Cluster analysis of finite element analysis and bone microarchitectural parameters identifies phenotypes with high fracture risk
Cluster analysis of finite element analysis and bone microarchitectural parameters identifies phenotypes with high fracture risk
High-resolution peripheral quantitative computed tomography (HRpQCT) is increasingly used for exploring associations between bone microarchitectural and finite element analysis (FEA) parameters and fracture. We hypothesised that combining bone microarchitectural parameters, geometry, BMD and FEA estimates of bone strength from HRpQCT may improve discrimination of fragility fractures. The analysis sample comprised of 359 participants (aged 72–81 years) from the Hertfordshire Cohort Study. Fracture history was determined by self-report and vertebral fracture assessment. Participants underwent HRpQCT scans of the distal radius and DXA scans of the proximal femur and lateral spine. Poisson regression with robust variance estimation was used to derive relative risks for the relationship between individual bone microarchitectural and FEA parameters and previous fracture. Cluster analysis of these parameters was then performed to identify phenotypes associated with fracture prevalence. Receiver operating characteristic analysis suggested that bone microarchitectural parameters improved fracture discrimination compared to aBMD alone, whereas further inclusion of FEA parameters resulted in minimal improvements. Cluster analysis (k-means) identified four clusters. The first had lower Young modulus, cortical thickness, cortical volumetric density and Von Mises stresses compared to the wider sample; fracture rates were only significantly greater among women (relative risk [95%CI] compared to lowest risk cluster: 2.55 [1.28, 5.07], p = 0.008). The second cluster in women had greater trabecular separation, lower trabecular volumetric density and lower trabecular load with an increase in fracture rate compared to lowest risk cluster (1.93 [0.98, 3.78], p = 0.057). These findings may help inform intervention strategies for the prevention and management of osteoporosis.
Westbury, Leo
5ed45df3-3df7-4bf9-bbad-07b63cd4b281
Shere, Clare
baded3fd-6bce-4a41-8a8c-aad48a8a7c6f
Edwards, Mark
06c1db44-4341-455e-8812-0ab4a1043828
Cooper, Cyrus
e05f5612-b493-4273-9b71-9e0ce32bdad6
Dennison, Elaine
ee647287-edb4-4392-8361-e59fd505b1d1
Ward, Kate A.
39bd4db1-c948-4e32-930e-7bec8deb54c7
Westbury, Leo
5ed45df3-3df7-4bf9-bbad-07b63cd4b281
Shere, Clare
baded3fd-6bce-4a41-8a8c-aad48a8a7c6f
Edwards, Mark
06c1db44-4341-455e-8812-0ab4a1043828
Cooper, Cyrus
e05f5612-b493-4273-9b71-9e0ce32bdad6
Dennison, Elaine
ee647287-edb4-4392-8361-e59fd505b1d1
Ward, Kate A.
39bd4db1-c948-4e32-930e-7bec8deb54c7
Westbury, Leo, Shere, Clare, Edwards, Mark, Cooper, Cyrus, Dennison, Elaine and Ward, Kate A.
(2019)
Cluster analysis of finite element analysis and bone microarchitectural parameters identifies phenotypes with high fracture risk.
Calcified Tissue International.
(doi:10.1007/s00223-019-00564-7).
Abstract
High-resolution peripheral quantitative computed tomography (HRpQCT) is increasingly used for exploring associations between bone microarchitectural and finite element analysis (FEA) parameters and fracture. We hypothesised that combining bone microarchitectural parameters, geometry, BMD and FEA estimates of bone strength from HRpQCT may improve discrimination of fragility fractures. The analysis sample comprised of 359 participants (aged 72–81 years) from the Hertfordshire Cohort Study. Fracture history was determined by self-report and vertebral fracture assessment. Participants underwent HRpQCT scans of the distal radius and DXA scans of the proximal femur and lateral spine. Poisson regression with robust variance estimation was used to derive relative risks for the relationship between individual bone microarchitectural and FEA parameters and previous fracture. Cluster analysis of these parameters was then performed to identify phenotypes associated with fracture prevalence. Receiver operating characteristic analysis suggested that bone microarchitectural parameters improved fracture discrimination compared to aBMD alone, whereas further inclusion of FEA parameters resulted in minimal improvements. Cluster analysis (k-means) identified four clusters. The first had lower Young modulus, cortical thickness, cortical volumetric density and Von Mises stresses compared to the wider sample; fracture rates were only significantly greater among women (relative risk [95%CI] compared to lowest risk cluster: 2.55 [1.28, 5.07], p = 0.008). The second cluster in women had greater trabecular separation, lower trabecular volumetric density and lower trabecular load with an increase in fracture rate compared to lowest risk cluster (1.93 [0.98, 3.78], p = 0.057). These findings may help inform intervention strategies for the prevention and management of osteoporosis.
Text
HCS cluster analyses (HRpQCT and FEA) with figure (final version)
- Accepted Manuscript
More information
Accepted/In Press date: 9 May 2019
e-pub ahead of print date: 11 June 2019
Identifiers
Local EPrints ID: 431581
URI: http://eprints.soton.ac.uk/id/eprint/431581
ISSN: 0171-967X
PURE UUID: b0462771-7d20-4cd6-b2a3-2508a6ddc062
Catalogue record
Date deposited: 10 Jun 2019 16:30
Last modified: 18 Mar 2024 05:04
Export record
Altmetrics
Contributors
Author:
Clare Shere
Author:
Mark Edwards
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics