Interobserver agreement in the staging of rectal cancer using endoscopic ultrasonography.

Abstract

In rectal tumors invasion of the rectal fat and perirectal lymph nodes are generally regarded as independent prognostic factors in most prospective series. There are no studies in the literature concerning interobserver agreement on the staging of rectal cancer by endorectal ultrasonography (EUS). The aim of the present study was to assess interobserver agreement using EUS in the TN staging of rectal cancer. Thirty-seven patients with rectal cancer were investigated at two centers using EUS as part of the pretherapeutic staging (Olympus EUM-3 or EUM-20). All examinations were videotaped and reviewed six months later by four independent observers who assessed the stage of the tumor (from uT1 to uT4) and lymphatic invasion on a blinded basis. When the tumor was assessed as uT3, the observers specified the degree of involvement of the rectal fat (in millimeters). Interobserver agreement was estimated using the kappa coefficient (k) and the intraclass correlation coefficient (ICC). Agreement was classed as poor (k < 0.40), fair to good (0.40 < or = k < 0.75) or excellent (k < or = 0.75). Agreement was fair for uT1 tumors (k = 0.40) and poor for uT2 tumors (k = 0.20). Agreement was good (k = 0.58; CI 0.51 to 0.65) for uT3 tumors; there was a significant interobserver correlation for the exact measure of the extent of rectal fat (ICC = 0.65). The agreement was also good (k = 0.54, CI 0.47 to 0.61) for metastatic lymph nodes. As in the case of esophageal cancer, interobserver agreement on the staging of uT2 tumors is poor with EUS. The evaluation of rectal tumors with a poor prognosis shows good interobserver agreement.