The diagnostic efficacy of natural orifice transluminal endoscopic surgery: is there a role in the intensive care unit?

Abstract

Evaluation of a potential source for abdominal sepsis in a critically ill patient can be challenging. With flexible endoscopy readily available in this setting, we sought to evaluate the diagnostic efficacy of a transgastric natural orifice transluminal endoscopic surgery (NOTES) peritoneoscopy vs. laparoscopic exploration in the identification of intra-abdominal pathology in a porcine model. In this acute study, 15 pigs were randomized to demonstrate 0 to 4 pathologic lesions: small bowel ischemia (SBI), small bowel perforation (SBP), colon perforation (CP), and gangrenous cholecystitis (GC). Two blinded surgical endoscopists were allowed 60 min to perform NOTES or laparoscopy (LAP) to correctly identify or exclude each lesion. A prototype endoscope (R-scope, Olympus, Inc), which enables independent instrument mobility, was used in the NOTES arm. When considering all lesions, LAP was more sensitive diagnostically than NOTES (77.4% vs. 61.3%) overall. LAP also displayed a slightly higher NPV compared with NOTES (79.4% vs. 70.7%). However, NOTES was 100% specific with 100% positive predictive value (PPV) compared with 93.1% and 92.3% with LAP, respectively. Individually, NOTES was found most sensitive with CP identification (87.5%) and least sensitive with SBP (37.5%). The sensitivity of NOTES for SBI and GC was 62.5% and 57.1%, respectively. The utilization of NOTES as a diagnostic tool may have an important role in the critically ill patient when operative intervention is highly morbid. Although it may be overall inferior diagnostically compared with laparoscopy, a positive identification was highly specific with a strong predictive value. Further investigation addressing an improved small bowel evaluation technique would be beneficial. A human trial of NOTES in the ICU utilizing the current technology would still initially mandate laparoscopic or open surgical confirmation and treatment.