Author:

Abstract:

Osteoarthritis (OA) is one of the most common hip joint diseases and with increasing age and obesity, the prevalence of OA is likely to become even higher. Many patients with end-stage hip OA are treated with total hip arthroplasty (THA). As the number of THA surgeries is increasing, also the number of revision surgeries is growing. To prevent bone-implant interface failure, it is important to detect and monitor patients at risk, in which the loading on the hip joint is very important. Besides that, mechanical factors, e.g. joint loading, are a risk factor for OA. However, most often hip pathology patients are evaluated using clinical measures, like abductor strength and dislocation rate, or in terms of kinematics and joint moments that only represent external loading and do not take into account any internal forces, like muscle forces. Hip contact forces are typically used to quantify internal hip joint loading. This measure accounts for the joint loading determined by the external forces, e.g. ground reaction forces, as well as internal forces, e.g. muscle forces. Hip contact forces can be measured using instrumented hip prostheses. A dataset containing measured hip contact forces of different subjects has previously been made publicly available. Although these instrumented hip prostheses provide a direct measure of hip contact forces, these measurements are rare and limited to subjects that received a THA. To be able to determine hip loading non-invasively in other subject groups, musculoskeletal modelling in combination with three dimensional motion analysis data, can be used to calculate hip contact forces in vivo. These model-based contact forces have been validated against measured contact forces from instrumented prostheses. Several studies show an overestimation of calculated hip contact forces compared to measured forces, while others find more comparable results. Different modelling choices, like including subject-specific detail and the use of different optimization methods to calculate muscle forces can contribute to the reported overestimation of calculated contact forces compared to measured forces. Also differences in movement kinematics and kinetics, based on which hip contact forces are calculated, will affect contact forces. It is important to consider the contribution of all of these factors when assessing the validity of calculated hip contact forces against a measured dataset. The aim of this PhD aim is twofold. First, the sensitivity of the calculated hip contact forces for specific aspects of the musculoskeletal modelling and dynamic simulation workflow is evaluated. Second, once the factors that influence calculated hip contact forces most are identified, insights are used to investigate hip loading in hip pathology patients. This results in the following research topics: Evaluate the effect of including different levels of subject-specific detail in musculoskeletal models on hip joint loading (studies I, II and III). Evaluate the effect of the optimization technique on calculated muscle and contact forces (study IV). Evaluate the effect of different gait patterns on hip joint loading (studies V and VI). Evaluate hip function in OA and THA patients (studies VII and VIII). The first four studies investigate the sensitivity of the calculated hip contact forces for different aspects of the musculoskeletal modelling workflow. In study I, we investigated the relative importance of including subject-specific geometrical detail in the musculoskeletal model versus the effect of an altered cost function definition on hip contact forces for healthy control subjects. We used computer tomography (CT) and magnetic resonance imaging (MRI) data to create seven model types that each contained a different level of subject-specific detail and simulated gait. Hip contact forces were calculated on the one hand using the standard simulation workflow and analyses implemented in OpenSim. On the other hand, the effect of including a term minimizing the hip contact forces in the optimization criterion underlying muscle force calculation was investigated. Results showed that inclusion of subject-specific detail had a dominant effect on the calculated hip contact forces, although the effect of the level of subject-specific detail varied. The MRI-based model that included subject-specific muscle paths and wrapping surfaces, to account for the effect of the hip capsule, resulted in contact forces that were most comparable to measured hip contact forces, both in magnitude and orientation. To generalize the effect of the subject-specific wrapping surfaces, the generic model was updated to include average MRI-based wrapping surfaces. The use of this model also brought contact forces closer to experimental values, but the effect was less pronounced compared to the MRI-based models. Including minimization of the hip contact forces into the cost function (SOmin) had only a limited effect on the contact forces. Specifically, the largest overestimations were not affected. Nevertheless, for the generic model, contact forces decreased on average more by using SOmin than by including wrapping surfaces, but the largest overestimations of the contact forces were not decreased. Therefore, inclusion of subject-specific geometrical detail in the model had a greater effect than altering the cost function definition. The effect of including subject-specific detail in the musculoskeletal model on hip contact forces was also investigated in hip OA and THA patients in study II. Besides that, the effect of subject-specific moment generating capacity of the musculoskeletal model was investigated for healthy control subjects as well as OA and THA patients. For all subjects a generic scaled and MRI model was created as well as the respective models including wrapping surfaces and gait was simulated. Maximal isometric muscle forces in the model were on the one hand statically scaled based on dynamometer measurements. On the other hand, muscle forces were functionally scaled based on the external joint moments required during gait and stair ascent and descent, an approach already successfully adopted for the knee. Static scaling decreased the maximal isometric muscle forces of control, OA and THA subjects and for all model types. As a result, the model