MR imaging differentiation of malignant soft tissue tumors from peripheral schwannomas with large size and heterogeneous signal intensity.

OBJECTIVE

To determine the value of MR imaging features at and near the margin of the tumor in differentiating peripheral schwannomas with large size (maximum diameter >5 cm) and heterogeneous appearance from malignant soft tissue tumors (MSTTs).

METHODS

We retrospectively reviewed MR images of 25 MSTTs and 15 peripheral schwannomas with heterogeneous appearance on MR imaging and maximum diameter ranged from 5 cm to 11 cm. The soft tissue masses were evaluated for split fat sign, bright rim sign, lobular shape (with two or more deep lobulations), peritumoral edema, and capsule. The Fisher's exact test was used to determine whether these imaging features differed significantly between schwannomas and MSTTs. A nonparametric Mann-Whitney test was used to compare the maximal extent of peritumoral edema of schwannomas and MSTTs. The optimal cutoff value of the maximal extent of peritumoral edema was calculated by means of receiver operating characteristic analysis for distinguishing between schwannomas and MSTTs. Interobserver agreement in the assessment of imaging features was evaluated using Cohen κ statistic and percentage agreement.

RESULTS

The split fat sign, bright rim sign were more common in schwannomas than in MSTTs (P<0.001). The incidence of lobular shape was higher for MSTTs than for schwannomas (P<0.001). The peritumoral edema was present more frequently in MSTTs than in schwannomas (P=0.001). The median (interquartile range) of the maximal peritumoral edema extent was 13 mm (9.5-17 mm) for schwannomas and 46 mm (24-75.5 mm) for MSTTs, respectively (P<0.001). The optimal cutoff value of the maximal peritumoral edema extent calculated to distinguish MSTTs from schwannomas was 18 mm, which had 100% sensitivity, 89% specificity and Youden index 0.89. The frequency of the presence of capsule was not statistically significant between schwannomas and MSTTs (P=0.107). Interobserver agreement as determined by Cohen κ statistic for the imaging features evaluated in the study was substantial to almost perfect (κ=0.694-0.942), with percentage agreements ranging 85-97.5%.

CONCLUSIONS

Large schwannomas with heterogeneous appearance on MR imaging can be distinguished from MSTTs by analyzing the imaging features at and near the margin of the mass. The presence of split fat sign and bright rim sign, and the absence of lobular shape and extensive edema (edema with maximal extent larger than 18 mm) favor the diagnosis of schwannoma, especially when two or more of this imaging features coexist.