Fractures are the leading cause of racehorse fatalities, accounting for 75-80% deaths in North America and Great Britain, and are the result of fatigue rather than trauma. More than 60% of fractures in the lower limbs occur in the third metacarpal (MC3) and proximal phalanx (P1) bones, and thus are the focus of this thesis. Although horses don’t reach full skeletal maturity until approximately 4 years old, horses begin racing at 2 years old, with little preparation in terms of training. Exercise at a young age has been shown to have lifelong benefits for bone health, and may provide a route to strengthen bones in areas prone to fracture and prevent fractures in racehorses.

There are many factors that influence whether an exercise intervention will result in optimal bone adaptation including duration of the intervention, speed, direction (running straight vs turning), and age of the participants. Computational models can be used to non-invasively predict the mechanical loading environment of bone in vivo and therefore provide a means for evaluating the effect of different exercise modes pre-clinically rather than adopting a trial and error approach. Subject-specific finite element models can be developed from computed tomography (CT) data using a density-modulus relationship that relates CT-derived density to a modulus obtained from mechanical tests. Several density-modulus relationships exist for horse bone using data from adult horses. However, juvenile equine bone tissue differs in terms of composition and structure from adult tissue and is therefore expected to differ mechanically. The overall goal of my research is to reduce fractures in racehorses by preparing their bones while young with targeted exercise interventions.

In Aim 1, I developed density-modulus relationships for the MC3 and P1 bones in both the longitudinal and transverse directions. Bones were collected from juvenile horses (age < 1 year), scanned using a clinical CT, cut into sections along the length of the bone, and cored using a drill press. The bone cores were tested in compression and the modulus was then related to the CT density. I found that the density-modulus relationships differed significantly by anatomical location (MC3 vs P1) and orientation (longitudinal vs transverse), indicating each bone and orientation required a unique density-modulus relationship. Additionally, the use of a density-modulus relationship developed with adult tissue on juvenile bone samples resulted in an 80% increase in error when predicting the modulus, indicating the sensitivity of the density-modulus relationship to subject age.

In Aim 2, I evaluated the relationship between microstructure, tissue density, and mechanical properties in the same juvenile trabecular bone samples using in Aim 1. MicroCT data (resolution = 144 µm) for each core was used to calculate microstructural measures, and the mechanical test data from Aim 1 was used to calculate modulus, yield stress, ultimate stress, yield strain, and ultimate strain for each sample. P1 samples, in both orientations, were more ductile and had significantly different microstructure compared to MC3 samples. Tissue density, bone volume fraction, and trabecular thickness were the strongest predictors of mechanical behavior, although the strength of these relationships varied by anatomical location and orientation of the samples. These data provide relationships between mechanical properties and microstructure that can be used for in vivo predictions of bone health, or to evaluate the effects of interventions (exercise or pharmacological) on trabecular microstructure of juvenile horses.

In Aim 3, I evaluated the effects of an exercise intervention on MC3 and P1 structure, density, and strength using repeated CT scans and virtual compression testing. Twelve juvenile horses were enrolled in the study (6 control, 6 exercise). Exercise horses underwent an 8 week long exercise intervention and were CT scanned before exercise, four weeks after exercise, and 16 weeks after exercise. Control horses were scanned at the same ages as exercise horses. Area fraction and density were evaluated in each cross-section along the length of the bone, as well as within each quadrant of each cross-section. Finite element (FE) models of each bone were used as virtual compression tests to evaluate whole bone stiffness as a measure of whole bone strength. I found no significant differences between control and exercise horses for any metric evaluated, at any time point. These results indicate either (1) the exercise intervention had no effect, or (2) the exercise intervention had a mild effect on bone parameters but the sample size was too small to identify statistical differences. This study is the first exercise intervention in juvenile horses to evaluate adaption along the entire length of the bone and to include a measure of whole bone strength (virtual compression testing). Additionally, this study provides an important lower threshold that future exercise interventions should exceed.

Overall, this thesis lays the foundation for using finite-element models to non-invasively assess exercise interventions in juvenile horses by (1) providing the necessary density-modulus relationships that accurately describe juvenile bone, (2) highlighting the influence of microstructure and tissue density on mechanical behavior and how those relationships change between neighboring anatomical locations, and (3) evaluating an exercise intervention using longitudinal CT scans and finite-element modeling.