IFHA Global Summit of Equine Safety and Technology: Fracture prediction and prevention

Abstract

Although the incidence of severe fractures in racehorses is relatively low, the impact on the welfare of individual animals and humans is often profound. The high visibility of these severe race day incidents also contributes to growing pressure from sectors of the media and prompts the wider public to question whether horse racing should have a social licence to operate. With most fractures in the racehorse occurring as an acute manifestation of a chronic repetitive overuse injury, there is tangible potential for timely intervention and prevention; something the industry must harness to improve racehorse welfare and demonstrate a culture of care of their equine athletes. Substantial progress has been made towards our understanding of fracture pathophysiology and in the development of modalities that can detect pathology that precedes fracture. However, there remains work to translate theoretical prediction of horses at greater risk of fracture to workable risk profiling mechanisms and to implement practical screening protocols.

To this end, in June 2024, a global, multidisciplinary group of expert researchers gathered at Woodbine Racecourse, Toronto, for a three-day discussion on how current research can be advanced and translated into action by developing a strategic plan, including the development of realistic screening programmes to identify high-risk horses. The workshop was included in the International Federation of Horseracing Authorities (IFHA) Global Summit on Equine Safety and Technology, an event sponsored by the Hong Kong Jockey Club Equine Welfare Research Foundation, Cornell University's Harry M. Zweig Memorial Fund for Equine Research and Woodbine Entertainment Group. The extensive mobility of the racehorse dictates risk profiling and fracture prevention as a global problem that requires international collaboration, but with the realisation that solutions may need to be localised to reflect inter-jurisdictional differences. This editorial serves to highlight the progress made, the key topics of discussion and the actionable items that can be taken forward from this meeting.

The incidence of fatality due to distal limb fractures in racing has been reported historically to range from 1 in every 70 to 4249 race starts,¹ dependent on race type, bone involved and geographical location. The incidence is much higher in races over obstacles, partly due to the additional trauma due to falls, although many fractures associated with this sport have similar pathology to those commonly seen in horses racing on the flat. Catastrophic fractures can be differentiated into two broad groups, which in turn reflect variation in the underlying pathophysiological process: those that develop soon after a horse enters training or returns to training following a rest period, and those that occur following a sustained period of work without rest.² In both scenarios, fracture results from accumulated bone damage. In the first, the damage accumulates rapidly when the horse is exposed to loads for which the bone has not yet adapted, whilst the latter is the result of prolonged, high-intensity exercise exceeding the capacity of adapted bone without opportunity for bone repair. In either case, fracture is normally the result of mechanical compromise exceeding a threshold. Recent work elegantly demonstrates subtle variations in the pathogenesis of one of the most common racing fractures (parasagittal fractures of the distal condyles of the third metacarpal or metatarsal bone), which can be linked to these different scenarios.³

Most fractures seen in racehorses show a high degree of consistency in terms of their location and morphology,⁴ often with evidence of pre-existing pathology at the site of fracture initiation and/or at the same anatomical location in the contralateral limb.⁵ This suggests consistent loading of bones between different horses and confirms a prodromal phase to fracture, associated with accumulation of microdamage and the associated repair response. Damage occurs in the bone matrix with repeated cyclic loads (strides) and unless a period of rest is instigated to allow repair, accumulation of damage and osteoclastic resorption of diseased matrix can result in areas of focal osteopenia. These foci can act as stress risers for larger cracks and complete fracture formation, even when the bone is only subjected to normal physiological loading.⁶ However, although the process is largely understood, and complete fracture is usually preceded by accumulation of damage that may be resolved with rest, there are challenges to utilising this information for identification of horses at risk of fracture. Fatigue is known to be a stochastic process, and when inherent individual biological variation is considered, combined with challenges in detecting microdamage in vivo it is extremely difficult to estimate the likely number of load cycles (strides) remaining prior to fatigue failure. Consequently, we need to utilise secondary indicators of bone fatigue to detect horses approaching their limits of skeletal structural integrity.

