A retrospective review of magnetic resonance imaging in assessing the extent of myometrial infiltration for patients with endometrial carcinoma.

The surgical treatment of endometrial malignancy is based on pre-operative assessment of differentiation, myometrial and lymph node infiltration. Studies have reported the use of magnetic resonance imaging (MRI) with conflicting results. This study aims to review the pre-operative accuracy of MRI against the histology in endometrial cancer within a district hospital. A total of 112 women were identified from January 1997 to December 2004. The post-operative FIGO staging showed 76.2% Stage I, 7.6% Stage II, 14.3% Stage III and 1.7% Stage IV. A total of 57% (n = 60) had a pre-operative MRI. The myometrial infiltration was difficult to interpret in 31 MRIs. The time between hysteroscopy and MRI scan in those cases was shorter (Mean 3 weeks, SD +/- 1.5) than in reports with a clear interpretation (4.4 weeks, SD +/- 1.6). Sensitivity for MRI to detect deep myometrial infiltration was 35.7% (5/14) and specificity 50% (23/46). Accuracy was 46.6% (28/60). The presented data suggests a poor predictability of myometrial invasion by MRI but the numbers are small and MRI is a rapidly evolving modality. The conclusions must be treated with caution and cannot easily be applied to large gynaecological cancer units in general.