lacked the capacity to generate the internal joint moments measured during functional activities, hence further results, i.e. hip contact forces, cannot be reliably calculated. Functional scaling decreased muscle forces less or even increased muscle forces compared to the unscaled models. Hip contact forces were comparable to the unscaled models. For THA patients, including MRI-based information decreased hip contact forces, specifically by including wrapping surfaces around the hip joint, as was also reported for control subjects in study I. For OA patients, changes in hip contact forces were less pronounced. The deviations of muscle generated moments from internal joint moments of the MRI-based models were not excessively increased compared to the generic model with unscaled muscle forces. This indicates that including MRI-based geometrical detail, without scaling muscle forces, results in a model that is strong enough to perform the measured functional tasks while hip contact forces are more comparable to measured hip contact forces. The importance of estimating body segment parameters (BSP) when calculating joint moments and muscle forces during gait was examined in study III. The segment mass, centre of mass (com) and inertial tensor of the left thigh, shank and foot were adjusted from 60% to 140% of the nominal value in steps of 10% both individually as well as for different combinations of BSP values. We found only a limited effect of BSP perturbation on inverse dynamics. The largest effect was found by perturbing the shank com. Further, the additional influence of a combined perturbation of parameters was only very small. On the other hand, muscle forces calculated using computed muscle control (CMC) were largely affected by changes in BSP, which resulted from the underlying forward integration. Indeed, post hoc analyses indicated that when using static optimization, a technique that is not based on forward integration, muscle forces were less affected by perturbations in BSP. In study IV, we investigated the effect of different optimization techniques on calculated muscle forces and the magnitude and orientation of resultant hip contact forces for gait and sit to stand using a scaled generic musculoskeletal model. To calculate muscle forces, four different optimization techniques were used, i.e. two different static optimization techniques (SO1 and SO2), computed muscle control (CMC) and the physiological inverse approach (PIA), after which muscle and hip contact forces were calculated. Both static optimization techniques showed the lowest hip contact forces for both gait and sit to stand. The additional constraints to include a physiological increase and decrease of muscle activation in time and the inclusion of passive muscle forces (SO2) did not majorly affect the hip contact forces compared to a standard SO formulation (SO1). In contrast, hip contact forces increased drastically when using CMC. These increased contact forces from CMC were potentially caused by higher muscle forces resulting from co-contraction of agonists and antagonists around the hip or the slightly poorer tracking of the net joint moment by the muscle moments. On the other hand, hip contact forces between the SO techniques and PIA were similar, which showed that the activation and contraction dynamics can be integrated without inducing an excessive overestimation of the hip contact forces as observed by CMC. The first four studies showed that the inclusion of subject-specific geometrical detail in the musculoskeletal model is most important in calculating hip contact forces. The muscle optimization technique, including minimization of the hip contact force into the cost function and subject-specific maximal isometric muscle forces all affect hip contact forces to a lesser extent. In the following four studies we investigated hip loading in hip pathology patients. In study V, we evaluated to what extent hip and pelvis kinematics can affect hip contact forces and aimed to identify gait patterns that can be used to effectively influence hip contact forces. In patients with knee OA, the relation between joint loading and kinematics has already been established and is used to deliberately reduce knee contact forces. For the hip, we therefore investigated the relation between hip contact forces and joint moments to determine whether hip contact forces can be estimated using external joint moments, as this is a more widely available parameter in clinical gait laboratories. We therefore systematically imposed perturbations on hip and pelvis kinematics. The hip adduction, rotation and flexion angles as well as the pelvis obliquity angle were perturbed by ±5°, both individually and in combination (resulting in 405 simulations, 81 simulations for each subject). The results showed that a change in frontal plane hip and pelvis kinematics had a dominant effect on the magnitude of hip contact forces. To decrease hip contact forces, the hip adduction angle should be decreased, while an increase in hip adduction will effectively increase the hip contact forces. Results also showed that the changes in hip contact forces were related to changes in hip moments, indicating that the hip adduction moment can be used to predict the contact forces independent of the specific kinematic strategy. However, at the second peak in hip contact force both the hip adduction and flexion moments are required to accurately predict hip contact forces. Therefore, gait training that focuses on decreasing hip adduction moments has the largest potential to reduce hip contact forces independent of the kinematic strategy used. On the other hand, increasing hip adduction moments has the largest potential to increase hip contact forces. Besides that the magnitude of the hip contact force is affected, also the orientation changes. Therefore, in study VI the effect of kinematics on the risk of antero-superior edge loading was investigated. We used the same synthetic dataset as for study V and assumed a prosthesis cup was placed within Lewinnek’s safe zone and calculated the angle between the edge of the cup and the hip contact force vector (HCF-edge angle). In addition to the effect of altered kinematics, also the effect of 4 different cup positions (an inclination angle of 30° and 50° and anteversion angle