Objective To clarify the effectiveness of second-look endoscopy (SLE) at preventing bleeding after gastric endoscopic submucosal dissection (ESD).
Design A multicentre prospective randomised controlled non-inferiority trial was conducted at five referral institutions across Japan. Patients with a solitary gastric neoplasm were enrolled. Exclusion criteria were previous oesophagogastric surgery or radiation therapy; perforation and the administration of antithrombotics, steroids or non-steroidal anti-inflammatory drugs. Patients were assigned to the SLE group or the non-SLE group by a computer-generated random sequence after ESD and were treated perioperatively with a proton pump inhibitor. SLE was performed one day after ESD. The primary endpoint was post-ESD bleeding, defined as an endoscopically proven haemorrhage. The trial had the power to detect a non-inferiority criterion of 7% between the groups.
Results From February 2012 to February 2013, 130 and 132 patients were assigned to the SLE and the non-SLE groups, respectively. All patients were included in the intention-to-treat analysis of the primary endpoint. Post-ESD bleeding occurred in seven patients with (5.4%) SLE and five patients with (3.8%) non-SLE (risk difference −1.6% (95% CI −6.7 to 3.5); pnon-inferiority<0.001), meeting the non-inferiority criterion. All 12 patients with post-ESD bleeding and one patient with a delayed perforation were successfully managed with conservative treatment.
Conclusions SLE after gastric ESD is not routinely recommended because it does not contribute to the prevention of post-ESD bleeding for patients with an average bleeding risk.
Trial registration number UMIN-CTR000007170.
- GASTRIC CANCER
- GASTRIC SURGERY
- GASTROINTESINAL ENDOSCOPY
- GASTROINTESTINAL BLEEDING
- SURGICAL COMPLICATIONS
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- GASTRIC CANCER
- GASTRIC SURGERY
- GASTROINTESINAL ENDOSCOPY
- GASTROINTESTINAL BLEEDING
- SURGICAL COMPLICATIONS
Significance of this study
What is already known on this subject?
Endoscopic submucosal dissection (ESD) is accepted as a standard treatment for intramucosal gastric neoplasms.
Post-ESD bleeding is a major complication that occurs in approximately 5% of patients.
To reduce the incidence of post-ESD bleeding, a scheduled second-look endoscopy (SLE) is empirically performed after gastric ESD.
What are the new findings?
The incidences of post-ESD bleeding were 5.4% and 3.8% in the SLE and non-SLE groups, respectively; therefore, non-inferiority of the non-SLE group compared with the SLE group was confirmed with an absolute risk difference of −1.6% (two-sided 95% CI −6.7% to 3.5%, one-sided pnon-inferiority<0.001).
In a comparison between the groups with and without prophylactic coagulation during SLE, the proportion of patients with post-ESD bleeding was similar even when prophylactic coagulation was performed.
Without performing SLEs after gastric ESD, we may save approximately 917 million yen per year in Japan alone.
How might it impact on clinical practice in the foreseeable future?
SLE after gastric ESD is not routinely recommended because it does not contribute to the prevention of post-ESD bleeding for patients with an average risk of bleeding.
Gastric cancer is one of the most common cancers worldwide, and approximately half of the total number of gastric cancer cases occur in East Asia, which has the highest estimated mortality rate (28.1 per 100 000 men and 13.0 per 100 000 women).1 Early diagnosis and treatment are both considered to be effective strategies to reduce the mortality from gastric cancer in Japan, where the prevalence of early gastric cancer exceeds that of advanced cases.2 Endoscopic treatment of early gastric cancer in Japan began in the 1980s. Specifically, the development of the endoscopic submucosal dissection (ESD) method established endoscopy as a standard treatment for early gastric cancer, and the method has been gradually spreading, especially within Asia. The Japanese Ministry of Health, Labour and Welfare officially reported that 3251 early gastric neoplasms were resected throughout Japan in June 2012 alone.3
Post-ESD bleeding is one of the most common adverse events related to the ESD procedure, and it occurs in approximately 5% of patients even with perioperative administration of proton pump inhibitors.4–9 Because performing a second-look endoscopy (SLE) after initial endoscopic haemostasis for peptic ulcer bleeding was suggested to reduce mortality and to improve clinical outcome,10 ,11 SLE was similarly thought to be effective at reducing the incidence of post-ESD bleeding and was empirically performed after gastric ESD. However, because of recent improvements in management strategies for bleeding peptic ulcers, including endoscopic haemostasis with haemoclips or a combination of an injection of epinephrine with thermal therapy and pharmacological therapy using proton pump inhibitors, the most recent data do not support the use of routine SLE.12 ,13 Accordingly, the international consensus recommendations published in 2010 do not endorse routine SLE for average-risk patients in clinical practice.14
According to our previous survey, SLE continues to be performed at most institutions,8 and our previous retrospective analysis suggested that the incidence of post-ESD bleeding was not significantly different before and after performing SLE, although it was not a direct comparative study.15 To address this controversy, we hypothesised that routine SLE would not be necessary after gastric ESD if bleeding and non-bleeding visible vessels were sufficiently treated during the ESD procedure and if perioperative proton pump inhibitors were administered. The aim of the current trial was to clarify the effectiveness of SLE at preventing bleeding after gastric ESD by performing a comparison between groups with and without SLE in clinical practice. Because SLE would not be performed on the day following ESD in the non-SLE group, treatment of the non-SLE group was less invasive than that in the SLE group. Therefore, the hypothesis that routine SLE would not be necessary would be proven by verifying the non-inferiority of the non-SLE group to the SLE group.
