OBJECTIVE: To evaluate ultrasonography and magnetic resonance imaging (MRI) performance in differentiating benign leiomyomas from malignant mesenchymal or mixed tumors (MMT) and smooth muscle tumors of uncertain malignant potential of the uterus (STUMP). DESIGN: Retrospective cohort study. SETTING: University hospital, France. POPULATION: One hundred and eight women who underwent imaging before surgery for uterine mesenchymal tumor (85 with benign leiomyomas and 23 with MMT/STUMP). METHODS: The ultrasonography reports were reviewed. Conventional, perfusion and diffusion MRI were blindly analyzed. Recursive partitioning analysis (RPA) was performed to construct diagnostic flowcharts. MAIN OUTCOME MEASURES: Accuracy of a diagnostic flowchart. RESULTS: At ultrasonography, single tumor, non-myometrial origin, absence of acoustic shadowing, thickened endometrium and ascites were associated with MMT/STUMP (p = 0.001, p < 0.001, p = 0.03, p < 0.0001 and p = 0.03, respectively). For conventional MRI, single tumor, non-myometrial origin, large tumor, poorly defined margins, thickened endometrium, peritoneal implants, intermediate or high signal intensity in T1 or T2 sequences, heterogeneous T1 signal, cystic alteration of the tumor and heterogeneity of the tumor's enhancement were significantly associated with MMT/STUMP. Perfusion weighted imaging and perfusion curve types were not discriminant. For diffusion weighted imaging, a high signal intensity at b = 1000 s/mm² was associated with MMT/STUMP (p < 0.001). RPA resulted in a model that ultimately included age, number of tumors and the aspect of the endometrium (both evaluated by MRI) and that had an area under the curve of 0.95. CONCLUSIONS: Simple criteria, such as single tumor, non-myometrial tumor, abnormal endometrium and age, should question the diagnosis of benign leiomyoma. MRI enhanced the sensitivity of detecting MMT/STUMP.
OBJECTIVE: To evaluate ultrasonography and magnetic resonance imaging (MRI) performance in differentiating benign leiomyomas from malignant mesenchymal or mixed tumors (MMT) and smooth muscle tumors of uncertain malignant potential of the uterus (STUMP). DESIGN: Retrospective cohort study. SETTING: University hospital, France. POPULATION: One hundred and eight women who underwent imaging before surgery for uterine mesenchymal tumor (85 with benign leiomyomas and 23 with MMT/STUMP). METHODS: The ultrasonography reports were reviewed. Conventional, perfusion and diffusion MRI were blindly analyzed. Recursive partitioning analysis (RPA) was performed to construct diagnostic flowcharts. MAIN OUTCOME MEASURES: Accuracy of a diagnostic flowchart. RESULTS: At ultrasonography, single tumor, non-myometrial origin, absence of acoustic shadowing, thickened endometrium and ascites were associated with MMT/STUMP (p = 0.001, p < 0.001, p = 0.03, p < 0.0001 and p = 0.03, respectively). For conventional MRI, single tumor, non-myometrial origin, large tumor, poorly defined margins, thickened endometrium, peritoneal implants, intermediate or high signal intensity in T1 or T2 sequences, heterogeneous T1 signal, cystic alteration of the tumor and heterogeneity of the tumor's enhancement were significantly associated with MMT/STUMP. Perfusion weighted imaging and perfusion curve types were not discriminant. For diffusion weighted imaging, a high signal intensity at b = 1000 s/mm² was associated with MMT/STUMP (p < 0.001). RPA resulted in a model that ultimately included age, number of tumors and the aspect of the endometrium (both evaluated by MRI) and that had an area under the curve of 0.95. CONCLUSIONS: Simple criteria, such as single tumor, non-myometrial tumor, abnormal endometrium and age, should question the diagnosis of benign leiomyoma. MRI enhanced the sensitivity of detecting MMT/STUMP.