Distinguishing soft tissue sarcomas of different histologic grades based on quantitative MR assessment of intratumoral heterogeneity

Abstract

PurposeTo explore the role of intratumoral heterogeneity on MRI assessed by histogram analysis in differentiating soft-tissue sarcomas (STS) of different grades. Materials and MethodsPatients with primary STS undergoing MRI prior to iatrogenic procedures were included retrospectively. The histologic grade was assigned according to Federation Nationale des Centres de Lutte Contre le Cancer grading system. T1WI and T2WI were normalized by dividing mean signal intensity (SI) of contralateral/near unaffected muscles. Contrast-enhanced T1WI was normalized by computing enhancement ratio (ER) map as (SIpost-SIpre)/SIpre×100, where SIpre and SIpost represent SI of each pixel before and after enhancement. A region of interest (ROI) was manually drawn to include entire tumor area on axial slice with largest tumor diameter. Mean, mode, standard deviation, kurtosis and skewness on ROIs were extracted with ImageJ software. ANOVA/Kruskal-Wallis test was used to determine the significance of differences. ROC curve was applied for statistically significant parameters. P value ≤0.05 was considered statistically significant. ResultsAmong involved 67 patients, 8 were assigned to grade 1, 38 to grade 2 and 21 to grade 3. Skewness (P = 0.022) and kurtosis (P = 0.035) on ER maps were significantly different among STS of different grades. The optimal cutoffs of skewness and kurtosis on ER maps were -0.488 (AUC[95% CI] 0.747[0.557–0.937]; sensitivity/specificity, 62.5%/86.4%) and 0.762 (AUC[95% CI] 0.684[0.548–0.821]; sensitivity/specificity, 76.2%/56.5%), respectively. ConclusionIntratumoral heterogeneity on MRI quantitatively displayed by histogram parameters can differentiate STS of different grades. Skewness and kurtosis on ER maps show the capacity.