Mechanical Failure Risk And Prognosis Index After Extensive Curettages Of Giant Cell Tumor Of Bone Around Knee

[Objectives]The aims of our study are(1)to explore the risk factors of mechanical failure(MF);(2)to figure out an index to evaluate this risk;(3)to select an optimal reconstruction strategy to reduce this risk.[Methods]We retrospectively reviewed 104 patients from Dec.2008 to Mar.2016,undergone extensive knee curettages in our institution.Radiographs and postoperative interviews were used to classified cases of MF.Relative factors(age,tumor location,the invaded area,etc.)were also collected and analyzed by SPSS software.Patients with complete preoperative CT image data were selected.The region of interest(ROI)was made via ITK-SNAP.According to the size of image grey value,the tumor area was classified into a range from 0 to 64.We adopt the method of texture analysis(GLCM gray level co-occurrence matrix,GLRLM gray-level run-length matrix and gray matrix GLSZM area size,neighborhood grayscale difference matrix NGTDM)to measure the tumor area.Wavelets analysis was used to deal with the original image respectively.And then we quantized the original image in the X,Y,Z three directions on wavelet change.The high wave and low wave signed H,L respectively.Therefore,one wave could be transfer into LLL,LLH,LHH,HHH,HHL,HLH,LHL,HLL eight aspects.[Results]Thick subchondral bony layer(p=0.006)and combined grafting of cement and bone(p=0.006)had lower risk of mechanical failure.Mechanical failure appeared to happen in femur(p=0.012)more easily.The ROC curve(AUC=0.722)reveals that less postoperative bony layer(? 3.3mm)is more likely to cause mechanical failure.The Kaplan-Meier survival curve showing increased survival in those patients after a combination grafting surgery(HR,3.799;p = 0.006).19 cases of mechanical reconstruction failure happened in the whole 46 cases of preoperative CT data.451 radiomics features were extracted from the 3D ROI of the patients.Mann-Whitney U test was used to evaluate these features.13 valuable features were found.And 4 features("LLL_GLRLM_HGRE","HLH_Mean","HLH_GLCM_cshad","HHL_GLRLM_LGRE")were selected by the LASSO method.Then 100 times of 4-fold cross validation was used for test the performance of this method.The overall performance of the combination of these four features was shown in ROC curve form(AUC=0.891,95%CI 0.764-0.963,the Youden index was 0.6784).[Conclusion]Based on our study results,combined grafting of cement and bone reduced the risk of mechanical failure in knee due to thin subchondral bone layer(SCB),especially in femur.Preoperative CT scan and SCB thickness can be used to predict this risk.