Prediction model for abnormal joint line obliquity after open wedge high tibial osteotomy using knee radiographic
Kritsada Sukha1, Witoon Thremthakanpon1,2, Burin Sutthapakti1
1Department of Orthopedics, Buddhachinaraj Hospital, Phitsanulok, Thailand.
Background:
The purpose of this study was to identify key preoperative radiographic parameters and develop a predictive model for abnormal knee joint line obliquity (KJLO) following medial open-wedge high tibial osteotomy (MOWHTO) in patients with medial compartment knee osteoarthritis.
Methods:
This retrospective observational cohort study included 81 patients (100 knees) treated with MOWHTO. All patients had symptomatic medial knee osteoarthritis with varus alignment and underwent preoperative and postoperative long-standing radiographs under a standardized protocol. Radiographic parameters including preoperative hip-knee angle (HKA), mechanical axis deviation, mechanical lateral distal femoral angle (mLDFA), medial proximal tibial angle (MPTA), joint line congruence angle (JLCA), and KJLO were measured. Multivariable logistic regression analysis was used to identify strong predictors of abnormal postoperative KJLO, defined as an angle > 4°. The final model was selected on the basis of backward elimination method. Model's discrimination performance was analyzed by using sensitivity, specificity, and area under the receiver operating characteristic curve (AuROC). Model stability was demonstrated by calibration plot and Hosmer-Lemeshow goodness of fit. Bootstrapping internal validation was performed to estimate model accuracy.
Results:
Of 81 patients (100 knees), there were 65 female and 16 male patients with an average age of 50.1 years and mean body mass index (BMI) of 27.8 kg/m2; 61 knees were found post-KJLO > 4°. According to the performance of logistic regression, sensitivity, specificity, and ROC were calculated to select the best formula for postoperative KJLO prediction. The final predictive model included preoperative HKA, MPTA, and KJLO, and the planned correction angle. The model demonstrated strong performance with an AuROC of 0.876 (95% CI 0.808-0.945). Calibration plot illustrated calibration-in-the-large (CITL) of 0.00, E:O of 1.00, slope of 1.00. Following internal validation, the model remained robust with AuROC of 0.875. The model demonstrated a sensitivity of 80.30%, specificity of 87.20%, positive predictive value of 90.70%, and negative predictive value of 73.90%.
Conclusions:
This predictive model incorporating preoperative HKA, MPTA, and KJLO, and the planned correction angle, demonstrates high internal validity and serves as a robust tool for individualized surgical planning in MOWHTO. By identifying high-risk patients preoperatively, surgeons can proactively tailor treatment strategies, such as modifying the target correction angle or opting for other surgical options.
Trial Registration:
TCTR20251003003.