Bilateral Risk Assessments of Surgery and Nonsurgery Contribute to Providing Optimal Management in Early Gastric Cancers after Noncurative Endoscopic Submucosal Dissection: A Multicenter Retrospective Study of 485 Patients

Eriko Koizumi,Osamu Goto,Kohei Takizawa,Yutaka Mitsunaga,Shu Hoteya,Waku Hatta,Atsushi Masamune,Satoshi Osawa,Hiroya Takeuchi,Sho Suzuki,Jun Omori,Go Ikeda,Tsugumi Habu,Yumiko Ishikawa,Kumiko Kirita,Hiroto Noda,Kazutoshi Higuchi,Takeshi Onda,Teppei Akimoto,Naohiko Akimoto,Mitsuru Kaise,Katsuhiko Iwakiri
DOI: https://doi.org/10.1159/000523972
2022-01-01
Digestion
Abstract:Background and Aims: Surgery is recommended in early gastric cancer (EGC) after noncurative endoscopic submucosal dissection (ESD), although observation can be an alternative. We aimed to develop a tailor-made treatment strategy for noncurative EGCs by comparing the lymph node metastasis risk (LNMR) and the surgical risk. Methods: We retrospectively identified 485 patients with differentiated-type, noncurative EGCs removed by ESD and classified them into two groups: a surgery-preferable group and an observation-preferable group, according to the clinical courses. Subsequently, LNMR and surgery-related death risk were assessed using a published scoring system and a risk calculator for gastrectomy, respectively. Finally, we investigated the optimal cutoff value of the risk difference (LNMR minus surgery-related death risk) to efficiently allocate these cases into either of two groups, surgery-preferable or observation-preferable. Results: In 485 patients (surgery in 322, observation in 163), 57 and 428 patients were classified into the surgery-preferable group and the observation-preferable group, respectively. The optimal cutoff value of the risk difference (LNMR minus surgery-related death risk) to allocate the cases to the two preferable groups was 7.85 with the highest area under the curve (0.689). When cases with >7.85 LNMR over the surgery-related death risk were allocated into the surgery-preferable group and vice versa, the discriminability was 73.2%, which was sufficiently higher than that in the clinical decision (44.5%). Conclusion: Personalized comparison of LNMR and surgery-related death risk is helpful to provide a favorable treatment option for each patient with EGCs after noncurative ESD.
What problem does this paper attempt to address?