Study design and participants
The current multicentre open-label prospective randomised controlled non-inferiority trial was undertaken in five referral institutions throughout Japan. Eligible patients were aged 20 years or older with adequate performance status (Eastern Cooperative Oncology Group 0–2) and normal major organ function; each patient also had histologically confirmed solitary adenocarcinoma or adenoma without lymph node and distant metastasis. The exclusion criteria were previous gastric surgery or gastric tube reconstruction, previous radiation therapy to the upper abdominal region, perforation and the administration of antithrombotic drugs, steroids or non-steroidal anti-inflammatory drugs (NSAIDs) during the perioperative period, which was defined as from 7 days before ESD until postoperative day (POD) 28.
The current trial complied with the Declaration of Helsinki, and the study protocol was approved by the institutional review board of each participating institution. All patients provided written informed consent prior to enrolment.
ESD was performed according to the standard ESD procedure in both the SLE and non-SLE groups. Briefly, the procedure consisted of the following: (1) marking of a circumferential region around the lesion; (2) submucosal injection of solution outside of the marked region; (3) mucosal incision outside of the marked region; (4) additional injection into the submucosa underneath the lesion; (5) submucosal dissection with a cutting device; (6) haemostasis of active bleeding and prophylactic coagulation of visible vessels on the mucosal defect with haemostatic forceps in soft coagulation mode or with clips during both submucosal dissection and at the final step of ESD and (7) retrieval of the specimen. The following choices were made at the discretion of the surgeons: cutting devices: dual knife (KD-650L, Olympus Medical Systems, Co., Tokyo, Japan), insulated-tip knife-2 (KD-611L, Olympus Medical Systems, Co.), SAFE Knife V (DK2518DV1, Fujifilm Medical, Tokyo, Japan) or Clutch Cutter (DP2618DT, Fujifilm Medical); coagulating devices: Coagrasper (FD-410LR, Olympus Medical Systems), Radial Jaw hot biopsy forceps (Boston Scientific Japan, Tokyo, Japan) or Pentax high-frequency haemostatic forceps (HDB2422, Pentax Medical, Tokyo, Japan); electrosurgical generators: VIO300D (ERBE Elektromedizin, GmbH, Tübingen, Germany), ICC200 (ERBE Elektromedizin, GmbH) or ESG-100 (Olympus Medical Systems); clips: EZ clip (HX-610-090/HX-610-090S/HX-610-135, Olympus Medical Systems) and submucosal injection solution: 0.2%–0.4% sodium hyaluronate or normal saline with or without 0.1% epinephrine.
SLE and perioperative management
Patients in the SLE group underwent the standard SLE procedure, which was defined as a scheduled endoscopy performed 1 day after ESD without any suspicion of post-ESD bleeding. When adherent clots (Forest type IIb) were observed on the post-ESD ulcer during SLE, the surgeon carefully checked whether visible vessels (Forest type IIa) existed after removing the clots.16 When visible vessels and/or active bleeding (Forest types Ia and Ib) were observed on the post-ESD ulcer, prophylactic coagulation or endoscopic haemostasis was performed with haemostatic forceps in soft coagulation mode or with clips until active bleeding and visible vessels were sufficiently treated. SLE was not performed in patients in the non-SLE group.
