Background: Accurate preoperative staging of primary rectal cancer is mandatory because the result may affect therapeutic decisions. Endoscopic ultrasonography (EUS) is considered the most accurate method for locoregional staging, but the issue of possible variations in the assessment of its accuracy related to technical aspects of pathologic staging has never been raised. The aim of this study was to assess EUS results as determined by two different methods of dissection of surgical specimens. Methods: Among all cases with primary rectal cancer staged with EUS from April 1991 to April 1997, 131 patients underwent surgery without preoperative radiotherapy; EUS results for nodal staging were compared with those obtained by pathology. Resected specimens were examined using two different techniques (conventional vs. special dissection). Results: There was a significant decrease in diagnostic accuracy of EUS according to pathologic technique. Overall accuracy, sensitivity, specificity, positive and negative predictive values for conventional versus special dissection were as follows: 74.6% vs. 43.3% (p = 0.0001), 67.8% vs. 21.8% (p = 0.0002), 79.1% vs. 67.8% (p = 0.14), 67.8% vs. 43.7% (p = 0.02), and 79.1% vs. 43.2% (p = 0.0003), respectively. EUS sensitivity according to size of metastatic lymph nodes was significantly lower for nodes smaller than 5 mm in diameter (p = 0.025) when special dissection was performed because of a larger number of lymph nodes harvested. Conclusions: Our findings raise concern about the results of EUS staging of lymph node metastases in rectal cancer. Further prospective studies on a node-by-node basis could clarify the real diagnostic yield of EUS.
ASJC Scopus subject areas