(P 05) COEFFICIENT OF VARIATION OF T2-WEIGHTED MRI PREDICTS THE PROGNOSIS OF MALIGNANT PERIPHERAL NERVE SHEATH TUMOR

Objective: Malignant peripheral nerve sheath tumors (MPNST) are rare and account for 2% of soft tissue sarcomas. To ameliorate clinical decision-making, it is important to establish several independent clinical factors associated with the survival of patients with MPNST. Therefore, we aimed to find an imaging-based prognostic factor of MPNST. An imaging-based predictor for the prognosis of MPNST has not been established. This study aimed to investigate whether non-enhancement MRI features, including measurement of the heterogeneity of the tumor with MR T2 imaging by calculating CV values, are associated with the prognosis of non-metastatic MPNST.

Methods: This retrospective study was conducted at four institutions. The inclusion criteria were as follows: histologically non-metastatic MPNST who underwent surgical resection. MRI was performed preoperatively. Non-enhancement MR images were evaluated for signal intensity heterogeneity at T1/2-weighted imaging, tumor margin definition at T1/2-weighted imaging, peritumoral edema at T2-weight imaging, and CV. Two observers were blinded to the clinical information and analyzed the MR images. The following were recorded according to the previous reports: (a) tissue layer (deep or superficial); (b) longest diameter (in millimeters); (c) SI heterogeneity at T1/2-weighted imaging categorized as homogeneous, < 50% heterogeneous, or ≥ 50% heterogeneous (when 0%, < 50%, or ≥ 50% of tumor volume showed areas of low, intermediate, and high SI, respectively); (d) tumor margin definition (poorly defined [> 75% of margin not clear], mixed margin [10%- 25% of margin not clear], or well defined [> 90% of margin clear]); (e) peritumoral edema at T2-weighted imaging, defined as high fluidlike signal at T2-weighted or short tau inversion-recovery sequences imaging with infiltrative and feathery borders that are distinguishable from the apparent tumor boarders (defined by a pseudo capsule or non-equivocal changes in SI of other sequence) and without mass effects (defined as modification of the usual morphology of the local anatomy categorized as no, limited, or extensive when involving the entire length and breadth of the muscle).The tumoral region of interest (ROI) was measured, which showed the maximum tumor diameter. The ROI in the tumor was hand-drawn and set with its surrounding tumors (Fig. 1). We measured the signal intensities of MR T2-weighted images and calculated the corresponding CV values using Philips DICOM Viewer R3.0-SP15. CV is defined as the ratio of the standard deviation to the mean. The associations between factors and overall survival (OS) were investigated via Kaplan–Meier method with log-rank tests and the Cox proportional hazards model.

Results: From April 2002 to August 2022, 77 patients with MPNST were identified. We then excluded patients with metastatic lesions and unplanned excision at initial presentation and 13 patients without surgical treatment and MRI. Finally, 42 patients were included (Fig. 2). Patients comprised 20 men and 22 women. The mean patient age was 50 years (range, 8–93 years), and the mean patient follow-up period was 54.7 months (range, 2–217 months). The median tumor diameter was 82.5 mm (range, 26–250 mm). Demographic and clinicopathologic characteristics of the 42 patients are shown in Table 1. The 5- year OS rate for the entire patient cohort was 55.9%. The mean CV value of MR T2 images was 0.2299± 0.1339 (SD) (range, 0.0381–0.8053). Applying receiver operating characteristics analysis, the optimal cut-off level for CV value was 0.137. This cut-off CV value was used for its stratification into high and low CV values. The results of univariate survival analyses of prognostic factors for OS are listed in Table 2. At multivariate survival analysis, a high CV value (hazard ratio = 3.63; 95% confidence interval = 1.16–16.0; p = 0.047) was identified as an independent predictor of OS. Interobserver and intraobserver reproducibility for the calculation of CV value was 0.99 (95% CI = 0.97-1.00) and 0.99 (95% CI = 0.99-1.00), respectively.

Conclusion: We focused on the heterogeneity of MPNST to evaluate an imaging-based prognostic factor. CV was used to reflect tumor heterogeneity because MRI protocols were diverse. The prognosis of MPNST could be predicted based on CV values, which do not require an additional imaging modality and are quite reproducible.