Preoperative staging of rectal carcinoma by endorectal ultrasound: is there a learning curve?

Abstract

Endorectal ultrasound (ERUS) is becoming an essential tool in the management of rectal cancer. However, accuracy in the assessment of disease staging may be dependent on operator experience. The aim of this study was to determine if a learning curve exists. From October 1999 to December 2004, all patients with rectal cancer had a pre-operative ERUS performed by a single radiologist. ERUS staging was compared with post-operative pathology findings using the tumour, node, metastases (TNM) classification. The accuracy of ERUS in tumour (T) and node (N) staging after each additional consecutive ten patients was calculated. One hundred and thirty one patients were investigated by ERUS, of which 36 were excluded, leaving 95 patients in the study (60 men). Overall accuracy for T staging was 71.6%. No improvement with experience was noted (p > 0.05). With regard to T staging, ERUS tended to overstage more frequently than understage (24.2 versus 4.2%). The sensitivity, specificity, positive predictive value and negative predictive value of uT3 staging were 96.6, 33.3, 70.4 and 85.7%, respectively. Overall accuracy of uN staging was 68.8%. ERUS tended to overstage nodal disease more frequently than understage (16.1 versus 15.1%). Sensitivity, specificity, positive predictive value and negative predictive value were calculated for ultrasound-detected nodal disease (73.2, 62.2, 74.5 and 60.5%, respectively). Nodal staging accuracy improved from 50% after assessment of 10 cases to 77% after 30 cases were examined. ERUS is an accurate method for staging rectal cancer pre-operatively. Accurate assessment of tumour stage can be achieved immediately by an experienced radiologist without specific training in ERUS. Nodal staging accuracy tends to improve with experience but reaches a plateau after 30 cases.