Bidirectional joint line convergence angle changes and their preoperative predictors 12 months after medial closing‐wedge distal femoral osteotomy for valgus knees: A single‐centre exploratory study
Yuki Kato, Ken Ichikawa, Hiroki Matsui, Shuzo Takazawa, Shunsuke Koga, Shin Yamada, Asami Kiuchi, Akira Ryu, Hiroshi OhuchiAbstract
Purpose
To characterize the signed 12‐month change in joint line convergence angle (JLCA) after medial closing‐wedge distal femoral osteotomy (MCWDFO) for valgus knees and assess associations with preoperative factors, particularly its preoperative value.
Methods
This retrospective single‐centre cohort included 62 knees treated with MCWDFO between October 2017 and December 2024 without additional osteotomy or ligament repair or reconstruction. The primary outcome was signed change (12‐month postoperative minus preoperative JLCA) on weight‐bearing full‐length radiographs. Univariable and multivariable linear regression analyses were performed. The primary model included preoperative JLCA, mechanical lateral distal femoral angle (mLDFA), mechanical medial proximal tibial angle (mMPTA), age, sex and body mass index (BMI). Cluster‐robust standard errors accounted for within‐patient dependence (eight bilateral patients). Internal performance was assessed using patient‐grouped 10‐fold cross‐validation (50 repeats), pooling out‐of‐fold predictions within each repeat.
Results
Preoperative JLCA was −1.13 ± 2.04° and the 12‐month value was −0.56 ± 1.61°, with postoperative values more concentrated near 0°. Mean signed change was 0.57 ± 1.90° (median, 0.3°; range, −4.6° to 6.9°). Preoperative JLCA showed the strongest association with signed change (coefficient, −0.56; 95% confidence interval, −0.78 to −0.34; p < 0.001). mMPTA was also independently associated, whereas mLDFA, age, sex and BMI were not. Primary and cross‐validated coefficients of determination were 0.602 and 0.474; cross‐validated root mean squared and mean absolute errors were 1.36° and 1.10°.
Conclusions
Signed change showed substantial between‐knee variability despite a small mean change. The association with preoperative JLCA should be interpreted cautiously because of mathematical coupling. The model showed moderate internal predictive performance and remains exploratory rather than a precise individual‐level prediction tool. External validation and formal radiographic reliability assessment are required before clinical implementation.
Level of Evidence
Level III, retrospective cohort study.