Patients consumed a liquid diet on POD 1 or POD 2; then, the diet changed daily to a soft meal by POD 5; and finally, patients without any complications were discharged from the hospital on POD 5 or POD 6. All patients took 10 mg of sodium rabeprazole once daily starting the day before ESD and for at least 4 weeks thereafter. All patients were observed at follow-up clinic visits for 4 weeks post-ESD. No follow-up endoscopy was performed during the 4-week follow-up period except for SLE in the SLE group and emergency endoscopy when post-ESD bleeding was suspected in both SLE and non-SLE groups.
The prespecified primary endpoint was the proportion of patients who experienced post-ESD bleeding, which was defined as haemorrhage with clinical symptoms and confirmed by emergency endoscopy from the time of the completion of ESD until POD 28. Clinical symptoms were defined as haematemesis, melaena or a decrease in haemoglobin of >2 g/dL since the patient's most recent laboratory test and emergency endoscopy was defined as endoscopy performed on a patient who had clinical symptoms under suspicion of post-ESD bleeding. Confirmation by emergency endoscopy would rule out cases such as haematemesis or melaena associated with intraoperative bleeding during ESD or a decrease in haemoglobin caused by a dilution effect due to drip infusion. Patients who underwent haemostasis for subclinical bleeding without any suspicion of post-ESD bleeding during SLE were not included in the number of patients with post-ESD bleeding. Secondary endpoints were the proportion of patients with post-ESD bleeding after POD 1, the effectiveness of the prophylactic coagulation that was performed in patients in the SLE group, the subgroup analysis for post-ESD bleeding and the percentage of patients who required a blood transfusion. Curative resection was defined as complete tumour removal with tumour-free resection margins and a negligible risk of lymph node metastasis.17
During hospitalisation, adverse events were evaluated daily with patient interviews and physical examinations. A complete blood cell count was assessed on POD 1 or if there were symptoms of bleeding. All adverse events after discharge were verified by an interview with the patient at a consultation 4 weeks after ESD. The responsible clinicians reviewed each patient's medical records and input data into a web-based electronic case record form (University Hospital Clinical Trial Alliance Clinical Research Supporting System, Clinical Research Support Centre, The University of Tokyo).
The current trial was powered for the assessment of non-inferiority of the non-SLE group compared with the SLE group with respect to the primary endpoint. On the basis of the previous large-scale clinical reports of more than 500 cases,4–8 we assumed that the general incidence of post-ESD bleeding would be 5%. According to the multicentre survey, post-ESD bleeding occurred in up to 11.3% of patients post-ESD, even when SLE was performed.8 Therefore, we assumed that the acceptable upper limit of the proportion of patients with post-ESD bleeding in the non-SLE group was 12%, and a non-inferiority margin in the difference in post-ESD bleeding between the SLE and non-SLE groups was set at 7%. With a one-sided type 1 error of 0.05, we calculated that 236 patients would yield a power of at least 80% to detect non-inferiority by the χ2 test; that is, the upper limit of a two-sided 90% CI of difference in post-ESD bleeding risk between the groups (non-SLE minus SLE) included a 7% increase with no more than a 20% chance under the equivalence assumption. Assuming that approximately 5% of patients would be lost to follow-up, the required sample size was determined to be 250 patients.
Randomisation and masking
The responsible clinicians at each institution enrolled patients who met the inclusion and exclusion criteria before ESD. After the completion of ESD, participants were allocated to either the SLE or non-SLE group at a 1:1 ratio via minimisation method using the following four stratification factors: institution, tumour location (antrum or corpus), tumour size (≤20 mm or >20 mm) and ulceration finding (present or absent).18 Briefly, our minimisation programme was provided via a web response system where the allocation sequence was computer-generated (University Hospital Clinical Trial Alliance Clinical Research Supporting System, Clinical Research Support Centre, The University of Tokyo). Through this minimisation programme, which could be accessed via internet by the responsible clinicians at each institution, the first patient was randomly allocated. Then, subsequent patient's allocation were sequentially determined so as to minimise the imbalance of the four stratification factors between the groups: (1) for each stratification factor, the number of previously allocated patients who had the same value as the patient to be allocated was counted in each group; (2) the sum of the four numbers was calculated in each group and (3) the patient was allocated with high probability (from 50% to 99% according to the difference between the groups) to the group with the lower score.19 The numbers in the algorithm were updated in the next allocation. We did not attempt to mask the patients or clinicians to the allocated treatment group.
