The Effects of Age, Weight and Femoral Shape on Cortical Thickness and Mass of the Hip
Category: Bone Biomechanics and Quality (Basic)
Poster Sessions, Presentation Number: SA0056
Session: Poster Session I & Poster Tours
Saturday, October 5, 2013 12:00 PM - 2:00 PM, Baltimore Convention Center, Discovery Hall-Hall C
* , UNITED KINGDOM, Andrew Gee, University of Cambridge, , GB, , UNITED KINGDOM
Several studies have linked local properties of the cortex, particularly thickness, to fracture risk (Johannesdottir, Bone 2011; Yang, JBMR 2012). It is also clear that bone strengthening drugs have localised effects on the femur (Poole, PLoS1 2011). Recent developments in image processing allow accurate estimates of cortical thickness down to 0.3mm (Treece, Med Image Anal 2010, 2012). However, interpretation of the resulting data requires an improved understanding of the effects of other covariates on such data.
In this study, we determine how a priori selected covariates are predictive of cortical thickness and mass. We have previously investigated the effect of age, weight and osteophytes (Turmezei, Insights Imaging 2013): here, the explanatory variables we examine are subject age and weight, and femur size and shape. We analysed QCT data from 268 women, aged 53 to 98 years and weight 40 to 103 kg, from two studies conducted at two sites. Combining studies in this manner increases the significance of the results, provided we allow for site as a confounding variable. After spatially aligning the cortical data onto a canonical model, we use statistical parametric mapping as implemented in the SurfStat package (Worsley, NeuroImage 2009) to analyse the results. The generalized linear model for cortical thickness or mass is 1 + age + weight + site + s1 + s2 + s3 + s4 + s5, where s1-s5 are the first five modes of shape variation derived from a statistical analysis of a much larger population. The results for age, weight and shape modes 1 (size) and 3 (head-neck angle) are shown in Figure 1.
Age (approx -1% per year) and weight (approx +0.5% per kg) are predictive of cortical properties over significant areas of the femur. There are also more localised effects associated with the shape modes. Observe how larger bones (mode 1) have thinner and lighter cortices: this might explain some of the process of causation by which taller people with larger femurs are more likely to fracture (Carballido-Gamio, JBMR 2013; Whitmarsh, Bone 2012). The results for modes 2–5 (not all shown) need interpreting with greater caution: they might not be caused by thickness and mass differences, but by misregistration in the spatial alignment, since such errors depend systematically on the individual's shape. The effects for mode 3 (head-neck angle) are more widespread and perhaps indicate a genuine reduction in the cortex with increasing head-neck angle.
* Presenting Authors(s):
, UNITED KINGDOM