Comparison and evaluation of risk factors for meningeal, pleural, and extrapleural solitary fibrous tumors: A clinicopathological study of 92 cases confirmed by STAT6 immunohistochemical staining.

Solitary fibrous tumors (SFTs) are an uncommon type of mesenchymal tumors that are presumably fibroblastic in nature. SFTs are translocation-associated neoplasms that can be consistently diagnosed through the evaluation of NAB2/STAT6 gene fusion. Currently, SFTs have a different grading system and criteria according to their primary sites, and the differences and similarities in SFTs according to their primary sites are still poorly understood. Therefore, we compared SFTs according to their primary sites and histologic appearance, and validated the current grading system of SFTs. A total of 92 patients (with 15 meningeal, 40 pleural, and 37 extrapleural SFTs) were evaluated. The patients with pleural SFTs (mean age: 60.2 years) showed a significantly increased age at diagnosis. Tumors with hemangiopericytoma-predominant morphology had significantly higher grades in the evaluation of several risk factors such as cellularity (P<0.001), pleomorphism (P=0.001), and mitotic activity (P<0.001). Consequently, hemangiopericytoma (HPC)-predominant tumors had a significantly higher recurrence rate. The meningeal SFT group had significantly higher proportion of the HPC-predominant histologic phenotype compared with pleural or extrapleural SFTs (66.67% vs. 5.00% or 18.92%, respectively; P<0.001). Consequently, meningeal SFTs showed significantly higher recurrence rates compared with pleural or extrapleural SFTs (33.33% vs. 12.50% or 2.70%, respectively; P=0.009). Regarding the evaluation of risk factors, a tumor size ≥10cm (P=0.017), a mitotic index ≥4/10 high power fields (HPFs) (P=0.001), high tumor cellularity (P=0.003), high nuclear pleomorphism (P=0.005), and tumor necrosis (P=0.004) were associated with both recurrence and disease-specific mortality. Upon evaluation of the usefulness of the criteria using previously described factors, the predictive model was on the borderline of validation. Of the five factors indicated in the log rank test, only a mitotic index ≥4/10 HPFs remained a significant factor in the multivariate Cox model.