晚期膝骨关节炎患者全膝关节置换术后股骨假体矢状位角度误差的影响因素分析OA

Objective To explore the factors influencing sagittal angle error of the femoral component following total knee arthroplasty(TKA)in patients with advanced knee osteoarthritis.Methods This retrospective study enrolled 120 patients with advanced knee osteoarthritis who underwent TKA in the Second Affiliated Hospital of Inner Mongolia Medical University from June 2021 to June 2024.Using a 4:1 ratio,patients were divided into a training set(n=96)and a test set(n=24)via stratified random sampling.Based on the sagittal femoral prosthesis flexion angle(FPFA)on the full-length radiograph at 6 months postoperatively,patients in training set were categorized into well-aligned group(n=42)and malaligned group(n=54).Clinical data,perioperative indicators and imaging parameters at different postoperative time points were compared between the two groups.Repeated measures analysis of variance was used to evaluate the trend of postoperative femoral radiographic parameters at different time points,and to analyze the time-group interaction effect.Six machine learning algorithms,including logistic regression,K-nearest neighbor,support vector machine,naive Bayes,multilayer perceptron,and extreme gradient boosting(XGBoost),were employed in training set to construct prediction models for sagittal angle error of femoral prosthesis after TKA.The performance of 6 machine learning models was evaluated and compared.Summary plots and feature dependence plots were drawn using R software to interpret the constructed machine learning models.Multiple linear regression analysis was used to analyze the relationship between FPFA and various influencing factors.Results Compared with well-aligned group,malaligned group had higher visual analogue scale score,whole blood high-shear viscosity,whole blood low-shear viscosity,plasma viscosity,red blood cell(RBC)aggregation index,fibrinogen and D-dimer levels,as well as a longer hospital stay.Conversely,the Hospital for Special Surgery score,range of motion,and RBC deformation index were significantly lower,while prothrombin time(PT)and activated partial thromboplastin time(APTT)were shorter(P<0.05).Compared with the values at 1 month postoperatively,the anterior femoral bowing angle,femoral lateral bending angle,and lateral angle of the femoral mechanical axis were increased in malaligned group at 3 and 6 months postoperatively(P<0.05);compared with 3 months postoperatively,these three angles were further increased in the malaligned group at 6 months postoperatively(P<0.05).Compared with well-aligned group,malaligned group showed a larger anterior femoral bowing angle at both 3 and 6 months postoperatively,and larger femoral lateral bending angle and lateral femoral mechanical axis angle at 6 months postoperatively(P<0.05).Recursive feature elimination with 5-fold cross-validation identified 7 optimal candidate variables for risk factors,namely ROM,RBC aggregation index,PT,APTT,anterior femoral bowing angle,femoral lateral bending angle,and lateral femoral mechanical axis angle.Finally,the XGBoost model was determined as the best machine learning model for predicting postoperative sagittal angle error of the femoral prosthesis,with a sensitivity of 0.845(95%CI 0.789-0.892),specificity of 0.801(95%CI 0.754-0.863),accuracy of 0.814(95%CI 0.762-0.885),and ROC-AUC of 0.812(95%CI 0.765-0.864).The SHAP summary plot showed that the candidate variables with significant effects on malalignment were,in order,anterior femoral bowing angle,ROM,lateral femoral mechanical axis angle,RBC aggregation index,PT,APTT,and femoral lateral bending angle.Analysis of the feature dependence plots of the top 6 SHAP values revealed that the risk of malalignment gradually increased with the increase of anterior femoral bowing angle,lateral femoral mechanical axis angle,and RBC aggregation index,the decrease of ROM,and the shortening of PT and APTT.The results of multiple linear regression analysis showed that anterior femoral bowing angle,lateral femoral mechanical axis angle,RBC aggregation index,and femoral lateral curve angle were independent risk factors for changes in sagittal FPFA on the full-length radiograph at 6 months postoperatively,while ROM,PT,and APTT were independent protective factors(P<0.05).Conclusion Anterior femoral bowing angle,lateral femoral mechanical axis angle,RBC aggregation index,femoral bending angle,ROM,PT,and APTT are influencing factors for sagittal angle error of the femoral prosthesis after TKA in patients with advanced knee osteoarthritis.