Use of Mendelian randomisation to examine the role of abnormal hip shape in the development of hip osteoarthritis

Context: Osteoarthritis (OA) is a type of joint degeneration that leads to pain and loss of function. Since OA is more common with increasing age, more people are becoming affected as society ages. My research focuses on hip OA, for which joint replacement surgery is the main treatment. In the UK in 2016, there were 80,000 hip replacements for OA costing the NHS £500 million. In a small minority, hip OA results from defective hip development in early life, leading to abnormal hip shape causing aberrant biomechanical forces that accelerate joint destruction. More subtle alterations in hip shape, harder to recognise using standard X-rays, have been suggested to contribute to hip OA development in the wider population. However, since early OA may itself affect hip shape, it's unclear whether hip shape changes reported in association with hip OA are causative factors or simply manifestations of early disease. Mendelian randomisation (MR), a statistical method where genetic factors are used to estimate the causal effect of any given factor independently of confounding, represents an ideal method to address this question. Although MR requires large datasets to ensure sufficient statistical power, the availability of hip DXA scans (a low radiation X-ray) in UK Biobank means that such an approach is now feasible.

Aims: I aim to establish whether alterations in hip shape reported to be associated with hip OA, including novel characteristics identified in my preliminary studies, are causal factors in hip OA development, as opposed to a consequence of the disease process. This will be achieved by performing MR analyses in 100,000 individuals from UK Biobank, with hip DXA scans linked to genetic data and information about hip OA.
1) I will extend automatic hip shape measurement to include new hip shape angles and models, using 100,000 DXA scans with UK Biobank
2) The same DXA scans will be graded for hip OA, using a further automated process also developed as part of my fellowship
3) I will describe associations between hip shape as characterised in (1), and hip OA as characterised in (2), as well as with other OA-related phenotypes including self-reported hip pain and risk of subsequent hip replacement
4) MR methods will be applied to establish whether associations between hip shape and hip OA described in (3) represent causal relationships

Methods: Under the separate Wellcome funded AUGMENT grant which my application will exploit, a method for automating placement of points around the outline of the hip joint (thigh bone and pelvis) will be developed and applied to UK Biobank hip DXA scans. I will use these marked up points in my fellowship, using different subsets and combinations to generate a range of aspects of hip shape, and to classify hip scans for OA according to the presence of osteophytes (bony protuberances) and joint space narrowing. A number of shape parameters thought to be associated with hip OA will be derived, including width relative to height of the upper thigh bone, the angle between the middle of the femoral head (thigh bone) and the edge of the acetabulum (pelvis), and the size of the lesser trochanter (a boney protusion on the inner aspect of the thigh bone).

Potential benefits: Identification of alterations in hip shape causing hip OA in the wider population will contribute to our understanding of the causes of OA, particularly the role of altered biomechanical forces. This knowledge will provide the basis for more effective preventative strategies for hip OA, including the role of bespoke exercises, physiotherapy and orthotics, designed to correct the abnormal joint loading predicted to follow from specific shape changes. By studying associations between hip morphology as evaluated by hip DXA and subsequent risk of hip replacement, my fellowship will also contribute to development of use of DXA scans, widely used to screen for osteoporosis, to additionally screen for hip OA.

I aim to establish whether there is a causal relationship between hip shape and hip OA:

(i) I will derive a range of hip shape parameters in 100,000 hip DXA scans from UK Biobank
(ii) I will ascertain radiographic hip OA in the same DXAs as in (i)
(iii) I will describe observational associations between hip shape measures derived in (i), and radiographic OA derived in (ii) and symptomatic OA
(iv) I will use MR to determine whether observational associations in (iii) represent a causal pathway between hip shape and hip OA

