Endoscopic transesophageal vs. thoracoscopic removal of mediastinal lymph nodes: a prospective randomized trial in a long term animal survival model.

Background and study aims: In cases where biopsies remain inconclusive, removal of mediastinal lymph nodes for further analysis requires surgical means. Natural orifice transluminal endoscopic surgery (NOTES) procedures allow incision/closure of the gut wall, which might enable endoscopic excision of pre-marked nodes. The aims of the current study were to investigate the feasibility, safety, and reproducibility of lymph node generation in an animal model to enable endoscopic ultrasound-guided (EUS) lymph node removal (ELR) using transesophageal NOTES access/closure and to compare this procedure with thoracoscopic lymph node removal (TLR) in a randomized long term survival animal study. Patients and methods: Lymph node creation using graphite injection was performed in 12 pigs. After randomization into ELR and TLR groups, lymph nodes were marked with newly developed anchors under EUS guidance and removed using either ELR or TLR. ELR included incision of the esophageal wall and closure after lymph node removal. The main outcome measures were success in lymph node generation, technical success of lymph node removal, complications, and comparability of ELR and TLR. Results: Generation of lymph nodes proved successful in all animals in 46 /48 sites injected (96 %). Anchors were placed through the selected nodes in a mean of 9.4 minutes. TLR and ELR were successful in all cases. One bleeding occurred during esophageal incision in ELR, which was stopped endoscopically. After lymph node removal, endoscopic suturing of the incision took a mean of 18 minutes. Procedure time was longer for ELR than TLR (mean 48 vs. 42 minutes). All animals survived the procedures. Autopsy after 4 weeks showed two thoracic wall abscesses in the TLR group and none in the ELR group. Microscopic analysis revealed well healed esophageal scars. Conclusion: ELR proved to be feasible in this limited sample size and complications were not observed more frequently in this group than in the TLR group.