Oncogenesis and classification of mixed-type liposarcoma: a radiological, histopathological and molecular biological...

Liposarcomas are separated into clinicopathological entities with a characteristic morphological spectrum and mutually exclusive genetic alterations. Therefore, the rare occurrence of cases with combined patterns of well-differentiated liposarcoma and myxoid liposarcoma designated as mixed-type liposarcoma pose a conceptual problem. Moreover, this feature may have consequences for treatment choice and prognosis. Here, we have dissected the molecular relation of tumor components in cases of mixed-type liposarcoma. On the basis of heterogeneous preoperative magnetic resonance image (MRI) features, eight cases of mixed-type liposarcoma were selected. Preoperative biopsy samples and resection specimens were analyzed including molecular and immunohistochemical analysis on all components. As controls, cases with homogeneous MRI features and uniform aspects of myxoid liposarcoma (n = 5), round cell liposarcoma (n = 5) and well-differentiated liposarcoma (n = 5) were studied. All patients with heterogeneous MRI features showed morphological components of myxoid liposarcoma and well-differentiated liposarcoma. Real-time polymerase chain reaction showed FUS-DDIT3 fusion in both components in five of eight cases in the absence (zero of five) of MDM2 and CDK4 amplification. In three of eight patients, MDM2 and/or CDK4 were overexpressed, and amplification was shown by multiplex ligation-dependent probe amplification (MLPA) in the absence of myxoid liposarcoma translocations. All control patients showed a molecular pattern consistent with their morphological features. Therefore, mixed-type liposarcomas should not be regarded as collision tumors, but as an extreme variant of the morphological spectrum within a single biological entity, explaining the biological contradiction of mixed-type liposarcoma. For treatment stratification, detailed classification including molecular support should be performed in tumors with heterogeneous MRI features.