Aberrant mechanical loading in knee osteoarthritis

Osteoarthritis (OA) is the most common form of arthritis and one of the most prevalent chronic joint disease that especially affects the knee joint. Roughly, knee OA affects 12% of people aged 60 years or older. Although a well-documented joint disorder, the cause of OA is not fully understood, and despite medical advances, there is no effective treatment for OA. Excessive knee loading and/or changes in the normal load bearing contact can produce joint damage and local degenerative alterations to the articular cartilage and, ultimately, contribute to the knee OA development. Knee adduction moment (KAM) is the most direct ‘mechanical’ parameter reported in medial knee OA because it is thought to represent medial knee loading and increased KAM has been associated with more pronounced medial knee OA. However, joint kinematics and moments only represent external loading and do not take into account the internal forces. Knee contact force (KCF), on the other hand, not only account for the external forces but also account for muscle and ligament forces. KCFs have been measured in vivo in individuals receiving instrumented total knee arthroplasty (TKA). However, although these approaches have provided valuable insight into knee loading, knee biomechanics is altered after TKA and it is, therefore, hard to infer the knee loading in individuals with knee OA from these measurements. Alternatively to joint moments and KCF measured from instrumented TKA, musculoskeletal modeling in combination with dynamic simulations of motions has shown to be useful in calculating KCF. Therefore, this PhD aims to evaluate knee joint loading assessed by calculating KCF using a musculoskeletal modeling workflow during common daily-living activities as walking and more demanding tasks, in patients with medial knee OA.

In study I, KCF were calculated and the relations of KCF with external knee adduction moments and/or flexion moments (KFM) were assessed during the stance phase of gait. Knee loading was evaluated in individuals with early medial knee OA, classified based on early joint degeneration on Magnetic Resonance Imaging (MRI) and compared to individuals with established medial knee OA and healthy subjects. Three-dimensional motion and ground reaction force data in subjects with medial early and established knee OA as well as in healthy subjects were used as input for a musculoskeletal model to calculate KAM and KFM, and KCF. Our results showed no significant differences between controls and subjects with early OA. In early OA patients, KAM significantly explained 69% of the variance associated with the first peak KCF but only KFM explained the second peak KCF. Multiple regression, combining KAM and KFM, leads to a better prediction of peak KCF than KAM or KFM alone. However, only 20% of the variance of the second peak KCF was explained by both moments in established OA. In conclusion, KCF are not increased in patients with early OA, suggesting that knee joint overload is a consequence of further joint degeneration in more advanced stages of OA. Additionally, our results clearly show that KAM is not sufficient to predict joint loading at the end of the stance, where KFM contributes substantially to loading, especially in early OA.

The effect of using an anatomical versus a functional axis of rotation (FAR) on the computation of KAM in healthy subjects and patients with knee OA was investigated in study II. In addition, this study reports the influence of calculating FAR using weight-bearing versus non-weight-bearing motion on KAM. Three musculoskeletal models were created with a different knee axis of rotation (AoR): transepicondylar axis (TEA); FAR calculated based on SARA algorithm using a weight-bearing motion (wFAR) and a non-weight-bearing motion (nwFAR). KAM was calculated during gait in subjects with early and established medial knee OA and compared to corresponding KAM of healthy subjects. Our results showed significant differences between the three groups in the first peak KAM when TEA was used (p = 0.038). However, these differences were no longer present when using a FAR. In subjects with established OA, KAM was significantly lower when using nwFAR compared to TEA and wFAR models. Therefore, we can conclude that the presence of excessive KAM in subjects with established knee OA depends on the definition of the AoR: anatomical versus functional. Therefore, attention should be paid to the AoR definition when comparing KAM in different studies on knee OA. In patients with end-stage knee OA where increased passive knee laxity is likely to exist, the use of weight-bearing motions should be considered to better reflect load-dependent knee kinematics.

Study III evaluates knee joint loading during gait and step-up-and-over tasks in subjects with early knee OA and established knee OA compared to healthy subjects. Subjects with varying degrees of medial compartment knee OA severity (early OA and established medial knee OA), and healthy controls performed gait and step-up-and-over tasks. Knee joint moments, the magnitude of KCF, contact pressures and center of pressures (CoP) on the medial compartment were analyzed for the three groups for both activities using a multi-body knee model with articular cartilage contact, 14 ligaments, and 6-DoF tibiofemoral and patellofemoral joints. Our findings showed that during gait the first peak of the medial KCF was significantly higher in patients with early (p = 0.048) and established knee OA (p = 0.001) compared to control subjects. Furthermore, the magnitude of the medial contact pressures and the CoP location were significantly different in both groups of patients compared to controls. External knee rotation moments (KRM) and external rotation angles were significantly higher during early stance in both patient groups (p < 0.001) compared to controls. During step-up-and-over, there was a high variability between the participants and no significant differences in KCF were observed between the groups. In conclusion, knee joint loading and kinematics were altered in patients with early and established knee OA, but only during gait. This is indicative that an excessive medial KCF and altered loading location, observed already in early OA patients, should be considered as risk factor for knee OA development.

Study IV assessed KCF and contact pressures during different stair negotiation strategies in individuals with medial knee OA. Motion analysis was performed in individuals with medial knee OA and healthy subjects while ascending and descending a staircase consisting of seven 17.2cm-height instrumented steps using different strategies: step-over-step (SOS) at controlled speed, and also SOS at self-selected speed and step-by-step (SBS). Trunk angles, knee joint moments, KCF, contact pressures and CoP on the medial compartment were calculated using a multi-body knee model. Our findings showed that at controlled speed patients significantly decreased medial and lateral peak KCF compared to controls during stair ascent but not during stair descent. During both stair ascent and descent, patients showed higher trunk flexion angles and lean the trunk more towards the leading leg than controls. In the self-selected condition, patients reduced gait speed and significantly reduced medial KCF during stair descent. By performing SBS instead of SOS, patients significantly increased the peak medial KCF during stair ascent and decreased contact pressures on the medial compartment during stair descent. Therefore, we can conclude that patients with medial OA use increased trunk flexion and lean to reduce knee joint loading during stair ascent. These changes were less effective during stair decent where reducing speed is more effective. Individuals should be recommended to use SOS during stair ascent and SBS during stair descent to reduce medial knee loading.

These studies showed, firstly, the importance of calculating the KCF in both medial and lateral knee compartments to better assess loading changes in individuals with varying levels of medial knee OA severities, especially those with early knee OA, during gait. The medial KCF provided a more sensitive metric to knee joint loading than external KAM or total KCF. Secondly, KAM was shown to be sensitive to the knee axis of rotation, indicating that differences between subject groups might be heavily dependent on the knee axis definition. Finally, different mechanisms used by patients with knee OA were identified during gait versus stair negotiation. Patients with knee OA use compensatory mechanisms during stair negotiation but not during gait.

Overall, this PhD contributed to a better description of knee joint loading in individuals with medial knee OA during gait, particularly those presenting early medial knee OA. The importance of assessing the KCF instead of only external moments in patients with knee OA, and compartmental KCF instead of only overall KCF in patients with early knee OA during gait were demonstrated. Furthermore, compensatory mechanisms used by patients with knee OA during highly demanding daily-living tasks were described. Finally, activities that might cause increased knee loading and those that cause decreased knee loading in individuals with medial knee OA were identified.