Staging uterine cervical carcinoma with low-field MR imaging

To assess the validity of low-field MR in staging cervical cancer compared to clinical staging.

A total of 95 women entered the study over a 3-year period. MR examinations with a 0.1 T resistive magnet using a body coil and clinical staging according to the FIGO recommendations (1988) were performed within 2 weeks from clinical diagnosis. T1- and T2-weighted sequences were obtained in transversal and sagittal acquisitions, and an additional T1 before and after contrast (randomisation to 0.1 or 0.3 mmol/kg b.w. gadodiamide). Treatment decisions on surgery or radiation therapy were made solely on the clinical staging.

Sixty-one patients were found to be eligible for surgery. In 5 women, the pathological results revealed a more advanced stage of the disease than assessed by clinical staging. MR correctly staged 4 of the 5 but otherwise tended to overstate the disease. Contrast enhancement significantly reduced this trend (p<0.05) regardless of the contrast medium dose used. Divided into two groups, an operable (less than stage 2b) and an inoperable group (more than stage 2a), the clinical staging correctly classified 57 patients (accuracy 92%) compared to 52 patients with MR using contrast enhancement (accuracy 84%). The specificity was no higher than 31%, whereas the reproducibility of the MR assessment was fairly good with kappa values around 0.65 for both intra- and inter-observer variations.

In the present set-up, clinical assessment was superior to low-field MR in staging cervical cancer. When using contrast enhancement, the staging accuracies of low-field MR were comparable to the ones reported for techniques with higher tesla values, whereas the specificity and reproducibility errors were lower. The method, therefore, needs to be optimised.