Localized cervical cancer (stage

Purpose: This study was undertaken to determine the accuracy of magnetic resonance (MR) imaging in the preoperative staging of patients with clinically localised cervical cancer eligible for less extensive surgery. Materials and methods: Fifty-three patients with biopsy-proven carcinoma of the uterine cervix and eligible for conservative surgery prospectively underwent MR imaging. Images were assessed for tumour site and size, infiltration of the cervical stroma, infiltration of vaginal fornices and relationship between the tumour and the internal os of the endocervical canal and the presence and dimensions of pelvic and lumboaortic lymph nodes (cutoff values 1 cm and 0.5 cm minimum axial diameter). MR imaging data were compared with the histopathological findings. Results: The endocervix was the site of origin of 25% (13/53) of the cervical tumours and the exocervix the site of origin of 75% (40/53). In the assessment of cervical stroma infiltration, there was agreement between MR imaging and histopathology in 75% of cases. MR imaging had 67% sensitivity, 92% specificity and 91% diagnostic accuracy in assessing infiltration of the vaginal fornices. In the evaluation of the infiltration of the internal os, MR imaging had 86% sensitivity, 93% specificity and 92% accuracy. In the assessment of the lymph nodes, when using a cutoff value of 1 cm, MR imaging had a sensitivity, specificity and diagnostic accuracy of 28%, 100% and 89%, respectively. With a cutoff value of 0.5 cm, MR imaging had a sensitivity, specificity and diagnostic accuracy of 33%, 92% and 83%, respectively. Conclusions: MR imaging had a high level of accuracy in the preoperative assessment of the extent of cervical tumour in patients eligible for conservative surgery. Accuracy is lower in the evaluation of the pelvic and lumboaortic lymph nodes. © 2009 Springer-Verlag Italia.