Effects of image-based personalization of knee varus deformity on medial and lateral knee contact forces during daily activities: a musculoskeletal modeling analysis on patients before and after high tibial osteotomy

High tibial osteotomy is a joint-preserving surgical technique for realigning the lower limb in patients with knee osteoarthritis and varus deformity, aiming to reduce medial knee loading and slow osteoarthritis progression. Knee contact force predictions via musculoskeletal models are affected by tibiofemoral alignment and knee contact locations. However, previous models assumed the entire varus deformity within the knee joint, neglecting bone deformities in the tibia and femur. This thesis aimed to (i) assess the accuracy of modeling tibiofemoral malalignment as a whole joint varus deformity, and (ii) analyze the effects of real high tibial osteotomy, which corrects tibial varus only, compared to a hypothetical correction involving both tibia and femur, on medial and lateral knee contact forces. The study included 25 patients with medial knee osteoarthritis and varus deformity, as part of a larger project at the Rizzoli Orthopedic Institute. Available data included full weight-bearing and Rosenberg 45° knee flexion radiographs, as well as motion capture data including 3D marker trajectories and ground reaction forces during walking, stair ascent and stair descent. A full-body musculoskeletal model was personalized for each patient and used in an inverse-dynamics optimization-based workflow in OpenSim to calculate knee contact forces during the three motor activities. To address the first aim, pre-surgical models were personalized image-based knee contact points, along with three levels of varus deformity, including tibial, joint and femoral deformities. Differences in knee contact forces during stance were evaluated using statistical parametric mapping with a non-parametric paired t-test. For the second aim, a subset of 7 post-surgical patients was analyzed, by comparing real correction on the tibia to a hypothetical correction involving both tibia and femur to achieve a neutral tibiofemoral alignment. Results showed that modeling knee varus malalignment as a whole joint deformity, without considering bone deformities, was generally accurate for predicting medial and lateral knee contact forces, with no significant differences across models during daily activities and maximum difference of 0.16 body-weight, apart from minor exceptions when the loads were low. High tibial osteotomy correcting both tibia and femur significantly differed from tibial correction alone only in extreme cases where high residual malalignment in the joint led to greater medial unloading and lateral loading shifting, with a maximum difference up to 1.72 body-weight for the lateral force. In conclusion, musculoskeletal models can reliably represent varus malalignment as a joint deformity for predicting knee contact force distribution. Single or double high tibial osteotomy procedures impact medial-lateral load distribution in specific cases. Future research should explore a larger sample and sensitivity analyses on bone deformity effects.