The primary endpoint was analysed according to the intention-to-treat (ITT) principle. A Dunnett–Gent test was used to analyse the non-inferiority of the primary endpoint. We censored patients from the Kaplan–Meier plots when they reached the endpoint or when they were lost to follow-up. We also analysed the time to reach the endpoint according to the Kaplan–Meier method and applied the log-rank test to compare the incidence of the endpoint between the two groups. We assessed the continuous variables with Welch's t test and categorical variables with Fisher's exact test or the Wilcoxon signed-rank test, as appropriate. We used a one-sided p value of <0.025 to indicate the statistical significance of the non-inferiority of the primary endpoint (post-ESD bleeding) against the test hypothesis of 7% risk difference. For other endpoints, we conducted ordinary statistical tests against the null hypothesis of equivalence between groups; therefore, a two-sided p value of <0.05 was considered to indicate statistical significance. All CIs were set at the 95% confidence level at the intersection of the non-significant hypothesis set of the upper 2.5% tests and the lower 2.5% tests. JMP V.9.03 (SAS Institute, Cary, North Carolina, USA) was used for the analyses. Data processing and statistical analyses were conducted by an independent statistician. All authors had access to the study data and have reviewed and approved the final manuscript. The trial is registered with UMIN-Clinical Trials Registry, number UMIN-CTR 000007170.
Between February 2012 and February 2013, 276 patients were enrolled in the trial. Seven patients (2.5%) had perforations during ESD, two (0.7%) were treated with endoscopic mucosal resection, two (0.7%) withdrew informed consent and three (1.1%) violated the exclusion criteria before randomisation (figure 1). We randomly assigned 262 patients to either the SLE group (n=130) or the non-SLE group (n=132). In the SLE group, two patients withdrew informed consent, and post-ESD bleeding occurred in two patients before SLE; thus, 126 (96.9%) of the 130 patients in the SLE group underwent SLE, whereas no patient underwent SLE in the non-SLE group. An additional two patients in the SLE group violated the exclusion criteria after randomisation due to the use of an antithrombotic medication or NSAID during the perioperative period. All 262 patients were followed up for at least 4 weeks after ESD and were included in the ITT analysis of the primary endpoint.
Baseline characteristics of the patients, the lesions and the procedures are shown in table 1. Longitudinal location, tumour size, ulceration finding and institution, which were predefined as stratification factors, were well-balanced between the groups. A total of 47 operators (16 fellows and 31 trainees) were involved in this study. Operator experience and other parameters were well-balanced between the groups. All 262 patients underwent ESD in an en bloc manner, and curative resection was achieved in 110 (84.6%) and 113 (85.6%) patients in the SLE and non-SLE groups, respectively.
A total of 12 (4.6%) patients—7 (5.4%) and 5 (3.8%) patients in the SLE and non-SLE groups, respectively (p=0.570)—reached the primary endpoint (table 2). Non-inferiority of the non-SLE group compared with the SLE group was confirmed with an absolute risk difference of −1.6% (two-sided 95% CI −6.7% to 3.5%, one-sided pnon-inferiority <0.001) according to ITT analysis of the primary endpoint (figure 2). Moreover, with a strict non-inferiority margin of 4%, the result of non-inferiority of the non-SLE group compared with the SLE group was unchanged (one-sided pnon-inferiority=0.015). In the SLE group, two patients withdrew informed consent before SLE and two patients violated the exclusion criteria (figure 3). To exclude the possibility that the protocol-mandated SLE allocation may have influenced the outcome, we performed a per-protocol analysis (126 (96.9%) patients in the SLE group and 132 (100%) patients in the non-SLE group). Post-ESD bleeding occurred in 7 (5.6%) out of 126 patients in the SLE group and in 5 (3.8%) out of 132 patients in the non-SLE group, which resulted in the non-inferiority of the non-SLE group with an absolute risk difference of −1.8% (two-sided 95% CI −7.0% to 3.4%, one-sided pnon-inferiority<0.001); this finding was consistent with the result of the ITT analysis of the primary endpoint (figure 2). The findings regarding post-ESD bleeding were also consistent across the subgroups (figure 4).
