Value of rectal MRI prior to endoscopic submucosal dissection (ESD): an exploratory study

To correlate rectal MRI findings with endoscopic margin status and depth of invasion in patients undergoing endoscopic submucosal dissection (ESD) for rectal adenomas and early rectal cancers.

Pre-treatment MRIs of patients with colonoscopy-detected polyps and early rectal cancer undergoing curative-intent ESD from 2018 to 2023 were re-interpreted by two radiologists (3- and 25-years' experience) blinded to outcomes. MRI features assessed included largest and smallest length, T2 signal intensity, degree of wall attachment, diffusion restriction and apparent diffusion coefficient values. The reference standard was histopathology. Associations between MRI features and outcomes were tested with Fisher's exact test and Wilcoxon rank sum test. Inter-rater agreement was assessed with kappa statistics and intraclass correlation coefficient.

In 21 patients (median age, 64 years [interquartile range, 54-75]; 12 (57 %) female), final ESD histopathology showed 12 (57 %) adenocarcinomas and 9 (43 %) adenomas. R0 resections were achieved in 11/21 ESD procedures (52.3 %, 95 % CI: 30-74 %). Depth of invasion was correct in 6/21 and 5/21 cases (29 % and 24 %, 95 % CI: 11-52 % and 8-47 %), for 2 readers, respectively. Significant associations with R0 resection for both readers included: smaller mass size (long axis; p = 0.005, 0.003, short axis; p = 0.018, 0.022) and lower degree of wall attachment (p = 0.001, 0.011). Agreement on these three measures was good to low (ICC = 0.87, 0.80, and 0.39), respectively.

In this hypothesis-generating exploration of MRI findings of adenomas and early rectal cancers undergoing ESD, MRI polyp size and degree of wall attachment showed strong negative associations with tumor-free resection margins.