Diseases of the colon and rectum 60(5) 488-496 doi 10.1097/DCR.0000000000000781
Adequate oncologic staging of rectal neoplasia is important for treatment and prognostic evaluation of the disease. Diagnostic methods such as endorectal ultrasound can assess rectal wall invasion and lymph node involvement.
The purpose of this study was to correlate findings of 3-dimensional endorectal ultrasound and pathologic diagnosis of extraperitoneal rectal tumors with regard to depth of rectal wall invasion, lymph node involvement, percentage of rectal circumference involvement, and tumor extension.
Consecutive patients with extraperitoneal rectal tumors were prospectively assessed by 3-dimensional endorectal ultrasound blind to other staging methods and pathologic diagnosis.
Patients who underwent endorectal ultrasound followed by surgery were included in the study.
The study was conducted at a single academic institution.
MAIN OUTCOME MEASURES
Sensitivity, specificity, positive and negative predictive values, area under curve, and κ coefficient between 3-dimensional endorectal ultrasound and pathologic diagnosis were determined. Intraclass correlation coefficient was calculated for tumor extension and percentage of rectal wall involvement.
Forty-four patients (27 women; mean age = 63.5 years) were evaluated between September 2010 and June 2014. Most lesions were malignant (72.7%). For depth of submucosal invasion, 3-dimensional endorectal ultrasound showed sensitivity of 77.3%, specificity of 86.4%, positive predictive value of 85.0%, a negative predictive value of 79.2%, and an area under curve of 0.82. The weighted κ coefficient for depth of rectal wall invasion staging was 0.67, and there was no agreement between 3-dimensional endorectal ultrasound and pathologic diagnosis for lymph node involvement (κ = -0.164). Intraclass correlation coefficient for lesion extension and percentage of rectal circumference involvement were 0.45 and 0.66. A better correlation between 3-dimensional endorectal ultrasound and pathologic diagnosis was observed in tumors <5 cm and with <50% of rectal wall involvement. LIMITATIONS
The relatively small sample size of patients with early rectal lesions referred directly for surgery could represent a potential selection bias.
Three-dimensional endorectal ultrasound was effective for determining rectal wall invasion and lesion extension in tumors <5 cm and with <50% of rectal wall invasion but was limited for detecting lymph node involvement in early rectal lesions.