| Objective To identify the risk factors for aseptic loosening after tumor knee prosthesis replacement,establish and verify the corresponding nomogram model,and help clinicians and patients to more accurately predict the incidence of aseptic loosening after tumor knee prosthesis replacement.Methods The clinical data of two bone tumor treatment centers(Tianjin hospital and the 960 Hospital of the PLA)who underwent tumor knee prosthesis replacement from August 2001 to September 2016 were collected retrospectively.Based on the common risk factors of loosening reported in the previous literature and the main risk factors in this study,the predictors needed to build the model were determined The risk factors related to aseptic loosening were grouped and analyzed by single factor and multiple factor Cox regression analysis: tumor location,prosthesis stem length,resection length,prosthesis movement mode,gender,age,whether cortical bone graft,customized or matched type,prosthesis stem diameter,prosthesis material,tumor type,activity intensity,BMI.The independent risk factors of aseptic loosening that can be used in the nomogram model are screened out,and the corresponding weighted scores are assigned to each prediction factor to build the nomogram model.According to the model,the 5-year and 10-year survival rates and median survival times of the patients can be calculated.C index is used to verify the differentiation of the model,and graph calibration method is used to verify the accuracy of the nomogram model to test the prediction reliability of the model.Results The mean time of aseptic loosening was 91 months(15-166 months).The survival rate of prosthesis without aseptic loosening was 92.5% in 5 years,85.0% in 10 years and 45.5% in 15 years.The incidence of aseptic loosening was 16.0%(19/119)in distal femoral prosthesis and 8.6%(5/58)in proximal tibial prosthesis.The incidence of aseptic loosening was 8.2%(13/158)and 57.9%(11/19),respectively.Single factor Cox regression analysis showed that the influence of osteotomy length(P=0.046)and prosthesis movement mode(P=0.001)on aseptic loosening was statistically significant,and the length of prosthesis handle(P=0.059)and tumor site(P=0.150)had a certain value in predicting aseptic loosening.Multivariate Cox regression analysis showed that tumor site(HR=3.99,95% CI 1.21-13.16,P=0.023),prosthesis length(HR=2.84,95% CI 1.13-7.12,P=0.026)and prosthesis movement mode(HR=4.11,95% CI 1.74-9.70),P=0.001)is an independent risk factor of aseptic loosening.Based on these three independent risk factors,a nomogram model is built to assign a weighted score to each factor.The total score of each factor is the total score of the patient.The higher the score,the higher the risk of aseptic loosening.According to the total scores of each patient,the 5-year and 10-year survival rates and median survival times of the prosthesis without aseptic loosening can be calculated.The C index of the model is 0.74(CI 95% 0.65～0.84),which indicates that the nomogram model has an acceptable prediction differentiation.According to the curve,the predicted survival rate of each time point is consistent with the actual survival rate of prosthesis.Conclusions In our study,tumor location,prosthesis length and prosthesis movement mode are independent risk factors of aseptic loosening after tumor type knee prosthesis replacement.We should pay enough attention to these three factors in clinical work.In addition,the nomogram model established in this study has a good accuracy for predicting aseptic loosening after tumor type knee prosthesis replacement,and can provide reliable individual prosthesis survival probability prediction for patients receiving prosthesis treatment. |
