PURPOSE: To review the reliability of computed tomographic (CT) and magnetic resonance (MR) imaging features in distinguishing lipoma and well-differentiated liposarcoma. MATERIALS AND METHODS: CT (n= 29) and MR (n = 40) images and radiographs (n = 28) of 60 patients with histologically verified fatty tumors (35 lipomas and 25 well-differentiated liposarcomas) were retrospectively reviewed in 31 females and 29 males (mean age, 56 years; age range, 1-88 years). Images were assessed for adipose tissue content, and non-fatty component was classified (thin and/or thick septa and nodular and/or globular components) as absent, mild, moderate, or pronounced. Also assessed were signal intensity and tissue attenuation of the fatty components and non-adipose elements. RESULTS: Statistically significant imaging features favoring a diagnosis of liposarcoma included lesion larger than 10 cm (P <.001), presence of thick septa (P =.001), presence of globular and/or nodular non-adipose areas (P =.003) or masses (P =.001), and lesion less than 75% fat (P <.001). The most statistically significant radiologic predictors of malignancy were male sex, presence of thick septa, and associated non-adipose masses, which increased the likelihood of malignancy by 13-, nine-, and 32-fold, respectively. Both lipoma and liposarcoma demonstrated thin septa and regions of increased signal intensity on fluid-sensitive MR images. CONCLUSION: A significant number of lipomas will have prominent non-adipose areas and will demonstrate an imaging appearance traditionally ascribed to well-differentiated liposarcoma. Features that suggest malignancy include increased patient age, large lesion size, presence of thick septa, presence of nodular and/or globular or non-adipose mass-like areas, and decreased percentage of fat composition.
PURPOSE: To review the reliability of computed tomographic (CT) and magnetic resonance (MR) imaging features in distinguishing lipoma and well-differentiated liposarcoma. MATERIALS AND METHODS: CT (n= 29) and MR (n = 40) images and radiographs (n = 28) of 60 patients with histologically verified fatty tumors (35 lipomas and 25 well-differentiated liposarcomas) were retrospectively reviewed in 31 females and 29 males (mean age, 56 years; age range, 1-88 years). Images were assessed for adipose tissue content, and non-fatty component was classified (thin and/or thick septa and nodular and/or globular components) as absent, mild, moderate, or pronounced. Also assessed were signal intensity and tissue attenuation of the fatty components and non-adipose elements. RESULTS: Statistically significant imaging features favoring a diagnosis of liposarcoma included lesion larger than 10 cm (P <.001), presence of thick septa (P =.001), presence of globular and/or nodular non-adipose areas (P =.003) or masses (P =.001), and lesion less than 75% fat (P <.001). The most statistically significant radiologic predictors of malignancy were male sex, presence of thick septa, and associated non-adipose masses, which increased the likelihood of malignancy by 13-, nine-, and 32-fold, respectively. Both lipoma and liposarcoma demonstrated thin septa and regions of increased signal intensity on fluid-sensitive MR images. CONCLUSION: A significant number of lipomas will have prominent non-adipose areas and will demonstrate an imaging appearance traditionally ascribed to well-differentiated liposarcoma. Features that suggest malignancy include increased patient age, large lesion size, presence of thick septa, presence of nodular and/or globular or non-adipose mass-like areas, and decreased percentage of fat composition.