Essentially, there are two main strategies to reducing fracture incidence: modification of extrinsic risk factors known to increase fracture risk, and identification of individual horses at increased risk of fracture so that they can be rested and/or managed appropriately to enable biological repair. Previous epidemiological studies combined have investigated around 300 factors for catastrophic musculoskeletal injury,² a subset of which have the potential to be modified to reduce fracture incidence rates. Recurring themes of discussion during the workshop were the potential for improved racetrack surface management and identification of low-risk training regimens.

Although the current body of evidence shows inconsistent results with regard to the influence of racetrack surface on musculoskeletal injury,⁷ this is probably a reflection of geographical variation in surface type and conditions and the parameters used for reporting. However, with racetrack surfaces open to modification and affecting all horses competing, their management has potential to reduce fracture risk. The workshop group discussed the goal to be characterisation of track surface properties that are optimal for the horse, and ways in which these could be applied globally; ensuring consistency so that horses race on surfaces to which they have become adapted during training. Of consideration when designing such a surface will be the need to minimise fetlock hyperextension; with a dirt surface having been shown to induce greater fetlock angles than a synthetic surface.^{8, 9} Consideration should also be given to the surface as a layered medium and how track surfaces are managed through hydration, aeration, harrowing and rolling^{10, 11} will also be important, together with impact to the skeleton via the horse–surface interaction, which may be modified by the horses' shoes. This plethora of interacting variables means studies to further our understanding of such relationships will be complicated, time-consuming and require significant investment and collaboration to ensure consistent reporting. However, such studies would have tangible potential to lead to significant improvements in equine safety and welfare for the racing industry.

Through our understanding of bone biology and fracture pathogenesis, we know that training intensity needs a balance between that required for protective skeletal adaptation and maintenance of cardiovascular fitness for racing performance, and excessive workload levels that cause injury. Accordingly, it is clear that both too little and too much work may result in fracture, but due to the complexity of training regimens and their interaction with resting protocols there are limitations on the advice we can provide on optimal training protocols. It is also worth noting that variation in response to training and ability to manage different intensities of exercise means that designing those optimal training protocols may remain elusive. What is clear is that the speed of exercise is a key consideration. A linear increase in speed has an exponential increase in impact on the skeleton in terms of using up its fatigue life; speed acting as a proxy for load on the bone.¹² From this we can infer that training at maximal gallop speed should be kept to the minimum required for skeletal adaptation and fitness; anything in excess of this potentially moves a horse into a higher risk category where fatigue life of the bone may be exceeded more readily by extrinsic factors. It is worth noting that in Victoria, Australia, it was found that trainers putting their horses through a lower volume of speed work reduced their catastrophic injury rates without affecting performance success.^{13, 14} This is an important message that must be conveyed to trainers and additional work to further elucidate the relationship between exercise intensity and fracture risk could allow evidence-based training to be carried out in the future.

The genetic basis of fracture risk, biomarkers and the use of wearable monitoring devices are all exciting areas of research that have shown significant advances in recent years and constituted a key component of workshop discussions. However, although they show promise as part of multi-layered screening approaches to identify horses at increased risk of imminent fracture, currently their application is in enhancing our understanding of the underlying pathological process to guide development of novel diagnostic, therapeutic and preventative options.

Distal limb fractures have been found to have a genetic element, with a heritability of 0.21 to 0.37 identified in Thoroughbred racehorses.¹⁵ A genome wide association study found significant genetic variation for fracture risk on four different chromosomes (9, 18, 22, 31) with three single nucleotide polymorphisms (SNPs) on chromosome 18 and one SNP on Chromosome 1 significantly associated with fracture.¹⁶ Additionally, the study of carpal chip fractures in Japan has identified specific areas of chromosome 18 where there may be a trade-off between racing performance and fracture risk.¹⁷ Although this heritability indicates that educated selective breeding could reduce fracture incidence via minimising the genetic susceptibility of the horses racing, overall, it is a complex trait where the interaction of environmental and genetic factors determines fracture status. Accordingly, despite polygenic risk scoring being currently optimised to identify horses at high genetic risk of fracture, the value is in further studies to understand why these horses are at higher risk.

