AbstractBackground Knee osteoarthritis (KOA) can produce pain, swelling, stiffness, and loss of joint function. Globally, KOA affects more than 10% of men and 18% of women aged over 60 years, with more than half reporting moderate to severe pain. Deep flexion activities are challenging for those with KOA, yet they are important both culturally and in our activities of daily living. Improving our knowledge of the kinematics of deep flexion can help understand the relationship of kinematics and KOA and may translate into better total knee replacement (TKR) design. Accurate in vivo 3D knee kinematics can be utilised in studying the relationship between function and knee motion. Kneeling is the deep flexion activity investigated in this research as it is one of the most challenging activities for people with KOA and TKR because it necessitates loading the knee into extreme deep flexion. The five aims of this deep flexion kneeling kinematics research were to 1) to quantify the current knowledge as determined by medical imaging, 2) to understand the effects of ageing on the kinematics of kneeling, 3) to measure and compare healthy and end-stage KOA cohorts, 4) to investigate the associations between knee kinematics and common KOA treatment outcome measures, and 5) to investigate the predictive capabilities of kneeling kinematics. Methods Our systematic review and meta-analysis were conducted using a computerised literature search and bibliography review without date restriction. Studies were included if they described deep flexion (flexion higher than 120°), and included the movements squatting, lunging and kneeling. The meta-analysis generated a pooled effect size using a random-effects model when heterogeneity was moderate to high. Participants were recruited with healthy knees (n=67) or end-stage osteoarthritis awaiting TKR (n=59). The data collected included six-degree-of-freedom (6DOF) knee kinematics, grade of osteoarthritis severity, patient-reported-outcome-measures (PROMs), clinical scores and functional test results. 6DOF kinematic data were collected by first recording both dynamic kneeling motion within a fluoroscope and a static CT image of the knee joint. The fluoroscope and the CT data were then spatially aligned and ‘registered’, frame by frame using Orthovis©; a (three dimension/two dimension) 3D/2D multi-modal image registration technology that accurately and non-invasively measures 6DOF knee kinematics. The motion was quantified using the Grood and Suntay reference system. The imaging data were used to perform biomechanical, rigid-body, dynamic analyses of kneeling. Four studies were undertaken using these data to ascertain the effect of age on kneeling kinematics, the effect of KOA, the relationship between kinematics and outcome measures and finally, the predictive capacity of kneeling kinematics with respect to KOA severity. To examine healthy ageing, 6DOF kneeling kinematics were compared for four healthy age-groups. Differences between the healthy and osteoarthritic groups were analysed with body-mass-index (BMI) as the covariate. To determine the associations between 6DOF kneeling kinematics and the variation of the PROMs, clinical and functional data, we used multiple step-wise linear regressions. Finally, a predictive model of KOA severity using kneeling kinematics was developed using data mining and machine learning following the Cross-Industry Standard Practice for Data Mining (CRISP-DM) protocol. The dataset was partitioned into training (60%), testing (20%), and validation (20%). Results The systematic review and meta-analysis included 12 studies (with 164 participants aged 25–61 years in vivo, and 69–93 years in vitro) in the analysis. In vivo measurement technologies included radiographs, open MRI and 2D/3D MRI or CT image-registration on fluoroscopy. In vitro methods utilised Microscribe. The meta-analysis found that between 120° and 140° flexion, there was internal tibial rotation and posterior translation of both the medial and lateral-femoral-condyles. There was high heterogeneity between squatting and lunging for medial-femoral-condyle translation, whereas the lateral-femoral-condyle had low heterogeneity, suggesting that only the medial-femoral-condyle translation is sensitive to the different activities Comparison of healthy age groups showed no differences between any of the age groups except that the 80+ group femurs were more varus and anterior at 110° and 120°. Also, after 120° flexion, the 80+ group rotated further and faster into valgus. Comparison of healthy and osteoarthritic participants showed that there were differences in kinematic positions, displacement and rate-of-change (/°flexion). At 100° flexion KOA femurs were more varus and externally rotated. Between 120° to maximum flexion the KOA femurs translated 5.8 (2.8, 8.9) mm less posteriorly, 1.3(0.6, 2.1) mm more superiorly and rotated 2.8° (0.8°, 4.7°) more externally. Knees with osteoarthritis had 12.8° (95% CI 8.6°, 17°) less maximum flexion in kneeling than healthy knees, moving into flexion. At maximum flexion, KOA femurs were more anterior and medial. The relationships between kneeling kinematics and the PROMs, clinical and function scores were moderate for the Oxford Knee Score (OKS) (53.2%), American Knee Society score (KSS) (52%) and pain visual-analogue-scale (painVAS) (48%). There was a moderate to small relationship with the associated quality of life (AQOL-8D), the timed-up-and-go (TUG) and remaining scores. Maximum flexion was strongly associated (85.3%) with femoral anterior position and internal/external rotation after 120° flexion and was the most significant contributor in all but the five-times-sit-to-stand-test. An 11° increase in maximum flexion explained minimally clinically important differences in many of the scores: including; 7 points of the OKS, 13 mm decrease on the painVAS, a 0.06 increase in the AQOL-8D, and a 1.3-sec decrease in the TUG time. Using kneeling kinematics, the classification and regression (CR&T) decision tree model delivered the highest accuracy and stability for training, testing and validating data. The performance of the model overall accuracy for training was 98.89%, testing was 100%, and validation 100%. Conclusion This research breaks new ground by using deep flexion kneeling kinematics to investigate healthy ageing, to determine kinematic changes due to osteoarthritis, to determine kinematic relationships with outcome measures, and to find a kinematic signature for KOA severity. We found very little kinematic impact of age, but real changes to knee kinematics with osteoarthritis in deep flexion. We found strong relationships between knee kinematics and clinical outcome measures. Finally, we developed a strong predictive model that can distinguish healthy, early and late stage KOA from kneeling kinematics. Understanding the relationships between osteoarthritis and kinematics may facilitate better non-surgical interventions.
|Date of Award||2019|
|Supervisor||Jennie Scarvell (Supervisor) & Phillip Newman (Supervisor)|
Healthy and Osteoarthritic Knee Kinematics: A 3D/2D Image-Registration Study of Kneeling
Galvin, C. (Author). 2019
Student thesis: Doctoral Thesis