Hybrid fluorescent magnetic gastrojejunostomy: an experimental feasibility study in the porcine model and human cadaver.

BACKGROUND

Laparoscopic gastrojejunostomies are time consuming and require a specific training. Alternatively, sutureless anastomosis can be achieved using endoscopically delivered magnetic rings. Our aim was to assess the feasibility and reproducibility of an endo-laparoscopic gastrojejunostomy technique, using a pair of magnets coated with a near-infrared fluorescent biocompatible polymeric material.

METHODS

Five pigs (3 acute and 2 survival models) and one human anatomical specimen were included. In the survival models, the distal ring was inserted into the duodenum endoscopically, and it was fixed to a thread clipped to the gastric mucosa. Twenty-four hours later, a two-port laparoscopy was performed using a near-infrared (NIR) laparoscope. The magnet position in the jejunum was detected with the transluminal fluorescence of the dye. Magnetic interaction with the metallic tip of the laparoscopic grasper allowed to capture the ring and bring the bowel loop to the future anastomotic site on the gastric wall. The proximal magnet was inserted into the stomach endoscopically and released when magnetic interaction started, allowing for a precise connection with the distal ring. The animals were followed up for 12 days and underwent control endoscopies and radiograms. In the acute animals, the anastomotic procedure was repeated 24 times. Finally, the procedure was performed in the human anatomical specimen.

RESULTS

There were no technical problems, and magnetic connection could be precisely directed at both the anterior and posterior gastric walls. No complications occurred during the survival period and the anastomoses were patent on day 5. Transluminal fluorescence enabled a rapid detection of the magnet.

CONCLUSIONS

Hybrid-reduced port magnetic gastrojejunostomy using a pair of fluorescently coated magnetic rings was feasible, reproducible, and easy to perform in both porcine and cadaver models.