Prediction of local recurrence in tenosynovial giant cell tumor of the knee: Based on preoperative MRI evaluation into disease subtypes and severity

Abstract

Objective Tenosynovial giant cell tumors (TSGCTs) of the knee differ in their clinical outcome according to disease subtypes and severity. The aim of this study was to determine the predictive MRI features related to local recurrence in TSGCT of the knee regarding disease subtypes and severity. Methods This retrospective study included 20 patients with pathology-proven TSGCT of the knee who underwent preoperative MRI and surgery from Jan. 2007 to Jan. 2022. The anatomical point of the lesion was determined with a knee mapping. And then MRI features related to disease subtype including nodularity (single vs. multinodular); margin (circumscribed vs. infiltrative); peripheral hypointenseity (present vs. absent); internal hypointensity reflecting hemosiderin deposition (speckled vs. granular) were assessed. Third, MRI features related to disease severity including involvement of bone, cartilage, and tendon were evaluated. MRI features for predicting local recurrence of TSGCT were tested using chi-square test and logistic regression analysis. Results Ten patients with diffuse-type TSGCT (D-TSGCT) and 10 patients with localized-type TSGCT (L-TSGCT) were included. There were six cases of local recurrence and all of them were D-TSGCT and none for L-TSGCT with statistical difference (P = 0.015). D-TSGCT that was direct risk factor for local recurrence showed more multinodular (80.0% vs. 10.0%; P = 0.007), infiltrative margin (90.0% vs. 10.0%; P = 0.002), and absent peripheral hypointensity (100.0% vs. 20.0%; P = 0.001) than L-TSGCT. Multivariate analysis showed infiltrative margin (odds ratio [OR], 81.0; P = 0.003) was independent MRI factor for D-TSGCT. Disease severity for risk of local recurrence included cartilage (66.7% vs. 7.1%; P = 0.024) and tendon (100.0% vs. 28.6%; P = 0.015) involvement compared to no local recurrence. Multivariate analysis showed tendon involvement (OR, 12.5; P = 0.042) was predictive MRI parameter for local recurrence. By combining tumor margin and tendon involvement, local recurrence was predicted sensitively on preoperative MRI (sensitivity, 100%; specificity, 50%; accuracy, 65%). Conclusion D-TSGCTs was associated with local recurrence and showed multinodularity infiltrative margin, and absent peripheral hypointensity. Disease severity including cartilage and tendon involvement was associated with local recurrence. Preoperative MRI evaluation by combining disease subtypes and severity can predict local recurrence sensitively.