of 5° and 25°) and 2 different coverage angles (180° and 168°) was evaluated. Results showed that factors related to the implant position and design largely affect edge loading. For cup positions within Lewinnek’s safe zone, a small anteversion angle (5°) never resulted in edge loading, while increased cup anteversion and inclination resulted in the smallest HCF-edge angles, indicative of edge loading. Although the analysis evaluated cups within the safe zone, several gait patterns and even some unperturbed gait patterns, presented a risk of edge loading. Also the coverage angle influenced edge loading, where a coverage angle of 168° increased the risk of edge loading. Apart from implant related factors, gait kinematics also affected the risk of edge loading. This is clinically relevant as gait kinematics are the sole factor that can be altered after THA to prevent edge loading. The risk of edge loading depended on the phase in the gait cycle. At initial double support there was no risk of edge loading for any of the imposed gait patterns. The highest risk of edge loading was found during terminal double support. Increased hip abduction, resulting in decreased hip contact force magnitude, and decreased hip extension, resulting in decreased risk on edge loading, were defined as gait strategies that could prevent edge loading. For studies VII and VIII, the workflow for the calculation of hip contact forces in patients with hip pathology was optimized based on the previous studies and the effect of hip pathology related kinematics on hip contact forces was analysed. In these studies the generic model including wrapping surfaces was used, based on studies I and II. Based on study IV, a static optimization was used to calculate muscle forces. In study VII a detailed analysis of the hip kinematics, kinetics, work, power, muscle activity and hip joint loading in OA patients was performed. The aim of this study was to define a detailed mechanical profile of patients with hip OA. Patients with hip OA and healthy control subjects performed gait analysis including electromyography (EMG) measurements. Kinematics, kinetics, work, power, as well as hip contact forces were calculated. Specifically frontal plane changes were found in patients with hip OA. Patients walked with reduced hip adduction angles and abduction moments as well as decreased power in the frontal plane. Also a large decrease in positive work of the hip abductors, representative for m. gluteus medius weakness, was found. Given the reported decrease in frontal plane power, the relative increase in muscle activity may reflect abductor muscle weakness. Furthermore, patients presented with reduced walking velocity and decreased maximal hip extension. The hip contact forces showed reduced hip loading in patients with hip OA, presenting reduced hip contact forces both at the first and second peak. This coincided with reduced hip adduction angles, hip abduction moments and frontal plane power. At the second peak, hip contact forces were even more reduced and coincided with the reduced hip flexion moment. Study V also showed that these gait strategies resulted in decreased hip loading. As a result of the adopted kinetic strategy, not only the magnitude of the hip contact forces decreased. The hip loading direction was also found to be more vertical in the sagittal and frontal plane, but more medial in the transversal plane. Overall, insights into compensatory mechanisms in hip OA patients were obtained, and the biomechanical strategies observed in these patients resulted in a decrease in hip loading. Study VIII evaluated the differences in hip joint loading between different surgical hip arthroplasty procedures, the direct anterior (DAA), direct lateral (DLA) and posterolateral (PLA) approaches in patients one year or more after surgery. The DAA is expected to result in reduced muscle damage and abductor dysfunction, which might result in a hip loading which is more comparable to healthy control subjects. Patients operated via the DAA received a THA, while DLA and PLA patients received a resurfacing hip arthroplasty (RHA). Gait trials were collected for all subjects. For control subjects and DAA patients also stair ascent and descent trials were collected, as these tasks are expected to better distinguish between groups. Results showed no differences between the patient groups one year or more after surgery. However, patients did show aberrant hip loading compared to healthy controls. In gait differences in hip contact forces were specifically found at terminal double support. Limited differences between controls and DAA patients were found during stair ascent and descent. Interestingly, the magnitude of the hip contact force was not larger for stair ascent and descent than for gait. This indicates that stair ascent and descent cannot be labelled as higher demanding tasks than gait in terms of magnitude of hip loading. In contrast, hip loading orientation did discriminate between DAA patients and controls during stair ascent (presenting a more vertical loading in DAA patients) and descent (presenting a more anterior loading in DAA patients). Since in gait no differences in load orientation were found between controls and patients, this might indicate that stair ascent and descent are more demanding when considering the load orientation. All these studies showed that hip pathology patients, both OA and hip arthroplasty patients, adopt a gait strategy that decreases hip loading. Besides that, we were able to identify gait strategies that affect the magnitude of hip loading, e.g. decreased loading by decreasing the hip adduction moment, and reduce the risk on edge loading, by decreasing hip extension. Overall, this PhD contributed to insights on several important factors affecting the calculation of hip contact forces. The insights resulted in the creation of a musculoskeletal model that included wrapping surfaces around the hip joint to include the effect of the hip capsule on the muscle paths. Insights of the first studies were used to optimize the workflow for calculating hip contact forces in patients with hip OA and after THA. These studies are relevant to be able to monitor hip loading in patients, which in future research can be used to investigate stresses and strains in bone and cartilage.