The timing of post-ESD bleeding is shown in figure 5. All 12 cases of post-ESD bleeding occurred prior to POD 15, and 4 (33.3%) of the 12 cases occurred within 24 h post-ESD. Post-ESD bleeding after POD 1 occurred in five (3.8%) patients in the SLE group and in three (2.3%) patients in the non-SLE group (p=0.499) (table 2). Post-ESD bleeding after POD 5, determined as part of an ancillary analysis, occurred in six patients: five (3.8%) in the SLE group and one (0.8%) in the non-SLE group (p=0.119). The time-to-event curve for post-ESD bleeding showed no statistically significant difference between the groups (p=0.549) (figure 6).
Prophylactic coagulation in the visible vessels was performed in 42 (33.3%) of the 126 patients in the SLE group during the SLE procedure (figure 3). From another ancillary analysis, we determined that post-ESD bleeding occurred in 3 (7.1%) of the 42 patients in whom prophylactic coagulation was performed during SLE and in 2 (2.4%) of the 84 patients who did not receive prophylactic coagulation (p=0.332) (table 3).
Other adverse events in each treatment group are shown in table 4. On the day after ESD, one patient with a 1.5 cm intramucosal early gastric neoplasm on the anterior wall of the gastric body presented with sudden epigastric pain and was found to have delayed perforation by an abdominal CT scan. All cases of post-ESD bleeding and delayed perforation were successfully managed with conservative treatment and without blood transfusions or surgery (table 2).
According to our supplementary analyses of risk factors for post-ESD bleeding, a resected specimen size >40 mm seemed to be a risk factor for post-ESD bleeding (see online supplementary tables S1 and S2).
In this multicentre, prospective, randomised controlled non-inferiority trial, the non-inferiority of omission of SLE after gastric ESD on the proportion of patients with post-ESD bleeding compared with the performance of SLE was demonstrated. In April 2013, a randomised clinical trial that evaluated similar outcomes suggested that no significant difference existed in post-ESD bleeding between patients in the SLE and non-SLE groups (p=0.66).20 However, the small sample size in that study (n=155 from a single centre) did not indicate a 10% difference in the proportion of patients with post-ESD bleeding between the groups (two-sided α=0.05, power=0.38). Accordingly, we are convinced that SLE after gastric ESD should not be routinely performed based on the results of our study.
With regards to the timing of post-ESD bleeding, it has been reported that 25%–75% of the cases of post-ESD bleeding occurred within 24 h after ESD.6 ,15 Similar to the previous reports, 4 (33%) of the 12 cases of post-ESD bleeding in the current trial occurred within 24 h post-ESD. Because SLE was performed on the day after ESD, we analysed the incidence of post-ESD bleeding after POD 1 as a secondary endpoint with which to assess the potential effects of SLE; however, the difference between the groups was not statistically significant. Interestingly, the proportion of patients with post-ESD bleeding after POD 5 was rather high in the SLE group compared with the non-SLE group, although this difference was not statistically significant.
Prophylactic coagulation was performed in one-third of the patients during SLE; most patients, even those in the SLE group, did not require prophylactic coagulation. The proportion of patients with post-ESD bleeding was similar in the groups with and without prophylactic coagulation during SLE, even when prophylactic coagulation was performed. The reason why the delayed bleeding still occurred, particularly in patients who underwent prophylactic coagulation, is unclear. Unrecognised arteries that were not coagulated during SLE, thick arteries that were not coagulated completely during SLE, air insufflation and prophylactic coagulation that may have induced tissue damage or necrosis during SLE may have contributed to the exposure of arteries on the base of the ulcer, which in turn contributed to the delayed bleeding. From this perspective, prophylactic coagulation during SLE cannot prevent post-ESD bleeding and may even increase the incidence of post-ESD bleeding.
With regards to cost-effectiveness, the Japanese national insurance system has set the fee for a diagnostic oesophagogastroduodenoscopy at 11 400 yen, whereas that for endoscopic haemostasis or coagulation it is 46 000 yen. Approximately, 40 000 ESDs are performed each year in Japan, according to the official report by the Japanese Ministry of Health, Labour and Welfare.3 Therefore, the total cost of SLE is estimated to be approximately 917 million yen per year, assuming that endoscopic haemostasis or coagulation is performed in one-third of all SLE cases. According to our findings, when unnecessary SLEs after gastric ESD are avoided, a substantial amount of money and human resources may be saved and the burden of endoscopy practices may be reserved for other purposes.
