Author(s):
Freitas, Patrícia ; Resende-Neves, Teresa ; Lameira, Pedro ; Costa, Marta ; Dias, Paulo ; Filipe, Juliana ; Ferreira, Joana ; Félix, Ana ; Cunha, Teresa Margarida
Date: 2024
Persistent ID: http://hdl.handle.net/10400.17/5194
Origin: Repositório do Centro Hospitalar de Lisboa Central, EPE
Subject(s): HSJ IMA; HCC IMA; HSM IMA; Female; Humans; Portugal; Necrosis; Diagnosis; Differential; Diffusion Magnetic Resonance Imaging; Leiomyoma* / pathology; Leiomyosarcoma* / diagnostic imaging; Leiomyosarcoma* / pathology; Magnetic Resonance Imaging / methods; Myometrium / pathology; Retrospective Studies; Smooth Muscle Tumor* / diagnostic imaging; Smooth Muscle Tumor* / pathology; Uterine Neoplasms* / pathology
Description
Purpose: To evaluate the magnetic resonance imaging (MRI) features that may help distinguish leiomyosarcomas from atypical leiomyomas (those presenting hyperintensity on T2-W images equal or superior to 50% compared to the myometrium). Materials and methods: The authors conducted a retrospective single-centre study that included a total of 57 women diagnosed with smooth muscle tumour of the uterus, who were evaluated with pelvic MRI, between January 2009 and March 2020. All cases had a histologically proven diagnosis (31 Atypical Leiomyomas-ALM; 26 Leiomyosarcomas-LMS). The MRI features evaluated in this study included: age at presentation, dimension, contours, intra-tumoral haemorrhagic areas, T2-WI heterogeneity, T2-WI dark areas, flow voids, cyst areas, necrosis, restriction on diffusion-weighted imaging (DWI), apparent diffusion coefficient (ADC) values, signal intensity and heterogeneity after contrast administration in T1-WI, presence and location of unenhanced areas. The association between the MRI characteristics and the histological subtype was evaluated using Chi-Square and ANOVA tests. Results: The MRI parameters that showed a statistically significance correlation with malignant histology and thus most strongly associated with LMS were found to be: irregular contours (p < 0.001), intra-tumoral haemorrhagic areas (p = 0.028), T2-WI dark areas (p = 0.016), high signal intensity after contrast administration (p = 0.005), necrosis (p = 0.001), central location for unenhanced areas (p = 0.026), and ADC value lower than 0.88 × 10-3 mm2/s (p = 0.002). Conclusion: With our work, we demonstrate the presence of seven MRI features that are statistically significant in differentiating between LMS and ALM.
