
www.giejournal.org