Intra- and interobserver agreement of rectal cancer staging with MRI.

Abstract

Reliable preoperative staging of rectal cancers is crucial for treatment decision making. To assess the intra- and inter-observer agreement of rectal cancer staging, including the sub-categories, with magnetic resonance imaging (MRI). The study includes 85 patients (35.3% women; mean age = 62.2 ± 11.2 years) who underwent MRI for rectal cancer staging between August 2020 and April 2021. All the stored images were evaluated independently by two radiologists with 10-15 years of experience. For intra-observer agreement, the evaluations were done two months apart. Analyses were made using kappa, prevalence and bias-adjusted kappa (PABAK), and intraclass correlation coefficient (ICC), where appropriate. There was a substantial inter-observer agreement for tumor localization (kappa = 0.665, PABAK = 0.682), mesorectal fascia invasion (kappa = 0.663, PABAK = 0.822), internal and external sphincter involvement (kappa 0.804 and 0.751, PABAK 0.859 and 0.929, respectively), and moderate to substantial agreement for M-staging (kappa = 0.451, PABAK = 0.742) and extramural vascular invasion (kappa = 0.569, PABAK = 0.741). There was also a good inter-observer agreement for T staging and N staging (ICC = 0.862, 95% confidence interval [CI] = 0.788-0.911; and ICC = 0.841, 95% CI = 0.595-0.922, respectively). As expected, intra-observer agreement was better than inter-observer agreement. Intra- and inter-observer agreement for MRI staging of rectal cancers using the structured reporting template is good.