Objective: To evaluate the utility of preoperative magnetic resonance imaging (MRI) in determining whether primary disease site is cervical or endometrial in patients with inconclusive preoperative endometrial biopsy.

Methods: We retrospectively identified all patients who underwent pelvic MRI and who had a preoperative diagnosis of cervical or endometrial cancer at MD Anderson Cancer Center between 1990 and 2006. The subset in which endometrial biopsy did not clarify the primary disease site was analyzed. Magnetic resonance imaging results were compared with postoperative histopathologic findings.

Results: A total of 168 patients who underwent MRI who had a preoperative diagnosis of cervical or endometrial cancer were identified. Of these patients, 51 had an inconclusive endometrial biopsy. Magnetic resonance imaging suggested an endometrial primary tumor without cervical invasion in 28 patients, of whom 21 (75%) actually had such a tumor and 7 had an endometrial primary tumor with cervical invasion. Magnetic resonance imaging suggested an endometrial primary tumor with cervical invasion in 3 patients, all of whom had such a tumor. Magnetic resonance imaging suggested a cervical primary tumor in 6 patients, of whom 5 had such a tumor and 1 had an endometrial primary tumor without cervical invasion. Magnetic resonance imaging was inconclusive (did not clarify primary disease site or no lesion visualized) in 14 (27%) of 51 patients, 6 of whom had an endocervical primary tumor or an endometrial tumor with cervical involvement. Overall, preoperative MRI was either inaccurate or unhelpful in 22 (43%) of 51 patients.

Conclusion: Preoperative MRI in patients with inconclusive endometrial biopsy is inaccurate or unhelpful in nearly half of patients.