Differentiating cellular leiomyoma from uterine sarcoma and atypical leiomyoma using multi-parametric MRI

OBJECTIVES: To evaluate the diagnostic performance of conventional magnetic resonance imaging (cMRI) combined with diffusion-weighted MRI (DWI) in discrimination of cellular leiomyoma, uterine sarcoma, and atypical leiomyoma. METHODS: This retrospective study enrolled 106 patients with uterine masses, including 51 cellular leiomyomas (CLs), 32 uterine sarcomas (USs) and 23 degenerated leiomyomas (LMs) confirmed by histopathologic examination. Clinical data and imaging findings were assessed. Chi-squared test for qualitative variables and one way ANOVA analysis for quantitative variables were performed. Logistic regression analysis and the receiver operating characteristic (ROC) analysis were performed to determine the cut-off point and diagnostic performances for significant numeric values or multiple models. RESULTS: Morphology (Odds ratio [OR] = 6.36) and margin (OR = 13.84) derived from cMRI were independent indicators for differentiating CLs from USs, and T2WI signal (OR = 0.23) were an independent indicator for differentiating CLs from degenerated LMs (all P < 0.05). The cutoff value of apparent diffusion coefficient (ADC) derived from DWI for differentiating CLs from USs was 839 ×10(-6) mm(2)/sec and was 1239 ×10(-6) mm(2)/sec for differentiating CLs from degenerated LMs. Compared with the use of cMRI features and ADC value alone, combination of independent indicators and ADC value achieved higher AUCs for both differentiations (all P < 0.05). CONCLUSIONS: cMRI is a reliable tool for differentiating CLs from USs and atypical leiomyoma, especially degenerated LMs. The combined use of cMRI and DWI can improve the differential diagnostic performance.