Methods: BoneFinder software, funded by the linked Wellcome AUGMENT grant will automatically plot points around the bony contours of 100,000 hip DXAs in UK Biobank. Having refined this method, I will generate a range of shape parameters previously found to be associated with hip OA, including alpha and CE angles (defining FAI and acetabular dysplasia). I will also apply novel statistical shape models, informed by analysis (as part of my fellowship) of the relationship between 2D and 3D hip shape, derived from hip DXA and CT scans respectively, in 3,500 MrOS participants. A classifier which defines osteophytes, joint space narrowing and with them radiographic OA will be developed in a subset of 1,000 manually graded DXA scans before being applied to all 100,000. Observational associations will be analysed between different hip shape parameters and radiographic OA, self-reported hip pain and risk of hip replacement. Following a GWAS, MR using genetic instruments for the different hip shape parameters will evaluate whether hip shape has a causal effect on hip OA.

Opportunities: Finding that altered hip shape leads to hip OA in the wider population will stimulate research into the role of abnormal biomechanical loading in hip OA, and novel prevention strategies aiming to correct this. Identification of hip DXA parameters predictive of hip replacement will provide a basis for using these scans to select those at risk of OA progression.

Patients: Assuming I find a causal relationship between hip shape and hip OA, those at risk of developing hip OA as a consequence of specific abnormalities in hip shape identified by my research, will benefit from the development of targeted interventions aiming to prevent hip OA within the next decade, achieved by correction of altered biomechanical forces responsible for accelerated joint degeneration. My research will also characterise associations between hip morphology assessed by DXA, and subsequent risk of hip replacement, irrespective of whether a causal relationship exists. In the midterm, application of these findings, leading to extended use of hip DXA to assess risk of OA as well as osteoporosis, will benefit substantial numbers of individuals with early hip OA, by identifying those at increased risk of progression to hip replacement. Increased uptake of general preventative measures like weight loss in high risk individuals identified in this way is expected to ameliorate progression and decrease morbidity associated with both loss of joint function and joint replacement surgery.

Charities: This fellowship will benefit charities such as Arthritis Research UK, whose mission is to reduce disability arising from musculoskeletal diseases. Relationships between specific hip shape and hip OA identified by my project, will provide Arthritis Research UK with new research opportunities to support, intended to examine the underlying biomechanical pathways in more detail, and develop and test interventions which target these. This fellowship will also benefit Arthritis Research UK, by increasing the impact of research it has recently supported (the AUGMENT project followed on from an Arthritis Research UK-funded project into the genetics of hip shape derived from DXA). In addition, by supporting my development as a clinician scientist, this fellowship will help to increase rheumatology research capacity in the UK, which is essential if the effectiveness of prevention and treatment strategies intended to reduce the burden of disability from musculoskeletal disease is to be improved significantly.

NHS: Any intervention to correct biomechanical disturbances contributing to hip OA as a result of altered hip shape, arising from my research, is likely to be initially developed and piloted by a research project funded by a charity such as Arthritis Research UK. Assuming these initial studies have a successful outcome, this project will also benefit the NIHR by providing an opportunity to support subsequent translational research studies intended to obtain evidence to underpin uptake across the NHS.

Commercial sector: This proposal will benefit manufacturers of DXA scanners and image processing software, who are likely to want to exploit the findings commercially. Several commercially available software packages are available which extract additional information about osteoporosis risk from DXA scans (eg hip structural analysis, trabecular bone score). My research may interest the same providers in extending use of DXA to assess risk of hip OA, based on software which extracts features of hip morphology associated with increased risk of progression to hip replacement. In addition, if hip shape parameters generated by my project are related to subsequent hip fracture risk, commercial providers may want to produce further packages which predict hip fractures more accurately, based on incorporation of these additional shape parameters.

Wider public: Total hip replacements currently cost the NHS £500 million annually. In the long term, through adoption of interventions intended to prevent OA in those with abnormal hip shape, and DXA-based screening strategies for identifying those at high risk of progression, reduced requirement for hip replacements arising from my research will benefit the wider public by reducing this cost, thereby freeing up valuable resources to be spent elsewhere.