A Crombe1, N Alberti2, E Stoeckle3, V Brouste4, X Buy2, J-M Coindre5, M Kind2. 1. Department of Radiology, Institut Bergonié, 229 cours de l'Argonne, 33076 Bordeaux Cedex, France. Electronic address: amandine.crombe@ens-lyon.fr. 2. Department of Radiology, Institut Bergonié, 229 cours de l'Argonne, 33076 Bordeaux Cedex, France. 3. Department of Surgery, Institut Bergonié, 229 cours de l'Argonne, 33076 Bordeaux Cedex, France. 4. Clinical and Epidemiological Research Unit, Institut Bergonié, 33000 Bordeaux, France. 5. Department of Pathology, Institut Bergonié, 229 cours de l'Argonne, 33076 Bordeaux Cedex, France.
Abstract
OBJECTIVES: To retrospectively evaluate the diagnostic performance of morphological signs observed on conventional magnetic resonance (MR) imaging to differentiate benign from malignant peripheral solid tumors of soft tissue with myxoid stroma. METHODS: MR images from 95 consecutive histopathologically proven tumors (26 benign and 69 malignant) of soft tissues with myxoid components were evaluated in our tertiary referral center. Two radiologists, blind to pathology results, independently reviewed conventional MR sequences including at least a) one T2-weighted sequence with or without fat suppression; b) one T1-weighted sequence without fat suppression; and c) one T1-weighted sequence with gadolinium-complex contrast enhancement and fat suppression. Multiple criteria were defined to analyze morphology, margins, architecture and tumor periphery and evaluated for each lesion. Intra- and inter-observer reproducibility and Odds ratios were calculated for each criterion. RESULTS: The most relevant and reproducible criteria to significantly predict malignancy were: (1) ill-defined tumor margins, (2) a hemorrhagic component, (3) intra-tumoral fat, (4) fibrosis and (5) the "tail sign". A lesion is classified as malignant if any of these 5 criteria is present, and benign if none of them are observed. Therefore, this combination provides a sensitivity of 92.9% and a specificity of 93.3%. CONCLUSION: Conventional MR imaging provides reproducible criteria that can be combined to differentiate between benign and malignant solid tumors of soft tissue with myxoid stroma.
OBJECTIVES: To retrospectively evaluate the diagnostic performance of morphological signs observed on conventional magnetic resonance (MR) imaging to differentiate benign from malignant peripheral solid tumors of soft tissue with myxoid stroma. METHODS: MR images from 95 consecutive histopathologically proven tumors (26 benign and 69 malignant) of soft tissues with myxoid components were evaluated in our tertiary referral center. Two radiologists, blind to pathology results, independently reviewed conventional MR sequences including at least a) one T2-weighted sequence with or without fat suppression; b) one T1-weighted sequence without fat suppression; and c) one T1-weighted sequence with gadolinium-complex contrast enhancement and fat suppression. Multiple criteria were defined to analyze morphology, margins, architecture and tumor periphery and evaluated for each lesion. Intra- and inter-observer reproducibility and Odds ratios were calculated for each criterion. RESULTS: The most relevant and reproducible criteria to significantly predict malignancy were: (1) ill-defined tumor margins, (2) a hemorrhagic component, (3) intra-tumoral fat, (4) fibrosis and (5) the "tail sign". A lesion is classified as malignant if any of these 5 criteria is present, and benign if none of them are observed. Therefore, this combination provides a sensitivity of 92.9% and a specificity of 93.3%. CONCLUSION: Conventional MR imaging provides reproducible criteria that can be combined to differentiate between benign and malignant solid tumors of soft tissue with myxoid stroma.
Authors: Brandon K K Fields; Natalie L Demirjian; Darryl H Hwang; Bino A Varghese; Steven Y Cen; Xiaomeng Lei; Bhushan Desai; Vinay Duddalwar; George R Matcuk Journal: Eur Radiol Date: 2021-04-23 Impact factor: 5.315