Longitudinal ambulatory measurements of gait abnormality in dystrophin-deficient dogs

Duchenne muscular dystrophy (DMD), a still incurable severe and generalized muscle disease in man, represents a therapeutic challenge because of the numerous obstacles it presents for systemic gene, cell or pharmacological therapies. Several of these issues, such as efficacy of studied drugs, or locoregional or systemic medication pathways, can be assessed in the dystrophin-deficient dog, a large animal model which mimics the human disease in many points, better than other animal models [1].

During pre-clinical trials, to assess a functional effect, these canine patients must undergo a clinical follow-up encompassing a qualitative and quantitative evaluation of their impaired functions, particularly the gait. Recently, two studies including ours have focused on gait analysis in adult dogs suffering from Golden retriever muscular dystrophy (GRMD), using two different tools [2, 3]. Both studies have demonstrated that GRMD dogs walk slower than healthy controls, and that several specific gait parameters were altered. The kinematic analysis of the hind limb has shown an increase of the stifle joint extension, and a decrease of the hock joint flexion, this last joint being moreover less mobile than in control dogs [2]. The accelerometry study we have conducted has shown less regular and powerful accelerations, a decrease of the stride length and frequency, and a redistribution of the power from the cranio-caudal to the medio-lateral axis [3].

These studies have demonstrated that gait analysis is feasible and repeatable in adult GRMD dogs and that these animals exhibit quantifiable gait impairments. However, the majority of the treatments that are tested in GRMD dogs are administered during the first months of life, at a period when the morphological changes due to growth as well as the clinical evolution rate are maximal [4‐6]. Therefore, we sought to longitudinally characterize the dynamic evolution of the gait parameters tested by an ambulatory gait analysis system based on 3D-accelerometers, in order to delineate the disease progression in the GRMD model over time. We aimed to address the question if the gait variables are able to reflect disease evolution according to the previous results obtained in control and dystrophin-deficient adults [3]. Another question was to determine if the gait analysis was able to quantitatively describe the well-known heterogeneity of the model [1, 7]. These points represent pre-requisites for the use of this method in pre-clinical trials. This is the first longitudinal study on gait analysis in growing GRMD dogs, and the first study using an ambulatory gait analysis system to monitor the canine gait evolution, during growth and disease progression.

This study has demonstrated the ability of 3D-accelerometry to early monitor the gait of GRMD dogs during growth and disease progression. It is the first study quantitatively describing sequential gait impairments in this canine model of DMD.

Early modifications of GRMD dogs' gait have been highlighted, at a stage when the gait could appear normal to a non-experienced observer, attesting to the sensitivity of the accelerometric method. Indeed as soon as at the age of two months, a dramatic lack of power was detected, resulting in a speed decrease, primarily explained by a drop in stride length. The decrease of stride frequency was a later event, secondarily contributing to the reduction of speed. These patterns of evolution are very close to those already described in DMD patients [12, 13].

Whatever the age of the dogs, the total power of GRMD dogs was shown to be strongly decreased, but varied with age in its three-axial composition. The decrease in total power was primarily harmonious among the three axes. Then, the relative dorso-ventral power dropped, followed by a later increase of the medio-lateral part of power. This last event has been described in adult GRMD dogs [3], but combined with a decrease of the cranio-caudal part of power, which was not present in growing puppies, even in 9 months-old animals. Some of the growing GRMD dogs were examined at the age of 12 months, and this test showed that within 3 months, the cranio-caudal relative power dropped and the dorso-ventral one normalized (data not shown). Briefly, during the first year of life of GRMD dogs, the gait first becomes less bouncing, maybe as a consequence of the early muscle weakness to support weight, then waddling maybe reflecting contractures and less balance, and finally less propelling and mobile.

One remarkable issue is the wide intra-individual variations of some variables between fortnightly sessions. If the repeatability between sessions has been demonstrated for adult clinically stabilized GRMD dogs [3], the same does obviously not hold true for clinically evolving and growing animals. It may be assumed that these variations are due to the fluctuating day-to-day general shape of the diseased animals during their first months of life. This inter-session variability also underlines the necessity of a careful clinical follow-up of these dogs and of frequent iterative examinations. The simplicity, the lightness and non-invasivity of the accelerometric device allow the repetition of tests over a short period of time.

