850 Prospective Blinded Study of a Magnetically Guided Gastric Capsule As Compared to Conventional Gastroscopy in Patients With Upper Abdominal Complaints

This in turn will improve efficiency in reading.Acknowledgements: This program was supported through National Institute of Health (NIH) grant 4R44DK091964. 850 Prospective Blinded Study of a Magnetically Guided Gastric Capsule As Compared to Conventional Gastroscopy in Patients With Upper Abdominal Complaints Naoki Hosoe*, Haruhiko Ogata, Kazuo Ohtsuka, Noriyuki Ogata, Keiichi Ikeda, Hiroyuki Aihara, Ulrike W. Denzer, Geoffroy Vanbiervliet, Bilal Hotayt, Mohamed H. Abdel-Hamid, Michel Greff, Toshifumi Hibi, Shin-Ei Kudo, Hisao Tajiri, Thomas RoSch, Xavier HeButerne, Jean-Francois Rey School of Medicine, Keio University, Tokyo, Japan; Showa University, Northern Yokohama Hospital, Yokohama, Japan; University Hospital of Medicine, Tokyo Medical and Dental University, Tokyo, Japan; The Jikei University School of Medicine, Tokyo, Japan; Department of Interdisciplinary Endoscopy, University Hospital Hamburg-Eppendorf, Hamburg, Germany; Department of Gastroenterology, Centre Hospitalier Universitaire de Nice, NIce, France; Institut Arnault Tzanck, NIce, France Background and aims Upper GI endoscopy is mostly performed under sedation and has a low yield of relevant gastric lesions in patients without alarm symptoms. A simpler filter tests such as capsule endoscopy would be helpful, but current passive capsules do not provide adequate gastric visualization. The current study evaluates a magnetically guided capsule endoscopy (MGCE) in patients with routine indications for gastroscopy. Methods Symptomatic patients from two French centers underwent subsequent MGCE and conventional gastroscopy, performed independently of each other and primarily in a blinded fashion; final gold standard was unblinded conventional gastroscopy with biopsy under propofol sedation. Main outcome was the accuracy (sensitivity/specificity) of MGCE for the diagnosis of major gastric lesions calculated on a patient basis, defined as those requiring conventional gastroscopy for biopsy or removal. Secondary outcomes were detection of minor lesions, factors influencing accuracy and complications. Results Of 214 eligible patients, 189 patients were included (105 male, 84 female, mean age 53 years); MGCE was performed by 9 examiners, conventional gastroscopy by 6. 23 major lesions were finally diagnosed in 21 patients. With a specificity of 94.1% (CI 89-97), MGCE sensitivity was only 61.9% (CI 34.5-77) and did not correlate with lesion localization, gastric luminal visibility, examiner case volume, or examination time (OR 0.715, p 0.03). Of the remaining patients (n 168), 94% had minor lesions, mostly diffuse and multiple; MGCE made a correct diagnosis in 88.1% (CI 82.2-92.6) of cases with a sensitivity of 89.2% (CI 83.3-93.6) and a specificity of 70.0% (CI 34.8-93.3); of confounding factors, only proximal location of findings was found to be significant. On subsequent interviews, patients preferred MGCE to conventional gastroscopy (VAS 1.17 (95% CI 1.11-1.23) vs. 1.67 (95%CI 1.55-1.79); p 0.001) and unequivocally favored MGCE (100%) for repeated examination. Conclusions In a prospective and strictly blinded study design, MGCE shows insufficient sensitivity for relevant focal gastric lesions despite clear patient preferences for this method. MGCE requires improvement of complete gastric visualization including preparation and capsule guidance.