Magnetic resonance imaging performance for diagnosis of ovarian torsion in pregnant women with stimulated ovaries.

Background: To determine if asymmetric ovarian edema on non-contrast MRI can be used to distinguish torsed from non-torsed stimulated ovaries in pregnant women.

Methods: In this retrospective study, our radiology database was searched for women who were pregnant and who had undergone ovarian stimulation and underwent MRI abdomen/pelvis from 1/2000-12/2012. At our institution, ultrasound is typically performed as a first line study for pregnant women with pelvic pain, with MR for those patients with indeterminate findings. 64 pregnant women (gestational age range 3-37 weeks) were included. MRI indication, prospective interpretation, operative diagnosis, and follow-up were recorded. Two blinded radiologists (with a third radiologist tie-breaker) independently measured and described the ovaries, including the likelihood of torsion. If one or both ovaries/adnexa had an underlying lesion such as a dermoid, cystadenoma, or abscess, the patient was excluded from size and signal intensity comparison ( = 14). For the remaining 50 women, comparison was made of the ovaries in women with normal ovaries ( = 27), stimulated ovaries without torsion ( = 11), non-stimulated ovaries with torsion ( = 3), and stimulated ovaries with torsion ( = 3). Patients with asymmetric ovarian edema without stimulation or torsion ( = 3) and with polycystic ovary syndrome ( = 3) were analyzed separately.

Results: Average normal ovarian length was 3.2 cm, compared to 4.5 cm for asymmetric edema and 5.6-8.8 cm for the other four groups. Average difference in greatest right and left ovarian diameter was 19% for normal ovaries compared to 24-37% for the other 5 groups. Asymmetric signal on T2-weighted imaging (T2WI) was seen in 12% (3/27) of normal ovaries compared to 9% (1/11) of stimulated patients without torsion, 33% (1/3) of patients with PCOS and 67% (2/3) of patients with torsion both without and with stimulation. The correct diagnosis of torsion was made prospectively in 5/6 cases but retrospectively in only 3/6 cases. In patients with stimulation, correct diagnosis of torsion was made in 2/3 cases prospectively (both with asymmetric T2 signal) and retrospectively in only 1/3 cases. In 13/64 patients, other acute gynecologic and non-gynecologic findings were diagnosed on MRI.

Conclusions: Enlarged edematous ovary can be seen with ovarian stimulation, ovarian torsion, or both. Although asymmetric ovarian edema occurred more frequently in patients with torsion than without, in pregnant patients with stimulated ovaries referred for MRI (typically after non-diagnostic ultrasound), ovarian torsion could not be confidently diagnosed or excluded retrospectively with non-contrast MRI.