A methodological issue relies on the fact that the animals self-select their speed of gait. Indeed the measurements give rise to comparisons of variables calculated from acceleration curves recorded over a wide range of speeds and with three types of gait. However, to the experience of teams working on GRMD dogs' locomotion [2, 3], this wide range of speed abilities would precisely make a standardisation of speed very difficult, if not impossible. As an example, if a standardized speed had been considered, the lowest speed measured in GRMD adults dogs, i.e. 0.26 m/s [3] should have been chosen, to allow all the ambulant dogs to perform the test. This protocol could not be retained as it is by far artificial and devoid of clinical sense, to force all the dogs to walk at this very low speed. Moreover, the habituation time required by the use of a treadmill on dogs would also probably be significant, and delay the beginning of the follow-up, which should be as early as possible, in a context of therapeutic trials. However, it is undeniable that most of the studied variables are influenced by the speed, and one may put forward that all the modifications observed in GRMD dogs are only due to speed decrease. The decreased relative force index proves that this is not true. This index was created in order to avoid bias due to different speeds on the total power, by dividing it by the speed. It is interesting to note that the evolution of this variable globally follows the evolution of the total power, attesting to both the variations of this variable and the differences with the healthy population are not only due to differences in speed of locomotion. Furthermore, the spontaneous speed of gait is known to be of clinical significance [14, 15] and for this reason it represents a relevant measure in GRMD dogs. Finally, one may analyse this issue of speed variations as a limitation, but we assume this can be seen as an advantage as well, because of the clinical pertinence provided by the self-selection of gait and speed. With the same intention to quantitatively assess a situation as spontaneous and physiological as possible, the preferred and more comfortable gait was selected, using the regularity index.

Attesting to the success of this process, the studied variables were, in a vast majority, correlated to the motor score, meaning that the degradation of the gait measured by accelerometry corresponds to the observed clinical evolution. In the same way as the clinical score, the gait variables reveal different evolution types. However, the relative SD observed among the GRMD population for the gait variables are larger than for the motor score (data not shown). This observation supports the argument for a higher sensitivity of the accelerometric measurements variables than the motor score to describe the dogs' locomotion ability. Moreover, if the motor score is a good index to quantify the general shape of the dog, it remains a non-objective evaluation, to use with caution, particularly in non double-blinded studies. However, accelerometry is the first fully objective and quantitative method of gait analysis, successfully assessed in clinically evolving GRMD dogs.

One of the objectives of this study was to assess if a longitudinal follow-up of one particular individual was possible. Examining more closely the results obtained dog per dog, different evolution profiles can be identified, allowing a quantitative description of the well-described heterogeneity of the disease [1, 7]. Notably, the three dogs having lost locomotion before the age of six months (Dchou, Dfois and Dsir) share the same characteristics of evolution. Within few weeks, their gait variables reached values comparable to or worse than the values of adult GRMD dogs. Thus, this disease-course bound to result in loss of locomotion, is not only more severe in outcome but also accelerated in its evolution. Moreover, at the first test session at 2 months, the values of stride frequency and of speed measured for these three dogs were among the lowest. This last point suggests that, as early as at the age of 2 months, their gait is already more impaired than the one of dogs who will not lose locomotion. Low stride frequency and speed values at 2 months could be considered as prognostic markers of poor outcome. The predictive value of these biomarkers needs a validation study with more animals. Concerning the 9 other dogs included in the study, different evolution patterns were also identifiable, in function of their capacity to perform a regular and maintained high frequency "bunny hopping" gallop. This "bunny hopping" gallop, also called bound, is used by large size mammals in acceleration and deceleration phases of a classical rotary gallop [16]. In the case of GRMD dogs, the maintenance of bound, a primitive high-frequency gait, could be an adaptive response to muscle weakness, to produce more power using two synchronized limbs, and avoiding a single limb weight support [16]. Two dogs, Dbrouille and Dalton, exhibited surprising evolution patterns. Indeed, after a period of degradation, they tended to improve their gait. This observation emphasizes the need to carefully interpret a slight improvement of gait in an animal participating in a treatment protocol.

In order to simplify the data interpretation, and because of the multitude of variables of interest, we applied the PCA multivariate exploratory method as a tool to position each dog in comparison to the others during time. We have chosen to project the puppies as supplementary observations on the PCA plane defined by the eight healthy and the 11 GRMD adult dogs of our previous study [3]. This has allowed analyzing the chronology of the relative trajectories of healthy and GRMD groups, and locating and following each animal at different ages on a reference PCA plane.

We proposed the use of the Euclidean distance from a control age-matched group as an overall variable to make gait analysis easier regarding data management and interpretation. This distance has already been described and used as a gait analysis index in kinematic studies [11, 17]. Our results regarding the evolution of this variable show that it can be used as a global gait index during pre-clinical trials, to improve the readability of the results.

This longitudinal study has allowed the description of gait pattern evolutions in GRMD dogs at a time when the disease becomes symptomatic, and has identified quantifiable and sequentially impaired gait variables. The ability of accelerometry to monitor gait evolution in dogs during several months, and to quantitatively reflect the heterogeneity of the disease with a good sensitivity has been demonstrated. The PCA analysis was proposed as a final data analysis method, making the interpretation of multivariate data as simple as their measure. Our study showed that a 3D-accelerometric gait analysis device is ready to be used as a simple and reliable gait follow-up tool during pre-clinical therapeutic trials, providing reliable biomarkers quantitatively showing the heterogeneity of the canine GRMD disease, a characteristic to take into account when evaluating this pre-clinical model, and which represents another feature in common with DMD.