Randomized, blinded comparison of transgastric, transcolonic, and laparoscopic peritoneoscopy for the detection of peritoneal metastases in a human cadaver model

Background: Natural orifice transluminal endoscopic surgery peritoneoscopy may be able to replace laparoscopic peritoneoscopy (LAP) for staging of GI malignancies if it is proven to be equally accurate and safe. Objective: To compare transgastric peritoneoscopy (TGP) and transcolonic peritoneoscopy (TCP) to LAP, pairwise, in a randomized, blinded (to location and number of beads) human cadaver model with simulated peritoneal metastases. Design: Metastases were simulated by 2.5-mm, color-coded beads, which were placed into the peritoneal cavity via an open approach. In previous porcine experiments, LAP resulted in a yield of 95%. By using a noninferiority design with a margin of equivalence of 15%, we needed a sample size of 34 beads for 80% power. Randomization was performed for number and location of beads. Eighteen experiments were performed on 6 fresh-frozen human cadavers. Setting: Experimental surgical laboratory. Intervention: LAP, TGP, and TCP were performed in randomized order by one of two surgeons/endoscopists blinded for location and number of beads. Main Outcome Measurements: Number of beads detected and touched. Results: LAP found and touched 33 beads (yield 97%), TGP 26 beads (76%; difference in yield vs LAP was -20.5 [95% CI, -26.3 to -9.27]), and TCP 29 beads (85%; difference in yield vs LAP was -11.8 [95% CI, -14.6 to 4.98]). Beads that were missed were mostly located at the inferior liver surface: TGP missed 6 of 9 of these beads (67%), TCP 4 of 9 (44%). Limitations: Cadaver model. Conclusion: In this prospective, blinded, comparative trial in a human cadaver model, TCP was comparable to LAP in detecting simulated metastases. TGP was inferior to LAP. Future development should focus on improved visualization of the inferior surface of the liver. (Gastrointest Endosc 2010;72:1027-33.)