Résumé
Purpose This study aimed to assess the reliability of rectal villous tumors staging between rectal MRI and histological examination used as the Gold Standard and to investigate causes for discrepancies. Methods The rectal 1.5 T MR scans of 40 patients followed for a histologically proven rectal villous adenoma were retrospectively included. Two independent experienced radiologists staged each tumor according to the TNM classification and described the occurrence of retraction of the rectal wall or spiculations within the associated mesorectum. A third radiologist collected tumor's morphological characteristics. Results Among the 40 villous tumors studied, 25 (63%) were non-invasive and 15 (37%) were invasive. The mean volume of tumors with spiculations and retraction was significantly greater (p < 0.05) compared to tumors without these characteristics. Spiculations and retraction of the rectal wall were observed regardless of the definitive histological stage and did not represent a malignancy criterion. A weak interobserver reliability [Gwet's AC2: 0.31 (0.04-0.57)] in T-staging was observed between the two readers. Reader 1 showed a high reliability [Gwet's AC2: 0.90 (0.81-0.99)] in T-staging between Histopathological examination and preoperative MRI. In the opposite, reader 2 showed a weak reliability [Gwet's AC2: 0.31 (0.03-0.58)] in T-staging. He overstaged all tumors (100%) with spiculations (p < 0.05). Conclusion MRI understaged rectal villous adenoma and was unable to detect degenerative criteria, along with slight Interobserver agreement. The typical worrisome signs of rectal tumor, such as retractions and spiculations, occurred in all stages and were responsible for misstaging in most cases, in particular bulky tumors.