An Observer Blinded Comparison of Magnetically Assisted Capsule Endoscopy and Conventional Upper Gastrointestinal Endoscopy in Upper GI Bleeding

Intubational endoscopy is uncomfortable for patients and incurs the risks of bleeding, perforation and sedation related complications. Gastroscopy is the 'gold-standard' investigation to identify upper gastrointestinal pathology but is poorly tolerated by patients. Moreover a large proportion of gastroscopies are normal or reveal insignificant pathology and despite efforts to produce pre-procedure scoring systems to grade the appropriateness of the procedure, correlation with diagnostic yield is poor. Capsule endoscopy is very well tolerated but is traditionally reserved for examination of the small bowel since it cannot be manipulated to view all areas of the capacious stomach. A capsule which could overcome this and be steered to examine all areas of the mucosal surface of the stomach as an alternative to intubational endoscopy would be advantageous to patients. Mirocam Navi (MC1000-WM, Intromedic Ltd, Seoul, Korea) is a small bowel capsule endoscope with an inclusion of magnetic material. Available accessories include a handheld magnet with which to gain some control over the capsule in the upper gastrointestinal tract. The capsule is CE marked (CE0120) with the handheld magnet classed as an accessory. In order to assess the feasibility of our hypothesis we conducted a study in an ex vivo porcine stomach model, commonly used for endoscopy training purposes. Different coloured/shaped beads were sewn into each major location of the stomach (cardia, fundus, greater and lesser curve, anterior and posterior wall, antrum and D1). The stomach was distended with 1000mls of water. Endoscopy was performed according to a set protocol. All stomach tags were identified in 87.2% (41/47) of examinations.

This was followed by a double blind, randomised controlled trial comparing magnetically assisted capsule endoscopy (MACE) with conventional upper gastrointestinal endoscopy to detect coloured beads in a porcine stomach model. Gastroscopy correctly identified 88% (79/90) beads, MACE correctly identified 89% (80/90) beads and thus is noninferior to gastroscopy in this setting (95% CI 82.76%-95.24%). Mean examination times for gastroscopy and MACE were 3.34 minutes and 9.90 minutes respectively. Clearly there are significant differences between a porcine model and live human stomach. However, these studies provide proof of the principle that a small bowel capsule endoscope can be manipulated to make targeted movements within the stomach. We now also have one completed human trial assessing whether the magnetic controller can be utilised to manipulate the small bowel capsule through the stomach into the duodenum to improve the completion rate of small bowel capsule examinations. Another trial is also near completion comparing the diagnostic ability ,within the stomach, of MACE compared to conventional gastroscopy in the context of recurrent iron deficiency anaemia.

In order to develop the protocol for our study, discussions were made with other health professionals (consultants, specialist nurses and ward nurses) who are involved in performing or reporting gastroscopy and small bowel capsule endoscopy examinations outside of the research team. Feedback was used to modify the study protocol so it could be as patient and user friendly as possible without compromising the overall aim to answer the research question.

Additionally the protocol has been externally peer reviewed and subsequent modifications made.

From our porcine and two human trials it is clear that to obtain reasonable views of the stomach to be able to guide the capsule the stomach needs to be distended. In our porcine study we achieved good results with 1000mls of water. We felt that this would be an acceptable amount for a patient to drink prior to magnetically assisted capsule endoscopy, and the two human trials we have completed to date have proven excellent tolerance to this volume of water.