Our trial does have some limitations. First, we excluded patients with a high risk of bleeding, such as those who had chronic renal failure or liver cirrhosis and who continued using antithrombotic drugs, anticoagulants, steroids or NSAIDs during the perioperative period. However, SLE may not be effective in this population either and may even increase the incidence of post-ESD bleeding, as observed in the current study, because SLE may not reduce the bleeding of unrecognised or thick arteries, and air insufflation with over-coagulation may induce tissue damage or necrosis during SLE. Second, we only enrolled patients who were undergoing gastric ESD in advanced, high-volume institutions in Japan. Gastric ESD is technically demanding; thus, the technical aspects of ESD may affect the risk of post-ESD bleeding in low-volume centres within and outside of Japan, as it has been reported that longer procedure times may affect post-ESD bleeding.7 ,21 ,22 Therefore, it is recommended that our findings be confirmed in other practical settings in future studies.
In conclusion, non-inferiority of the non-SLE group compared with the SLE group with respect to the incidence of post-ESD bleeding was demonstrated. When bleeding and non-bleeding visible vessels are sufficiently treated during the ESD procedure and perioperative proton pump inhibitors are administered, SLE after gastric ESD is not routinely recommended for patients without high bleeding risks because SLE does not contribute to the prevention of post-ESD bleeding. Drugs and prophylactic coagulation during SLE may even increase the incidence of post-ESD bleeding.
We thank Tomohiro Shinozaki (Department of Biostatistics, School of Public Health, Graduate School of Medicine, The University of Tokyo) for his insightful comments and advice for analyses in this study; we also thank Shigetaka Yoshinaga, Satoru Nonaka, Seiichiro Abe, Yasuhiro Ohno, Tomoyuki Odagaki, Atsushi Yagishita, Yusuke Yoda, Tomoji Kato, Satoshi Ono, Keiko Niimi, Yosuke Tsuji, Itsuko Asada-Hirayama, Maki Konno-Shimizu, Rie Matsuda, Chihiro Minatsuki,Yoshiki Sakaguchi, Yu Takahahi and Chiemi Nakayama for the enrolment of patients and data collection; and finally, we thank Kanako Yoshida for data entry.
Collaborators Tomohiro Shinozaki; Shigetaka Yoshinaga; Satoru Nonaka; Seiichiro Abe; Yasuhiro Ohno; Tomoyuki Odagaki; Atsushi Yagishita; Yusuke Yoda; Tomoji Kato; Satoshi Ono; Keiko Niimi; Yosuke Tsuji; Itsuko Asada-Hirayama; Maki Konno-Shimizu; Rie Matsuda; Chihiro Minatsuki; Yoshiki Sakaguchi; Yu Takahahi; Chiemi Nakayama; Kanako Yoshida.
Contributors MF, NU and SM were responsible for the concept of the trial and the formulation of the study design, which all authors subsequently accepted. MF and SM were responsible for the management of the trial and for the writing of the report. TS was responsible for data management and for the implementation of the statistical analysis. SK and NY contributed to the trial concept and design. The initial draft was written by MF and SM and was revised based on comments from all of the other authors. All authors contributed to the enrolment of patients and to the data collection, reviewed the final submitted report and agreed with its contents. The authors take full responsibility for the report.
Funding The current trial was self-funded.
Competing interests MF has received unrestricted grants from Astellas Pharmaceutical, Takeda Pharmaceutical, Zeria Pharmaceutical, Otsuka Pharmaceutical, AstraZeneca Pharmaceutical, Dainihon-Sumitomo Pharmaceutical, Taiho Pharmaceutical, Ajinomoto Pharmaceutical, Daiichi-Sankyo Pharmaceutical, Aska Pharmaceutical, and Eisai for his department; non-financial support from HOYA Pentax, Olympus Medical Systems, and Fujifilm for his department; and honoraria from Olympus Medical Systems, HOYA Pentax, Eisai, Daiichi-Sankyo Pharmaceutical, AstraZeneca Pharmaceutical, Aska Pharmaceutical, Taisho-Toyama Pharmaceutical, Otsuka Pharmaceutical, Zeria Pharmaceutical, Takeda Pharmaceutical, Astellas Pharmaceutical, Johnson & Johnson, Ajinomoto Pharmaceutical, Amco, Novartis Pharmaceutical, Boston Scientific, and Boehringer-Ingelheim outside the submitted work. The other authors declare that they have no conflicts of interest.
Ethics approval IRB of the participating institutions.
Provenance and peer review Not commissioned; externally peer reviewed.