Role of magnetic resonance imaging (MRI) in early cervical cancer.

The aim of this study was to prospectively determine if invasive carcinoma of the cervix can be ruled out by negative magnetic resonance image (MRI) findings. Eligible patients were those who were scheduled to undergo either hysterectomy or conization because of invasive or non-invasive lesions of the cervix. T2-weighted and T1-dynamic enhancement images were reviewed by two radiologists who had no information on these patients except for the preoperative diagnosis. Clinical information regarding location of the disease was not given. At least 12 longitudinal sections prepared from the surgically removed cervix were reviewed by a pathologist who was not provided with MRI information. The correlation between MRI and histological findings was analyzed in terms of depth of invasion. Seventy cases were entered and 66 were evaluable. Sensitivities of T2 and dynamic MRI for invasive and non-invasive disease by radiologist 1 were 0.8824 and 0.8235, respectively, and those by radiologist 2 were 0.6296 and 0.7647, respectively. Specificities were 0.8776 and 0.8571, respectively, for radiologist 1 and 0.8974 and 0.8367, respectively, for radiologist 2. When sensitivities and specificities were calculated for prediction of non-invasive plus microinvasive lesions < or = 5 mm vs. invasive lesions > 5 mm, specificities of T2 and dynamic MRI became 1.0000 for both radiologists. Negative findings on T2 and/or dynamic MRI indicated non-invasive or early invasive cervical diseases < or = 5 mm. The dynamic technique provided only limited additional value in the detection of early invasive cervical carcinoma.