Diagnostic accuracy of magnetic resonance imaging in assessing placental adhesion disorder in patients with placenta previa: Correlation with histological findings

Objective: To evaluate MRI accuracy in assessing placental adhesion disorders (PAD) in patients with placenta previa correlating imaging results with histological findings. Materials and methods: Sixty-one patients who underwent abdomino-pelvic magnetic resonance imaging (MRI) for ultrasound suspicion of PAD were prospectively evaluated. T1- and T2-weighted images, with and without fat suppression, were obtained in the three conventional planes using a 1.5 T MRI scanner. MRI accuracy to evaluate the presence of PAD was assessed on the basis of the occurrence of the following abnormal MRI signs: 1) intraplacental dark bands; 2) focal interruption of myometrial border; 3) intraplacental abnormal vascularity; 4) uterine bulging; 5) tenting of the bladder and/or 6) direct visualization of adjacent tissues invasion only in case of percretism. Imaging results were classified as suggestive or not of PAD using histological data as standard of reference; two methods of imaging analysis were used represented by the presence of at least one (Method A) or two (Method B) abnormal MRI signs; the correlation between the presence of each abnormal MRI sign of PAD and the corresponding histological finding was also assessed. Results: The accuracy, as the area under the receiver operating characteristic curve, was significantly (p = 0.001) higher for Method B (0.92, C.I. 95%: 0.82-0.97) compared to Method A (0.764, C.I. 95%: 0.64-0.86). Among the abnormal MRI signs, intraplacental dark bands and focal interruption of myometrial border were those highly correlated with histological proof of PAD (? > 0.71, p < 0.001, for both); as result, a modified version of Method B (Method C) was identified considering as criterion for PAD the combined presence of the two abnormal MRI signs highly correlated with histologically proven PAD; however, the accuracy of Method C was significantly (p = 0.005) lower (0.80, C.I. 95%: 0.67-0.89) than Method B and comparable to Method A. Conclusions: MRI is a useful imaging technique to assess PAD in patients with placenta previa; in particular, the presence of at least two among all the abnormal MRI signs represents the most accurate criterion (Method B) to identify PAD. Although intraplacental dark bands and focal interruption of myometrial border showed the highest correlation with histological proof of PAD as well as this association was the most frequent in PAD, the combination of these latter MRI signs along with other abnormal signs should be considered diagnostic for PAD.