The performance of magnetic resonance imaging in early cervical carcinoma: a long-term experience

We report our long-term experience of performance of magnetic resonance imaging (MRI) in localizing cervical tumor, assessing tumor size, staging, and lymph node infiltration in patients with early cervical cancer. MRI of 150 patients with early carcinoma between 1995-2005 was retrospectively reviewed. Tumor location, size, tumor distance from internal os, parametrial invasion, myometrial invasion, lymph node size, and location were documented. All patients underwent surgery, pelvic lymphadenectomy, and histological correlation of MRI findings. For staging, MRI and histopathology had kappa value of 0.89. For parametrial invasion, MRI had specificity, negative predictive value (NPV) of 97% and 100%, respectively. For tumor size, MRI and histology had mean difference of -0.9 mm with 95% limits of agreement between -12.6 to +13 mm. In tumors greater than 10 mm, mean difference was 0.3 mm and limits of agreement were -7.5 to +7.9 mm. For internal os involvement sensitivity, specificity, positive predictive value (PPV) and NPV were 90%, 98%, 86%, 98%. respectively. For myometrial invasion sensitivity, specificity, NPV, and PPV were 100%, 99%, 88%, 100%, respectively. Incidence of nodal metastases was 2.9%. On a per-patient basis, sensitivity, specificity for nodal involvement was 37% and 92% and on node-by-node basis, sensitivity and specificity of MRI was 27% and 99%, respectively. Our study confirms MRI is highly accurate in localizing cervical tumor, excluding parametrial invasion, confirming myometrial and internal os invasion. MRI is therefore useful in selecting patients for surgery and mandatory in patients for fertility-preserving surgery. Using accepted size criteria for nodal involvement, MRI is insensitive and currently will not avoid need for pelvic lymphadenectomy.