Parameters derived from DXA-based structural engineering models of the proximal femur as a biomarker for hip fracture

Osteoporosis affects hundreds of thousands of people, mostly women, in the UK and they are at a high risk of broken hips (hip fracture) as a result. Hip fracture usually leads to a loss of independence and can even increase the risk of dying in the very elderly.

Currently the risk of hip fracture is assessed based on how dense the thigh bone is using x-rays. This method is not totally accurate since other factors, such as the size and shape of the bone and how the bone is distributed, also affect the bone strength.

We aim to interpret x-rays of the hip using engineering techniques to take account of its size and structure in order to predict the strength of the thigh bone. A large number of subjects will be involved to ensure the study is reliable and useful. We hope to find ways to better identify patients at high risk of hip fracture so that they can be treated appropriately.

The risk assessment for osteoporotic hip fracture is currently based on bone mineral density (BMD) as measured by dual-energy x-ray absorptiometry (DXA). However, this approach only explains about 65% of bone strength and low BMD is only found in 30%-70% fracture cases since factors such as bone geometry and tissue spatial distribution also determine the strength. Structural engineering model (SEM) of the proximal femur integrates all the geometry and bone distribution information embedded in DXA scans, therefore has the potential to improve greatly the bone strength assessment and fracture prediction. Although this potential has been demonstrated in cadaver studies the use of SEM as diagnostic tools for hip fracture has not yet been fully evaluated in large, well-characterised studies.
We therefore propose a clinical study to determine parameters from DXA-based SEMs that are highly effective as diagnostic tools for hip fracture. We will pool our previous studies on postmenopausal women to form a case (n=250) and 2 control groups (n=1000 each), one matched with age, another matched with age, height and weight. We will use part of the groups to determine the best predictors, and test these predictors on the other part of the groups. We will determine the precision and accuracy of a new and innovative 3D DXA technology in diagnosing postmenopausal osteoporosis and hip fracture. We will use data from TRIO study (200 young and 150 postmenopausal osteoporotic women) for osteoporosis assessment, and recruit 100 postmenopausal women, half with and half without hip fracture and matched with age, for hip fracture prediction. Using well-established methods, we will generate three 2D and two 3D SEMs of the proximal femur from the traditional and 3D DXA scans. The well-known hip structure analysis parameters, stress/strain and strength during sideway fall will be calculated. We will perform logistic regression followed by receiver operating characteristics analysis to identify the best predictors. Population attributable risk will be calculated. We will compare the performance between 2D and 3D SEMs.
The new images collected and SEMs will be in the public domain. Our industrial collaborator will have 6 months lead time to the images and SEMs and will implement the software on its devices so that the approach can be used widely. The results will be published in peer-reviewed journals and presented in learned society and patient interest-group meetings. We expect this research to attract considerable attention among those involved in clinical decision-making (such as NICE).