IN SILICO-INZICHTEN IN RISICOFACTOREN VAN OVERBELASTING BIJ KNIEARTROSE: IMPLICATIES VOOR REVALIDATIE

Knee osteoarthritis (KOA) is a multifactorial joint disease characterized by progressive cartilage degeneration, pain, and impaired mobility. Mechanical overloading of the knee joint—particularly in the medial compartment—is widely recognized as a key contributor to both the onset and progression of KOA. However, the precise interplay between anatomical variation, joint stability, muscle coordination, and movement strategies in determining joint loading remains insufficiently understood. Furthermore, while gait retraining is a promising non-invasive therapy for reducing joint load, its effectiveness is often limited because generic guidelines overlook patient-specific factors such as joint geometry, ligament integrity, or neuromuscular control. This PhD project aimed to clarify how these factors influence medial compartment loading in KOA, providing the basis for in silico-informed, personalized non-invasive interventions, with a specific focus on gait retraining. This thesis was structured around five key research objectives, each addressed in dedicated studies using experimental data alongside musculoskeletal modeling and simulation approaches. Objective 1 examined which anatomical features predispose individuals with KOA to medial compartment overloading. Using a physics-based musculoskeletal modeling framework enriched with SSM-derived geometries, we systematically assessed the effect of population-based anatomical variations on joint loading. Contrary to our hypothesis, alignment variations in the sagittal and transverse planes (e.g., posterior tibial translation and external tibial rotation) had a greater effect on medial loading than frontal plane variations. Additionally, smaller femoral condyles and narrowed joint spaces were associated with increased contact pressures, even in the absence of altered gait patterns. These findings challenge the conventional focus on frontal plane alignment (e.g., varus) and highlight the importance of patient-specific joint geometry for accurate internal joint load estimation. Objective 2 investigated how ligament laxity—a common characteristic of KOA—affects medial compartment loading under overloading conditions, integrating both anatomical features and gait patterns linked to increased loading. Using a probabilistic modeling approach, we simulated increased laxity by modifying ligament stiffness and slack length parameters. Results confirmed that ligament integrity plays a crucial role in load distribution, as laxity led to excessive non-sagittal joint motion and a consistent anteromedial shift in the center-of-pressure during gait. This shift increased loading on thinner cartilage regions, thereby elevating degeneration risk. Interestingly, compensatory muscle activation did not adequately counteract passive instability, indicating that ligament-driven mechanical changes can override the intended benefits of gait retraining. These results underscore the need for interventions that also address passive joint stabilization, in addition to altering movement patterns. Objective 3 assessed whether patients with early-stage KOA exhibit neuromuscular adaptations that influence joint loading. Using muscle synergy analysis and co-contraction metrics during walking, we found that overall muscle coordination patterns were similar to those of healthy controls. However, a subset of KOA patients showed altered hamstring-quadriceps co-contraction, potentially as a compensatory response to perceived instability, despite minimal reported pain. While these neuromuscular changes did not significantly affect medial compartment contact pressures or gait kinematics after accounting for gait speed, the observed individual variability suggests that early, subtle neuromuscular changes may be present in some patients. Such changes could offer a therapeutic window before structural joint damage progresses. Objective 4 evaluated whether predictive gait simulations can inform more effective gait retraining strategies than population-derived synthetic gait patterns. Using a predictive simulation framework that minimized knee joint loading, we generated subject-specific gait patterns and compared them to gait variations extracted via principal component analysis (PCA) from an existing KOA dataset. Both approaches successfully reduced medial compartment contact force, but each had different strengths: predictive simulations were more effective during the second peak of the gait cycle, while PCA-derived patterns performed better during the first peak and aligned more closely with natural gait behaviors. These findings support a hybrid clinical model beginning with generalized PCA-based gait strategies, then refining them with personalized simulations informed by patient-specific joint anatomy and mechanics. Objective 5 was a proof-of-concept study testing whether a synthetic gait pattern optimized to reduce knee loading could be taught, learned, and retained by KOA patients. After a brief retraining session, participants successfully adopted a new walking pattern and retained it for at least one week. With additional practice, this pattern could more closely approximate the desired target gait pattern. Importantly, musculoskeletal modeling confirmed that the learned gait pattern reduced medial compartment loading, although effects were highly subject-specific, with some participants showing increases. This study demonstrates the feasibility of translating in-silico optimized gait modifications into clinical practice and highlights the potential of personalized, data-driven rehabilitation strategies. Collectively, these studies reveal that medial knee joint loading in KOA results from a complex interplay between anatomy, ligament integrity, neuromuscular control, and movement strategy. The findings consistently emphasize the need for patient-specific assessments and interventions. This thesis establishes a robust computational foundation for personalized rehabilitation in KOA, showing that in-silico methods can identify high-risk anatomical features, estimate personalized loading outcomes, and optimize gait patterns for maximal therapeutic benefit. Ultimately, this work supports a paradigm shift toward precision biomechanics in KOA management, where patient-specific digital twins can guide real-time, adaptive treatment plans to reduce joint loading.