Studies integrating multiple genomic techniques and in vitro cell culture models have begun to identify novel genes that are associated with fracture risk. These genes are associated with biological processes and pathways that may result in changes to bone mineralisation and the extracellular matrix.¹⁸

The translation of this genetic information into studies to elucidate their practical impact, are currently hampered by lack of engagement and understanding from the industry. We need to make the message clear that it is not about preventing high-risk horses from racing but identifying them so that further scrutiny can keep them safe and, over time, breeding a racehorse that can sustain the demands of training. Added to this is the recognition that in humans there is good evidence for the impact of intra-uterine and post-natal environmental exposures on the long-term health and disease risk of offspring; maternal diet, smoking and physical activity, for example, affect bone mineral acquisition in utero, whilst low birth weight and poor childhood growth have been linked to increased risk of hip fracture in later life.¹⁹ The identification of similar associations in horses could further guide breeding practices and early life management to optimise the long-term health and athletic potential of our Thoroughbred athletes.

In principle, biomarkers are an ideal method through which large numbers of horses could be feasibly screened to identify those at higher risk of fracture but, so far, results have proven mixed. Micro-RNAs (miRNAs) have been reported to be affected by exercise and mechanical loading in both horses and humans, but no specific miRNAs have been linked to stress fracture in equine studies and, even in large scale human studies of osteoporotic patients, no consistent association between miRNA profile and fracture occurrence has been detected.²⁰ Whole blood messenger RNAs (mRNAs) have also been investigated as potential biomarkers, with success reported previously.²¹ However, initial analysis of ongoing work using pre-race/pre-injury samples from horses with and without fractures suggests further work in this area is still necessary. Furthermore, classic protein biomarkers of bone formation and bone resorption have shown conflicting results between studies but have revealed greater understanding of bone pathophysiology. For example, it was hypothesised that international horses arriving for the Spring Carnival in Melbourne underwent significant osteoclastic bone resorption during quarantine that could put them at increased risk of fracture. However, investigation of classic biomarkers of bone formation and resorption did not reveal evidence of significant bone resorption in these horses during quarantine, but increased net bone formation (identified through increased serum osteocalcin) and persistent high cortisol, an indicator of stress, was detected.²² While the utility of biomarkers as early screening tools for identifying horses at risk for injury remains to be demonstrated, at a minimum, it is likely that they will continue to provide vital information regarding the underlying pathological changes seen with fractures.

Surveillance and comprehensive monitoring are essential to making progress in understanding and developing regulatory tools. So far, epidemiological studies have primarily provided retrospective information at the population level, but wearables will allow longitudinal data collection at the level of the individual horse; this being more relevant to application in risk profiling and screening strategies. Most wearables currently in use are biometric devices, able to capture speed and stride length, acceleration and some performance parameters, the latter of which will act as an incentive for trainer participation.

When it comes to monitoring, there are two groups of horses that would be valuable to identify through use of wearables: those at high risk of fracture due to excessive rapid accumulation of workload, and those with sub-clinical injury that may be indicated by reducing speed and stride length. Indeed, studies using Tasmanian racing data demonstrated that for every 0.1 m/s decrease in speed and each 10 cm reduction in stride length, estimated risk of injury increased 1.18× and 1.11×, respectively.²³ However, so far studies have only demonstrated proof of concept and the substantial inter-horse variation indicates that wearables would be of greatest value if used continuously; capturing an individual horse's stride characteristics over time so that any deviation from normal could be easily detected. To progress from detection, to prediction, of injury requires validation of the wearable devices in use and collection of data over significant periods of time. These training and workload data must then be made available to researchers in conjunction with injury data so that associations can be identified and used to develop algorithms that could be used for prediction.

