Kohei Takizawa1, 14, Noriya Uedo2, 14, Takuji Gotoda3, 14, Takuji Akamatsu4, 14, Hisashi Doyama5, 14, Osamu Dohi6, 14, Koichi Kawabe7, 14, Shoichi Kayaba8, 14, Yoshimasa Miura9, 14, Hiroya Mizutani10, 14, Haruhisa Suzuki11, 14, Satoshi Ono12, 14, Tatsuya Mikami13, 14, Hideki Ishikawa5, 14

1Shizuoka Cancer Center, Shizuoka, Japan, 2Osaka International Cancer Institute, Osaka, Japan, 3Nihon University School of Medicine, Tokyo, Japan, 4Japanese Red Cross Wakayama Medical Center, Wakayama, Japan, 5Ishikawa Prefectural Central Hospital, Ishikawa, Japan, 6Kyoto Prefectural University of Medicine, Kyoto, Japan, 7Fukaya Red Cross Hospital, Fukaya, Japan, 8Iwate Prefectural Isawa Hospital, Iwate, Japan, 9Jichi Medical University Hospital, Tochigi, Japan, 10The University of Tokyo Hospital, Tokyo, Japan, 11National Cancer Center Hospital, Tokyo, Japan, 12Chiba-Nishi General Hospital, Chiba, Japan, 13Hirosaki University Hospital, Hirosaki, Japan, 14The SET-UP trial group, Tokyo, Japan

Background/Aims: Current Japanese guidelines recommend interrupting low-dose aspirin (LDA) during peri-procedural period of endoscopic submucosal dissection (ESD) for early gastric cancer in low thrombotic risk patients. However, if continuation of LDA during peri-ESD period does not increase bleeding risk significantly, ESD can be performed under sustained anti-thrombotic effect of LDA.

Methods: The study was designed as a prospective randomized controlled non-inferiority trial. Patients with a gastric mucosal neoplasm who were taking LDA were enrolled. After adjustment by tumor location, size, and presence of ulcer findings, the patients were randomly allocated to continuation group (CONT-G) in which ESD is undergone under continuation of LDA, or to interruption group (INT-G) in which ESD is performed with interruption of LDA. Primary endpoint was rate of bleeding that was defined as sum of massive intraprocedural bleeding and delayed bleeding. Required sample size was a total of 560 (non-inferiority margin: 5%).

Results: Between June 2014 and March 2020, 560 patients were enrolled from 66 institutions. After exclusion of 9 cases, 552 patients (432 men, median age of 75 years old) were randomized into each group. In the intent-to-treat analysis, the bleeding rate in CONT-G was 11.2% and that in INT-G was 7.61%, showing the risk difference of 3.62% (95% CI: -0.48-7.73%, p-value for non-inferiority: 0.29). In the per protocol analysis, the bleeding rate in CONT-G was 12.0% and that in INT-G was 7.31%, showing the risk difference of 4.66% (95% CI: 0.41-8.92%, p-value for non-inferiority: 0.45). One patient in CONT-G had cerebral infarction 2 days after ESD. One patient in CONT-G developed delayed bleeding 7 days after ESD, required surgery, and had cerebral infarction after the surgery.

Conclusion: We could not demonstrate non-inferiority ESD under continuation of LDA, thus interruption of LDA during peri-gastric ESD period is recommended in patients with low thrombotic risk.

Keywords: ESD, complication, bleeding, low dose aspirin, gastric neoplasia