Accuracy of Baseline Magnetic Resonance Imaging for Staging Rectal Cancer Patients Proceeding Directly to Surgery
Elasma Milanzi, Rachel Pelly, Ian Hayes, Peter Gibbs, Ian Faragher, Jeanette C. Reece
Abstract
BACKGROUND AND OBJECTIVES: High-resolution magnetic resonance imaging (MRI) accuracy for staging preoperative rectal cancer varies across studies. We examined MRI accuracy for T- and N-staging of rectal cancer compared with final histopathology of the resected specimen in a large Australian cohort who did not receive neoadjuvant therapy or radiation. METHODS: Retrospective analysis of prospectively-collected clinical data from 153 rectal adenocarcinomas locally staged by high-resolution MRI between January 2012 and December 2019 that did not undergo chemoradiotherapy or radiation before surgery. T- and N-stage agreement between MRI and final histopathology was assessed using Kappa statistic. Agreement at each T-stage was evaluated using log-linear modeling. N-staging accuracy was examined using positive and negative predictive values. RESULTS: Overall agreement between MRI and final histopathology for T-stage and N-stage was 55% and 65%, respectively. Kappa statistic found higher agreement between MRI and final histopathology for T-staging (κ = 0.33) versus N-staging (κ = 0.18). MRI correctly assessed 91% of T1 tumors, 43% of T2 tumors, 65% of T3 tumors, and 80% of T4 tumors. MRI accuracy was higher for N-negative tumors (74.1%) than for N-positive tumors (44.4%). CONCLUSION: MRI is moderately accurate at staging T1, T3, and T4 rectal tumors but caution when staging tumors as T2 is advised. Greater accuracy for staging N-negative versus N-positive tumors is indicated.