Background: Because endoscopic ultrasound (EUS) accuracy for staging gastrointestinal tract tumors is limited by many factors, this study was designed to analyze potential sources of error in the EUS staging of colorectal carcinoma.
Methods: All patients referred for EUS evaluation of colorectal carcinoma were staged prospectively by one ultrasonographer and retrospectively by two others with EUS videotape recordings. Pathologic staging was done independently in a blinded fashion. Deceptive pathologic features were defined for T staging by presence of inflammation extending beyond tumor or microscopic spread without inflammation extending to a level consistent with the next stage, and for N staging by large (> or = 10 mm) benign lymph nodes or small (< 10 mm) malignant lymph nodes.
Results: Of 22 patients entered into the study, an inflammatory reaction around microscopic tumor spread thought to actually enhance detection by EUS was present in 57.1% of cases. Nine deceptive pathologic lesions were present in 36.4% (8 of 22) of patients (5 T stage, 4 N stage lesions). Of 40 T and N stage mistakes made by the three physicians, 45% were made in the presence and 55% in the absence of deceptive pathologic lesions. Accuracy increased significantly from the presence to absence of deceptive pathologic lesions, from 53.3% to 83.7% (p = 0.029) for T stage, and 8.3% to 73. 1% for N stage (p = 0.0001). Confidence of T staging correlated significantly with accuracy, increasing from 63.3% when unsure to 88. 2% with staging certainty (p = 0.017), an effect not seen for N staging.
Conclusions: Inflammation and desmoplasia around colorectal carcinoma are often present, but may actually enhance EUS detection of microscopic tumor spread. Deceptive pathologic lesions are present in only one third of patients, but account for almost half (45%) of the errors in T and N staging by EUS. Diagnostic accuracy for EUS was increased with confidence in T stage assessment (but not N stage) and in the absence of deceptive pathologic lesions. Errors in interpretation still accounted for the majority of mistakes (55%) made in EUS staging of colorectal carcinoma.