With significant progress having already been made in risk factor identification and pathological understanding, the next step is to harness this information to develop a multimodal screening strategy to identify horses at high risk of imminent fracture. However, with fracture overall being a rare outcome, historically statistical models had only obtained predictive values of roughly 65%²⁴ and have not proven robust enough to make regulatory decisions at an individual level, although they have been used to inform policy. To improve the predictive ability of such models it is likely that testing in series will be the solution²⁵; using a series of tests to rule out horses at each level, thus reducing the denominator and artificially increasing the prevalence and as a consequence the associated positive predictive value at each step. This approach was outlined by the FRAT Group at a meeting in Newmarket 4 years ago,²⁵ but we now need to act and make this a reality. To do so will require large volumes of data to be collected from longitudinal studies over substantial time periods and, ideally, including many regions and racing jurisdictions. Indeed, a study in Hong Kong which was the first to make individual horse veterinary histories available for screening use, improved model predictive ability to near 85%,²⁶ thereby emphasising the value of additional horse-level information.

Although this is a key research aim, what should be remembered is that making predictions is not without risk in itself and models need to be validated prior to their implementation. Additionally, racing as an industry must understand that no such screening programme can ever be 100% accurate due to uncertainty in factors that we are unable to measure or are yet to identify. Accordingly, the risk will never be null, and it must be decided what level of risk the racing industry and the general public can tolerate; a lower risk tolerance leading to more horses being prevented from racing unnecessarily, but a high-risk tolerance potentially allowing more fractures to occur that could have been prevented. Whilst this work progresses in the background, it is important that action is taken now in areas where improvements are already being realised.²⁷

It is likely that imaging will form a significant element of any screening programme but we are limited by our inability to directly detect microdamage, instead relying on the identification of secondary changes such as subchondral bone sclerosis, focal lysis and general bone activity. Racing Victoria is currently the only jurisdiction with an active pre-race screening programme that was introduced following six Melbourne Cup horse deaths in only a decade, all of which were international horses. Involving three-dimensional imaging of all four fetlocks before travel in advance of the race meet, followed by CT scan of them on arrival at the track,²⁸ the programme has had a substantial impact with no serious injuries occurring since its introduction. In California, advanced imaging, more specifically positron emission tomography (PET) and magnetic resonance imaging, is increasingly used by regulatory veterinarians as a clearance to race tool, when concerns are raised from physical examination or gait abnormalities. Over 1000 horses have been imaged with PET scan at Santa Anita in the last 5 years with the support of a subsidised imaging programme available at the racetrack. Realistically, unless heavily subsidised, advanced imaging techniques are unlikely to be widely available as a screening modality for all bar the highest profile racing events. The group consensus was that more widespread screening across the general racehorse population may be an achievable intervention in the longer term. Radiography, like all imaging systems, has its limitations.²⁹ However, despite these, it is affordable and readily available. Once again reiterating findings at the previous FRAT meeting,²⁵ it was decided that diligence to ensure best clinical practice and obtainment of high-quality radiographs of relevant projections could have a significant screening impact. The full multi-level screening protocol may not be ready for implementation, but increased relevant imaging could lead to tangible benefits tomorrow; condylar fractures certainly being a subset that could be reduced by such an intervention.

In light of the meeting discussions, it is clear that there are different pathways to bone failure, and the complexity of the problem must be understood if the goal of accurate prediction and prevention is to be realised. To this end, the following short- and long-term actions were identified which need to be addressed by immediate collaborative research and regulatory efforts:

All of these aims focus on accurate surveillance and data collection that can then be translated into the building of prediction and prevention strategies; only through sustained effort and collaboration can accurate results then be transformed into tangible improvements to racehorse safety.

It must not be forgotten that considerable progress in the area of fracture prevention has already been made and we currently have available tools that can be implemented to make significant impact. Indeed, the fracture reduction rates seen in Melbourne, California and New York following introduction of increased scrutiny and a culture of safety, is evidence that fractures are preventable and change is possible. This workshop has forged collaborations and opened discussions but efforts need to be sustained and translated into tangible actions to realise the desirable outcomes.

The International Federation of Horseracing Authorities (IFHA) Global Summit on Equine Safety and Technology was sponsored by the Hong Kong Jockey Club Equine Welfare Research Foundation, Cornell University's Harry M. Zweig Memorial Fund for Equine Research and Woodbine Entertainment Group.

Victoria A. Colgate: Writing – original draft. The FRAT Group II: Writing – review and editing. Christopher M. Riggs: Conceptualization; writing – review and editing.