Population variations in ankle complex structure using machine-learning 3D-MRI statistical shape modelling and implications for Osteoarthritis : the Newcastle thousand families study cohort (medical imaging)

Abstract

Background: Osteoarthritis (OA) is the most common joint condition. Although risk factors are well-recognised, the underlying pathogenesis is not fully understood. OA may be linked to biomechanics and force distribution during activities, with the ankle playing a crucial role in absorbing and transferring lower-limb forces. The ankle is structurally complex and understudied. Further understanding of ankle morphology could lead to better prevention of, and treatments for OA. This study aims to reconstruct Three-dimensional Statstical Shape Models (3D-SSM) from ankle MRIs, identify key ankle morphological features and explore their inter-relation to OA in the Newcastle Thousand Families birth cohort. Methods: Prospectively collected data from 206 cohort participants aged 62 was used: 3D SSM were built using ankle MRI images to quantify morphological features (bone shape/area, and joint space widths (JSW)) of ankle bones and joints. Sex differences, presence/absence of knee/hip OA, and associations between quantified morphological features and other clinical variables (body anthropometry, bone mineral density (BMD), and self-reported questionnaire data) were analysed. Results: This study found significant sex differences in ankle morphology, with females having smaller JSW, bone area, and thinner bone shapes. Knee OA showed a pes-plan foot bone shape while the hip OA had pes-cavus. Positive associations were found between ankle JSW and BMD. Subtalar posterior JSW showed a positive association with knee JSW. Several ankle JSWs showed a positive association with hip JSW in males only. Participants with hip/knee OA have smaller ankle JSWs compared to others. Conclusions: This is the first study using novel machine-learning techniques simplifying morphological complexity to examine ankle structure and its association with OA, using a population birth cohort. Results showed significant associations between ankle morphology and OA. This indicates ankle morphology may have direct effects on knee and hip OA through gait and mechanical force distribution. However, further studies are needed to confirm this hypothesis, which could be tested in subsequent reviews of this cohort.